Organizational reform and the rise of trust funds: Lessons from the World Bank

Abstract

Over the past two decades, earmarked funding to international development organizations through special-purpose trust funds has increased greatly. This paper studies the incentives for trust funds from the perspective of multilateral agencies, notably the World Bank. A theoretically intriguing type of funds are so-called “pass-on funds,” in which one unit hosts the fund, then passes on its resources to another type of unit for implementation. Each unit has different preferences for the specific types of activities to be supported by the fund. Interviews with World Bank staff and complementary documents demonstrate the rationale for pass-on funds and the associated division of labor between fundraising network units and implementing regional units. While pass-on funds reflect an efficient division of labor between functionally specialized units, they increase the misalignment between sector-specific global priorities and country-specific needs. Organizational reform drove the sudden explosion of pass-on funds around the millennium turn, facilitated by growing availability of donor monies for specific sectors and by lenient internal regulation. Organizational reform undermined budget autonomy of sector units, causing those units to seek new funds in their areas of expertise. A number of reform features also reduced administrative budgets of country units, increasing their demand for pass-on funding grants. The results contribute to the emerging literature on earmarked funding and highlight the need to consider international organizations as heterogeneous actors.

1 Introduction

Trust funds, channeling grant resources to international development organizations for specific development purposes, have become an important tool in the global aid architecture. Trust funds are popular with donors because of their tremendous flexibility, compared to traditional sources of multilateral finance (e.g., OECD 2011; Reinsberg et al. 2016; Eichenauer and Reinsberg 2017).

While trust funds have proliferated in the last twenty years, their governance arrangements also have become increasingly complex. An exemplary case of such complexity is represented by the so-called “pass-on funds” at the World Bank. Unlike trust funds that individual units raise to support their own work program, pass-on funds entail a division of labor within the multilateral bureaucracy, with a global sector unit raising the funds, managing the overall program, and relating to donors, while a country unit implements the programs financed by these funds. This feature raises important theoretical questions. How do pass-on funds emerge? What makes this institutional arrangement beneficial to the relevant units? Which conditions sustain the implied division of labor? A related question pertains to timing. Why did pass-on funds take off suddenly around the millennium turn?

This paper answers these questions by studying the related historical developments at the World Bank. While focused on the World Bank, the present analysis may be generalized to all international development organizations. All major international development organizations consist of headquarters and field offices, with headquarters often developing sector-specific programs and field offices implementing these program-supported activities at the country level. At the World Bank, “Sector Managing Units” (SMUs) prepare specific projects, while “Country Managing Units” (CMUs) interact with recipient countries and manage project portfolios in multiple sectors. This division of labor delivers quality projects relevant to specific country needs, but functional specialization also perpetuates differences among different units. Country units emphasize specific country needs while sector units also promote their sectors irrespective of geographical boundaries.

In addition to describing recent trust fund history and development of pass-on funds, this paper explores goal conflicts and bargaining processes in functionally differentiated organizations (e.g., Barnett and Finnemore 2004; Weaver 2008; Graham 2013), links between organizational structure and organizational performance (e.g., Barnett and Finnemore 1999; Caughey et al. 2009; Bauer and Ege 2016), and determinants of trust funds. Previous studies on the determinants of trust funds emphasize the donor perspective, some arguing that trust funds allow donors to regain control over unfaithful agents (Sridhar and Woods 2013), others focusing on divisions among donors to explain why trust funds have proliferated (Graham 2015; Bayram and Graham 2016; Reinsberg et al. 2016). Motivations for multilateral agencies to establish trust funds have not been systematically analyzed, though they have been discussed cursorily (e.g., Michaelowa and Michaelowa 2011; Mahn 2012; Reinsberg et al. 2015).

A focus on pass-on funds facilitates more thorough analysis of agency motivations and generates new explanations for trust funds beyond the obvious explanation that agencies use the funds to grow their budgets. Organizational reforms can play a key role by generating powerful incentives to create new funds. For example, the “Strategic Compact” reforms initiated by the World Bank in 1997 undermined budget autonomy of sector units, which encouraged these units to use pass-on funds to regain autonomy. While the rapid growth of pass-on funds may help the World Bank increase its position against multilateral competitors, it most likely undermined ongoing efforts to increase the relevance of development projects for specific country priorities. Because pass-on funds represent sector interests at the global level, they inherently ignore specific country interests (IEG 2011).

My analysis draws primarily on qualitative evidence from semi-structured interviews conducted at the World Bank headquarters in 2013, augmented by trust fund program documents, annual budget reports, and evaluation studies from the World Bank. My choice of method recognizes that the creation of pass-on funds as a specific form of financing is a singular event, but once established, pass-on funds involve repeated interactions between host units and implementing units that can be studied quantitatively. There is an apparent link between the reasons for establishment of pass-on funds and patterns governing their use. Web Appendix A provides additional evidence consistent with the findings presented here.¹

I proceed as follows. Section 2 discusses historical developments of trust funds at the World Bank, introduces the phenomenon of pass-on funds, and specifies research questions. Section 3 reviews theoretical literature on trust funds, identifies current gaps, and demonstrates how analyzing pass-on funds enhances our understanding of international organizations. Section 4 explores potential explanations for the rise of pass-on funds. Section 5 presents original empirical analysis drawing on the case of the World Bank. Section 6 discusses the consequences of pass-on funds for the functioning of the World Bank and probes the potential for generalization beyond the World Bank context. It also reassesses some prevailing arguments in the relevant theoretical literature. Section 7 concludes.

2 The rise of pass-on funds at the World Bank

Pass-on funds are a special type of trust fund involving a particular division of labor between at least two units of an international development organization. One unit hosts the fund, maintaining donor relations and allocating program funds to specific projects, and at least one other unit implements specific projects supported by the fund.

Pass-on funds are common in larger multilateral agencies such as the United Nations which comprise a bundle of functionally specialized entities. Under the pooled funding mechanism, administered by the Multi-Partner Trust Fund Office, the United Nations Development Program allocates pass-on fund resources to multiple entities that implement resources according to their own operating procedures.² The World Bank often uses the term “global funds” to characterize pass-on funds, because their sector programs often have global scope. However, “pass-on fund” is a more appropriate term as it emphasizes the institutional arrangement among distinct organizational units.³ This institutional arrangement involves the so-called “network anchors” as hosts and the regional units as implementers. The reverse arrangement in host-user relations is rare. As shown in Fig. 1, regional units primarily use trust funds received from other units, but also raise their own funds that cannot benefit other units. Conversely, network units mostly host funds intended for implementation in another unit, while also raising funds to support their own work program.⁴ These patterns indicate a division of labor by which network units host trust funds and pass them on to regional units, which motivates the pass-on funds terminology. The patterns also reflect the country-based assistance model of the World Bank.

