Coordination at the Edge of the Empire: The Delegation of Headquarters Functions through Regional Management Mandates Journal of International Management

In this paper, we conceptualise the role of headquarters not as bound to a unique physical location but as a set of functions distributed across the multinational enterprise. Specifically, we investigate the phenomenon of delegating headquarters functions to local subsidiaries in the form of regional management mandates (RMMs), a new concept not addressed in prior research. Our work contributes to the literature by building a comprehensive typology of headquarters functions performed at various levels of the multinational enterprise and by developing propositions about the potential benefits and risks of using RMMs, drawing on various theoretical perspectives and a rich empirical case study. Our findings suggest that the use of regional management mandates offers several benefits for the MNE: (1) the ability to balance integration and responsiveness at levels below the efficient scale for dedicated regional headquarters; (2) the exploitation of local operational expertise on a regional level; (3) relieving headquarters of the burden of monitoring remote peripheral agents. At the same time, we also find evidence of risks and dangers associated with the use of regional management mandates. Our study develops a strong foundation for future research on the distributed nature of roles and functions of headquarters, while providing a novel perspective on how headquarters functions are enacted at the regional level. about the potential benefits and drawbacks of using regional management mandates. Our findings suggest that RMMs can offer considerable advantages over establishing dedicated RHQs in small regions: (1) balance between integration and responsiveness at levels below the efficient scale for dedicated RHQs; (2) the exploitation of local operational expertise on a regional level; and (3) relieving headquarters of the burden of monitoring remote peripheral agents. At the same time, there are also potential risks in using RMMs to manage ‘ satellite ’ subsidiaries at the edge of the MNE's reach: (1) the question of legitimacy and acceptance of operational subsidiaries as intermediate authorities; and (2) issues arising from the lack of visibility of ground-level organisational tensions at the top of the MNE. Through our exploration of these issues, we provide a potentially fruitful new perspective on how HQ functions are implemented at the regional level, and contribute to the discussion about the core role and purpose of corporate headquarters in the contemporary MNE. implications for regional management warrant closer investigation. We discuss these next. (1) the question of legitimacy and acceptance of operational subsidiaries as intermediate principals; and (2) issues arising from the lack of visibility of ground-level organisational tensions at the top of the MNE. These findings serve as a reminder of the importance of the concepts of authority and visibility, which are traditional hallmarks of headquarters and regional headquarters as actors in the overall corporate network. They also highlight potential avenues for further investigations, such as an empirical comparison between the functions of dedicated RHQs and operating subsidiaries with RMMs; an examination of RMMs from a micro-political and power perspective; as well as a sensemaking and sensegiving approach to RMMs.


Introduction
In the past decades, multinational enterprises (MNEs) have increasingly come to be seen as differentiated networks where the roles of individual units vary according to their specific capabilities and the internal and external networks they are embedded in (Andersson and Forsgren, 2000;Doz and Prahalad, 1991;Ghoshal and Bartlett, 1990;Nohria and Ghoshal, 1997). Within this network view, much of the focus has been on subsidiaries and their increasingly complex roles inside the MNE (Bartlett and Ghoshal, 1989), but in recent years, we have begun to witness a growing scholarly interest in the role played and value added by headquarters (HQ) (Ambos and Mahnke, 2010;Andersson and Holm, 2010;Egelhoff, 2010).
Despite this growing interest, few studies have examined in a systematic manner what headquarters actually do, and conceptualisations of HQ roles have yet to be linked to specific activities performed. The HQ literature also appears somewhat disconnected from the literature on regional headquarters (RHQs), despite the fact that both of them address the same core question: how to effectively govern complex, differentiated organisations operating in complex, differentiated environments. While progress is being made in connecting these perspectives by arguing that HQ functions can be performed at various levels of the MNE (Goold and Campbell, 2002;Kähäri et al., 2010;, the regional management literature remains largely under-theorised and short on detail about how HQ functions are actually performed at the regional level. In particular, little attention has been paid to the nature and characteristics of units that perform HQ functions at a regional level-for example, whether these units are fully dedicated to their regional role or are required to perform other roles as well-and how such characteristics may affect their ability to fulfil their roles effectively. In this paper, we build on the view that in complex MNEs, HQ functions may be distributed across the organisation and are not necessarily performed by dedicated RHQ units only (Lasserre, 1996;Schütte, 1997). We examine the delegation of HQ functions to local operating subsidiaries, a phenomenon that we first encountered in the field and termed regional management mandates (RMMs). Despite being used by MNEs in various parts of the world, this phenomenon has yet to be properly addressed in the literature. We believe there are strong reasons why RMMs should be treated as conceptually different from dedicated RHQs: in particular, their potential cost and efficiency advantages over RHQs. We define an RHQ as an administratively focused entity which dedicates all of its time and resources to performing regional HQ roles. In contrast, a RMM denotes a mandate given to a profit-oriented local subsidiary that is mandated to dedicate some of its time and resources to performing regional HQ roles. We argue that this difference in focus is likely to have an impact on how delegated HQ functions may be performed and warrants further investigation. Our paper explores two research questions, using a case study of Unilever and its cluster of tightly coupled subsidiaries in Hungary, Croatia and Slovenia: RQ1: What are the key functions of headquarters, and to what extent can they be delegated to operating subsidiaries in the form of regional management mandates? RQ2: What are the benefits and drawbacks of delegating these functions to operating subsidiaries, as opposed to dedicated regional headquarters?
Our study makes two major contributions. Firstly, we develop a typology of HQ functions performed at three different levels of the MNE, thus extending existing approaches-which have tended to focus on a single level-into a more integrative classification. Secondly, we draw on three interlinked theoretical perspectives as well as a rich empirical case to develop propositions about the potential benefits and drawbacks of using regional management mandates. Our findings suggest that RMMs can offer considerable advantages over establishing dedicated RHQs in small regions: (1) balance between integration and responsiveness at levels below the efficient scale for dedicated RHQs; (2) the exploitation of local operational expertise on a regional level; and (3) relieving headquarters of the burden of monitoring remote peripheral agents. At the same time, there are also potential risks in using RMMs to manage 'satellite' subsidiaries at the edge of the MNE's reach: (1) the question of legitimacy and acceptance of operational subsidiaries as intermediate authorities; and (2) issues arising from the lack of visibility of ground-level organisational tensions at the top of the MNE. Through our exploration of these issues, we provide a potentially fruitful new perspective on how HQ functions are implemented at the regional level, and contribute to the discussion about the core role and purpose of corporate headquarters in the contemporary MNE.

The roles, functions and tasks of headquarters in the MNE
In early MNE research, corporate HQ was thought of as the default locus of hierarchical decision making and control (Bower, 1970;Chandler, 1962), while the specific functions of HQ were rarely examined in depth. More recently, MNEs have been characterised as complex differentiated networks, explicitly recognising that many of their strategic capabilities and resources reside at the subsidiary level (Doz and Prahalad, 1991;Ghoshal and Bartlett, 1990;Nohria and Ghoshal, 1997). This has inevitably raised questions about what value is added by HQ (Ambos and Mahnke, 2010;Egelhoff, 2010;Tallman and Koza, 2010). A review of the literature reveals a variety of perspectives: HQ may be viewed as a performer of basic corporate functions such as legal representation and governance; an orchestrator of resources, knowledge and attention; a manager of information, employees and conflict resolution; and overall coordinator (Ambos and Birkinshaw, 2010;Ambos and Mahnke, 2010;Bartlett and Ghoshal, 1989;Ciabuschi et al., 2010;Ciabuschi et al., 2011;Collis et al., 2007;Dellestrand and Kappen, 2011;Doz and Prahalad, 1984;Foss and Pedersen, 2002;Goold and Campbell, 2002;Hamel and Prahalad, 1983;Lai Hong et al., 2006;O'Donnell, 2000).
The HQ's overarching purpose (administering the enterprise as a whole and maintaining continued profitability) is achieved by fulfilling two broad roles: a 'positive' entrepreneurial role focused on value creation and a 'negative' integrative role focused on loss prevention (Chandler, 1991;Foss, 1997). Many scholars note the dual roles of HQ and also cite various HQ tasks as exampleshowever, the latter tend to be aggregated into the former anecdotally or using a broad brush. What is missing is a mid-level link between the two roles and the plethora of tasks, which would allow us to categorise HQ activities more systematically. Table 1 represents a step towards closing this gap. Here, we synthesise the existing literature on HQ activities and develop a typology with three levels of aggregation: roles, functions and tasks. At the highest level, we distinguish between an entrepreneurial (value-adding) role and an integrative (loss-preventing) role based on Chandler (1991). We note that the distinction is far from clear-cut-arguably, HQ 'creates value by preventing loss' (Foss, 1997, p.314), so to make it clearer, we conceptualise the two roles in terms of creation versus management. Thus, the entrepreneurial role is broadly defined as 'creating something new' and the integrative role is broadly defined as 'managing something existing'. Below this level, we identify ten mid-level functions Table 1 Roles, functions and sample tasks of corporate headquarters (HQ).

