International bureaucracies and their influence on policy-making: a review of empirical evidence

ABSTRACT Although we find considerable literature on international organizations and their bureaucratic interior, there has been little effort to systematically synthesize empirical research across the different academic disciplines examining how international bureaucracies affect policy-making at an international level. This contribution reviews existing research on the policy influence of international bureaucracies published during the past 50 years. Applying a keyword-based search strategy allows us to identify a core body of 83 books and articles. We find a general consensus in the literature that international bureaucrats do influence policy-making, though this influence varies with the political salience and scope of the decision at question. Yet there is still much disagreement about other context factors, including mechanisms and behavioural assumptions. The contribution advances the state of the art by extracting major disputes – mostly linked to diverging disciplinary perspectives – and existing gaps in the literature, and by suggesting areas for future research.


Introduction
At the 1919 Paris Peace Conference, the Allied victors of World War I convened to discuss one of the most significant alterations of the international system since the Westphalian Peace. Apart from specifying the terms of peace with Germany in the Treaty of Versailles, the conference merged into the League of Nations. By deciding in favour of an independent League secretariat, global powers for the first time defined the 'international as a space distinct from the sum total of member states' interests' (Sending 2014: 338-9). Surprisingly, this historic instance of reallocating authority has not been limited to technical matters, such as conference support functions. Having agreed on legal rules and bureaucratic procedures to govern this space and having decided that international civil servants should act towards international peace and stability, member states reshaped international peace and security governance, thus touching upon the very essence of national sovereignty. Doing so, international relations expanded from the realm of international law and diplomacy to international bureaucracy.
Who are these international bureaucrats, how are they organized, and how do they affect global policies? For almost 100 years, the bureaucratic dimension of global governance has eluded mainstream academic interest. This lacuna is surprising, given that the role of public administration in policymaking was among the core themes that originally defined the discipline of public administration in a national context (Aberbach et al. 1981;Page 1992). Compared to their national counterparts, bureaucracies in international organizations (IOs) have less resources and enforcement powers at their disposal, but their accountability to multiple principals (often bureaucrats themselves) and their operation in the absence of an established legal framework puts them in a potentially very powerful position (Mailick 1970). Thus, the study of international bureaucracies and their involvement in policy-making beyond the nation-state not only promises new answers to old questions addressed by scholars of public administration, but has already shown to fruitfully supplement a state-centric perspective in international relations research (Abbott and Snidal 1998;Reinalda and Verbeek 1998).
With some exceptions in the 1970s, the tide has only recently begun to turn since Michael Barnett and Martha Finnemore's (2004) bureaucratic explanation of powers and pathologies encouraged more scholars to unpack the 'black box' of IOs (Ege 2016;Hawkins et al. 2006;Hooghe and Marks 2015). This literature cuts across the disciplinary boundaries of international relations, public administration, comparative politics and policy analysis, and is diverse in its research objectives and approaches. Previous state of the art reviews have focused on the emergence of international bureaucracies as a new research field (Liese and Weinlich 2006), their conception from different disciplinary (Ege and Bauer 2013;Venzke 2010) and epistemological perspectives (Bauer and Weinlich 2011), their capacities for internal learning (Benner et al. 2009) and reform (Geri 2001), as well as their influence on climate negotiations (Bauer et al. 2009) and on policy-transfer in member states (Busch 2015). Moreover, Busch (2014) focused on the independent influence of international public administration (IPAs). By and large, these reviews agree that IPAs matter and that they are autonomous actors with some degree of influence on global public policy.
In this contribution, we apply a formal approach to review empirical evidence on the influence of IPAs on global policy-making. This allows us to broaden the scope of previous reviews that drew their conclusions from literature on a specific phenomenon (such as learning) or on a specific policy field (such as environment or security). Applying a systematic keyword-based analysis allows us to identify 83 empirical articles and books published by major academic journals and publishers between 1965 and 2014. These texts constitute the core body of IPA literature.
