The architecture of organizations as missed opportunity in political research

Döhler, Marian Article — Published Version The architecture of organizations as missed opportunity in political research Public Administration Provided in Cooperation with: John Wiley & Sons Suggested Citation: Döhler, Marian (2020) : The architecture of organizations as missed opportunity in political research, Public Administration, ISSN 1467-9299, Blackwell Publishing Ltd, Oxford, UK, Vol. 98, Iss. 4, pp. 891-904, https://doi.org/10.1111/padm.12669 This Version is available at: https://hdl.handle.net/10419/230160 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ ORIGINAL ARTICLE The architecture of organizations as missed opportunity in political research Marian Döhler Institute of Political Science, Leibniz University Hannover, Hannover, Germany Correspondence Marian Döhler, Institute of Political Science, Leibniz University Hannover, Schneiderberg 50, 30167 Hannover, Germany. Email: [email protected]over.de Abstract Few scholars in political science or public administration deny that organization matters. However, the rare application of an organizational perspective in research suggests otherwise. To revive the awareness of organizations as unit or level of analysis, attributed and generic properties are distinguished. Attributed properties, which dominate research, assign functions and patterns of behaviour to an organization. In contrast, generic properties refer to the constitutive elements of organizations that take effect before attributed properties. This article takes a closer look at four generic properties by examining their often-implicit use in current political science and public administration research. The aim is to demonstrate that the formal dimension, the goals, the expertise of personnel and organizational boundaries exert an independent influence on the output of political or public sector organizations. 1|INTRODUCTION: DOES ORGANIZATION STILL MATTER? The role and relevance of organizations in political science and public administration research are beset by paradox. On the one hand, few researchers deny that organization matters. On the other hand, however, considering the organization as unit or level of analysis is rare due to the dominance of a phenotypic perspective that perceives organizations as parties, interest groups, agencies, universities, and so on. The genotypic perspective, which puts the organizational dimension at the forefront, seems to be left to other disciplines, such as management, organization studies or institutional economics. The distinction between the two perspectives is profound. The phenotypic perspective assumes that organizations act as they do because of properties in addition to being an organization. Received: 21 May 2019 Revised: 28 March 2020 Accepted: 8 April 2020 DOI: 10.1111/padm.12669 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Author. Public Administration published by John Wiley & Sons Ltd. Public Admin. 2020;98:891–904. wileyonlinelibrary.com/journal/padm 891 According to this perspective, parties are motivated by vote-, officeor policy-seeking, interest groups by the preferences of their members, and government agencies by law, and so on. There is nothing wrong with this analytical predisposition that is also predominant in the literature as long as this does not lead to distorted empirical observations. At the same time, some important limitations of taking such a perspective need to be considered. What gets lost in the dominant analytical perspective are generic properties, such as the formal structure or boundaries of organizations, which have an effect before attributed properties. Either these variables are taken for granted or have ‘become epiphenomenal’(Stinchcombe 2001, p. 2). Consequently, the impact of generic properties is underrated or even overlooked, thus leaving a blind spot in explaining the performance, output or behaviour of political or public sector organizations. But does this blind spot truly exist, and if so, does it matter? When considering the decisive impact of the internal structure for decisions and policy output in military and intelligence operations (Allison 1971; Hammond 2010) or the field of high-risk technologies (Heimann 2010), the question must be answered in the affirmative. Going beyond special domains, this article suggests a general distinction between generic and attributed organizational properties. The aim is to demonstrate that generic properties play a more critical role in policy output than is commonly considered in political science and public administration. Generic properties denote essential elements of any organization, including their internal structure, their boundaries, or routines and procedures (Scott and Davis 2007, pp. 19–24). By contrast, attributed properties signify the political or social functions of an organization, such as interest representation, service delivery or the constitution of a group based on religion (churches or charities) or political beliefs (parties). Broadly speaking, the attention of the phenotypic perspective is on attributed properties so that the organizational dimension remains a black box. Typical are notions such as ‘international bureaucracy’or ‘nongovernmental organization’that, more or less implicitly, pretend to explain organizational behaviour and output. By contrast, the genotypic perspective, to be elaborated subsequently, looks behind labels to get to the effects of generic properties, which form the basic, though variable, components of an organization regardless of its type, age or size. To be sure, an organizational perspective is not entirely absent from the existing literature. Classics such as the work of James