Splitting the Check: Bargaining Over Counterpart Commitments in World Bank Projects Matthew S. Winters University of Illinois at Urbana-Champaign [email protected] Jaclyn D. Streitfeld University of Illinois at Urbana-Champaign [email protected] November 2012 When an international development agency enters into a project agreement with a foreign aid-receiving country, the agreement often requires the recipient country to provide a certain level of counterpart funding from its own budget for the project. In contrast to much of the aid allocation literature, which views donors as making unilateral decisions about how much funding countries should receive, we use counterpart commitments to provide a theoretical perspective on bargaining between an international aid agency and an aid-receiving state over the financing of development projects. Then we use an original dataset to look at the variation in counterpart commitments across a sample of 1,565 World Bank investment projects from the 2000s. The data reveal a continuum of financing arrangements, including a significant number of World Bank projects that are country-driven projects in which the Bank plays only a minor financing role, an arrangement that again challenges commonly-held perceptions of World Bank projects. We also observe a shift in the relevant predictors of counterpart commitments following a 2004 policy change at the World Bank. Key Words: bargaining; international institutions; World Bank; development

Previous versions of this paper were presented at the 2011 Midwest Political Science Association and American Political Science Association Annual Meetings and the 2012 Political-Economy of International Organizations Conference. Thanks to Xinyuan Dai, Simone Dietrich, Desha Girod, Chris Humphrey, Christopher Kilby, Espen Moe and Robert Pahre for comments and to Martha Ainsworth, William Bernhard, Daniel Kaufmann, Phil Keefer, Steve Knack, Aart Kraay, Aymeric-Albin Meyer, Denyse Morin, Vikram Nehru, Vivek Srivastava and John Underwood for useful discussions.

Much of the literature on foreign aid allocation places donors in a privileged position, assuming that donors make decisions about funding commitments and that aid recipients accept these decisions and receive the aid. If this is the case, however, then it is puzzling that relatively powerful middleincome countries, such as Brazil, India and China, would continue to make use of funding from international development institutions. In reality, aid projects result from negotiation between development partners and the governments in developing countries. Although there is a literature from the 1990s that explores donor-recipient bargaining over conditionality in structural adjustment programs (Mosley 1998; Mosley, Harrigan and Toye 1991; Killick 1998) and a literature that thinks very generally about contracts between donors and recipient governments (Svensson 2000, 2003; Azam and Laffont 2003), there is little research that tries to explain how the design of specific investment projects results from negotiation and bargaining.1 Better understanding negotiations between donors and recipients will help us accomplish at least three objectives. First, to the extent that we can identify recipient-driven projects, our results pose a challenge to the dominant donor-driven perspective underlying the study of aid allocation. Second, thinking from a bargaining perspective can guide future studies of aid allocation. Third, understanding the way in which the content of specific projects is determined is essential for thinking about the outcomes of these projects and aid effectiveness in general. Having a better understanding of how particular development projects come to have the features that they do will help us better assess the contributions that foreign aid makes to poverty alleviation and economic growth in developing countries and why it makes that contribution. In this paper, we think about donor-recipient negotiations in the context of counterpart funding requirements found in investment projects with which the World Bank is involved. The World Bank lends money to poor and middle-income countries to help them meet their development objectives. 1

In a recent paper, Bayer, Marcoux and Urpelainen (2012) look at donor-state bargaining in the context of Global Environment Facility projects in the 1990s and 2000s.

1

Although much attention has been paid to World Bank structural adjustment lending,2 the majority of Bank lending is for specific investment projects. Over the period 2000 – 2010, two-thirds of total World Bank lending financed investment projects. Among the poorest borrowers – those who receive credits from the Bank’s concessional lending wing, the International Development Association (IDA) – the ratio was even higher: over three-quarters of IDA financing went to investment projects. Within these investment projects, the World Bank often is only one of several financing entities. In many cases – over 80 percent of projects – the borrowing country itself agrees to make financial contributions to the investment project.3 As the World Bank commits money to build schools, upgrade irrigation infrastructure, train nurses or provide microloans to entrepreneurs, the Bank usually does so in the context of an agreement where the aid-receiving country agrees to supply some portion of this funding. This is also true of investment projects funded by other multilateral (and some bilateral) donors. What influences the size of these counterpart commitments? We imagine two bargaining scenarios – one in which the impetus for the project under consideration originates with the World Bank and one in which the project originates with the recipient country. For negotiations under either scenario, we suggest a set of factors that will influence the attitudes of the World Bank and the recipient country toward the amount of counterpart funding that the project will include.

2

Now known as development policy lending, structural adjustment lending is lending meant to catalyze policy reforms. That is, it does not go to fund specific investments. Investment projects may also have reform components – agreements often describe specific policies that the government should undertake to support the development objectives of the project – but they differ from policy lending in that the project agreement describes specific goods and services that will be financed by the project funding. 3 A substantial number of projects – about 20 percent – include funding from other bilateral or multilateral donors. Other bilateral or multilateral donors also may finance complementary projects based on negotiations with the World Bank and/or the aid-receiving state. Whereas we can observe cofinancing by other donors within the projects in our dataset, we know of no data source that would allow us to observe the financing of complementary projects.

2

We then examine patterns in counterpart financing from the 2000s, using an original dataset compiled from the World Bank’s Projects Database.4 The data reveal that there are a set of World Bank projects that are country-originated in which the World Bank plays a relatively minor financing role. The data also reflect differential patterns that result from a general shift in the World Bank’s policy toward counterpart funding in 2004. We anticipate that the World Bank’s perception of the borrower’s commitment to the project may be influenced by certain characteristics of the recipient-country’s government, driving the World Bank to request certain levels of counterpart funding. We also argue that economic and political constraints may influence the recipient-country’s capacity to demand lower counterpart commitments from the Bank. We find that counterpart financing increases with World Bank financing and with other sources of external project financing, that countries with more effective governments are more likely to pay for a greater share of investment projects, that democracies make smaller counterpart commitments than non-democracies in the post-2004 period and that wealthier countries and countries which have borrowed more money from the World Bank make larger counterpart commitments in the post-2004 period. There is some evidence that countries of geostrategic interest to the United States make smaller counterpart commitments.

Contracting Foreign Aid Much of the theoretical literature on aid contracting starts from the assumption that donors are benevolently concerned with poverty reduction and economic development.5 The models in this body

4

The portal to the Projects Database can be found at http://go.worldbank.org/0FRO32VEI0. The empirical literature on aid allocation has convincingly shown how wrong this assumption is, demonstrating that international donors consistently privilege their own geostrategic and commercial interests or historical relationships with particular countries over need when allocating aid (McKinlay and Little 1979; Maizels and Nissanke 1984; Alesina and Dollar 2000; Berthélemy and Tichit 2004; Berthélemy 2006; Bueno de Mesquita and Smith 2007, 2009). The influence of developed country strategic and commercial interests has been shown to extend to multilateral donors, such as the World Bank and the IMF (Frey and Schneider 1986; Thacker 1999; Stone 5