Overall, the contribution of this study is threefold. After a brief conceptual and methodological primer, we first provide a descriptive overview of the most salient patterns in empirical research on the influence of IPAs along three focus areas (institutional design, policy formulation and policy implementation). Second, we identify which determinants scholars have employed to explain bureaucratic influence. Third, by considering patterns in the disciplinary origin of IPA research in our sample, we argue that a conceptual gulf between public administration and international relations research is responsible for the fragmented state of the art. In the absence of a common heuristic, scholars emphasize different aspects of the social reality within IOs, impeding coherent findings on IPA influence on policymaking.

Methodology
A glance at the various foci and findings of previous reviews shows that research on international bureaucracies is considerably diverse. Scholars have looked at IPAs from different disciplinary angles and with varying epistemological lenses. Against this background, the aim of our literature search was to provide an interdisciplinary perspective and identify those journal articles and monographs characterized by a pronounced focus on the bureaucracy of IOs. In a second step, this sample of publications was then analysed for its content.
We relied on two major literature databases to screen the pertinent academic debates for relevant publications. First, WorldCat®, the world's largest library catalogue, was used to retrieve relevant monographs from the collections of tens of thousands of libraries worldwide. We did not include edited volumes and handbooks in our sample because we assume that sooner or later the findings presented in these formats will appear in journal articles and monographs. 1 Second, we used Thomson Reuters' Web of Science™ (WoS) subscription-based citation database to supplement our sample with relevant academic articles published in international peer-reviewed journals.
References to the policy role of international bureaucracies can be found in publications that consider bureaucratic features of IOs, as well as in studies with a genuine focus on administration in the global sphere. In consequence, we performed two separate keyword searches for each database (details on keywords and search restrictions are available as supplemental material to this contribution). Since we tried to be as inclusive as possible with the specification of search parameters, our initial search yielded 281 publications, including many works that referenced international bureaucracies only marginally. After eliminating duplications, literature reviews and highly normative contributions, we coded the centrality of the administration in each study based on information in the abstract or summary of the publication. Only studies in which international bureaucracy was either the prime focus or an important explanatory factor were considered relevant. The resulting list of publications included 26 books and 57 journal articles (see supplemental material). In order to indicate that a publication cited in the remainder of this contribution is part of the sample, the respective author and year of publication are printed in italics. Owing to the comprehensiveness of the literature reviewed, however, these publications are not listed in the reference section of this contribution.
Applying a formal strategy to identify relevant literature has limitations. First, it can hardly be claimed that the selection procedure applied here allows us to analyse the full population of pertinent social science scholarship. Instead, our review provides a rather conservative assessment of the research output in the field because publications are selected only if one of the relevant search terms was mentioned in the title or abstract of the respective publication or in the database-specific keyword section. Furthermore, our sample underestimates research efforts in the form of articles because the WoS database includes only the most established and highly ranked journals, at the neglect of new and innovative ones. Despite these caveats regarding the absolute size of research output, a formalized approach allows us to assess the relative prioritization within this body of literature. Because defining objective criteria reduces biases inherent in personal and necessarily selective perspectives on the topic, our sample can be considered indicative of the broader multidisciplinary debate about the involvement of international bureaucracies in matters of public policy. Accordingly, we are confident that this mapping of the pertinent literature (and the conclusions drawn from it) reflects the general trend, patterns, gaps and controversies in research studying international bureaucracies at large. Figure 1 provides an overview of the works in our sample according to their respective year of publication. It shows that the debate gained substantial momentum after 2005. In the first 40 years only 22 articles and books were published, but this number almost tripled in the decade after 2005. While it is difficult to identify the immediate causes of such an increase, this observation would certainly seem to substantiate arguments that Barnett and Finnemore's Rules for the World (2004) marked a seminal turning point in the debate toward a more explicit recognition of the bureaucratic branch of IOs (see Ege and Bauer 2013;Liese and Weinlich 2006).
