ANALYTICAL Connecting International Relations and Public Administration: Toward A Joint Research Agenda for the Study of International Bureaucracy

The recent debate on administrative bodies in international organizations has brought forward multiple theoretical perspectives, analytical frameworks, and methodological approaches. Despite these efforts to advance knowledge on these actors, the research program on international public administrations (IPAs) has missed out on two important opportunities: re- ﬂection on scholarship in international relations (IR) and public administration and synergies between these disciplinary perspectives. Against this backdrop, the essay is a discussion of the literature on IPAs in IR and public administration. We found inﬂuence, authority, and autonomy of inter- national bureaucracies have been widely addressed and helped to better understand the agency of such non-state actors in global policy-making. Less attention has been given to the crucial macro-level context of politics for administrative bodies, despite the importance in IR and public admin- istration scholarship. We propose a focus on agency and politics as future avenues for a comprehensive, joint research agenda for international bu- reaucracies. El reciente debate sobre organismos Le récent débat sur les organes administratifs des organisations in- ternationales a mis en avant plusieurs perspectives théoriques, cadres analytiques et approches méthodologiques. Malgré ces efforts pour faire progresser la connaissance de ces acteurs, le Programme de recherche sur les administrations publiques internationales a manqué deux oppor- tunités majeures : une réﬂexion sur les recherches en relations internationales et administration publique ainsi que sur les synergies entre ces perspectives des disciplines. Cet essai s’appuie sur cette toile de fond pour établir une discussion au sujet de la littérature abordant les admin- istrations publiques internationales dans les domaines des relations internationales et de l’administration publique. Nous avons constaté que l’inﬂuence, l’autorité et l’autonomie des bureaucraties internationales avaient été largement abordées, ce qui permettait de mieux comprendre le pouvoir de tels acteurs non-étatiques dans l’établissement des politiques internationales. Toutefois, malgré son importance dans les recherches en relations internationales et administration publique, une moins grande attention a été accordée au contexte macro des politiques des organes ad- ministratifs alors qu’il est crucial. Nous proposons de mettre l’accent sur le pouvoir et les politiques comme pistes futures pour un programme de recherche conjoint complet sur les bureaucraties internationales.

Le récent débat sur les organes administratifs des organisations internationales a mis en avant plusieurs perspectives théoriques, cadres analytiques et approches méthodologiques. Malgré ces efforts pour faire progresser la connaissance de ces acteurs, le Programme de recherche sur les administrations publiques internationales a manqué deux opportunités majeures : une réflexion sur les recherches en relations internationales et administration publique ainsi que sur les synergies entre ces perspectives des disciplines. Cet essai s'appuie sur cette toile de fond pour établir une discussion au sujet de la littérature abordant les administrations publiques internationales dans les domaines des relations internationales et de l'administration publique. Nous avons constaté que l'influence, l'autorité et l'autonomie des bureaucraties internationales avaient été largement abordées, ce qui permettait de mieux comprendre le pouvoir de tels acteurs non-étatiques dans l'établissement des politiques internationales. Toutefois, malgré son importance dans les recherches en relations internationales et administration publique, une moins grande attention a été accordée au contexte macro des politiques des organes administratifs alors qu'il est crucial. Nous proposons de mettre l'accent sur le pouvoir et les politiques comme pistes futures pour un programme de recherche conjoint complet sur les bureaucraties internationales.