3

of research then demonstrate the ways in which incentive structures in the principal-agent relationship between donors and aid-receiving states limit the effectiveness of foreign aid in achieving these ends. Much of this literature portrays a Samaritan’s dilemma in which donors’ desire to provide assistance to impoverished populations creates aid dependency and limits the effectiveness of aid: although donors want to alleviate poverty, governments want to continue receiving foreign aid and therefore lack incentives to see that the aid is used for the purpose of poverty alleviation. Since the threat to withdraw aid is not credible (because this would mean interrupting the flow of goods and services to impoverished populations), aid flows continue despite the fact that development objectives are not being met (Svensson 2000; Gibson et al. 2005; Williamson 2010). This situation is compounded by the intense “disbursement culture” that prevails within aid agencies. Bureaucratic incentives within aid organizations reward complete disbursement of available budgets rather than the design of effective programs. This culture additionally hinders the ability of a donor to credibly threaten the withdrawal of aid in the event that it has been used inappropriately and therefore sustains large aid flows without development progress (Mosley, Harrigan and Toye 1991; Killick 1998; Ostrom et al. 2001; Easterly 2002; Svensson 2003; de Renzio et al. 2005; Woods 2006). In this literature, aid is conceptualized as a transfer from an international agency to a domestic government with the latter making a decision about whether to use the money in line with the purposes for which it has been given or not. Although this picture does not necessarily contradict the types of decisions made by country governments when they receive foreign aid, we want to flag the fact that decisions about aid provision tend to be seen from the donor’s perspective without much sense of negotiations leading up to the point at which there is an agreement by the donor to give money to a recipient country. 2004; Barro and Lee 2005; Andersen, Hansen and Markussen 2006; Fleck and Kilby 2006; Dreher and Jensen 2007; Dreher, Sturm and Vreeland 2009), although some authors argue that multilateral donors are less likely to be driven by geostrategic interests and more likely to give aid for the purposes of encouraging development and economic growth (Maizels and Nissanke 1984; Girod 2008; Steele 2012).

4

In this paper, we study the common scenario in which a domestic government not only receives an international transfer but also agrees to contribute its own resources to an investment project. We assert that the mix of domestic and international funding is the outcome of a bargaining process. This specific type of bargain has not been addressed in the theoretical or empirical literature to date. We draw attention to counterpart funding as an important component of aid contracts and argue that, for some projects, the level of counterpart funding indicates the fact that the development project originated with the aid-receiving country, a pathway to aid flows that has been underacknowledged and understudied in the literature. We admit that counterpart funding commitments suffer from credibility problems. That is, in the same way as foreign funding might be diverted for other purposes or reform objectives not be implemented, contracted-upon counterpart commitments might not materialize. Indeed one of the most frequently cited problems in the implementation reports issued by the World Bank at the conclusion of projects is exactly the failure of governments to provide agreed-upon counterpart funding (Winters 2011; see also World Bank 2004a). Nonetheless, just as project lending still happens and just as conditionalities are still written into policy lending, agreements are still reached in project negotiations with regard to the size of counterpart contributions. We turn now to discussing what might determine the size of these contributions.

Bargaining Over Who Pays: What Explains the Size of Counterpart Commitments in Development Projects? There are two sides that participate in the negotiation of an international development project – development agency staff and borrowing-country technocrats. Both aid agency staff and recipientcountry technocrats have preferences over aid projects that involve a mix of aid agency money and

5

recipient-country counterpart funding. Either the aid agency or the recipient country might initiate a project, and an agreement might either be feasible or not.6 We depict these four scenarios in Figure 1.7

Donor Proposes Project with Counterpart Funding Donor Loan Amount

Recipient Country Proposes Project with Counterpart Funding

UGov UD

Donor Loan Amount

UGov UD

A A

Counterpart Funding

Counterpart Funding

Donor Proposes Project; Recipient Government Not Interested Donor Loan Amount

Recipient Country Proposes Project; Donor Not Interested Donor Loan Amount

UGov

UGov

UD

Counterpart Funding

UD

Counterpart Funding

Figure 1. Possible Scenarios in Donor-Recipient Country Bargaining over Project Design

In the upper left hand corner, the project is initiated by the donor. The donor is willing to dedicate some money to the project even if the recipient country is not willing to dedicate any. As the country becomes willing to provide money for the project, the donor is willing to provide additional foreign assistance, although the marginal amount is decreasing as project size increases. Therefore an 6

“Recipient country” remains shorthand for a variety of actors within a government. A given project might be the product of negotiations with a single ministry or with multiple ministries. Therefore, we can think of four cases of how projects might be initiated: (1) one government ministry approaches the donor about a pet project; (2) the government more broadly conceived approaches the donor about a project generally identified as contributing to development; (3) the donor puts pressure on the government broadly conceived to consider a project; or (4) the donor puts pressure on particular government allies to consider a project. 7 We gratefully acknowledge Christopher Kilby’s advice on how to depict the bargaining space.

6

agreement can be reached that includes some donor funding and some counterpart funding. Alternatively, as pictured in the upper right hand corner, there may be a project initiated by the government, where it will undertake the investment on a certain scale even if the donor provides no money. Then as the donor provides more money, the government also is willing to provide additional funding. Both the donor and the government have some maximum amount of money that they are willing to contribute to the project. Based on the decreasing marginal utility of contributing additional amounts of one’s own money, the two sides reach agreement on a certain mix of recipient-government and donor funding. While the utility curves for both of these scenarios should be similar, in projects initiated by the recipient-government the country displays interest in the project from the start and, as a result, is willing to agree to a higher maximum amount of counterpart commitment during bargaining than in donor-proposed projects. In the lower left hand corner, the project is initiated by the donor. The government is only willing to contribute a small amount of money – it quickly reaches its tolerance or capacity for counterpart funding – and therefore the donor opts to undertake a relatively small project with no counterpart funding involved. In the lower right hand corner, the government proposes the project and is willing to contribute a certain amount and no more, but this amount is not enough for the donor to be ready to finance part of the project. In this case, the government finances a relatively small project on its own without any donor funding or with funding from a more amenable donor.8 What are the factors that influence the willingness of an aid agency to provide more or less funding for a given amount of counterpart funding? And conversely what are the factors that influence the willingness of an aid-receiving state to agree to more or less counterpart funding for a given level of international aid? What are the project or country characteristics that would lead to different response curves? 8

The dataset that we use includes only World Bank projects for which negotiations were completed, so our analysis does not include cases that fit the scenario in the lower right hand corner.

7

On the side of the donor, we offer three general reasons why the donor would want greater counterpart financing for a given level of donor financing: (1) concerns about the country’s commitment to development; (2) concerns about the project; and (3) general or country-specific budget constraints within the donor agency. On the side of the borrower, we propose three general reasons why the country would demand greater donor financing for a given level of counterpart financing: (1) lack of interest in the project; (2) government budget constraints; and (3) bargaining leverage, that is, an ability to be more demanding. Finally, we discuss a category of development projects in which recipientcountry financing makes up the majority of project funding; for those cases, we believe that the World Bank’s involvement is motivated largely by the country’s desire for World Bank expertise, such that the bargaining scenarios portrayed above may not apply. For these very high borrower-interest cases, for example, we expect that the recipient-country would not demand greater donor financing. Donor Concerns about Country Commitment to Development. The most commonly cited reason among World Bank task team leaders for why World Bank projects have counterpart financing is that the country’s financial contribution is taken as a sign of a general commitment to the project (author’s interviews, January 2012). By getting the recipient country to literally buy into the project financially, the logic goes, the government then will be more likely to support the development objectives of the project in other ways. For example, in a project in the Republic of the Congo (Project ID 106975), the Project Appraisal Document (PAD) says, the proposed project is consistent with the Bank’s strategic role of seeking to leverage its small envelope in each new project with significant amounts from the Government in order to obtain the desired critical mass. The Bank’s strategic approach under the proposed operation is therefore to leverage a relatively small IDA amount (US$25.5 million) with US$100 million from the Government for moving the reform agenda in the urban, water and electricity sectors (World Bank 2010a). In countries that are viewed as already committed to development, we expect to see lower levels of counterpart financing, since there is less of a need to demand that the government show its commitment to development. A 2007 operational memorandum about a subset of World Bank 8

borrowers without established financing parameters shows World Bank thinking along these lines; it notes that “when there is sufficient evidence that a borrower is committed to a project or investment program, the country director may make exceptions to the borrower’s minimum 10 percent cost sharing,” which otherwise would be required for projects in such countries (World Bank 2007). Following literature that proposes that well-governed countries are better at promoting development and making use of foreign aid (Burnside and Dollar 2000, 2004) and that democracies are better at providing public goods for their citizens (Lake and Baum 2001; Baum and Lake 2003), we hypothesize that there should be less of a concern with commitment to development in well-governed countries and democracies.9 In these types of countries, therefore, we anticipate smaller counterpart commitments, on average, for a given level of donor funding.10 This view of counterpart funding as a sign of borrowing country commitment is changing at the World Bank. In 2004, the Bank revised its policy framework such that, for many countries, counterpart commitments were no longer required for any given project, although the Bank expected selectivity from its staff in the use of this ability to be the sole source of funding for projects (World Bank 2004a, 2005).11 As can be seen in Figure 2, the proportion of World Bank projects with any counterpart

