Bridging the citizen gap: Bureaucratic representation and knowledge linkage in (international) public administration

Funding information Deutsche Forschungsgemeinschaft, Grant/Award Number: EC 506/1-1 Abstract Bureaucratic representation theory holds that civil servants are not “neutral” in a Weberian sense. Bureaucrats are thought to “actively” represent their communities by trying to make them better off. This article proposes an alternative understanding of individual behavior in representation that emphasizes knowledge sharing instead of patronage, but leads to similar outcomes: Their societal background provides officials with advanced social knowledge about the group(s) they represent, including both informational knowledge (facts about culture, history, politics) and relational knowledge (how people interact). Bureaucratic knowledge linkage is the process of sharing information and managing relations internally and with citizens. An extreme case serves to illustrate knowledge linkage empirically: Survey data from an international organization yield high levels of knowledge asymmetries within staff bodies and subsequent observation of knowledge linkage mechanisms. In generalizing findings, the risks (knowledge distortions) and benefits (attaining public value) of knowledge linkage are discussed for both international and domestic administrations. Received: 16 May 2019 Revised: 31 December 2019 Accepted: 10 February 2020

behavior in representation that emphasizes knowledge sharing instead of patronage, but leads to similar outcomes: Their societal background provides officials with advanced social knowledge about the group(s) they represent, including both informational knowledge (facts about culture, history, politics) and relational knowledge (how people interact). Bureaucratic knowledge linkage is the process of sharing information and managing relations internally and with citizens. An extreme case serves to illustrate knowledge linkage empirically: Survey data from an international organization yield high levels of knowledge asymmetries within staff bodies and subsequent observation of knowledge linkage mechanisms. In generalizing findings, the risks (knowledge distortions) and benefits (attaining public value) of knowledge linkage are discussed for both international and domestic administrations.

| INTRODUCTION
There is increasing agreement that the creation of public value should be as much a normative guiding principle for public administration as the need for effectiveness and efficiency (Denhardt & Denhardt, 2015;Moore, 1995). Such an orientation toward the common good, however, requires the administration to know about a diverse people's concerns and preferences. This article asks how the knowledge bureaucrats possess due to their social origin affects administrative actions, and how recognition of such knowledge linkage alters our theoretical understanding of individual behavior in the public workforce.
Public Administration literature has long studied questions of diversity and how societal groups get reflected in the demographic composition of the public sector workforce. This literature, going back to the work of Kingsley (1944), is known as bureaucratic representation theory. It distinguishes a passive form of representation, in the sense of descriptive similarity between the demographic makeup of society and the administrative workforce, and an active form, which is about individual "representative" behavior and its substantive consequences. The focus of this article is on individual behavior and what happens during "active representation." A key assumption of active bureaucratic representation is that civil servants represent the interests of the societal group(s) they belong to, by actively trying to make these groups better off (Meier, 1993;Mosher, 1968;Selden, Brudney, & Kellough, 1998). There is a rich "macroscopic" literature on the societal consequences of such assumed workplace behavior, showing for instance less discrimination against minority groups as minority employment increases (e.g., Bradbury & Kellough, 2011;Grissom, Nicholson-Crotty, & Nicholson-Crotty, 2009;Meier & Nicholson-Crotty, 2006;Wilkins & Williams, 2008). But there is almost no individuallevel data to sustain claims about clientelistic behavior. Furthermore, active interest representation also violates core normative premises that guide most public service systems around the world, namely the Weberian idea of public servants who act neutrally and without regard for persons (sine ira et studio) as the apolitical tool of government (Weber, 1922(Weber, , 1968; see also Peters, Maravi c, & Schröter, 2013;Seibel, 2010).
This article aims to build theory by proposing an alternative understanding of individual behavior in bureaucratic representation. The core argument is that individuals "represent" by brokering knowledge about societal groups, rather than engaging in clientelistic behavior. The article draws on ideas and evidence from literature on organizational boundary spanning and street-level bureaucracy (Durose, 2009;Quick & Feldman, 2014) as well as bureaucratic representation in non-Western states (Fernandez, Koma, & Lee, 2018) and international organizations (Badache, 2019;Christensen et al., 2017;Gravier, 2013;Murdoch, Trondal, & Geys, 2016) to theorize bureaucratic knowledge linkage as the process by which bureaucrats-who possess advanced knowledge about a social environment which is affected by policy-share information and broker relations between their organization and that environment. Building on basic ontological categories, knowledge linkage is conceptualized as consisting of four mechanisms, differentiated according to the nature of knowledge that is being shared (objective informational or subjective relational knowledge) and whether linkage affects the administration or its environment: (a) the gathering of information/intelligence, (b) the dissemination of information, (c) the mitigation of coworker prejudices, and (d) the facilitation of local access. The article proposes that knowledge linkage by individual bureaucrats constitutes a more accurate understanding of dominant individual behavior in a diverse bureaucracy than active representation as clientelism.
An extreme case setting serves to illustrate linkage empirically. While social knowledge asymmetries constitute a general phenomenon in any administration that serves a diverse people, they are extreme in an international public administration (IPA)-that is, the bureaucracy of international organizations such as the United Nations (Bauer, Knill, & Eckhard, 2017;Eckhard & Ege, 2016;Rittberger, Zangl, Kruck, & Dijkstra, 2019;Trondal, Marcussen, Larsson, & Veggeland, 2010). IPAs are most well-known for the implementation of international development assistance, humanitarian aid, or peace operations, to name just a few examples. In such settings, globally recruited international bureaucrats are expatriates who lack knowledge about host state societies and their people. This is why international organizations also recruit national staffers from the area where they set up office. These national staffers "represent" local citizens in an international administration. They can do so in a clientelistic sense, but they are also a most-likely case for knowledge linkage between international organizations and the national environment.