Introduction
Under the umbrella term of international public administrations (IPAs), 1 research disciplines and subdisciplines have shared lessons and challenges of studying IPAs but have remained separated in their analytical endeavors. Recently, scholars discuss how to conceptualize the role of international bureaucracies from a public administration (PA) perspective (Bauer, Knill, and Eckhard 2017;Bauer et al. 2019;Christensen and Yesilkagit 2019). Hence, concepts such as bureaucratic autonomy, administrative behavior, and the politics-administration nexus have come under renewed scrutiny in the context of IPAs. This paper seeks to contribute to IPA scholarship by identifying and assessing reference points in both international relations (IR) and PA to ascertain what is needed to encourage an interconnected, scholarly debate and research agenda for understanding IPAs. Despite the common interest in IPAs, scholars have been occupied with "determin [ing] to which disciplinary tradition a study may best be attributed" (Bauer and Ege 2013, 141), rather than to integrating PA and IR perspectives into a coherent research agenda. While we do not claim to entirely achieve this aim in this article, we lay out avenues toward it and identify reference points of intersections in order to encourage critical reflection of and between scholarly disciplines.
From an IR perspective, interest in IPAs reflects the shift of research from whether international organizations (IOs) matter for global governance toward how they matter (e.g., Barnett and Finnemore 2004;Johnson 2013Johnson , 2014. Frameworks have been developed to link delegation theory and sociological institutionalism to address IPAs' role in global public policy (e.g., Betsill and Corell 2001;Arts and Verschuren 1999) from a distinct IR perspective (e.g., Weinlich 2014). The prevailing opinion among IR scholars has remained, however, that IR theories and concepts are of limited use for the study of IPAs (Biermann 2017).
Despite efforts of IR scholars to explore and theorize the role of IPAs for international and transnational policy-and law-making (already Liese and Weinlich 2006), PA scholars suspect a "failure" in properly tackling IPAs due to their allegedly "fuzzy conceptualization of bureaucratic influence and disregard for the relevance of carefully establishing bureaucratic policy preferences" (Ege, Bauer, and Wagner 2020, 552). Accordingly, their research suggests conceptual perspectives from domestic PA research for understanding IPAs' influence at the international level (e.g., Bauer et al. 2019). Recent contributions have been criticized by fellow PA scholars, especially regarding "under-developed" theoretical foundations, neglect of existing findings in PA scholarship, and limited acknowledgment of politics to understand IPAs (Christensen and Yesilkagit 2019, 947).
Contrasting promises and pitfalls in IR and PA perspectives in IPA scholarship show the necessity for more collaboration to advance knowledge and to provide crucial focal points to build resilient bridges, yet the pledge for acknowledgment and use of theory pluralism is hardly new. In an overview article, Bauer and Weinlich (2011) identified concept cross-fertilization as one of the key avenues for future IPA research in IR (see also below). The debate has not come very far since then, partly because of the dynamics of how scholarly debates and the corresponding theoretical foundations and methodological orientations progress in the discipline. Some foundational assumptions do not travel well across contexts. Another reason is the lack of thorough engagement in identifying the most suitable points for bridging.
With this essay, we promote interconnected research by stressing two key concepts-agency and politics-which we regard as suitable bridges for linking research on IPAs to a joint research agenda. By reflecting on these focal points of reference, we facilitate broad and mutually satisfying conversations across multiple scholarly communities. The essay first summarizes how IR and PA scholarship has so far approached IPAs and their role in global public policy. We discuss the disciplines' key questions and how they have proceeded in answering them and take stock of achievements and challenges. Afterward, we discuss conceptualizations of IPAs' agency and the immanent relationship with politics as bridges that connect IR and PA research agendas and as potential starting points for identifying and promoting mutually beneficial conversations. Lastly, we discuss avenues for bridging of major explanatory perspectives for comprehensive future research on IPAs.

The IR Perspective: IPAs as IOs or Non-state Actors
The emergence and development of the debate on IPAs can be summarized as a shift from scholars asking what PA can do for IPA research to PA scholars asking what IPA research can do for PA. What has been largely neglected is what IR has to offer to this debate. IR is regularly mentioned as one discipline that can inform IPAs, alongside PA and organizational sociology, but oftentimes without further elaboration on the research this interest has already produced. Almost a decade after Bauer and Weinlich (2011) proposed three research avenues 2 to bring IPA research from an emerging subfield to the core of IR, some progress has been made, while challenges remain. In this section, we describe the accomplishments and the persisting challenges in IPA research from an IR perspective.
The discipline's turn from studying IR as interactions between nation states to global governance reflects that more actors matter for governance processes.
Besides non-governmental actors, such as non-governmental organizations (NGOs), corporations, and transnational networks, scholars have explained and analyzed the autonomous role of IOs for global governance (Koch 2009;Zürn 2018). International organizations are no longer treated as a forum used by powerful states to achieve their interests (see Dingwerth, Kerwer, and Nölke 2009, 16-18). Instead, scholarship has opened the black box of IOs to understand and explain their authority in global governance. With the interest in entities that make an IO comes the question of how to approach IPAs. Most early IR scholarship of IPAs looked at them as nonstate actors (see Bauer and Weinlich 2011), but recently there have been acknowledgments that they share more features with governments than NGOs or multinational companies.

IPAs as Actors in Global Governance
The interest in IO entities has brought IPAs as "governance actors" to the attention of IR scholars (Herold 2019). Barnett and Finnemore's (1999) work on IOs as bureaucracies opened the door to studying IOs as actors able to formulate policies, even if these policies are contrary to the interest of the majority of the IO's member states. By confronting rationalist perspectives on IOs, these scholars demonstrated the static nature that principal-agent approaches ascribe to IOs is insufficient to explain the partially autonomous behavior of IOs. Additionally, IOs differ in their degrees of autonomy and their organizational environments, demanding varying behaviors and responsiveness (Barnett and Finnemore 1999, 703 ff.). The analysis of IO bureaucracies also addresses their abilities to exercise authority through setting "rules for the world" (Barnett and Finnemore 2004). The production and possession of knowledge through compiling, interpreting, and giving meaning to information is therefore regarded as crucial for them to exert influence on international policy-making (Barnett and Finnemore 2004, 6-7).
Following Barnett and Finnemore's pioneering work, IR scholars have made important contributions to the study of international bureaucracies by exploring the internal dynamics and functions of IOs (Biermann and Siebenhüner 2009;Chwieroth 2013;Park and Vetterlein 2010;Reinalda and Verbeek 2014;Hawkins 2006). Explanations for bureaucratic autonomy come from principal-agent, constructivist, and psychological perspectives. Those who follow a rational choice, understanding of principals and agents assume information asymmetries and conflicts of interest between IPAs and their IOs' member states (da Conceição-Heldt 2013). Another strand of the literature is built on the Bourdieu-inspired framework for organizational analysis (Emirbayer and Johnson 2008) and emphasizes social conditions for the (re-)production of power relations in and between organizations and their fields to explain organizational change (Schwindenhammer 2017;Vetterlein and Moschella 2014). Sociological institutionalism has proven well suited to explain organizational "pathologies" (Barnett and Finnemore 1999), i.e., the unintended behaviors of organizations. These studies have in common that they ask what the bureaucratic apparatuses of IOs contribute to the IO's role in global governance. Several research projects have used this question as the starting point to build datasets on IOs and IO bodies, allowing for comparative studies on the functions of IPAs in IOs (Tallberg et al. 2013;Sommerer and Tallberg 2015;Hooghe and Marks 2015;Hooghe, Marks, and Lenz 2019).