9

In general, there has been a concern at the World Bank with tying aid allocation to borrowing countries’ development performance since at least the 1969 Pearson Commission Report (see the discussion in Easterly 2007). In particular, since the 1998 Assessing Aid volume, which argued that countries with good policies and institutions make better use of foreign aid (as per the results in Burnside and Dollar 2000), the World Bank has emphasized allocating aid to those countries that have a certain set of allegedly beneficial institutions and policies. Most notably, this can be seen in the Performance-Based Allocation System used to divvy up IDA funding (see the discussion in Winters 2010). The observed reality seems to match the stated intent to some extent: a series of studies find some evidence – although not overwhelming evidence – that governance characteristics predict aid allocation patterns at the World Bank and for IDA funding in particular (Neumayer 2003; Dollar and Levin 2006; Easterly 2007; Winters 2010). 10 This mirrors the expectations in Winters (2010) that the World Bank should use more programmatic lending and more nationwide investment lending in better-governed countries. In that paper, good governance is positively associated with increased nationwide investment lending for both the IDA and IBRD, whereas good governance is associated with increased programmatic lending only for the IBRD and surprisingly with decreased programmatic lending among IDA borrowers. 11 Prior to 2005, the World Bank imposed cost sharing limits that ranged from 50 percent to 90 percent of project costs based on the borrowing country’s per capita income. Under these limits, the World Bank expected that every project would include at least a minimum counterpart contribution of 10 percent of the project costs to

9

financing included dropped steeply following the rule change. Several task team leaders at the Bank were quite positive about this shift, noting that there are superior ways to judge a country’s commitment to development and arguing that projects could operate more efficiently if they were not subject to delays because of the host government’s failure to provide counterpart funding (author’s interviews, January 2012).

0.8 0.7 0.5

0.6

Proportionof Projects

0.9

1.0

Proportion of IBRD/IDA Projects with Counterpart Financing

2000

2002

2004

2006

2008

2010

Year

Figure 2. Prevalence of Counterpart Funding in World Bank Projects. Source: authors’ calculations.

Donor Concerns about the Project. For some types of projects, there may be more skepticism on the donor’s side about the development benefits of the project. For this reason, the donor may demand more counterpart financing, putting more of the responsibility for the project onto the shoulders of the demonstrate borrower commitment. As of 2005, the World Bank had established country financing parameters under which the Bank could provide 100 percent of project costs for 85 countries. For two other countries with established financing parameters, the Bank’s country financing parameters still required a 10 percent contribution for any given project (World Bank 2005).

10

recipient country. We have limited ability to assess the extent to which this is true across our data, but we explore this possibility by looking at variation in counterpart funding across different sectors. We expect that public administration and finance sector projects will be viewed by the World Bank as lowerrisk than large infrastructure projects and, as a result, will require lower levels of counterpart funding. Donor Budget Constraints. A donor may require more counterpart funding in a project simply because of its own budget constraints. Within the World Bank, this is most likely to apply in the case of IDA lending, where there is a finite budget that is allocated across IDA recipients by formula. IDA borrowers wishing to undertake larger projects may simply be required to contribute more of their own funds. Recipient Interest in the Project. Across different development projects, there will be varying levels of country interest. When the World Bank develops a Country Assistance Strategy prior to negotiating specific projects, it uses the borrower’s “own vision for its development” as a starting point and then, following consultations with a variety of domestic actors, identifies areas where the Bank can have the largest impact on development.12 The result is that the World Bank will sometimes propose projects that meet its own development goals for the country or the goals of ministries or civil society actors that are not party to the project negotiations. For projects in which the country is less interested or feels less of an imperative to prioritize, the level of counterpart funding for a given level of donor funding is likely to be smaller. However, we have no way of systematically measuring the recipient’s level of interest in a project. Recipient Budget Constraints. Wealthier countries have more funds available with which to make counterpart commitments, and so the opportunity costs of doing so are lower. Therefore, a country’s level of economic development (as measured by its GDP per capita) should positively predict counterpart funding. External debt, on the other hand, raises the opportunity costs of providing 12

See, the World Bank’s “Country Assistance Strategy” webpage, available at: http://go.worldbank.org/4M75BI76J0 (accessed 20 July 2012).

11

counterpart funding and so should reduce a country’s willingness to do so. If a donor is interested in the overall stability of its borrowing countries, it should be cognizant of these budget constraints and demand less domestic funds for projects in debt-burdened countries. Recipient Country Bargaining Leverage. Countries that have greater bargaining leverage should be able to demand more donor funding for a given level of counterpart funding. This leverage might come from several sources. First, countries that have access to a larger variety of financing sources should be able to demand more funds from any given donor. Countries that are more reliant on a particular donor have fewer outside options to which they can turn and therefore might have to make greater contributions to investment projects for the same levels of donor funding.13 Domestic or international politics may also influence bargaining leverage. World Bank projects are like international treaties and therefore often require parliamentary approval, implying that projects might be derailed by domestic political institutions even after agreement has been reached at the international level.14 To the extent that there are more veto players in a country who might be able to stop a World Bank project from gaining full approval, we expect to see fewer counterpart commitments, since veto players might complain about the amount of financing being requested of the borrowing country.15 In terms of international politics, if an aid-receiving country is geostrategically important, such that the United States or other developed country principals at the World Bank might bring their influence to bear on aid agency technocrats, we expect that the country will be able to negotiate less 13

A robust finding from the conditionality literature is that countries with access to alternative sources of financing have been most able to resist harsh conditions in their structural adjustment programs. Mosley, Harrigan and Toye, for instance, find that “large countries with a great deal of scope to borrow elsewhere” are those with the loosest conditionality (1991: 105), and Killick identifies “a clear inverse relationship between the use of conditionality and recipient governments’ access to alternative sources of financing” (1998: 12). 14 In December 2009, for instance, the Nigerian House of Representatives rejected a $300 million World Bank loan, saying that the president had not sufficiently consulted with the National Assembly before agreeing to the loan (Nzeshi 2009). 15 Vreeland (2003) examines the way in which a single veto player influences the design and negotiation of IMF programs. In a study of IMF conditionality, Stone (2008) finds less conditionality in countries where there is a powerful domestic opposition.