Observational data are drawn from a staff survey conducted by the United Nations (UN) secretariat in 2012 of approximately 20,000 civilian staff working in 16 UN peace operations. The analysis demonstrates that national staffers are the most crucial intermediary of local knowledge and that they have access to different and apparently more important knowledge sources than their international colleagues. Furthermore, analysis of several hundred narrative open question responses illustrates the presence of all four knowledge linkage mechanisms.
Knowledge linkage by national staffers in implementing IPAs is not surprising per se. In fact, it is often even part of formal job descriptions. But it is precisely the nature of a most-likely case that a presumed mechanism is more pronounced (Levy, 2008). In generalizing findings beyond the UN/IPA context, the article offers three theoretical contributions: First, an administration that seeks to orientate toward the common good can know about the interests and preferences of a diverse people if it represents societal groups. Knowledge linkage by representing bureaucrats bridges the gap between citizens and the state.
Second, knowledge linkage constitutes an alternative mechanism of bureaucratic representation. It belongs to representative bureaucracy theory because it captures the substantive effects of a descriptively (passively) representative bureaucratic workforce. But it is also distinctive from active representation because it does not necessarily imply a "minority representative role" (Selden et al., 1998, p. 717) or that officials "press for the interests and desires" of those represented (Mosher, 1968, p. 14). Thus, whereas "active" representation is at odds with a Weberian understanding of Public Administration (Meier, 2018, p. 42), linkage representation as knowledge brokerage is not.
Third, advanced social knowledge constitutes a source of power and misuse of such power a risk for organizations. The article explores the risks and benefits of knowledge linkage and the moderating conditions under which knowledge linkage may turn into interest representation that harms public value rather than creating it.
The next section reviews the literature on representative bureaucracy and proposes a conceptualization of social knowledge and knowledge linkage. Section 3 discusses case selection, data, and methods. Section 4 presents and discusses the empirical data. Section 5 offers a discussion on the generalization of knowledge linkage theory. "without regard for persons" (Weber, 1968, p. 975) or "sine ira et studio" (without anger and passion, Weber, 1968, p. 975). The idea that public service employees should treat all citizens as equal, irrespective of their demographic background, has become an important legal principle that governs most democratic administrative systems around the globe (Peters et al., 2013;Seibel, 2010). These ideas, however, seem to be at odds with an entire subfield of Public Administration literature-namely bureaucratic representation theory (for a discussion on this, see Meier, 2018, p. 42). The theory suggests that bureaucrats' beliefs or values shaped through socialization and demographic backgrounds affect their job-related actions (Krislov, 1974;Mosher, 1968). If a public service organization achieves descriptive similarity between its workforce and the demographic makeup of society (representing all groups), bureaucrats are expected to act in a clientelistic way by serving their communities. The resulting macro-effects, such as less discrimination against minorities, have been frequently observed. But there is very little empirical evidence regarding individual behavior. Several recent studies nurture doubts about the dominance of interest representation. Based on these, an alternative understanding of individual behavior in representation is proposed below; a mechanism that causes similar macro-effects as previously reported but does not imply individual favoritism and is thus more in line with the legal reality of most administrative systems and also Weber's ideas.

| State of the art in representative bureaucracy literature
The original idea of representative bureaucracy goes back to Kingsley (1944) who wrote about the integration of the working class in the 20th-century British civil service. The theory argues that "a bureaucracy recruited from all segments of societies will produce policies that are democratic in the sense that they are generally responsive to the desires of the public" (Meier, 1993, p. 2). For conceptual clarification, later studies distinguished a passive and an active form of representation. Passive representation refers to the condition of descriptive similarity between the demographic makeup of society and the demographic makeup of the public service. Studies highlight that such passive representation can come with symbolic values, such as an increased acceptance of agency actions among all societal groups or a perceived equality of access (Gade & Wilkins, 2013;Meier & Nicholson-Crotty, 2006;Riccucci & Ryzin, 2017).
Active representation refers to individual behavior in representation. It is the main mechanism by which a passively representative bureaucracy, through individual action, is thought to become more responsive to the desires of the public. Under active representation, representing bureaucrats act in the interests of their clients by making these clients better off (Meier, 1993;Selden et al., 1998). As Mosher wrote, active representation is the process by which "individuals (or administrators) are expected to press for the interests and desires of those whom they are presumed to represent, whether they be the whole people or some segment of the people" (Mosher, 1968, p. 14). And Thompson (1967, p. 203) specified that "behavior actively represents a (…) community when it increases the wealth, prestige, or other advantages associated with belonging to that [community]." Whereas active representation implies strong claims about individual behavior, such behavior is in fact rarely observed empirically (for an important exception, see Selden et al., 1998). Instead, following up on Meier's (1993) groundbreaking work, a large number of studies analyze administrative outputs at the macro-level. Researchers typically match variation in passive representation, such as minority employment shares, with variation in the extent to which administrative outputs benefit minority groups. These outputs are assumed to be caused by minority bureaucrats who actively ensure that the interests of their groups are accounted for (e.g., Bradbury & Kellough, 2011;Grissom et al., 2009;Meier & Nicholson-Crotty, 2006;Wilkins & Williams, 2008).