Which IOs are studied most prominently? An argument often raised in this context is that scholars are interested in only a small number of 'usual suspects' (Haftel and Thompson 2006: 254). Although IOs from various policy areas are studied, Figure 2 substantiates that the empirical basis of IPA research has remained limited in contrast to the over 260 IOs mentioned by the Yearbook of International Organizations (Union of International Associations 2013).
Empirical evidence: IPA influence from institutional design to policy implementation IPAs constitute an integral part of an IO; most often referred to as its secretariat, which is distinct from the political parts of the IO (i.e., the bodies for member state representation). IPAs, sometimes also called international bureaucracies in the sociological tradition, can be defined as 'hierarchically organized group[s] of international civil servants with a given mandate, resources, identifiable boundaries, and a set of formal rules of procedures' (Biermann and Siebenhüner 2009: 37; see also Bauer and Weinlich 2011: 252). For political scientists, it is the independent influence of IPAs that is of primary interest. We use the term 'influence' as synonymous to 'having an effect ' (e.g., Bierman and Siebenhüner 2009: 40;Liese and Weinlich 2006: 504). Our review focuses on IPA actions with an (observable and attributable) effect on IO policy-making. To systematize the diverse range of empirical phenomena studied under the umbrella term 'policy-making', we follow an approach suggested by Rittberger et al. (2012) and Reinalda and Verbeek (2004), among others. These authors understand international organizations as problem-processing systems. Accordingly, policy-making is the process whereby the components of these systems interact (with each other and their environment) to transform societal demands into policy solutions in the form of organizational decisions (Easton 1957;Luhmann 1968;Scott 2003). Empirically, IPA influence becomes a relevant analytical category for studying three distinct products of this process: (1) decisions concerning the organization itself (institutional design); (2) decisions formulated at the interface of member states and IO bureaucracy (policy formulation); and (3) decisions that take place throughout the execution of IO policies (policy implementation). In line with this, we classify existing literature according to its dependent variables. 2 It is fairly straightforward to delimit matters of institutional design and the formulation of substantial IO policies as explanandum. Owing to the multilevel structure of global governance (see Benz et al. 2016), policy implementation applies to different phenomena: Firstly, when a secretariat participates in negotiations outside the legislative structures of its own IO (such as the Millennium Development Goalsfor such a research agenda see, for example, Stone [2008]); secondly, when an IO secretariat executes a policy mandate via its own projects, programmes, or missions (such as United Nations [UN] or European Union [EU] peace operations; see, for example, Caplan [2005] and Eckhard [2016a]); and thirdly, when member states implement decisions taken at the level of an IO, with the secretariat being responsible for supervision and co-ordination (Joachim et al. 2008).
Based on this simple taxonomy of organizational policy-making, we classified previous work on the policy role of international bureaucracies into three clusters. 3 The first cluster studies 'institutional design'. To varying degrees, authors are concerned with decisions that trigger change within an IO. Putting organizational and administrative institutional features centre stage, researchers seek to describe and explain the characteristics of what we initially defined as the problem-processing system. With 30 books and articles, the bulk of studies falls into this category. Half of them address one of two questions: First, the design and change of international organizations (e.g., Hanrieder 2014; Johnson 2013a, 2013b; Johnson and Urpelainen 2014; Mouritzen 1990; Weaver and Leiteritz 2005), and second, the composition or growth of the international civil service (e.g., Jordan 1981Jordan , 1991Reymond and Mailick 1986;Vaubel et al. 2007). 4 In the third group of studies, authors such as Brandsma (2012), Grigorescu (2010), Weiss (1975Weiss ( , 1982Weiss ( , 2008Weiss ( , 2010 and Conceição-Heldt (2013) investigate the relation between politicians (member state representatives) and bureaucrats, such as their respective influence on decision-making, internal control mechanisms, bureaucratic authority and the autonomy of IPAs.
While a coherent and distinct bureaucratic interest is easy to assume when bureaucratic resources and powers are concerned (Niskanen 1971), it is more difficult to theorize the interest of bureaucrats regarding substantial policies. This is the concern of the second cluster: policy decisions taken within the legal framework of an IO. With only nine articles and eight books, the 'policy formulation' cluster attracted the least scholarly interest. Most  (Benner et al. 2011;Hirschmann 2012;Schroeder 2013). Others place emphasis on the question of organizational pathologies (Barnett andFinnemore 1999, 2004) and performance (Elsig 2010).