12

counterpart funding for the same level of World Bank funding. The threat of external pressure on the Bank shifts the bargaining space in favor of the country. An extensive literature finds that the interests of important developed countries predict patterns in overall World Bank aid allocation (Frey and Schneider 1986; Andersen, Hansen and Markussen 2006; Fleck and Kilby 2006; Dreher, Sturm and Vreeland 2009). Stone (2008) finds evidence that the United States uses its informal power to reduce the level of IMF conditionality for certain countries, while Stone (2002) finds that strategically important countries are less likely to suffer from punishment if they fail to live up to their commitments to the IMF.16 Soliciting Donor Expertise. As we will show in the data, there exist a significant number of World Bank projects in which the majority of the financing comes from the recipient country itself. For instance, in September 2010, the World Bank agreed to provide $200 million worth of support to the Brazilian social welfare program Bolsa Familia. Although $200 million is a substantial loan, it is less than 1.5 percent of the project’s expected $15.2 billion cost during the period of World Bank involvement. Why has Brazil chosen to finance this small portion of the program using World Bank money? We believe that the reason lies mostly in Brazil’s desire to engage World Bank expertise in issues of program monitoring and evaluation. Where the World Bank is truly being used as a development bank, countries may approach the World Bank needing only a limited amount of additional financing in order to realize an investment project or else wanting the World Bank to contribute only its technical expertise rather than a large amount of funding. It is also possible that countries want World Bank involvement in order to provide a

16

Winters (2010), on the other hand, finds only mixed support for the hypothesis that strategic interest variables will predict project design. That paper presents some evidence that those countries that vote with the United States at the United Nations are more likely to receive national projects. On the other hand, countries that engage in significant levels of trade with the United States are less likely to receive fungible programmatic lending.

13

“good housekeeping seal of approval” to a project.17 Where this is the case, it is no longer true that bargaining leverage is going to influence the desired level of donor funding for a certain level of counterpart funding. Here the desire is for donor involvement more than for donor financing. In Bank documentation for some of the projects with very large levels of host government financing, there is clear evidence that the projects originated in and are owned by the borrowing country. Mexico pays for 95 percent of the $26.9 billion Social Protection System in Health Project. The Project Appraisal Document explicitly states that the Mexican government is interested in the World Bank’s technical input on the Mexican Popular Health Insurance program, based on the World Bank’s involvement with other insurance systems around the world (World Bank, 2010b). Similarly, China has implemented two railways projects in which the state contribution is more than 85 percent of the total project costs of $5.9 and $1.8 billion. The Project Appraisal Documents for these projects note that the government desired that the railway remain substantially financially independent and that China’s decision to seek Bank participation was based upon access to the Bank’s technical expertise gained from previous involvement with railway construction around the world (World Bank 2004b, 2009). We did not find any similar project-ownership language in a random sample of 35 PADs for projects in which the borrowing country supplied less than 30 percent of total project costs. Confounding our hypothesis that the World Bank is likely to seek more counterpart financing from poorly governed countries as a sign of commitment to development, it may be the case that wellgoverned countries are more likely to come to the World Bank with these types of projects in hand. Insofar as these countries might already have allocated funding for the development project in question, perhaps they are only looking for limited amounts of World Bank funding or are looking for the nonmonetary benefits of World Bank participation in the project. The PAD for Morocco’s National Initiative

17

This, for instance, was the case with the Chad-Cameroon Oil Pipeline Project. Although the World Bank funded less than three percent of construction costs, the consortium of oil companies involved refused to let the project go forward without World Bank involvement (Winters and Gould 2011).

14

for Human Development Support Project notes that six months prior to requesting financial assistance from the World Bank, the government requested technical support for the design and implementation of the program and that this analytical assistance contributed to their decision to select the World Bank as the key partner for the project (World Bank 2006). If countries with highly effective governments are more likely to prepare projects for which they want only additional financing or technical expertise from the Bank, then, in contrast to the hypothesis we offered above, governance might positively predict counterpart financing in projects with which the World Bank is involved.

Observing Bargaining through Counterpart Commitments: Description of the Data To assemble data on counterpart funding, we downloaded the entire World Bank Projects Database as of the end of the 2010 calendar year. We retained only those projects classified as “IBRD/IDA” projects, which represent the subset where the Bank is the least constrained in its decision making (as compared to those projects that are part of the Global Environment Facility or other World Bank-administered trust funds). These projects represent 86 percent of entries in the database. We then eliminated all development policy lending after an initial review of such projects found that they do not include counterpart commitments from borrowing countries.18 Eliminating pipeline projects, dropped projects and projects that listed no lending amount, we were left with 9,726 IBRD and IDA projects dating back to 1948. For this paper, we use a random sample of 1,565 projects (out of 2,633) from the years 2000-2010. For each project, we recorded from the Project Appraisal Document (PAD) the amount of project funds coming from both government and non-government sources in the aid-receiving country. When the PAD was unavailable, we made use of other documents found in the projects database (usually the Project Paper). We also recorded this same data from the “financial tab” found in the 18

This makes sense in terms of policy lending going directly into the national treasury without having a dedicated purpose.

15

Projects Database. The PAD-derived data reflect counterpart commitments at the time of project approval by the World Bank Board of Directors. According to the Projects Database, the financial tab data are supposed to match the information contained in the PAD.19 However, we find discrepant reporting of the counterpart amounts between the two sources for about 20 percent of cases. We comment on these discrepancies in the online appendix. Because of the way that additional funding for projects is recorded in the project database, a number of our observations are not new projects but rather supplementary funds for existing projects. These instances of supplementary funding often include counterpart funding; in the online appendix, we provide results from analyses where we have removed these cases, since the decision making about these project ledger entries may be different from initial decision making about new projects. In the next section, we provide descriptive results at the project-level using two outcome variables: an indicator for whether or not a project has any counterpart funding commitment at all and then the ratio of counterpart funding to the total project size. The total project size includes the World Bank’s commitment, funding from other multilateral and bilateral donors and the counterpart commitments if any. The descriptive statistics reveal a variety of counterpart funding relationships, providing evidence of both country-led World Bank projects and the bargaining relationship that exists between the World Bank and its borrower countries. In the subsequent section, we use multivariate regression to identify the key predictors of levels of counterpart funding.

Country-Driven Projects and Bargaining Over Counterpart Funds Of the 1,565 projects in our dataset, three-quarters of them feature some level of counterpart funding at approval. On average, almost one-quarter of project financing is expected to come from

19

The tab reports, “Total project cost includes funding from World Bank and non-bank sources in US$ millions. Active and Closed projects show commitment at Board approval. It does not reflect any cancellations.”

16

within the borrowing country. Figure 3 and Figure 4 show the way in which counterpart funding commitments are distributed across different project types. In Figure 3, we show the relationship between total project size and the relative amount of counterpart financing. There is a positive association between the two with larger projects requiring more financing from sources within the borrowing country. First, the plot makes clear that there are some very large projects where the borrowers supply almost all of the financing. The countries involved in these projects are some of the biggest World Bank borrowers: Argentina, Brazil, China, Mexico, India and Indonesia. As described above, these projects are borrower-driven projects in which the World Bank contributes only a very small portion of the financing and likely has been brought into the project because of the value of its expertise or its seal of approval. Second, the plot shows the range of outcomes that are reached with regard to financing in World Bank projects. Across different borrowers, different lending wings of the World Bank and different project sizes, we see a variety of levels of counterpart funding. Visualizing the data through this plot, it is clear that there is no set standard for how a World Bank project looks in terms of counterpart funding – a variety of outcomes are reached in the negotiations that go on between the World Bank and its borrower countries. The plot also makes clear the level of variation possible within individual countries. India has had similarly-sized investment projects where only a small portion of the money (less than 20 percent) comes from the World Bank and where a large portion of the money (over 80 percent) comes from the World Bank. Figure 4 shows the proportion of projects with any counterpart funding and the average percent of project funding that come from borrowing-country counterparts across the World Bank’s classifications of project type. In infrastructure-heavy sectors, such as water and sanitation, transportation, and energy and mining, we find both higher proportions of projects with counterpart

17

funding and higher levels of counterpart funding within those projects.20 Projects that support programming in the information and communications, finance and public administration sectors, on the other hand, are less likely to involve counterpart financing and involve somewhat smaller amounts when they do. The agriculture, fishing and forestry sector is an interesting case. Although projects in this sector are the most likely type of projects to include counterpart financing, the amount of financing is relatively less than for most other sectors. Therefore, in our first glimpse at the data, we find that infrastructure-heavy projects are likely to involve larger amounts of counterpart funding and that the costliest projects with which the World Bank is involved are often majority-financed by sources within the borrowing country. Most importantly, we find a range of levels of counterpart financing. For instance, 10 percent of all projects in the dataset are majority-financed by the borrowing country, while 18 percent of all projects in the dataset are solely financed by the World Bank. That the borrower financing in a World Bank project can range from none to almost the entire project shows that World Bank projects are being created in different ways and with different outcomes reached through negotiations between the Bank and its borrowers.