Recent studies lament an incomplete empirical picture. Kennedy (2014, p. 398;similar, Lim, 2006) argued that the scope of inquiry "has been limited". Nicholson-Crotty et al. (2016, p. 754) wrote that "studies examine only organisation-level data, and, at the aggregate level, they all produce identical findings." A literature review of empirical studies of the past 15 years confirms the "macroscopic" state of the art. Organizational-level data and quantitative covariational methods dominate. There is in fact not a single study with a qualitative, more mechanism-oriented research design at the individual level (Figure 1), which hints to a research gap about individual behavior in bureaucratic representation.
In addition to active representation as clientelism or favoritism, there could be other individual-level mechanisms that occur because of individuals' social background and affect administrative actions. Several studies nurture this claim. In a conceptual article, Lim (2006) distinguished between bureaucratic behavior that directly benefits minority groups and that which indirectly benefits those groups when representing bureaucrats alter the behavior of other bureaucrats. More recently, Riccucci and Ryzin (2017) showed that passive representation may foster whether clients and citizens cooperate and comply with government (see also, Gade & Wilkins, 2013;Meier & Nicholson-Crotty, 2006). Fernandez et al. (2018) found several mechanisms in a study on policy implementation by street-level administrations in South Africa. They showed that passively representative implementing offices perform better. While evidence for active interest representation was found, performance improvements were also caused by better communication, the mitigation of biases, and more positive citizen reaction. Gravier (2013) found that the EU Commission exploits the nationality of its civil servants to establish informal ties with national administrations, what she describes as linkage representation. The implications for bureaucratic representation theory is that there could be bureaucratic behavior that is tied to an individual's social background and affects the outputs produced by an administration, but is not active representation in the sense of clientelism.

| The role of social knowledge in bureaucracy: Insights from street-level bureaucracy
Social knowledge can be broadly defined as knowledge about societal groups, such as the concepts, ideas, and intellectual activities that give groups their identity, including both worldview assumptions and specific topics and issues (Eyerman & Jamison, 1991, p. 3;Tajfel, 1981). This definition is broad enough to apply to smaller groups, such as social movements, but also to religious groups or even nations. The possession of specific social (group) knowledge is what distinguishes members of a group from outsiders.
The existence of social knowledge as a distinctive category of bureaucratic knowledge is not well recognized. A century ago, Max Weber only distinguished technical knowledge (Fachwissen) and knowledge bureaucrats acquire through the conduct of governance (Dienstwissen) (Weber, 1922, p. 129). Nowadays, bureaucracies in democratic systems are expected to serve the people (Denhardt & Denhardt, 2015) and to provide public value (Moore, 1995), which presupposes knowledge about what societal groups need and want. Apart from top-down political representation and bottom-up citizen participation, administrations can also gather such insights though informal knowledge sharing by individual bureaucrats in a representative public service.
Literature on street-level bureaucracy has been most vocal regarding the importance of local social knowledge (Lipsky, 1980). Durose (2009;Durose et al., 2016) explicitly recognized that frontline workers possess "high levels of 'local knowledge' built through their lived experience" and that they use such knowledge "in order to engage with the community" (Durose, 2009, pp. 985, 991). In organization research, Yanow (2004, p. 12) describes such "local knowledge" as "the very mundane, yet expert understanding of and practical reasoning about local conditions derived from lived experience". Literature on boundary spanners or brokers generally emphasizes that organizational boundaries can be porous and tenuous and that individual boundary spanners translate across social (group) differences (Quick & Feldman, 2014;Tushman, 1977).
Passive representation increases the range of social knowledge sources in an organization. Single employees should therefore engage in knowledge linkage activities on the basis of their advanced insights. Knowledge linkage refers to the process by which bureaucrats-who possess advanced knowledge about a social environment which is affected by policy-share information and broker relations between their organization and that environment. After specifying the associated mechanisms of knowledge linkage behavior in the next section, observable implications are derived.

| Conceptualizing knowledge linkage behavior
How exactly does knowledge linkage affect bureaucratic decision making and administrative outputs? To specify the exact mechanisms, a distinction must be made between the type of social knowledge and the direction of knowledge linkage: First, drawing on Karl Popper's (1978) ontological work on knowledge about the world, two basic categories of knowledge can be differentiated: the subjective world of individual experience, emotions and thoughts; and the world of objective knowledge, such as scientific theory. This distinction extends to knowledge about social groups, which can be separated into subjective relational knowledge regarding communication and the management of interpersonal relationships between individuals; as well as objective informational knowledge, such as facts about culture, history, politics, and the interests of social groups. Relational knowledge refers to the assumptions individuals have about how to interact with others and is most relevant when people from inside an organization directly engage with outsiders such as citizens. Informational knowledge may be also relevant in communicative settings as the content of exchange. But it is most distinctive when bureaucrats apply assumptions about the social world when they draft policies, concepts, standard operating procedures, or concrete action plans.
As for the second dimention, linking bureaucrats can have either an inward directed effect on the bureaucracy or it can be outward directed and affect the behavior of citizens. This distinction draws on boundary-spanning literature, which describes boundary spanning as two-directional, aiming at the organization itself or the environment (van Meerkerk & Edelenbos, 2014).
