“The silent revolution:” How the staff exercise informal governance over IMF lending

Abstract

This paper examines how the staff exercise informal governance over lending decisions of the International Monetary Fund (IMF or Fund). The essential component of designing any IMF program, assessing the extent to which a borrowing country is likely to fulfill its policy commitments, is based partly on informal staff judgments subject to informal incentives and normative orientations not dictated by formal rules and procedures. Moreover, when country officials are unable to commit to policy goals of the IMF, the IMF staff may bypass the formal channel of policy dialogue through informal contacts and negotiations with more like-minded actors outside the policymaking process. Exercising informal governance in these ways, the staff are motived by informal career advancement incentives and normative orientations associated with the organization’s culture to provide favorable treatment to borrowers composed of policy teams sympathetic toward their policy goals. The presence of these sympathetic interlocutors provides the staff both with greater confidence a lending program will achieve success and an opportunity to support officials who share their policy beliefs. I assess these arguments using a new dataset that proxies shared policy beliefs based on the professional characteristics of IMF staff and developing country officials. The evidence supports these arguments: larger loan commitments are extended to countries where government officials and the Fund staff share similar professional training. The analysis implies informal governance operates in IOs not just via state influence but also through the evolving makeup, incentive structure, and normative orientations of their staffs.

1 Introduction

What explains the lending decisions of the International Monetary Fund? The International Monetary Fund (hereafter IMF or Fund) is often assumed to be a technocracy, with formal rules and procedures motivating lending decisions on the basis of the severity of the economic problems facing borrowing countries (Drazen 2002; Martin 2006:142). Increasingly, however, a number of scholars have argued the IMF is also a political organization, with powerful countries, including but not limited to the United States, intervening in lending decisions through informal processes to further their geopolitical and financial interests (Broz and Hawes 2006; Copelovitch 2010; Oatley and Yackee 2004; Stone 2002, 2008, 2011). Yet IMF lending decisions are not shaped by these influences alone.

This paper focuses on how these influences co-exist with staff informal governance over IMF lending decisions that manifests itself in several ways. While the staff prepare IMF programs within a formal delegation process, in practice the essential component of any program; assessing the extent to which a borrowing country is likely to fulfill its policy commitments, is based partly on informal staff judgments subject to informal influences not dictated by formal criteria. These influences in the form of informal incentives and normative orientations shape staff judgements and lead them to provide favorable treatment to borrowers composed on policy teams sympathetic toward their policy goals. The staff also may pursue actions that go outside formal organizational rules and procedures. When country officials are unable to commit to policy goals of the IMF, the IMF staff may bypass the formal channel of policy dialogue through informal contacts and negotiations with more like-minded actors outside the policymaking process.

Informal governance can arise within the IMF whenever informal practices outside formal rules and procedures help determine outcomes. For Stone (2011), informal governance emerges from the interaction of three sources of power: structural power, representing the outside options available to the leading state, formal rules and procedures, which set the policy of the organization and create the parameters through which informal influence is exercised, and informal influence, which provides the leading state with the means to shape the essential features of organizational outcomes when its interests are exceptionally intense. I offer a view of informal governance that shares these features, but also broadens our understanding of it with a focus on the staff and the ideational dimension of structural power.

Like Stone, I see staff informal governance as arising from informal influence within a set of formal rules and procedures. The formal delegation process creates the parameters within which informal staff influence is exercised in much the same way formal voting rights shape the way in which informal state influence operates. In neither case are formal rules and arrangements discarded fully. Instead, the staff and governments work within and around these formal rules and arrangements to shape organizational behavior. This is does not mean that everything constitutes informal governance and that formal governance never happens. On the contrary, it suggests the need to consider governance within IOs as often representing a hybrid of both forms rather than representing an either-or dichotomy (Stone 2011). Put differently, informal and formal governance often operate synergistically.

In Stone’s account, a crucial aspect of informal governance comes from the structural power associated with the outside options available to the leading member state and the externalities this generates for other member states. While the account here recognizes and, indeed, provides evidence consistent with such statist influence, it also points to the ideational dimension of staff interactions as a crucial but thus far neglected aspect of informal governance. This ideational dimension represents, as Strange (1988) argues, a form of structural power that comes from the development and deployment of authoritative modes of interpreting the world. Where the IMF staff possess some autonomy from member states, this means that their policy beliefs partly shape the way in which the organization is informally governed (Barnett and Finnemore 2004; Martin 2006). These policy beliefs may be consistent with the interests of leading member states but this does not mean they are attributable solely to them. As Woods (2006:56) observes, “[The] set of ideas [shared by the Fund staff] is not a direct reflection of the interests of the most powerful members of the organization, even though powerful members get to influence it.”

This line of argument suggests the definition of informal governance needs to be expanded to accommodate both this ideational dimension of structural power as well as informal staff influence, even when such influence emerges within the parameters of a formal delegation process. This broader definition is consistent with much recent work on governance in the European Union, which reveals how informal practices and policy beliefs – including policy networks, cultural orientations, and unwritten rules and routines of the type featured in this account – shape political outcomes even in the context of formal rules (Christiansen and Piattoni 2003; Christiansen and Neuhold 2012). In developing this argument, this paper shows that staff informal governance can explain a surprising amount of the variation in the way the IMF treats borrowers.

The argument here combines rationalist and constructivist insights by emphasizing informal career advancement incentives as well as normative orientations associated with the Fund’s organizational culture. These factors shape the way the staff manage the complexity they face when negotiating a lending program and judging the extent to which a borrower will fulfill its policy commitments. I argue these informal incentives and orientations motivate the staff to provide favorable treatment to certain types of borrowers; in particular, those composed of policy teams with sympathetic interlocutors.

