Angling for Influence: Institutional Proliferation in Development Banking

Angling for Inﬂuence: Institutional Proliferation in Development Banking

Tyler Pratt∗ Yale University December 6, 2019

∗Assistant Professor, Department of Political Science, Yale University. Email: [email protected]. I am grateful to Eric Arias, Sarah Bush, Christina Davis, Julia Gray, Kosuke Imai, Robert Keohane, Amanda Ken- nard, James Lee, Melissa Lee, Christoph Mikulaschek, Julia Morse, and Kelsey Pratt for valuable feedback on this project. Abstract Why do states build new international organizations (IOs) in issue areas where many institutions already exist? Prevailing theories of institutional creation emphasize their ability to resolve market failures, but adding new IOs can increase state uncertainty and rule inconsistency. I argue that institutional proliferation can be explained by the failure of existing IOs to adapt to shifts in the distribution of state power. States expect decision-making rules within IOs to reﬂect their material power; when it does not, they construct new organizations that provide them with greater institutional control. To test this argument, I examine the proliferation of multilateral development banks since 1944. I leverage a natural experiment rooted in the allocation of World Bank votes at Bretton Woods to show that the probability of institutional proliferation is higher when power is misaligned in existing institutions. My results suggest that conﬂict over shifts in global power contribute to the fragmentation of global governance. 1 Introduction

Since the end of the World War II, states have constructed international institutions at a breakneck pace. The number of formal international organizations (IOs) grew from less than a hundred in 1950 to over 300 by the year 2000. In the same time period, the volume of multilateral treaties registered with the United Nations increased from roughly 500 to 4,000. This growth has generated significant crowding of governance institutions in issue areas like trade, human rights, and finance, where multiple IOs compete for authority to regulate state behavior. Recently, the proliferation of multilateral development banks has drawn particular scrutiny, as new organizations like the Asian Infrastructure Investment Bank (AIIB) and the New Development Bank (NDB) challenge the primacy of established institutions. The proliferation of IOs is meaningful because it has the potential to undermine international cooperation. International institutions are believed to play an important role in world politics, managing interdependence and facilitating bargaining among states. But the fragmentation of global governance across multiple IOs often leads to coordination problems, redundancy, and rule conflict (Raustiala & Victor, 2004; Alter & Meunier, 2009).1 Lamenting the proliferation of development aid institutions, a large conference of development ministers argued that “the effectiveness of aid is reduced when there are too many duplicating initiatives” and cautioned against “creating separate new channels that risk further fragmentation” (Accra Agenda for Action, 2008). Given the challenges associated with overlapping IOs, why do states continue to construct them? Classic theories of IO formation argue that states build IOs to reduce transaction costs, overcome market failures, and capture gains from cooperation (Keohane, 1984). This rational institutionalist account provides a convincing explanation of IO construction when

1While some argue that institutional crowding does not necessarily undermine cooperation, at the very least sustained proliferation demands a high level of policy coordination among IOs (Gehring & Faude, 2014; Abbott et al. , 2015; Pratt, 2018).

1 few prior institutions exist (such as the formation of the World Bank in 1944), but is less compelling when many institutions are already present (e.g., the creation of the AIIB in 2015). In the latter context, institutional proliferation raises transaction costs by increasing uncertainty and introducing multiple bargaining venues. Rather than a rational response to an underlying cooperation problem, I argue that institutional proliferation often stems from a contest for bargaining power among states. States seek decision-making power in IOs because cooperation involves distributional conflicts. Con- trol over institutions allows states to push negotiations toward their favored outcome. States that are dissatisfied with their influence — often because existing institutions have failed to adapt to changes in the global distribution of power — build new institutions in which they have greater control. Overlapping IOs are a byproduct of state attempts to increase their influence over multilateral outcomes. This argument suggests that institutional proliferation is often rooted in contestation associated with power transitions. When IOs are created, member states typically design formal and informal rules so that multilateral influence reflects states’ underlying material power (Gruber, 2000; Odell, 2000; Stone, 2011). As the distribution of state power shifts, however, institutions do not smoothly adapt (Kaya, 2015; Lipscy, 2017). A power misalignment emerges when a state’s influence within the institution is not commensurate with its unilateral power. States engage in institutional proliferation as part of a strategy to rectify these misalignments. To test the link between power misalignment and institutional proliferation, I examine the creation of multilateral development banks. This regime grew from a single development bank in 1944 (the World Bank) to at least twenty-eight overlapping institutions in 2019. These organizations are important actors in world politics, disbursing over $400 billion in development finance since the year 2000.2 The issue area has a clear functional rationale for

2Calculations reflect OECD data on total flows from international financial institutions.

2 institutionalized cooperation: the coordination of global development finance efforts. It has also experience a substantial amount of proliferation that is difficult to explain on purely functional grounds.3 The power misalignment argument suggests that states are more likely to construct development banks when their influence in existing bodies – and especially the central institution (the World Bank) — falls short of their underlying material power. This very logic was prevalent in media and policy debates surrounding the creation of the Chinese-led AIIB in 2015, which frequently cited Chinese under-representation in the World Bank and IMF. Systematic testing of the argument, however, has been limited. Here, I examine whether the historical pattern of development bank proliferation is consistent with such an explanation. To do so, I collect an original data set on power misalignment in the World Bank and state participation in the founding of alternative development banks. Estimating the effect of power misalignment presents a difficult identification challenge, since state influence in international institutions is politically determined. States’ decision- making power in IOs is shaped by confounding variables like alliance commitments and state preferences over the policy domain governed by the IO. I adopt two approaches to mitigate endogeneity. First, I leverage a natural experiment that occurred during negotiations pre- ceding the World Bank’s creation. To satisfy a political promise to its wartime allies, the United States made an abrupt change to the formula used to allocate vote shares at the 1944 Bretton Woods conference. The formula change represents an exogenous shock to the influence of some World Bank member states, which I use to identify the causal effect of power misalignment on the creation of new development banks. In a second set of tests, I examine dynamic shifts in power misalignment that emerged after the World Bank’s founding. These specifications assess whether a state is more likely

3In a recent paper examining growth in multilateral development banks (MDBs), Kellerman (2019) argues that “the proliferation of MDBs has resulted in an ineﬃcient duplication of international institutions with overlapping functions” (107).

3 to participate in development bank proliferation as its economic growth outpaces its formal vote share in the World Bank. To minimize differences between over- and under-represented states, I employ covariate balancing propensity score matching (Imai & Ratkovic, 2014). Both sets of tests support the theory: states are more likely to construct new banks when power is misaligned in the World Bank. Under-representation in the World Bank by a single percentage point makes a state 6.8% more likely to participate in the creation of new development banks. These results establish the importance of power misalignment in spurring networks of overlapping IOs. The paper contributes to at least two distinct literatures in international relations. By tying IO proliferation to misalignments in power, I build on scholarship examining power transitions and conflict over adaptation of the international order. Over fifty years ago, Or- ganski (1968) highlighted the potential for shifts in state power to engender dissatisfaction and interstate conflict. Gilpin (1981) similarly examined how changes in the distribution of power generate disequilibrium and the potential for systemic war. Subsequent work illuminates how power shifts affect the survival and vitality of existing international institutions (Ikenberry, 2001). Recent studies detail how power transitions are manifested in prominent economic IOs, including the World Bank (Kaya, 2015; Zangl et al. , 2016; Lipscy, 2017). This paper draws on the concepts and theoretical mechanisms common to this literature to explain a new phenomenon – institutional proliferation and overlap – that increasingly characterizes contemporary global governance. In addition, the paper contributes to recent work on the sources and consequences of international regime complexity (Raustiala & Victor, 2004; Alter & Raustiala, 2018). Much of this literature focuses on the implications of institutional crowding for interstate bargaining and cooperation (Alter & Meunier, 2009; Davis, 2009; Keohane & Victor, 2011; Gehring & Faude, 2014). Fewer studies offer a theory of regime complex emergence: an explanation

4 for why and when states construct overlapping IOs.4 The paper complements recent work that privileges power relations in international regime complexes (Drezner, 2009; Henning, 2017; Pratt, 2018), and it applies rigorous, quantitative empirical testing to a predominantly qualitative literature. The following section outlines the theoretical logic linking power misalignment to institutional proliferation, in which states create overlapping IOs to rectify imbalances in multilateral inﬂuence. Section 3 describes the regime for development lending, presents the paper’s primary hypothesis, and describes the dataset of multilateral development bank proliferation. Section 4 presents empirical results that identify a strong eﬀect of power misalignment on the creation of new development banks, and section 5 concludes.

2 Inﬂuence and Institutional Proliferation

The main argument of this paper is that state competition for bargaining power, rather than an attempt to maximize gains from cooperation, often drives states to construct new IOs. Formal and informal IO decision-making procedures are typically designed to align multilateral inﬂuence with states’ underlying material power. When that alignment is perturbed – either by changes in IO procedures or shifts in state power – under-represented states will demand institutional reform. If the misalignment persists, dissatisﬁed states will construct a new, overlapping institution which provides them with greater control.

2.1 Why Power Misalignments Emerge and Persist

Although this paper focuses on the effect of power misalignment, it is worth briefly consid- ering how such discrepancies emerge in the first place. Power misalignments are a common

4Exceptions include Morse & Keohane (2014), who include “competitive regime creation” in their study of contested multilateralism, and Kellerman (2019), who explicitly examines the proliferation of multilateral development banks.