Fig. 1

Host-user relations in World Bank trust funds. Notes: Figure shows the relative distribution of trust fund grants during FY 2002-13 aggregated over all regional units and all network units. Bar labels indicate total numbers of trust-funded grants. Source:World Bank (2013c)

At the World Bank, the growth of has increased rapidly. In 1999, there were virtually zero pass-on funds, by 2005 there were 40, and by 2010 the number peaked at 110 (Fig. 2). Furthermore, their significance in the overall trust fund business is considerable. Over fiscal years (FY) 2002–13 — the longest time span for which data are available — 37 % of all donor contributions went to global funds, which rely on pass-on funding mechanisms.⁵ An individual pass-on fund may support multiple activities, thus there are many more pass-on fund supported activities than there are pass-on funds. For example, in FY 2010, the Bank had 110 active pass-on funds, but all active pass-on funds together disbursed 2,071 individual grants (World Bank 2013d).

Fig. 2

The evolution of pass-on funds at the World Bank. Notes: Thick lines show the total number of active pass-on funds minus terminated funds in a fiscal year by unit type. Individual dots of corresponding color show unit-specific data. The figure excludes trust funds involving the International Finance Corporation (IFC). Source:World Bank (2013d)

Pass-on funds funds are also significant in terms of grant volumes. The total size of pass-on funds in FY 2004 was USD 750 million, which doubled to USD 1.5 billion in FY 2011. This figure is almost half the expenditure from IBRD/IDA trust funds over this period. The USD 1.5 billion of pass-on funds in FY 2011 was roughly 15 % of the total grant amount committed by the World Bank to its client countries that year (as the grant volume equaled USD 10 billion in FY 2011).

The rapid expansion of pass-on funds contrasts with the more gradual development of other types of funds at the World Bank. While pass-on funds are a novel phenomenon, trust funds in general are not. In fact, the World Bank and other multilateral organizations have long-standing experience with trust funds to support specific development initiatives. The World Bank established its first trust fund in 1960, when several donors sought to co-finance the Indus Basin Project in Pakistan.⁶ Since 1960, trust funds grew slowly and were primarily vehicles for the Bank to participate in partnership programs, for example the Consultative Group on International Agricultural Research (CGIAR) in the 1970s. In the mid-1980s, the World Bank started establishing a number of consultant trust funds with bilateral donors (World Bank 2005a: 1). In the mid-1990s, the World Bank created trust funds to address emerging development issues, for example post-conflict reconstruction, good governance, and environmental protection (Barakat, Rzeszut, and Martin 2012: 13).

In many of these trust funds, individual sector or regional units raised their own resources to support their operational work. They are called “own funds” in this paper. This is a broad term for a range of funds with different purposes. Own funds have steadily grown for decades and continue to be a significant element of grant funding at the World Bank and other multilateral organizations. Own funds allow individual units to expand their work program and hence will be established as the opportunity arises. A given unit has much flexibility in the design of own funds, which implies that they remain close to actual needs while having a lenient institutional structure (Reinsberg 2016).

In contrast, the benefits from pass-on funds may not be reaped without certain costs. Focusing on trust funds that support country operations, program effectiveness hinges on the type of funding source, notably, whether the country unit raises its own fund or whether it is supported by a pass-on fund. An independent evaluation reports that “[s]ingle-country trust funds are generally adapted to country situations and work well; this is less true for global funds that support multiple countries — many of which poorly integrate and coordinate their support at the country level” (IEG 2011: 25). From the perspective of World Bank management, pass-on funds are double-edged swords: They help the organization stay relevant on global issues despite its country-focused assistance model (IEG 2011: 9), but the funds may also undermine efforts in other priority areas such as increasing the relevance of World Bank operations for recipient countries. It may not be a coincidence that long after the World Bank decentralized its operations to field offices, the alignment of operations with country needs remains remarkably poor (IEG 2011: viii). Pass-on funds warrant closer examination partly because they can alter how multilateral agencies work. The significant size and sudden growth of these funds in the World Bank require explanation and provide the empirical basis of this paper.

In addition, pass-on funds raise interesting theoretical questions. Why do pass-on funds exist at all? Which factors explain the sudden explosion of pass-on funds at the millennium turn? The first question touches on the rationale for pass-on funds. While this question can be studied from a donor perspective, I employ an intra-agency perspective in order to better understand the incentives for the individual units involved. It appears to be puzzling that implementing units rely on pass-on funds rather than raising own funds to support their activities, given that units could better target the funds to their specific needs if the funds were housed in their own unit. Moreover, it is unknown what motivates host units to raise trust funds that they do not use to implement programs themselves. The case of pass-on funds hence seems to reflect an interesting case of functional differentiation at multilateral organizations and increased intra-organizational bargaining.

The second question concerns the timing of pass-on funds. Which factors came together at the millennium turn that explain the sudden takeoff of pass-on funds? It is clear that the usefulness of the funds alone does not explain the timing of their sudden growth; however, the factors underlying this new trend are yet unexplored. The goals of this paper are to establish the general case for pass-on funds, to trace the origins of pass-on funds, and to explore resulting bigger lessons for international organizations theory.

3 Theoretical perspectives

While my paper relates directly to the emerging literature on trust funds, I elaborate more broadly on debates in international relations, organizational theory, and the public choice of international bureaucracies for which the study of pass-on funds may bring about new insights.

Trust funds are a new mechanism of multilateral development finance. From an empirical perspective, the new multi-bi aid dataset provides opportunities to study trends in earmarked aid for a prolonged time period (Eichenauer and Reinsberg 2017). From a theoretical perspective, the rise of trust funds can be studied within a principal-agent framework. The principals are the member states, acting collectively through the governing board or individually as donors, while the agent represents the multilateral bureaucracy (e.g., Nielson and Tierney 2003; Hawkins et al. 2006; Johnson and Urpelainen 2014).

Trust fund research using the principal-agent framework is primarily from the donor perspective (e.g., Sridhar and Woods 2013; Eichenauer and Hug 2015; Graham 2015). For example, Sridhar and Woods (2013) coin the term “Trojan multilateralism” to refer to the risk of global funds undermining universal multilateralism, and conclude these funds exist because of the problem of agency slack. By earmarking their financial contributions, donors reduce the policy discretion of their agent and hence increase donors’ collective control. For example, donors influenced the World Health Organization by using the Global Fund to Fight Aids, Tuberculosis and Malaria. Others consider goal conflicts among the principals themselves to be the main driver for earmarked funding. Graham (2016) argues that earmarked contributions are the joint product of divergent preferences regarding policies and financial costs of multilateral activities. Moreover, Eichenauer and Hug (2015) study the choice of funding channel when special-purpose trust funds become available, taking into account the role of voting rules at multilateral organizations. Finally, Reinsberg et al. (2016) study donor participation decisions in World Bank trust funds and present evidence to suggest that individual donors trade off gains from burden- sharing in funds with more donors against gains in control in funds with fewer donors. The present paper provides a complementary perspective to these studies by exploring potential explanations for trust funds that explicitly focus on the multilateral agency perspective. This change of perspective provides some new insights, for example that trust funds rarely result from unfaithful agents and more often result from unfaithful principals.