Finally, scholars have focused on 'policy implementation', asking how factors associated with the actions of international bureaucrats affect the level of global governance more generally. Within our data set, 16 journal articles and 13 books fall into this clustera small number compared to the arguably vast amount of research looking at global governance processes without considering the bureaucracy as an explanatory factor (and thus not covered by our sample). Those that consider the bureaucracy as analytically relevant fall into three groups. First, seven studies address questions of policy-transfer to member states (e.g., Finnemore 1993; Vilella 1999; see also Busch 2015). The general tenor of these studies is that IO influence often happens behind the scenes (e.g., Mathiason 2007; see Jörgens et al. 2016) and that IO bureaucrats rarely risk open conflict with member states (Finnemore 1993; see Knill et al. 2016). The second group, consisting of 14 publications, studies IPA influence on global policies; i.e., policies negotiated outside the legal framework of only one IO (e.g., Biermann and Siebenhüner 2009; Nay 2012; Xu and Weller 2008). Finally, eight articles and books look at the implementation activities carried out by the secretariat itself. Gest and Grigorescu (2010), for instance, study the determinants of IO co-operation; while Graham (2014) finds that fragmented agents are more difficult to control, causing IO bureaucrats to drift from their mandate.

Determinants and context factors of influence
Almost all authors observe administrative influence on IO policies in at least one of their cases (but see McLaren 1980). Their findings regarding influence are, however, by no means based on a coherent research interest and are often a side-product rather than the actual objective of most studies. Although some authors explicitly focus on the influence of IPAs as a dependent variable (e.g., Biermann and Siebenhüner 2009;Trondal 2011;Widerberg and van Laerhoven 2014;Xu andWeller 2004, 2008), the lion's share of research addresses the shape and change of specific policies or institutional features as its central puzzle. Because of the lack of an objective reference point or yardstick for administrative influence, only a few comparative studies are able to compare the bureaucratic footprint across organizations (e.g., Biermann and Siebenhüner 2009; Elsig 2010). Thus, instead of assessing the size of influence, authors in the sample identify particular administrative characteristics (often linked to an underlying causal mechanism) that are found to lead to orin a more probabilistic readingincrease or limit secretarial influence. 5 First, several studies point to the expertise and authority of international bureaucrats (experience, control over information, neutrality) and the entrepreneurial activity of their senior leadership (e.g., Barnett and Finnemore 2004;Biermann and Siebenhüner 2009;Harman 2011;Jinnah 2011;Morgan and Shahjahan 2014;Nay 2011;Xu and Weller 2004). Second, structural aspects of the administration (such as hierarchy, personal selection, specialization and distance from political decision-makers) are also found to shape the capability of IPAs to influence policy-making (Xu andWeller 2004, 2008). Third, a commonly applied strategy to trace mechanisms of influence is to distinguish between different types of influence. Biermann and Siebenhüner (2009), for instance, differentiate between three types (see also Nay 2012; Widerberg and van Laerhoven 2014). 'Cognitive influence' captures the capacity of the bureaucracy to gather, shape and disseminate information and knowledge. Bureaucratic influence on policies themselves (called 'prescriptive or normative influence') and influence during policy implementation (called 'technical or executive influence') supplement the typology. These studies suggest that the 'bureaucratic footprint' is most pronounced in the earlier stages of policy-making. Elsig (2010) supports this finding by pointing out that the further the involvement of the administration shifts from institutional designs to policy formulation and implementation, the less pronounced the bureaucratic influence and the more important political and environmental factors become (see also Weinlich 2014: 218).