Entrepreneurial (Creating/Value-Adding) Role Integrative (Managing/Loss-Preventing) Role
Function 1: Strategic Leadership, Planning and Direction Function 6: Monitoring, Control and Governance • Define business portfolio, growth paths and goals for the corporation as a whole (Benito et al., 2011;Chandler, 1991;Foss, 1997;Goold and Campbell, 2002) • Perform periodic planning cycles linked to long-term strategic plans (Chandler, 1991) • Appoint subsidiary mandates and managerial roles (Foss, 1997;Hamel and Prahalad, 1983) • Monitor how decision-making rights are exercised, supervise operations and evaluate performance (Chandler, 1991;Collis et al., 2007) • Structure the relationship between organisational units (Doz and Prahalad, 1984) • Manage budgets, data, managers and conflict resolution (Bower, 1970;Chandler, 1991;Doz and Prahalad, 1984;Lai Hong et al., 2006) and reduce opportunism (Foss, 1997) Ghoshal, 1989;Bower, 1970;Chandler, 1991), • Deploy these centrally developed or procured resources and assets to subsidiaries in different locations (Collis et al., 2007;Dellestrand and Kappen, 2011;Foss, 1997;Foss and Pedersen, 2002) • Leverage and direct knowledge across the firm, whatever its source (Ciabuschi et al., 2010;Foss, 1997) by providing structure, processes and incentives for knowledge creation and sharing among subsidiaries (Ambos and Mahnke, 2010;Ciabuschi et al., 2011) • Develop intra-firm IT platforms for knowledge sharing (Ciabuschi et al., 2010) • Strategically manage human resources (Taylor et al., 1996)  • Build external relationships with various stakeholder groups (Birkinshaw et al., 2006) • Identify and incorporate significant innovation into company strategy (Egelhoff, 2010) • Conduct corporate-wide discovery processes (Foss, 1997) • Act as the company's representative and fulfil legal requirements (Bower, 1970;Collis et al., 2007;Goold and Campbell, 2002) • Act as a mediator with government agencies and public bodies (Chandler, 1991) and with customers • Handle investor relations (Bower, 1970) Function 4: Driving Organisational Adaptation Function 9: Coordination and Harmonisation • Foster continuous self-renewal of the corporation (Tallman and Koza, 2010) • Initiate and assist adaptation processes throughout the organisation (Ambos and Mahnke, 2010; • Orchestrate, drive and institutionalise strategic change (Goold and Campbell, 2002;Tallman and Koza, 2010) • Identify economies of scale and scope; pool resources and internal services (Chandler, 1991;Collis et al., 2007;Egelhoff, 2010;Hamel and Prahalad, 1983; • Oversee co-ordination along the value chain, across countries and business functions (Bartlett and Ghoshal, 1989;Goold and Campbell, 2002) Function 5: Attention and Signalling Function 10: Integration and Facilitation of Inter-unit Linkages • Recognise and give credit to subsidiaries for their contribution to the organisation as a whole (Ambos and Birkinshaw, 2010) • Provide positive attention, thus creating a signalling effect (Ambos and Birkinshaw, 2010) which can enhance individual commitment and motivation • Encourage vertical integration and foster shared norms, values and corporate ethos among HQ and subsidiaries (Foss, 1997;O'Donnell, 2000) • Encourage lateral integration through facilitating contact among managers of different subsidiaries (Ciabuschi et al., 2010;O'Donnell, 2000) • Develop and implement tight coupling within the organisation (Egelhoff, 2010) representing key elements of the two roles, based on aggregating the various HQ tasks cited in the literature. Hence, functions denote the major elements of the two roles, as well as serving to aggregate tasks (i.e. the specific activities performed by HQ). Under the entrepreneurial role, a key function of corporate HQ is strategic leadership, planning and direction (Function 1). This function adds value by identifying appropriate goals and targets for the MNE, creating a short-, medium-and long-term path to be followed and promoting stability and coherence. Function 2 concerns the development, acquisition and deployment of key resources (tangible and intangible), which enable the MNE to work towards the strategic goals defined in Function 1. Function 3 is concerned with seeking and exploiting new opportunities, building relationships with external actors and bringing ideas into the MNE to create innovation. Function 4 focuses on driving organisational adaptation, initiating strategic change in response to shifting external demands that allows the MNE to renew itself and battle strategic inertia that can threaten its viability in the long run. Finally, Function 5 adds value by paying attention and signalling commitment to employees dispersed around the globe, thus stimulating their motivation to engage in behaviours that promote the MNE's overall interests.
With regards to the HQ's integrative role, Function 6 (monitoring, control and governance) is essential. This function allows the MNE control over how resources and decision-making rights are employed, and helps to minimise the risk of subsidiary managers acting against the MNE's overall interests. Function 7 focuses on managing resources and promoting knowledge sharing within the MNE (see Bartlett and Ghoshal, 1989;Kogut and Zander, 1993;Minbaeva et al., 2003;Taylor et al., 1996). Function 8 (representation and mediation) is largely externally facing and includes tasks such as filing accounts, negotiating with governments and customers. Function 9 involves coordinating and harmonising geographically and functionally dispersed activities, allowing the MNE to simplify its operations and exploit cost savings along the value chain. Function 10 (integration and facilitation of inter-unit linkages) is closely related to Function 9, however, we see a subtle but important difference in that coordination tends to be 'imposed' on subsidiaries, whereas integration is 'encouraged'. Hence, Function 9 tends to involve formal mechanisms, while Function 10 is more reliant on informal or semi-formal mechanisms. With regards to Table 1, it must be noted that the complexity of functions means that there may well be overlaps at the level of functions and even roles. Nonetheless, we believe our typology provides a valuable foundation for a more informed discussion of the value added by HQ, since it is based on a systematic logic of 'creating' versus 'managing' to identify distinct clusters of tasks.
In recent years, scholars have argued that as MNEs grow ever larger and more complex, HQ roles and functions are becoming increasingly distributed throughout the whole organisation and many of them are enacted by regional headquarters or RHQs (Benito et al., 2011;Birkinshaw et al., 2006;Goold and Campbell, 2002;Kähäri et al., 2010;. In the next section, we investigate the applicability of our typology of HQ roles at the regional level.