A key debate in studies of international organization relates to the sources of autonomy of multilateral agencies. Recent literature makes significant steps to measure autonomy more precisely (e.g., Brown 2010; Conceição-Heldt 2013; Bauer and Ege 2016). For an agent to be autonomous with respect to the principal(s), it needs to have autonomous preferences as well as autonomous resources (e.g. Barnett and Finnemore 2004; Cortell and Peterson 2006; Caughey et al. 2009; Bauer and Ege 2016). It is an open debate whether or not trust funds increase the autonomy of multilateral agents. From a donor perspective, earmarked funding constrains the agent, especially when the same donors that provide earmarked funding simultaneously reduce their core funding (e.g., UN 2012; Sridhar and Woods 2013; Browne and Weiss 2014). From an agency perspective, however, earmarked funding multiplies the available choices, especially when agencies manage to broaden the donor base. In that way, agencies can exploit differences among the donors for their own benefit (e.g., Bryner 1987; Wilson 1989; Lyne et al. 2006). Unless donors establish controls costly to them — contract precision, appointment procedures, monitoring agents, “shadow bureaucracies” (e.g., Nielson and Tierney 2003; Dijkstra 2015; Parizek 2016) — agent autonomy grows with the resources under control. Therefore, it is widely assumed that bureaucracies attempt to maximize their budgets and grow their staff (e.g., Frey 1984; Vaubel 1986; Vaubel et al. 2007). Some researchers argue that agents want independence from their principals wherever possible (e.g., Reinalda and Verbeek 2004; Hawkins and Jacoby 2006; Johnson 2014).

In this debate, autonomy refers to autonomy of a single agent versus its principals. This ignores that such autonomy may not be the same for every unit inside a multilateral bureaucracy. In general, different units within the same organization may have different preferences (Barnett and Finnemore 1999: 724). Different units might also have different access to external resources. For both reasons, considering agencies as heterogeneous actors may be necessary in some contexts for the concept of agent autonomy to be sensible.

In addition, the struggle for autonomy may not be limited to the principal-agent relationship. In fact, this struggle also occurs within a collective agent. While it had long been argued in the domestic context that bureaucrats strive for power, prestige, security, and loyalty (Downs 1967; Fiorina and Noll 1979; Breton and Wintrobe 1982), the power struggles within international bureaucracies tend to be neglected. As an exception, Graham (2013) provides case evidence from the World Health Organization that headquarters in Geneva were unable to control local field offices, causing country programs in health systems to be poorly connected to the donors’ goals embraced at headquarters. No study of trust funds exists to date that uses these insights, leaving important gaps in our understanding of trust funds. Current explanations do not adequately account for the emergence of certain types of trust funds. In this paper, I provide an explanation for the rise of pass-on funds by studying their implications on the intra-organizational patterns of authority.

Recent literature particularly from organizational theory emphasizes the role of organizational structure (e.g., Gehring 2009; Trondal 2013; Bauer and Ege 2016). Organizational structures that provide for more local input may be crucial for project performance. Honig (2015) shows that such features have a strong impact precisely when local conditions are highly unpredictable, for example in fragile states. Trondal (2013) shows that organization by purpose engenders a logic of portfolio, while organization by process engenders a logic of hierarchy, and both behavioral logics have different implications on performance. However, demonstrating that organizational structure influences autonomy faces the challenge of endogeneity, which might exist because states design institutions rationally with a view to desired outcomes (e.g., Abbott and Snidal 1998; Koremenos et al. 2001). Nonetheless, organizational theory is useful as it can inform the debate on the determinants of institutional design. Budgetary pressure may lead to reorganization and specifically more centralization in the budget procedure, whether it was anticipated by the principal or not (Goetz and Patz 2016). More generally, there is a debate on the determinants of (de)centralization. For instance, agencies may centralize to avoid “siloization” and hence facilitate the flow of information across functionally specialized units (Trondal 2012: 435). Conversely, decentralization may reap the benefits of functional specialization. Considering the insights from organizational theory, I will argue that the institutional arrangement of pass-on funds follows a functional logic.

4 Potential explanations for pass-on funds

The puzzle of pass-on funds has two parts, one on the existence of pass-on funds and the other on their timing. For the static part of the puzzle, I draw on a functionalist argument. For the dynamic part, I take the principal-agent framework as a point of departure, discussing both donor-centered explanations and an agency-centered explanation.

Consider the question why pass-on funds exist at all. Certain units have comparative advantages in certain tasks and hence specialize in those; for example, sector units have expertise in global themes, and country units have unique country knowledge. These facts alone might explain why each type of unit raises its own funds in their relevant area of expertise, but they cannot explain the division of labor in pass-on funds. Organizational theory suggests that the process-driven allocation of responsibilities establishes hierarchy (Trondal 2013). Arguably, such hierarchy will only be acceptable in the presence of mutual benefits: Sector staff have better access to global funding than regional units and regional units have better capacities for implementation than sector units.⁷

In the timing of the rise of pass-on funds, two perspectives can be distinguished within the principal-agent framework. While the donor perspective relates pass-on funds to specific donor motivations, the agent perspective emphasizes the multilateral bureaucracy’s motivations. From the donor perspective, one potential explanation for pass-on funds relates to the collective desire to increase control over the bureaucratic agent. Theoretically, unfaithfulness of the agent is a risk of delegation (e.g., Niskanen 1973; Vaubel 1986; Dunleavy 1991), and agent control mechanisms abound (e.g., Nielson and Tierney 2003; Lake and McCubbins 2006; Dijkstra 2015; Parizek 2016). Empirically, the creation of the Global Fund (to Fight Aids, Tuberculosis, and Malaria) can be rationalized by the desire of the donors to control the World Health Organization (Sridhar and Woods 2013). By analogy, one could argue that a pass-on fund is a control mechanism that relies on the disciplining effect of competition for funds. This strategy suggests that principals could create one big thematic fund and let country offices compete with good projects rather than delegating authority directly to country offices. For this control mechanism to be effective, donors would need to replace direct delegation to country offices (which is also difficult to screen) with indirect delegation via pass-on funds. An observable implication of this mechanism would be a structural break in the number of own funds or even a reduction in the number of own funds hosted at the regional units. However, available evidence suggests the contrary. Own funds in the regional units have continued to grow, even after the emergence of pass-on funds (see Web Appendix B2).

