More control–less agency slack? Principal control and the risk of agency slack in international organizations

Jankauskas, Vytautas; Knill, Christoph; Bayerlein, Louisa Article — Published Version More control–less agency slack? Principal control and the risk of agency slack in international organizations Regulation & Governance Provided in Cooperation with: John Wiley & Sons Suggested Citation: Jankauskas, Vytautas; Knill, Christoph; Bayerlein, Louisa (2023) : More control– less agency slack? Principal control and the risk of agency slack in international organizations, Regulation & Governance, ISSN 1748-5991, John Wiley & Sons Australia, Ltd, Melbourne, Vol. 18, Iss. 4, pp. 1171-1192, https://doi.org/10.1111/rego.12525 This Version is available at: https://hdl.handle.net/10419/306094 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/4.0/ More control–less agency slack? Principal control and the risk of agency slack in international organizations Vytautas Jankauskas Department of Political & Social Sciences, Zeppelin University, Friedrichshafen, Germany Christoph Knill LMU Munich, Institute of Political Science, Munich, Germany Louisa Bayerlein Department of Political Science, European University Institute, Fiesole, Italy Abstract Principal-agent theorizing is based on the idea of a linear inverse relationship between principal control and agency slack: the higher the control over the agent, the less likely is the agent to slack. In this paper, we challenge this assumption by explicitly taking the varying nature of agents into account. While control may reduce the agent’s room for maneuver, it does not explain the extent to which different agents are inclined to put efforts in circumventing these obstacles. Focusing on international organizations (IOs), we measure member states’as principals’control over IO administrations as their agents as well as the latter’s intrinsic propensity to slack across eight major IOs. The analysis shows that low control by the principal is not necessarily associated with run-away agents, whereas high control is not necessarily associated with servant-like agents. The assumed control–slack relationship can thus be distorted and determining an ideal level of control is not possible without considering the agent’s entrepreneurialism. Keywords: agency slack, control, international organization, international public administration, principal-agent theory. 1. Introduction Over the past decades, the principal-agent (PA) framework has evolved as one of the main analytical tools to study international organizations (IOs) and global governance. Member states as principals delegate authority to IO administrations as their agents to achieve delegation benefits such as efficiency, credibility, or legitimacy (Hawkins et al., 2006a:13–20). However, agents are assumed to develop diverging interests and so the key concern for PA literature is how to ensure agents’faithfulness toward their principals and reduce the risk of agency slack, that is, an “independent action [from the agent] that is undesired by the principal”(Hawkins et al., 2006a: 8). So far, the conventional explanation for this key delegation conundrum is based on the assumed linear inverse relationship between principal control and the risk of agency slack: the higher the control over the agent, the less likely the agent should be to slack (Abbott et al., 2016; Conceiç˜ ao-Heldt, 2013; Hawkins et al., 2006b; Heldt, 2017; Kassim & Menon, 2003; Kiewiet & McCubbins, 1991; Nielson & Tierney, 2003; Sobol, 2016). Of course, member states also need to consider the costs of control itself, but the general assumption is that agents as self-serving actors will slack if not properly controlled. In this paper, we challenge this core PA assumption arguing that it should not always hold, because the intrinsic nature of the agent and how it actually behaves may vary despite formal characteristics of the delegation contract. While control may reduce the agent’s room for maneuver, it does not explain the extent to which different agents are inclined to put efforts in circumventing these obstacles. There is thus a missing element in Correspondence: Vytautas Jankauskas, Zeppelin University, Am Seemooser Horn 20, 88045 Friedrichshafen, Deutschland, Germany. Email: [email protected] Accepted for publication 7 April 2023. © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. Regulation & Governance (2024) 18, 1171–1192 doi:10.1111/rego.12525 conventional PA theorizing, and it relates to the oftentimes neglected varying nature of the agent. Indeed, PA theory is known of its predominant focus on the principal side of the relationship (Tamm & Snidal, 2014: 138). Hence, we test the assumed inverse relationship between control and the risk of agency slack empirically, focusing on international public administrations (IPAs)—IO secretariats—as agents and IO member states as their (collective) principals. Drawing on primary IO documents and original interview data, we conduct a congruence analysis measuring both member states’control and IPAs’risk of agency slack across eight major IOs. To provide a more in-depth picture of our measurement, we zoom into four cases displaying variation, focusing on the IMF, NATO, ILO, and UNHCR administrations. For one, our findings show that IPAs display pronounced differences in their risk of agency slack as indicated by the prevalence of behavioral routines in these organizations. While some IO administrations develop highly entrepreneurial routines demonstrating high risk to slack, others remain servant-like—regardless of the formal principal control under which they operate. Importantly, we thus find that the observed risks of agency slack are not always congruent with the expected levels of member states’control over these administrations. These findings make a threefold contribution. Theoretically, the paper scrutinizes PA theory showing that the assumed inverse relationship between the level of principal control and the risk of agency slack can be distorted. Differently put, the risk of agency slack is not a mere by-product of (missing) principal control but is also related to the intrinsic nature of the agent. Two implications stand out from our findings. First, not all agents demonstrate high risk of agency slack if not harshly controlled. Second, control does not always automatically constrain the agent’s behavior reducing its propensity to slack. This way, the article calls to put more focus on the agent side of the delegation relationship and contributes to existing accounts scrutinizing agents’and IPAs’characteristics (see Barnett & Finnemore, 1999,2004; Bauer et al., 2017; Chwieroth, 2008; Ege, 2020; Hawkins & Jacoby, 2006; Jankauskas, 2022). It also offers venues for future research which seeks to determine adequate levels of institutional control in (international) organizations (Abbott et al., 2020; Long & Sitkin, 2018; Stone, 2011). Conceptually, the paper introduces a novel way to grasp the risk of agency slack by looking at the agent’s behavioral routines. In contrast to existing accounts focusing on what an IPA could do given the formal structure, we go a step further by looking at how the IPA actually behaves beyond the formal constraints. To capture these agent’s behavioral dispositions, we focus on organizational routines guiding IPAs’behavior, that is, informal and collectively internalized practices and processes characterizing the way an IPA engages in IO policymaking (Bayerlein et al., 2020; Knill, 2001; Knill et al., 2019; Mintrom & Norman, 2009). In this regard, IPAs can display a more or less pronounced entrepreneurial orientation. As we later discuss, the more entrepreneurial an IPA is, the higher the risk it will engage in a behavior that might be unwanted by member states. Empirically, we offer two separate original measures of both the extent of member states’control and IPAs’ risk of agency slack across eight major IOs. Especially the latter lacks comparative empirical evidence in IO research. The additional in-depth analysis reveals puzzling cases for PA theory, pointing to the IMF where high risk of agency slack exists despite high levels of control as well as NATO where low risk of agency slack prevails despite low levels of control. In the following, we first revisit the core of PA theory and the assumed inverse relationship between principal control and the risk of agency slack (Section 2). In Sections 3and 4, we outline our measurement concepts and operationalization for both concepts putting our theoretical challenge of the PA assumptions to an empirical plausibility probe (Section 5). The findings are discussed and concluded in Section 6. 