But what are these political and environmental factors? In one of the earlier programmatic papers on international secretariats, Liese and Weinlich (2006: 515) provided a list of relevant context factors. Two major kinds of context factors can be distinguished: the underlying policy and the preferences and control capacities of member states. These factors also feature prominently within our sample. First, the assumption that IPAs wield more influence when technical issues ('low politics') and complex problems are concerned (Cox and Jacobsen 1973) has been substantiated by later studies (Bohne 2010;Biermann and Siebenhüner 2009;Xu andWeller 2004, 2008). In line with this, studies in our sample regularly argue that the more politicized a decision or task, the less bureaucratic influence is found (see Second, the preferences and control capacities of member states matter for influence. For example, bureaucratic influence shrinks under conditions of rule-based delegation (in contrast to administrative discretion) and political oversight (Best 2012; Conceição-Heldt 2013), in particular over organizational timing and sequencing (Goetz 2014). Policy-related preferences are relevant, but findings are inconclusive as to whether heterogeneity among principals increases or decreases bureaucratic influence (Copelovitch 2010;Jinnah 2011;Xu and Weller 2004). In addition to Liese and Weinlich (2006), Busch (2014: 55) also refers to the relevance of situational factors. In our sample, two studies emphasize policy uncertainty (Broome 2012) and organizational crisis (Chorev 2012) as facilitators of IPA influence.

Gaps and disputes
International bureaucrats matter. Although scholars have tackled a variety of policy-related phenomena at different analytical levels, this finding is their smallest common denominator (see also Busch 2014). Our review yields a list of administrative characteristics and context factors that have been found to affect administrative influence. We should note that these do not comprise an integrated theory of IPA influence, but rather, provide only snapshots often based on single case studies. As Table 2 demonstrates, scholars address IPAs from different disciplinary backgrounds. Scholars of international relations are more actively involved in the debate (42 publications), while proponents of public administration have published only 33 studies on the subject. 6 Table 2 further highlights that the disciplinary angle seems to affect which phenomena scholars address. In terms of studying institutional design and individual bureaucrats, public administration is slightly more active. When it comes to policy formulation and implementation, however, public administration scholarship contributes only 12 studies, while the output within international relations is more than twice as high (27 publications).
Public administration and international relations ask different questions and focus on different aspects along the three main areas of policy-making. Explanatory factors commonly applied in public administration researchsuch as bureaucratic rules, organizational structures and management processeslose relevance when the analytical perspective shifts upwards along these three areas of policy-making. The explanation of institutional design is the most prominent cluster of research in our sample (30 publications). In explanations of policy formulation and implementation, however, member-state influence and bureaucratic agency matter more. None of the studies in our sample explain policy formulation and implementation with primary reference to the secretariat's internal structure and processes. Arguably, this finding may be partially explained by the fact that internal administrative features are empirically less relevant during the later stages of the policy process. We suspect, however, that this imbalance also has epistemological reasons, such as a particular preference for actorcentred explanations among international relations scholars. Addressing this more systematically, the next sections show that there is a high degree of theoretical diversity in IPA research.

Theoretical roots of research on IPA influence
IPA research is part of a larger neo-institutionalist school of thought that is united by the core assumption that 'organization matters' (Hall and Tylor 1996;March and Olsen 1984;Scharpf 1977). Within this broad paradigm, three theoretical viewpoints dominate. First, actor-centred approaches, the most frequently applied explanation, highlight the influence of crucial individuals located at the upper echelons of the secretariat (especially the executive head and senior management). These executives are identified as being able to promote change of a structural or content-related nature from within the administration by acting as policy entrepreneurs ( In the second group, several studies apply a structural perspective to explain policy-making (see Egeberg 1999;Hammond 1986). Such studies challenge the idea that 'the administrative behavior of staff within international bureaucracies is profoundly shaped by the legal mandates of IOs' (Trondal 2011: 797). Instead, structural features such as the vertical and horizontal specialization of secretariats are found to determine individual action, which, in turn, affects policy substance (Biermann and Siebenhüner 2009;Trondal 2011;Xu and Weller 2008; see also Bauer and Ege 2016;Eckhard 2016b). Others refer to organizational cultures as relevant structural features of the bureaucracy (Sarfaty 2012).