Another donor-centered explanation relies on preference divergence within the collective principal. By using pass-on funds, a donor with diverging preferences can reinforce a specific sector in which it has a salient interest and which the hosting unit addresses through the trust fund. While the donor actually aims at gaining leverage in the intergovernmental policy discussions, a by-product of its strategy is the empowerment of a specific sector unit inside the agency. This mechanism has two observable implications. One would be a similar rise in the number of own funds in the relevant sector units that are being empowered. Another would be a positive relationship between measures of salient donor interest in a given sector and the number of pass-on funds in the same sector. The available evidence for these implications is limited though. The growth of own funds in the network units started in the mid-1990s and halted in FY 2003 (see Figure B1 in Web Appendix B2). Moreover, there appears to be no robust relationship between trust fund membership and chairmanship in the relevant working group at the Organization for Economic Cooperation and Development (Reinsberg et al. 2016).⁸

From a multilateral agency perspective, pass-on funds may be a result of “defunct delegation.” This possibility follows the recent turn in principal-agent studies to focus on the behavior of the principals. Defunct delegation can occur in cases of principal shirking, principal drift, and principal subversion — opportunistic behavior by individual members of the collective principal (Elsig 2011: 498). For example, principal shirking occurs when member states do not pay their dues. This is a primary concern for agencies because their survival depends on financial contributions (e.g., Pfeffer and Salancik 1978; Barnett and Coleman 2005; Vaubel 2006). The other two defects are equally intimidating because they may reduce the perceived legitimacy of the agent, which is required to ensure the long-term viability of the agency (e.g., Haas 1964; Suchman 1995; Weaver 2007). For these reasons, the agent will respond to these defects because they threaten its autonomy. It is well known that not only states shape the environment for multilateral agencies but also certain multilaterals sometimes define the environment for states (Brechin and Ness 2013; Johnson 2014; Johnson and Urpelainen 2014). To regain autonomy, agencies can diversify their resource base (Reinalda and Verbeek 2004: 232), build alliances with non-state actors (Hawkins and Jacoby 2006), and form cartels with some competitors (Vaubel 1986), while reducing resource flows to other direct competitors (Posner 2009).

Someone must act on behalf of the agency. If agencies are assumed to be unitary actors, who acts does not matter, but a high level of actorness is required. This makes unitary-actor explanations for the emergence of pass-on funds unlikely, as there would need to be a corresponding authoritative decision by someone who represents the agency. There is no evidence of such decision. A more plausible explanation can be formulated if multilateral agencies are understood as collective agents (e.g., Barnett and Finnemore 2004; Weaver 2008; Graham 2013). In that case, pass-on funds emerge as the product of an interaction between different units. Each unit faces incentives to perpetuate the implied division of labor — network units raising funds and managing the programs, and regional units implementing them. Supportive evidence would need to identify the changing circumstances that render these funds profitable for both units.

My above discussion suggests that agency-centered explanations for pass-on funds are a priori more convincing than donor-centered explanations, especially when considering pass-on funds as the result of individual optimizing decisions of different units rather than the planned response of the agency leadership. In summary, my favorite explanation has two parts. The static part uses the functionalist argument that pass-on funds are an efficient way to divide the labor based on a process logic. The dynamic part builds on the public-choice argument that individual units struggle for their autonomy, and argues that both units faced changes in their environment that rendered pass-on funds profitable to them.

In the following section, I provide empirical support for my theoretical explanation. Given that the rise of pass-on funds at the World Bank is a singular event, I need to rely on qualitative evidence. Specifically, I draw on semi-structured staff interviews at World Bank headquarters during summer 2013, with a few follow-up interviews in 2014.⁹ Complementary data include documents such as trust fund program documents, annual budget reports, and evaluation studies.

Web Appendix A includes a quantitative analysis of determinants underlying the use of pass-on funds, based on a formal bargaining model between a network unit and a regional unit. The results provide further statistical evidence suggesting that the struggle for autonomy of individual units (as proxied by their budgets) and external constraints facing the units are the factors driving the use of pass-on funds. The statistical analysis supports my claims regarding the incentives underpinning pass-on funds, but does not generate qualitatively new insights, and therefore appears in the Web Appendix.

5 Empirical analysis

The evidence presented in this paper draws on the case of the World Bank. Following an introduction to its institutional structure, the subsequent parts of the analysis address the research questions of this paper — the rationale for pass-on funds and factors driving their timing.

5.1 Organizational reforms at the World Bank

The World Bank is a multilateral agency that promotes development mainly through the International Bank for Reconstruction and Development (IBRD) and the International Development Association (IDA). Multilateral policies are broadly determined by the member states, which convene every three years in the Board of Governors, and on a regular basis in the Board of Executive Directors. The Board of Executive Directors approves the IBRD/IDA projects as prepared by Bank staff.

The World Bank bureaucracy includes a president (appointed by the member states) and over 10,000 staff in headquarters and country offices. It has a complex structure that changed a few times in recent history. A persistent feature of the World Bank has been the so-called “matrix structure,” existing since 1987. Since then no institutional reform — not even the latest one in 2014 — touched this constitutive feature (see Web Appendix B1 for details on organizational reforms at the World Bank).

The matrix structure governs the interactions of two functionally specialized types of units in each of the six regional Vice Presidential Units (VPUs) of the World Bank (Fig. 3). In each regional unit, headed by a regional Vice President, there are several Country Managing Units (CMUs), headed by country directors, and Sector Managing Units (SMUs), headed by sector directors. While CMUs are responsible for the overall country strategy, SMUs cover all technical aspects of project design for a group of countries.

Fig. 3

A simplified representation of the “matrix structure”. Notes: This simplified structure shows the interaction between one CMU and SMU (and higher hierarchical levels). Blank areas correspond to headquarters, and gray-shaded areas indicate the country level. Dotted lines reflect accountability relationships

A matrix structure has the benefit of functional differentiation, where CMUs develop expertise in convening services and managing portfolios, and SMUs develop expertise on technical aspects of individual projects. As a result, regional units can be expected to deliver quality projects also relevant for country needs. A disadvantage of the matrix structure is the risk that projects may not fully benefit from global knowledge at the sector level, because SMUs work in their respective country departments without sufficient access to sector-specific knowledge at the global level.

“Network anchors” are sector-wide knowledge hubs detached from individual regions, which collate sector-specific knowledge at the global level. They somewhat alleviate problems of “siloization.” The World Bank introduced network anchors in its FY 1997 reform (also known as the “Strategic Compact”). Since mid-2014, following the latest organizational reform, the network anchors no longer exist formally, and the sector-specific global work is carried out by hierarchically organized network units called “Global Practices.” As the new Global Practices have their own heads (rather than a Sector Board governing network affairs collectively), network units are more powerfully represented than regional units in the new structure with the goal of enhanced Bank-wide learning (Stoiljkovic and Hansen 2014).

None of the most recent institutional changes alters key organizational features established by the Strategic Compact. One such key feature is the coexistence of CMUs and SMUs in any given regional unit and the dual accountability of SMUs to regional and network heads. Moreover, there is continuity with respect to pass-on funds, which continue to be administered by network units at headquarter level while being routed through SMUs for implementation.