2. Principal control and the risk of agency slack in delegation relationships Delegation is defined as “a conditional grant of authority from a principal to an agent that empowers the latter to act on behalf of the former”(Hawkins et al., 2006a: 7). The resulting relationship structure is characterized by power and information asymmetries. While the ultimate power remains with the principal who can rescind the agent’s authority, the expert knowledge favors the agent. Essentially, PA theory assumes interest divergence between the principal and the agent, the latter assumed to be a self-interested utility maximizing actor. Although Hawkins et al. (2006a: 7) argue that PA theory “does not imply any particular assumptions about the preferences of the actors,”the authors at the same time assume that these preferences diverge, so there is a “natural and perhaps inevitable conflict of interest between the parties” © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1172 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK (Hawkins et al., 2006a: 24; see also Kiewiet & McCubbins, 1991: 24). Cortell and Peterson (2006: 256) also conclude that “PA approaches generally take agents’preferences as given and assume they will conflict with those of principals.”Indeed, if there was no preference divergence, PA models would “lose their power” (Miller, 2005: 205). This divergence creates incentives for the agent to pursue its own agenda and slack, that is, engage in an “independent action […] that is undesired by the principal”(Hawkins et al., 2006a: 8). At the same time, the embedded information asymmetry creates opportunities for the agent to slack (Kiewiet & McCubbins, 1991). Hence, the risk of agency slack is assumed to be structurally inherent to all PA relationships. The key concern of the principal, then, is how to minimize the risk of agency slack. In this regard, PA theory assumes that the main instrument for the principal is to establish control tools. Control is defined as mechanisms used by the principal to observe the agent’s behavior and direct it toward the principal’s interests (Kiewiet & McCubbins, 1991; Miller, 2005). These mechanisms are supposed to reduce the agent’s autonomy and thus “the range of potential independent action available to an agent”(Hawkins et al., 2006a: 8). The central goal of PA theory has therefore been to clarify “how particular institutional forms can be used to increase the likelihood of compliant behaviour by bureaucrats”(Huber & Shipan, 2002: 27), and, conversely, “minimize actual [agency] losses through control mechanisms designed into the P-A contract”(Hawkins & Jacoby, 2006: 226). As Hawkins et al. (2006a: 31) put it, “the principals can always reduce slack by tightening oversight.”Or in more relative terms, “the more closely the principal monitors the agent’s behaviour […]–the lower the risk of agency slack” (Abbott et al., 2016: 723; Kassim & Menon, 2003; Lake & McCubbins, 2006). In theory, there should thus be a linear inverse relationship between control and the risk of agency slack: the tighter the former, the lower the latter (Fig. 1). Given that agency slack is seen as a direct function of principal control, it is hardly surprising that much scholarship on IO delegation puts its predominant focus on the principal’s side of the relationship (Tamm & Snidal, 2014: 138). Yet, taking agents and their potentially varying nature seriously has profound consequences. First, it challenges the assumption that agents will slack if only given the opportunity to do so. From a conventional perspective on delegation, the extent to which a given agent might develop and act upon diverging preferences is a constant that is seldomly explicitly theorized. If agents’intrinsic slack propensity varies, however, a loosely controlled agent does not necessarily need to slack, while a tightly controlled agent may not automatically become “a perfect handmaiden”(Cortell & Peterson, 2006). Second, even when controlled similarly tightly, agents may vary in whether and to what extent they try to work around their principals’rules. Indeed, while formal control arrangements are primarily explained against the background of principals’preferences, institutional path-dependencies, and functionalist reasoning (Hawkins et al., 2006a; Koremenos et al., 2001), the risk of agency slack relates to agents themselves, that is, their intrinsic propensity to engage in slacking behavior. As we argue in the following section, such slack propensity is rooted in agents’behavioral routines which develop over time depending on factors like common professional beliefs and normative orientations of organizational staff (esprit de corps) as well as administrative perceptions of external challenges related to organizational fields or political threats (Bayerlein et al., 2020; Knill et al., 2019). In this regard, a number of studies have purposefully drawn attention to agents’characteristics (see, for instance, Barnett & Finnemore, 1999,2004; Elsig, 2011). Nascent IPA research has shown that—over time—IO administrations develop a life of their own beyond the IO’s delegation contract and controls that bind them Figure 1 Assumed inverse relationship between the risk of agency slack and control according to principal-agent theory. © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1173 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. (Bauer et al., 2017; Chwieroth, 2008; Eckhard & Ege, 2016; Ege, 2020; Knill et al., 2019). While these studies clearly speak against a principalor state-centric perspective when studying IOs, a comprehensive empirical test of the core P-A assumption regarding slack–control relationship is still needed. After all, if we accept the argument that we cannot deduce the risk of agency slack from control alone, the implication is that the inverse relationship between the level of control and agents’slack propensity should be tested rather than taken as given. In the sections to come, we investigate these issues empirically. 3. Conceptualizing and operationalizing the risk of agency slack Grasping the risk of agency slack conceptually is not an easy feat. Especially if we do not want to treat it as a mere by-product of (missing) control. To determine the risk of agency slack, we must turn our attention to the agents themselves. Agency slack as such, that is, agents’actions undesired by the principal, can only be observed after the fact (Conceiç˜ ao-Heldt, 2013; Heldt et al., 2022). From the outside, we can only know for sure whether an agent slacked after observing whether there has been a policy output that the principal did not desire (see, e.g., Cortell & Peterson, 2006). Additionally, there is the problem of observational equivalence (Dür & Elsig, 2011). When looking not at the outputs as such but whether the agent’s behavior had repercussions, we cannot say for sure if control and sanctioning mechanisms were not triggered because the agent slacked “smartly” or because the anticipation of sanctions prevented slacking (Delreux & Adriaensen, 2017). Also, to capture agency slack empirically or conceptually, the researcher would need to infer the (collective) principal’s specific preferences—as well as the agent’s—in order to determine how much of a gap there is between them. This, in turn, would depend on case-by-case preference measures for separate policy issues. Therefore, in PA theorizing (and for our purposes here) the focus is on the risk of agency slack, that is, the likelihood that an agent will deviate from the member states’instructions. Following studies that disentangled IOs highlighting their bureaucracies as the actual agents (Barnett & Finnemore, 1999,2004; Biermann & Siebenhüner, 2009; Chorev, 2012; Chwieroth, 2008), we focus on prevalent behavioral routines of IO administrations to capture their propensity to slack. Influenced by their organizational environment and functional pressures, IPAs establish sticky informal working routines through socialization. 