Q. Wilson and Harold Seidman have pointed to the relevance of organizational structures and tasks (Seidman and Gilmour 1986, p. 15; Wilson 1989, pp. 24–26). Organizational properties have long been highlighted as crucial for the policy capacity of interest groups (most notably: Schmitter and Streeck 1981, pp. 140–82; Halpin 2014). More recently, Morten Egeberg (1999; Egeberg and Trondal 2018) has emphasized horizontal and vertical specialization, age and location as key variables for analysing bureaucracies. Apart from the authors just mentioned, generic properties mostly remain subordinated to the phenotypic perspective so that they are not seen analytically. This may be less true for public administration. But the once pronounced focus on organizational properties is nowadays either deported to the less relevant classics overview in textbooks, or increasingly superseded by research topics that are easier to measure through external observation, such as performance, reputation or accountability (see the overview by Bach and Wegrich 2018). Therefore, this discipline would also benefit from revitalizing an organization-centred perspective. To gain further insight into the effects generic properties generate for the policy output, the subsequent analysis proceeds as follows. In the next section, four generic properties are isolated from a broader set, and the claim is substantiated that these units do not receive sufficient attention in research. The third section takes a closer look at existing research findings to devise an organization-centred research perspective. Based on this synthesis, the final section proposes an analytical framework for future research. 2|WHY GENERIC ORGANIZATIONAL PROPERTIES BECAME ‘EPIPHENOMENAL’ Even if a final list of generic properties is lacking, an established consensus in organization research would suggest the formal organization (Blau and Scott 1963, p. 2) as the most general category. This property includes the 892 DÖHLER organizational form, which expresses the vertical and horizontal specialization, and determines varying degrees of hierarchy. Second, each organization has a boundary with a varying ‘degree of permeability’(Blau and Scott 1963, p. 194). Aside from this, organizational goals often discussed along with jurisdictions or missions in the public sector (Wilson 1989, pp. 32–49) are the third relevant generic property. Finally, members or employees of an organization represent the fourth property, which again, can take on different forms such as academic qualifications, professional norms or particular attitudes. A common denominator is that these four properties serve as a kind of hardware on which the software, such as culture, routines or organizational memory, is running. To sum up, the current state of the art in political science and, to a large extent, also public administration seems to be a continuation of the existing tendency to downplay or bypass the organizational dimension. This is less due to a lack of relevance than being caused by attention cycles in research: the first is the new institutionalism, in which looking at institutions is often confused with an organization-centred perspective. Second, two disciplines from which political science and public administration heavily borrowed to conceptualize and analyse organizations, that is, organization theory and the ‘new economics of organization’(Moe 1984), have distanced themselves from an organization-centred view. While the institutional school of thought, flourishing in several versions since the 1980s, has refreshed political science research considerably, it simultaneously includes an analytical weakness as a random selection of authoritative accounts reveals (Hall and Taylor 1996; Pierson and Skocpol 2002; Mahoney and Thelen 2010; V. A. Schmidt 2010). None of these well-written and thoughtful synopses demonstrates an effort to draw a line between institutions and organizations. Organizations appear here and there, but the role and effects of the organizational dimension remain underexplored. Again, it may be asked why should this be an analytical weakness. A side-glance at a neighbouring discipline helps to answer this question. Unlike political science and public administration researchers, sociologists agree about the particular meanings of institutions and organizations. Ever since Philip Selznick's pioneering contribution, institutions have been perceived to emerge if organizations are ‘infuse[d] with value beyond the technical requirements of the task at hand’ (Selznick 1957, p. 17). In other words, organizations become institutionalized if they are pervaded by the cultural norms or values operating in their environment (Scott 1995, p. 31). No such comparative distinction is regularly used in political science or public administration research. Institutions are mostly defined as action-enabling or -restricting rule systems ‘that structure the course of action that a set of actors may choose’(Scharpf 1997, p. 38). While organizations, next to individuals, are recognized as political actors, their internal structure and operations are rarely considered a relevant dimension for analysis. Despite its name, the Research Committee on the Structure and Organization of Government, a subgroup of the International Political Science Association, is a typical representative of this perspective. In a programmatic paper, structure and organization are lumped together into one category without further ado (Peters 1985). Using organization, structure and institution as interchangeable terms has been evident ever since (for a notable exception see Ansell et al. 2017). In a discipline in which the research subjects stretch from individuals, such as voters or