Informal career advancement incentives motivate the staff to negotiate large programs that are likely to be successful. The ability of the staff to achieve success is not simply a matter of technocratic considerations; IMF staff also must deal with complex political dynamics that can cause a program to fail. The Fund’s lengthy and complicated lending experience has led the staff to conclude that success often depends on their ability to find and work with sympathetic domestic interlocutors who embrace the organization’s policy goals (Woods 2006). Rather than being simply technocratic, these policy goals also reflect the normative orientations found in the IMF’s organizational culture, which is heavily influenced by common staff professional training from Anglo-American economic departments (Barnett and Finnemore 2004; Chwieroth 2010a). This important socializing experience has helped to instill the Fund staff with a particular way of forming policy judgments. As a result, the Fund staff tend to seek out interlocutors whose academic credentials and professional training have instilled similar beliefs. The presence of sympathetic interlocutors provides the staff with greater confidence in achieving success and an opportunity to support policy teams that share their policy beliefs.

Other things being equal, one would expect the Fund to reward such interlocutors with favorable financial assistance. I assess this argument using a new dataset that captures the professional training characteristics of over 300 IMF staff members and over 1,000 officials from 44 developing countries. The evidence strongly suggests that sympathetic interlocutors matter: larger loans are extended to countries where government officials and the Fund staff share similar professional training.

This paper is not the first to suggest the importance of sympathetic interlocutors for the IMF. Indeed, James Boughton (2001), the Fund’s official historian, suggests that a “silent revolution” in developing countries, in which individuals with similar professional characteristics as the Fund staff emerged in top policymaking positions, helped to underpin successful IMF policy reform efforts in the 1980s. Yet much of the evidence tends to be anecdotal and impressionistic at best, unsupported by systematic and cross-national analysis. This paper seeks to rectify this shortcoming by assessing how the configuration of the Fund’s domestic interlocutors can informally influence the size of the loan it extends to borrowers.

The rest of the paper proceeds as follows. The first section outlines arguments. The second section provides the empirical analysis. The final section explores implications.

2 The Argument

The existing literature provides some important insights into IMF lending. The IMF is a technocracy of sorts in that economic models and formal organizational rules and procedures shape the organization’s basic approach to designing programs. However, this view of the IMF misses out on critical political aspects of its behavior. Those that see the IMF as a political organization shine important light on informal state influence but miss out on critical informal influences motivating staff behavior. To better understand how the IMF operates, we therefore need to investigate how the staff navigate the incentives they face as members of an organization and how they think about the world.

2.1 The IMF as an Autonomous Agent

The IMF staff, which grew from about 500 in 1962 to 2,610 in 2012 (de Vries 1985:1010; IMF 2012), have a great deal of authority over the design of IMF programs.¹ In designing and negotiating programs, the staff have a significant agenda-setting capacity that developed and solidified in the 1950s and early 1960s. Although country representatives on the IMF Executive Board are responsible for approving all programs, these Board directors consider only those programs that the staff have designed and presented to them. Should the staff anticipate a negative reaction from the Board, it can refuse to submit it for approval, providing it with significant gate-keeping power. Moreover, since the Board has higher turnover than the IMF staff, and thus less institutional expertise, this increases the likelihood that members will defer to staff judgments (Martin 2006:145–147).

To be sure, the staff recognize the concerns and preferences of the Board directors, and some powerful states are, in selected cases, capable of advancing their interests through informal contacts with staff throughout the loan negotiation process (Stone 2008, 2011). Indeed, some directors and government officials may, on occasion, become informally involved in drafting staff proposals, particularly when there is a program for a country where clear economic and geopolitical interests are at stake. But most programs are the product of the staff, particularly when member state interests diverge, when powerful countries oppose the status quo, and, importantly, when powerful countries believe staff policy goals are aligned with their long-term preferences (Martin 2006).

Because directors and government officials do not see confidential documents that are critical to loan negotiations, such as the mission briefing papers that determine the parameters of the staff’s negotiating position and the back-to-the-office reports that survey the progress of negotiations, they cannot easily influence the content of staff prescriptions before they take place. Therefore, in most instances, member states can do little more than direct the staff to encourage, or not, a particular policy in the future, but they can do little to alter the content of programs under consideration. As a result, even though directors are formally empowered to veto a particular program, in practice they have rarely done so, generally confining their interventions to minor changes to staff proposals (Southard 1979:7). The Board’s limited ownership over the positions the staff take in loan negotiations has historically been and continues to be an issue of concern for many directors, especially those representing constituencies that lack the analytical and human resources to monitor staff proposals and to advance alternatives (IMF 1999:13, 34; Martin 2006:7).

2.2 Formal Rules and Informal Staff Influence and Motivations

A formalized process, as outlined in the IMF Articles of Agreement and organizational procedures, governs access to IMF resources (Mussa and Savastano 1999). In principle each member state has the right to participate in a program when it faces balance of payments problems. The request for financial assistance comes from the member state and then enters a bureaucratic process governed by formal organizational procedures.

The staff from regionally organized area departments has primary responsibility for design of programs. When a request for financial assistance is made, the area department staff draw up a blueprint that contains a preliminary assessment of the central elements of the adjustment program and the size of the IMF loan. A mission briefing paper summarizing the blueprint is then prepared and circulated for comments to other departments. After incorporating comments from other departments and receiving approval from IMF management (the Managing Director and Deputy Managing Directors), the staff mission is dispatched to the member capital to negotiate a program. The outcome of these negotiations is a letter of intent that specifies the country’s proposed adjustment program.

Back at IMF headquarters, the mission team prepares a staff report that includes an account of discussions with country officials, details of the agreed policy adjustments, and the size of the loan. The report and letter of intent are then circulated to other departments for review, who check that the proposed loan is consistent with IMF procedures and offer their views, which are often critical, about its risks. A revised draft is then submitted to management for clearance, who then takes it to the Board for approval. Although management has the final decision on the size and design of the loan, it generally makes no changes to the features agreed by the area department mission (Mussa and Savastano 1999:12).

The size of any loan proposed by the staff is shaped by country-specific quotas that specify the amount of access to IMF resources a country is permitted under ordinary circumstances. Formal rules ordinarily limit cumulative access to IMF resources to 300 % of a country’s quota, but such limits can be waived during exceptional circumstances, as they were during the Asian crisis in 1997 and the recent global financial crisis.² According to formal organizational procedures, the staff determine the size of loan based on calculations of the financing gap in the balance of payments. Yet these calculations, which are essential to the success of the program and play a central part in Board discussions, are based on a number of uncertain projections about policy adjustments in the borrowing country, resource commitments from other official sources, and the reaction of private financial markets. The response of private financial markets is a crucial element of the success of IMF programs, and this response depends in large part on the credibility of the country’s intended policy adjustments (Stone 2002; Copelovitch 2010).