Autonomy, Influence, and Authority
Interest in IPAs is connected to key questions in the disciplines of autonomy, influence, and authority in global governance. Research asks for the aim at the level of IPAs' autonomy, how this autonomy affects their authority, and how actors in the IO influence its agenda and decision-making. Findings demonstrate IPAs' qualities as "managers of global change" (Biermann and Siebenhüner 2009) and policy entrepreneurs (Oksamytna 2018). Many IR scholars interested in non-state actors are familiar with the application of "modest" proposals and attempts to study IPAs (Ege, Bauer, and Wagner 2020). Despite state-centric foundations of the discipline, roles of nonstate actors, such as NGOs, epistemic communities, multinational corporations, or bureaucracies in IOs, have produced analytical frameworks to address how to measure such actors' influence on policy-and law-making 3 . Influence can be approached as a zero-sum-game, whereby state and nonstate actors compete an outcome. For the study of IPAs' influence, it may be useful to understand nonstate actors' influence as "political influence" and expect "goal-achievement and political influence do not coincide, as Player A might achieve his goal due to the intervention of others, due to external events, or due to autonomous developments" (Arts and Verschuren 1999, 413). Influence is possible even when actors are not present in decision-making to exert influence directly, such as in intergovernmental deliberations where bureaucrats are not present.
IR offer a range of research designs and methods for detecting political influence (see, for example, Dür 2008; Betsill and Corell 2008;Arts and Verschuren 1999;Cox and Jacobson 1973). Research projects on transnational actors and their influence in IOs have emphasized structural explanatory factors, such as formal access or political opportunities, and have included less visible activities, access points, and resources for empirical analyses (e.g., Tallberg et al. 2013;Betsill and Corell 2001). A methodological solution to study the informal activities of bureaucrats lies in ethnographic methods that have provided insights into the bureaucratic authority of, for example, ASEAN (Nair 2020). A focus on formal and informal practices of IPAs also helps to explain variations in de facto and de jure authority (Busch and Liese 2017).
Studying IPAs means engaging with entities set up by sovereign nation states. Hence, many explanations for an international bureaucracy's autonomy or influence are derived from explanations for IO agency. Emphasizing distinct practices of bureaucratic entities, scholars have conceptualized IPAs as actors themselves. Frameworks for the study of IPAs can contribute to theoretically and practically significant findings, e.g., integrative studies combining delegation theory and constructivism (Weinlich 2014). Furthermore, international political economy scholarship has used data on the bureaucratic input and output of IOs to assess IOs' role in global public policy. These studies in international political economy have also focused on a particular subset of IOs, namely mostly those with mandates in global financial and economic policy, such as the WTO, the IMF, the World Bank, the UN, the EU, and others, to address political-economy issues, including how they are organized and governed (Patz and Goetz 2019;Vaubel, Dreher, and Solyu 2007).