20

Note that the variation in whether or not projects have any counterpart funding comes from 2004 and later; in the pre-2004 data, 97 percent of all projects in the dataset include counterpart financing.

18

Figure 3. Project Size and Counterpart Financing. IBRD projects in blue; IDA projects in green; blend projects in purple. The trend line is a locally-weighted regression line.

Figure 4. Counterpart Funding by Project Sector. Sector is defined as the modal sector to which project financing went; if no single mode existed, the project was labeled as multisectoral. The right-hand plot includes only projects with positive levels of counterpart financing.

19

Empirical Predictors of the Size of Counterpart Financing Having shown that there exists substantial variation in the levels of counterpart funding across World Bank projects, we turn in this section to identifying some of the characteristics that might predict this variation. In particular, we are interested in governance and democracy as proxies for a country’s commitment to development, GDP per capita and debt-to-GNI ratio as proxies for a country’s budget constraint, ODA-to-GNI ratio as a proxy for the outside options available to the borrowing country, the level of outstanding loans to the World Bank and the part of the Bank in which the loan originates as proxies for the Bank’s budget constraint, and a measure of institutional veto players within the country as a proxy for negotiating leverage that might result from more domestic hand-tying (Putnam 1988). At the end of the section, we examine whether geostrategic variables affect the bargaining leverage of recipient states. In order to retain as much information as possible, we use multilevel models of projects nested within country-years. We include both project-level and country-year-level variables as predictors of the logarithmic transformation of the total amount of counterpart funding found in a given project. On the right hand side of the equation, we include measures of the World Bank’s contribution to the project and other donor contributions to the project, both logarithmically transformed. In order to address the threat of bias induced by correlation between the varying intercepts in the multilevel model and the project-level predictor variables, we follow the specification recommended in Zorn (2001) and Bartels (2008): at the country-year level, we include group averages for these two main project-level predictor variables and then we include demeaned versions of those variables at the project-level (see also Bafumi and Gelman 2006). This reduces the threat of bias in the multilevel estimation and means that the project-level coefficients are estimated entirely off of within-country-year variation, whereas the grouplevel averages represent the effect estimated off of between-country-year variation. The estimating equations are:

20

""""""""""""""""""""""""""""""""""" Bank Amount !  logcounterpart funds   β  β logWorld Bank Amount    logWorld

"""""""""""""""""""""""""""""""""""" β# logOther Donor Amount    logOther Donor Amount !  β' IBRD  β* Blend  +  , β  -  - """""""""""""""""""""""""""""""""" logWorld Bank Amount   -# """"""""""""""""""""""""""""""""""" logOther Donor Amount.   -' Governance  -* Democracy  -2 logGDP per capita   -5 Governance  -6 log 7  -; log 7

Debt : GNI 

ODA :  -< logOutstanding World Bank Loans   - Veto Players  η GNI 

where β is a varying intercept, + is a set of project-level sector fixed effects, , is a normally-

distributed project-level error term and η is a normally-distributed country-year-level error term. All of the variables are defined in the appendix. With regard to the governance variable, we use the government effectiveness measure from the Worldwide Governance Indicators. This variable “captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies.”21 The measure, therefore, seems to best capture the idea of a country’s being committed to development. In addition, in simple regressions predicting the amount of counterpart funding in a given project using the various Worldwide Governance Indicators and controlling for the total project size, all six indicators are significant positive predictors of counterpart funding.22 When all six Worldwide Governance Indicators are included together (along with total project size) in a regression predicting counterpart funding, government effectiveness is statistically significant and has the largest coefficient;

21

Worldwide Governance Indicators, “Government Effectiveness,” http://info.worldbank.org/governance/wgi/pdf/ge.pdf (accessed 2 March 2012). 22 The high collinearity among the indicators is well known, so this is not surprising. The results of these regressions can be found in the online appendix.

21

political stability and control of corruption are also statistically significant, although the latter is a surprisingly negative predictor of counterpart funding. Table 1 presents the results from a series of multilevel models predicting the amount of counterpart funding in a given World Bank project. In the first three columns, we make use of all the projects in the dataset, whereas in the second set of three columns, we use only the subset of projects that were majority-financed by the Bank (possibly in combination with other donors), removing from the data any projects that had greater than 50 percent country financing. The first column in each set of three regressions pools all observations together before and after the 2004 rule change, whereas the second column looks at only those projects approved before the rule change and the third column looks at only those projects approved after the rule change. The first thing that the table reveals is that counterpart commitments rise as World Bank commitments rise. For a project in any given country-year, a one percent increase in the World Bank commitment to the project leads to a slightly greater than one percent increase in counterpart commitments. As described in the theoretical section, the willingness of each partner to contribute money to the project rises as the other entity contributes more money to the project. In addition, across country-years, we find that more World Bank lending, on average, in a given country-year increases the likely amount of counterpart funds for any project in that country-year. The table also reveals major differences between the pre-2004 period and the later period. In the pre-2004 data (as seen in columns 2 and 5), almost none of the predictors are significant. Variation in the size of the World Bank contribution – both within a country-year and between country-years – positively predicts counterpart commitments in this period, as done the between-group variation in the amount of funding coming from other donors. However, none of the country-year characteristics, except governance, significantly predict counterpart commitments in the pre-2004 data. On the other hand, several country-year variables become significant predictors in the post-2003 period and often

22

with different signs from the estimates in the earlier period. In addition, in the post-2003 period between-country variation in average World Bank funding ceases to be a significant predictor of counterpart funding; the effect of variation in Bank funding on counterpart funding is a within-country phenomenon. This reflects the new homogeneity across countries in terms of the financing parameters that the Bank has established for them. This difference across periods reflects the increasing discretion that World Bank project managers had when considering counterpart funding after the 2004 rule change. In the post-2003 period, wealthier countries are asked to contribute more in counterpart funding, with a one percent increase in GDP per capita corresponding to a 1.5 percent increase in counterpart funds, suggesting a new responsiveness to government budget constraints. Countries that have more outstanding World Bank loans also are asked to contribute more in counterpart funding, suggesting a new responsiveness to donor budget constraints. In the post-2003 period, IBRD loans involve lower counterpart commitments as compared to IDA loans, suggesting that the Bank negotiates harder bargains in the case of IDA lending, where resources are more scarce. The relatively small and insignificant coefficients on blend projects suggest that these projects are treated more like IDA projects. Also, a number of the sector indicators, which are not significant in the earlier period become significant in the later period: public administration, communications and finance projects all involve less counterpart funding relative to the baseline of agriculture projects (results not reported). There is also some evidence that more heavily indebted countries are asked to make larger counterpart commitments, which is contrary to our expectations that the opportunity costs of contributing counterpart funds are larger for such countries. And there is some evidence that counterpart commitments are smaller for countries with higher levels of aid dependence, which matches with our expectation that these countries have more outside options besides the World Bank and can therefore negotiate better bargains.