These projections, though developed within a formalized process, are not dictated by formal criteria alone. Rather, they are partly informal in origin and are subject to informal motivations. As Randall Stone (2011:137, emphasis added) suggests, “in essence, these [staff] projections are judgments about what is politically feasible.” Without a formal set of criteria to make these judgments, the staff are provided some discretion to set policy informally based on their assessment of the extent to which a borrower will likely fulfill its policy commitments.

If staff judgments do not derive from formal criteria alone, how then are they formed? Public choice scholarship suggests the IMF staff face bureaucratic incentives to engage in rent-seeking behavior to maximize their power, autonomy, and budgets (Vaubel 1996; Dreher and Vaubel 2004). From this perspective, subject to a financial constraint, these bureaucratic incentives should induce staff judgments in favor of proposing larger loans regardless of the country’s situation.

However, others scholars, including some offering a broader public choice analysis of the IMF, suggest the staff also face career advancement incentives to be more selective in forming judgments (Willett 2002; Woods 2006). Promotions within the IMF are linked to a number of different factors, including years of service and quality of technical skills (Momani 2005). Although there is no formal criterion stipulating as such, there is also an implicit but widespread view within the organization that advancement within the IMF depends on negotiating large and successful programs.³ Program failure, especially where the Fund makes a large financial commitment, can not only damage the career prospects and professional integrity of the staff who designed it, but also the IMF’s reputation and the legitimacy of the principles that have been applied in crafting it. The staff are therefore not necessarily interested in large programs per se, but rather large programs that are likely to be assessed as successful.

Because program success often depends on the ability of the staff to find and work with government officials who embrace their policy goals, informal career advancement incentives lead the Fund staff to prefer to work with sympathetic interlocutors in whom they place greater confidence in meeting their policy commitments. As Miles Kahler (1992:125) observes, “Consensual knowledge, agreement on the underlying features of the national economy and on the policy prescriptions best suited for adjustment, may explain the likelihood of cooperative outcomes between national governments and IFIs [international financial institutions].”⁴

How does the staff select their preferred interlocutors? Here organizational culture, which can be broadly defined as the set of collectively shared beliefs that shape how the staff interpret and understand their environment, select and process information, and arrive at particular decisions, plays an important role. A critical determinant of the IMF’s culture, which historically rests on a theoretical core stressing market efficiency and rationality, is the professional training profile of the staff, which is heavily skewed toward Anglo-American economics departments (Babb 2003; Barnett and Finnemore 2004; Chwieroth 2010a, b; Nelson 2009). Although it would be an overstatement to suggest that economics training leads the Fund staff to share one distinct set of beliefs, this common professional training does help instill a common set of general assumptions about how things are to be done within the Fund that are more or less shared by all staff. The staff draw on this training—which instills both shared technical knowledge based on positive models of economic behavior and normative conceptualizations about appropriate policy—to diagnose economic problems and to form policy judgments.⁵ As Paul DiMaggio and Walter Powell (1983:153) observe, “[Those] drawn from the same universities and filtered on a common set of attributes…will tend to view problems in a similar fashion, see the same policies, procedures and structures as normatively sanctioned and legitimated, and approach decisions in much the same way.”

Since this important socializing experience helps create shared policy beliefs, I argue the IMF staff seek out interlocutors in borrowing countries with similar academic credentials and professional training. Similar professional characteristics help to create a small, but significant, cadre of government officials who are inclined to share the IMF’s policy goals and therefore a stronger commitment to the policies it promotes. Similar professional characteristics thus aid in producing the shared policy beliefs that Kahler finds so critical to the success of IMF programs.

Several IMF officials and close observers of the Fund echo these arguments. Certainly, Boughton’s discussion of the “silent revolution” attests to the importance of such factors. Similarly, Ngaire Woods (2006:72), in her study of the IMF and the World Bank, concludes that success often depends on the presence of sympathetic interlocutors: “Where the Fund and Bank staff share technical expertise, methodology, and an orthodox economist’s understandings of problems and solutions with officials in a borrowing country, their capacity to transmit (or reinforce) ideas is heightened.”

Some observers go even further, claiming that the presence of sympathetic interlocutors constitutes a necessary condition for IMF success. Kahler (1992:127) observes that, “close alignment between a cadre of national economic technocrats and the IFIs [international financial institutions] seems to have been a prerequisite for agreement [on reform].” Stephan Haggard (1985:184, 186) similarly concludes:

In the absence of a countervailing ‘stabilizing cadre’ capable of articulating the long-term political, as well as economic, benefits of adjustment, programs are likely to fail. The core of this cadre is a cohesive group of sympathetic economic technocrats forming the domestic half of a transnational coalition with the Fund. When these groups are absent or politically marginal, commitment falters…These networks, perhaps more than resources per se, are the political bases for the power and influence of these organizations [IMF and World Bank]. The existence of an IMF-sympathetic ‘stabilizing cadre’ within the state appears to be a prerequisite for program success.

From their lengthy and complicated lending experience Fund officials have come to recognize both the importance of sympathetic interlocutors as well as the IMF’s capacity to facilitate their emergence and empowerment. According to David Finch (1989), a former long-serving senior staff member, as early as the 1950s the IMF staff realized the importance of working with sympathetic interlocutors to achieve success. Similarly, historian Harold James (1996:133) observes that, “[O]ne of the major functions of the IMF has been concerned with the transmission of ideas…by bolstering the position of reformers in the bureaucratic structures, usually the finance ministry or central bank.” Where the Fund initially finds none available, it often seeks to create sympathetic interlocutors via technical assistance, further underscoring their importance for the institution’s success (Kahler 1992:129, 1993:377; Woods 2006).