5 feature of international institutions. States’ underlying material power and institutional influence are often well aligned when an organization is initially created. Misalignments emerge over time when there are large shifts in states’ underlying material power and the IO does not adapt its decision-making procedures in a commensurate manner. The constant evolution of member state capacity, combined with the slow pace of institutional reform, result in a lag between changes in members’ material power and IO decision-making procedures. Many misalignments are corrected in a short time period, forestalling the threat of institutional proliferation. Others persist and grow more severe, generating intense dissatisfaction among under-represented states. A growing literature examines why international institutions fail to accommodate member state demands. Much of this works envisions institutional adaptation as a rational bargaining process, subject to well-known sources of bargaining failure. Morse & Keohane (2014) and Zangl et al. (2016), for example, point to states’ inability to resolve information asymmetries as a common source of institutional bargaining failure. Others argue that adaptation is stymied when an IO is captured by bureaucrats that are opposed to reform (Hawkins et al. , 2006; Urpelainen & de Graaf, 2013; Johnson, 2014). Three recent contributions specifically interrogate when multilateral development banks adjust to the demands of dissatisfied member states. Kaya (2015) and Kellerman (2019) argue that development finance institutions have particular characteristics that limit adaptation. For example, the ability of the International Bank for Reconstruction and Develop- ment (IBRD) to raise funds from private financial markets insulates it from member state demands, compared to IOs that rely on immediate state financing (Kaya, 2015). Kaya also cites the path dependence of IBRD voting rules that stems from its original relationship with the IMF – a historical process detailed in Section 4. Due to these institutional features, power misalignments in the IBRD have been larger and more persistent than in the IMF. Kellerman (2019) similarly argues that funding constraints limit states’ ability to impose reforms on development banks. Many states that seek to challenge existing development

6 banks often have poor credit and limited capital reserves, forcing them to invite outside partners when they create overlapping institutions. These partners insist that the new IO clings to status quo modes of operation, preventing much of the institutional adaptation initially sought by the dissatisfied coalition. Lipscy (2017) also examines power misalignment in the IBRD. His analysis challenges the results of Kaya (2015), demonstrating that the IBRD has been more accommodating to shifts in member state power than the IMF.5 To explain IBRD adaptation, Lipscy credits the large number of competing development finance institutions. States demanding greater influence in the IBRD can leverage other development banks as outside options, strengthening their hand in negotiations.6 This argument suggests that the broader institutional environment, rather than the design features of individual IOs, determines when power misalignments are resolved via bargaining. If Lipscy is correct, power misalignments in the IBRD should become smaller over time as states layer competing development banks onto the issue area.

2.2 Misalignment and Proliferation

Whatever the underlying cause, a failure of institutional adaptation is a necessary condition for power misalignments to emerge and persist in IOs. The logic linking power misalignment to institutional proliferation begins with two assumptions. The first is that states prefer greater influence over the activities and policy decisions of international organizations. States value IO influence for several reasons. The primary benefit is increased bargaining power over the specific issues negotiated within an IO. IOs may help reach efficiency-enhancing bargains, but member states often disagree on “which point along the Pareto frontier should

5Differences in the empirical findings may stem from the time periods examined by each author. Kaya (2015) analyzes institutional adaptation by the IBRD and IMF in the 2008-2010 reforms; Lipscy (2017) examines the period 1975-2014. 6Lipscy (2017) argues that features exogenous to the issue area (“policy area” or “network effects”) make it feasible for states to construct multiple development banks alongisde the IBRD and only one crisis finance institution, the IMF.

7 be chosen” (Krasner, 1991, 43). Greater influence increases the probability that multilateral negotiations will produce outcomes that reflect a state’s policy preferences. Several second-order benefits compound states’ desire for control. These include the ability to steer cooperative benefits associated with the IO, from trade flows to military protection, to a state’s allies. In effect, states can exploit influence in IOs for broader geopolitical leverage (Fleck & Kilby, 2006; Kuziemko & Werker, 2006; Davis & Pratt, 2019). Another second-order benefit is the prestige associated with institutional control. States – and particularly rising powers – associate influence in IOs with status in world politics (Paul et al. , 2014). The degree of control awarded to member states is one clear indicator of their position in the hierarchy of world powers. These forces provide ample motivation for states to seek greater influence when possible. The second assumption is that states expect the distribution of multilateral influence to reflect members’ underlying material power. The notion that states use unilateral power as a reference point is consistent with extant research on negotiation and bargaining power in IOs. Gruber’s concept of “go-it-alone power” (2000), Odell’s “best alternative to a negotiated agreement” (2000), and Stone’s “structural power” (2011) all link states’ influence in IOs to their ability to act outside the institution. The assumption also finds support in diplomatic and policy debates regarding the appropriate distribution of IO decision-making power. Under-represented countries and their supporters routinely argue that the allocation of multilateral influence should be brought into alignment with states’ unilateral power. They often explicitly cite measures of material power when making their case. For example, a 2015 editorial lamented that “China has a 3.8% voting share in the IMF and World Bank, even though it accounts for more than 12% of world GDP” (Sheng & Geng, 2015). While these assumptions will not hold true for all countries in all cases, they are nonethe- less useful building blocks for theorizing about institutional proliferation. Because states expect decision-making procedures to reflect their capacity to act outside the IO, power

8 misalignments generate dissatisfaction among under-represented states. These states will agitate for greater control in the institution. Sometimes, their demands will be satisfied as IO decision-making power is redistributed in their favor.7 Often, however, the IO will fail to adapt, and disadvantaged states will take disruptive action to rectify power imbalances. Specifically, they may create entirely new IOs with more favorable distributions of influence to compete with existing bodies. Put simply, states strategically construct new IOs to increase their influence over multilateral outcomes. The independent variable in the theory is the degree of alignment between a state’s influence in multilateral institutions and its underlying material power. The variable includes two components which require formal definition. I define “underlying material power” as the national resources available to help a country achieve its desired outcome via unilateral action. The relevant national resource will vary by issue area. For security institutions, the key resource is often military strength; for trade and financial institutions, it is economic capacity. I define “influence in existing institutions” as the institutional procedures that help a state achieve outcomes via multilateral action. These procedures determine how the preferences of member states are aggregated into an institutional outcome, such as a treaty, development loan, or authorization of military force. The procedures may be formal or informal, though in the empirical analysis below I focus on states’ formal vote shares in IOs. The dependent variable, institutional proliferation, is defined as state participation in the creation of IOs that overlap with existing institutions. The independent and dependent variables are linked through a functional logic. Under-represented states engage in institutional proliferation to rectify power misalignments. This implies that new institutions must bestow its creators with additional influence in the regime. I argue that new IOs provide influence in two ways. First, proliferating states design new institutions to give themselves greater

7Keohane & Nye (1977) argue that the rules within an international regime will be renegotiated when bargaining power becomes misaligned.

9 decision-making power than they have in existing IOs. When the Asian Development Bank (ADB) was created in 1966, for example, Japan was awarded more than 20% of formal vote shares (compared to less than 3% in the World Bank). This level of control reflects Japan’s status as a founding member and architect of the new institution. Second, IO proliferation reshapes influence more broadly by offering some states an additional outside option during multilateral negotiations. A state with a credible outside option gains bargaining leverage, shifting negotiation outcomes in its favor (Hirschman, 1970; Voeten, 2001; Schneider, 2011). Other scholars have noted how institutional exit options can potentially alter bargaining power among states (Helfer, 2004), including among development banks (Lipscy, 2017).8 According to this logic, the creation of the ADB should grant Japan additional influence over lending decisions in the World Bank, since it can credibly threaten to shift some proposed programs to the ADB. If executed successfully, institutional proliferation generates additional influence in both the new IO and in legacy institutions. However, the strategy also entails costs that constrain its use by states. The first is the difficulty of amassing a coalition of states to join a new organization. Institutional proliferation is not a unilateral act; it requires the participation of multiple states. The operational success and perceived legitimacy of a new IO grows as it attracts more members, increasing the need to amass a large coalition. Powerful states can buy off collaborators through concessions and side payments, but constructing a new organization is significantly easier if there is a large set of potential partners that are dissatisfied with the current regime. A second constraint is the efficiency costs associated with institutional proliferation. Efficiency costs represent the erosion of cooperative gains that occur as governance of an issue area is fragmented across multiple IOs. These costs vary in size and incidence across

8Countries do not usually “exit” IOs in the sense of relinquishing their membership. They can shift negotiations on speciﬁc issues to another fora, however, which will generate additional bargaining leverage.

10 issue areas. The proliferation of trade institutions, for example, generates relatively small efficiency costs. New trade treaties do not undercut cooperative gains from existing trade institutions, at least from the perspective of states participating in the new treaty. A new institution may lead to trade diversion, but these losses are imposed on states left out of the new agreement. Because the states engaging in institutional proliferation bear no losses, efficiency costs are not a strong constraint on the proliferation of trade agreements. In other issue areas, the creation of new IOs can create costs for proliferators. The construction of new development banks, for example, tends to shift power from lending states (i.e., those that provide funding for the institution) to borrowing states. Each additional bank provides borrowers with another potential lender, facilitating forum-shopping and undermining the monopoly power of donor states. Since proliferating states usually become the primary lenders in a new development bank, they indirectly bear some of these costs.9 Because different issue areas feature disparate returns to fragmentation, we should expect significant issue-area variation in proliferation. When deciding whether to create overlapping IOs, states weigh the benefits of greater control against the material and efficiency costs of a new institution. Power misalignments affect this calculus by increasing the perceived value of multilateral influence. When states become sufficiently dissatisfied, they are willing to bear the material and efficiency costs – in effect, sacrificing cooperative gains – in order to strengthen their influence over multilateral outcomes.

2.3 Theoretical contribution

The logic of power misalignment builds on a long-standing literature examining power sharing in multilateral institutions. Gruber (2000) argues dominant states can dictate outcomes

9Lipscy (2015) argues that the proliferation of development banks does not necessarily undermine the function of the regime, compared to other issue like balance of payments lending.