1 Yet, stickiness does not mean that administrative routines are static features of an organization. Stickiness does not completely exclude the possibility of change. We rather assume that changes in organizational routines are less likely than for less institutionalized features of an organization, and only unfold over longer periods of time. Variation in organizational routines implies that some IPAs regularly push for their own agenda, while others willingly follow member states’commands and orientate toward mandate implementation (Bayerlein et al., 2020; Biermann & Siebenhüner, 2009; Jankauskas, 2022; Knill et al., 2019). Such behavioral routines, or administrative styles (Knill, 2001; Knill et al., 2019:85–86), can be defined as “an institutionalized informal modus operandi that materializes as a guiding principle over time and by repetitions, routinization, and subsequent internalization.” This relates to a wider notion of “organizational culture”but is different from the basic macro-level understanding of historically emerged administrative systems (Barnett & Finnemore, 1999,2004; Chwieroth, 2010) or roles and behaviors of IPA individuals at the micro-level (Yi-chong & Weller, 2008). Conceived of as standard informal operating procedures, administrative styles refer to the lesser explored meso-level of behavioral routines. 2 These informal patterns capture what makes public administrations “tick”when it comes to their role in initiating, formulating, and implementing the policies of their respective IOs. Such behavioral patterns “correspond to an ideal typical characterization of IPAs’‘styles’as entrepreneurial or servant-like”(Knill et al., 2019: 85). In general, IPA behavior can be shaped primarily by the goal of fulfilling their principals’wishes and needs (prevalence of servant routines) or reflect more entrepreneurial orientations toward developing and pursuing own bureaucratic interests to strengthen institutional consolidation and enhance policy effectiveness (prevalence of entrepreneurial routines). 3 Accordingly, the extent to which an IPA displays more or less entrepreneurial orientations has strong implications for its likelihood to engage in agency slack. While a less entrepreneurial, servant-like IPA can be generally expected to routinely align its behavior with the preferences of the principal even in the absence of tight control, a more entrepreneurial IPA is more likely to follow its own interests and work around even tight political oversight. 4 © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1174 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK Yet, one might argue that the risk of agency slack cannot be judged by merely looking at behavioral routines of the agent, but in addition has to take into account the degree of autonomy given to the agents by design (Barnett & Finnemore, 2004). If an IPA is designed as being entrepreneurial from outset, as it is the case for the EU Commission, for instance (Knill et al., 2016), entrepreneurial routines should hence hardy come with agency slack. Yet, apart from the fact that the supranational EU Commission is fundamentally different from conventional IPAs, the mere presence of entrepreneurial orientations comes with a generally higher risk for agency slack, even if increases of highly autonomous IPAs. In other words, the more entrepreneurial routines are dominant, the higher is the probability that IPA behavior differs from the principals’intentions. Approaching the risk of agency slack via IPAs’entrepreneurial routines has several benefits. It does not capture actual agency slack as a fait accompli but looks at the propensity for such outputs to emerge out of IPAs’ day-to-day work. Thereby, it grasps IPAs’behavior beyond their structural features and institutional characteristics, such as mandates and control. It does not discriminate between different issues the IPA works on and casespecific principal or agent preferences (which hinders comparative empirical research) but considers general IPA working procedures throughout the cycles of typical IO policymaking. It does not assess preference divergence or IOs’outputs as such, but the basis for “own bureaucratic”goals to be formed and routinely pursued. We can thus capture how IPAs fulfill their tasks as opposed to (and in relation to) their mandate which specifies what they are asked to do. As noted above, such entrepreneurial routines emerge over long time periods and—once established—tend to be highly sticky. IPAs’entrepreneurialism is thus not determined by the initial mandate and a set-up of an IO, but it is developed independently as an IPA’s“life of its own.”This allows us to measure IPAs’ entrepreneurialism and hence the risk of agency slack separately and then compare its levels with the extent of control. In measuring the level of IPAs’entrepreneurial behavior, we draw on six indicators (Table 1) which are commonly employed for grasping administrative routines (Kingdon, 1984; Knill et al., 2019; Mintrom & Norman, 2009). These indicators cover core informal administrative routines during all stages of policy-making, that is, policy initiation, drafting, and implementation. Less entrepreneurial, servant-like administrations follow a more passive and reactive role in each of the activities, operating “by the books”and refraining from attempts to intervene with politics beyond their formal mandate. More entrepreneurial administrations, by contrast, develop more active informal routines along the six policy-making activities, thereby increasing the chance of deviation from the formal delegation contract. First, during policy initiation, the analytical focus typically lies on an administration’s efforts to identify and shape new policy problems (issue emergence). This relates to the origin of IO policies—while more entrepreneurial IPAs tend to develop own policy items (hence, potentially deviating from the initial delegation contract), in IOs with less ambitious administrations, policies usually emerge from the outside of the IO bureaucracy (topdown). Moreover, IPAs differ in their efforts to understand the principal’s preferences during policy initiation phase (mapping of political space). As studies on the European Commission demonstrate (see, e.g., Knill Table 1 Indicators for measuring international public administrations’(IPAs) entrepreneurialism Indicator for IPA activity Description Issue emergence IPA identifies new policy problems on its own and initiates own (bureaucratic) solutions and procedures Mapping of political space IPA uses political facilitation mechanisms to bypass political “red lines”when initiating own policies Support mobilization IPA uses mobilization activities to gain support for its policy proposals from political or societal actors Political anticipation IPA anticipates member state interests to ensure its policies circumvent political discord during policy drafting Strategic use of bureaucratic powers IPA uses its