5.2 The rationale for pass-on funds

5.2.1 The demand side of pass-on funds

Pass-on funds allow regional units (CMUs and SMUs) to expand their activities. Pass-on funds are very flexible because they can provide program funding to support recipient-executed activities while also covering expenses for staff activities related to project preparation, appraisal, and supervision. In the former case, pass-on funds are similar to IBRD/IDA resources while involving a potentially higher degree of concessionality to the recipient country. Trust funds are always grants, IDA funding includes concessional loans and grants, and IBRD loans are non-concessional. By “sweetening the deal” for recipient countries, trust funds enable CMUs to boost their lending operations by giving countries the capacity to borrow and easing the terms of Bank assistance. For example, “[i]n Indonesia, which is reluctant to borrow IBRD funds for technical assistance, as much as 23 % of the trust fund portfolio is dedicated to such work [...]” (IEG 2011: 36). In addition, thematic funds often provide funding for global public goods and thereby provide country staff with increased access to government officials with whom they did not interact beforehand, hence increasing the potential for new projects. Indeed, pass-on fund grants are “[...] an incentive to trigger internal actions within countries and also provide the Bank with a way to engage with line ministries in middle-income countries” (IEG 2011: 35).¹⁰

Besides financing recipient-executed activities, pass-on funds are a substitute for the administrative budget. An important example is the Policy and Human Resources Development (PHRD) trust fund. Supported by Japan and housed by the Concessional Finance and Partnerships unit, the trust fund provides resources to develop lending projects and to enhance their quality. As stated in its annual report, “[p]roject success can often be related to the [...] solid project preparation, which would not have been possible without PHRD support” (PHRD 2012: 8). In one project discussed in the report, “the PHRD grant was instrumental in the preparation and approval [...] by the World Bank Board of Executive Directors” (PHRD 2012: 10). The quote demonstrates that an individual project preparation grant may be valuable, and pass-on funds are more likely to supply these grants than other trust funds (see Fig. 4). When pass-on funds finance analytical work that at first sight does not seem to be useful for regional units, they may still be beneficial because those funds are “[...] sometimes fungible when paying for staff designated to work on the implementation of Bank projects.” If they do so, “there are pressures to access trust funds for economic and sector work [...] in order to preserve Bank budget for project appraisal and supervision” (IEG 2011: 75).

Fig. 4

Purposes of pass-on funds and own funds in the regional units. Notes: Purposes of trust fund in the regional units by origin of funds ? pass-on funds versus own funds, based on the number of grants between FY 2003-12. Other purposes not shown include administrative funds, consultancies, debt service, financial intermediation, and other innovative financing. Source:World Bank (2013c)

The considerable benefits that pass-on funds accrue to regional units may still not be enough to explain their actual use. Could a regional unit not realize these benefits by raising its own funds? To be sure, by raising its own funds, a unit can often well align the fund with its own priorities. However, there are limits as to which impacts these trust funds can achieve. Own funds in the regional units often co-finance individual projects, often with the support of a single donor. Apart from a few single-recipient flagship funds such as the Afghanistan Reconstruction Trust Fund (ARTF), these regional funds tend to be hardly visible for the larger donor community and hence do not attract huge amounts of contributions. If a regional unit decides to raise its own fund, it also faces administrative costs due to the maintenance of donor relations and trust fund secretariat services, which may overburden the capacities especially of the smaller units. Finally, country directors often are unable to access those donor officials in the donor capitals who make significant aid budget decisions. Therefore, pass-on funds may be a better option under some circumstances. They often mobilize large amounts of resources for specific themes that may also be relevant in a given recipient country. Their global scope increases attention from a larger pool of donors, which implies larger fund sizes. Indeed, global funds on average are five times larger than country-specific funds, soliciting contributions of about USD 200 million per fund — this difference is strongly statistically significant (calculations based on World Bank 2013d). This provides the case for pass-on funds as an additional funding mechanism besides own funds. In addition, own funds often do not provide (enough) funding for purposes such as project preparation. Half of all grants from own funds co-finance country operations, while pass-on funds concentrate most of their grants on technical assistance, followed by co-financing and project preparation (Fig. 4).

However, pass-on funds are not costless to the regional users, especially to CMUs. The funds creep into the autonomy of regional units by altering the sector composition of their country activities. This drawback becomes particularly salient when the pass-on funds support sector activities that are not a priority for the CMU and its recipient government. The Independent Evaluation Group (IEG) states that “[...] while virtually all of the trust-funded technical assistance and economic and sector work deals with issues relevant to the country, the work is not always of the highest priority, would not have been done using the Bank’s budget, and does not always receive management attention and quality assurance [...]” (IEG 2011: 37). In fact, “[i]t is staff in those sectors that are crowded out of the CAS [Country Assistance Strategy] or who do not accept that theirs is not a sector expected to feature active Bank engagement in the country who then go looking for trust-fund funding (since this is not made available by the CMU). It is this sort of entrepreneurial activity that leaves counterparts complaining about studies not relevant to the country and studies that lie outside donor coordination mechanisms” (IEG 2011: 37). Another manager asked for an IEG study criticized the Carbon Funds — for causing “[...] a thousand flowers to bloom,” and that the funds have become “an operational mess, supporting things that we should not be doing” (IEG 2011: 36). During my own interviews, one official gave an example of a study intended to advance the knowledge base of a global sector fund, but it turned out that it did not have any concrete value to the client country.¹¹ The intrusion of pass-on funds into country priorities clearly is a significant challenge.¹²

A related concern refers to the potentially adverse quality implications of pass-on fund activities. Since the funds do not align well with regional units’ priorities, regional managers tend to neglect quality control. Several interviewees highlighted this problem. For example, one Operations Officer explained that there is “de-facto a problem of ownership [...]” and a “higher chance of disconnect and that certain people are not put in the loops.”¹³ A regional Trust Fund Coordinator claimed: “In our unit, all trust fund projects must be approved through the country director [...], including the funding applications for pass-on funds,” adding that “other regional units do not necessarily have such rules.”¹⁴ My evidence on lack of ownership of pass-on funds also bears out in the IEG evaluation. According to the evaluators, supervision quality remains low because the country unit has little ownership in these funds, arguing that “[...] the smaller funds, especially those managed by the networks, seem to be ‘off the radar screen’ of the country management unit once they are approved” (IEG 2011: 61).

A final problem yet unmentioned refers to transaction costs. The same Trust Fund Coordinator whose unit had strict senior approval rules argued that a task team leader may submit an application “but does not automatically get the funding,” suggesting that pass-on funds may create costs to the regions even when they do not support any operation. This echoes earlier work showing that trust funds increase internal fundraising costs for World Bank units (Reinsberg 2016).¹⁵

5.2.2 The supply side of pass-on funds

Network anchors benefit from pass-on funds for at least two reasons. The first is that these funds tend to be sizable and therefore generate significant fee income that a network can use to pursue its priority activities independently from country demand (World Bank 2007: 4). If network staff attract funds from donors, they obtain tangible rewards (such as pay rises and job promotions) from becoming more operationally relevant and psychic rewards from being operationally relevant and not just a anchor unit bureaucrat. The second reason is that thematic funds allow network staff to promote their specific thematic interests, even if a sector is not of greatest priority to country units. Pass-on funds therefore enable network units to tilt the portfolios of regional activities toward their own preferred sectors. In addition, they also involve co-benefits for network units themselves by supporting their analytical work at the global level.¹⁶