Third, norm-centred approaches (mostly rooted in social-constructivist thinking) apply a similar strategy but operate from the assumption that actors and institutions are highly intertwined (see Wendt 1987). Thus, instead of tracing influence back to individuals or internal structures, they emphasize that international secretariats function as bureaucratic organizations in a Weberian sense. By establishing themselves as an authority that defines 'what counts' (Morgan and Shahjahan 2014: 192), international bureaucracies wield influence by classifying information, fixing meanings and diffusing norms (Barnett andFinnemore 1999, 2004;Finnemore 1993Finnemore , 1996Nay 2012). According to Barnett and|Finnemore (1999, 2004), this authority enables IPAs to reach substantially beyond their rational-legal character and delegated tasks. Instead, it is their superior moral authority and specialized knowledge vis-à-vis member states that bestows IO bureaucrats with power. Thus, the availability of unique bureaucratic knowledge (Gest and Grigorescu 2010;Hirschmann 2012;Johnson 2013aJohnson , 2013bJohnson and Urpelainen 2014;Widerberg and van Laerhoven 2014;Xu and Weller 2004) is a power resource that is prominently used in both actor-and norm-centred approaches.

Individual behaviour and bureaucratic influence
Researchers agree that the international bureaucrat is an entrepreneur rather than a passive servant facilitating state interaction. Beyond this question, however, the literature is far from united in terms of its understanding of these multilateral bureaucratic élites. Although not consistently linked to the underlying theoretical roots, three images of the underlying behavioural logic of international officials prevail. While the individual portraits in the sample are, of course, more nuanced than such a simple taxonomy suggests, international civil servants are generally seen as either neutral professionals committed to the mandate of the organization, self-interested individuals, or actors subject to outside interests.
In line with the normative ideal of the impartial international civil servant, as coined by the first Secretary-General of the League of Nations, Eric Drummond, some researchers find that international civil servants do 'serve the system' (Xu and Weller 2008: 49). Biermann and Siebenhüner (2009) also found that IPAs should be seen as problem-solvers primarily interested in realizing the founding mandate of their organizations (see also Widerberg and van Laerhoven 2014). Especially in the context of the European Union, scholars argue that EU officials' subscription to the European idea shapes their behaviour (Juncos and Pomorska 2013; see also Hooghe 2005).
The second perspective views bureaucrats as (at least partly) self-interested individuals. For instance, Bauer (2012: 500) confirms that the 'expectation of fair career treatment' is among the best predictors of whether EU bureaucrats will support management reforms. Rationalist scholars such as Vaubel et al. (2007: 276) also argue that 'self-interested utility-maximizing behaviour' leads to 'international bureaucrats who have a vested interest in the expansion of their organization' (see also Copelovitch 2010; cf. Nielson and Tierney 2003). Acknowledging the relevance of this basic human tendency for bureaucratic behaviour, Weiss (1982Weiss ( : 301, 1975) was among the first to debunk the ideal of the fully benevolent international bureaucrat. In his seminal assessment of the empirical reality in the UN system, Weiss found that personal ambition for status within their organization is what makes IO bureaucrats tick.
Partially overlapping with the previous argument, the third perspective links bureaucratic behaviour to outside political and societal interests. International bureaucrats do not act in a political void. Instead, as Weiss (1982) argues, they act mainly in ways that please powerful member states' interests a phenomenon that, in a national context, is known as professional or functional politicization of the administration (Mayntz and Derlien 1989). Stone (2011: 224) also observes that '[i]nfluential states manipulate the rules, insist on privileged treatment for their own interests, and exploit their control of the agenda'. And Park (2010) traces changes in World Bank policies back to the lobbying of transnational environmental advocacy networks.