bureaucratic powers in the policy implementation phase to foster or defend bureaucratic interests Policy promotion IPA engages in capacity-building and policy promotion of its policy agenda and implementation © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1175 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. et al., 2016), administrations can be highly strategic in dealing with their political principals at a very early stage (e.g., through informal diplomacy). This, in turn, may allow IPAs to formulate policy proposals that favor their preferences yet without crossing political “red lines.”Finally, IPAs vary in their involvement in mobilizing necessary support for their initiatives from other actors (support mobilization). More entrepreneurial IPAs tend to build coalitions with relevant stakeholders (e.g., other IOs or NGOs) to increase external support for their policy proposals (Stone, 2008). Second, during the stage of policy drafting, IPAs vary as to how responsive to member states’preferences they are (political anticipation). In line with functional politicization arguments (see Aberbach et al., 1981), less entrepreneurial administrations tend to avoid politically sensitive issues, formulating policy proposals in a more conservative manner without risking political drawbacks. More entrepreneurial IPAs, by contrast, put more effort in strategic anticipation of member states’preferences, which allows them to exploit potential interest heterogeneity and frame or re-frame relevant policies accordingly (see Hustedt & Salomonsen, 2014). Third, IPAs also differ in their behavioral routines during the implementation stage. For one, IO administrations can employ different approaches as to how they use their bureaucratic powers vis-à-vis the member states (strategic use of bureaucratic powers). More entrepreneurial administrations may actively use their bureaucratic powers such as sanctioning the member states to defend their bureaucratic interests during policy implementation. Other administrations, on the contrary, may seek to avoid any potential conflicts with their political principals and remain more cautious as to how they interpret and use their delegated authority. Furthermore, IPAs may differ in how actively they promote their policies to build pressure for their implementation (policy promotion). Here, more entrepreneurial administrations regularly engage in information dissemination and branding, sometimes even enlisting third-party intermediaries to advocate for their policy targets (see Abbott et al., 2015). More servant-like administrations see less need to actively promote their agenda. In line with our research interest, the underlying question is thus how active an IPA’s behavioral routines in the described activities are. If an IPA engages in a certain activity on a regular basis, the indicator value is high and coded 1; if its engagement is occasional or uneven across organizational units, the indicator value is medium and coded 0.5; if an IPA’s engagement is rare and passive, the indicator value is low and coded 0. For instance, if an IPA actively interacts with third parties (e.g., NGOs) and uses communicative strategies to mobilize their support for its own proposals, the chance that it will leverage its authority and resources regardless of states’preferences increases (see Abbott et al., 2015). Accordingly, the indicator “Support Mobilization”would be coded 1. Similarly, if an IPA puts significant efforts in regular identification of new policy problems and then actively initiates its own bureaucratic solutions, the risk of agency slack increases as the IPA is better able to steer the IO agenda toward its own bureaucratic preferences. In such cases, the indicator “Issue Emergence”is coded 1. Together, the values of six indicators constitute an additive index of IPA entrepreneurialism. While high values reflect entrepreneurial IPA routines which are likely to deviate from the principal’s wishes, low indicator values reflect servant-like agent behavior and thus lower risk of agency slack. The different indicators are treated as equally significant, as we have no prior theoretical knowledge on their exact causal weight. We assume complete substitutability among the indicators as the underlying logic of concept formation (Goertz, 2006). While the substitutability of different dimensions might blur conceptual boundaries, concepts of so-called “family resemblance”have a higher capacity to capture the complexity of political and social reality (Barrenechea & Castillo, 2019). For more details on measurement, see Table A1 in the Appendix. 4. Conceptualizing and operationalizing IPA control The challenge of conceptualizing and measuring principals’control over agents has been approached from different angles, yet most studies focused either on a single type of control or conducted case studies without offering a generalizable metric of principal control. Regarding the former, Parízek (2017), for instance, focused specifically on staff nationalities, while Grigorescu (2010) measured bureaucratic oversight. Regarding the latter, Nielson and Tierney (2003) described in detail how member states controlled the World Bank, yet their control assessment remains case-specific. Against this backdrop, we propose an additive control index which allows for comparison of member states’ control across IOs. Conceptually, scholars usually differentiate between three main types of principal control: © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1176 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK oversight and monitoring, screening and selection, and sanctioning (Conceiç˜ ao-Heldt, 2013; Hawkins et al., 2006b; Heldt, 2017; Kiewiet & McCubbins, 1991; Nielson & Tierney, 2003). To operationalize the extent to which these control types are employed in a given IO, we disaggregate them into distinct control mechanisms (indicators), which may be either specific and direct or broader and indirect. 5 First, to measure direct oversight and monitoring, we include two indicators which provide for variation across IOs and account for reporting lines. For one, oversight can be conducted through centralized evaluation units, which assess IO performance. These units assess IO performance, highlight both successes and failures and thereby fulfill learning and accountability functions. Evaluation reports are usually discussed by IO management and member states. Research shows that varying reporting lines (to member states vs. IPA) affect the extent to which evaluation function serves member states’control interests (Eckhard & Jankauskas, 2020). Hence, the indicator value is coded 1 if evaluation unit reports directly to member states (as in the IMF), 0.6 in case of dual reporting as in the FAO, or 0.3 when reporting only to the administration head as in the IOM (see also Eckhard & Jankauskas, 2019). 