Persisting Challenges in IR Research on IPAs
From an IR perspective, the research program on IPAs therefore addresses one of the discipline's core questions: Who can act in global governance? Similar to early studies of IOs as actors in global governance, IR scholarship on IPAs has had a focus on units of the IO and its bureaucracy when analyzing IPAs as actors. Several characteristics of these actors have been acknowledged to varying degrees.
First, boundaries between the IO and the IPA can be blurry. It is often not explicitly acknowledged where an IO ends and an IPA starts (Bauer and Weinlich 2011, 251). As all IOs are different in terms of membership and purposes, all IOs consist of different bodies with different functions that consist of and interact with different actors and thus do not exert influence on international politics and international law equally. This is a conceptual problem that follows empirical reality. For example, while the ozone secretariat, as an administrative body, employs seventeen staff members who represent seven countries, it is housed by UNEP, which has been subject to studies on its bureaucracy (e.g., Bauer 2009;Johnson and Urpelainen 2014, 195ff.).
Second, in comparative and large-N projects, researchers have often treated IOs as actors in global governance as homogenous (Hooghe, Marks, and Lenz 2019;Zürn et al. 2015). However, the secretariat is composed of many individuals (Littoz-Monnet 2017; Yi-Chong and Weller 2008; see also below), who are highly trained and often share the same educational backgrounds as other key players in their policy sector, such as global finance and banking or global health. Individual choices in a given bureaucratic culture have different effects on domestic policies (Barnett 2002). As Hanrieder (2015) demonstrated, a WHO reform was possible because the culture of IO agency was established with an entrepreneurial bureaucrat, as opposed to an impersonal civil servant. Under favorable conditions, individual bureaucrats can successfully advocate for innovation and become policy entrepreneurs in rather risk-adverse settings like peacekeeping (Oksamytna 2018). Bureaucratic entrepreneurship may further develop organizational resources to expand into new domains (Littoz-Monnet 2020). Politics of staffing in IOs (Parízek 2017) gives further importance to individual agency (Bode 2015), a research program which has been strengthened and expanded through highlighting emotions and microsociological perspectives to explain bureaucrats' power (Nair 2020).
Third, IPAs interact with other IPAs and actors, often on joint policy programs, posing challenges for comparing IPAs' influence across IOs and policy fields. Overlap of policies across IOs and regime complexity (Finnemore 1996;Gehring and Faude 2014) support their policy claims (by adding legitimacy) and also undermine their authority (e.g., when these claims come from several IPAs). Furthermore, IOs often "outlive the bureaucrats and partnerships" (Johnson 2016, 759) that were shaping the programs at one point in time.
Fourth, individuals working in IPAs are often embedded in their own networks and epistemic communities, some loosely connected to the IO, resulting in exclusive networks that can shape policy outcomes autonomously (Ban, Seabrooke, and Freitas 2016;Tsingou 2014). Empirical studies have shown how important the composition of IPAs is for an IO's recognition as unbiased by domestic decision-makers (Parízek 2017;Heinzel et al. 2020).
IR studies of IPAs have alternated between two poles: bureaucracies as synonymous with IOs or as autonomous nonstate actors. Both approaches include risks of overlooking dynamics in the bureaucracy and with their organizational environment, e.g., relationships with other entities or to what extent IPAs as administrative bodies should be analytically differentiated from political bodies (Elsig 2010). While progress has been made, we have found a strong connection to PA research and the identification of mutually relevant focal points for further empirical analyses may renew some empirical answers and dispose conceptual perspectives.

The PA Perspective: Under-Theorized, Over-Operationalized, and Non-Politicized
The recent PA debate on IPAs has produced many contributions in a short period of time, mostly in agreement and support with each other (special issue of JEPP 2016, 23(7); Bauer, Knill, and Eckhard 2017;Knill and Bauer 2018). Some contestation of the use of existing theories and concepts from PA scholarship at domestic level has emerged (Christensen and Yesilkagit 2019;Bauer et al. 2019).

(Mostly Old) Problems of Current Theoretical Perspectives
Most recent PA studies on IPAs have stressed the distinct nature of IPAs by assigning them the status sui generis or by suggesting novel typologies and operationalizations Downloaded from https://academic.oup.com/isr/advance-article/doi/10.1093/isr/viaa097/6109791 by guest on 29 January 2021 of PA key concepts to fit IPAs in (Bauer and Ege 2016;Ege 2017), yet follow-up on this key observation is sparse. Instead, the literature often revokes this understanding when selecting and applying theoretical and conceptual arguments from PA scholarship without discussing the consequences of such a concept transfer. To some extent, this scholarly strategy is meaningful and reasonable: PA research benefits from the application and adaption of conceptual perspectives from adjacent fields and the contribution of its findings to these fields; however, some basic assumptions of national PA research may not hold for IPAs and need to be assessed to determine the consequences of a concept transfer. So far, PA contributions in IPA research have had a focus on delegation theory (with the exception of Christensen and Yesilkagit 2019; see below).
The two basic claims of delegation theory have been discussed among PA scholars. The asymmetry of information between the principal and the agent is a claim widely regarded to be the reason why political executives delegate tasks to the permanent bureaucracy. In contrast, conflict of interest between principals and agents is a claim that awaits empirical verification. In numerous surveys of public officials in and across domestic bureaucracies, researchers have studied the motivation of these public employees (Christensen and Laegreid 2020; Van de Walle and Jilke 2014; Van de Walle, Steijn, and Jilke 2015) and general willingness to accomplish demands from their superiors (Aberbach, Putnam, and Rockman 1981). Few studies have shown deviant behavior as "critical loyalty," and the studies have limited external validity (e.g., Ebinger and Jochheim 2009). There is mixed evidence as to whether bureaucrats favor maximizing utility over following preferences of their political masters.
Public administration scholars have suggested alternative concepts to explain relationships between political executives and permanent officials. Among these, stewardship theory and its emphasis on prosocial and intrinsic motivation has gained scholarly attention (Davis, Schoorman, and Donaldson 1997;see Schillemans 2013). Some studies have applied the concept of public service bargains (PSB; Hood and Lodge 2006; see, e.g., special issue of Public Policy and Administration 2011, 26(2)), also because this concept captures empirical variation in politicianbureaucrat relations. In the PA debate on IPAs, Christensen and Yesilkagit (2019) proposed to apply the PSB perspective as an alternative conceptualization for studying IPAs' autonomy and behaviors. It is meaningful to depart from the dominance of delegation theory if national PA scholars have collected enough evidence to question the applicability of principal-agent approaches to national bureaucracies. One may argue IPAs differ from national bureaucracies enough to apply delegation theory, yet the following two cautions may substantiate critiques of delegation theory.
National administrations face a single political principal, namely a cabinet minister exercising responsibility for bureaucratic apparatuses vis-à-vis parliament, yet the political principal is a single actor with identifiable and often predictable preferences for permanent bureaucracy. Public administration concepts and theories on bureaucratic autonomy, behavior, and influence are based on this predictability of the political principal's preferences. In contrast, IPAs serve a composite political principal (see Johnson 2013) who formulate policy preferences that can be flexible, prone to contestation, and less predictable. In contrast to multiple principals who lead to preference heterogeneity and thus enable agents' moral hazard, composite political principals may not yield similar behavioral effects. They often express preference dispersion and thus add pacifying to IPAs' core activities: IPAs are likely to engage in preference building for their political principal. Any conflict in the composite political principal limits achieving the IO's organizational goals. Scholars are unsure whether this coincides with moral hazard and agency shirking or whether it motivates an IPA as an agent to align preferences to their principal's preferences in order to fulfill the pacifying role.
Principals may turn the delegation at a domestic level into a multiple-principal setting (Kiewiet and McCubbins 1991, 27). Although the IPA debate also identifies multiple principals (see Lyne, Nielson, and Tierney 2006), relevant differences exist. In national settings, bureaucratic organizations engage with non-political principals, such as interest groups, courts, or the media. International PAs may develop similar interactions with interest groups and other non-governmental actors engaged in policy areas covered by their IO's mandate, but arguably courts or media actors are of less concern. In contrast, donors become crucial principals in their delegation settings.
IPAs as agents have differences from national bureaucracies. Most IPAs recruit staff with short-term contracts and job rotation, thus staff experience less permanency in their positions than national officials. The classic bureaucratic advantage over political executives-their long-term orientation for gaining expertise-has been absent in IPAs: IPA officials cannot wait for leadership turnover to prepare policy shifts for their preferences nor can they turn into grey eminences sitting on arcane knowledge upon which politicians may depend. Moreover, IO member states also rely on national administrations as agents supporting their activities. These shadow bureaucracies (see Dijkstra 2015) contribute to a multiple-agent setting with potential consequences of free riding or sabotage (Huber and Shipan 2000). The parallel or shadow bureaucratic support structure, particularly the resulting interactions between national administrations and IPAs, has been neglected, and there is a need to substantiate claims about IPAs' autonomy and influence from a delegation theory perspective.
Delegation theory has an emphasis on contract design to reduce moral hazard ex ante (Kiewiet and McCubbins 1991). In comparison to national administrations, most IPAs are created and designed alongside policy programs. For example, UNEP is the UN environmental program and at the same time an administrative body assigned to exercise it. Consequently, the contract design of these IPAs is likely to entail compromises over policy-and structure-related motives. We know very little about the effects of these agency designs for the delegation relationship and the agent's behavior therein.

