Practice coordination by principles: a contemporary MNC approach to coordinating global practices

Purpose – In the mainstream international business literature on multinational corporations (MNCs), an authoritative central headquarter (HQ) that transfers standardised practices to its subsidiaries remains the norm.This study aims to explore howMNCscoordinate theirmanagement practicesthrough principles. Design/methodology/approach – The paper draws on empirical ﬁ ndings from a qualitative in-depth single case study based on evidence-rich qualitative data including observations from how a high-tech MNC headquarteredin Sweden coordinates itsdevelopment practices. Findings – An alternative informal coordination approach (i.e. coordination by principles) is identi ﬁ ed. Additionally,antecedents andimplications of theapproach are presented. Practical implications – Coordination by Principles may facilitate the internalisation of practices and be a feasible compromise between context adaptation and traditional standardisation, particularly for MNCs with highly heterogeneous researchanddevelopment operations. Originality/value – This paper highlights the importance of acknowledging that ﬁ rm practices often are based on management ideas that HQs adopt to prevent loss of legitimacy. As such, this study contributes to the scarce literature that critically questions the assumption that HQs solely transfer practices to subsidiaries to improve subsidiary ef ﬁ ciency andperformance. is conducted in 30 di ﬀ erent sites that work towards di ﬀ erent types of customers, with an array of technologies and systems in di ﬀ erent life cycle phases. Some develop products in the shape of hardware, software or both and some [develop] services. Imagine that these 30 sites are confronted with the question “ What is the problem ” ? Everybody gets that the answers won ’ t be identical – why would the solutions be identical? (HQ Manager)


Introduction
Coordinating heterogeneous and geographically distant subsidiaries is an essential part of International Management, as the successful implementation of international strategies strongly depends on adequate coordination of the multinational corporation (MNC)'s dispersed activities (Andersson and Forsgren, 1996). Hence, in the field of international business (IB), implementation, alignment and coordination of practices within an MNC's farflung international operations are well-researched topics (Chiang et al., 2017;Yu and Zaheer, 2010). Most studies describe an approach in which headquarters (HQs) replicates, standardises and transfers practices, often imposing them on its global network of subsidiaries. These studies, however, conclude that successful transfer of practices is difficult (Bjerregaard et al., 2016;Kostova, 1999;Morschett et al., 2015) [1]. Although there has been much discussion of the de-centred or "networked" nature of contemporary MNCs (Andersson and Holm, 2010;Egelhoff and Wolf, 2017;Ferner et al., 2012), an authoritative central HQ that transfers standardised practices remains the norm in mainstream IB literature on MNC practices (Fuchs, 2020;Gutierrez-Huerter O et al., 2020). The discrepancies between the literature on MNC practices and the contemporary view of the MNC are especially problematic in light of the growing critical perspective on power and politics within MNCs, where the HQ is suggested to have little ability or interest to exert full control. According to this perspective, the MNC is viewed as a "contested terrain" and the HQ's ability to impose hierarchical control and globally standardised organisation structures and practices are significantly limited (Morgan and Kristensen, 2009;Mudambi et al., 2014).
Furthermore, firm practices are seldom purely internal innovations; they are increasingly based on management ideas such as Lean and Agile (Sturdy et al., 2019). The stream of literature that focusses on management ideas highlights their fashion-like character where new ideas gain rapid popularity and equally rapid fade from view only to be replaced by new fashions (Abrahamson, 1996). As management ideas disseminate and project notions of efficiency, neutrality and universality, they turn into powerful and (nearly) taken-forgranted "regimes of truth". Hence, management ideas are historically inscribed and socially constructed institutions, and therefore neither neutral nor de-politicised (Salles-Djelic, 2019). Despite the literature often pointing out the role of MNCs as both consumers and disseminators of management ideas (Bort and Kieser, 2019;Kern et al., 2019), studies on management ideas in the MNC-context remain scarce. With a few exceptions, the mainstream body of IB literature on practice transfer has shown little concern for whether disseminated practices have been based on external management ideas. Similarly, Sturdy et al. (2019) report that the field of management ideas also has neglected MNCs. Hence, studies on how contemporary MNCs interact with management ideas remain scarce.