When negotiations with country officials become difficult, the staff may bypass the formal channel of policy dialogue through informal contacts and negotiations with sympathetic interlocutors outside the policymaking process in an effort to strengthen their domestic political position against those unable to commit to IMF policy goals. The case of Chile in the early 1980s illustrates such informal processes. Following a wave of radical market reforms engineered largely by a group of University of Chicago-trained technocrats, Chile suffered an economic crisis. In 1983 and 1984 these technocrats, due to mounting domestic pressure, were replaced by nationalist entrepreneurs, including the new finance minister Luis Escobar Cerda, who then promptly raised tariffs and took expansionary measures to stimulate the economy. The IMF opposed these expansionary measures, and, when agreement became impossible, it suspended formal discussions for financial support. The staff then sought out sympathetic interlocutors outside the finance ministry, such as Hernán Büchi, then superintendent of Chile’s banks and a supporter of many of the IMF’s policy goals. Bypassing Escobar Cerda, the IMF staff held informal discussions with Büchi, which had begun in late 1982, through 1984. These discussions with Büchi, who had studied economics at Columbia University, formed the basis of an economic program that would result in a $750 million (113 % of quota) multi-year IMF program and would return Chile to the path of market reform when, in 1985, following poor results from the expansionary measures, Büchi became finance minister (Teichman 2001:78–82). Examples such as this one from Chile are part of well-documented pattern of informal interaction in the IMF’s history in which the IMF staff has suspended negotiations or hardened their position with unsympathetic officials, while at the same time making informal contacts with likeminded officials outside the policymaking process in an effort to empower these actors.⁶

Exercising informal governance in these ways, the staff are thus likely to provide favorable treatment to countries with sympathetic interlocutors. When a country requests a lending program the IMF must make a judgment about the extent to which a borrower is likely to fulfill its policy commitments and thus result in success. Here the presence of sympathetic interlocutors makes this task easier since it suggests that there are likeminded officials in the borrowing country who are likely to follow the Fund’s preferred policy goals.

Countries where sympathetic interlocutors are present offer, in all likelihood, a greater chance of success than those without them, and larger programs, by offering greater bargaining leverage and by helping to reduce the severity of the required economic adjustments, are likely to offer greater empowerment of sympathetic interlocutors than smaller programs. Countries with sympathetic interlocutors may therefore be seen as worth the risk of extending larger loans. Working with sympathetic interlocutors thus fits with the staff’s informal career advancement incentives, since it enhances the prospect of crafting a large and successful program. The staff’s desire to work with sympathetic interlocutors also fits with their informal normative orientation as economists seeking to influence the direction of the world economy and the economies of individual borrowers according to a particular set of theoretical principles, that is, those broadly in line with the organization’s cultural predispositions.

In supporting sympathetic interlocutors the argument here is not that the staff lend blindly to those who share their beliefs. The staff are aware that country officials, even those sympathetic to their policy goals, may still have credibility problems because of domestic political constraints (Vreeland 2003). Yet the staff’s training as economists leads them to prioritize designing programs based on policy beliefs, to which other similarly-trained economists are often sympathetic, aimed at maximizing efficiency, with less consideration given to political feasibility. The IMF’s organizational culture, which emphasizes first-best policy prescriptions, reinforces the tendency of the staff to prioritize efficiency over political feasibility (Momani 2005; Chwieroth 2010a:34–40). Indeed, as one internal evaluation of IMF operations suggests, the staff often “fails to take into account existing political constraints, or is so optimistic about the ability of governments to overcome them that it does not consider second-best policy choices that would be consistent both with the maintenance of macroeconomic stability and country-specific political realities” (IMF 2004:12).⁷ The staff, though cognizant of domestic constraints, also may be overly confident in the efficacy of loans as instruments to enhance the domestic negotiating position of sympathetic interlocutors against their opponents (Woods 2006). As a result, the staff often end up displaying what Graham Bird (2004) calls “over-optimism” about the ability and capacity of sympathetic interlocutors to overcome domestic political constraints.

It is also important to point out that I am not arguing that similar training will necessarily lead the Fund staff and their interlocutors to see eye-to-eye on every detail of a particular program. This is clearly not the case. In some cases interlocutors, though broadly sympathetic to the IMF’s position, may pursue more extreme policy goals (Fang and Stone 2012).

In other cases, the presence of interlocutors with similar professional characteristics as the Fund staff can engender an improvement in a country’s technical and analytical capacities, thereby potentially diminishing the Fund’s influence. Indeed, the silent revolution brought with it a remarkable change in the intellectual calibre of the Fund’s interlocutors. As a result, the Fund’s discussions with government officials often became more complex and even-handed. The staff no longer possessed the intellectual strength to assert and press their views on some governments. The vast improvement in the quality of the Fund’s interlocutors meant that some governments were less likely to be deferential to the Fund’s technical expertise. But, on average, policy consensus tends to be stronger in cases where the Fund staff and their interlocutors share similar professional characteristics than in cases where they do not.

2.3 IMF Management and Member States

How are management and the Board likely to view staff recommendations for larger loans for countries with sympathetic interlocutors? Both should offer their support. Indeed, former First Deputy Managing Director Stanley Fischer (1997:26) has suggested that the Fund purposefully seeks out such interlocutors when negotiating programs. But IMF management, which increasingly, especially in important cases, is involved in program design throughout the negotiating process also faces incentives to seek excessive insurance against failure in all cases, not just those where sympathetic interlocutors are present. This may bias management to err on the side of larger programs for all countries (Willett 2002:69).

The Board, for its part, is also likely to look favorably upon programs negotiated with sympathetic interlocutors. But geopolitical and financial interests may lead powerful countries to err on the side of larger programs for strategically important countries, not just those where sympathetic interlocutors are present. As a result, the approach favored by the IMF and its major shareholders can diverge, particularly when geopolitical and financial interests are at stake; with the Fund on occasion refusing to support governments that demonstrate an inability to commit to its policy’s goals, while powerful countries pursue their interests even when they lead them to support programs where sympathetic interlocutors are not present.