23

Regarding borrower commitment to development, we find different results across the two proxies that we have included. Government effectiveness, as we described above, positively predicts counterpart funding – the opposite of our expectations. This is true even when looking only at the set of projects that are majority Bank-financed (in columns 4 through 6). On the other hand, democracy – in the overall data and in the recent period – negatively predicts counterpart funding for any given country year, although the effect does not obtain conventional significance levels.23 Insofar as democratic governments are understood to be dedicated to development, the negative coefficient on democracy can be taken as a sign that such commitment to development leads to greater World Bank willingness to fund projects given smaller counterpart commitments. However, the lack of significance for this variable means that we cannot put much stock in these results. With regard to governance, there clearly seems to be an indication that more effective governments contribute more counterpart funds to World Bank projects – the variable is positive and marginally significant even in the pre-2004 period. As described above, we think that these governments might simply be better at preparing projects for the World Bank to fund: they come to the World Bank with a well-developed project proposal that already involves domestic funding, and the World Bank agrees to join the project. That this relationship holds even in the set of projects that are majority-financed by the World Bank suggests that this is true even for projects where the country is not simply seeking expertise or a seal of approval but rather has particular financing needs that it is looking to meet. Our multivariate analysis reveals several important patterns in the determinants of counterpart funding. First, we see evidence for the basic trade-off that our theoretical model assumes: counterpart

23

There is a concern about including both democracy and governance in the same estimating equation, given the high correlation between them: the effect of one variable might be estimated off only a small (and potentially unusual) set of cases after controlling for the other. That does not appear to be the case here. The online appendix shows that the coefficient estimates are relatively stable when including or not including one variable or the other.

24

funding increases with World Bank funding. Each side involved in negotiating a World Bank project puts in more money as the other side does. Second, we see the effect of the 2004 rule change. Whereas levels of counterpart funding in the first period in the data are not obviously related to the country characteristics that we have included in the model, in the second period, once World Bank officials have been granted more discretion over the requirements for counterpart commitments, we observe a clear set of correlates of counterpart financing. In particular, we find that wealthier countries and countries with more outstanding debt to the World Bank contribute more of their own resources to a project, and we find some evidence that more aid-dependent countries contribute less of their own resources. Finally, we find evidence that better governed countries contribute more funding to development projects in their borders with World Bank involvement; we think that this is due to these countries being more likely to come to the World Bank with shovel-ready projects to which they are willing to make a contribution.

25

DV: Log(Counterpart Funding)

Project-Level Variables Log(World Bank Amount), Demeaned Log(Other Donors Amount), Demeaned IBRD (0/1) Blend (0/1)

(1) All Projects

(2) FY99-FY03

(3) FY04-FY10

(4) Majority BankFinanced Projects

1.32***

1.00***

1.33***

(0.26)

(0.37)

(0.32)

0.05

-0.02

(0.03)

(0.04)

Log(Debt/GNI)

(6) BankFinanced FY04-FY10

1.09***

0.71*

1.06***

(0.32)

(0.43)

(0.39)

0.08*

0.03

-0.04

0.08

(0.04)

(0.04)

(0.05)

(0.05)

-1.64**

-0.17

-1.53

-1.80**

-0.24

-1.72

(0.76)

(0.77)

(0.99)

(0.81)

(0.82)

(1.07)

-0.00

0.22

-0.07

0.13

-0.28

0.37

(1.16)

(1.77)

(1.36)

(1.26)

(1.87)

1.16*

1.61*

2.29***

1.07

1.62*

(0.65)

(0.86)

(0.69)

(0.68)

(0.93)

0.45

-1.08

-0.74

0.32

-0.86

(1.26) Country-Year-Level Variables Government 2.16*** Effectiveness (0.64) Democracy (0/1) -0.88* Log(GDP per capita, PPP USD)

(5) BankFinanced FY99-FY03

(0.53)

(0.55)

(0.70)

(0.55)

(0.56)

(0.74)

0.87*

-0.36

1.51**

0.92*

-0.15

1.54**

(0.50)

(0.50)

(0.66)

(0.53)

(0.53)

(0.70)

0.93***

-0.20

0.37

0.99***

-0.37

0.36

(0.35)

(0.44)

(0.46)

(0.37)

(0.49)

(0.48)

Log(ODA/GNI)

-0.62**

-0.37

-0.51

-0.50*

-0.21

-0.35

(0.27)

(0.29)

(0.36)

(0.29)

(0.31)

(0.39)

Log(Outstanding World Bank Loans)

0.46**

-0.15

0.70**

0.39

-0.27

0.72**

(0.23)

(0.21)

(0.32)

(0.25)

(0.22)

(0.35)

0.05

-0.15

0.08

0.06

0.02

0.08

Veto Players

(0.12)

(0.20)

(0.14)

(0.13)

(0.21)

(0.15)

Log(World Bank Amount), Average

0.64**

0.60**

0.41

0.62**

0.58**

0.35

(0.28)

(0.25)

(0.40)

(0.30)

(0.27)

(0.42)

Log(Other Donors Amount), Average

0.18***

0.10***

0.18***

0.18***

0.11***

0.17***

(0.04) 910 471 0.24 0.08 0.22

(0.04) 260 149 0.21 0.15 0.24

(0.06) 650 322 0.27 0.08 0.23

(0.04) 803 449 0.20 0.03 0.19

(0.04) 232 142 0.17 0.15 0.21

(0.06) 571 307 0.23 0.05 0.20

Projects Country-Years 2 Overall R 2 Within R 2 Between R

Table 1. Multilevel Analysis of Predictors of Counterpart Funding. Data comes from the Project Appraisal Document (PAD), Project Paper or similar pre-implementation documentation. Projects are clustered in country-years, and a varying intercept is estimated for each country-year. All regressions include unreported sector indicators. * - p < 0.10; ** - p < 0.05; *** - p < 0.01.

26

How Deep Does External Influence Go? The Strategic Manipulation of Counterpart Commitments Previous literature has found evidence of U.S. influence at the World Bank when it comes to the Bank’s overall patterns of aid allocation (Frey and Schneider 1986; Andersen, Hansen and Markussen 2006; Fleck and Kilby 2006; Dreher, Sturm and Vreeland 2009). However, there is less evidence that this influence extends to the level of project design (Mosley, Harrigan and Toye 1991; Winters 2010); that is, the U.S. may be able to influence which countries are getting money, but decisions about how exactly they get that money are made by World Bank staff.24 Therefore, on the one hand, we might expect that countries of strategic interest to the U.S. would be able to leverage their geostrategic relationship to demand smaller counterpart commitments in World Bank projects, such that they would be receiving aid money more cheaply (i.e. with fewer conditions). But on the other hand, we might expect World Bank staff to maintain operational independence when it comes to decisions on counterpart financing, such that strategic interest variables that predict overall World Bank aid allocation do not, in fact, predict counterpart commitments. We use five variables to identify countries in which the United States might have a particular interest and thereby examine the role of U.S. influence at the World Bank when it comes to decisions on counterpart funding. In a series of regression models, we proxy for U.S. interest with the amount of ODA and the amount of military assistance that the United States sends to a given World Bank borrower (as reported to the OECD-DAC and in the U.S. Greenbook respectively), with a measure of the similarity of voting at the United Nations (a measure based on Barro and Lee (2005) from the updated dataset associated with Dreher and Sturm (2006)), and with measures of U.S. exports to and total trade with a given World Bank borrower (data from the OECD web site). We include these strategic variables in each of the six models from Table 1. The inclusion of the strategic variables does not change any of the core results reported above. Government effectiveness remains a significant, positive predictor of 24

That said, there is evidence that U.S. influence mediates the severity of punishment for non-compliance with World Bank (and IMF) agreements (Mosley, Harrigan and Toye 1991; Stone 2002; Gould and Winters 2007).

27

counterpart funding. We also see the same patterns of wealthier countries and countries with more outstanding debt to the Bank contributing larger counterpart commitments and more aid-dependent countries contributing less funding towards projects. Therefore, we only present coefficient results for the strategic interest variables in the table below.