Based on the discussion above, this study contributes to the literature on MNCs' global coordination of practices as it shows that in addition to the traditional scenario of authoritarian HQs transferring more or less standardised practices, some contemporary MNCs coordinate their practices globally through principles borrowed from management ideas. Specifically, this study explores how a high-tech MNC headquartered in Sweden coordinates its software development practices at a global level using principles from the management idea of Agile development. We analyse our empirical findings, conceptualised as Coordination by Principles, by posing two research questions: Q1. What makes an HQ use a Coordination by Principles approach? And Q2. What are the implications of such an approach in terms of internalisation and implementation at the subsidiaries?
By going beyond the prevailing monolithic description of practices in the IB literature and incorporating the multidimensional characteristic of practices in our theory-building efforts, this study offers a second contribution by showing how principles can influence the coordination of a global practice within an MNC.
In the next section, we present insights from the IB literature discussing the notion of practice transfer within MNCs and management ideas. Due to the abductive approach of this study but also due to space limitations, we discuss the concepts and insights that are of relevance in our study, rather than providing a comprehensive review of the research topics covered. Thereafter, we discuss the methodological choices made during the study, followed by a portrayal of the case. Subsequently, we present the empirical findings and depict how the firm used principles to coordinate its practices globally. We discuss why an MNC might apply the conceptualised approach and its advantages and disadvantages. We conclude by summarising our findings and discussing our contributions and implications for further research.
2. Theoretical framework 2.1 The dissemination of practices within the multinational corporation As mentioned by Björkman and Lervik (2007), research within the IB field has long sought to delineate how MNCs align, coordinate and/or transfer management practices, predominately human resource management (HRM) practices (Edwards et al., 2007). In the multinational context, the transfer and diffusion of HRM practices can be classified into three types or directions: forward (from the parent to subsidiaries); horizontal (amongst subsidiaries); and reverse (from subsidiaries to the parent) (Edwards and Ferner, 2004). Although the terms "transfer" and "diffusion" are often used interchangeably in IB, some authors (Chiang et al., 2017) suggest that distinctions should be made to yield more nuanced findings. Szulanski (1996), for example, understands diffusion as the flow of practices that gradually disseminate or occur in many different steps, whereas transfer is a one-time event.
A wide-ranging view in the literature is that transferring management practices within the MNC is problematic (Becker-Ritterspach et al., 2010), often resulting in different outcomes across the globe (Gamble, 2010;Szulanski et al., 2003). Kostova and Roth (2002) conceptualise practice adoption in two dimensionsimplementation and internalisation. Implementation is expressed in the objective behaviours and actions required or implied by the practice and internalisation is the state in which the employees at the recipient unit view the practice as valuable for the unit and become committed to the practice. An implemented but not internalised practice faces the risk of being decoupled or only adopted ceremonially (Bjerregaard et al., 2016;Meyer and Rowan, 1977). Lozeau et al. (2002) further found that decoupling was more likely if coercion (i.e. imposition of practice by HQ) was the diffusion mechanism as disallowing local adaptation could reduce the motivation of local management (Morschett et al., 2015). In contrast, adapting management practices in line with different contextual specificities, preferably using informal and subtle coordination mechanisms (Martinez and Jarillo, 1989), can reduce decoupling and increase acceptance of the practice (Ansari et al., 2014). Other factors that influence the degree of local adaptation or mechanisms used when transferring a management practice are the resources available to HQ management to identify and implement its practices (Kostova, 1999), the perceived necessity or desire to diffuse standardised practices (Edwards and Rees, 2006) and the subsidiary's dependence on its HQ (Farndale et al., 2010). The same is true for the subsidiary's strategic role within the MNC (Bouquet and Birkinshaw, 2008) and the size and performance of the subsidiary (Ferner et al., 2011).