11 in IOs because of their capacity to act alone. Stone (2011) describes how powerful states cede formal control but exert informal influence at pivotal moments. Ikenberry (2001) claims that prudent great powers intentionally relinquish multilateral influence to gain the acquiescence of weaker states in a broader international order. Gilpin (1981) explicitly examines how misalignment in the international system can generate hegemonic wars that rebalance the international order. A similar logic drives states concerned about their influence in IOs to proliferate new institutions.10 As this work has demonstrated, control of multilateral institutions is carefully distributed among member states to maintain their support for the regime. When power sharing arrangements become inequitable, or are perceived as mechanisms of great power domination, support for the institution breaks down. The argument also contributes to theoretical efforts distinguishing the process of institutional proliferation, in which states layer overlapping IOs one on another, from de novo institutional formation (e.g., Kellerman 2019). The dominant theory of institutional formation is the functional logic developed by Keohane (1984): states construct institutions to lower transaction costs and facilitate cooperation. This account struggles to explain IO proliferation, since cooperative outcomes often suffer as issue areas become crowded with institutions.11 Existing theories of IO proliferation emphasize competing preferences among member states (Urpelainen & de Graaf, 2013; Morse & Keohane, 2014). However, states may have motives to create overlapping IOs even when existing organizations operate in a manner consistent with their preferences. Influence over multilateral outcomes brings states unique benefits, including prestige and the ability to extract side payments, that cannot be achieved without control. The power misalignment account illuminates how concerns about

10Henning (2017) also links concerns about control to institutional overlap, though in his account states are concerned with limiting agency drift among institutions, not power imbalances among member states. 11For example, Raustiala & Victor (2004) describe “legal inconsistencies” in the regime complex for plant genetic resources (300), Helfer (2009) ﬁnds intellectual property institutions adopting a “competing regulatory approach,” (40), Davis (2009) notes “the potential for contradictory legal rulings” among trade institutions (25), and Pratt (2018) highlights conﬂicting standards in counterterrorism IOs.

12 inﬂuence encourage institutional proliferation even when states concur on the basic goals of the regime.

3 The Multilateral Development Lending Regime

To test the effect of power misalignment on institutional proliferation, I examine the creation of multilateral development banks since 1944. Development lending is both a substantively important and a puzzling case for testing the effect of power misalignment. It is a highly salient issue area with a substantial amount of proliferation. States have created dozens of global, regional, and sub-regional multilateral development banks since the World Bank was established in 1944. At the same time, it is difficult to explain the pattern of proliferation on purely functional grounds. The crowding of development banks induces inefficiencies as institutions engage in redundant efforts to screen proposals, negotiate with borrowing countries, and audit funded projects. Development banks have recognized these problems and spend significant time and effort attempting to coordinate activities. And while proliferating states often claim that new banks will fill unmet development needs in a drastic departure from prior institutions, they often closely replicate the activities of existing bodies. If states’ primary goal in development lending is to maximize cooperative gains, we should not expect a significant amount of institutional proliferation. However, a large literature examining the politics of the World Bank and international financial institutions (IFIs) suggests that states have more self-interested goals. Two findings from this research program are relevant for understanding institutional proliferation. First, IFIs are inherently political institutions; they distribute finance on the basis of “high politics” at least as much as technical need (Frey & Schneider, 1986; Thacker, 1999; Stone, 2011; Dreher et al. , 2009; Copelovitch, 2010; Kersting & Kilby, 2016). Second, influence in IFIs is contested and highly sought after by states (Krasner, 1981; Zangl et al. , 2016). In both the World Bank and regional development banks, states use their influence in support of broader foreign policy

13 goals (Fleck & Kilby, 2006; Lim & Vreeland, 2013). States also trade influence in the World Bank to buy votes in other multilateral instituions (Dreher & Sturm, 2012). The findings from this literature confirm a key assumption of the power misalignment theory: states care deeply about their influence in IFIs and will seek to augment that influence when possible. Finally, development banking provides several advantages for large-N empirical tests. The presence of a focal institution (the World Bank) that distributes formal voting shares to members facilitates the measurement of states’ influence in the existing regime. The issue area also provides an opportunity for causal identification rooted in the allocation of vote shares at Bretton Woods, the multilateral conference that created the World Bank and International Monetary Fund (discussed in Section 4.2).

3.1 Evolution of the Regime Complex

Multilateral development lending began in 1944, when states created the International Bank for Reconstruction and Development (IBRD), commonly known as the World Bank. The main impetus for the World Bank was the need to finance European reconstruction after World War II. Over time, the Bank shifted its emphasis from post-war reconstruction to economic development, and it became primarily a source of development finance for less developed countries. Decision-making power in the Bank is determined by states’ formal vote shares, which are distributed unequally among member states. These vote shares are tied to the capital contributions that states are required to make to the Bank, though in later years much of the Bank’s capital came from private finance rather than state contributions. For the first decade of its existence, the World Bank was the only large multilateral development lending institution. Beginning in the mid-1950s, however, coalitions of states began to construct additional development banks. Many of these early banks were associated with new or existing IOs. In 1956, for example, members of the Council of Europe created a development bank of their own. Two years later, signatories of the Treaty of Rome created

14 the European Investment Bank (EIB). The debate over the creation of the EIB revealed how concerns over influence shaped the decision to construct the new bank. France, the most vocal advocate of the EIB, argued that without a bank of their own European states would “soon depend entirely on the United States,” the dominant vote holder in the World Bank (Bussiére,2008, 32). Notably, the EIB was strongly opposed by the United Kingdom, which enjoyed the second largest vote share in the World Bank, on the grounds that it “would have duplicated the Washington institution” (Bussiére,2008, 33). In 1959, states in the Western Hemisphere created the Inter-American Development Bank (IADB), followed the next year by the Central American Bank for Economic Integration. Architects of the IADB cited the “low level of representation of Latin American countries in the existing financial institutions” as a primary justification for the new bank (Diaz-Bonilla & del Campo, 2011, 59). Two other large regional institutions, the African and Asian Development Banks, were created in 1965 and 1966, respectively. These banks focused their lending activities on specific geographic regions, though membership was generally not restricted to regional states.. Like the World Bank, these institutions typically employed weighted decision-making rules that allocated unequal influence to member states. However, the distribution of influence in the new banks often departed significantly from the World Bank. In the Asian Development Bank, for example, regional powers sought to limit Western influence over the institution’s operation. According to Wilson, “The Asian feeling at the time was that the World Bank was dominated by ’Anglo-Saxons’ – by the Western powers led by the United States and Britain...The ambition was therefore to have the Americans not as the largest shareholder but as equal with Japan” (1987, 6). By 1993, states had created at least twenty-four IOs that participated in development lending alongside the World Bank. These included sub-regional institutions like the Arab Fund for Economic and Social Development, Caribbean Development Bank, and Nordic De- velopment Fund, as well as development banks emanating from existing institutions (e.g., the

15 OPEC Fund for International Development). Institutional proliferation increased again in the late 2000s, as many developing countries led a series of efforts to build banks which gave them greater control over lending decisions. Kazakhstan and Russia created the Eurasian Development Bank in 2006. In 2013, the BRICS countries founded the New Development Bank, intended “as an alternative to the existing US-dominated World Bank.”12 The following year, 21 Asian states joined a Chinese-led effort to create the Asian Infrastructure and Investment Bank (AIIB), focused on infrastructure lending in Asia. Despite a lobbying cam- paign by the United States to prevent its allies from joining the AIIB, thirty-six additional countries (including the United Kingdom, Australia and many European states) signed onto the new bank as founding members. Table 1 summarizes the observed cases of institutional proliferation in the regime complex for development lending. The table displays all multilateral development lending institutions and their dates of formation. In the final column, I list the number of founding members who joined each bank in the year of its creation.

3.2 Measurement and Data

I argue that states are motivated to build new multilateral development banks due to power misalignments in existing institutions — i.e., a divergence between states’ inﬂuence in the development lending regime and the underlying distribution of material power. When member states are unable to win greater inﬂuence in the existing regime, they face a choice. They can continue to support the current institution despite their lack of control. Or they can

12“About the New Development Bank,” http://ndbbrics.org

16 Institution Date Founders International Bank for Reconstruction and Development 1944 26 Council of Europe Development Bank 1956 14 European Investment Bank 1958 6 Inter-American Development Bank 1959 18 Central American Bank for Economic Integration 1960 5 African Development Bank 1965 24 Asian Development Bank 1966 27 East African Development Bank 1967 3 Arab Fund for Economic and Social Development 1968 11 Andean Development Corporation 1968 4 Caribbean Development Bank 1970 6 Islamic Development Bank 1973 27 West African Development Bank 1973 7 Development Bank of Central African States 1975 4 Arab Bank for Economic Development in Africa 1975 17 Development Bank of the Great Lakes States 1976 3 OPEC Fund for International Development 1976 13 Nordic Investment Bank 1976 5 International Fund for Agricultural Development 1977 84 Eastern and Southern African Trade and Development Bank 1985 14 Nordic Development Fund 1989 5 European Bank for Reconstruction and Development 1991 38 Black Sea Trade and Development Bank 1992 11 North American Development Bank 1993 2 Economic Cooperation Organization Trade and Development Bank 2005 3 Eurasian Development Bank 2006 3 New Development Bank 2013 5 Asian Infrastructure Investment Bank 2015 57

Table 1: Proliferation of Multilateral Development Banks

17 begin the costly process of constructing a new bank where institutional control more accurately reflects economic power. The theory implies that states are more likely to make the latter choice when they confront larger power misalignments. This logic underpins the Power Misalignment Hypothesis tested below: states are more likely to build new development banks when their underlying material power exceeds their influence in existing institutions. The empirical tests will examine whether power misalignment predicts the creation of new development lending institutions. To test this question, I collect data on the proliferation of development banks as well as states’ influence in the central development lending institution, the World Bank. The dependent variable is state participation in development bank proliferation. To operationalize proliferation, I identify twenty-seven unique development banks that were created after the establishment of the World Bank. I then construct a dichotomous variable, Institutional Proliferation, which takes a value of one if a state becomes a founding member of a new development bank. A state qualifies as a founding member if it joins a new bank in its first year. Appendix Table A1 shows a list of founders for each bank. Institutional proliferation occurs in 4.4% of 9,020 state-year observations. The independent variable is power misalignment in existing institutions. This variable requires measuring two components: states’ influence in the multilateral development lending regime and their underlying material power. To measure multilateral influence, I collect data on states’ formal vote shares in the World Bank. The World Bank is the clear focal institution in the regime. It is the largest multilateral development bank in terms of lending, personnel, and bureaucratic expertise.13 Vote shares in the Bank therefore provide a reasonable measure of states’ influence over multilateral development lending. Vote shares vary significantly both cross-nationally and within countries over time. Figure 1 displays the distribution of World Bank vote shares in 2014 for the 30 most influential states. The United States is the dominant power in the institution, with slightly over 15%

13Figure A1 in the Appendix shows the World Bank’s share of total development assistance since 1970.