Regional management in MNEs
The regional management literature harks back to authors such as Heenan and Perlmutter (1979), who first identified the trend for MNEs with regiocentric orientations-i.e. MNEs that differentiate their strategies by regional blocks such as the 'triad' regions of Europe, North America and Asia. Regions can be defined at various geographic levels, from the triad regions to small groups of countries, and even intra-country regions. Over time, 'region' as a level of analysis has been increasingly incorporated into academic discussions about strategy and structure (see Ghemawat, 2005;Morrison et al., 1991;Rugman and Verbeke, 2004), and the literature on regional management-although still somewhat limited-is growing steadily.
Regional management in MNEs most commonly involves the establishment of regional headquarters (RHQs) as an extra layer of management between the global and local (country) levels. Large-scale studies on RHQs are scarce, but existing research (Enright, 2005a, b) suggests that RHQs are widely used by MNEs. The literature covers three broad themes: (1) how MNEs define regions and pick locations for RHQs (2) what functions RHQs fulfil and (3) how the roles of RHQs change over time (Daniels, 1987;Enright, 2005a, b;Ghemawat, 2005;Kähäri et al., 2010;Lasserre, 1996;Lehrer and Asakawa, 1999;Paik and Sohn, 2004;Schütte, 1997;Yeung et al., 2001). Our examination of the literature leads us to concur with Piekkari et al (2010) that much of the RHQ literature is predominantly managerial and descriptive, and the classification of the functions performed by RHQs is incomplete and under-theorised. Using our original typology presented in Table 1, we synthesise the extant literature on RHQ functions in Table 2.
Most of the tasks identified in Table from the regional management literature essentially represent regional versions of the same tasks found at the HQ level (e.g. regional strategic planning, monitoring, etc). However, we also see some differences, notably in Function 7: when performed at the regional level, this function involves not only external representation (liaising with customers, governments etc.), but also internal representation (liaising between HQ and local units), which creates an additional dimension. Meanwhile, we did not find a mention of tasks relating to driving organisational adaptation (Function 4). While this function is explicitly noted by Piekkari et al (2010) who examine HQ roles both at global and regional levels, they suggest that initiating organisational adaptation is the primary responsibility of corporate HQ, rather than RHQs. As such, it appears that not all HQ functions may be delegated to RHQs.
Very few authors note the possibility of HQ functions being delegated to local operating subsidiaries-as opposed to dedicated RHQs-despite practical evidence to the contrary (Businessline, 2004;Rees, 1996). When discussing HQ functions being performed at regional levels, most studies do not distinguish between dedicated RHQs and local subsidiaries with regional responsibilities. The latter-which we term regional management mandates (RMMs)-is at most briefly noted (Kähäri et al., 2010;Lasserre, 1996;Schütte, 1997). We argue that there is an important distinction between delegating HQ functions to administratively oriented units (RHQs) and to profit-oriented subsidiaries (RMMs), whose implications for regional management warrant closer investigation. We discuss these next. Table 2 Roles, functions and sample tasks of regional headquarters (RHQs).

Function 2: Resource Development, Acquisition and Deployment Function 7: Resource and Knowledge Management
• Effective development of globally trained human resources (Paik and Sohn, 2004) • Create and design regional products, services and business processes (Enright, 2005b) • Improve information flow within the MNE (Paik and Sohn, 2004) • Benchmarking and spreading best practice within the region (Schütte, 1997) • Regional IT management (Enright, 2005b) • Manage senior personnel (Enright, 2005b)

Function 5: Attention and Signalling Function 10: Integration and Facilitation of Inter-unit Linkages
• Signal commitment to subsidiaries in the region (Lasserre, 1996;Yeung et al., 2001) • Foster synergy among operations (Daniels, 1987) • Manage interdependence among units (Paik and Sohn, 2004) 2.3. Delegating headquarters functions to operating subsidiaries When trying to understand or explain complex problems, scholars have often noted the merits of combining theories (Doz and Prahalad, 1991;Eisenhardt, 1989). Since the delegation of HQ functions to operating subsidiaries via RMMs is a complex phenomenon which is yet to be explored in the literature, we chose three different but closely linked theoretical lenses to frame our empirical investigation: contingency theory, information processing theory and agency theory. All three perspectives have been used or been suggested for use in studies of headquarters and regional management (see Ambos and Mahnke, 2010;Egelhoff, 2010;Foss, 1997;Paik and Sohn, 2004;. In addition, they build on each other in a natural manner, sharing fundamental concepts such as 'fit', 'information' and 'structure', which allows a comprehensive picture to emerge from the findings derived using each of the three perspectives.

Contingency perspective
At its heart, the key argument of contingency theory is that an organisation's structure must follow its strategy, which itself must follow environmental imperatives (Chandler, 1962;Lawrence and Lorsch, 1967). Hence, the optimal structure for any organisation is contingent on its specific strategy within its specific environment, and optimal performance is contingent on the 'fit' between environment, strategy and structure. The finding by Lawrence and Lorsch (1967) that organisations must balance differentiation and integration to be successful can be seen as a direct precursor to the integration-responsiveness (IR) framework developed in the 1980s (Bartlett and Ghoshal, 1989;Prahalad and Doz, 1987). While the original IR framework was focused on global integration versus local responsiveness, scholars soon highlighted the role of IR pressures at the regional level, and their impact on MNE strategy and structure (Ghemawat, 2005;Lehrer and Asakawa, 1999;Rugman and Verbeke, 2004). Arguably, a regional strategy holds the ideal balance between integration and responsiveness: it keeps costs low by exploiting regional scale economies, while earning high revenues by adapting to regional differences in business environments and consumer preferences (Daniels, 1987;Morrison and Roth, 1992;Morrison et al., 1991;Paik and Sohn, 2004;Schütte, 1997).
To the extent that the MNE chooses to pursue a regional strategy, it is likely to develop some form of regional structure to match, such as clustering subsidiaries into regional groups based on geographical proximity and institutional, cultural or market similarities (see Egelhoff, 1982;Ghemawat, 2005). While this is most commonly done by establishing dedicated RHQs, Lasserre (1996) notes that building and maintaining a dedicated office (either in a separate physical location or within the HQ) and employing regionallevel directors to look after a particular region is a high-cost solution. As an alternative, he suggests that certain regional management tasks may be fulfilled more cost-effectively by mandating existing local subsidiaries as management centres, particularly in small or fragmented markets. This idea is also echoed in Schütte's (1997) notion of 'virtual regional headquarters'. Under this set-up, RHQ functions are fulfilled by local directors and managers on top of their local activities. In theory, this should not incur significant extra costs (provided the local subsidiary has sufficient resources to fulfil the extra functions), which allows this solution to be employed at levels of regional aggregation where the cost of a fully fledged RHQ would be prohibitive. We argue that this allows the MNE to achieve a more fine-grained balance between integration and responsiveness.

Information processing perspective
The information processing approach was developed around the same time as contingency theory began to gain currency, and explicitly builds on it (Egelhoff, 1982;Galbraith, 1973;Tushman and Nadler, 1978). Information processing refers to 'the gathering, interpreting and synthesis of information in the context of organisational decision making' (Tushman and Nadler, 1978, p.614), and is a vital but resource-intensive component of managing organisations. The theory distinguishes between different types of information processing: strategic (high-level, broad, long-term) versus tactical (low-level, narrow, short-term) and posits that different structures facilitate one while restricting the other (Egelhoff, 1982). Organisational effectiveness is thus a function of the fit between information processing requirements (shaped by environmental conditions) and information processing capabilities (determined by features of organisational design). To survive, firms must develop information processing mechanisms capable of dealing with complexity and uncertainty from various internal and external sources (Egelhoff, 2010;Galbraith, 1973;Tushman and Nadler, 1978).
Information processing theory is well suited to study HQ functions and RHQs and has been used by several scholars (Collis et al., 2007;Egelhoff, 2010;. Due to the growth in the geographical reach of MNEs, the processing of large volumes of diverse local information requires increasingly tailored information processing capabilities. While the corporate HQ and dedicated RHQs tend to be good at the processing of high-level strategic information, they are less effective at dealing with low-level tactical information. In particular, as Ghemawat (2005) notes, administratively oriented RHQs tend to suffer from weak links to subsidiaries' operating activities. This may be seen as a result of an insufficient fit between their information processing capabilities (largely strategic) and the information processing requirements of local environments (largely tactical). Hence, to reduce the risk of missed opportunities, inappropriate products or resources wasted on ill-advised interventions, there is a need to increase discretion at the operating level (Hamel and Prahalad, 1983), while also coordinating information processing to capture scale economies in routine tasks such as payroll processing (Collis et al., 2007). The delegation of HQ functions to local operating subsidiaries presents a solution to this problem. Since the primary role of local directors and managers is to engage in business activities on their own market, it is reasonable to expect that they will be able to link their day-to-day operating experience to higher-level functions delegated to them. As such, the knowledge accumulated in key individuals in a local subsidiary can be exploited on a regional level without removing them from their primary role on the local market, thus allowing more efficient information processing.