While network anchors are motivated to raise pass-on funds, they are also best placed to solicit large trust funds for global themes from bilateral donors. Donor officials who are interested in particular sectors do not reach out to the field but rather to network staff in Bank headquarters. For a donor, making a network anchor the sole point of contact for a global fund can (1) reduce its transactions costs, and incidentally the Banks — the minimum size requirement may even force it to provide one larger TF instead of several region- or country-specific trust funds; (2) use the network anchors expertise — the donor may trust it to use the funds efficiently more than it trusts anyone else to; (3) ensure cross-border costs and benefits are taken into account that CMUs might ignore.¹⁷

Network staff at the Bank can thus present themselves as the natural interlocutors for donor representatives in donor countries’ capitals. As one Trust Fund Coordinator from a large network unit explained: “There is a headquarter advantage in terms of capacities and preferences. We are based in D.C. [...] other VPUs have regional offices, so they often do not have the capacity and the incentive to negotiate hard with donors.”¹⁸ In their effort to raise new pass-on funds, networks are most likely to focus on bilateral donors with large global budgets, donors that share their sector interests, and larger donor countries which tend to have broad country lists.¹⁹

A glance at the contribution data for pass-on funds supports these claims. In terms of cumulative contributions over FY 2002-13, the four most important donors of pass-on funds were the United Kingdom, Norway, the Netherlands, and Spain. Engagement strategies vary considerably across these donors: The United Kingdom and the Netherlands also have vast country-specific contributions and hence are the primary financiers of regional units’ own funds. Their contribution shares to pass-on funds are below 20 %. In contrast, Norway and Spain are most engaged in pass-on funds as compared to the other donor countries. This squares well with the common wisdom that these two countries have salient sector interests and thereby prefer to ally with the network units at the World Bank. For example, Spain created a “MDG Achievement Fund” at the United Nations to help advance social development in 2007. Also, Norway has strong interests in themes like gender equality, good governance, and human rights.²⁰

In the same way as pass-on funds involve challenges for regional users, they may also be a risk for the network units which own them. One Program Officer of a well-established global program stated that “country directors do not look what is best for the program [...]”²¹ — implying that they are rather concerned with aligning trust funds and Country Assistance Strategies. Asked about the most likely implications of pass-on funds, a Controller argued that these funds could improve internal cooperation, although communication across departments could sometimes be burdensome.²²

In summary, the analysis thus far lends support to the functionalist argument that pass-on funds and their underlying division of labor are efficient as they reflect the respective comparative advantages of each type of unit. Network anchors host the funds as they are better able to access global donor resources, while regional units have implementing expertise at local levels. This division of tasks along the main processes creates a new hierarchy in which networks gain authority over regions. For the World Bank as a whole, the funds only seem beneficial when they align with country priorities and hence avoid problems of insufficient ownership on the part of regional users.

5.3 Organizational reform and the rise of pass-on funds

5.3.1 Organizational reform threatened the autonomy of sector units

Yet to be explored are the particular factors that drove the sudden explosion of pass-on funds around the millennium. In fact, the adoption of the Strategic Compact and the rise of pass-on funds are not coincidental. As argued below, the Strategic Compact can be seen as the single most important explanation for the rise of trust funds. The adoption of the reform followed a series of setbacks for the World Bank in its relations with the external environment since the end of the Cold War. In particular, a key concern was the technical quality of its project work. An internal evaluation of the matrix structure stated that “technical staff worked primarily in country department silos and the technical quality of [...] project work deteriorated” (World Bank 2012: 115). In addition, development projects were not always perceived as relevant for country needs. An influential report accused the Bank of a “pervasive lending culture” (Wapenhans 1992: 12). The Bank also was perceived as losing sight of the poor (Smyth 2011: 983) and as sluggish in responding to new demands from donors, but arguably these demands proliferated after the end of the Cold War (Weaver 2007: 500). Major donors lost its faith in the Bank and threatened to withhold contributions unless Bank management implemented organizational reforms (Thibodeau 1996).

To address these concerns, the World Bank under the leadership of president James Wolfensohn launched a three-year reform program dubbed the “Strategic Compact” in 1997. The reform expedited the devolution of country activities to the field. In addition, it aimed at reinvigorating the organization as a “knowledge bank” to better reflect the development needs in the globalization era. To achieve the goal of a knowledge bank and more specifically to facilitate Bank-wide learning, the reform established so-called “network anchors” and sector boards to link all SMUs across the Bank and hence make available global knowledge to country teams. In addition, the reform established dual reporting relationships and country budgets allocated by CMUs to different SMUs. After the restructuring, it was no longer possible for SMUs to deploy budget resources without approval by CMUs, which sought to reduce “supply-driven behavior” by SMUs, hence aligning World Bank operations more closely with country demand (World Bank 2001).

The World Bank itself recognized the significance of this specific feature of the reform. According to an internal report, “it is this feature, and not the creation of the networks [...], or decentralization,” that makes the reform “qualitatively different from those that came earlier” (World Bank 2012a: 116). The transfer of budget autonomy into the hands of CMUs posed a threat to the autonomy of SMUs. As one experienced official put it: “Why should CMUs buy expert knowledge from the SMUs?” — obviously, SMUs remain collectively relevant as providers of sector expertise, but not an individual SMU whose sector is no longer deemed relevant by a given CMU. Therefore, according to the official, while the historical division between CMUs and SMUs is reasonable, it translates into “budget fights.” The consequence of budget pressure on the SMUs was that SMUs developed additional resources through trust funds. But this bears the risk of “satellite operations,” activities not aligned with Country Assistance Strategies, without “the Country Director knowing about trust fund missions in the country.”²³

According to another official, sector staff “were desperately looking for their sense of being.” As a result, “a pathological trust fund culture emerged, where TTLs [task team leaders] chased even after ten thousand dollars and labeled this a great success for their cause.”²⁴ This echoes statements in the IEG evaluation report that individual staff members were shopping around for trust fund monies, oftentimes leading to studies that were not considered as relevant for country needs (IEG 2011: 37).

5.3.2 Network anchors start filling resource gaps for sector units

At the same time as SMUs lost their budget autonomy, their global pendants — the newly established “network anchors” — started raising large funds for thematic priorities. These funds provided SMUs with an opportunity to tap new resources independent of the demand from CMUs. As the funds were formally hosted by network anchors but passed on to SMUs for local projects, they are called “pass-on funds.” As a consequence of pass-on funds, the authority of network units (and the autonomy of SMUs vis-à-vis CMUs) inside the World Bank increased.

The networks could monetize on the increased donor attention toward social development since the Millennium Declaration, adopted in September 2000. Allying with sector specialists in donor capitals, network staff mobilized “huge buckets of money.”²⁵ This statement from a Trust Fund Coordinator can be further substantiated by a look at the bilateral aid allocations of the ten largest donors of trust funds. Without suggesting a one-to-one relationship between the bilateral aid allocations and trust fund contributions, it is still remarkable that bilateral funding for global purposes became more abundant since the mid-1990s. In the four years between 1997 and 2000 only, global-purpose aid more than doubled. Thereafter, funding volumes fluctuated around USD 8 billion. Key donors that expanded their global allocations around that time were Germany, the Netherlands, the United States, and also the European Commission.²⁶ Suggesting a growing awareness among donors for global issues, these figures hence illustrate that the network units seized a unique opportunity to build up their trust-funded chiefdoms for global themes. An internal report reaches a similar conclusion, arguing that the newly created network anchors served the growing demand for sector-specific global development expertise, though this was not at all foreseen at the time the Strategic Compact was conceived (World Bank 2012a: 119).