Insufficient Attention to Operationalization?
Another caveat on recent PA contributions to the IPA debate is the emergence of novel operationalizations for classic PA concepts, such as autonomy or influence. Generally, it is necessary to critically reflect on long-standing standards in operationalization and to innovate measurements due to the availability of novel methods of data collection and analysis. It is likewise necessary to articulate advantages of reinventing these measures. Recent PA studies on IPAs with an explicit aim to introduce novel operationalizations have had little emphasis on the findings and "settlements" of debates among domestic PA scholars over the measurement and operationalization of concepts such as bureaucratic autonomy (e.g., Ege 2017). Their contributions thus miss out on acknowledging the research already engaged in these operationalization debates, which is a necessity of the scientific method: Weak measurements or difficulties in applying operationalizations provide sound arguments for novel measures yet these contributions to the IPA debate would benefit from engaging more strongly with the existing literature and show the transferal weakness they seek to overcome. Given that several seminal contributions on how to operationalize and measure key research interests in the IPA debate such as bureaucratic autonomy and influence have been applied to EU agencies in the past (e.g., Groenleer 2009; Wood 2018; see also Rittberger and Wonka 2011), we know administrative bodies at supranational level can be successfully and convincingly studied with measures developed in domestic PA debates.
The lacking application of existing operationalizations may be related to limits in empirical data necessary to follow the measures available for domestic public administrations. Several seminal contributions in national PA debates on bureaucratic autonomy and influence have been based on attitudinal data collected via surveys of bureaucrats (see e.g., Verhoest et al. 2012; George, Van de Walle, and Hammerschmid 2019). To build on a similar richness of data to study IPAs' autonomy and influence will take time and effort. One may therefore argue we must be patient and collect more attitudinal data on IPAs until we can truly follow and test operationalizations developed for these bureaucratic organizations at domestic level. However, there are alternative data available for IPAs that outperform the data on their domestic counterparts in terms of validity and availability (see below). More importantly, more sensible applications of existing operationalizations for ongoing IPA research would enable retesting theoretical arguments, which eventually may inspire theory development for both domestic and IPA research.