6 Also, direct oversight can be conducted through advisory oversight units which oversee IOs’general functioning and, similarlytoevaluationunits,reportprimarily either to member states (coded 1), both member states and the IPA (coded 0.6), or only the IPA head (coded 0.3). To measure more indirect oversight, we focus on the functioning of IO executive governing bodies which make key policymaking and oversight decisions. The more often they meet and discuss issues with each other and with the IPA, the more careful their attention to IPAs’day-to-day functioning and the more intensive IPAs’reporting to member states as part of these meetings. In this line, we code the frequency of governing body meetings (see Table 2). For instance, if a governing body meets weekly, oversight is higher than if such meetings occur only twice a year (like in the IOM). Since control issues are only one part of such meetings’agenda, we treat the intensity of governing body meetings as more indirect control tool. Second, to measure direct screening efforts in IOs, we code formal member states’involvement in appointing IPA staff. Selection of favorable staff members is a well-established control mechanism in PA literature (Hawkins et al., 2006b). If the head of the organization appoints their team in consultation with member states, control is higher (coded 1) than if the IPA head appoints the team unilaterally (coded 0). Similar to staff appointment, yet less directly, member states may seek to control IO administration staff using nationality (see Parízek, 2017). For this, member states may seek to install their nationality staff into the administration. If, however, staff nationalities are equally distributed across executive governing body member states, such control potential diminishes. To measure indirect screening in IOs, we thus consider the extent to which few member states in executive governing bodies have monopolized nationality ties in IPA staff. In particular, we use the ratio of (professional) staff nationals of five main state donors relative to the total (professional) staff. The higher the percentage, the higher the indirect control over IO staff by the (biggest) member states. Finally, to measure member states’direct sanctioning capabilities, we consider formal obstacles for them to agree on sanctioning the IPA. Generally, in order to sanction the IPA, a collective decision among the membership is needed (e.g., to close an IO’sfield mission). The easier it is for member states to make such decisions, the better this control tool functions. While consensus-based decisionmaking might be easily hindered by a single state, majority voting simplifies such procedures. Hence, such control is higher in IOs where the voting system in the governing body is weighted (coded 1), because fewer states can make the decision. In contrast, control is coded 0.5 if unweighted majority voting is needed and 0 if voting is consensus-based. Regarding indirect sanctioning, we consider an IPA’sfinancial fragility—the extent to which an IPA is dependent on member states’flexible funding, coded as a ratio of earmarked (unassessed) contributions to the total IO budget. While regular budget cuts are comparatively hard to implement due to the initial delegation contract, earmarked contributions can be easily withdrawn. The more an IPA depends on such contributions, the higher the potential for such sanctioning. Table 2summarizes the control indicators. Indicator values range from 0 to 1 and are unweighted to allow comparison across control dimensions. 7 For further details on operationalization and measurement results, see Tables A2 and A3 in the Appendix. © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1177 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. 5. Empirical analysis: Principal control and the risk of agency slack in eight IOs To probe the plausibility of our argument that principal control and the risk of agency slack are not necessarily inversely related, we measure both principal control and IPAs’entrepreneurialism in eight IOs: the Food and Agriculture Organization of the United Nations (FAO), the International Labour Organization (ILO), the International Monetary Fund (IMF), the International Organization for Migration (IOM), the United Nations Environment Programme (UNEP), the United Nations High Commissioner for Refugees (UNHCR), the North Atlantic Treaty Organization (NATO), and the World Health Organization (WHO). Together, these organizations constitute a sample representative for the population of major IOs; they differ across various categories such as policy fields, staff sizes, or budgets, deemed relevant in comparative IO studies (Bayerlein et al., 2020; Biermann & Siebenhüner, 2009; Hooghe & Marks, 2015) (see Table A4 in the Appendix). In line with the congruence method (Beach & Pedersen, 2016: 269ff; George & Bennett, 2005: 75; Levy, 2008: 6f), we test whether the values of both factors align with the described PA expectations: whether high member state control over the IPA is congruent with low risk of agency slack measured via IPA’s entrepreneurialism and vice versa. In terms of data collection, for the measurement of IPAs’entrepreneurialism, we draw on 131 semi-structured expert interviews, conducted with IO staff in 2016 and 2017. Each interview lasted about 45 minutes, ranging from minimum 11 to maximum 23 interviews per IO from the sample (see Table A5 in the Appendix). Semistructured interviews followed the list of indicators presented above, were recorded and transcribed. While only officials in IO headquarters were interviewed, different IO departments and three hierarchical levels were covered. Interview partners were asked to describe the IPA’s working procedures and routines along the six indicators (see Table A6 in the Appendix for interview questionnaire). In line with the “negotiated agreement approach” Table 2 Conceptualization and operationalization of control over an international public administration (IPA) Type of control Direct Indirect Oversight and monitoring Evaluation: existence and reporting line 1: Head of evaluation reports to member states (MS) 0.6: Head of evaluation reports to both MS and the head of IPA 0.3: Head of evaluation reports to the head of IPA 0: No evaluation unit Advisory oversight: existence and reporting line 1: Head of advisory oversight unit reports to MS 0.6: Head of advisory oversight unit reports to both MS and the head of IPA 0.3: Head of advisory oversight unit reports to the head of IPA 0: No advisory oversight unit Governing body (Executive Board) meetings: intensity 1: Board meets at least weekly 0.6: Not weekly, but at least three times a year 0.3: Board meets two times a year 0: Board meets once a year Screening and selection IPA staff appointment: MS’involvement 1: Head of the IPA appoints his/her team in consultation with MS 0: Head of the IPA appoints his/her team unilaterally IPA staff nationality: MS dominance Ratio of (professional) staff nationals of five main donor states relative to total (professional) staff, financed by regular budget and other funds Sanctioning Formal obstacles for sanctioning: voting system in Executive Boards 1: Weighted voting system 0.5: Unweighted majority voting 0: Consensus Threat of sanctions: financial fragility Ratio of earmarked (unassessed) contributions to total budget © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1178 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK Conceiç˜ ao-Heldt, E. d. (2013). Do agents “run amok”? A comparison of agency slack in the EU and US trade policy in the Doha round. Journal of Comparative Policy Analysis: Research and Practice, 15(1), 21–36. Cortell, A. P., & Peterson, S. (2006). Dutiful agents, rogue actors, or both? Staffing, voting rules, and slack in the WHO and WTO. In D. G. Hawkins, D. A. Lake, D. L. Nielson, & M. J. Tierney (Eds.), Delegation and Agency in International Organizations (pp. 255–280). Cambridge University Press. Delreux, T., & Adriaensen, J. (2017). Introduction: Use and limitations of the principal-agent model in studying the European Union. In T. Delreux & J. Adriaensen (Eds.), The principal agent model and the European Union (pp. 1–34). Palgrave Macmillan. Douglas, W. A., Ferguson, J. P., & Klett, E. (2004). An effective confluence of forces in support of workers’rights: ILO standards, US trade laws, unions, and NGOs. Human Rights Quarterly, 26(2), 273–299. Dür, A., & Elsig, M. (2011). Principals, agents, and the European Union’s foreign economic policies. Journal of European Public Policy, 18(3), 323–338. Eckhard, S., & Ege, J. (2016). International bureaucracies and their influence on policy-making: A review of empirical evidence. Journal of European Public Policy, 23(7), 960–978. Eckhard, S., & Jankauskas, V. (2019). The politics of evaluation in international organizations: A comparative