Agency perspective
Agency theory is linked to both contingency theory and information processing theory. Its basic assumptions are bounded rationality, information asymmetry and goal incongruence between principal and agent, and it is concerned with the optimal structuring of control relationships in order to minimise agency costs (agent behaviour that is not in the best interest of the principal) and managing the trade-off between the cost of monitoring behaviour and the cost of monitoring outcomes (Eisenhardt, 1989;Fama, 1980;Jensen and Meckling, 1976). MNE governance can be aptly conceptualised in terms of agency relationships where the HQ (principal) seeks to maintain control over subsidiaries (agents) that would generally prefer to act autonomously (Paterson and Brock, 2002). Subsidiary behaviour may be monitored by using expatriates (distributed agents) to supervise and report on the day-to-day activities of the subsidiary, while outcomes can be monitored by setting sales and financial targets for the subsidiary to meet. Monitoring outcomes is generally low-cost, but may not be a reliable indicator of effort when the subsidiary's tasks are complex, while monitoring behaviour tends to be more effective in complex operating environments, but at a much higher cost.
With regards to HQ functions, maintaining control becomes more difficult as the number of divisions and markets served increases, as there are more numerous and diverse agents for HQ to manage and monitoring behaviour becomes increasingly costly, especially in peripheral subsidiaries (Benito et al., 2011). By delegating monitoring and coordination functions to RHQs or 'intermediate parents', HQ can reduce the number of agents it needs to manage directly (Goold and Campbell, 2002). However, due to their high cost, dedicated RHQs tend to have limited staff who must divide their attention amongst several subsidiaries, meaning they are constrained in their ability to monitor peripheral subsidiaries' behaviour on a day-to-day basis. By appointing this 'intermediate principal' role to an operating subsidiary instead, HQ can access a larger pool of employees who are able to monitor the behaviour of agents in other subsidiaries. Given that these employees are already engaged in operating activities, the information asymmetry between those doing the monitoring and those being monitored is reduced, thus reducing monitoring costs.
Our theoretical discussion shows that the delegation of HQ functions to operating units through RMMs can hold advantages over the use of dedicated RHQs. As noted before, there is evidence from the practitioner literature that this is considered a viable option by some MNEs-however, to date, no academic research has been conducted to provide empirical evidence for this, or to investigate which of the HQ functions and tasks noted in Tables 1 and 2 may be performed by operating subsidiaries. We now turn to investigate this empirically.

Methodology and context
For our empirical study, we use qualitative interview data collected from a larger project investigating hierarchical relationships and knowledge sharing amongst subsidiaries of Unilever, the Anglo-Dutch fast-moving consumer goods (FMCG) manufacturer. The overall project compares and contrasts the perspectives of three different subsidiaries; however, in this paper we concentrate only on the perspective of Unilever Hungary (ULH) as the intermediate principal appointed by Unilever HQ and  2  WL1  3  Total  20  20  Total  20  Total  20  Total  20 Total 20 a Due to the size and structural complexity of Unilever, the definition of work levels is complicated and contains overlaps. As a rough guide, work levels are as follows: WL6 = global executive board, inc. directors of business lines and 'triad' regions; WL5 = regional/cluster-level chairmen; WL4 = cluster-level directors; WL3 = functional department or country unit managers; WL2/WL1 = operational managers/assistant managers. b At the time of data collection. c Retired but remains a member of the company's supervisory board. holder of a regional management mandate (RMM), since our research questions are focused on what HQ functions are performed 'on the ground' and how.

Case study methodology
Given the lack of prior academic attention to regional management mandates, an exploratory approach was the most suitable for our study. Our methodological approach is best described as abductive: it starts with theory, but involves constant interplay between theory and data in order to develop and refine propositions which offer the most plausible explanation for our observations (Dew, 2007;Peirce, 1960). We conduct an in-depth single case study (see Stake, 1995;Yin, 2003) of Unilever Hungary, which holds a RMM to manage two of Unilever's peripheral 'satellite' subsidiaries in Croatia and Slovenia. Our time frame is defined as the period from 1997 to 2007, starting with Unilever Hungary's RMM appointment and finishing just as the subsidiary began to prepare for a significant reorganisation.

Data collection
Our primary data collection took place over the course of two years between 2005 and 2007. Although the organisational structure of the focal units has changed significantly since that time, we present our findings in the context of the specific organisational structure in place at the time that our data was collected: a country cluster comprising three countries and reporting to Unilever's European HQ. Most of our primary data was constructed from retrospective accounts by respondents, gathered through semi-structured interviews and respondents' sketches. This data was supplemented and triangulated through the use of publicly available data sources such as websites and reports. To present a sense of the three subsidiaries' differing operating contexts, key indicators and features of the three countries' business environments are shown in Table 3.
Our analysis is based on interviews with eighteen respondents from Unilever Hungary and two from the CEE regional level. Some respondents granted us repeat interviews, taking the total number of interviews used to twenty-five. We interviewed respondents at all but the very highest work level, which allowed us to explore RMM involvement at various operational levels ( Table 4).
All interviews were conducted by the same member of the research team, who is bilingual in Hungarian and English. Interviews at Unilever Hungary were conducted in Hungarian, allowing greater freedom of expression, rapport-building and richness of the data (see Welch and Piekkari, 2006). The two interviews at the CEE level were conducted in English with nonnative speakers whose command of English was at an extremely high level, given that it is Unilever's corporate language. As such, it may be said that linguistically, all interviews were conducted on a 'neutral platform' (Welch and Piekkari, 2006). Interviews ranged from 30 to 120 minutes in length, and all but one were recorded and transcribed. Hungarian-language interviews were translated into English by the same researcher, who was able to scrutinise potential meanings in both languages as well as closely evoke the interview context during translation. Being familiar with 'everyday speak' (metaphors, social expressions, historical and cultural references etc.) as well as 'company speak' (operating terms, value statements etc.) in both languages helped to generate authentic and well-contextualised translations; and the translation process itself allowed deeper meanings to emerge. Of course, the 'problematic of translation' (Welch and Piekkari, 2006) could not be avoided: the interpretation element of translation means that a risk of potential bias can never be completely eliminated. We minimised this risk by regularly comparing translated accounts of events and phenomena with accounts in the English-language interviews, and by revising and refining expressions in the translated data over time during data analysis.

Data analysis
We used the NVivo software and two basic strategies to analyse our qualitative data (Richards, 2005). First, we used topic coding, which followed a data-driven logic and allowed us to categorise and make sense of the rich data we collected about topics such as Unilever's complex organisational structure and the different demands of the three countries' business environments. Secondly, we used analytical coding, which involved coding the data into an evolving structure according to the main theoretical concepts and ideas identified from our literature review. Our overall analytical process relied on the theoretical inputs into our research as well as on our empirical data. Our findings are presented in the next section.