DV: Log(Counterpart Funding)

(1)

(2)

(3)

All Projects

FY99-FY03

FY04-FY10

-0.66* (0.40) -0.11 (0.10) 5.42 (3.63) -0.47 (0.41) -0.03 (0.37)

0.28 (0.45) 0.03 (0.11) -1.98 (4.02) 0.45 (0.39) 0.36 (0.35)

-0.96* (0.52) -0.12 (0.13) -4.42 (5.25) -0.48 (0.56) -0.04 (0.51)

(4) Majority BankFinanced Projects

(5)

(6)

BankFinanced FY99-FY03

BankFinanced FY04-FY10

0.70 (0.49) 0.15 (0.12) -1.25 (4.16) 0.42 (0.42) 0.31 (0.38)

-1.08* (0.56) -0.21 (0.14) -3.05 (5.50) -0.73 (0.58) -0.27 (0.53)

Strategic Variables Log(U.S. ODA) Log(U.S. Military) U.S.-U.N. Voting Similarity Log(U.S. Exports) Log(U.S. Trade)

-0.69 (0.43) -0.15 (0.10) 7.72** (3.79) -0.66 (0.43) -0.22 (0.40)

Table 2. Results for Strategic Interest Variables in Multilevel Analysis of Predictors of Counterpart Funding. * - p < 0.10; ** - p < 0.05; *** - p < 0.01 Coefficient estimates come from regressions corresponding to columns (1) through (6) in Table 1.

In the FY04-FY10 period, when World Bank officials gain more discretion over project design, all of the indicators of U.S. affinity are negatively signed, suggesting that countries regarded as important by the United States made smaller counterpart commitments on average than other countries (even after controlling for the amount of other donor financing in a given project). This is true for all projects and for majority-Bank-financed projects. Only the coefficient on U.S. ODA rises to the level of statistical significance. The relationship between U.S. ODA and counterpart commitments means that when World Bank officials have more discretion over counterpart commitments, borrowers who receive more development assistance from the U.S. make smaller counterpart commitments. This result may indicate that the World Bank officials are accepting U.S. participation in development in the borrowing country as an indicator of the country’s overall commitment to implementing development reforms, or it could

28

be evidence of successful U.S. lobbying at the World Bank on behalf of favored countries for them to contribute less of their own resources to projects. There is also some evidence in the overall data (for either all projects or majority Bank-financed projects) that countries that vote with the United States at the United Nations make larger counterpart commitments to World Bank projects. Oddly, this relationship becomes negative in either of the timedivided subsets of the data. It is not clear what is causing this inconsistency in estimation.

Summary: Patterns in Splitting the Check on Development Projects Foreign aid projects are negotiated between donors and recipients. In the extant literature, there often appears an implicit assumption that these negotiations are rather one-sided, with donors giving what they want and recipients accepting whatever they get. We question this donor-dominant view by looking at the levels of counterpart funding in World Bank projects and proposing factors that would influence the preferences of the donor and the recipient state for certain levels of counterpart funding. We argue that the World Bank is likely to want more counterpart funds from poorly governed and non-democratic states because of the need for a sign of commitment to development among these states; we believe that both sides might have varying levels of interest in the project and varying budget constraints; and finally, we propose that recipient states might have more bargaining power where they have outside options for foreign assistance or are of strategic importance to the Bank’s rich-country principals. Our descriptive results clearly show evidence of bargaining over counterpart commitments: there is no single formula that determines the level of counterpart funding, and we can observe in the data both projects with no counterpart funding and where the country is financing almost the entire project. These latter projects are clear examples of country-driven borrowing from the World Bank,

29

perhaps motivated more by a desire for expertise or the Bank’s seal of approval than by a need for capital from the Bank. The results from our multivariate analysis reveal that an extra dollar of World Bank money does, in fact, require an extra dollar of counterpart funding on average – again suggesting a bargaining relationship between the Bank and its borrowers. Perhaps more importantly, we see that country characteristics have become more relevant predictors of counterpart funding since a 2004 rule change at the World Bank gave task team leaders more leeway to choose levels of country financing. Since 2004, wealthier countries and countries with outstanding World Bank loans contribute more counterpart funding. Countries that are borrowing from the IDA, where there is a hard budget line for how much countries can borrow, also contribute more counterpart funding. There is some evidence that democracies contribute lower levels of counterpart funding to World Bank projects, perhaps as a result of their presumed commitment to development. Wellgoverned countries, on the other hand and against some of our initial expectations, contribute more money. We believe that well-governed countries are more likely to come to the Bank with shovel-ready projects for which they are already willing to commit a certain level of financing. This relationship appears to have been the same – although less intense – before the 2004 rule change. When looking at variables proxying for the level of U.S. interest in a given country, we find that levels of U.S. aid are a negative and marginally significant predictor of counterpart funding, particularly in the post-2003 period. Other proxies are also negatively signed for the post-2003 period, perhaps providing evidence that the United States can, in fact, influence project design decisions at the World Bank. Because of the lack of statistical significance, however, we are cautious about putting too much emphasis on these results. In conclusion, we have shown new evidence of the bargaining relationship that goes on between the World Bank and its borrowers over one element of project design. Whereas previous

30

studies have looked at the way in which overall aid flows react to borrower characteristics, we are able to show some theoretically and empirically consistent patterns in terms of how project design elements react to borrower characteristics and to other characteristics of the project. That we provide evidence of borrowing countries having some leverage with the World Bank serves as a rejoinder to those who generally assume a donor-driven model of aid allocation.

31

Works Cited Alesina, Alberto and David Dollar. 2000. “Who Gives Foreign Aid to Whom and Why?” Journal of Economic Growth 5.1: 33-63. Andersen, Thomas Barnebeck, Henrik Hansen and Thomas Markussen. 2006. “U.S. Politics and World Bank IDA-Lending,” Journal of Development Studies 42.5: 772-94. Azam, Jean-Paul and Jean-Jacques Laffont. 2003. “Contracting for Aid,” Journal of Development Economics 70: 25-58. Barro, Robert J. and Jong-Wha Lee. 2005. “IMF Programs: Who is Chosen and What are the Effects?” Journal of Monetary Economics 52: 1245-69. Baum, Matthew A. and David A. Lake. 2003. “The Political Economy of Growth: Democracy and Human Capital,” American Journal of Political Science 47.3 (April): 333-47. Bayer, Patrick, Christopher Marcoux and Johannes Urpelainen. 2012. “When International Organizations Bargain: Theory and Evidence,” Columbia University, New York. Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer, and Patrick Walsh. 2001. "New tools in comparative political economy: The Database of Political Institutions," World Bank Economic Review 15:1, 165-176. Berthélemy, Jean-Claude. 2006. “Bilateral Donors’ Interest vs. Recipients’ Development Motives in Aid Allocation: Do All Donors Behave the Same?” Review of Development Economics 10.2: 179-94. Berthélemy, Jean-Claude and Ariane Tichit. 2004. “Bilateral Donors’ Aid Allocation Decisions – A ThreeDimensional Panel Analysis,” International Review of Economics and Finance 13: 253-74. Bueno de Mesquita, Bruce and Alastair Smith. 2007. “Foreign Aid and Policy Concessions,” Journal of Conflict Resolution 51.2: 251-84. -----. 2009. “A Political Economy of Aid,” International Organization 63.2: 309-40. Burnside, Craig and David Dollar. 2000. “Aid, Policies, and Growth,” American Economic Review 90.4: 847-68. -----. 2004. “Aid, Policies, and Growth: Revisiting the Evidence,” World Bank Policy Research Working Paper 3251, Washington, D.C. Cheibub, Jose Antonio, Jennifer Gandhi and James Raymond Vreeland. 2010. “Democracy and Dictatorship Revisited," Public Choice 143.1-2: 67-101. de Renzio, Paolo with David Booth, Andrew Rogerson and Zaza Curran. 2005. “Incentives for Harmonization and Alignment in Aid Agencies,” Overseas Development Institute Working Paper No. 248, London. Dollar, David and Victoria Levin. 2006. “The Increasing Selectivity of Foreign Aid, 1984-2003,” World Development 34.12: 2034-46. Dreher, Axel and Nathan M. Jensen. 2007. “Independent Actor or Agent? An Empirical Analysis of the Impact of U.S. Interests on International Monetary Fund Conditions,” Journal of Law and Economics 50.1: 105-24. Dreher, Axel, Jan-Egbert Sturm and James Raymond Vreeland. 2009. “Development Aid and International Politics: Does Membership on the U.N. Security Council Influence World Bank Decisions?” Journal of Development Economics 88.1: 1-18. Easterly, William. 2002. “The Cartel of Good Intentions: The Problem of Bureaucracy in Foreign Aid,” Journal of Policy Reform 5.4: 223-50. -----. 2007. “Are Aid Agencies Improving?” Economic Policy 22.52: 633-78. Fleck, Robert K. and Christopher Kilby. 2006. “World Bank Independence: A Model and Statistical Analysis of U.S. Influence,” Review of Development Economics 10.2: 224-40. Frey, Bruno and Friedrich Schneider. 1986. “Competing Models of International Lending Activity,” Journal of Development Economics 20.2: 225-45. 32