The Neglect of Politics
Many PA scholars in the IPA debate have addressed organizational interactions with the IO and its member states, etc., as they analyze IPAs' bureaucratic autonomy, influence, and administrative capacity. Few scholars have stressed the need to add a distinct explanatory perspective on politics (Christensen and Yesilkagit 2019), yet the proposed PSB approach may be less suitable for grasping the distinct politicsadministration nexus in IOs (see also below). The key claim is relations between politicians and bureaucrats are bargains over competency, loyalty, and rewards. All three aspects require a careful reassessment in the context of IPAs. Competency may be the same for IPAs and national bureaucracies but conditions such as the high staff rotation and turnover may limit the generation of competencies such as organizational skills. IPAs are headed by composite political masters, so there is no clear recipient of loyalty with whom bureaucrats could bargain. IPAs may have loyalty, but it is likely to the organization and its goals, rather than any politicians. Lastly, rewards for bureaucratic careers in international bureaucracies vary considerably across IPAs (arguably more than across national ministries or agencies) because of differences in statutory legislations and frameworks for staff members. In national bureaucracies, a single national civil service code often regulates extrinsic rewards, next to other intrinsic rewards available (Dahlström and Lapuente 2017). It is difficult to establish which features may be regarded as rewards genuine for all IPA staff members to be seen as an exchangeable mean in a PSB.
The PSB approach has been argued to be "sensitive to the political context in which civil servants operate" (Christensen and Yesilkagit 2019, 950). Politicians and thus politics is part of all eight ideal-typed PSBs (Hood and Lodge 2006, 21), yet political context features have not got much attention. Hood and Lodge focused on party turnover in government to explain why bargain types emerge and why regular political turnover is beneficial for adopting agency-type PSBs (Hood and Lodge 2006, Chapter 3.2). This is a valid argument but not a conceptualization of how politics matters for bureaucratic behavior.
Party competition as a crucial context feature of bureaucratic organizations at domestic level has not been conceptualized, instead referred to as part of the bargaining context and as relevant for understanding changes in the rewards system for the civil service that trigger PSB changes (Hood and Lodge 2006, 38, 45). The variation in government types and single versus multi-party governments at domestic level correspond to distinct PSB dynamics and bargains over competency, loyalty, and rewards. The PSB approach is nevertheless a fruitful perspective to capture the variety of politics-bureaucracy relations and shifts therein (Hood and Lodge 2006, Chapter 7.3). It also provides an alternative conceptualization of bureaucratic behavior to delegation theory; however, its immediate applicability to IPAs requires answering similar questions as noted above for delegation theory (i.e., Who is the politician in that bargain? What consequences arise from them being a composite actor?), and it needs further conceptualization of the PSB key means of competency, loyalty, and rewards. Moreover, PSB offers a nuanced view of why politics is interlinked with bureaucratic behavior but is limited in explaining how politics matters.
The overall conceptual and empirical neglect for the political context of IPAs is striking because it deprives PA scholarship of the possibility to benefit from advances in IR. The PA debate at domestic level has often stressed bureaucracies operate in political environments (March and Olsen 1989) without providing further considerations of the peculiar nature and effects of politics for bureaucracies, but this lack of interest does not release a PA perspective on IPAs from reflecting on the nature of IPAs' political environment and its consequences for bureaucratic autonomy, administrative behavior, and influence in global policy-making.

Future Avenues: Agency and Politics
The current IR debate on IPAs has not achieved its full explanatory strength and potential yet, whereas PA has pushed conceptual perspectives and novel operationalizations but has received critical feedback. Both research communities share an empirical interest and, to some extent, theoretical perspectives. For example, delegation theory and sociological institutionalist arguments have been applied to administrative actors at international and domestic levels. The comparative assessments of dominant perspectives in IR and PA research in debates over IPAs have shown various points of reference where scholarly debates intersect and may provide starting points for connecting research questions and supporting mutual conversations. Across both scholarly communities, themes of agency and politics reoccur, and we argue that these are crucial points to explicate how IR and PA approaches may be reconciled and to discuss with what consequences for future research under a joint agenda.