On the basis of these two dimensions, four mechanisms of knowledge linkage are distinguishable (Table 1): First, informational knowledge highlights that organizations are in constant demand of acquiring and sharing information. Internally, organizations tend to be stagnant, and employees tend to stick to established routines. Knowledge linkage, by contrast, implies the introduction of new ideas and lateral thinking (Dorado, 2005, p. 397) but also the ability to adjust policy or programs to distinctive local conditions. Externally, linkage enables the provision of outside actors with information about bureaucratic intentions and plans that should be important to raise the social acceptance of bureaucratic actions (Denhardt & Denhardt, 2015). This means that linking bureaucrats can either (a) gather information/intelligence about social groups or they (b) disseminate information from the bureaucracy to citizens. Second, relational knowledge highlights that bureaucracies need to acquire an understanding of people outside their own boundaries "to acknowledge and value difference in terms of culture, mind-set, profession, role and 'gaze'" (Williams, 2002, p. 110). In this vein, linking bureaucrats can use their knowledge about group customs and culture to (c) mitigate prejudices among their fellow bureaucrats (see also Lim's (2006) suggestion that bureaucrats can sensitize other employees)the internal effect-or they (d) facilitate access to outside communities to "create a new, shared domain" (Quick & Feldman, 2014, p. 677)-the outward effect. 1

| Empirical expectations
The basic observable components of knowledge linkage theory are summarized in Figure 2. The condition is that passive representation prevails, in the sense of administrative workforce diversity that mirrors salient demographic features of society. As a consequence, individual civil  Table 1 should be observable. This leads to the following propositions: Proposition (1): Under the condition of workforce diversity, substantive differences in possession of social knowledge prevail between bureaucrats; Proposition (2): Bureaucrats with advanced social knowledge engage in knowledge linkage behavior (four knowledge linkage mechanisms).

| DATA AND METHODS
In methodological terms, extreme or most-likely cases constitute ideal settings for theory building because the social mechanisms of interest are most pronounced (e.g., Levy, 2008, pp. 5, 12). While knowledge asymmetries constitute a general phenomenon in any administration that serves a diverse people, they are extreme in an international public administration (Bauer et al., 2017;Eckhard & Ege, 2016;Rittberger et al., 2019;Trondal et al., 2010). At the country level, where IPA officials interact most frequently with citizens, they implement a large range of activities in development assistance, humanitarian aid, or peacekeeping, such as technical assistance, mentoring and advising, or the provision of public safety in the case of UN police or military peacekeepers.
In terms of representation, the most salient group cleavage in IPA country offices is between international and national officials. The international staff comprises individuals of many nationalities who are usually foreigners when deployed to a country office. They should lack local social knowledge. The national staff, by contrast, consists of local residents with a career linked to the hiring office. National officials "represent" their people, but they also possess local social knowledge that is indispensable for an IPA to adjust to the local context (Eckhard, 2019;Parízek, 2017). National staffers should consequently engage in knowledge linkage.
Among the various policy fields of international organizations, this study looks at peacebuilding, which is the major operational activity of the UN secretariat. In 2012, around 20,000 civilian staff worked in 16 UN peace operations alongside military peacekeepers. 2 International officials in UN peace operations are formally called the international professional staff (IPS). They are substantive peacekeeping experts, such as civil affairs officers or human rights specialists. Once incumbents go through more junior positions (P1-P3), they assume management and leadership tasks (P4-P5 and D1-D2) (CEB/2014/HLCM/HR/21, Table 6A). By contrast and as shown in Table 2, locally recruited officials fall into two categories: national professional F I G U R E 2 Observable elements of knowledge linkage officers (NPOs) and general service (GS). NPOs perform substantive tasks such as political affairs officer, legal officer, medical officer, or humanitarian affairs officer, usually under IPS supervision. GS staff fulfill technical or support functions, such as administrative, secretarial, and clerical tasks (G1-G3); higher paygrades also perform substantive tasks, such as program, research, and community liaison assistance (G4-G5). 3 To observe linkage empirically, responses to an internal staff survey conducted by the UN secretariat in 2012/13 are analyzed. 4 The survey designers were interested in how peace operations gather information on "local perceptions," defined as "the ways in which situations, events and dynamics relevant to the conflict, the peace process or the peacekeeping mission's mandate are regarded by local people" (DPKO/DFS, 2013, p. 11). This definition comes close to social knowledge as understood here.