Context and empirical case
Unilever is one of the world's largest manufacturers of fast-moving consumer goods (FMCG), with an annual turnover of around €44bn (Unilever, 2010). The company manages a highly diversified portfolio of branded products (foods, toiletries and household cleaners) and operates in a sector characterised by high pressure for local responsiveness (differing consumer tastes, especially in the foods business) as well as for global integration (major advantages from leveraging brands such as Cornetto, Dove or Domestos). During our data collection (2005)(2006)(2007)), Unilever's regional structure consisted of various levels: three 'triads', six regions and numerous sub-regions/country clusters. Our research focuses on the sub-region/country cluster headed by Unilever Hungary within the Central and Eastern European (CEE) region, which includes Hungary, Croatia and Slovenia.
Unilever has been present in the CEE region since the early twentieth century. Its factories in CEE countries were nationalised following the Second World War, forcing the company to exit. From the 1970s, Unilever re-entered several CEE markets through licensing and exporting, largely managed by the Unilever Export unit in Bristol, UK. After 1990, as CEE countries began to open up to FDI, Unilever started to enter CEE countries directly and sequentially through acquisitions and greenfield investments, often 'piggybacking' on its existing ventures. It should be noted that between 1997 and 1999, our focal cluster also contained other socalled South Central European (SCE) countries (Bulgaria, Albania and former Yugoslav countries) and Unilever Hungary played an active role in establishing local operations in all of these. However, in 1999 all but the Croatian and Slovenian satellite units were transferred to the cluster led by Unilever Romania. Hence, although there are some important references to the SCE region in our findings, the majority of our findings focus on Unilever Hungary's RMM as it existed between 1999 and 2007: Unilever Hungary as cluster leader and Unilever Croatia and Unilever Slovenia as 'satellite subsidiaries'. Another thing that needs to be noted here is that most of our interviews came from one of two large operational divisions: marketing and sales/trade marketing, 2 which provided an interesting contrast in terms of integration / responsiveness (see Section 4.1.2). This data was supplemented by interviews with general business unit directors and managers in support functions such as HR.

Empirical findings
4.1. HQ functions performed as part of a regional management mandate To answer our first research question (to what extent can HQ functions be delegated to operating subsidiaries), we asked our Hungarian respondents to define how they viewed their regional management mandate. Most respondents were in agreement about the overall responsibilities involved-in the words of a Hungarian manager expatriated to Slovenia: "What constitutes the official mandate of Unilever Hungary in the region? Well, it is to manage Unilever's brands and categories according to central directives, managing the three countries relatively as one and exploiting synergies." We then asked about their specific tasks and interactions with the Croatian and Slovenian units to find evidence of performing entrepreneurial and/or integrative roles.

Entrepreneurial role
We find evidence that several of the functions that constitute the headquarters' entrepreneurial role can be regularly performed at the level of operating subsidiaries. In Unilever Hungary (ULH), these include strategic direction (Function 1), with respondents indicating that the Slovenian and Croatian businesses are driven from Hungary 'on an extended market base', where the Hungarian unit receives strategic plans and frameworks from HQ and has the responsibility of 'translating' these to the three countries. Although respondents emphasised that this strategic planning process happened with the involvement of their Croatian and Slovenian colleagues, the authority to sign off project plans rested with managers and directors at ULH, underlining its role in driving strategic planning in the context of the country cluster.
Our data suggests that ULH was responsible for HR recruitment in all three countries, and historically it had been engaged in resource development on a regional level (Function 2). Between 1998 and 2004, the Nyírbátor factory in Eastern Hungary housed a regional innovation centre (RIC) which created new products for regional markets (such as a thick, economical Cif cleaning paste for low-income consumers). In addition, the Hungarian HR team created new processes for the region (career path models, leadership development training packages), some of which were later transferred to HQ. However, due to Unilever's centralisation strategy from 2004 onwards, this resource development function has been significantly reduced in ULH.
We noted the key role of ULH in seeking and exploiting new opportunities (Function 3). This function was crucial at the early stages of Unilever's entry into the South Central Europe region (1997)(1998)(1999), when ULH was tasked with taking over arm's length distributor agreements from Unilever Export BV and building up new, dedicated local operations in the region, involving the establishment of legal entities and kick-starting business operations while piggybacking on existing contacts and agreements in the region. The entrepreneurial nature of these tasks is clear from this account of the former SCE director: "We had to create the infrastructure in every single country, building on the heritage…to start a business structure that fitted in with the maturity, openness or potential of that market. So we had to form a market entry, business and infrastructural strategy and manage its implementation, in a way that it could work with the required level of autonomy, but it would operate in the Unilever way." Despite evidence of new entry and market-building, we did not find evidence of initiating or driving organisational adaptation in the region (Function 4). While the organisational strategy and structure of ULH and the two satellite subsidiaries evolved and adapted over time, these changes were primarily driven by HQ strategy and directives. We believe that this is in line with Piekkari et al's (2010) assertion that the initiation and driving of organisational adaptation is the primary function of the corporate headquarters, rather than RHQs or subsidiaries with RMMs.
Similarly, we did not find evidence of tasks relating to Function 5 (attention and signalling), which is hardly surprising. The literature suggests that HQ attention is highly valued by subsidiaries and the establishment of RHQs is seen as a signal of commitment to a region (Ambos and Birkinshaw, 2010;Lasserre, 1996;Yeung et al., 2001). In contrast, the use of RMMs reduces the amount of direct HQ or RHQ attention (as the satellite subsidiaries report to the RMM subsidiary, they have limited contact with HQ or triad-level RHQs) and signals that the region (in this case, the cluster of Hungary, Croatia and Slovenia) is not considered important enough to have a dedicated RHQ committed to it. As the value of Function 5 lies in conveying the headquarters' interest in local units, it makes sense that this interest cannot be substituted by appointing relatively low-level local staff to provide attention.

Integrative role
We find evidence that through its RMM, Unilever Hungary performs an important integrative role at the cluster level. This entails a strong element of monitoring and control (Function 6), as the board of ULH is formally accountable for the performance of the Croatian and Slovenian subsidiaries and is responsible for ensuring that all three countries conform to Unilever's increasingly centralised processes. Our data suggests that this task is clear and unambiguous for employees at the highest work levels: the cluster chairman (W5) and the business unit directors (WL4) are required to visit the Croatian and Slovenian subsidiaries on a regular basis and conduct strategic reviews/performance evaluations every 3-6 months. However, at lower work levels, monitoring and reporting responsibilities vary by functional department: in marketing, middle managers (WL3) monitor their Croatian and Slovenian colleagues' activities and performance closely, meeting monthly to 'put every brand in [their] category under the microscope'; while in sales and trade marketing, middle managers are not formally accountable for the performance of Croatian and Slovenian employees who report to Hungarian managers only informally. This difference seems to stem from the fact that marketing lends itself well to standardisation, while sales and trade marketing require a high degree of embeddedness in local business and retail networks. Overall, ULH's formal responsibility for the financial results of all three subsidiaries alleviates the information processing burden faced by HQ, as HQ monitors only the aggregate performance of the cluster.
With regards to resource and knowledge management (Function 7), we find evidence that Unilever Hungary's RMM involves human resource management (administration, career counselling and management) as well as considerable knowledge sharing with the two subsidiaries under its watch. This includes both less intense forms of knowledge sharing (phone and email exchanges, benchmark visits, trainings, workshops) and more intense forms (team-building, expatriation, joint projects, job swaps, coaching). For example, when the German retailer Lidl was entering Slovenia in 2006, six months after it had entered the Hungarian market, ULH transferred all the information, documentation and experience it had gathered from dealing with Lidl to allow the Slovenian subsidiary to respond more quickly and efficiently. More intense forms of knowledge sharing also took place, for example, a Hungarian trade marketing manager spent seven months in 2000 coaching newly appointed staff in Slovenia and Croatia where the trade marketing function was less developed at the time. The depth and breadth of knowledge sharing within the cluster can be illustrated by this quote from a business development manager: "We would put together a presentation for them…it could be very interesting for them to see what's going on in Hungary, how our whole organisation is structured, where the customer care team is integrated into the operations, what new things there are…then they go on trade visits where they can see Tesco's operations [which do not exist in Croatia or Slovenia]." In particular, we note that while less intense forms of knowledge sharing are found across the entire CEE region, and even upwards on the European stage, more intense forms of knowledge sharing appear to be reserved for use within the cluster only. This is most likely because they require greater commitment and incur higher resource and opportunity costs, which can only be justified by the formal accountability aspect of Unilever Hungary's RMM.
ULH acts as a representative of the two satellite subsidiaries towards CEE, European and global headquarters (Function 8), acting as a regional liaison with HQ and attending higher-level meetings on behalf of the three countries. As a result, there is limited contact between HQ and satellite subsidiaries. We find that in ULH's case, this mediation function is geared primarily towards internal (rather than external) relationships. One reason for this may be that as a local operational subsidiary, ULH lacks the clout to deal with governments and customers on a regional basis, and is better suited to the management of internal relations.
Our data also suggests that Unilever Hungary's RMM involves a considerable amount of cross-border coordination and harmonisation (Function 9). Hungary, Croatia and Slovenia 'have a single back office system, a single SAP, a single board of directors', and support functions such as finance, IT and other back office operations are performed at the Hungarian subsidiary for all three countries. In addition, we found plenty of evidence of resource pooling in the form of joint procurement of promotional and pointof-sale items (promotional gifts, branded display shelves, advertising banners) geared towards cost efficiency. Similarly to monitoring, coordination and harmonisation is stronger in marketing and weaker in sales and trade marketing where the significant differences in the three countries' retail structures limit coordination possibilities.
We find that even in functions where formal harmonisation is not possible, considerable effort is put into informal integration (Function 10). Middle and top managers in sales and trade marketing reveal that even though they have no formal control over Croatia and Slovenia, they engage in regular operational support, training and external workshops. The relatively informal nature of these integration efforts is evident from a trade marketing manager's account: "We would go there or invite them here, look through the projects and try to help, but…they did in fact learn by themselves what was worth doing given the local conditions…If I thought about who my team were, Croatia and Slovenia were not part of it in such an organic way." Table 5 Roles, functions and sample tasks included in regional management mandates (RMMs).