IPAs' Agency for Authority and Autonomy
To achieve bureaucratic influence, either ascribed or self-claimed, international bureaucracies need agency. We observed IR and PA perspectives differ in how they define agency, and we argue this contributes to the weak mutual recognition of scholarship on IPAs. Key questions for the study are: How is agency gained? What happens when agency is contested? Can agency be lost or transformed? Does it matter who contests agency?
The field of IR has approached agency with a focus on the dualistic relationship of structure and agency (Wendt 1987(Wendt , 1999. Depending on research interests and theoretical standpoints, agency concepts have ranged from political (Zanotti 2017;Arts and Verschuren 1999) to strategic (Littoz-Monnet 2017) to complex (Elsig 2010) to moral (Erskine 2003) agency and the observation of multiplicity of agency as possessing several agencements (Bueger 2019).
A focus on performance has broadened theorizations of agency in IR (for an overview, see Braun, Schindler, and Wille 2019) and empirically exposed international bureaucratic performance as gendered and class dependent (Nair 2020). PA and public management scholars have discussed bureaucratic performance inter alia as goal attainment, they acknowledged goal ambiguity as a key feature of public organizations, and proposed several performance taxonomies, indicators, and measures (Meyer and Zucker 1989;Kearney and Berman 1999;Lapuente and Van de Walle 2020). The IR debate focuses on role identification and attainment by bureaucratic actors at the international level, applying the notion of performance to a well-known empirical phenomena discussed as politicization and representative bureaucracy for domestic administrative actors (Putnam 1973;Aberbach, Putnam, and Rockman 1981;Meier 1975Meier , 2019. Whether or not IPAs are understood as administrative units serving IOs or as the sum of their individual civil servants, IPAs' agency depends on principals. While domestic bureaucracies can make use of access channels to exert policy influence, an IPA is more limited: There is no separation of powers at the international level; hence, there are no other enforcement mechanisms that IPAs could make use of (see below). Even where IPAs are autonomous by institutional design, budget constraints by their principals affect agency, especially when earmarked contributions predetermine which policies are supported or which and how many positions are created (Patz and Goetz 2019). Yet, IPAs may engage in generating expert knowledge and broker their expertise in order to excel their policy influence (see Dellmuth, Gustafsson, and Kural 2020).
Moreover, IPA staff may show more autonomous attitudes than their domestic counterparts and IOs compete for qualified bureaucrats to guarantee their vitality (Gray 2018). International officials are well educated and highly trained and need to adapt to new environments quickly by moving regularly between the IO's headquarters, field offices, and other IOs. Accordingly, executive heads and top officials in IOs are vividly studied (Hall and Woods 2018;Bode 2015). Individual sacrifices that come with careers in IPAs can enhance an individual's perception of having a right to have a say in the matters of the IO in which they work, instead of unquestioningly carrying out administrative tasks. Studies have shown IOs have effects on value acquisitions of individuals, triggering the socialization of IPA staff at the international level (Checkel 2005;Murdoch et al. 2019). The tension between individual ambitions and constraints in acting on behalf of the bureaucracy should be considered in future IR scholarship on IPAs to gain knowledge on who really acts.
Public administration research has regarded IPAs as a population of bureaucratic organizations and has often turned agency into an equivalent of autonomy. Many PA contributions have had a focus on IPAs' bureaucratic autonomy and apply a very classic question in PA research at domestic onto the international level. More importantly, this application depicts their understanding of agency in the context of IPAs so far: Agency is expressed via autonomy, or rather, when autonomy is present. As a consequence, agency is relational: Either PA scholars apply the principal-agent approach and implicitly take the principal into account (along with the overall delegation arrangement providing the agent with their discretion to act upon their agency) or they follow alternative approaches, such as the PSB perspective, referring to a composite arena of politicians toward which bureaucratic actors articulate their agency and thus exert influence in policy-making. The PA literature on domestic bureaucracies has made a similar step into discussing the agency of bureaucratic actors in a relational manner, yet in an opposite conceptual interest, namely for bureaucratic accountability and reputation. This debate dismantles actors toward which bureaucracies are accountable and reputable-thus implicitly also toward which they seek to exert their agency (see Bovens 2010;Busuioc and Lodge 2016).
In sum, IR and PA scholars have promoted different terms when stressing IPAs' agency, which is also empirically pursued when IR authors have stressed de facto and de jure authority (Busch and Liese 2017) and PA authors have discussed de facto and de jure autonomy (Christensen and Yesilkagit 2019;Bauer et al. 2019). These terminologies reflect common ground and can be partly related to disciplinary debates on other actors in global governance: The IR tradition has had a focus on studying how sovereign nation states claim and recognize authority at the international level, whereas the PA tradition has regarded bureaucracies as actors at the final step of a democratic chain of delegation that comes with certain autonomy. Therefore, these two conceptual and partly terminological foci on agency offer reconciliation as the common prerogative is shared by IR and PA scholars: IPAs are actors in their own right and they create, maintain, use, and lose their agency vis-à-vis their IO and the member states organized in and other principals crucial to their mission and mandate.