The survey was sent to all civilian peacekeeping staff in November 2012. Survey participants could skip individual questions which is why N varies between 306 and 2,061 for each question (online appendix 2). Respondents were predominantly male (73.3%) and served in a total of 16 peace operations (96.9%). Respondents are broadly representative of the three main staff groups in the UN system. 5 The below analysis is based on descriptive statistics. But to control whether distributions between groups are significantly different, standard nonparametric distribution tests (Kruskal-Wallis Test and Pairwise Wilcoxon Test) were conducted for each of the group categories (gender (Q1), mission unit (Q2), mission level (Q4), employment category (Q5), employment level (Q6) in combination with the questions analyzed (online appendix 3). 6 According to theory (see Figure 2), the empirical expectations are that (a) there should be substantive differences in the possession of or access to local social knowledge between international and national officials; and that (b) national staff should engage in knowledge linkage activities. Corresponding to these expectations, responses to two single-item survey questions are presented next: a closed question asking UN staffers about the knowledge sources they use in their daily work (Q14). This allows for scrutinizing of whether national staffers take on an outstanding role as knowledge source, or not. Analysis of answers to an open survey question about the impact of local knowledge on mission activities (Q40) allows for scrutinizing the presence of all four theorized knowledge linkage mechanisms. Narrative accounts of linkage mechanisms by respondents in their survey answers would classify as "smoking gun" evidence (Collier, 2011, p. 825).  (1): Under the condition of workforce diversity, substantive differences in possession of social knowledge prevail between bureaucrats The data illustrate that there are indeed crucial differences in the possession of local social knowledge among staff groups. As expected, national staffers seem to possess advanced social knowledge, both in terms of the knowledge they possess individually, but also in terms of their access to local actors that are relevant stakeholders of peace operations and peace processes. First, Figure 3 highlights the outstanding relevance of national staff as knowledge brokers by showing answers to the question, "Who are your interlocutors when gathering information on local perceptions?" For the majority of respondents, the national staff is the most relevant single information source. Sixty-six percent indicated that they refer to their national colleagues frequently or all the time. Official contacts with the local government rank only second. There are also statistically significant differences between staff groups as international staffers are more likely to ask national colleagues for local information, compared to other staff groups (online appendix 3).
Second, Figure 4 takes the same data presented in Figure 3 but compares the knowledge sources of international and national staff. To compare groups, the share of answers as displayed in Figure 3 was calculated separately for international (IPS) and local (GS + NPO) officials. Figure 4 displays the absolute difference between the sum of the values for "frequently" F I G U R E 3 Answers to the question "Who are your interlocutors when gathering information on local perceptions?" Source: Online appendix 2, Q14-A1-Q14-A22, multiple answers possible; N = 380. GS (34%) are slightly underrepresented and NPO (18%) overrepresented (35% international staff; 13% other). The answer option A15 on "National colleagues" was answered 361 times, with 13 (3%) answering "not at all," 101 (27%) "sometimes," 130 (34%) "frequently," 109 (29%) "all the time," and 8 (2%) "not applicable" and "all the time" for both groups. It shows that national staffers are much better linked to local security actors, inhabitants, and traditional leaders. By contrast, internationals interact much more with civil society organizations, local coordination bodies, people addressing them, as well as their own household staff. Bearing in mind that peace operations act in insecure and highly politicized conflict environments, the contacts of the national staff seem of much higher relevance for the execution of the mandate compared to the contact of international officials.
Overall, the data presented in Figures 3 and 4 clearly indicate differences in the substantive knowledge sources of staff groups. Linkage by national staffers should therefore provide their organization with much deeper insights on local social knowledge than what internationals could obtain alone. The next question therefore is whether it is possible to also observe empirically the actual knowledge linkage mechanisms as theorized earlier.

| Proposition (2): Bureaucrats with advanced social knowledge engage in knowledge linkage behavior (four knowledge linkage mechanisms)
The expectation is that national staffers engage in knowledge linkage behavior. Answers to an open survey question (Q40) provide such insights. Respondents were asked to reflect, in their own words, on situations in which awareness of local perceptions positively or negatively affected their work. All 306 answers were coded with respect to one of the four mechanisms of linkage. 7 Beginning with the dimension of relational knowledge linkage, to which 107 statements predominantly referred, below paragraphs discuss each of the four knowledge linkage mechanisms. Figure 3 answer shares between national (<0) and international (>0) staff respondents.

F I G U R E 4 Difference in
Source: See Figure 3 The first mechanism is the mitigation of coworker prejudices (50 statements). International staffers in UN peace operations are typically experts with a lot of cultural training. But they are still foreigners to the countries in which they work. One international police officer who was deployed to the UN mission in Liberia explained the problem: "My culture is not at all the same than in Liberia. Therefore, even if as Police Officer I am dealing with Police Officer, it is vital to take into account their perceptions for all activities I am developing with them. If not I will for sure fail. What is working in my country might not work here …" (ID: 634884007870949861). One national official in the UN mission in South Sudan additionally wrote that "if the international staffs don't respect the diversity and the people of the host nation it will directly contribute to misunderstanding and turn the local citizen against any task to carried out by the mission" (ID: 634883961742344692). Another national staffer wrote that "ignoring the perceptions can lead to a big misunderstanding as I personally witnessed fight and physical assault on International Staff" (ID: 634883966303669698). National staffers who know the mentality and political sensitivities of conflict countries are therefore crucial to mitigate biases among their international colleagues, who tend to hold the more senior leadership positions and decisionmaking powers. In a "smoking gun" example, one international staffer from the UN mission in DR Congo explained how local colleagues affected his cultural awareness and behavior: "Because CLAs [a national staff category] told me that greetings are done lengthily and I should shake the hand of everyone and I always do it, I was well liked by the local people and they told me they are happy with me" (ID: 634884147995288259). Another agreed that "[b]eing aware of local customs always has a positive impact in my work … It enables us to adapt our interactions to achieve a better contact with population at a local level" (ID: 634885023766393031, UN mission in Côte d'Ivoire).