Function 1: Strategic Leadership, Planning and Direction Function 6: Monitoring, Control and Governance
• Overall strategic direction and leadership within the region • Translation of central strategy to the regional context • Perform planning in the context of regional opportunities and constraints • Monitor regional market, competitors and operations and report to HQ • Evaluate satellite subsidiaries' performance on a regular basis and authorise new projects (mostly in marketing, less so in sales and trade marketing) • Perform bookkeeping, enforce and monitor the implementation of central directives • Ease information processing burden on HQ by taking over direct contact with satellite units

Function 2: Resource Development, Acquisition and Deployment Function 7: Resource and Knowledge Management
• HR recruitment within the region • Create products for regional markets (Regional Innovation Centre) and new HRM processes for the region (career path models, leadership development training packages) • Development function has been in decline since 2004 due to corporate-wide centralisation • HR administration, career counselling and career management for employees • Passing on information, best practices and local innovations within the region, exploit existing relationships with regional retailers • Loose forms of knowledge sharing (regular phone/email contact, benchmark visits, trainings, workshops) • Intense forms of knowledge sharing (team-building, expatriation, job swaps, coaching)

Function 3: Seeking and Exploiting New Opportunities Function 8: Representation and Mediation
• Build dedicated local operations in new markets from scratch • Detect new opportunities in advertising/marketing/sales that can be exploited in the region • Regional liaison centre for parent company

Function 4: Driving Organisational Adaptation Function 9: Coordination and Harmonisation
While we found evidence of implementing and assisting organisational adaptation through coordination, knowledge sharing and operational assistance (integrative/managing role), we did not find evidence of initiating or driving organisational adaptation in a regional context -the primary initiator of organisational adaptation appears to be corporate HQ • Standardise marketing activities on a country cluster level (in line with MNE strategy) • Coordinate activities across borders, shared services such as IT, finance, logistics etc.
• Pool resources for cost savings (joint procurement of promotional/advertising materials) • Assist satellite subsidiaries in conforming to new corporate processes/structures

Function 5: Attention and Signalling Function 10: Integration and Facilitation of Inter-unit Linkages
No evidence found -on the contrary, the use of RMMs reduces direct HQ attention and sends the signal that peripheral subsidiaries are not important enough to warrant direct contact with HQ or an RHQ • Encourage informal integration even where complete harmonisation is not possible due to local differences (sales and trade marketing) • Operational support and advice as needed, such as tailored training/workshops Overall, we find strong evidence that Unilever Hungary performs a strong integrative role at the level of the cluster. The entrepreneurial and integrative roles we found as part of ULH's regional management mandate are summarised in Table 5. Next, we revisit our second research question to examine the benefits of delegating HQ functions to operating subsidiaries. 4.2. Potential benefits and drawbacks of regional management mandates 4.2.1. Integration-responsiveness advantages of regional management mandates In Section 2.3.1, we argued that regional management mandates add value to the MNE by allowing a fit between regional strategy and structure at a level where the costs of a dedicated RHQ would be prohibitive. Empirically, we find evidence for both parts of this argument (fit as well as cost efficiency). The reason that Unilever developed a cluster structure based on RMMs is perhaps most succinctly summed up by the CEE regional director: "Some of the countries wouldn't justify by themselves a separate structure…I mean, they didn't have the critical mass that would afford a fully fledged company, but had sufficient potential to require focus. So we basically went to a structure where we could do both at the same time, so have something that makes some economic sense, but at the same time, have sufficient focus, so that we can develop the markets." In other words, from a contingency perspective, markets such as Slovenia and Croatia are considered important enough to merit a locally differentiated strategy, but their small size means they cannot support a matching structure. As a business unit director notes: 'In a small country, obviously, there is a lack of critical mass, the size that allows it to sustain a particular organisation'. In such a case, entrusting small local units to the care of a larger, geographically close and well-established unit such as Unilever Hungary allows integration at the cluster level while also preserving responsiveness at the local level, which may be the best strategy-structure fit under these circumstances. In addition, as respondents noted, Unilever's multinational customers and competitors are similarly organised in the region, therefore matching their structure adds value by allowing Unilever to negotiate more efficiently at various levels.
With regards to our cost efficiency argument, although we did not have access to respondents at the HQ level who could confirm whether RMMs indeed represent cost savings over establishing RHQs at the same level, we find indirect evidence that establishing an RMM does not require significant resource investment from HQ (unlike establishing a RHQ). In the case of Unilever Hungary, strong existing financial and human resources and self-sufficiency were repeatedly cited as the key reason why the subsidiary received its RMM, which would suggest that RMMs are given to subsidiaries which are able to perform regional HQ functions with no need for added investment. Therefore, it would be logical to deduce that RMMs are regarded by the company as a cost-effective use of resources. More directly, our data suggests that cost savings arising from the use of existing resources for monitoring, coordination and knowledge transfer are an important aspect of Unilever Hungary's RMM. The value added by resource pooling, in particular, was emphasised by several respondents. As a marketing manager notes, 'without the size and strength of the Hungarian business, everything is a lot more expensive for [Unilever Croatia and Unilever Slovenia]'. Given the small size of these subsidiaries, it is no wonder that scale economies achieved through cluster-level resource pooling are highly valued: according to a business development manager, 'there is a lower level of profitability in the whole of Central and Eastern Europe, because only two countries, the Russians and the Poles, have an efficient scale'. At the same time, the value added by maintaining knowledgeable local teams in each country is also strongly emphasised, which suggests that there are limits to the potential for cross-country integration. On the whole, our data yields qualitative evidence that the use of RMMs at the subregional level allows Unilever to fine-tune responsiveness to diverse local markets, as well as exploit cost savings and operational scale economies across borders. This leads us to suggest the following proposition: Proposition 1. Delegating HQ functions to operating subsidiaries through RMMs adds value to the MNE by allowing a balance between integration (resource pooling and shared services) and responsiveness (local operating teams) at a regional level below the efficient scale for a dedicated RHQ.