18 USA Japan China Germany UK France India Canada Italy Saudi Arabia Russia Spain Netherlands Brazil Switzerland Belgium Iran South Korea Australia Turkey Venezuela Mexico Indonesia Argentina Sweden Denmark South Africa Kuwait Austria Norway

0.00 0.05 0.10 0.15

World Bank Vote Power (% of Total Votes)

Figure 1: World Bank Vote Share, 2014. The figure displays the 30 states with highest share of formal votes in 2014. Data are from World Bank annual reports. of its vote share. Appendix Figure A2 depicts how votes have changed over the last 35 years for select states. Countries that experienced rapid economic growth tended to receive more votes in this period (e.g., China and Japan) while others saw their relative influence reduced (United States, United Kingdom). As a measure of underlying material power, I use states’ share of global Gross Domestic Product (GDP). GDP is a common measure of economic power and a reasonable proxy for states’ unilateral ability to provide development finance. Policymakers and scholars who criticize the power structure of economic institutions, including the World Bank, often reference states’ share of global GDP (e.g., Sheng & Geng 2015).

19 The independent variable, power misalignment, measures the difference between a state’s influence in the existing regime and its underlying material power. Figure 2 depicts this relationship by plotting 2014 World Bank vote share (Y-axis) for a subset of states against their share of global GDP (X-axis). As the figure shows, some states (e.g., Turkey, France) have influence in the World Bank that is well aligned with their underlying economic capacity. Others appear to “punch above their weight,” with vote power outstripping their economic might (e.g., Saudi Arabia, the Netherlands). A few states are less influential in the World Bank relative to their economic power (e.g., China, Mexico). These are the states I expect to engage in institutional proliferation.

I use the data displayed in Figure 2 to create the variable Vote Power Misalignment. It calculates the difference between state i’s share of total GDP among Bank members in year t and its share of Bank votes in the same year: GDP shareit − World Bank vote shareit. Higher values (> 0) indicate that a state is under-represented: its underlying economic power exceeds its influence in the World Bank. For example, Brazil represented 3.1% of Bank members’ GDP in 2014, but only received 1.7% of votes in the organization, resulting in a Vote Power Misalignment of 1.4. I expect this variable to have a positive effect on participation in institutional proliferation. While my argument prioritizes concerns about power misalignment, this is not the only reason that states might seek to construct alternative development banks. For example, demand for new institutions may stem from state preferences over more immediate policy outcomes, rather than the institutional procedures that generate those outcomes. If policy dissatisfaction is a primary driver of institutional proliferation, states that object to the World Bank’s pattern of lending will be more likely to create new development banks. By the same logic, states will be less likely to engage in institutional proliferation as long as lending decisions conform to their preferences. Concerns about policy and power are not mutually exclusive, and it is likely that the

20 ● Japan 0.08 0.06

● China ● Germany ● ●

0.04 France UK

India World Bank Vote Power Bank Vote World ●

Saudi Arabia ● Netherlands● Italy ● Belgium 0.02 ● ● Turkey Brazil ● South Africa ● ● Mexico 0.00

0.00 0.02 0.04 0.06 0.08

Relative GDP

Figure 2: World Bank Vote Share vs. Share of GDP, 2014. Select states are plotted according to their share of World Bank votes (Y-axis) and global GDP (X-axis). policy dissatisfaction mechanism operates alongside the power misalignment logic.14 In the empirical analysis, I control for two possible dimensions of state satisfaction with World Bank lending decisions. First, states may care about their own access to development finance. In this scenario, a coalition of states whose development needs are insufficiently addressed by existing institutions construct a new bank to fill the gap. To account for this possibility, I control for the (logged) annual flow of development assistance disbursed to each state by

14Because policy outcomes are partly a function of the distribution of inﬂuence in multilateral institutions, power misalignment is causally prior to policy dissatisfaction. Controlling for policy satisfaction therefore risks inducing post-treatment bias. In the tables below, I present speciﬁcations with and without these controls.

21 existing institutions.15 Since many development banks have a regional focus, I also control for total aid given to the state’s geographic region.16 In addition to demand for finance, an excess supply of capital could lead states with large financial reserves to form new banks to disburse finance. I control for the supply of capital with a measure of annual outgoing bilateral aid flows. Second, states may prefer that World Bank loans are disbursed to their allies. Exist- ing work emphasizes geopolitical contestation over World Bank lending patterns (Frey & Schneider, 1986; Thacker, 1999). If a high percentage of World Bank financing flows to a state’s allies, it should be less likely to engage in institutional proliferation. As World Bank loans depart from a state’s ideal distribution, the probability of proliferation will increase. To account for states’ geopolitical preferences, I control for the proportion of World Bank loans that are distributed to a state’s formal allies.17 Additional controls address state-level features that influence the probability of institutional proliferation. I include both GDP and GDP per capita to account for variation in states’ capacity to construct new institutions. An indicator for democratic political institutions reflects the higher propensity of democracies to create new IOs. Measures of states’ national military capability control for the potential use of coercion to build a coalition of proliferating states. I account for the possibility that states form development banks to facilitate regional economic integration by controlling for the proportion of each state’s total trade that is conducted with regional partners.18 I also include fixed effects for each region to account for varying baseline propensities of some regions to form new banks. I include a count of development banks previously joined to address the extent to which states have

15Data on aid from existing institutions comes from the AidData dataset (Tierney et al. , 2011). 16Regional categories are from the World Bank. 17Alliances are coded according to the Correlates of War Formal Alliances dataset. 18I include an analogous variable at the region level (the proportion of trade that is conducted intra-region).

22 already broadened their membership across the regime complex. Finally, I account for time dependence with a count of years since a state last engaged in institutional proliferation, along with a cubic polynomial (Carter & Signorino, 2010). Despite the inclusion of these controls, endogeneity remains a significant threat to infer- ence. World Bank votes are not randomly assigned; they arise from a bargaining process inextricably linked to states’ political power and preferences for development lending. Unob- served factors likely influence both vote power misalignment and states’ propensity to create new development banks. This could generate bias in either direction, depending on how votes are allocated. If powerful states bargain for greater influence and can also more easily construct new institutions, regression estimates will be biased in a negative direction. On the other hand, if states that are devoted to World Bank lending practices are given more influence in the Bank, and these same states resist the creation of new development banks, estimates will be biased in a positive direction. To reduce the risk of a spurious finding, I adopt two strategies that leverage different sources of variation in power misalignment. I begin with an instrumental variable approach that exploits the initial power misalignment at the time of the World Bank’s founding. To implement this strategy, I identify a natural experiment that occurred during negotiations regarding the formation of the World Bank. Late in this process, the Bank’s architects switched formulas for allocating votes to member states, providing plausibly exogenous variation in

Vote Power Misalignment among many countries. This strategy requires restricting the sample of states, but accurately estimates the causal effect of power misalignment in the presence of unobserved confounders. Having established a causal relationship between power misalignment and institutional proliferation, I turn to a fixed effects specification. This approach leverages dynamic shifts in power misalignment that emerge from differential rates of economic growth among World Bank members. The fixed effects model represents a tradeoff. It lacks the clean causal

23 identiﬁcation of the instrumental variable analysis, though it can still account for sources of bias that are constant at the state and year level. Notably, it allows us to examine the relationship between power misalignment and the creation of new banks in a more precise way and among a broader sample of states. Both speciﬁcations yield results consistent with the logic of power misalignment.

4 Empirical Tests

4.1 IV: Vote Share Allocation at Bretton Woods

Serious planning for the institution that would become the World Bank started in the early 1940s, in the midst of the second World War. American policymakers, anticipating the need for European reconstruction after the war, drafted designs for a “Bank for Recon- struction and Development of the United and Associated Nations” in 1942 (White, 1942). Primary responsibility for planning the bank, as well as an International Stabilization Fund (which would become the IMF), was given to Department of Treasury oﬃcial Harry Dexter White. White’s inﬂuence on the Bank was profound; his thinking shaped its goals, structure, and decision-making procedures. From 1942 until the Bretton Woods Conference in July 1944, White worked closely, and often contentiously, with his British counterpart, the noted economist John Maynard Keynes, to put plans for the bank into action (Steil, 2013). White’s professional correspondence provides a detailed picture of the Bank’s origins, including his plans to distribute vote shares among member states. Initial drafts of the bank, prepared in a series of memos for Treasury Secretary Morgenthau, describe a highly technical institution, with each member state assigned a mathematically determined share of stock in the bank.19 The initial formula for distributing shares was simple: each member

19States with more shares of stock were required to commit greater capital resources, and they were also given greater inﬂuence over lending decisions via increased vote power.