5.3.3 Organizational reform created demand for pass-on funds

A number of features in the Strategic Compact also led to incentives for CMUs to use pass-on funds to support their operations. In fact, CMUs became ever less resistant against pass-on funds because nearly all features of the reform increased the pressure on their administrative budgets.

Administrative budgets are indispensable for delivering development projects. In particular, the administrative budgets covers Bank expenditure related to project preparation (prior to approval by the Board), and project supervision (once the projects are activated). Three features of the Strategic Compact squeezed administrative budgets: increasing safeguard requirements, the evolution of country portfolios toward high-cost environments, and the increasing devolution of country activities. I discuss each of them in turn.

First, since the mid-1990s, projects involved growing exigencies on safeguards. As one project manager explained for an IEG review: “I have plenty of examples of technical specialists that I have not been able to bring on mission [...] because there is no money in my supervision budget after hiring fiduciary, procurement, and safeguard specialists [...],” while other respondents confirmed that on supervision missions “compliance specialists” tend to crowd out sector specialists (IEG 2014: 49). The reason for these elevated requirements was a consequence of a series of alleged failures since the late 1980s. In 1991, the Bank had to revamp its environmental safeguards following United States pressure and protests from environmentalists in the realm of the Narmada dam project in India (Weaver 2008: 23). Later, the Bank also revised its social-economic safeguards, responding to criticism on the alleged failure of “structural adjustment policies” (Danaher 1994; Weaver 2008; Jolly 2014).

Second, regional units also faced higher costs as they engaged in more difficult country contexts, notably post-conflict states (World Bank 2005a: 35), which need a lot of technical assistance without offering the immediate prospect of taking up a loan from the Bank. After the end of the Cold War and the related shift in donor motives in aid allocation, the World Bank also committed to direct IDA funding toward poorer clients and to Africa, which “resulted in faster growth of expenditures” (World Bank 2005b: 38).

Finally, following accusations of a pervasive “lending culture” (Wapenhans 1992: 12), the World Bank reorganized its operations and devolved country operations to the field. While in FY 1996, not a single country director was located in the field, already three quarters of the country directors were dislocated to the field in FY 2005 (World Bank 2005b: 34). The built-up of local infrastructure was costly, which reduced the administrative budget being available for other purposes. In the spirit of the Paris Declaration, the Bank also sought to increase country ownership through “greater participation in formulating country development assistance strategies,” which likewise increased costs for the regional units to be covered from their administrative budgets (World Bank 2005b: 33).

In light of these challenges, regional units were less reluctant than ever before to turn to pass-on funds in the hope for additional resources. Raising their own trust funds at the field level was no alternative because such funds primarily co-finance program activities and thus cannot be used for technical assistance and project preparation.

In summary, my case study has established the rationale for pass-on funds in general and related their sudden rise to the organizational reforms of the Strategic Compact. As argued more generally in an IEG report, “[a]ccording to the vast majority of those interviewed, budget constraints are [...] perhaps the most important reason why staff pursue trust funds. They referred to ‘the budget squeeze,’ the work program ‘increasing tremendously while the Bank budget has remained flat,’ and the risk that their unit ‘would be out of business without trust funds”’ (IEG 2011: 75). Organizational reform and budget pressures are not mutually exclusive explanations. The reform can be seen as the trigger for the sudden emergence of pass-on funds, and the effect of the reform worked mainly through its budgetary implications for the two units involved. For the SMUs, the reform took away their independent budgets and hence made them seek out new funding opportunities, notably the individual grants from pass-on funds. This partially restored their sense of being. CMUs were ever less reluctant to accept the intrusion of pass-on funds into their programs While the regional units were able to expand their work program, because the reform entailed features that squeezed their own administrative budgets and hence pass-on funds promised to ease this constraint.

6 Theoretical implications

Given the novelty of pass-on funds and their rapid growth over the past decade, the primary motivation for this paper is empirical. However, pass-on funds raise interesting theoretical questions regarding the functional differentiation of international bureaucracies and the kinds of incentives facing different bureaucratic actors. Following previous studies on trust funds, my subsequent discussion focuses on potential lessons within the principal-agent framework.

With respect to the emergent literature on trust funds, my analysis provides a complementary perspective to most existing studies that focus on the donor perspective (e.g., Sridhar and Woods 2013; Graham 2015; Eichenauer and Reinsberg 2017). An exclusive donor perspective may not be fruitful when trying to explain intra-organizational patterns of authority as implied by pass-on funds. In fact, there is no clear evidence that pass-on funds directly relate to interests among the principals, neither due to a collective desire to gain control over the agent nor due to preference divergence among the principals.

While the focus on the agent holds much promise, it is further necessary to disaggregate the agent. To my knowledge, only one study in the principal-agent framework examines the implications of a fragmented agent on its overall faithfulness (Graham 2013). Moreover, autonomy is often thought to refer to the principal-agent relationship. However, struggles for autonomy exist within collective agents. My case study of pass-on funds shows that different units must cater to different needs — CMUs take the local needs into account, while network units think globally and promote specific themes without necessarily appreciating local circumstances. For an actor to be autonomous, having an autonomous will is not enough — autonomous resources are required to pursue autonomous preferences (Cortell and Peterson 2006; Caughey et al. 2009; Bauer and Ege 2016). As pass-on funds give budget authority to network units and hence potentially reduce the policy discretion of CMUs, the decision for CMUs whether to accept pass-on funding is not unconditional. The tradeoffs of pass-on funds only become clear when considering autonomy between individual units inside the multilateral bureaucracy rather than autonomy between the entire bureaucracy and its member states.

My case analysis also echoes recent arguments that unfaithful principals rather than unfaithful agents may be responsible for changes in organizational behavior (Elsig 2011). The case study has offered some evidence of principal shirking: The threat of some powerful donors to withhold core contributions was a key motivation for World Bank president James Wolfensohn to launch the Strategic Compact (Weaver 2008: 22) that triggered the observed dynamic of pass-on funds through its implications on administrative budgets. Moreover, there is evidence of principal drift: The growing demands on the World Bank were not matched with corresponding increases in resources (Linn 1998), leading to growing frustration by donors about lack of progress (Smyth 2011: 983); this likely expedited the organizational reform. Finally, there is evidence of principal subversion: Some donors openly challenged the collective consensus and pushed the Bank into new areas, as evidenced by the Bank-Netherlands Partnership Program and the activities of the Nordic countries in the area of human rights (see also, Reinsberg 2016; Reinsberg et al. 2015, 2016).