Consequences of bureaucratic influence
Diverging views regarding the goals of individual bureaucrats leaves the literature similarly divided on the question of which effects to expect with regard to the substance of policy outputs or outcomes. Although IPA influence is generally assessed as dependent on the context conditions mentioned above, its effects on global public policy are detectable. Yet, whether these effects are desirable in the sense that they enable IOs to pursue their mandate more effectively remains contested. Bierman and Siebenhüner (2009), as well as Xu and Weller (2008), are among those convinced that, without international bureaucrats as independent actors at the negotiation table, global problems could not be approached, or at least not in an effective manner. As Xu and Weller (2008: 49) put it: 'the contribution of the WTO's Secretariat is essential to success in multilateral trade negotiations.' Authors in a norm-centred tradition are more ambivalent. Bureaucrats can be influential, to be sure, but these authors note that influence can also be pathological: the same characteristics that lend power to international civil servants may also render them 'unresponsive to their environments, obsessed with their own rules at the expense of primary missions, and ultimately lead to inefficient, self-defeating behavior' (Barnett and Finnemore 1999: 700). In line with this, Vaubel et al. (2007) have pointed to IOs' (uncontrolled) growth in size as having a pathological effect on global politics. It is not surprising in light of these deficits that authors have criticized IPA influence as undemocratic (Verweij and Josling 2003). Indeed, it is this very absence of democratic legitimacy that is used to characterize IO policy as pathologicalfor example, in UN peace operations where the local population refuses to accept international rules over their own government structures (Zaum 2006).

Conclusions and outlook
In this contribution, we have reviewed empirical investigations of the influence of international bureaucracies on policy-making. In contrast to previous literature reviews, we applied a formal search strategy that we deem particularly adequate for such an interdisciplinary research field. We confirmed the main findings of previous state of the art reviews, which captured the literature's general tenor that international bureaucracy matters. Additionally, we presented an original mapping of empirical evidence on IPA influence, as well as an overview of the determinants of bureaucratic influence and the fragmentation of the research field. Overall, bureaucratic influence seems particularly significant in the early stages of policy-making, if issues of 'low politics' are concerned, and on decisions that affect an organization's own institutional design. Apart from that, authors' conclusions regarding the relevance of single determinants of administrative influence are inconsistent.
Our review confirms that the majority of studies focus on the secretariats of the most prominent organizations, such as the UN, the EU, and the Bretton Woods institutions. By identifying the frequency with which each organization is studied, this claim is put on a more solid empirical basis. In a similar vein, the present study finds that with currently only 12 studies with more than four cases (i.e., 14 per cent of the sample), the claim that the field would benefit from more comparative assessments (Bauer et al. 2009;Busch 2014;Ege and Bauer 2013) is understandable. In recent years, however, there has been a visible trend towards more empirical diversity and more comparative research. Thus, we see these shortcomings as a decreasingly problematic issue within the more recent debate.
By contrast, another problem identified in this review is unlikely to vanish. It is particularly evident that IPA research is fragmented. Two disciplines dominate the field: international relations and public administration. International relations research has traditionally considered IOs from the perspective of multilateral negotiations among powerful states who act in their own rational interest. In asking 'how institutions matter in shaping the behavior of important actors in world politics' (Martin and Simmons 1998: 729), empirical research on IPA has advanced the discipline's focus. Yet, in asking these questions, international relations has also ventured into the theoretical grounds of domestic politics and public administration research (see also Ellis 2010). As long as scholars (from both disciplines) apply what has been referred to as 'x-centered research designs' (Ganghof 2005) in order to investigate the causal effects of single factors on policy-making, such a development is not problematic. For instance, if the aim is to determine if a particular factor matters for policy-making, the two disciplinary approaches can be seen as fruitfully complementing each other. If scholars turn to 'y-centered research designs,' however, in order to study all 'causes of effects' of a certain policy, the diversity found in the current research landscape may become an impediment to theoretical progress. Y-centred designs are applied, for instance, to identify the conditions under which bureaucratic influence occurs. This requires the ex-ante selection of theoretically relevant factors; a process that tends to end differently depending on one's disciplinary background. In view of the increasing convergence of research questions and a shift towards a common empirical interest, a common theoretical framework to study the role of IPAs in