political leaders, to largescale issues like governments, federalism or international relations, analytical attention is difficult to focus on the level of organizations. Even more so since parties, interest groups or government agencies are mainly categorized as collective actors, the interaction and external effects of which absorb enough attention to stifle questions about the organizational level as an action generator. In addition, researchers have become disillusioned by the difficulty of attributing ‘intra-structural factors’(Scharpf 1982, p. 92) to political outputs, especially when they need to be separated from institutional influences. Although the next section will show a more positive picture in this respect, the dominating intuition in political science is that generic organizational properties are negligible because they will be diluted on their way up to the political context. Oddly enough, the usual suspects have failed to deliver ammunition for a counter-argument or at least support for an organization-centred perspective. Over two decades ago, Hirsch and Lounsbury (1997) defended organization theory against the new institutionalism in sociology. While not rejecting the idea of isomorphism in organizational fields overall, the authors worried about replacing interests, conflicts and action inside and between organizations DÖHLER 893 with the overarching question ‘how is it that so many organizations look the same?’(Hirsch and Lounsbury 1997, p. 81). The epistemic consequences of this redirection extend far beyond organization theory and sociology. A straightforward legacy is an environment-dominates-everything view in which the intra-organizational dimension hardly occurs. Rather than looking at the impact organizations have on their environment, the main question is how their environment influences organizations. Even if organizations are assumed to be ‘sites and drivers of social action’ (Powell and Brandtner 2016), they are pictured as ‘porous’,‘exposed to a social environment’and ‘busy responding to and negotiating external pressures’(Powell and Brandtner 2016, p. 270). As a consequence, organizations appear ‘as little more than instantiations of their environment’(King et al. 2010, p. 298). A less direct but parallel assault on recognizing generic properties is the widespread ‘fear of the formal’(du Gay and Lopdrup-Hjorth 2016), which also emerged out of the critique of classical organization theory that gave birth to the new institutionalism. It became fashionable to depict organizations as ambiguous, porous or hybrid but in no ways formal or hierarchical (Hammond 1990; King et al. 2010; Lopdrup-Hjorth 2015). Stinchcombe explains the aversion to formality as being ‘misconceived as consisting mainly of its pathologies’(Stinchcombe 2001, p. 3). Despite all its positive effects, formality became a scapegoat not only in the popular management literature. Getting rid of formality and ‘destroying the hierarchy’(a consultants' catchphrase, quoted in du Gay and Lopdrup-Hjorth 2016, p. 21) became companion slogans. In political science, a great deal of decrying the formal dimension of organizations, especially hierarchy and clear-cut boundaries, can be ascribed to an influential book by March and Olsen (1989). Building on the Simonian tradition, this book reads like an extended advocacy of the garbage can model, including the ‘organized anarchy’image through which the concept opposed formal structures and procedures. In March and Olsen's broad understanding of institutions (Sjöblom 1993, pp. 399–401), distinguishing them from organizations is superfluous as they do little more than reflect the institutional environment. This conceptual stretching contributed to an understanding of institutions that crowded out organizations as a distinct level of analysis. While organization and management theorists once considered organizations as entities with functions, identities and intentionality that trumps individual preferences and—to a certain extent—even environmental forces, the new economics of organization ‘erased the frame of reference through which formal organization had initially been conceived’(du Gay and Lopdrup-Hjorth 2016 p. 18). Instead, individuals, acting upon subject expected utility, take centre stage. Agency theory is a popular representative of this school of thought. Even though the early protagonists claimed agency theory to be an organization theory (Jensen and Meckling 1976; Jensen 1983), their actual focus was on individuals engaged in a ‘nexus of contracts’. Hence, the focus on dyadic contractual relations diverts attention away from the broad range of mechanisms working inside organizations (Mitnick 1992, p. 88). Although political science and public administration do not suffer from an ‘organization phobia’(LopdrupHjorth 2015) such as that diagnosed for sociology or organization theory, a weak understanding of organizations dominates, especially in political research. What is implied is the firm belief that a catch-all view of institutions and the phenotypic perspective suffice to cover policy-relevant aspects of organizational life. While this is a very rough description, it serves the purpose of sensitizing the reader to a dominating direction in research that drives attention away from an organization-centred view. Those analyses that nevertheless consider the organizational dimension mostly do so in an implicit or even accidental fashion. Thus, a unique lens is necessary to make these findings visible. Instead of the common bird's-eye view of organizations, an inside-out perspective is subsequently adopted. 