24 would contribute “2 percent of its estimated national annual income” and receive “50 votes plus one vote for each share of stock held” (White, 1942). This formula for allocating votes had a clear basis in states’ economic power: those with a higher national income would contribute more capital and therefore enjoy more influence in the organization. As White began to negotiate the terms of the bank with US allies, it became clear that the technical formula would have to bend to political realities. The UK, which was comparatively limited in national income but a had a large trade volume, lobbied the United States to add international trade as an element in the vote share formula (White, 1943). White fielded similar requests from China and the USSR, the other two major US allies in the war effort. To complicate matters, US officials decided it would be too onerous to negotiate separate vote shares in the World Bank and IMF; instead, they would come up with a single distribution of decision-making power that would apply to both institutions. As negotiations proceeded, U.S. officials realized that the effort would not succeed without first achieving a politically feasible distribution of votes among the “Big Four” allies: the US, UK, USSR, and China. Accordingly, the contributions of the Big Four were negotiated at the highest political level. The United States would have the largest capital quota of approximately $2.9 billion, the UK half that amount, and the USSR and China slightly less. As in White’s initial draft of the bank, voting power was tied directly to these capital contributions. Having reached a difficult agreement among these fundamentally important states, White sought to limit negotiations with the other 40 states who would attend the Bretton Woods conference. His strategy for doing so was identify a “scientific formula” to govern the allocation of votes. Finding a formula that respected the prior political decision among Big Four, however, proved challenging. White assigned this task to a Treasury aide named Raymond Mikesell (Mikesell, 1994). Mikesell details the assignment from White in his memoir of Bretton Woods. White instructed Mikesell to construct a formula using four variables: national income, foreign

25 trade (exports and imports), gold reserves, and dollar holdings. He “gave no instruction on the weights to be used,” but insisted the formula accurately reflect the agreed upon vote shares for the Big Four (Mikesell, 1994, p. 22). Mikesell went through “dozens of trials” (p. 22) to find a formula that satisfied these constraints. Eventually, he developed one that White deemed close enough to the political agreement forged among the Big Four. Unfortunately, the formula was difficult to justify on any rational basis. According to the Mikesell formula, a country’s capital quota in the World Bank would be calculated by first taking the sum of 4 quantities: 2 percent of national income, 5 percent of gold and dollar holdings, 10 percent of average imports, and 10 percent of the maximum variation in exports from 1934-1938. This sum was then multiplied by the ratio of average exports from 1934-1938 to national income in 1940 to get a state’s final quota (Mikesell, 1994, p. 23). Stone (2011) describes the formula as an attempt by the United States “to cloak in technocratic calculations its political judgments about what share of control it was necessary to cede to each of the great powers in order to secure their participation” (53). While it accomplished this task, the formula also entailed a tradeoff: it arbitrarily shifted the allocation of votes for member states outside the Big Four allies. White and Mikesell were aware of this fact and took pains to hide the details of the formula from potential member states. Their concern is apparent in White’s personal correspondence. Responding to a memo from Mikesell detailing a proposed vote formula, White scribbles in the margins: “Deny it ever existed!” (Mikesell, 1943). As the US and its allies prepared for the Bretton Woods conference, they circulated the proposed distribution of vote shares to participating states, but withheld Mikesell’s formula.20 At Bretton Woods, Mikesell’s quotas were used as starting points for negotiation. States were permitted to issue protests, and some succesfully lobbied for increases in vote power

20Mikesell reports that despite his insistence that vote shares were derived scientiﬁcally, delegates “were intelligent enough to know that the process was more political than scientiﬁc” (Mikesell, 1994, p. 23).

26 over their initial allocation. It appears that deviations from Mikesell’s formula were fairly limited, however. One important reason was that the United States maintained tight control of the “Committee on Quotas” at the conference. The committee was chaired by Treasury official Fred Vinson and advised by Mikesell himself (Mikesell, 1994). This historical episode reveals the high degree of salience that states placed on maximizing their influence in the new institution. For some states, it also provides a source of plausibly exogenous variation in the assignment of World Bank vote shares. In the analysis below, I exploit the “randomness” of Mikesell’s formula, calculating the change in vote shares that arose by shifting from the original formula (2% of national income) to the convoluted one that was ultimately used at Bretton Woods. This shift in votes emerged due to a separate political bargain among the Big Four allies, and thus was clearly endogenous to the power and interests of those states. For most other World Bank members, however, the shift in influence was arbitrary. Through no effort of their own, some states were awarded more influence in the bank than would be justified by their underlying material power. Others saw their influence reduced. Because vote shares in the World Bank are path dependent, this shift in vote allocation had a lingering effect on their control of the institution. The initial change in vote shares can therefore be used as an instrument for states’ bargaining power in the World Bank.

IV Assumptions

The instrumental variables approach entails three assumptions. The ﬁrst, exogeneity, requires that the instrument’s eﬀect on the explanatory variable is independent of potential outcomes. Here, the assumption requires that the “extra” vote power a state received (or lost) due to the Mikesell formula is unrelated to determinants of development bank proliferation, such as states’ political power or preferences for development lending. This is clearly not true for the Big Four allies, whose political power prompted the new vote formula.

27 For other states, however, the change in votes arising from Mikesell’s formula is plausibly exogenous to political considerations. Two pieces of evidence support the exogeneity assumption. The first is the historical record, which establishes that White and Mikesell searched for a vote quota formula with a single goal in mind: to match the political promise made to the Big Four allies. Mikesell is clear that the only instructions White provided were related to these four states; other considerations were ignored as he adjusted the formula to achieve the correct weighting among the US, UK, USSR, and China.21 The fact that the formula was hidden from other states at Bretton Woods affirms the notion that their interests were not reflected in its construction. Second, the data reveal no significant correlation between the change in votes and measures of political or economic power. States who gained influence from the Mikesell formula and those that lost votes exhibit no significant differences in economic power (GDP), level of development (GDP per capita), military power (CINC score), or political regime (Polity score) (see Figure A3 in the Appendix). Notably, the exogeneity assumption does not require that the ultimate allocation of votes was apolitical. In practice, the Mikesell formula was used as a guideline to start negotiations at Bretton Woods, with actual votes deviating based on subsequent bargaining (Stone, 2011). That later deviation is undoubtedly a function of state power and is excluded from the instrument, which only exploits the initial change in votes arising from the shift to Mikesell’s formula. The second IV assumption, the exclusion restriction, requires that the instrument affects the outcome only through its impact on the independent variable. This means that

21Mikesell (1994, p. 22-23) states: “White gave no instructions on the weights to be used, but I was to give the United States a quota of approximately $2.9 billion; the United Kingdom (including its colonies), about half the U.S. quota; the Soviet Union, an amount just under that of the United Kingdom; and China, somewhat less...I confess to having exercised a certain amount of freedom in making these estimates in order to achieve the predetermined quotas. I went through dozens of trials, using diﬀerent weights and combinations of trade data before reaching a formula that satisﬁed most of White’s objectives.”

28 the change in votes due to Mikesell’s formula must aﬀect states’ propensity to build new banks by altering power misalignment, rather some alternative causal pathway. This assumption cannot be tested formally, though I use sensitivity analysis to ensure the results are robust to reasonable violations. However, the construction of the instrument and independent variable help allay some common concerns about the exclusion restriction. The primary eﬀect of the instrument (shift in vote shares) is to change the independent variable

(Vote Power Misalignment), because the former is a direct mathematical input into the latter. A violation of the exclusion restriction would require that votes gained or lost due to Mikesell’s formula generate an effect separate from states’ influence in the World Bank. Finally, the third assumption (monotonicity) requires that the instrument affects all observations in the same direction. This assumption is violated if an increase in vote power decreases the propensity of some states to engage in institutional proliferation, while increasing the propensity of others. In this context, it is unlikely that states become more dissatisfied with the World Bank after being given greater control within the institution.

IV Results

To identify the causal eﬀect of Vote Power Misalignment, I use a two-stage least squares (2SLS) model with the following estimating equations for each stage. For state i in year t:

Vote Power Misalignmentit = α1 + γ1Formula Shifti + γ2Xi,t−1 + it (1) ˆ Institutional Proliferationit = α2 + β1Vote Power Misalignmenti,t−1 + β2Xi,t−1 + δit (2) where Xit is the vector of control variables discussed in the previous section.

Results are presented in Table 2. Column 1 displays a model with baseline control variables and the key independent variable, Vote Power Misalignment, instrumented by the shift in votes due to Mikesell’s formula. All models include region ﬁxed eﬀects, a time

29 polynomial, and the four component variables in Mikesell’s vote formula.22 Including these component variables guards against the possibility that Mikesell strategically altered the formula to aﬀect states other than the Big Four, though the historical record suggests this is unlikely.

The estimated effect of Vote Power Misalignment on institutional proliferation is positive and statistically significant in Model 1, confirming expectations that states are more likely to create new development banks when their authority in the World Bank falls short of their underlying economic power. The effect is substantively large: a one standard deviation increase in misalignment causes a 6.1 percentage point increase in the probability of institutional proliferation, more than doubling the baseline rate of 4.4%. Among the control variables, states with higher GDP are more likely to construct new development banks, as

22National income (1940), average exports (1934-1938), gold and dollar reserves (1940), and maximum variation in exports (1934-1938).

30 Dependent variable: Institutional Proliferation (1) (2) (3) 2SLS 2SLS Reduced Form Vote Power Misalignment 5.487∗∗∗ 8.357∗∗ (1.879) (4.075)

Mikesell Vote Shift −11.673∗∗∗ (6.188)

GDP 0.035∗∗∗ 0.046∗∗∗ 0.036∗∗∗ (0.008) (0.017) (0.010)

GDP per capita −0.016∗ −0.018 0.003 (0.008) (0.016) (0.009)

Polity 0.001 0.002∗ 0.001 (0.001) (0.001) (0.001)

Military Capacity −8.776∗∗∗ −12.802∗∗∗ −4.270∗∗∗ (2.605) (5.518) (1.010)

Loans to Allies −0.026 −0.038 (0.074) (0.063)

MDBs joined −0.023∗∗∗ −0.030∗∗∗ (0.005) (006)

Aid Received 0.002∗∗∗ 0.002∗∗∗ (0.001) (0.001)

Aid Given 0.001 0.002∗∗ (0.001) (0.001)

Regional Trade Integration 0.054 0.048 (0.072) (0.060)

Observations 2,056 1,963 1,963 States 34 34 34 F Statistic 18.649 23.139

Table 2: Eﬀect of Vote Power Misalignment on Institutional Proliferation. Stan- dard errors are clustered by country. ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.

31 well as states with lower levels of development (GDP per capita) and military capacity.