Information processing advantages of regional management mandates
In Section 2.3.2, we suggested that delegating HQ functions to operating subsidiaries adds value by allowing a closer fit between information processing requirements and capabilities. We find evidence for this, particularly with regards to the entrepreneurial task of developing local operations throughout the South Central Europe region (as detailed in Section 4.1.1). As we argued before, HQ tends to have strong capabilities for strategic information processing, such as developing the MNE's overall business and product portfolio and identifying potential synergies across the globe. Meanwhile, tactical information processing capabilities, such as responding to specific local market situations or tailoring promotions to particular audiences, tend to reside more at the subsidiary level. This is echoed by Unilever Hungary's vice-chairman at the time of allocating the task of developing and overseeing regional operations to Unilever Hungary in 1997: "The centre did not have all the information on what was the best way to develop these countries. We were deeply involved at that time in the decision-making process, and we came to the conclusion that doing it out of Hungary with the existing resources, with our knowledge and additional resources…was the best way of [developing operations]." In this case, it was acknowledged that as an operating subsidiary with accumulated experience of trading in a postsocialist business environment (as opposed to an RHQ that would not have this kind of operational experience), Unilever Hungary possessed more relevant tactical information processing capabilities to set up operations in CEE markets than could be found at HQ. Thus, it made sense to exploit the close fit between the information processing capabilities of ULH and the information processing requirements of the regional environment, instead of HQ or an RHQ taking the task of establishing new subsidiaries upon itself. As a trade marketing manager noted, Unilever Hungary was 'armed with adequate knowledge, having got over 'childhood diseases'…all this experience was accumulated there, it simply had to be implemented [in other markets]'.
Looking beyond the establishment of new operations and focusing on the ongoing management of satellite subsidiaries, we find further evidence of tactical information processing capabilities at the subsidiary level which are successfully exploited at the cluster level. ULH's operational expertise in the highly competitive, fast-growing Hungarian market was seen by respondents as vital to manage the satellite subsidiaries in Croatia and Slovenia, particularly with regards to effective knowledge sharing. As we highlighted in Section 4.1.2, knowledge sharing is an important aspect of Unilever Hungary's RMM, and the subsidiary's accumulated expertise is acknowledged and sought after by the Slovenian and Croatian units because it was developed in the unique regional environment of Central and Eastern Europe. As the Hungarian sales director points out, the CEE retail environment has developed in a very different manner from Western Europe where local chains with long histories tend to dominate national markets: "In Germany, it is the German [retailers], in the UK it is the British ones…Well, what happens in Central and Eastern Europe? Like some kind of birds' nesting place, all the dominant retail chains are present, and each of them brings from its own country its national business culture, its own set of expectations, its own system of solutions and its own strategy. We have to live up to this, which is a challenge in itself." Thus, it may be said that the unique information processing requirements of dealing with a variety of multinational retailers (Tesco, Lidl, Auchan etc.) as well as with local/regional chains and buying groups (CBA, Reál) foster the development of specialised and wide-ranging information processing capabilities. We find strong evidence that through Unilever Hungary's RMM, these capabilities are being exploited and passed on to Unilever Slovenia and Unilever Croatia. This allows us to generate a second proposition: Proposition 2. Delegating HQ functions to operating subsidiaries through RMMs adds value to the MNE by exploiting and sharing tactical information processing capabilities, such as accumulated local operational expertise, on a regional level.

Agency advantages and disadvantages of regional management mandates
In Section 2.3.3, we argue that delegating HQ functions through regional management mandates adds value to the MNE by reducing the cost of monitoring peripheral agents' behaviour. In the context of our case, Unilever Hungary can be viewed as an agent of HQ and intermediate principal to Unilever Croatia and Unilever Slovenia, which are satellite subsidiaries seen as peripheral agents operating in markets that are of low strategic importance for the MNE. Although our empirical investigation was limited by our lack of access to Unilever HQ, we found indications that allocating RMMs to subsidiaries such as Unilever Hungary can reduce monitoring costs for HQ. As a business unit director noted, 'an organisation can operate more efficiently if, instead of twenty-something CEE countries, the headquarters has to negotiate with and communicate through just five cluster-leading ones, so…the task is to create clusters'. As noted in Section 4.1.2, Unilever Hungary fulfils a strong monitoring and control function: it is held responsible for ensuring the timely and accurate implementation of central directives in the cluster, and bears the costs of monitoring the two satellite units through day-to-day contact and regular visits between the countries. The relatively small geographical distance and close operating links between the three subsidiaries means that information asymmetry is reduced, which allows the monitoring function to be performed by ULH more efficiently than by geographically and organisationally distant principals such as HQ or higher-level RHQs. This finding allows us to introduce the following proposition: Proposition 3a. Delegating HQ functions to operating subsidiaries through RMMs adds value to the MNE by transferring the cost of monitoring peripheral agents' behaviour from the central principal (HQ) to an intermediate principal which is able to monitor peripheral agents' behaviour more closely.
In addition to the potential benefits of using RMMs, our data also revealed potential risks that appear to stem from the nature of the principal-agent relationship between Unilever HQ and Unilever Hungary. Despite the fact that every respondent we talked to was aware of Unilever Hungary's mandate to oversee and manage operations in Croatia and Slovenia, and most of them were actively involved in this task, we found a notable lack of clarity at the operational level about what the RMM actually entailed. In the words of a Hungarian trade marketing manager, 'even Unilever could not define exactly how these two countries were connected to us'. When probing a senior level director expatriated by HQ to Hungary about whether the RMM arrangement was formally recorded, the response was: "That is not the Unilever style. The decision is taken and then executed. Of course, we had discussions, there was an iteration process…there was a formal decision, but there was nothing written at all. That is a normal development process inside of Unilever." From an agency perspective, this highlights the existence of incomplete contracts between HQ and subsidiaries (see Foss, 1997) and raises intriguing questions about how principals can ultimately monitor the fulfilment of duties that have not been explicitly defined or recorded. When asked about HQ mechanisms to monitor how well RMMs are fulfilled, the same director cited financial performance (monitoring outcomes) and levels of knowledge and professionalism in the satellite units (monitoring behaviour). Arguably, these are relatively crude measurements for such a complex task as an RMM, raising questions about the extent to which the principal is able-or even willing-to monitor the behaviour of agents in complex contexts.
Resulting from this incompleteness of the RMM contract between Unilever HQ and Unilever Hungary, we identified a set of underlying tensions that can be helpfully conceptualised in agency terms. As noted in Section 4.1.2, in the marketing division, ULH has strict control over the Slovenian and Croatian units through formal reporting lines. Yet, in the sales and trade marketing division, ULH's 'intermediate principal' role is not clearly defined or enforced. Given that HQ, the ultimate principal does not interact with the peripheral units, there is a risk of agency costs such as moral hazard. That is, satellite subsidiaries may end up acting in ways that are not in the best interest of the MNE because their behaviour is only monitored by an intermediate principal that is not fully recognised as having legitimate authority (ULH), not by the principal who has the ultimate authority (HQ). This problem is noted by a trade marketing manager: "Often, if we ask [Croatian/Slovenian colleagues] to implement something because it is expected of us by the European management-since they are not directly in touch with the European management, we are the 'watershed' or the 'buffer'we are often the ones being confronted because they do not feel as obliged, or things don't happen as they should." Tensions such as this highlight the problem of principal-agent relationships where the principal lacks full legitimacy, and its authority can therefore be challenged by the agent. In this case, from the satellite subsidiaries' perspective, Unilever Hungary is an appointed principal, but at the same time it may be viewed as 'just another subsidiary', which can reduce its perceived authority. Alluding to this issue, a marketing manager remarked that 'the Croatians somehow… intellectually loathed the fact that…these Hungarians were sitting on a throne above them and were telling them how to run the business.' Issues such as this can be linked to current debates about the distribution of power and authority within MNEs (see Bouquet and Birkinshaw, 2008;Dörrenbächer and Geppert, 2006) and allow us to present a new proposition relating to the potential risks of using RMMs for regional management: Proposition 3b. Delegating HQ functions to operating subsidiaries through RMMs can reduce the effectiveness of behavioural monitoring and increase the risk of moral hazard, unless the intermediate principal is accepted as fully legitimate by peripheral agents.
We also note that inserting extra layers of regional management between the ultimate principal and the MNE's lowest-level agents can also have negative consequences for the latter. Due to Unilever Hungary acting as their principal, employees at Unilever Croatia and Unilever Slovenia have limited opportunity to engage in direct contact with European-level activities, or in informal lateral communication outside the cluster. The resulting information bottlenecks (the distortion or loss of information 'cascaded down' from HQ through ULH) as well as the isolation of the satellite subsidiaries (see Monteiro et al., 2008), was noted by several respondents and acknowledged by a marketing manager as a constant source of tension between the subsidiaries: "Due to the fact that…it is Hungary who is present at the CEE and European meetings, the Croatians and Slovenians feel cut off from the world, since they only receive the materials and things in our interpretation…and this causes some tensions." Of course, the kinds of principal-agent tensions discussed above can be found in any HQ-subsidiary relationship and are not limited to subsidiaries with RMMs and their satellite subsidiaries. Nonetheless, what is remarkable about the tensions we note in this particular relationship is that they are taking place at the periphery of the MNE-the 'edge of the empire'-so far removed from the central locus of decision making that they become virtually invisible to corporate HQ. Thus, in a large, complex firm such as Unilever-given the number of regional management layers between the HQ and operating units-it may appear from the headquarters' perspective that delegating monitoring, coordination and other functions to operating units 'solves' coordination problems, whereas in reality, these problems are merely shifted down to a level that is no longer visible to central decisionmakers. As a result, problems may fester at the periphery, potentially leading to subversion and resistance, with no awareness or intervention from the headquarters. This forms the basis of our final proposition: Proposition 3c. Delegating HQ functions to operating subsidiaries through RMMs does not eliminate coordination problems within the MNE (information bottlenecks, subsidiary isolation and resistance), but merely shifts them to a level where they are less visible to HQ and are therefore less likely to be tackled.
Arguably, these peripheral tensions may be a necessary price to pay in order for HQ to be able to cope with the information processing and decision making demands of its role. However, we speculate that by paying more attention to clarifying and formalising principal-agent roles at all levels of the MNE, the headquarters may be able to reduce the potential for conflict at the 'edge of the empire' and make the overall organisational network run more smoothly.