global governance is overdue. For the time being, the fragmented nature of IPA research prevents us from drawing definite conclusions on bureaucratic influence and its determinants The fields of public administration and (neo-institutional) international relations each have their strengths and weaknesses. Both share the assumption that societal phenomena of change and persistency are irreducible to rational political behaviour or environmental factors. Instead, policy-making in and through international secretariats implies that the structures, rules, interests and characteristics of those working in these administrative bodies have an independent effect on policy substance. Similar to the discipline of policy analysis in the domestic context (Howlett et al. 2009), it is the comparative strength of international relations to ask about the entelechy of distinct global policies and the linkages between institutions, actor-driven political processes, and policy content. In contrast, it is the strength of public administration to ask about the functionality, structure, and interaction of the administration with the political bodies of public organizations (Pierre 1995: 9-13). As Renate Mayntz (2009: 10) has written, this makes public administrationamong all the strands of research that have emerged from classical organizational theorythe most useful source for findings and analytical approaches to be applied to the study of IPAs. Given also that domestic policy scholarship has benefited tremendously from a 'bureaucratic perspective' (Kaufman 1960(Kaufman : 1992, working towards a true integration of international relations and public administration approaches could be equally beneficial. Research on the European Commission may serve as an example of what the study of IPAs under a common theoretical framework could look like (Scharpf 1997;Trondal 2007).
Conversely, little is known about the challenges that traditional public administration research faces when its unit of analysis 'goes international'. So far, it is largely unclear what this means for the development of the discipline and we are only starting to realize the extent to which the upward delegation of powers calls into question traditional approaches and concepts of public administration research. Looking at the similarities and differences between how national and international administration operate is a good starting point. More importantly, however, investigating the consequences of these differences for the politico-administrative system of IOs and the decisions they produce remains the actual challenge.
Notes 1. To be sure, this does not mean that we deem publications in edited volumes less relevant. On the contrary, edited volumes have served as important fora for the advancement of research on internal features of IOs and their bureaucracies (e.g., Dingwerth et al. 2009;Kim et al. 2014;Reinalda and Verbeek 2004). As an exception to this general rule, we included the volumes edited by Biermann and Siebenhüner (2009) and Cox and Jacobsen (1973). In contrast to conventional volumes, these two books assemble the results of joint research efforts in a coherent manner and are thus particularly interesting for our purposes. 2. Importantly, not all of the studies are designed to identify a causal link between a dependent and one or more independent variables. Thus, the term 'dependent variable' (or explanandum) is, in fact, too narrow in this context. Nevertheless, since the expression 'dependent variable' is regularly used to denote the central analytical phenomenon of empirical research more broadly, we use this terminology in the course of the paper. 3. Our search included seven journal articles that aim to describe and explain the attitudes and role behaviour of individual civil servants. Since such a focus is not directly relevant for assessing the policy-making role of IPAs, the table refers to the reduced sample of 76 publications. 4. In order to indicate that a publication is part of the sample, the respective author and year of publication are printed in italics. 5. To be sure, the term 'mechanism' usually refers to causal processes through which influence operates. The reference to particular 'characteristics' of administrations, in contrast, is more prominent in variable-oriented (comparative) approaches. Since our goals is to identify the different determinants of influence (no matter whether they are identified by means of process tracing or cross-case comparison), we do not further differentiate between diverging understandings of these two notions here. 6. Eight publications come from related disciplines such as history, anthropology or economics. The disciplinary origin of a publication is coded by considering the classification specified by the respective data base. In WoS, public administration research includes the research areas 'Public Administration' and 'Law and Government'. International relations is a separate research area. For monographs, we coded the disciplinary origin according to the target discipline of the book series or, if this was not possible, according to the academic positions or denomination the authors hold at their home institution.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the German Research Foundation under Grant FOR #1745.

Notes on contributors
Steffen Eckhard is a senior researcher at the University of Munich (LMU) and a nonresident fellow at the Global Public Policy Institute (GPPi) in Berlin.
Jörn Ege is a research fellow at the German Research Institute for Public Administration Speyer.