Whilst there has been much discussion of the de-centred or networked nature of contemporary MNCs (Andersson and Holm, 2010), an authoritative central HQ remains the norm in the above mainstream IB literature (Ferner et al., 2012). However, recent research has emphasised that the structures of MNCs have become more complex and fragmented (Mudambi and Navarra, 2004). In contrast to mainstream IB, there is a growing critical perspective on power and politics within MNCs that offers a deeper understanding of micropolitical processes and power relations in the MNC based on a more bottom-up and actorcentred approach (Dörrenbächer and Geppert, 2017). This bottom-up trajectory suggests that we study MNCs as "contested terrains" (Edwards and Bélanger, 2009) or as "societies" (Morgan and Kristensen, 2009) with not just one central power unit but diverse power centres each with a different set of influential actors and where subsidiaries perpetually bargain with HQs that have little ability or interest to exert full control. Although studies on power within the MNC primarily focus on the power of subsidiaries (Bouquet and Birkinshaw, 2008;Dörrenbächer and Gammelgaard, 2011;Rego and Steger, 2019), few studies focus on how the HQs manage their subsidiaries without formal authority (the right to decide).
The IB literature has also treated management practices as monoliths without considering the intrinsic components that constitute these practices. However, management practices are multidimensional by nature (Yu and Zaheer, 2010), consisting of people, products and processes (Kedia and Bhagat, 1988) or having a technical or social nature (Kostova, 1999). Yu and Zaheer (2010) argue that considering the underlying dimensions of a management practice is important because different dimensions have different characteristics, which require different mechanisms for adaptation. Ansari et al. (2014) even suggest that management practices should be designed to facilitate transfer and local adaptation. Their line of thinking is similar to Winter and Szulanski's (2001) concept of an arrow corethe attributes of a practice that are replicable and add value according to the business model. As long as the arrow core is transferred, such a partial transfer of practices will give the anticipated and desired results.
Similarly, Lillrank (1995) suggests a three-category model for identifying the different aspects of a practice that should be considered when transferring practices, a perspective used in this study. In an effort to explain why diffusion of management practices is often slow and complicated on an inter-firm diffusion level across countries, Lillrank (1995) argues that the transfer of management practices includes three components: principles, organisational structures and tools. Principles consist of the normative values concerning what is important. From the perspective of Agile, this means, for example, frequent deliveries, teamwork and responding to change rather than following a plan. Organisational structures refer to the organisational structures required for the practice. For example, a practice focussing on flexibility requires an organisation to handle short cycle times and low inventory levels. In addition, this component includes organisational roles such as cross-disciplinary teams, project sponsors and team leaders. Tools refer to practical aides such as statistical quality control and software. These three components are complementary and interdependent, so the successful transfer of one component requires support from the other two components. Because the component Tools has rather clear-cut applications requiring little abstraction, it is applicable in most types of organisations. In contrast, the component organisational structures require a high level of abstraction, making it difficult to transfer, particularly when the geographical and cultural distances are large. The larger the variations between a management practice's institutional context of origin and destination, the more meaning is lost due to misunderstandings. The model by Lillrank (1995) is relevant to this study, as it specifically concerns translations of management ideas across institutional contexts. Further, it has been shown that the model can aid understanding of the transfer process within the MNC context, e.g. in a study by Rolfsen et al. (2014) the model by Lillrank was used to explain the unintended consequences from implementation of Lean at a Norwegian subsidiary of a German MNC. Firms rarely develop their management practices in the house; in many cases, management practices are based on fashionable management ideas such as Lean or Agile (Sturdy et al., 2019). Bort (2015) makes a temporal distinction between management ideas and practices: When an organisation has implemented a management idea, it becomes a management practice. Agile development, for example, can be perceived as both a management idea and a management practice when implemented (Sturdy et al., 2019).