The second relational linkage mechanism refers to interpersonal communication that affects missions' ability to act in a local environment, what is termed the facilitation of local access (57 statements). Many of the communication obstacles mentioned in the survey responses are cultural, having to do with the traditional ways in which local people interact. One national staffer who worked as interpreter for the UN mission in DR Congo mentioned a relatively banal example of everyday interactions on the streets, in which a local citizen wanted to give the military patrol he accompanied information about a hidden marijuana field: "As I know the mentality of people to share some confidence, I request the team leader if the team can contribute to give 10 USD as a reward. Four of us agreed to give money. And the guy opened his heart and lead us to the hidden place to find the marijuana field" (ID: 634883972786994568). Yet another example was given by an international staffer at the UN mission in Côte d'Ivoire, who explained how his national staff team member helped facilitating local access in a police investigation on sexual crimes: Knowing that the area we went in was against UN thinking we were partisan of the new regime. For the culture knowledge I took my assistant who is Ivorian and who speaks the same language to break the ice. At the end after multiple talks the rape case which was being concealed amicably was reported to us and the victim ended up filing the case to police and the perpetrator being arrested… (ID: 634883966328200477).
The second knowledge dimension is informational knowledge. Other than in relational knowledge that is about interpersonal relations and communication, associated mechanisms refer to abstract information that is relevant for bureaucratic concepts as well as outside perceptions. Overall, 140 respondents (56.7%) mentioned the two associated mechanisms in their answers.
First, respondents in the category of information gathering (93 responses) referred to facts or events that are relevant for mission planning or conduct. Respondents emphasized that missions must know about "traditional land and boundary disputes in rural communities" and "local traditional secret society festivities in local communities" (national staff respondent from the UN mission in Liberia; ID: 634886493822214692). But local information can also be crucial for immediate mandate implementation. For example, the UN mission in DR Congo has a mandate to protect civilians. In one case, a local employee had provided information about a location where civilians were at risk that enabled military peacekeepers to act: The CLA [community liaison assistant, the author] gathered information on fighting between Raia Mutomboki local combatants and the armed forces of the DR Congo in Shabunda Territory. This incident provoked a mass displacement of civilians to the neighboring Nzibira locality. Once alerted on this incident, the CLA shared the info with the Contingent which decided to carry a patrol towards the axis (ID: 634883970346410038).
In another case, a national employee in the UN mission in Darfur reported a "smoking gun" example in which he prevented the mission to interfere in a domestic political struggle. The context was an international colleague who had planned to conduct a capacity-building activity with members of parliament who, at the time, led an investigation against a local politician that had attracted much media attention: "I advised to postpone the activity until the issue was sorted out to avoid that the capacity building of oversight actors would be misperceived as a politically-motivated activity" (ID: 634884022029153836). And another example shows how a mission without reliable social knowledge sources was manipulated by one side of a local conflict. An international official from the UN mission in Western Sahara detailed a situation in which "biased information [was] provided by the Moroccan government on where we could rent housing to ensure that we were not mixing with the local Saharoui Population but were held under the control of the Moroccan security forces" (ID: 634883988595472537). It becomes clear from these examples how important local knowledge is for an international presence. In such politicized environments, even the decision where to set up offices has mandate-related implications. Without the possibility to triangulate information from as many sources as possible, missions are severely restricted in their ability to execute a mandate.
The second linkage mechanism is the dissemination of information (47 respondents) from missions to local communities. In one example, an international official at the UN mission in East Timor referred to a series of Townhall events that were organized by national staff, to distribute information on upcoming elections in the country: "The large amount of local staff in UNMIT through the relevant staff associations held discussions with the leadership and regular Townhalls were held. All in all, this contributed to open dialogue and information flow related to the elections" (ID: 634888875470419081). In another example for the UN mission in South Sudan, a national official details how they helped reducing prejudices about the international presence. It shows how preposterous the misconception in the local population can sometimes be and how crucial it is for missions to effectively communicate about their true intentions: People were think that peacekeepers are brings HIV/AIDS to the people of the new nation South Sudan, but after some time people realize that peacekeeper bring a very good policy of preventing HIV/AIDS, and stigma and discrimination … when the local staff went out to train the local community (ID: 634884028557958296, original language).
The above examples show that national staff engage both in relational and informational linkage. This goes both ways: The local people trust the national staff because they perceive them as their kin. This enhances access for an outside peace mission and it facilitates information dissemination. And if international officials rely on their local colleagues, they gain a crucial information source and cultural ambassadors who can improve personal interactions between civil servants of an international organization and the local people. All these examples highlight the potential of knowledge linkage to be instrumentally valuable for the UN itself and, to the extent that the UN serves the interests of local societies, to also enhance public value.

| Discussion of findings and implications
The case of UN peace operations constitutes a most-likely setting for social knowledge asymmetries in the administrative workforce, depending on whether officials are host state nationals or come from abroad. The expectation was, first, that the two staff groups should differ in the possession of local social knowledge; and second, that those familiar with the local environment should engage in linkage activities.
Consistent with expectations, the first finding is that knowledge differences exist. International officials lack local social knowledge and national officials fill the gap. They are also much better connected to critical local information sources, such as security actors. Linkage is not only executed by those the UN actually employed for the very purpose of information gathering (National Professional Officers with substantive functions). Linkage activities could also be attributed to General Service support staff, such as interpreters and drivers. In background interviews, international officials said that it is more the extent of interaction between staffers than the job profile that defines knowledge linkage activities. Drivers and household staff are often mentioned as important local interlocutors. 8 The second finding is that all four previously conceptualized linkage mechanisms could be observed. The open survey question offered a unique opportunity to analyze "smoking gun" evidence from many mission environments. The answers support the expectation that linkage refers to both informational and relational knowledge and is two-directional in nature. It is important to bear in mind that the advantage of interpreting open survey question answers is nonreactivity, that is, that questions "do not cue respondents to think of particular causes or treatments" (Iyengar, 1996, p. 64) that risks overestimating latent effects. Without being offered a precise definition of the term "local perceptions," respondents reported in their own words how local knowledge positively or negatively affects their work. All (meaningful) narrative answers could be attributed to one of four linkage mechanisms, which suggests that the concept is comprehensive.