Politics and IPAs
IR researchers have acknowledged politics at international level as a distinct political context, which led to the creation of IPAs: IPAs exist because member states as principals decided to engage in politics on a global scale (Abbott and Snidal 1998;Hawkins 2006). At the same time, IR scholarship has emphasized the role of domestic politics for IR as a two-level game, revealing interdependencies between member states engaged in domestic and international politics (Putnam 1988;Pevehouse 2002). Following the understanding of the relevance of politics, IR scholars have studied how IPAs' behavior as (strategic) actors in global governance is shaped by IO member states and how far IPAs are able to shape the domestic politics of the members (Avant, Finnemore, and Sell 2010;Rosenau and Czempiel 1992) and how their performance in the field is constrained by politicians (Honig 2019). Similarly, IPAs are situated less into a stable system of checks and balances than their national counterparts (Moloney and Rosenbloom 2020). Politics at the international level goes beyond democratic politics, so anarchy and its characteristics are recognized when studying the relations and interplay of sovereign state actors in global public policy-and the role of non-state actors therein. Moreover, member states differ considerably with regard to their quality of (domestic) democracy, and therefore IOs and their IPAs face a distinct and ever-changing mix of democratic and anarchic dynamics as context to their bureaucratic activities and behavior. Lastly, scholarship on international parliamentary assemblies shows that the national members of parliament (MP) therein contribute their role and experience in domestic parliamentary control in order to extend the legitimacy of the distinct IO, to improve the control over their own government's activities within this IO-and highlight interdependencies between these control patterns over executive activities at the international and domestic level (Kraft-Kasack 2008; Malang 2019; Verdoes 2019; Lipps 2020).
Most PA research on domestic bureaucracies has acknowledged the importance of politics but has neglected distinct effects of contextual political determinants for administrative actors and behavior, such as political systems or types of party competition except studies on political appointees or on formal structural changes of central government organizations (see Page and Wright 1999;Dahlström and Lapuente 2017;Fleischer, Bertels, and Schulze-Gabrechten 2018). For international bureaucracies, scholars have pledged to disentangle the relationship of bureaucracy and politics (Ege, Bauer, and Wagner 2020), but so far few attempts have been made to do so. These attempts have been limited to asserting politics matters yet are rather limited as to how they matter (see above on IPA research applying the PSB perspective). Therefore, the international political context of IPAs allows PA scholarship to benefit from a closer look into the findings of IR studies: In the absence of party politics and regular elections, IPAs have less political and ideological control for their role and operations (Weller and Xu 2015). At the same time, member states' interactions are regularly more contested than most cabinet dynamics at the national level and donors as decisive "parallel principals" do not exist in the democratic chain of delegation at national level. Lastly, international bureaucracies and national bureaucracies interact, thus developing transnational administrative settings (e.g., Egeberg and Trondal 2009;Heidbreder 2015;Stone and Ladi 2015) in which political signals and inputs primarily addressing domestic bureaucratic actors may spill over to their international counterparts within these transnational arrangements. Hence, we advocate for more research that takes the Downloaded from https://academic.oup.com/isr/advance-article/doi/10.1093/isr/viaa097/6109791 by guest on 29 January 2021 politics of international bureaucracies into account and benefits from IR and PA perspectives while doing so.

Outlook: A Comprehensive View on Agency and Politics
IPA scholarship in IR and PA has been primarily concerned with identifying gaps and missing pieces in their own research agendas, rather than discussing potential points of intersection suitable for strong, mutual scholarly conversations. This paper provides a nuanced look at the current state of IR and PA research on IPAs to initiate a joint research agenda. Reflection on the disciplines' commonalities and peculiarities shows focal points for bridges and future avenues to foster mutual conversations. Despite varying starting points and research interests of the fields, our discussion shows two conceptual common denominators suitable as points of reference for future research: agency and politics.
One may argue in favor of more research acknowledging the variety of agency in the realm of IPAs. This could be done by including different levels at which individual and collective actors perform this agency through distinct practices in order to accomplish policy influence, authority, and autonomy. Future research could focus on the relational rather than the substantial nature of agency (Hofferberth 2018), emphasizing the need for comprehensive empirical accounts that situate IPAs next to their counterparts in national administrations and to other nonstate actors at domestic level and explore their relational dynamics. Moreover, the closer analysis of the microlevel of IPA staffs and their attitudes allows testing other crucial concepts in domestic PA debates such as the motivation of these IPA staff members or their core attitudes toward the peculiar nature of their international political working context. At the international level, a focus on varieties of agency also raises questions of representation and of IPAs' capabilities in "unmaking" inequalities in international institutions (Fehl and Freistein 2020). Under which conditions do international bureaucrats decide to act as policy entrepreneurs and when do they (deliberately) decide to support the status quo? Do individual qualities or interpersonal relationships of their staffs determine the scope of an IPAs' entrepreneurship? What role do epistemic communities or transnational advocacy networks play for bureaucratic attempts for innovation?
One may also argue in favor of analytical frameworks sensitive to agency and politics that are capable of assessing potential interactions between IPAs and other actors or their counterparts in IOs. Interactions between IPAs and NGOs, for example, can range from conflict to cooperation (Johnson 2016), prompting the question how these dynamics unfold. Studying networks as bureaucratic forms of cooperation may provide such opportunities for joint IR and PA research (Lecy, Mergel, and Schmitz 2013). Another opportunity for more integrated research may be a stronger focus on preferences. On the one hand, preferences are formed to articulate agency, and agency is needed to express preferences. On the other hand, politics signal, exchange, and mediate preferences, and lead to compromises that reflect a distinct balance between preferences. A promising analytical avenue forward in future IPA research is therefore to aim for conceptualizing preferences in the realm of IPAs in light of agency and politics. Do IPAs formulate and express organizational preferences next to policy preferences? Are these preferences more dominant in the implementation of existing programs or when IPAs engage with principals over future priorities and preferences?
As a last consequence, a joint research agenda that considers the distinct characteristics of IPAs may enable addressing of questions of generalizability and equifinality. Notable attempts have been made to systemize IPAs along different dimensions (Knill and Bauer 2018;Bauer, Knill, and Eckhard 2017), but the challenge remains to account for the variety of IPAs as organizational actors and as the sum of their staff. This challenge also encourages us to make better use of existing data and reflections on the applicability and validity of different methods available. Given the peculiarities of the political context (and the diffusion of accountability therein), we have comparatively more quantitative data available on the organization and outputs of IPAs than for most domestic bureaucracies. At the same time, the rigorous filing and digitalization of these data sources over time allow for unique data analyses. Hence, we should make use of these data for IPA research and for reflecting on the applicability of theoretical arguments that originated in IR and domestic PA research and allow for their renewed empirical testing.