3|LESSONS FROM THE ARCHITECTURE OF ORGANIZATIONS IN POLITICS Two specifications guide the following analysis: the first concerns the criteria for selecting research. Here, examples are introduced that are either emblematic in the way they link generic properties with outcomes or because they stress the role of generic properties more clearly than the mainstream. Second, even though it would be desirable to present a uniform perspective, the analytical lens ingrained in the existing literature cannot be reversed. For the sake 894 DÖHLER of consistency, generic properties are conceptualized as variables that transform inputs into outputs. Since we cannot expect these variables to exist in isolation, interaction effects receive particular attention. Figure 1 serves a dual purpose: as a map to guide the reader through the following subsections and as a proposed framework for an organization-centred analysis. Part of this framework is the distinction between organizations as the level of analysis and generic properties as units of analysis. Because boundaries can be more or less permeable, hierarchies can be steep or flat, etc., the variations of generic properties, which determine their impact on the policy output, not covered entirely in Figure 1, receive special attention. The four generic properties are presented separately, while recognizing that interaction can increase or weaken their effects. 3.1 |The form(al dimension) of organizations As stressed earlier, the formal structure has been crowded out for decades (Whittington 2006, p. 812; LopdrupHjorth 2015) by the informal dimension combined with all sorts of ‘soft’drivers of organizational behaviour and change. Nevertheless, a strong argument for the relevance of this property is the ‘impossibility of a neutral hierarchy’ (Hammond and Thomas 1989; similarly Seidman and Gilmore 1986, p. 15) according to which any structure causes a bias for information processing inside organizations. To claim impossibility seems bold enough. However, since the theorem does not rule out other influencing factors, it leaves open the extent to which the formal and the informal dimensions are relevant. It may be challenging to determine the specific effects of both aspects, but that does not justify neglecting the former. Research on federal agencies in the US offers a first approximation for the effects of formal organizational factors. The literature emphasizes the creation or restructuring of government agencies as well as the ensuing problem of political control. Moe's theory of ‘structural choice’(Moe 1990), a well-known treatment of this subject, focuses on organizational design as a process in which the political forces that create and delineate bureaucracies are FIGURE 1 An organization-centred framework for analysis DÖHLER 895 paramount. A whole array of instruments, such as decision deadlines, promoting professionals over political appointees or vice versa, or giving contested functions ‘a safe location’(Moe 1990, p. 137) inside the government apparatus are identified that policy-makers use either to protect or to cripple an agency (Moe 1990, pp. 136–38; Romano 2019). While this perspective stresses the role of political preferences during the creation of government agencies (Bendor and Hammond 2010, p. 641), it does not explicitly pretend that the architecture of organizations makes a difference. However, assuming that organizational design remains without consequences would render the political struggle that surrounds the creation or reform of government agencies, that is more pronounced in the US, purely symbolic. Empirical evidence that organization matters stems from two strands of research: the effects of varying degrees of hierarchy and the impact of goals on organizational output (see the next section). An analysis of the threatened armed conflict between Turkey and Greece in 1996 shows how an insufficiently organized Greek administrative apparatus impeded a proper crisis assessment. Although a formal hierarchy was in place, the flow of information from different government and military units was not appropriately ‘resynthesized and evaluated’(Jacobides 2007, p. 468), mainly due to excessive organizational fragmentation. As a result, the hierarchy lost its function as a ‘provider of templates’(Jacobides 2007, p. 467). Information from fragmented sources remained biased and was difficult to correct by exercising authority. Similarly, Zegart, in her analysis of the failure of US intelligence to prevent the 9/11 terrorist attacks, points to the internal structure of the CIA as a root cause of deficient information processing. The separation of the Directorate of Operations, who liked to adorn itself with a ‘cowboy’image (Zegart 2009, p. 67), from the cerebral-intellectual Directorate of Intelligence artificially split analysis from action. The subsequent ‘agency parochialism’(Zegart 2009, p. 67) impeded the synthesis of crucial information, which under better functioning coordination might have presented a clearer picture of terrorist threats. Both cases show the difficulty of finding the proper doses of hierarchy. Too much of it tends to hamper the upward flow of information, whereas a lack is prone to operational subgoals that subvert the overall mission. Next to organizational failures, policy-driven changes, occasionally camouflaged as efficiency-enhancing reforms, have attracted attention. Pollitt's aptly titled study Manipulating the Machine (Pollitt 1984) is one of the first pieces to scrutinize the changing organization and jurisdictions of (British) ministerial departments. Similarly, Campbell stressed the role of structures in executive behaviour (Campbell and Szablowski 1979; Campbell 1983). While these studies were designed as fact-finding or to look at the variety of government organizations, the US debate has a focus on the political