study of stakeholder influence potential. Evaluation, 25(1), 62–79. Eckhard, S., & Jankauskas, V. (2020). Explaining the political use of evaluation in international organizations. Policy Sciences, 53(4), 667–695. Ege, J. (2020). What international bureaucrats (really) want: Administrative preferences in international organization research. Global Governance: A Review of Multilateralism and International Organizations, 26(4), 577–600. Elsig, M. (2011). Principal–agent theory and the World Trade Organization: Complex agency and ‘missing delegation’. European Journal of International Relations, 17(3), 495–517. George, A. L., & Bennett, A. (2005). Case studies and theory development in the social sciences. MIT Press. Goertz, G. (2006). Social science concepts: A user’s guide. Princeton University Press. Gould, E. R. (2006). Delegating IMF conditionality: Understanding variations in control and conformity. In D. G. Hawkins, D. A. Lake, D. L. Nielson, & M. J. Tierney (Eds.), Delegation and agency in international organizations (pp. 281–311). Cambridge University Press. Grigorescu, A. (2010). The spread of bureaucratic oversight mechanisms across intergovernmental organizations. International Studies Quarterly, 54(3), 871–886. Hall, N. (2016). Displacement, development, and climate change: International organizations moving beyond their mandates. Routledge. Hawkins, D. G., & Jacoby, W. (2006). How agents matter. In D. G. Hawkins, D. A. Lake, D. L. Nielson, & M. J. Tierney (Eds.), Delegation and agency in international organizations (pp. 199–228). Cambridge University Press. Hawkins, D. G., Lake, D. A., Nielson, D. L., & Tierney, M. J. (Eds.). (2006b). Delegation and agency in international organizations. Cambridge University Press. Hawkins, D., Lake, D. A., Nielson, D. L., & Tierney, M. J. (2006a). Delegation under anarchy: States, international organizations, and principal-agent theory. In D. G. Hawkins, D. A. Lake, D. L. Nielson, & M. J. Tierney (Eds.), Delegation and agency in international organizations (pp. 3–38). Cambridge University Press. Heldt, E. (2017). Regaining control of errant agents? Agency slack at the European Commission and the World Health Organization. Cooperation and Conflict, 52(4), 469–484. Heldt, E. C., Mello, P. A., Novoselova, A., & Serrano Oswald, O. R. (2022). When do international organizations engage in agency slack? A qualitative comparative analysis of United Nations institutions. Global Studies Quarterly, 2(3), 1–15. Honig, D. (2019). When reporting undermines performance: The costs of politically constrained organizational autonomy in foreign aid implementation. International Organization, 73(1), 171–201. Hooghe, L., & Marks, G. (2015). Delegation and pooling in international organizations. The Review of International Organizations, 10(3), 305–328. Huber, J. D., & Shipan, C. R. (2002). Deliberate discretion? The institutional foundations of bureaucratic autonomy. Cambridge University Press. Hustedt, T., & Salomonsen, H. H. (2014). Ensuring political responsiveness: Politicization mechanisms in ministerial bureaucracies. International Review of Administrative Sciences, 80(4), 746–765. ILO. (2016). Compendium of rules applicable to the governing body of the International Labour Office. ILO. (2017). ILO Evaluation Policy: GB.331/PFA/8. ILO. (2018a). Core voluntary funding (RBSA) for ILO development cooperation. https://www.ilo.org/wcmsp5/groups/ public/—dgreports/—exrel/documents/publication/wcms_317826.pdf ILO. (2018b). Staff regulations. ILO. (2019). Independent Oversight Advisory Committee. https://www.ilo.org/public/english/edmas/ioac/index.htm IMF. (2015). Terms of reference for the Independent Evaluation Office of the International Monetary Fund. IMF. (2017). Distribution of professional and managerial staff by nationality. https://www.imf.org/external/pubs/ft/ar/2017/ eng/assets/ar17-web-table-3-1.pdf IMF. (2018). The External Audit Committee (EAC). https://www.imf.org/external/np/eac/eng/eac.htm IMF. (2019a). How the IMF makes decisions. https://www.imf.org/en/About/Factsheets/Sheets/2016/07/27/15/24/How-theIMF-Makes-Decisions IMF. (2019b). Rules and regulations of the International Monetary Fund: N-staff regulations. https://www.imf.org/external/ pubs/ft/bl/rr14.htm IMF. (2019c). Where the IMF gets its money. https://www.imf.org/en/About/Factsheets/Where-the-IMF-Gets-Its-Money © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1185 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. Jankauskas, V. (2022). Delegation and stewardship in international organizations. Journal of European Public Policy, 29(4), 568–588. Kassim, H., & Menon, A. (2003). The principal-agent approach and the study of the European Union: Promise unfulfilled? Journal of European Public Policy, 10(1), 121–139. Kiewiet, D. R., & McCubbins, M. D. (1991). The logic of delegation: Congressional parties and the appropriations process. University of Chicago Press. Kingdon, J. W. (1984). Agendas, alternatives and public policies. Little, Brown and Co. Knill, C. (2001). The Europeanisation of national administrations: Patterns of institutional change and persistence. Cambridge University Press. Knill, C., Bayerlein, L., Enkler, J., & Grohs, S. (2019). Bureaucratic influence and administrative styles in international organizations. Review of International Organizations, 14(1), 83–106. Knill, C., Eckhard, S., & Grohs, S. (2016). Administrative styles in the European Commission and the OSCE secretariat: Striking similarities despite different organizational settings. Journal of European Public Policy, 23(7), 1057–1076. Koremenos, B., Lipson, C., & Snidal, D. (2001). The rational design of international institutions. International Organization, 55(4), 761–799. Lake, D. A., & McCubbins, M. D. (2006). The logic of delegation to international organizations. In D. G. Hawkins, D. A. Lake, D. L. Nielson, & M. J. Tierney (Eds.), Delegation and agency in international organizations (pp. 341–368). Cambridge University Press. Levy, J. S. (2008). Case studies: Types, designs, and logics of inference. Conflict Management and Peace Science, 25(1), 1–18. Long, C. P., & Sitkin, S. B. (2018). Control–trust dynamics in organizations: Identifying shared perspectives and charting conceptual fault lines. Academy of Management Annals, 12(2), 725–751. Mayer, S. (2014). Introduction: NATO as an organization and bureaucracy. In S. Mayer (Ed.), NATO’s post-cold war politics: The changing provision of security (pp. 1–27). Springer. Miller, G. J. (2005). The political evolution of principal-agent models. Annual Review of Political Science, 8(1), 203–225. Mintrom, M., & Norman, P. (2009). Policy entrepreneurship and policy change. Policy Studies Journal, 37(4), 649–667. NATO. (2016a). NATO Secretary General. https://www.nato.int/cps/en/natolive/topics_50094.htm NATO. (2016b). Resource Policy and Planning Board. https://www.nato.int/cps/en/natohq/topics_67653.htm NATO. (2017). North Atlantic Council. https://www.nato.int/cps/ic/natohq/topics_49763.htm Nielson, D. L., & Tierney, M. J. (2003). Delegation to international organizations: Agency theory and World Bank environmental reform. International Organization, 57(2), 241–276. Parízek, M. (2017). Control, soft information, and the politics of international organizations staffing. The Review of International Organizations, 12(4), 559–583. Sobol, M. (2016). Principal-agent analysis and pathological delegation: The (almost) untold story. Governance, 29(3), 335–350. Stone, D. (2008). Global public policy, transnational policy communities, and their networks. Policy Studies Journal, 36(1), 19–38. Stone, R. W. (2011). Controlling institutions. Cambridge University Press. Tamm, H., & Snidal, D. (2014). Rational choice and principal-agent theory. In T. G. Weiss & R. Wilkinson (Eds.), International organization and global governance (pp. 132–143). Routledge. UNCHR. (2016). UNHCR