A limitation is that the data do not allow to systematically gauge the relative impact of linkage mechanisms on administrative outputs. Simply put, it does not show whether linkage serves the common good and increases public value. But the data provide some insights on this question, indicating where future research should be headed: All above examples highlight the potential of knowledge linkage to be of value for the UN administration and its mandate. Relational aspects of knowledge linkage improved communication and prevented misunderstandings on cultural ground. Informational aspects of knowledge linkage enhanced operational awareness and enabled missions to act upon mandated tasks. For instance, in the Darfur example, advice by a national staffer helped preventing the UN to get caught up in a domestic power struggle; and in South Sudan local staffers helped to correct rumors and misconceptions about the activities of the UN presence.
Despite their beneficial potential, these examples also indicate that there may be a fine line between linkage that serves the interests of the UN and ultimately enhances public value-and the risk that individuals exploit knowledge asymmetries to selectively share social knowledge, or to alter its content. In such cases, rather than to occur instead of active interest representation, linkage would enable clientelism or patronage.
On the one hand, none of the survey data and narrative accounts collected among all UN peace missions included instances of clientelism, despite a question that asked directly about negative effects of local perceptions. On the other hand, however, the UN itself seems to be wary of such risks. An internal guidance document warns that "relying solely on a limited number of established contacts can restrict missions' ability to understand the broad spectrum of local perceptions and can result in bias" (DPKO/DFS, 2014, p. 65). And a previous study by this author on UN peacebuilding in Kosovo reports a setting in which local staffers from one ethnic group refused to implement affirmative policies designed to serve another (opposed) ethnic group (Eckhard, 2014). But that study also highlights the UN's failure to employ sufficient numbers of staffers from all ethnic groups, thereby risking one-sided action.
Whether or not linkage enables clientelism in a systematic manner seems to depend on the way administrations handle their internal knowledge asymmetries. A first relevant moderator of knowledge linkage is which "locals" actually get represented in the national staff. Language and educational requirements mean that local staffers tend to be recruited from national elites whose knowledge does not necessarily represent all locals. There may also be biases when missions fail to employ all societal groups equally, such as ethnic communities. Survey data hint that equal recruitment is important, not only to tap into a diverse array of perspectives, but also for the acceptance of missions among local populations as one national staffer in the UN Mission in South Sudan reports: "Whilst on mission in Africa, I realized that employing local staff had a big impact on how our work would be done, by this I mean the different tribes or groups, by hiring too many of one group would alienate another which would hamper our operations in the field" (ID: 634884026657824742).
Staff relations seem to be another moderator of knowledge linkage. In several instances, national staffers criticized that their international colleagues did not take their opinions seriously. For instance, an employee at the UN mission in Congo complained that "repeated claims, including one from a senior national staff, have been largely underplayed or ignored entirely" (ID: 634884037350471180); and a national employee at the UN mission in Haiti criticized that "[m]ission management ignore most of the times inputs on local perceptions and rather trust their sources, that are normally based on opinions given by ex-pats or diplomats close to the senior administration" (ID: 634884110029639948).
Overall, linkage promises crucial benefits for the operations of international organizations who deploy expatriate staff to countries about which these people lack informational and relational knowledge. Knowledge deficits can be compensated through employment of national staffers who engage in knowledge linkage. There are, however, important moderators, including those mentioned earlier (recruitment practices, staff relations), but also whether organizations employ alternative knowledge sources and learning tools (Benner, Mergenthaler, & Rotmann, 2011) or the ability of country operations to act autonomously upon local information in the first place (Campbell, 2018).

| CONCLUSION: BUREAUCRATIC REPRESENTATION AS KNOWLEDGE LINKAGE
This article theorized knowledge linkage as the process by which bureaucrats-who possess advanced knowledge about a social environment which is affected by policy-share information and broker relations between their organization and that environment. It is perhaps of little surprise that there are such knowledge differences between international officials and national staff groups in UN peace operations. Thus the question is, whether the theory also applies to other areas in which international public administrations act and, what is more, to domestic administrations more generally.
It is theoretically plausible to answer these questions in the affirmative. Linkage should be important for other international organization tasks and policy fields. Peacekeeping clearly is a distinctive activity in the sense of the attached security risks and the need for precise military intelligence and good relations with local populations. But this does not mean that peacekeeping is specific when compared to other international organization field level tasks, such as development assistance or humanitarian aid. Quoting Séverine Autesserre (2014), who asked why foreign interventions in general achieve very little despite tremendous financial and personnel resources, a major problem is their "imposition of foreign ways of thinking and working" (Autesserre, 2014, p. 98) and an "attendant disregard of local ideas" (Autesserre, 2014, p. 97). Knowledge linkage by national staffers offers a potential solution in that regard. Furthermore, linkage should be relevant when international organizations prioritize policy-formulation, rather than implementation. In such settings, however, the inward mechanisms of linkage (gathering information/intelligence and mitigating bureaucratic prejudices) bear greater relevance than the outward dimension (cf. Gravier, 2013).