The different approaches that the IMF and the U.S. took to Argentina in the 1980s vividly illustrate the possibility for divergence. In the mid-1980s the IMF was reluctant to support “heterodox” stabilization programs that employed wage and price controls. But Argentine officials, by appealing to concerns over the systemic implications of default, enlisted U.S. officials to press the IMF to lend (Kaufman 1989:400–401). On this occasion the IMF bent to the contrasting approach of its most powerful shareholder, but, interestingly, it refused to augment the level of financial assistance available to Argentina (James 1996:377–378). The failure of Argentina to commit fully to the policy goals of the IMF contributed to its hesitation to augment its financial commitment.

In 1988, Argentine officials again sought IMF support for a heterodox program. But this time the Fund insisted on orthodox measures as a precondition for progress on IMF lending and debt negotiations with commercial banks. But Argentine officials refused, arguing that debt relief should come first. Yet despite U.S. pressure on the Fund to support Argentina, which included a direct appeal from the Treasury Secretary, the IMF refused to lend, citing a history of failed programs and an unwillingness of government officials to commit to the Fund’s policy goals.

However, unlike Argentina in the 1980s, in many countries the influence and inclination to lend from powerful countries and the staff may co-exit and be difficult to disentangle. Some may interpret this co-existence as a symptom of influence of powerful countries. The strong version of this argument would be that powerful countries push the staff to reward countries with “good” government officials, while the weak version of this argument would be that the staff recognize the preferences of powerful shareholders and thus craft IMF programs in accordance with these preferences. Yet neither version is entirely convincing.

While powerful shareholders may place additional political pressure on the IMF to craft generous programs for strategically important countries where sympathetic interlocutors are present (such as Mexico in the 1980s and 1990s), such cases tend to be the exception rather than the rule. As stated earlier, powerful shareholders do not typically actively intervene in the design of loan programs. Routine intervention would, in fact, be counterproductive, as it would undermine the benefits of delegation and the legitimacy of the IMF as a multilateral organization. Moreover, as principal-agent theorists emphasize, powerful shareholders have put in place various control mechanisms to help ensure staff behavior reflects their long-term preferences, so in equilibrium there is little need to closely monitor staff behavior and good reason to permit the staff a degree of autonomy and discretion. All of this suggests that the strong version of the symptoms of influence argument overstates the degree to which staff lending decisions result directly from shareholder pressure. Indeed, as Stone (2008:593) observes, “In ordinary times, the United States and other shareholders have no compelling interest in intervening in the details of conditionality.”

From a borrowing country’s perspective it also makes little sense to employ on a routine basis close relations with powerful shareholders in order to secure more generous loans. Shareholder pressure only becomes operative when the borrowing country places a high priority on securing IMF financing. But borrowing country influence with powerful shareholders is a valuable and limited resource, which is likely only to be drawn upon when the stakes are particularly high. To the extent that the IMF is inclined to look favorably upon sympathetic interlocutors for the reasons I suggest, then there should be little compelling reason for borrowing countries to employ their influence with powerful shareholders to place additional pressure on the Fund.

The weak version of the symptoms of influence argument is problematic in that it attributes too much of IMF lending decisions to shareholder preferences and not enough to autonomous staff behavior. Certainly, the Fund staff is keenly aware of the preferences of their powerful shareholders and they rarely stake out a contrary position. Nonetheless, though powerful shareholders may favor crafting larger loans for countries with “good” government officials, unless these shareholders actively intervene in the design of a loan we would be attributing too much influence to these actors and not enough to the staff. Indeed, adherents to the weak version of the symptoms of influence argument set too high of a bar for identifying autonomous behavior.

Autonomous staff behavior is present not only when IOs overcome opposition from powerful shareholders or when they act contrary to their interests; it also manifests itself in more subtle ways. Though loans for countries with sympathetic interlocutors may help powerful shareholders advance their interest in rewarding “good” government officials, unless these shareholders actively pressure the Fund, such lending decisions should be seen as a form of autonomous behavior (Martin 2006). In such cases the IMF is developing policy that does not stem directly from explicit shareholder demands but nevertheless helps these shareholders realize their interests. Thus, even in cases where statist and staff influence co-exist the staff can be seen as exercising autonomous behavior by deciding how, among the several possible ways they could advance shareholder interests, these interests are best served. Such behavior, as Michael Barnett and Martha Finnemore (2004:28) observe, might be called “autonomy by design.” Thus, rather than being seen as a “symptoms of influence” the decision to lend to sympathetic interlocutors can often be seen as a “symptom of autonomy.”

3 Data, Methods, and Analysis

In this section, I assess whether sympathetic interlocutors lead to larger IMF loan commitments. The statistical findings provide strong support for the argument. Where the Fund staff and a borrowing country policy team share similar professional characteristics, the IMF is likely to provide larger loan commitments.

The data set comprises annual data on 143 IMF loans extended to 29 developing countries from 1975 to 1998 under the Stand-By and Extended Fund Facility programs.⁸ The dependent variable is the ratio of the total amount committed when a new program is agreed and the borrowing country’s quota in the IMF.⁹ While there is an extensive literature that seeks to explain the pattern of IMF program participation, the loan commitment ratio is the more theoretically appropriate dependent variable. In principle, each member state is entitled to participate in an IMF program and thus requests are rarely rejected. However, the loan commitment ratio is much more susceptible to the influences discussed. While informal staff influence may also extend to the design of conditionality, I focus here on the size of the loan because, as Stone (2011:136) suggests, “the amount of access is one of the most intensely political decisions” within the IMF and because, as opposed to the design of conditionality, it is a decision where the staff are expected to have less control (Copelovitch 2010:73). Moreover, the size of the loan is generally more strongly linked than the design of conditionality with the theoretical emphasis here on the likelihood of program success, and thus more susceptible to the informal motivations outlined above (Stone 2011:136; Ivanova et al. 2003).

The amount of loans outstanding has fluctuated significantly over time, with peaks observed during the debt crisis of the 1980s and during the emerging market crises of the late 1990s. The size of loans in the dataset varies between a minimum of nearly 15 % of quota to a maximum of nearly 1940 %. Since these data are positively skewed, I use the natural log of the loan ratio.