control of agencies. The separation of powers provides fertile ground for presidents and Congress to clash on this issue. The fact that almost every imaginable setscrew is engaged to get a grip on federal agencies is a clear indication for the policy relevance of generic properties. An often-used example is the Environmental Protection Agency, which was celebrated as a potent regulator in the 1970s but fell under the bus a decade later. The Reagan administration implemented an outspoken deregulation agenda by tightening centralized control, introducing procedural requirements, or appointing regulatory-critical executives. Thus, priority setting and regulatory enforcement at the field-office level were weakened to the point that the agencies almost lost their clout (Moe 1989; Weiland and Vos 2002, pp. 103–05). Organizational ‘forms’, which are widely discussed in management studies, represent an important aspect of the formal structure. Whereas structure refers to an indefinite number of hard-wired features, organizational forms, such as the unitary, the multidivisional (M) or the holding form, categorize organizations by a combination of functional specialization and graduation of hierarchy. The M-form, in which self-contained divisions manage a particular product or service, is often credited with superior efficiency. In the public sector, coordination problems, and operational subgoals, as in the CIA case, must be added (Döhler 2017; with an application to interest groups, see Halpin 2014). To avoid being accused of wasting resources or fuelling bureaucratization, politicians nevertheless tend to locate new tasks in existing agencies instead of creating new ones (some empirical evidence is provided by White and Dunleavy 2010). This seems to drive the proliferation of the multidivisional form in the public sector. Despite its shortcomings, the M-form also allows cross-fertilization of policies at the intra-organizational level (see also the next section). 896 DÖHLER A recent study about the effects of Europeanization on the capacity of German and Austrian ministerial departments demonstrates that merging dispersed tasks into a single division results in more substantiated policy statements (Lichtmannegger 2018, pp. 120–23). This improvement is achieved by allowing interrelated EU issues to be worked out in a specialized context. Thus, the location of tasks inside a government organization has a perceptible influence on the policy output (see also Schout 1999). The formal structure also reveals its effects if a task is moved, especially if the new parent organization has a competing or contradictory jurisdiction. Such an effect could be observed in Germany when the responsibility for consumer protection, including food safety, was transferred to the Department of Agriculture (Janning 2011, pp. 203–04, 281–87). Despite facing powerful agricultural interests, which are traditionally represented inside the department, resistance against food safety regulation decreased as pressures to compromise in-house proved to be stronger than quarrelling with another ministry. The effects of merger-based reorganizations also became visible with the short-lived fusion of two German ministerial departments that traditionally oppose each other. Merging the Department of Economics and the Department of Labour into a single ‘superdepartment’in early 2003 aimed at better coordination between labour market policy and social assistance. Furthermore, it was expected to smooth out departmental egoisms, and eventually even to ‘define a new policy domain’(Frankfurter Allgemeine Zeitung 8 October 2002). After the merger was reversed in 2005, two effects showed up. First, ministerial bureaucrats from both camps developed a mutual understanding of how to solve conflicts internally instead of via the cabinet table. Second, and probably more importantly, was to sideline traditional, often labour union-related employees from the Department of Labour by mixing them with the business-oriented Department of Economics staff. Most observers regard this as an important prerequisite for the far-reaching social policy reform during the second Red-Green Coalition (M. G. Schmidt 2007, p. 306; Schiller 2010). This episode not only corroborates that an in-house exchange of information facilitates a mutual adjustment of antagonistic policies. Rather, tight organizational coupling, which changes the balance between divisions, can significantly redirect the policy output. 3.2 |Goals and missions According to Wilson (1989 pp. 32–35), goals are a higher-level category from which detailed tasks of an organization are derived. However, as research does not follow consistent wording, other terms such as mission, often with a normative connotation, appear as well. Goals in political science and public administration research are regarded either through the institutional choice perspective or as a condition for bureaucratic performance. As diverging interests need to be aligned when defining or redefining the mandate of a government agency, goal conflicts and goal ambiguity are well-recognized problems (Selznick 1949; Wilson 1989, pp. 32–49), but frequently remain detached from the organizational dimension. Without the explicit intention of doing so, more recent studies have unveiled the organizational underpinning of this generic property. In the control-of-bureaucracies debate, goal conflicts result from a situation in which multiple principals try to hardwire their preferences into the agency structure or jurisdiction (Moe 1990, pp. 136–38). One group of politicians might try to limit bureaucratic authority, while their opponents strive to broaden the agencies' mandate (Romano 2019). Besides, both camps make efforts to protect