Linkage should also be relevant in domestic settings. Knowledge deficits about societal groups prevail in any bureaucratic organization that serves a diverse people. Imagine a minority group school teacher who tells other teachers at her school how to adjust lectures to the cultural background of ethnic minorities, thereby improving their learning success. Or, imagine a female police officer who anticipates the ambiguities of domestic violence against women that causes victims to respond more accurately to a police investigation, leading to more arrests in such cases. Or consider a transport department planning a highway through an entire country; the ministry lacks knowledge about and access to local residents, but bureaucrats who originate from these regions possess just such information and personal contacts, which facilitates the planning task. Examples such as those represent everyday challenges in public administration which is why knowledge linkage should be highly relevant in any domestic administration, providing an alternative causal pathway to explain the macro effects observed in bureaucratic representation studies on the link between minority teacher employment and minority pupil grades (for instance, Roch and Edwards, 2017) or on the link between women police officer employment and domestic violence arrest rates (for instance, Andrews and Johnston Miller 2013).
This finding has implications for literature on bureaucratic representation that tends to equalize the process of active representation with the observation of administrative outputs at the macro level, such as benefits for certain social groups. Instead, active representation is about individual behavior and there is very little direct empirical evidence that clientelism, favoritism, or patronage indeed prevail systematically in administrative systems around the world. Yes, social origin makes a difference for how a task is executed. But these differences may be more related to the social knowledge bureaucrats possess, rather than individual attempts to cause group benefits. Knowledge linkage theory does not challenge the basic existence of active interest representation, but it suggests that its prevalence may be significantly less pronounced than previously assumed.
Whether the same macro effect (better services for minority groups) is caused by active interest representation or knowledge linkage should make a huge difference for practitioners. After all, clientelistic behavior by civil servants goes directly against the ethical and legal frameworks governing democratic administrative systems. By contrast, bureaucratic representation as knowledge linkage is not only in line with Max Weber's ideas on bureaucracy, it also highlights that administrations could attain public value by increasing social group diversity in the civil service. It would be important, therefore, to study the mechanisms of knowledge linkage in a broader range of empirical settings, bearing in mind in particular the factors moderating the impact of linkage behavior on substantive decisions.
ACKNOWLEDGMENT I thank Marco Donati at the UN Department for Peacekeeping Operations for sharing the data used in the empirical part of the article; as well as several other officials in international organizations who provided background information. Previous versions of this article have been presented at the 2017 ECPR conference in Oslo, the 2017 MPSA conference (host of Ken Meier's representative bureaucracy workshop) in Chicago, and the 2018 representative bureaucracy workshop at the University of Adger, Norway, among others. For their comments and help during these and other occasions I owe my thanks to Magali Gravier, Zuzana Murdoch, Fanny Badache, Christoph Knill, Ken Meier, Bernhard Zangl, Wolfgang Seibel, Vytautas Jankauskas, Alexa Lenz, Ronny Patz, Yves Steinebach, and three anonymous reviewers from Governance. I would also like to thank all members of the DFG-funded Research Unit International Public Administration (www. ipa-research.com) for inspiration, new insights, and great discussions.
ORCID Steffen Eckhard https://orcid.org/0000-0002-5320-0730 ENDNOTES 1 Knowledge linkage puts in a broader context some of the new mechanisms previously proposed in bureaucratic representation literature. For instance, Fernandez, Koma and Lee (2018, p. 13 f.) observe five mechanisms by which representing bureaucrats cause performance effects: (a) favoritism toward their community, (b) a more empathic understanding of local concerns, (c) better communication, (d) mitigating biases and discrimination, and (e) better citizen reaction. Apart from (a), which describes interest representation, the others dovetail neatly with knowledge linkage mechanisms. 2 Alongside peacekeeping as a primarily military task, civilian staff in peace operations contribute "to reintegrating former combatants into the civilian economy; strengthening the rule of law (…); providing technical assistance for democratic development (…); and promoting conflict resolution and reconciliation techniques" (S/2000/809, paragraph 13). 3 At field-level duty stations, the overwhelming majority of GS staff are host state nationals. Background telephone interviews with an UN human resource expert on February 27, 2018 and an expert in the Department for Peacekeeping Operations on February 27, 2019, both based in New York. 4 The data and survey questionnaire were generously shared by Marco Donati, one of the survey designers, upon e-mail request in 2017. The views expressed in the survey responses represent exclusively the personal opinion of each UN staffer. 5 With 31% IPS, 11% NPO, 46% GS, and 13% others. GS are thus slightly underrepresented (54% in the entire UN system, see Table 2). The category "others" includes mainly military and police staff as the open answers indicate. These are higher ranks and logistics officers who obtained an official e-mail address as part of their duties. 6 Reproduction data can be retrieved from the author's personal website. 7 There were 306 narrative answers of varying length. Fifty-nine answers could not be classified, either because the response consisted of text fragments or because respondents wrote only about the importance of local social knowledge without explaining how it impacts their work. The remaining 247 answers were coded with respect to one of the four mechanisms of linkage. Although the question did not ask directly about national staff, they are clearly identifiable as acting subjects in 30 responses. The remaining 217 answers also report linkage mechanisms, only that the acting individual could not be identified from the text because respondents referred to "their mission" or "their unit." Overall respondents to Q40 are 40% international staff, 32% GS, 14% NPO, 14% other; respondents mentioning national staff are 31% are international staff, 41% GS, 28% NPO and 0% other.