7 Conclusion

Over the past two decades, earmarked funding to international development organizations has grown massively. The financial mechanism through which external donors contribute these earmarked resources are so-called trust funds. The World Bank as possibly the most important trustee organization has a long-standing tradition in the management of trust funds that dates back to the 1960s. Until the mid-1990s, the prevailing type of trust funds were so-called “own funds,” by which individual units in the multilateral bureaucracy raised trust funds to support their own work program.

Since the turn of the millennium, however, a new type of trust funds has emerged. In so-called “pass-on funds,” one unit hosts the trust fund while another unit inside the organization implements the individual activities under the trust fund. In the case of the World Bank, this division of labor has taken a characteristic shape: The so-called “network anchors,” which exist only since an organizational reform in FY 1997, host large funds for global themes, for example fragile states, gender equality, and climate change, and pass them on for use in the “regional units” and their individual country offices.

Theoretically, the rise of pass-on funds reflects an interesting setting. The first question that I address in this paper is why these funds exist at all. Following organizational theory, my paper shows that this process-driven specialization reflects the respective comparative advantages of network units to access global funding sources and the regional units to implement projects on the ground. This is less trivial than it looks at first sight given that pass-on funds also establish a new hierarchy that puts regions at a disadvantage. Unlike “own funds,” which match the priority issues of a unit, “pass-on funds” may lead to misalignment between global priorities (as promoted by the network anchor) and country-specific needs (as promoted by the regional unit and particularly the country office) (see also, IEG 2011: viii). As compared to own sources of funding, the regional unit hence faces a tradeoff between obtaining additional resources and staying relevant to country needs when deciding whether or not to accept pass-on funding.

A second question pertains to the timing of pass-on funds. Using evidence from semi-structured interviews at the World Bank and complementary document sources, I identify the Strategic Compact — a three-year organizational reform that reorganized budget authority between the relevant units — as the key driver of pass-on funds. On the one hand, the network units had a particular incentive to do so since the reform moved budget authority into the hands of country managers and thereby created pressures to regain budget autonomy among sector staff. At the same time, bilateral donors increased their funding for global themes, which was a unique opportunity for the network units to start building their own chiefdoms supported through these donor monies. On the other hand, regional units became more desperate for the project preparation budgets provided by pass-on funds as they saw their own administrative budgets decline as a result of costly devolution, a higher share of difficult countries in their portfolios, and increasing requirements on safeguards.

These findings can claim generalizability beyond the period under investigation and beyond the World Bank context with some minor adaptations. Since 1997, the bread-and-butter business of the World Bank — working with recipient countries and managing projects — operates under a matrix structure that intertwines country units and sector units and that gives budget authority to the country directors (World Bank 2012a: 111). In 2014, the World Bank underwent an organization reform that established Global Practices in an attempt to break the institutional primacy of the regions and encourage global learning within each sector. However, the reform did not alter the above matrix structure within each regional unit. Country directors continue to hold de-facto budget autonomy as their specific demands determine which sectors feature in country assistance strategies. The key feature that was created with the introduction of the 1997 reform therefore persists even under the revised structure. My own interviews suggest that staff do not perceive to be affected with respect to their incentives vis-à-vis trust funds. Therefore, the issue of pass-on funds remains acute, which implies a continued need to balance sector interests and country needs at the World Bank.

Beyond the World Bank case, my analysis may apply to a broader set of agencies. Though this paper is about the World Bank, its analysis is relevant beyond this specific case because the pass-on fund arrangement is common in all major functionally differentiated international organizations. Such large organizations consists of headquarters and field offices, each having different preferences (despite shared goals) and hence there is a struggle for autonomy and a need for bargaining. For example, one official from the International Labor Organization (ILO) privately said: “Rainmakers in the various multilateral agencies actively use the funds [...] to position themselves and their departments and to build their careers.” There hence seems to be the risk that trust funds become a strategic device by which individual units seek to enhance their position within a multilateral agency; actual relevance for recipient-country needs, project quality, development impact, and cost effectiveness may get lost on the way. Future research may look into these issues, specifically also in a United Nations context where pass-on funds formally exist since the establishment of the Multi-Partner Trust Fund Office in 2003.²⁷

My study also has potential applicability even beyond intra-organizational interactions. The specific situation in which one agent hosts funds and another agent requests support from the funds occurs a lot in the multilateral development system, at least since the rise of so-called “multi-actor funds” (Heimans 2004), for example the Global Environment Facility, the GAVI Alliance, and the Global Fund to Fight Aids, Malaria and Tuberculosis. These multilateral institutions raise huge amounts of donor monies while not having own implementation structures. To implement their programs, these institutions rely on the established international development organizations, such as the World Bank and other bureaucracies. In these settings, my analysis can shed light on the conditions under which a given global fund chooses a specific implementing organization. Prominent research has already been done on the case of the Global Environment Facility (e.g., Bayer et al. 2014), and this paper further contributes to this important strand of research. In an increasingly complex world plagued by many global problems, the division of labor among different multilateral organizations will further increase and hence these settings of inter-agency bargaining deserve closer attention.

From a theoretical perspective, the findings of this paper have immediate implications for the emerging literature on earmarked funding at international development organizations. Related studies forward donor-centric explanations for the emergence of earmarked contributions (e.g., Graham 2015; Reinsberg et al. 2015). In these studies, disagreement among the donors over the substantive activities of international organizations holds a key to explain earmarking patterns (e.g., Eichenauer and Hug 2015; Graham 2016; Reinsberg et al. 2016).

While these studies have advanced the field in significant ways, this paper provides a complementary perspective by analyzing the incentives for multilateral agencies to pursue extra-budgetary funding through trust funds. The literature thus far has explored these incentives only for specific issue areas, for example Carbon Finance at the World Bank (Michaelowa and Michaelowa 2011). Most literature often does not question the unitary-actor assumption on the multilateral bureaucracy. This overlooks that different units may have different preferences and that they need independent resources to pursue these preferences. This echoes with public choice arguments that bureaucrats (here aggregated into cohesive units) have an incentive to grow their budgets in order to increase their autonomy. As an important caveat, however, this autonomy is defined with respect to competing units, not with respect to the principal, as usual within principal-agent studies.

In a broader view, the paper reinforces arguments about the active role of bureaucrats in institutional design (Johnson 2014), given that the creation of trust funds can be seen as such an instance of institutional design led by bureaucratic entrepreneurs. My work shows that bureaucratic influence on institutional design may itself be the result of a strategic interaction between organizational units. Hence, the set of scope conditions under which we see bureaucrats to enter the design stage may need to be extended to also include factors such as resource competition, thematic alignment, and inter-unit trust, in addition to factors that focus more on the interests of member states.

Within studies of bureaucratic politics and intra-organizational bargaining, more research could be done to integrate the donors as a strategic actor, without sacrificing the heterogeneous actor perspective on multilateral agencies. More specifically for the case of the World Bank, an open issue refers to the rise of own funds as evident from the descriptive statistics and how it relates to the rise of pass-on funds. This issue was deliberately left aside in this paper but it would be worthwhile to look into this at greater detail.