At the heart of my argument is the notion that the IMF’s loan commitment depends in part on shared policy goals between the IMF area department staff and their interlocutors in the borrowing country, which are typically the finance minister and the head of the central bank. Building on earlier work on the socializing experience of common professional training, I develop new data using the educational background of the IMF staff and borrowing country officials as a proxy for shared policy beliefs.¹⁰ Extensive research, much of it using surveys, has shown professional training in Anglo-American economics departments to be an important socializing experience in generating shared policy beliefs (Chwieroth 2010a; Colander 2008; Colander and Klamer 1987; Klamer and Colander 1990; Fourcade 2009). While consensus does not extend to all areas, these surveys show that belief heterogeneity tends to be less severe among Anglo-American-trained economists, particularly when it comes to a theoretical core stressing market efficiency and rationality, compared to that between American-American-trained economists and those trained elsewhere.

For the IMF I created a sample of 287 area department staff members from 1946 to 1998 from telephone directories found in the IMF Archives and supplemented by the IMF Communications Department. These data enable a reasonable reconstruction of the area department chain of command responsible for program design. For borrowers I created a sample of 1173 chiefs of government, finance ministers, and heads of the central bank from 44 developing countries from 1969 to 1998.¹¹ I identify economic policymakers from the Current World Leaders Almanac, the CIA’s Chiefs of State and Cabinet Members of Foreign Governments Directory and Keesings Record of World Events, and the websites of the finance ministries and central banks of various countries. Responses to emails and faxes from some finance ministries and central banks supplemented these sources. Data on whether an IMF staff member or borrowing country official obtained an Master’s degree or Ph.D. from an American or British economics department were obtained from Digital Dissertations, Index to Theses in Great Britain and Ireland, LexisNexis Executive, and various documentary sources (such as International Who’s Who and Who’s Who in Central Banking). ¹²

Figure 1 plots the mean proportion by regional coverage of Anglo-American trained economists in IMF area departments from 1946 to 1998. It reveals some interesting variation over time. Until the 1980s the area department covering Latin America contained the highest proportion of Anglo-American-trained economists. This likely reflected an early organizational practice of the Fund, since discarded to avoid conflicts of interest, which sought to place a region’s nationals in their respective area departments.¹³ Because such a large number of the Fund staff at the time were American nationals it is not surprising to find such a large proportion of Anglo-American-trained economists in this department.

Fig. 1

Professional characteristics of IMF staff, 1946-1998. The values for each for the period 1946-1949 is zero

In the 1970s there was a decline in the overall prevalence of Anglo-American-trained economists working in the area departments. Yet this trend reversed itself in the 1980s as Anglo-American economists becoming increasingly represented across all area departments, with the largest concentrations in those covering Asia and Europe. These trends, Sarah Babb (2003:19) speculates, were likely due to the Fund’s tendency in the early to mid-1970s to recruit more staff with public financial sector work experience than economists before reversing course in the late 1970s.¹⁴

Figure 2 plots the mean proportion by region of IMF interlocutors who were Anglo-American trained economists. Figure 2 reveals that a silent revolution of sorts did occur in the 1980 and 1990s in a number of developing countries. Prior to the 1980s less than a tenth of the Fund’s interlocutors in the sample of countries included in the analysis were Anglo-American trained economists. The increase in the proportion of sympathetic interlocutors since the 1980s has been particularly sharp in Asia and Latin America. This proportion also has increased dramatically in Europe, but one should be cautious in reaching conclusions from this finding because the regional sample is limited to two countries (Israel and Turkey). Historically, the data show that the Fund has encountered the fewest sympathetic interlocutors in sub-Saharan Africa, which could help account for its relative lack of success in the region.

Fig. 2

Professional characteristics of IMF interlocutors, 1969–1998

I use these data to create three variables: (1) Anglo-American staff, which captures the proportion of area department staff who received professional training in economics in the United States or the United Kingdom; (2) Anglo-American policy team, which captures the proportion of borrowing country officials who received professional training in economics in the United States or the United Kingdom; and (3) Sympathetic interlocutors, which is then the interaction effect of these two variables. This latter variable is the primary independent variable. All other things being equal, the presence of borrowing country interlocutors who share similar professional training as the IMF staff should lead to larger IMF loan commitments.

By themselves, the Anglo-American staff and policy team variables assess how IMF loan commitments are affected when the other component of the interaction term equals zero; that is, they capture in part how IMF lending decisions are shaped where the staff and their interlocutors share dissimilar policy goals. Where the IMF does not encounter sympathetic interlocutors, it may be more cautious in extending loans. The uninteracted Anglo-American policy team is statistically of little importance to the analysis because the IMF area department staff variable never equals zero in the sample (Braumoeller 2004).

Admittedly, common professional training is only one of the possible ways that shared policy beliefs can emerge. In addition, as suggested earlier, it also may not necessarily lead the Fund staff and government officials to agree on the details of a program. Both possibilities should make it more difficult to uncover a positive relationship, as they should result in a negative bias against the hypothesized relationship.

In the statistical analysis I also control for alternative explanations featured in the literature on IMF lending. The first set of control variables includes a number of economic factors commonly featured in the literature on IMF lending. Economic variables from the World Bank’s World Development Indicators are used to capture a country’s reserve position, monetary conditions, overall debt, and debt profile.¹⁵ I also include a separate indicator from Reinhart and Rogoff (2008) that captures the presence of a banking crisis. To proxy global financial conditions and the scarcity of private international capital, I include the nominal U.S. Treasury bill rate found in the IMF’s International Financial Statistics (IFS). If more formalized technocratic considerations alone motivate the design of IMF programs then countries with more severe economic problems should receive more generous loans.

The second set of controls captures political influences on IMF lending. In a number of studies, the evidence suggests that the U.S. exercises the greatest influence over the IMF as its leading shareholder, though its influence tends to be exercised informally and limited to countries where its geopolitical and financial interests are at stake. To assess the intensity of U.S. geopolitical interests, I follow others (Broz and Hawes 2006; Oatley and Yackee 2004) in using United Nations (U.N.) voting affinity scores for countries vis-à-vis the United States. This measure ranges from −1 to 1, with higher values indicating closer geopolitical relations. The Fund’s lending decisions can also be shaped by the extent to which a crisis in a given country may imperil U.S. financial institutions. Following the convention in the literature (Broz and Hawes 2006; Stone 2008; 2011), I proxy this influence using country lending exposure by U.S. commercial banks. The BIS (2007) provides these data. If the U.S. intervenes to further its geopolitical and financial interests, politically influentially and financially important countries should receive larger loans. As a control for domestic political constraints I include the natural log of the number of veto players (CHECKS), which Vreeland (2003) finds to be an important determinant of IMF lending decisions.