Column 2 adds the remaining controls. The effect of Vote Power Misalignment increases in magnitude compared to Model 1. To demonstrate the effect size, suppose the World Bank offered China more vote power in 2012, changing its Vote Power Misalignment from the observed value (0.078) to a level equivalent to Japan (−0.013). The results indicate that this additional influence would make China 76% less likely to create a new development bank (as China did shortly thereafter with the Asian Infrastructure Investment Bank). Per- haps surprisingly, states’ geopolitical preferences over the distribution of World Bank loans

(Loans to Allies) has no significant association with institutional proliferation. Demo- cratic states and recipients of foreign aid are slightly more likely to build new development banks, while membership in many existing banks depresses the likelihood of proliferation. Finally, Column 3 presents a linear model that estimates the direct effect of the Mikesell formula shift on the probability of institutional proliferation. This reduced form model will not accurately recover the effect size of Vote Power Misalignment, but it demonstrates that an exogenous infusion of vote power reduces the likelihood of institutional proliferation. The results suggest that a one percentage point increase in World Bank vote share decreases the probability of institutional proliferation by 11.7 percentage points. One potential concern is that dynamic changes in World Bank votes after 1944 render the Mikesell formula irrelevant in later years. In the current application, this would mean that the initial shift in vote shares is a weak instrument, unable to predict states’ influence in subsequent years. However, diagnostic tests indicate the instrument is strongly correlated with Vote Power Misalignment (F-statistic = 18.649). I also calculate Anderson-Rubin standard errors, which are robust to weak instruments, with no changes in the results. The results are robust to the removal of each development bank from the sample. Follow- ing Conley et al. (2012), I perform sensitivity analysis to confirm that the results are robust

32 to minor violations of the exclusion restriction.23 Finally, I examine the proliferation of other international institutions as a placebo test. If the theory is correct, power misalignment in the World Bank should affect states’ propensity to form new development banks; there is no reason for Bank misalignment to be associated with the proliferation of security IOs or trade agreements. Appendix Table A3 replicates the fully specified model with the proliferation of these institutions as outcome variables. As expected, Vote Power Misalignment has no significant effect.

4.2 Fixed Eﬀects Models

The strength of the IV strategy is internal validity; it accurately estimates the eﬀect of

Vote Power Misalignment in the presence of unobserved confounders. However, this approach entails three important limitations. First, the sample is restricted to conform to the exogeneity assumption. In the models above, I excluding the Big Four states and others for which contemporaneous data on the Mikesell formula components is unavailable.24 An ideal test would include more states and particularly those, like China, that have recently engaged in institutional proliferation. Second, the analysis does not exploit dynamic variation in power misalignment. Because the instrument reﬂects a single cross-sectional shock to vote shares, it cannot examine power misalignments that emerged in the years after Bretton Woods. Finally, the IV speciﬁcation assumes institutional proliferation is informed only by power misalignment as measured annually in the World Bank. While this is a reasonable assumption given the prominence of the Bank, it precludes testing of more complex dynamic

23Sensitivity tests estimate that the instrument can have a direct eﬀect of up to +−1.03 on the outcome with substantively identical results. This is equivalent to each standard deviation increase in vote share received from the Mikesell formula directly aﬀecting the probability of institutional proliferation by approximately 1 percentage point. 24States included in the IV sample are: Argentina, Australia, Belgium, Bolivia, Brazil, Canada, Chile, Colombia, Costa Rica, Czechoslovakia, Denmark, Ecuador, Egypt, El Salvador, Finland, Germany, Greece, Italy, Japan, Mexico, Netherlands, New Zealand, Norway, Peru, Philippines, Poland, Portugal, South Africa, Sweden, Thailand, Turkey, Uruguay, Venezuela, and Yugoslavia.

33 eﬀects or broader conceptions of power misalignment. For example, states may weigh their power in the growing number of multilateral development banks that emerge alongside the World Bank throughout the period. China’s role in creating the AIIB, as well as Japan’s decision to reject participation in it, may be driven by their representation in the Asian Development bank as well as the World Bank. To address these limitations, I turn to a set of non-instrumented regression models

with state and year fixed effects. These models leverage dynamic shifts in Vote Power Misalignment while accounting for unobserved time- and state-invariant confounders. Al- though fixed effects cannot address state-specific, time-varying confounders, they do allow us to overcome the constraints discussed above. I use linear probability models to maintain comparability with the 2SLS specifications presented in Table 2.25 Table 3 presents the results of the fixed effects models. Column 1 shows estimates for a model that includes the full set of controls from Table 2, Col. 2 along with state and

year fixed effects. The coefficient for Vote Power Misalignment is positive and statistically significant at the 0.1 level, demonstrating that the effect is not limited to the 34 states in the 2SLS sample. Substantively, the coefficient indicates that a one standard deviation increase

in Vote Power Misalignment leads to a 1.5 percentage point increase in the likelihood of institutional proliferation. The fixed effects estimate is significantly attenuated compared to the IV results (0.95 vs 8.36), a possible sign of bias due to systematic differences between states that are under- and over-represented in the Bank. To minimize heterogeneity in these groups, I adjust the sample via matching. I employ the covariate balancing propensity score matching technique described in Imai & Ratkovic (2014). This approach combines the modeling of treatment assignment common in other propensity score methods with the optimization of covariate

25Logit models produce similar results, but exhibit signs of over-ﬁtting and occasional convergence problems.

34 Dependent variable: Institutional Proliferation (1) (2) (3) (4) Vote Power Misalignment 0.953∗ 2.538∗∗ 2.222∗∗ 2.248∗∗ (0.576) (0.997) (1.038) (1.096)

Years Increasing 0.003∗∗∗ 0.003∗∗∗ Misalignment (0.001) (0.001)

Vote Power Misalignment 0.141 (Regional Banks) (0.164)

GDP 0.017 0.002 0.004 0.003 (0.017) (0.027) (0.027) (0.026)

GDP per capita −0.021 −0.015 −0.017 −0.016 (0.016) (0.023) (0.023) (0.023)

Polity −0.001 −0.0003 −0.0002 −0.0002 (0.001) (0.001) (0.001) (0.001)

Military Capacity 0.083 −0.342 −0.413 −0.386 (0.629) (2.153) (2.133) (2.121)

Loans to Allies −0.040 0.026 0.024 (0.049) (0.059) (0.060) (0.060)

MDBs joined −0.088∗∗∗ −0.080∗∗∗ 0.079∗∗∗ −0.079∗∗∗ (0.009) (0.011) (0.013) (0.011)

Aid Received 0.002 0.001 0.001 0.001 (0.0005) (0.001) (0.001) (0.001)

Aid Given 0.001 0.001 0.001∗ 0.002∗ (0.001) (0.001) (0.001) (0.001)

Regional Trade Integration −0.010 −0.006 0.0004 −0.0002 (0.083) (0.094) (0.094) (0.095)

Observations 7,231 7,231 7,222 7,212

Table 3: Dynamic Change in Vote Power Misalignment. Standard errors are clustered by country. All models include a time polynomial, state and year fixed effects, ingoing and outgoing aid flows, regional trade dependence, and number of members in existing banks (not shown). ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.

35 balance, ensuring the treatment and control groups look substantively similar.26 Column 2 displays results for the same ﬁxed eﬀects model after weighting observations based on

covariate balanced propensity scores. Vote Power Misalignment has a substantively larger, statistically significant effect on institutional proliferation in this specification. The next model continues to use the matched sample and begins to relax the myopic view of power misalignment represented in prior specifications. It was previously assumed that states only consider their degree of World Bank representation in the previous year when deciding whether to construct new development banks. In reality, temporary misalignments that are quickly corrected are unlikely to generate the same dissatisfaction as enduring misalignments that grow worse over time. The specification in Column 3 adds a variable that counts the number of years each state has experienced growing misalignment in the bank.

The coefficient for Years Increasing Misalignment is small but statistically significant. It suggests that each year in which a state’s misalignment in the Bank worsens, the probability of institutional proliferation increases by 0.3 percentage points. That holds true even after conditioning on the absolute level of misalignment, which retains its statistically and substantively significant effect in Column 3. Finally, the last model further expands the conception of power misalignment to include states’ representation in other multilateral development banks. It incorporates a separate measure of power misalignment in the four largest regional development banks.27 Kellerman (2019) argues that these banks also create an incentive for continued proliferation due to a tendency to revert to status quo modes of governance, frustrating their founding members. If so, we should expect a positive and significant estimate for the regional power misalignment

26I adopt the matching technique for continuous treatment described in Fong et al. (2018a) and imple- mented in the CBPS package in R (Fong et al. , 2018b). See Appendix Figure A4 for a comparison of covariate balance before and after the matching procedure. 27Regional development banks included in this model are the European Investment Bank, Inter-American Development Bank, African Development Bank, and Asian Development Bank. I collect data for years in which the banks provide publicly available annual reports listing vote shares.

36 variable. Column 4 displays the results of this model. The coefficient is in the expected direction, but fails to achieve statistical significance at conventional levels. The effect of power misalignment in the World Bank remains statistically significant in this specification.