The power dynamics of management ideas
Building extensively on neo-institutional theory, recent research has focussed on the global dissemination of new management ideas. This literature highlights the fashion-like character of this process, with new ideas gaining rapid popularity and then equally rapidly fading from view only to be replaced by new fashions (Abrahamson, 1996). As management ideas spread, projecting ideas about efficiency, neutrality and universality, they turn into powerful and (nearly) taken-for-granted "regimes of truth" (Salles-Djelic, 2019). However, management and management ideas are not the natural habitats of humanity. Instead, they are historically inscribed and socially constructed institutions, and therefore neither neutral nor de-politicised. An important aspect concerning the dissemination of management ideas is the power dynamics between societal actors such as management gurus, business schools and the business press, which determine what constitutes legitimate business knowledge, and therefore what becomes fashionable. O'Mahoney and Sturdy (2016), for example, use McKinsey as a case to show how the dissemination and translation of management ideas are enabled and constrained by the exercise of power. Management ideas can also shift the power distribution within organisations. Birkinshaw (2017) reports that the management idea of Agile can be difficult for executives at established companies, as it shifts power away from those at the top to those closest to the action. Studies on management ideas have been criticised for neglecting MNCs and the power dynamics between their internal actors (Sturdy et al., 2019). Furthermore, Neo-institutional theory, which much of the literature on management ideas is based, has been subject to criticism for underplaying and taking for granted the broader power dynamics on a macro-level (Ferner and Tempel, 2006;Kostova et al., 2008), for instance, the dominance effect put forward by Smith and Meiksins (1995). Smith and Meiksins (1995), taking a macro-perspective, argue that which practices are perceived as best and most likely to disseminate, depends on the practices' country of origin and its position in the international economic system. Their theoretical frameworksystem, society and dominance effectsexplains how organisations are shaped by macro factors including the nature of the capitalist economic system, the societal features of an organisation's location and the characteristics of the economically dominant nation and MNCs. The theory includes a global dominance effect, which recognises a (shifting) hierarchy amongst nations, allowing some of them to influence the developments of others. Hence, the international economic system ultimately determines what constitutes best practices and shapes how these practices are disseminated. Dominance effects also shape systems of meaning as the dominant power becomes hegemonic. On the one hand, dominance effects create a presumption by actors from dominant countries and firms that their practices have superior efficacy, providing a motive for transfer. On the other hand, dominance effects may be shared by actors in the host country, including policymakers, subsidiary managers and workforces, increasing receptiveness to adoption. Thus, the transfer is smoothed because practices are accorded legitimacy by a wide range of MNC actors and are regarded as global "best practices". As a result, ideas flow mostly from the most economically dominant nations. That is, mostly neoliberal ideas from western countries, especially the USA, are promoted and spread (Smith and Meiksins, 1995).
Despite the typically de-centred or network-based nature of the contemporary MNC, the extant debate on the coordination of practices within MNCs is based on studies with predominantly hierarchical MNCs with authoritative central HQs. Furthermore, we argue that an overly simplified depiction of practices as monoliths has been applied and that the above literature neglects intrinsic components of practices. This paper shows how the principles of management practices play an important role in the coordination of a global practice within a contemporary MNC.

Research context
This study focusses on the research and development (R&D) organisation of a global hightech firm headquartered in Sweden (the MNC). At the time of the study, this MNC's globally distributed R&D organisation consisted of approximately 30 development sites located in different subsidiaries. Whilst there were some foreign nationals, the vast majority of the employees at the subsidiaries were locals, for example, at a subsidiary in China with approximately 400 employees a small minority of only 1% were of non-Chinese origin. The orientation of the MNC was primarily geocentric (Perlmutter, 1969), as the subsidiaries' focus was on worldwide, as well as local objectives and an international mindset was aimed for in HQ decisions. However, the MNC's orientation in terms of practices was more polycentric, as the subsidiaries held a high level of practice autonomy and the HQ perceived practice heterogeneity to some extent as necessary. Software development, one of the firm's core processes, had in the past years become informed by the management idea of Agile development.
In contrast to the more well-known management idea Lean production, which focusses on efficiency, Agile development aims to make the software development process more flexible and customer-oriented. Compared to the traditional sequential waterfall model for software development (Rigby et al., 2016), Agile development gives more power to the developer, is less coercive, is more team-oriented and aims to develop in an iterative manner. For example, in Agile development, programmers have increased autonomy and closely collaborate with customers. This tilt in power balance towards programmers has contributed to Agile gaining traction within the programming community. Similar to how Lean development has become the de facto standard in the automotive industry, Agile development has become the de facto standard in software development (Meyer, 2014). According to Lillrank (1995), management ideas such as Agile development can be divided into three components: principles, organisational structures and tools. The principles of the Agile management idea are formulated in a document called the Agile Manifesto (Agilemanifesto.org).