The third set of controls includes proxies for organizational imperatives identified by public choice scholars that may lead the Fund to provide more generous loans when it has more resources to lend and when its members are considering whether to increase the organization’s resources via a quota review. I use two variables from Dreher and Vaubel (2004) to assess these possibilities. One variable, the IMF’s liquidity ratio, divides the sum of outstanding loan commitments by its total quota resources. The other variable is a dummy variable that takes on the value of one in the years a quota review was underway.

Since the sample of countries participating in IMF programs is systematically different from the overall population, statistical analyses of IMF loan commitments must address the issue of non-random selection (Steinwand and Stone 2008). I address this issue using propensity score matching. The basic intuition is to match each country-year under an IMF program (a “treated” observation) with a country-year without an IMF program (a “control” observation) based on the observed covariates that are as close to identical as possible. This “nearest-neighbor” strategy provides a propensity score for each observation ranging from zero to one that captures the predicted probability of IMF program participation. Inclusion of this propensity score in the loan commitment models helps to minimize selection bias (Ho et al. 2007).

In both the loan commitment and program participation specifications, I also control for temporal dependence. While tests indicate that serial correlation is not present in the loan commitment specification, I account for a country’s past history with the IMF by including a dummy variable that takes the value of one when a country is already under an IMF program.¹⁶ In the binary program participation specifications, where the dependent variable takes on a value of one if a country received an IMF loan in a given year, I control for temporal dependence using the country-specific number of years since the last IMF program, its square, and its cube (Carter and Signorino 2010).¹⁷

For the loan commitment specification, I ran ordinary least squares regression with panel corrected standard errors.¹⁸ I use a logit analysis to generate the propensity scores included in the subsequent loan commitment specification. Table 1 presents the results. The commercial bank exposure variable restricts the time-series to 1983 to 1998, thereby reducing the sample size from 143 to 100. I therefore estimate two sets of models. In Models 1–4, I exclude the commercial bank exposure variable and analyze data from 1975–1998. Models 5–8 then add the commercial bank exposure variable. For each set of models I first present a baseline model (Models 1, 3, 5, and 7), in which I exclude the professional characteristics variables. I then introduce the professional characteristics variables in subsequent models (Models 2, 4, 6, and 8).

Table 1 Covariates of IMF loan commitments, 1975–1998

Macroeconomic factors are found to be critical determinants of IMF program participation. Countries with lower levels of reserves relative to imports (Models 1 and 2), smaller money supplies relative to reserves (Models 1, 2, 5 and 6), and facing higher U.S. Treasury bill rates (Model 1) are more likely to enter IMF programs.¹⁹ Quota reviews are found in Models 1 and 2 to lower program participation, suggesting the staff may decrease lending to enhance the perception that they are a responsible manager of IMF resources. The negative coefficient on years since last IMF program cubed in Model 6 provides some evidence for recidivist tendencies. Sympathetic interlocutors and U.S. geopolitical and financial interest do not appear significant at the selection stage.²⁰

The remaining models in Table 1 present alternative specifications of the loan commitment regressions. As expected, in both models where it is included the sympathetic interlocutor interaction effect is positively and significantly related to IMF loan commitments. The coefficient on Anglo-American staff is negative, as expected, and achieves marginal significance in Model 8.

Importantly, the goodness-of-fit statistics – R², the Bayesian information criterion (BIC), and Akaike information criterion (AIC) – clearly indicate that the inclusion of the professional characteristics variables explain a larger share of the overall variation in IMF loan commitments than the baseline models.²¹ A series of F tests confirmed that models including the professional characteristics variables provide a better fit than the baseline models. This evidence strengthens the conclusion that staff informal governance is a critical, though hitherto neglected, important determinant of IMF lending. Controlling for non-random selection and a range of different explanations, the evidence suggests that borrowing country policy teams staffed with sympathetic interlocutors receive better treatment from the IMF.

Figure 3 illustrates the conditional effect of Anglo-American policy team on IMF loan commitments. Using the results from Model 8 this figure graphs the coefficient on Anglo-American policy team as Anglo-American staff varies from its minimum to maximum values. The figures also include a histogram illustrating the distribution of the Anglo-American staff variable. As hypothesized, Figure 3 illustrates the marginal effect of Anglo-American policy team is significant and changes signs at both ends of the distribution. At the left end of the graph, differing beliefs between the policy team and staff lead to smaller loans, whereas the commonality of beliefs leads to larger loans at the right end of the graph. Put differently, where the IMF staff and their interlocutors share similar professional characteristics, the Fund is likely to offer larger loans.

Fig. 3

Loan size (Amount/Quota)

At higher levels of Anglo-Americanization within the relevant area department, the staff increasingly defines orthodoxy in terms of a particular set of theoretical principles. In such circumstances, the absence of sympathetic interlocutors in a borrowing country may lead the staff to be normatively oriented against such a government, sceptical of its commitment to the IMF’s policy goals, and thus less willing to treat it favorably. On the other hand, greater belief homogeneity within particular area departments likely induces the staff to be normatively oriented toward interlocutors who share their professional characteristics and to have greater confidence that such interlocutors are more likely to produce successful programs. The result is a willingness to provide more generous loan commitments.

Using the results from Model 8, Table 2 presents substantive quantities of interest illustrating the effect of a one standard deviation increase in Anglo-American policy team, conditional on varying levels of Anglo-American policy staff, holding all other variables constant at the means. As the quantities illustrate, a one standard deviation increase in Anglo-American policy team, which is roughly equivalent to adding one additional sympathetic interlocutor, leads to substantial rises in IMF loan size but only at intermediate and high levels of Anglo-American staff. This effect is found to exceed that exerted from a one standard deviation increase (3 %) in US bank exposure.