5 Discussion and Future Research

This paper argues that states are more likely to create new IOs when they believe their influence in existing institutions is constrained by outdated rules. Rather than attempting to maximize cooperative benefits, states often build institutions as part of a competition for influence in multilateral organizations. IOs that fail to adapt to changes in states’ underlying material power risk spurring the proliferation of additional institutions, fragmenting global governance and potentially creating obstacles to effective cooperation. Power misalignment in existing IOs is one important pathway through which states de- cide to build multiple, overlapping institutions. It provides a rationale for perhaps the most significant trend in international cooperation over the last several decades: the increasing crowding of governance institutions in many issue areas. This trend has only recently be- gun to attract scholarly attention, and the power misalignment theory offers an important complement to preference-based explanations that have been advanced in the literature. A notable implication of the theory is that we will observe more institutional proliferation than is strictly optimal for cooperation. As the distribution of state power evolves, regime complexes are likely to experience continued fragmentation and its attendant coordination challenges. This expectation stands opposed to the “rational design” literature (Koremenos et al. , 2001), which argues that international regimes are designed to efficiently address the underlying cooperation problem states confront. I find empirical support for the effect of power misalignment in the issue area of development lending. By leveraging a unique natural experiment, I demonstrate that initial imbalances in multilateral bargaining power in the World Bank encouraged states to con-

37 struct new development banks. Subsequent power misalignment that emerged due to shifts in state capacity have had a similar effect, especially when misalignment persists over many years. The substantive effect of power misalignment is large in magnitude, significantly increasing the probability of institutional proliferation among disadvantaged states. The analysis has several implications for our understanding of international financial institutions. First, newly established institutions offer a source of leverage that provide states with influence in IFIs. These findings are consistent with work portraying regional development banks as viable outside options, strengthen their members’ hands in negotiations (Lipscy, 2015). Understanding how states exert influence over World Bank lending therefore requires knowing who wields power in other development banks. Second, concerns about severe discord between new banks and existing IFIs may be overstated. If states establish a new institution to provide bargaining leverage – and not to challenge operational practices in existing bodies – the institutions may interact more harmoniously than is commonly expected. To serve as an outside option, new institutions must be at least partial substitutes for the original IO. This implies they will not function in fundamentally different ways than existing institutions. The basic mission and operation of new banks should be consistent with long-standing institutions like the World Bank. These implications extend beyond development banking to other salient issue areas in world politics. Other major institutions, including the European Union and the UN Security Council, distribute influence unequally to member states. If these IOs fail to adapt to shifts in state power – a common criticism of the Security Council in particular – they will likely face increasing dissatisfaction and disengagement from under-represented member states. While it is hard to imagine the creation of a competitor to the Security Council, states could construct new IOs that encroach on particular issues or competencies of the UNSC. More broadly, periods of rapid change in the distribution of power are likely to be accompanied by intense bursts of institutional proliferation. The waning power of the US and Europe

38 and the rise of some developing states may be partly responsible for the recent ‘Cambrian explosion’ of institutions that is reshaping the architecture of global governance (Keohane & Victor, 2011).

39 References

Abbott, Kenneth W., Genschel, Philipp, Zangl, Bernhard, & Snidal, Duncan. 2015. Inter- national Organizations as Orchestrators. Cambridge University Press.

Accra Agenda for Action. 2008 (September). 3rd High-Level Forum on Aid Eﬀectiveness.

Alter, Karen J., & Meunier, Sophie. 2009. The politics of international regime complexity. Perspectives on politics, 7(01), 13–24.

Alter, Karen J, & Raustiala, Kal. 2018. The Rise of International Regime Complexity. Annual Review of Law and Social Science, 14, 329–349.

Bussi´ere, Eric.` 2008. The bank of the European Union: the EIB, 1958-2008. European Investment Bank.

Carter, David B., & Signorino, Curtis S. 2010. Back to the future: Modeling time dependence in binary data. Political Analysis, mpq013.

Conley, Timothy G, Hansen, Christian B, & Rossi, Peter E. 2012. Plausibly exogenous. Review of Economics and Statistics, 94(1), 260–272.

Copelovitch, Mark S. 2010. Master or servant? Common agency and the political economy of IMF lending. International Studies Quarterly, 54(1), 49–77.

Davis, Christina, & Pratt, Tyler. 2019. The Forces of Attraction: How Security Interests

Shape Membership in Economic Institutions. Available at https://scholar.harvard. edu/files/cldavis/files/davispratt_forces_july2018.pdf.

Davis, Christina L. 2009. Overlapping institutions in trade policy. Perspectives on politics, 7(01), 25–31.

Diaz-Bonilla, Eugenio, & del Campo, Victoria. 2011. A Long and Winding Road: The Creation of the Inter American Development Bank. Lulu. com.

Dreher, Axel, & Sturm, Jan-Egbert. 2012. Do the IMF and the World Bank inﬂuence voting in the UN General Assembly? Public Choice, 151(1-2), 363–397.

40 Dreher, Axel, Sturm, Jan-Egbert, & Vreeland, James Raymond. 2009. Development aid and international politics: Does membership on the UN Security Council inﬂuence World Bank decisions? Journal of Development Economics, 88(1), 1–18.

Drezner, Daniel W. 2009. The power and peril of international regime complexity. Perspec- tives on politics, 7(01), 65–70.

D¨ur,Andreas, Baccini, Leonardo, & Elsig, Manfred. 2014. The Design of International Trade Agreements. The Review of International Organizations, 9(3), 353–375.

Fleck, Robert K, & Kilby, Christopher. 2006. World Bank independence: A model and statistical analysis of US inﬂuence. Review of Development Economics, 10(2), 224–240.

Fong, Christian, Hazlett, Chad, Imai, Kosuke, et al. . 2018a. Covariate Balancing Propensity Score for a Continuous Treatment: Application to the Eﬃcacy of Political Advertisements. The Annals of Applied Statistics, 12(1), 156–177.

Fong, Christian, Ratkovic, Marc, & Imai, Kosuke. 2018b. Package ‘CBPS’.

Frey, Bruno S, & Schneider, Friedrich. 1986. Competing models of international lending activity. Journal of Development Economics, 20(2), 225–245.

Gehring, Thomas, & Faude, Benjamin. 2014. A theory of emerging order within institutional complexes: How competition among regulatory international institutions leads to institutional adaptation and division of labor. The Review of International Organizations, 9(4), 471–498.

Gilpin, Robert. 1981. War and change in world politics. Cambridge University Press.

Gruber, Lloyd. 2000. Ruling the world: Power politics and the rise of supranational institutions. Princeton University Press.

Hawkins, Darren G., Lake, David A., Nielson, Daniel L., & Tierney, Michael J. 2006. Dele- gation and agency in international organizations. Cambridge University Press.

Helfer, Laurence R. 2004. Regime shifting: the TRIPS agreement and new dynamics of

41 international intellectual property lawmaking. Yale J.Int’l L., 29, 1.

Helfer, Laurence R. 2009. Regime shifting in the international intellectual property system. Perspectives on Politics, 7(01), 39–44.

Henning, C Randall. 2017. Tangled Governance: International Regime Complexity, the Troika, and the Euro Crisis. Oxford University Press.

Hirschman, Albert O. 1970. Exit, voice, and loyalty: Responses to decline in ﬁrms, organizations, and states. Vol. 25. Harvard university press.

Ikenberry, G. John. 2001. After Victory: Institutions, Strategic Restraint, and the Rebuilding of Order after Major Wars. Princeton University Press.

Imai, Kosuke, & Ratkovic, Marc. 2014. Covariate Balancing Propensity Score. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 76(1), 243–263.

Johnson, Tana. 2014. Organizational progeny: Why governments are losing control over the proliferating structures of global governance. Oxford University Press, USA.

Kaya, Ayse. 2015. Power and Global Economic Institutions. Cambridge University Press.

Kellerman, Miles. 2019. The Proliferation of Multilateral Development Banks. The Review of International Organizations, 14(1), 107–145.

Keohane, Robert. 1984. 0.(1984) After Hegemony. Cooperation and Discord in the World Political Economy, Princeton.

Keohane, Robert O., & Victor, David G. 2011. The Regime Complex for Climate Change. Perspectives on Politics, 9(01), 7–23.

Keohane, Robert Owen, & Nye, Joseph S. 1977. Power and interdependence: World politics in transition. Little, Brown Boston.

Kersting, Erasmus K, & Kilby, Christopher. 2016. With a little help from my friends: Global electioneering and World Bank lending. Journal of Development Economics, 121, 153–165.

Koremenos, Barbara, Lipson, Charles, & Snidal, Duncan. 2001. The rational design of

42 international institutions. International organization, 55(04), 761–799.

Krasner, Stephen D. 1981. Power structures and regional development banks. International organization, 35(2), 303–328.

Krasner, Stephen D. 1991. Global communications and national power: Life on the Pareto frontier. World Politics, 43(03), 336–366.

Kuziemko, Ilyana, & Werker, Eric. 2006. How Much Is a Seat on the Security Council Worth? Foreign Aid and Bribery at the United Nations. Journal of Political Economy, 114(5), pp. 905–930.

Lim, Daniel Yew Mao, & Vreeland, James Raymond. 2013. Regional organizations and international politics: Japanese inﬂuence over the Asian Development Bank and the UN Security Council. World Politics, 65(01), 34–72.

Lipscy, Phillip Y. 2015. Explaining Institutional Change: Policy Areas, Outside Options, and the Bretton Woods Institutions. American Journal of Political Science, 59(2), 341–356.

Lipscy, Phillip Y. 2017. Renegotiating the World Order: Institutional Change in International Relations. Cambridge University Press.

Mikesell, Raymond F. 1943. “Approximate Figures ”. Harry Dexter White Papers, Box 6, Folder 6; Public Policy Papers, Department of Rare Books and Special Collections, Princeton University Library.

Mikesell, Raymond F. 1994. The Bretton Woods Debates: A Memoir, Essays in International Finance, No. 192, Princeton University. International Finance Section.

Morse, Julia C., & Keohane, Robert O. 2014. Contested multilateralism. The Review of International Organizations, 1–28.

Odell, John S. 2000. Negotiating the World Economy. Cornell University Press.

Organski, Alfred FK. 1968. World Politics.

Paul, Thazha V, Larson, Deborah Welch, & Wohlforth, William C. 2014. Status in world

43 politics. Cambridge University Press.

Pratt, Tyler. 2018. Deference and Hierarchy in International Regime Complexes. Interna- tional Organization, 72(3), 561–590.

Raustiala, Kal, & Victor, David G. 2004. The regime complex for plant genetic resources. International organization, 277–309.

Schneider, Christina J. 2011. Weak states and institutionalized bargaining power in international organizations. International Studies Quarterly, 55(2), 331–355.

Sheng, Andrew, & Geng, Xiao. 2015 (April). The AIIB and Global

Governance. https://www.project-syndicate.org/commentary/ aiib-china-global-leadership-by-andrew-sheng-and-geng-xiao-2015-04? barrier=accesspaylog.

Steil, Benn. 2013. The battle of Bretton Woods: John Maynard Keynes, Harry Dexter White, and the making of a new world order. Princeton University Press.

Stone, Randall W. 2011. Controlling institutions: International organizations and the global economy. Cambridge University Press.

Thacker, Strom C. 1999. The high politics of IMF lending. World politics, 52(01), 38–75.