Methodological choice
During another study that we conducted at the MNC, we realised that the MNC coordinated its management practice in a novel approach that had been neglected in the literature. As we had good access to the firm, we decided to explore how the MNC coordinated its software development practices through a set of principles. Hence, the exploratory nature of the study led us to use a single-case study approach (Easton, 2010;Siggelkow, 2007;Yin, 2003). Once we decided to study the approach in detail, we made the first round of data collection focussing on developing a chronological reconstruction of the trajectory of decisions that led to the new form of practice coordination within the MNC (Langley, 1999). During this first round of interviews (marked in bold in Table 1), the respondents were asked to tell their story of the introduction of the Agile management practice with a focus on the various steps taken.
In the next step of our abductive journey, we reviewed the literature on coordination of management practices within MNCs in more detail (summarised in 2.1). From our first round of interviews, we realised that the different components of management practices played an important part in our case, especially principles. However, the extant literature in IB on the coordination and transfer of management practices is predominately based on a simplistic perception of management practices as monoliths, ignoring their intrinsic components. Thus, we broadened our search for relevant literature and learned that some scholars in related disciplines such as Economic Geography and Organisational Studies, understand practices to consist of several components. We became particularly inspired by Lillrank (1995), who argues that a management practice consists of three components (principles, organisational structures and tools), which differ in their degree of transferability. Drawing on Lillrank (1995) and the insights from our first round of interviews, we started to elaborate on an initial defining understanding of the Coordination by Principles approach used by the studied MNC. Next, we used the second round of interviews to refine and enrich our evolving conceptualisation of Coordination by Principles and explored the motives and implications of the approach. Hence, the questions were more specific as they focussed on understanding why the MNC used a Coordination by Principle approach and its implications.
Here, insights from our detour to the literature on MNC practice management were essential. Following Dubois and Gadde (2002), we stopped the data collection and searched for useful theories after we felt that we had a final conceptualisation that matched the theory with the data.
As noted above, we collected the data in two main rounds. In total, we interviewed 36 respondents covering all major roles in the R&D organisation. We also held shorter followup interviews with several respondents. Hence, 46 interviews were conducted. To obtain an industry perspective on software development practices, we also interviewed two senior software developers outside the MNC. The focus of each interview was largely determined by the responses and emergent themes, but an interview guideline provided the core questions and was adjusted continuously. The interviews were conducted in either English, Shanghainese, Mandarin or Swedish, lasted between 55 and 80 min and were all recorded with permission. Irrespective of the language, all interviews were translated into English (resulting in approximately 380 pages of transcription) in preparation for collective analysis in Nvivo. Reflection notes were taken during each interview and observation. Table 1 summaries the conducted interviews, including the role of the interviewees, their locations (due to anonymisation requirements only broadly depicted) and interview date.
In addition, observations were conducted at two subsidiaries, one in China and one in Sweden. At each site, one of the authors was provided a work desk between two Agile teams. During six weeks in spring 2016 (Sweden: two days per week for three months in spring 2016; Shanghai: five days per week for three weeks in March 2016), the author shadowed the daily work of the teams and attended all their meetings (Gherardi, 2006). The ability to both collect the data and analyse it in its real setting offered the opportunity to think of aspects that might not have been possible to identify solely through interviews (Barley and Kunda, 2001). This combined observation method (i.e. combining interviews with overt observations) can be compared to what Czarniawska (2007) refers to as shadowing people during their daily work. The use of ethnographic approaches in researching management practices in the context of the MNC is also suggested by Diana Sharpe (2018). Our third source of data was internal documents from the firm's intranet (256 pages of material). Because the study is partly retrospective, the three sources, as well as a large number of interviews allowed for triangulation, reducing the risk of posthoc rationalisation of previous actions, thoughts and decisions, as well as problems related to memory (Bryman and Bell, 2011). Furthermore, to strengthen the study's reliability, respondent validation (van de Ven and Poole, 1990) was applied by discussing the findings with change agents at the company as work progressed.