Conclusions
In this paper, we have built on the recent line of enquiry focusing on the distribution of headquarters functions throughout the multinational enterprise, taking Piekkari et al's (2010) conceptualisation of headquarters as a set of distributed activities rather than bricks and mortar as our starting point. We extended this view by highlighting the important distinction between HQ functions being distributed to dedicated regional headquarters and HQ functions being distributed to local operating subsidiaries in the form of regional management mandates (RMMs). We first encountered the latter phenomenon-previously unexplored in the academic literature-in the field, following in a long tradition of international business research where theorising is inspired by empirical observation of real-life phenomena. Building on our initial observations in the field, our theoretical and empirical investigations were driven by two broad research questions: "What are the key functions of headquarters, and to what extent can they be delegated to operating subsidiaries in the form of regional management mandates?" and "What are the benefits and drawbacks of delegating these functions to operating subsidiaries, as opposed to dedicated regional headquarters?" Our study is particularly timely in that-for all the literature on HQ activities (roles, functions and tasks)-there is a dearth of understanding on how these are performed at different organisational levels, on top of some evident confusion over what the relevant terms really mean and how they are applied throughout the organisation.
One of the main contributions of this paper has been to develop an integrated typology of headquarters' activities (roles, functions and tasks), which-to the best of our knowledge-is the first to systematically compare how HQ activities are performed at three different organisational levels. Our study suggests that most HQ functions can, at least to some extent, be distributed to lower levels of the organisation, with the apparent exception of driving organisational adaptation and institutionalising strategic change (Function 4 in our typology). This finding seems to be in line with other recent studies that suggest this function is the most important responsibility of corporate HQ, which may explain why it is kept at the top Tallman and Koza, 2010). Nonetheless, it is possible that some aspects of this function could in fact be distributed to regional levels, but they simply happen not to be delegated in our case company or in the (arguably limited) regional management literature. The inherent difficulty in trying to reduce such a complex phenomenon to a relatively simple typology also means that there remains a lot more to be learnt about the core roles, functions and tasks of headquarters.
With regards to the choice between regional headquarters (RHQs) and regional management mandates (RMMs), our study suggests that most of the functions performed by dedicated RHQs can also be performed by operational subsidiaries with RMMs, potentially with greater efficiency and at a lower cost. However, there is one function which requires a dedicated regional unit and cannot be delegated to existing local units: attention and signalling (Function 5). The establishment of an RHQ signifies the HQ's attention and commitment to a specific region in a way that establishing operational, profit-oriented units in a country simply cannot-as such, we believe that this finding is likely to hold not only in our case but also in general.
The other main contribution of our study is a set of propositions regarding the potential benefits and drawbacks of using RMMs for regional management. These propositions suggest that RMMs can offer considerable advantages over establishing dedicated RHQs in small regions: (1) balance between integration and responsiveness at levels below the efficient scale for dedicated RHQs; (2) the exploitation of local operational expertise on a regional level; and (3) relieving headquarters of the burden of monitoring remote peripheral agents. In complex MNEs such as Unilever, the HQ is so far removed from many of its local markets that there is a risk of it being 'everywhere and nowhere', and we argue that RMMs form an important part of the toolkit of MNEs to try and develop effectively in new, peripheral or transitional markets. The ability to distribute HQ functions through RMMs allows MNEs to come as close to local markets as possible, without becoming what Bartlett and Ghoshal (1989) call multidomestic MNEs. Historically, this is exactly the type of MNE that Unilever had been, before moving towards greater integration in order to remain competitive in the global fast-moving consumer goods market. Unilever's particular administrative heritage may be a reason why the company has chosen to use RMMs as a way to preserve local responsiveness while also pursuing regional and global integration, and it would be interesting to compare our findings with future studies on the use of RMMs in firms with different administrative heritages (cf. Bartlett and Ghoshal, 1989).
At the same time, our findings also suggest that there are potential risks in using RMMs to manage peripheral 'satellite' subsidiaries at the edge of the MNE's reach: (1) the question of legitimacy and acceptance of operational subsidiaries as intermediate principals; and (2) issues arising from the lack of visibility of ground-level organisational tensions at the top of the MNE. These findings serve as a reminder of the importance of the concepts of authority and visibility, which are traditional hallmarks of headquarters and regional headquarters as actors in the overall corporate network. They also highlight potential avenues for further investigations, such as an empirical comparison between the functions of dedicated RHQs and operating subsidiaries with RMMs; an examination of RMMs from a micro-political and power perspective; as well as a sensemaking and sensegiving approach to RMMs. Together, our typologies and propositions provide a foundation for further research on the distributed functions of headquarters and the use of RMMs for regional management. In-process methodologies on organisational innovation that capture the change and evolution of RMMs, and link these to both the internal and external context of the MNE, would be particularly valuable in exploring our ideas further and enable the development of a dynamic perspective on RMMs. In taking the ideas developed in this paper forward, it would also be particularly useful to conduct comparative studies across different industry contexts. Our study has been limited to the fast-moving consumer goods (FMCG) manufacturing industry, which may constrain the extent to which our findings may be transferred to other industries. However, given the interdependence between the FMCG manufacturing industry and the grocery retail industry, it would not be surprising to find similar regional management structures in the latter, so our findings may be at least partly transferable. Since our work highlights the vital importance of combining local responsiveness with regional integration in the FMCG manufacturing context, our findings can conceivably inform studies on other industries characterised by similar pressures, such as banking or hospitality. Hence, we believe that further research on distributing HQ functions through RMMs in other sectors is strongly warranted. Ultimately, our study underlines the importance of future work on the roles and functions of headquarters, and encourages looking beyond the question of physical location towards questions about the locus of decision-making roles and functions in the contemporary MNE-contributing to what we believe is a very promising and fruitful avenue for future research.