Table 2 First differences, Anglo-American policy team and U.S. Bank exposure

However, the effect of exceptionally intense U.S. financial interests is likely to be substantively more important than exceptionally close alignment between the policy goals of the IMF staff and their interlocutors. Using the U.S. bank exposure data for Mexico in February 1995, which Stone (2011:145) reports as 18 %, and setting all other variables to their means, the model predicts an IMF financial commitment of 675 % of quota. On the other hand, using the values from the case of Indonesia in 1997 (where Anglo-American policy team = 66.6 and Anglo-American staff = 65) and setting all other variables to their means, the model predicts an IMF financial commitment of 260 % of quota.

In the sample there are 25 cases in which Anglo-American policy team and Anglo-American policy staff take on high values. These cases cluster in Latin America (13 cases) and Asia (9 cases), with Mexico (1995), Thailand (1997, and Indonesia (1997 and 1998) as recent examples.²² Since these cases are somewhat collinear with U.S. financial and geopolitical interests, it is difficult to tease apart the relative importance of statist and staff influence. U.S. financial interests were intense (one standard deviation above their mean) in four of these 25 cases.²³ In such cases informal statist and staff influence co-exist, and, like Indonesia in 1997–1998, the relative importance of statist versus staff influence is hard to ascertain because U.S. preferences fit so closely with those of the IMF staff as well as management (Stone 2011:170–173; IEO 2003:12–13).

U.S. geopolitical interests, as measured by the U.N. voting affinity measure, are intense (one standard deviation above their mean) in six of the 25 cases, including two cases where U.S. financial interests are also intense.²⁴ Informal statist and staff influence also likely co-exists in countries, such as Pakistan (1994, 1995, 1997) and the Philippines (1984, 1986, 1991), which play important roles in US foreign policy but generally vote against it in the U.N. General Assembly (Stone 2011:165–166). Among the 25 cases it is easier to conclude that staff influence is likely substantively more important than statist influence in those, such as Sri Lanka (1983), where the interests of powerful countries are less intense.

Turning to the remaining explanations, variables associated with more formalized technocratic considerations receive some support in Models 3, 4, 7 and 8, which suggest larger loan commitments are associated with smaller money supplies relative to reserves, larger public debt burdens, short-term debt profiles, banking crises, and higher U.S. interest rates. Support for political considerations is mixed. On the one hand, there is consistent support in Models 7 and 8 confirming previous research that U.S. financial interests impact IMF lending decisions. Countries where there is significant exposure by U.S. commercial banks receive more generous loans. On the other hand, there is no support for the view that countries of geopolitical value to the U.S., as measured by the U.N. voting affinity measure, receive more generous loans from the IMF.

There is some evidence that domestic political constraints have an effect on IMF loan size, with the veto player measure positive and significant in Models 3 and 4. Models 7 and 8 provide weak evidence that a country already receiving IMF credit under a pre-existing program is likely to receive a smaller loan. In contrast to what public choice arguments would predict, Model 4 suggests the IMF lends more generously when its liquidity ratio is higher. Quota reviews are found in Models 7 to result in smaller loans, suggesting once again that the staff may seek to restrain lending to enhance the perception that they are a responsible manager of IMF resources. Lastly, the propensity score is found to be significant in Models 7 and 8.

Ultimately, these results provide strong confirmation that IMF loan commitments depend in part on the extent to which staff and borrowing country officials share common professional characteristics. This finding is in line with the arguments made here about the importance of staff informal governance of IMF lending decisions. Others suggest that IMF staff influence tends to limited to those countries of lesser geopolitical and financial importance. While the staff may exercise a preponderant influence in such countries, the findings here suggest formalized technocratic considerations and informal state and staff influence co-exist within the IMF, even in high-profile cases where the interests of powerful states are particularly intense.

4 Conclusion

Although rationalists and constructivists tell us to expect considerable autonomy for IO staff, our understanding of how the staff will behave given this autonomy remains limited. With respect to scholarship on the IMF, inadequate attention has thus far been given to how the staff exercises informal governance over its lending decisions. Rather than viewing IMF lending decisions as beholden to formal technocratic considerations or informal member state influence alone, I have argued that informal staff influence, motivated by informal incentives and orientations, also plays an important role. The presence of sympathetic interlocutors in borrowing country policy teams provides the staff both with greater confidence a lending program will achieve success and an opportunity to support officials who share their commitment to a particular set of policy beliefs. The statistical evidence suggested that these career advancement incentives and normative orientations, which weigh heavily on staff judgments and lead them to prefer to work with sympathetic interlocutors, do shape IMF lending outcomes.

A clear implication of this analysis is that a fuller understanding of how IOs work and evolve requires close attention to the evolving makeup, incentive structure, and normative orientations of their staffs. Much of the existing literature has yet to devote much attention to such important features of IO behavior. The rational design approach implied IO behavior was largely a function of the institutional features crafted by states (Koremenos et al. 2001). The recent behavioral turn in the study of IOs has brought to the light the importance of informal governance, but has remained decidedly state-centric. The findings reported here suggest these understandings of IO behavior are incomplete. The behavior of IOs is driven not just by formal rules or informal state influence but also by how their staffs exercise informal influence and respond to informal incentives and normative orientations. Even though it may partly occur within a formal delegation process, our conceputalization of informal governance needs to be expanded to accommodate these dimensions of informal staff influence and structural power.

This finding not only speaks to theoretical debates over IO behavior, but also to current debates over IMF reform. The results provide evidence that the staff provide favorable treatment to government officials with similar professional characteristics. While such behavior is understandable given their incentives and normative orientations, it also serves to elevate certain policy goals over others, to engender blind spots in staff analysis, and to downplay “local knowledge,” thus generating resentment toward IMF orthodoxy and challenges to the organization’s legitimacy. The recent global financial crisis has only served to heighten such negative perceptions. To the extent these criticisms have some basis; these findings should add impetus to efforts to broaden the Fund staff’s recruitment base and to expand the array of actors with which the staff engage in their negotiations and consultations.