policy on evaluation: UNHCR/HCP/2016/2. UNHCR. (1950). Statute of the Office of the United Nations High Commissioner for Refugees: General Assembly Resolution 428 (V) of 14 December 1950. UNHCR. (2016). Rules of procedure: A/AC.96/187/Rev.8. https://www.unhcr.org/4201eb512 UNHCR. (2019a). Applications for Independent Audit and Oversight Committee (IAOC). https://www.unhcr.org/5a7337d44 UNHCR. (2019b). Global appeal: 2019 update. Vaubel, R. (1996). Bureaucracy at the IMF and the World Bank: A comparison of the evidence. The World Economy, 19(2), 195–210. Weisband, E. (2000). Discursive multilateralism: Global benchmarks, shame, and learning in the ILO labor standards monitoring regime. International Studies Quarterly, 44(4), 643–666. Yi-chong, X., & Weller, P. (2008). ‘To be, but not to be seen’: Exploring the impact of international civil servants. Public Administration, 86(1), 35–51. © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1186 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK APPENDIX The risk of IPA agency slack: Operationalization Table A1 provides an overview of indicators and their operationalization used to measure the extent to which IPAs have developed entrepreneurial administrative routines. The risk of IPA agency slack: Measurement results Based on our interview data (Section A.5 in the Appendix) and the coding scheme (Table A1), we qualitatively assessed the extent to which IPAs are involved in one of the six activities. As qualitative assessments, they naturally remain more a matter of degree rather than a sharp distinction. The final measure for IPA entrepreneurialism is the sum of six indicators’values (total score). Table A1 Indicators and operationalization of the risk of IPA agency slack Indicator Phase Operationalization Issue emergence Policy initiation Regular and active policy initiation from within the bureaucracy: High (1) Occasional policy initiation from within the bureaucracy without clear pattern: Medium (0.5) Usually policy initiation outside the bureaucracy: Low (0) Mapping of political space Policy initiation Regular, active mapping activities (i.e., bureaucratic procedures to early identify member states’preferences toward bureaucratic initiatives, e.g., via consultations or informal working groups): High (1) Occasional mapping activities with no clear pattern: Medium (0.5) No mapping activities: Low (0) Support mobilization Policy initiation Regular and active mobilization activities, active campaigns to gain external support for bureaucratic initiatives: High (1) Occasional mobilization activities without clear pattern: Medium (0.5) No mobilization activities: Low (0) Political anticipation Policy drafting Regular and active identification of bureaucratic policies’political implications to ensure their success: High (1) Occasional anticipation without clear pattern: Medium (0.5) Routinely policy drafting without distinct anticipation of bureaucratic policies’ political implications: Low (0) Strategic use of bureaucratic powers Policy implementation Strategic use of bureaucratic powers, especially in conflict situations with member states (e.g., enter into open conflict or deliberate conflict avoidance in order not to undermine member state delegation): High (1) Occasionally strategic use of bureaucratic powers: Medium (0.5) No strategic use of bureaucratic powers: Low (0) Policy promotion Policy implementation Strong and regular efforts by the bureaucracy promote its agenda and preferred policy implementation (e.g., via mobilization of societal support, the close relationships with stakeholders, interest groups, or nat. administrations): High (1) Occasional efforts to promote its policy agenda and preferred policy implementation: Medium (0.5) No active efforts to promote its policy agenda and preferred policy implementation: Low (0) © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1187 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. IPA control: Operationalization and measurement results Table A3 provides an overview of our principal control measurement results per IO and per indicator. Table A2 Measurement results of the risk of IPA agency slack (data are based on interview data) Indicator Measurement of the risk of IPA agency slack based on entrepreneurialism in their behavioral routines ILO NATO WHO IOM FAO UNEP IMF UNHCR Mapping of the political space Low Medium Medium High High High High High Support mobilization Low Low Medium High High Medium Medium High Issue emergence Low Low Low Low Medium High High High Political anticipation Medium High High High High High High High Strategic use of formal powers Medium Low Low High High Medium High High Policy promotion Low Low High Low Medium High High High Total score 1 1.5 3 4 5 5 5.5 6 Label scores: high (1), medium (0.5), and low (0). Table A3 Operationalization and results of member states’control over IPAs measurement Type of control Indicator Direct oversight and monitoring Evaluation: existence and reporting line 1: Head of evaluation reports to member states (MS) 0.6: Head of evaluation reports to both MS and the head of IPA 0.3: Head of evaluation reports to the head of IPA 0: No evaluation unit IMF: Independent Evaluation Office reports to the Executive Board (1) IOM: Head of Evaluation reports to Inspector General who reports to Director General (0.3) ILO: Director of Evaluation reports to Director General (0.3) FAO: Head of Evaluation Office (OED) reports to Director General and Governing Bodies (0.6) WHO: Evaluation Office reports to Director General (0.3) UNHCR: Head of Evaluation Service reports to High Commissioner (0.3) UNEP: Evaluation Office reports to Executive Director (0.3) NATO: No internal evaluation unit (0) Direct oversight and monitoring Advisory oversight: Existence and reporting line 1: Head of advisory oversight unit reports to MS 0.6: Head of advisory oversight unit reports to both MS and the head of IPA 0.3: Head of advisory oversight unit reports to the head of IPA 0: No advisory oversight unit IMF: External Audit Committee reports to the Board of Governors (1) IOM: Audit and Oversight Advisory Committee reports to the Council (1) ILO: Independent Oversight and Advisory Committee reports to both Governing Body and Director General (0.6) FAO: Audit Committee advises Director General (0.3) WHO: Independent Expert Advisory Oversight Committee reports to Executive Board (1) UNHCR: Independent Audit and Oversight Committee reports to High Commissioner and Executive Committee (0.6) UNEP: No comparable unit identified (0) NATO: Resource Policy and Planning Board reports to the Council as senior advisory body (1) (Continues) © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1188 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK Table A3 Continued Type of control Indicator Indirect oversight and monitoring Governing body (Executive Board) meetings: Intensity 1: Board meets at least weekly 0.6: Not weekly, but at least three times a year 0.3: Board meets two times a year 0: Board meets once a year IMF: In-house Executive Board (1) IOM: Standing Committee meets twice a year (0.3) ILO: Governing Body meets three times a year (0.6) FAO: Council holds at least five sessions in biennium (0.6) WHO: Executive Board meets two times a year (0.3) UNHCR: Executive Committee’s Standing Committee meets three times a year (0.6) UNEP: Committee of Permanent Representatives meets at least four times every year (0.6) NATO: North Atlantic Council meets at least once a week (1) Direct screening and selection IPA staff appointment: MS’involvement 1: IPA head appoints his/her (management) team in consultation with MS 0: IPA head appoints his/her (management) team unilaterally IMF: The Managing Director shall inform the Executive Board at least two weeks in advance of any action to appoint or dismiss any person to or from a position graded