their legislative achievements from future changes. While these considerations are rather general, the debate on regulation provides empirical evidence of how mixed or contradictory goals affect the policy output. Regulatory agencies are often burdened with conflicting missions, such as consumer protection, industry promotion and stimulating competition, in domains as different as pharmaceuticals, energy or transport (Biber 2009; Döhler 2011; Kovacic and Hyman 2013). It takes little imagination to comprehend how this can result in internal tensions. Research interest on the impact of goal conflicts is typically raised by bureaucratic underperformance or failure. A recent example is the Minerals and Mining Service, a US agency that was dissolved shortly after the Deepwater Horizon oil spill. At first sight, a strange mixture of goals brought about the ‘regulatory laxity’(Carrigan 2017, p. 88) DÖHLER 897 that preceded the Deepwater Horizon disaster. The Minerals and Mining Service was in charge of collecting taxes on oil and gas extraction and, at the same time, was responsible for the safety and environmental inspection of offshore drilling. Therefore, it seems evident that both goals would run into conflict. Carrigan, however, sounds a note of caution on this interpretation. His analysis reveals that agency problems were not caused primarily by goal conflicts. Rather, a lack of coordination between the separate divisions responsible for tax collection and regulation, combined with an underfunding of the regulatory activity, subverted the agency's operations. By highlighting the ‘importance of task synchronization’(Carrigan 2017, p. 230), which—due to a deficient flow of information—remained subpar, this case does not render goal conflicts irrelevant. The upshot is that task coordination, which appears at the interorganizational as well as at the intra-organizational level (Nou 2015; Döhler 2017), may sound like a milder version of goal conflicts but can have as severe consequences for the policy output. The second variant of mixed goals is a jurisdiction embracing more than one sector of the economy. This agency design, often labelled a multi-utility or multisector regulator, especially applies to Britain and Germany (Böllhoff 2005), but is also present in US agencies such as the Food and Drug Administration. The pros and cons of a single integrated financial service regulator (Herring and Carmassi 2008), not only discussed in academic circles, underlines the policy relevance of jurisdictions. Majone (1997) argues that single-purpose regulators are better suited to avoid goal conflicts and to benefit from specialization. On the other hand, cross-sectoral jurisdictions may also yield positive effects. The Federal Network Agency (Bundesnetzagentur), the most prominent German regulator, illustrates both effects. In 1998, the agency started to regulate telecommunication and postal services. For practical rather than political reasons, energy and rail regulation were added in 2005 and 2006. An internal measure triggered by this expansion was staff rotation, introduced not only to prevent capture but also to circulate regulatory knowledge between sector-focused divisions (Schnitker 2009, p. 151). Although there is no validated evidence, it can be supposed that a knowledge transfer from the advanced regulatory regime in telecommunications to the lesselaborated rail sector would strengthen the regulatory grip of the latter. Despite the declared interest of politicians in the whole range of instruments to control bureaucracies, the potential effects of mixed goals largely remain out of sight during the creation or reallocation of government functions. If problems arise in practice, they are considered a bureaucratic failure; if a policy portfolio turns out to be beneficial, even if this is only due to internal adjustments, it is taken for granted. Although research seems to follow the lack of explicit political debates, the previous section confirms the relevance of goal conflicts and (insufficiently coordinated) policy portfolios for agency performance. 3.3 |Professionals and experts in bureaucratic politics The fundamental importance of human resources for organizational performance and behaviour has created a vast, almost insurmountable, volume of research. Thus, the subsequent focus is on professionals or experts as a smaller segment of personnel whose influence on policy output is more readily proved. Not surprisingly, economists and lawyers working inside governments or international organizations have generated the most interest. The rise of economists to expert prominence, especially as scientific advisers, is plausibly considered to make this discipline a driving force in shaping policy ideas (Fourcade et al. 2015; Christensen 2017). Research on this topic is rather thin when the position of economists inside policy-making organizations is considered, even though the relevance of organizational switchboards is at least recognized (Hirschman and Berman 2014, p. 790). Froeb et al. (2009) are among the few scholars to discuss the relationship between formal structure and experts. As lawyers and economists adhere to disciplinary knowledge that is context-resistant, their impact on the policy output should be regulated by their integration into agency decision-making. Both a concentration on a functional structure and a more decentralized assignment into a divisional structure exist in reality. By incorporating organizational tasks, predictions about their influence are possible. In the case of broad or several missions, the power of economic expertise increases if assigned to a functional division. By contrast, in single-task organizations, concentration in one 898 DÖHLER