Gibson, Clark C., Krister Andersson, Elinor Ostrom and Sujai Shivakumar. 2005. The Samaritan’s Dilemma: The Political Economy of Development Aid. New York: Oxford University Press. Girod, Desha. 2008. “Cut from the Same Cloth? Bilateral vs. Multilateral Aid,” paper presented at the annual meeting of the International Political Economy Society, Philadephia. Gould, John A. and Matthew S. Winters. 2007. “An Obsolescing Bargain in Chad: Explaining Shifts in Leverage Between the Government and the World Bank,” Business and Politics 9.2. Killick, Tony. 1998. Aid and the Political Economy of Policy Change. With Ramani Gunatilaka and Ana Marr. New York: Routledge. Lake, David A. and Matthew A. Baum. 2001. “The Invisible Hand of Democracy: Political Control and the Provision of Public Services,” Comparative Political Studies 34.6: 587-621. Maizels, Alfred and Machiko Nissanke. 1984. “Motivations for Aid to Developing Countries,” World Development 12: 879-900. McKinlay, Robert D. and Richard Little. 1979. “The U.S. Aid Relationship: A Test of the Recipient Need and Donor Interest Models,” Political Studies 27: 236-50. Mosley, Paul. 1998. “On Persuading a Leopard to Change His Spots: Optimal Strategies for Donors and Recipients of Conditional Aid,” in Robert H. Bates, ed. Toward a Political Economy of Development: A Rational Choice Perspective. Los Angeles: University of California Press. Mosley, Paul, Jane Harrigan and John Toye. 1991. Aid and Power: The World Bank and Policy-Based Lending. Vol. 1. New York: Routledge. Neumayer, Eric. 2003. The Pattern of Aid Giving: The Impact of Good Governance on Development Assistance. New York: Routledge. Nzeshi, Onwuka. 2009. “House Rejects US$300 Million World Bank Loan,” This Day [Lagos, Nigeria], 10 December. Ostrom, Elinor with Clark Gibson, Sujai Shivakumar and Krister Andersson. 2001. “Aid, Incentives, and Sustainability: An Institutional Analysis of Development Cooperation,” Sida Studies in Evaluation 02/01, Stockholm. Putnam, Robert D. 1988. “Diplomacy and Domestic Politics: The Logic of Two-Level Games,” International Organization 42.3: 427-60. Steele, Carie A. 2012. “Disease Control and Donor Priorities: The Political Economy of Development Aid for Health,” paper presented at the annual meeting of the European Political Science Association, Berlin, June. Stone, Randall W. 2002. Lending Credibility: The International Monetary Fund and the Post-Communist Transition. Princeton: Princeton University Press. -----. 2004. “The Political Economy of IMF Lending in Africa,” American Political Science Review 98.4: 577-91 -----. 2008. “The Scope of IMF Conditionality,” International Organization 62: 589-620. Svensson, Jakob. 2000. “When is Foreign Aid Policy Credible? Aid Dependence and Conditionality,” Journal of Development Economics 61: 61-84. -----. 2003. “Why Conditional Aid Does Not Work and What Can Be Done About It?” Journal of Development Economics 70: 381-402. Thacker, Strom C. 1999. “The High Politics of IMF Lending,” World Politics 52.1: 38-75. Vreeland, James. 2003. The IMF and Economic Development. New York: Cambridge University Press. Williamson, Claudia R. 2010. “Exploring the Failure of Foreign Aid: The Role of Incentives and Information,” Review of Austrian Economics 23.1: 17-33. Winters, Matthew S. 2010. “Choosing to Target: What Types of Countries Get Different Types of World Bank Projects,” World Politics 62.3: 422-58. -----. 2011. “Targeted Aid and Capture in World Bank Projects,” unpublished manuscript, University of Illinois at Urbana-Champaign. 33

----- and John Gould. 2011. “Betting on Oil: The World Bank’s Attempt to Promote Accountability in Chad,” Global Governance 17: 226-46. Woods, Ngaire. 2006. The Globalizers: The IMF, the World Bank and Their Borrowers. Ithaca: Cornell University Press. World Bank. 1998. Assessing Aid: What Works, What Doesn’t, and Why. New York: Oxford University Press. -----. 2004a. “Eligibility of Expenditures in World Bank Lending: A New Policy Framework,” Operations Policy and Country Services, 26 March, available at http://www1.worldbank.org/operations/eligibility/documents/March26ExpenditureEligibilityBo ardPaper.pdf (accessed 28 February 2012). -----. 2004b. “Project Appraisal Document for Second National Railways Project,” 28 May, available at http://wwwwds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2004/06/09/000090341_20 040609112506/Rendered/PDF/29067.pdf (accessed 28 February 2012). -----. 2005. “Eligibility of Expenditures in World Bank Financing: FY05 Report on Implementation Experience,” Operations Policy and Country Services, 3 October, available at http://siteresources.worldbank.org/INTOPEELI/641683601132754290708/20734079/ExpenditureEligibilityFY05AnnualReport.pdf (accessed 28 February 2012). -----. 2006. “Project Appraisal Document for National Initiative for Human Development Support Project,” 14 November, available at http://wwwwds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2006/11/22/000310607_20 061122101032/Rendered/PDF/36973.pdf (accessed 3 March, 2012). -----. 2007. “Operational Memorandum: Specific Expenditure Eligibility and Cost Sharing Requirements for Investment Projects in Countries Without Approved Country Financing Parameters,” available at http://go.worldbank.org/51SUT8YBC0 (accessed 28 February 2012). -----. 2009. “Project Appraisal Document for Nanguang Railway Project,” 29 May, available at http://wwwwds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2009/06/09/000333037_20 090609010825/Rendered/PDF/481320PAD0P112101Official0Use0Only1.pdf (accessed 28 February 2012). -----. 2010a. “Project Appraisal Document for a Water, Electricity and Urban Development Project,” 1 March, available at http://wwwwds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2010/03/09/000334955_20 100309011630/Rendered/PDF/529950PAD0P106101Official0Use0Only1.pdf (accessed 6 March 2012). -----. 2010b. “Project Appraisal Document for Social Protection System in Health Project,” 23 February, available at http://wwwwds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2010/03/08/000333037_20 100308233341/Rendered/PDF/521420PAD0P116101Official0Use0Only1.pdf (accessed 3 March 2012).

34