equal to or above that of a division chief (1) IOM: The Director General shall appoint the staff of the Administration in accordance with the staff regulations adopted by the Council (1) ILO: Appointments to vacancies of Deputy Director-General, Assistant Director-General and Treasurer and Financial Comptroller shall be made by the Director-General after consultation with the Officers of the Governing Body (1) FAO: The power of appointment of staff members rests with the Director-General (0) WHO: All staff members are subject to the authority of the Director-General and to assignment by him to any of the activities or offices of the World Health Organization (0) UNHCR: High Commissioner appoints his/her team unilaterally (0) UNEP: Acting Executive Director appoints his/her team unilaterally (0) NATO: International staff are either recruited directly by the Organization or seconded by their governments and each appointment is approved by the Secretary-General (0) Indirect screening and selection IPA staff nationality: MS dominance Ratio of (professional) staff nationals of five main donor states relative to total (professional) staff, financed by regular budget and other funds IMF: 0.36 IOM: 0.24 ILO: 0.33 FAO: 0.3 WHO: 0.24 UNHCR: 0.17 UNEP: 0.21 NATO: 0.48 Direct sanctioning Formal obstacles for sanctioning: Voting system in Executive Boards 1: Weighted voting system 0.5: Unweighted majority voting 0: Consensus IMF: Weighted voting (1) IOM: Majority voting (0.5) ILO: Usually consensus (0) FAO: Majority voting (0.5) (Continues) © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1189 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. Overview of diverse general characteristics across IOs from the sample Table A3 Continued Type of control Indicator WHO: Majority voting (0.5) UNHCR: Majority voting (0.5) UNEP: Majority voting (0.5) NATO: Consensus (0) Indirect sanctioning Threat of sanctions: Financial fragility Ratio of earmarked (unassessed) contributions to total budget IMF: 0.51 IOM: 0.98 ILO: 0.42 FAO: 0.61 WHO: 0.72 UNHCR: 0.65 UNEP: 0.79 NATO: 0.12 Measurement is based on data from official IO websites and IO annual reports (2018–2019). Table A4 Diversity in general characteristics across selected IOs FAO ILO IMF IOM UNEP UNHCR NATO WHO Founding year 1945 1919 1945 1951 1972 1993 1949 1948 Policy area Agriculture Labor Finance Migration Environment Asylum, migration Defense Health Location headquarter Rome (IT) Geneva (CH) Washington (USA) Geneva (CH) Nairobi (KE) Geneva (CH) Brussels (BE) Geneva (CH) Number of Member States 191 184 187 132 58 79 28 194 Total budget in billions $ 2.736 0.723 624.5 1.369 0.801 1.992 0.28 4.540 % of non-assessed budgetary contributions 69% 44% 42% 97% 68% 99% n/a 83% No. of staff (general and support staff, excl. short-term consultants) 3069 3011 2421 7699 910 6314 1200 8660 Based on IO websites and annual reports (2018–2019). 1 Source: Policy fields manually coded based on official IO websites (http://www.un.org/en/sections/about-un/fundsprogrammes-specialized-agencies-andothers/index.html); total and decentralized staff (outside HQ) calculated based on UN System Chief Executives Board for Coordination data (UN CEB) (https://www.unsystem.org/content/un-system-hrstatisticsreport-2018) and IO annual reports; budget data calculated based on UN CEB or IO websites (http://www.unsystem.org/con tent/FS-A00-01). © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1190 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK Interview data and questionnaire The interviewees were asked about internal operating procedures in their IOs along the six indicators (see Table A1). Upon completion of the interview process, the interviews were transcribed by the authors. To determine the intensity to which the administration is involved in the six activities, we assessed the consistency of interviewees’statements and gave score values according to Table A1. In rare cases of conflicting statements, we triangulated them with secondary scholarly literature to decide upon the value of the indicator (see Table A2). Table A5 Overview of conducted interviews across IOs and interviewee positions IO Senior managers Middle managers Analysts, technical officers Total NATO 2 5 5 12 FAO 3 1 19 23 ILO 1 3 12 16 UNEP 3 4 15 22 IOM 1 3 12 16 UNHCR 0 4 12 16 IMF 1 6 8 15 WHO 4 4 3 11 Total 131 Please note that the “Senior managers”category refers to officials at the level of Secretaryor Director-General, directors, their deputies, and similar. The “Middle managers”category refers to team leaders, heads of units, and similar. The “Analysts and technical officers”category refers to various specialists at the technical level of administrations such as policy specialists or technical officials. We identified the respondents through IO websites and primary documents such as annual reports. Table A6 Interview questionnaire used for semi-structured interviews with IPA staff Phase Indicator Open question Closed question Test questions Initiation (Please describe how your organization is involved in the very first steps of policy development) Issue emergence When you are thinking about the angle of a new policy, is the administration involved in the identification of new issues and solutions, and if so, how exactly? Did your administration try to actively identify issues and propose solutions in the case of policy X, or was this left to other actors? Have you made the experience that your administration is working on issues before they receive member states’ attention and support or are these preconditions before your administration becomes involved? Mapping of political space When you are thinking about the first steps in the development of a new policy, do you consider member states position on that policy, why? How was member states’ support for the development of policy X organized and how were the interests of the member states addressed? Does the administration regularly consider potential political “red lines”and how to bypass them when initiating own policies? Support mobilization When you are initiating new policies, do you seek support from actors other than member states? How important was the external support for policy X, and how was it secured? Does the administration regularly highlight the extent of external support toward its new policy to member states and if so, why? (Continues) © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd. 1191 PRINCIPAL CONTROL AND RISK OF AGENCY SLACK V. Jankauskas et al. Table A6 Continued Phase Indicator Open question Closed question Test questions Drafting (Could you describe a typical policy drafting process?) Political anticipation When you are drafting new policies, do you consider member states’ interests toward that policy, why? Did member states’ interests affect how policy X was drafted and if so, how? Are you aware of any instances when knowing member states’interests helped the administration to draft a policy so it favors administrative preferences? Implementation (Could you please name the most common outputs your organization produces?) Strategic use of bureaucratic powers How does the administration defend its interests when implementing policies? Do you make use of bureaucratic powers to push for favorable policy implementation or do you rather avoid risking backlash from the member states? Is that a general approach applied to many policies, or does it vary between output and/or unit? Policy promotion How does the administration promote its policies during implementation phase? In the case of policy X, did you try to promote your agenda in implementation and if so, why and how? Does your administration decide by itself on how to promote policy implementation? © 2023 The Authors. Regulation & Governance published by John Wiley & Sons Australia, Ltd.1192 V. Jankauskas et al. PRINCIPAL CONTROL AND RISK OF AGENCY SLACK