Tierney, Michael J, Nielson, Daniel L, Hawkins, Darren G, Roberts, J Timmons, Findley, Michael G, Powers, Ryan M, Parks, Bradley, Wilson, Sven E, & Hicks, Robert L. 2011. More dollars than sense: Reﬁning our knowledge of development ﬁnance using AidData. World Development, 39(11), 1891–1906.

Urpelainen, Johannes, & de Graaf, Thijs Van. 2013. Your Place or Mine? Institutional Capture and the Creation of Overlapping International Institutions. British Journal of Political Science.

Voeten, Erik. 2001. Outside options and the logic of Security Council action. American Political Science Review, 95(4), 845–858.

44 White, Harry D. 1942. UN Stabilization Fund and a Bank for Reconstruction and Develop- ment of the United and Associated Nations. Harry Dexter White Papers, Box 6, Folder 6; Public Policy Papers, Department of Rare Books and Special Collections, Princeton University Library.

White, Harry D. 1943. “Preliminary Draft Outline of a Proposal for a UN Bank for Recon- struction and Development ”. Harry Dexter White Papers, Box 6, Folder 6; Public Policy Papers, Department of Rare Books and Special Collections, Princeton University Library.

Wilson, Dick. 1987. A bank for half the world: the story of the Asian Development Bank, 1966-1986. Asian Development Bank Manila.

Zangl, Bernhard, Heußner, Frederick, Kruck, Andreas, & Lanzend¨orfer,Xenia. 2016. Im- perfect Adaptation: How the WTO and the IMF Adjust to Shifting Power Distributions Among Their Members. The Review of International Organizations, 11(2), 171–196.

45 Appendix

Institution Year States CEDB 1956 Belgium, Denmark, France, Greece, Iceland, Luxembourg, Norway, Netherlands, UK, Italy, Turkey, Austria, Germany, Sweden EIB 1958 Belgium, France, Luxembourg, Netherlands, Italy, Germany IDB 1959 Bolivia, Brazil, Chile, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, USA, Colombia, Haiti, Argentina CABEI 1960 Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua AfDB 1965 Algeria, Benin, Cameroon, DR Congo, Congo, Egypt, Ethiopia, Ghana, Guinea, Kenya, Liberia, Mali, Mauritania, Morocco, Niger, Nigeria, Rwanda, Senegal, Sierra Leone, Somalia, Sudan, Tanzania, Togo, Tunisia AsDB 1966 Afghanistan, Australia, Austria, Belgium, Canada, Denmark, USA, Finland, Germany, India, Indonesia, Italy, Japan, South Korea, UK, Laos, Malaysia, Nepal, Netherlands, New Zealand, Norway, Pakistan, Philippines, Sri Lanka, Sweden, Thailand, Vietnam EADB 1967 Kenya, Tanzania, Uganda AFESD 1968 Algeria, Egypt, Jordan, Kuwait, Lebanon, Libya, Morocco, Saudi Arabia, Sudan, Syria, Tunisia CAF 1968 Bolivia, Chile, Colombia, Ecuador, Peru, Venezuela CDB 1970 Canada, Guyana, Jamaica, Trinidad & Tobago, USA IsDB 1973 Afghanistan, Algeria, Bahrain, Bangladesh, Cameroon, Egypt, Guinea, Indonesia, Jordan, Kuwait, Libya, Malaysia, Mauritania, Morocco, Niger, Oman, Pakistan, Qatar, Saudi Arabia, Senegal, Somalia, Yemen, Sudan, Syria, Tunisia, Turkey, UAE WADB 1973 Benin, Burkina Faso, Cote d’Ivoire, Mali, Niger, Senegal, Togo BDEAC 1975 Cameroon, Central African Republic, Congo, Gabon ABEDA 1975 Algeria, Bahrain, Egypt, Iraq, Jordan, Kuwait, Lebanon, Libya, Mauritania, Morocco, Oman, Qatar, Saudi Arabia, Sudan, Syria, Tunisia, UAE

Table A1: Coding of Institutional Proliferation

46 Institution Year States DBGLS 1976 Burundi, DR Congo, Rwanda OFID 1976 Algeria, Ecuador, Gabon, Indonesia, Iran, Iraq, Kuwait, Libya, Nigeria, Qatar, Saudi Arabia, UAE, Venezuela NIB 1976 Denmark, Finland, Iceland, Norway, Sweden IFAD 1977 Australia, Austria, Bangladesh, Belgium, Benin, Bolivia, Botswana, Burkina Faso, Cameroon, Canada, Cape Verde, Chad, Comoros, DR Congo, Cuba, Cyprus, Denmark, Djibouti, Dominican Republic, Ecuador, Egypt, El Salvador, Ethiopia, Finland, France, Gambia, Germany, Ghana, Guinea, Guyana, Haiti, Honduras, India, Indonesia, Iran, Iraq, Ireland, Italy, Jamaica, Japan, Kenya, Kuwait, Lesotho, Libya, Luxembourg, Malawi, Mali, Malta, Mexico, Morocco, Netherlands, New Zealand, Nicaragua, Niger, Nigeria, Norway, Pakistan, Panama, Peru, Philippines, Qatar, Romania, Rwanda, Saudi Arabia, Senegal, Sierra Leone, Somalia, Sri Lanka, Sudan, Swaziland, Sweden, Switzerland, Tanzania, Thailand, Tunisia, UAE, Turkey, Uganda, UK, Uruguay, USA, Venezuela, Samoa, Zambia ESATDB 1985 Burundi, Djibouti, DR Congo, Ethiopia, Kenya, Madagascar, Malawi, Mauritius, Rwanda, Swaziland, Sudan, Uganda, Zambia, Zimbabwe NDF 1989 Denmark, Finland, Iceland, Norway, Sweden EBRD 1991 Albania, Australia, Austria, Belgium, Bulgaria, Canada, Cyprus, Czechoslovakia, Denmark, Egypt, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, South Korea, Liechtenstein, Luxembourg, Malta, Mexico, Morocco, Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Spain, Sweden, Switzerland, Turkey, UK, USA BSTDB 1992 Albania, Armenia, Azerbaijan, Bulgaria, Greece, Georgia, Moldova Romania, Russia, Turkey, Ukraine NADB 1993 United States, Mexico ECOTDB 2005 Iran, Pakistan, Turkey EuADB 2006 Russia, Kazakhstan, Tajikistan

Table A1 (cont): Coding of Institutional Proliferation

47 Institution Year States NDB 2013 Brazil, Russia, India, China, South Africa AIIB 2015 Australia, Austria, Azerbaijan, Bangladesh, Brazil, Brunei, Cambodia, China, Denmark, Egypt, Finland, France, Georgia, Germany, Iceland, India, Indonesia, Iran, Israel, Italy, Jordan, Kazakhstan, South Korea, Kuwait, Kyrgyzstan, Laos, Luxembourg, Malaysia, Maldives, Malta, Mongolia, Myanmar, Nepal, Netherlands, New Zealand, Norway, Oman, Pakistan, Philippines, Poland, Portugal, Qatar, Russia, Saudi Arabia, South Africa, Spain, Sri Lanka, Sweden, Switzerland, Tajikistan, Thailand, Turkey, UAE, UK, Uzbekistan, Vietnam

Table A1 (cont): Coding of Institutional Proliferation

48 Dependent variable: Vote Power Misalignment Mikesell Vote Shift −1.794∗∗∗ −2.002∗∗∗ (0.448) (0.416)

Region Fixed Eﬀects

Mikesell Component Variables

Time Polynomial

Full Controls

Observations 1,898 1,456 Adjusted R2 0.765 0.813 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Table A2: First Stage Results. Results of the first stage model examining the effect of the instrument (Quota Change) on Vote Power Misalignment. Time polynomial not shown. Statistical significance is denoted by: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.

49 Dependent variable: Institutional Proliferation

Security IOs Trade Agreements Vote Power Misalignment −1.225 −5.989 (2.365) (7.140)

Region Fixed Eﬀects

Mikesell Component Variables

Time Polynomial

Full Controls

Observations 1,898 1,456 Adjusted R2 0.765 0.813 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Table A3: Placebo Tests. Models test the effect of Vote Power Misalignment in the World Bank on the proliferation of security organizations (Column 1) and preferential trade agreements (Column 2). Security IOs are identified using the classification scheme described in Davis & Pratt (2019). Trade treaties are identified from the Design of Trade Agreements (DESTA) project (Dür et al. , 2014). The specification is identical to the full 2SLS regression model in Table 2. Statistical significance is denoted by: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01.

50 0.5 0.4 0.3 0.2 % Aid from World Bank % Aid from World 0.1 0.0

1970 1980 1990 2000 2010

Year Figure A1: World Bank Share of Development Assistance. Calculations by author using data from AidData.

51 China Japan South Korea Iran Spain Saudi Arabia Mexico Turkey Italy Nigeria South Africa Belgium Peru Tanzania Uganda India Greece Bangladesh Kuwait Pakistan Argentina Canada Sweden Germany UK USA

−0.06 −0.04 −0.02 0.00 0.02 0.04 0.06

Change in World Bank Vote Power, 1980−2014

Figure A2: Change in World Bank Vote Share, 1980-2014. The ﬁgure shows changes in share of formal inﬂuence within the World Bank for select states over the period 1980-2014.

52 GDP ● p = 0.64

GDP per capita ● p = 0.21

Military Power ● p = 0.25

Polity ● p = 0.19

−2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 2.0

Figure A3: Covariate Balance for States Affected by Mikesell Formula. Stan- dardized mean differences and 95% confidence intervals for “winners” vs. “losers” from the Mikesell formula. Data are from 1947 (earliest available year after the switch to the Mikesell formula).

53 ● ● ●● ● ● Unweighted

● ● ●● ●● CBPS Weighted

−0.6 −0.4 −0.2 0.0 0.2 0.4 0.6

Pearson Correlation

Figure A4: Covariate Balance for States Aﬀected by Mikesell Formula. The ﬁgure displays the weighted Pearson correlation between the treatment and six covariates (GDP, GDP per capita, Polity, Military Capacity, Aid Received, and Aid Given) before and after post-matching weights are assigned. After matching, covariates have very little correlation with Vote Power Misalignment.