The analysis of our collected data can be divided into two mains steps. Firstly, using Nvivo, we chronologically coded our findings from our first round of data collection as we wanted to write a case narrative and construct a timeline of the process that led to the existing development practices and how they were coordinated. As mentioned above, this initial data also provided us with insights that we used to sketch a preliminary conceptualisation of the HQ's approach to coordinating practices. In our second round of interviews, we asked specific questions to gradually refine the preliminary conceptualisation. The second step of our analysis took place after the second round of interviews where several cycles of data coding were performed by one of the authors. Analysis progressed from descriptive codes (Lillrank's (1995) three components) to aggregating them to broader concepts and emerging themes, which clustered around the antecedents and implications discussed in Section 5.

Empirical findings: Coordination by principles
Spurred by the need to manage an increasingly heterogeneous global R&D organisation, the HQ was forced to abandon the prevailing software development process that was based on a globally standardised project management model -Plug. Indeed, with the introduction of Plug more than 30 years earlier, the number of global R&D sites had multiplied and the heterogeneity of the MNC, in terms of both technology and local culture, had increased because of growth in the product portfolio. Each R&D site applied different technologies when developing hardware, software or a combination of both and served different types of customer categories. As mentioned by several R&D managers at the HQ, they also searched for a common denominator of the R&D organisation's heterogeneous operations, asking themselves "What has been and what will be permanent in what we do"? The HQ concluded that large-scale systems development was the answer to this question and that such a development practice required flexibility to cater to the heterogeneity of the R&D organisation: Basically, I don't give a damn how people work. However, I do have an opinion on how their mindset and their values are manifested based on our principles. We should all value the same things such as optimisation of flow. There, I have much less understanding of variations. I see this as a part of our corporate culture that we manifest through principles. (HQ Manager) In 2014, when more than a handful of subsidiaries already had replaced Plug with Agile practices on their own initiative, the HQ decided that Agile was the way forward. Based on the consensus that the R&D organisation should have one what and many hows, HQ, together with representatives from the divisions, developed a framework called the Product Development Principles. These 16 principles aimed to unify all the R&D sites' individual development practices. The idea was to let the R&D sites be autonomous regarding the hows whilst being united on the what: One may see the product development principles as a foundation for our development practices. Based on this foundation, each division adds practices that are specific to them and each sub-unit then further adds or specifies practices, as long as the principles are applied. That's how we want it to work. (HQ Manager) As the HQ's product development principles were translations of the Agile principles, they called the change process the Agile transformation. By this time, Agile had been diffused on a global level and was quickly becoming the new industry norm in software development and a handful of subsidiaries had independently already adopted Agile. Despite the evident similarities between the firm's new development principles and the principles of Agile, however, HQ deliberately excluded references to Agile and other management ideas so as to avoid Agile becoming an end in itself. Below is a list of some of the principles that the HQ formulated for successful product development: [. . .] we continuously optimise the flow of value.
[. . .] we use architecture as enablers for high quality and modularity.
[. . .] we continuously reflect on our results to improve our products, flows and ways of working.
[. . .] we develop working products in short cycles, with fast feedback loops.
[. . .] we work in empowered cross-functional teams, with an end-to-end view.
[. . .] we are structured and systematic in everything we do.

(from internal documents of the firm)
Basing the principles on the Agile idea, which had been established as an industry norm, proved beneficial. Whilst few respondents were aware of HQ's development principles, there was a high awareness of the principles in the Agile manifesto. By basing the product development principles on a well-known management idea, HQ avoided competing for institutional logic (Thornton and Ocasio, 2008) and could spend less time and effort on internalising its principles. In this case, knowledge on how to work according to Agile practices did not disseminate internally through the MNC network from the HQ or from any subsidiary, for example, through people travelling between the sites. Instead, as the Agile management idea had diffused globally each site was able to learn Agile development from local consultants and training firms. Hence, knowledge of the new practice flowed into each subsidiary from their respective local context. This knowledge flow stands in contrast to traditional practice transfer within MNCs where knowledge of new practices flows internally, often from HQs to subsidiaries.
Considering Agile's growing legitimacy in the industry and considering that several subsidiaries had already taken the initiative to switch to Agile, one may question whether HQ actually had a choice of whether or not to adopt Agile. HQ managers were not as receptive to Agile as was the R&D organisation as a whole. One HQ manager described some developers as "Agile fanaticspeople who are more interested in working according to Agile for the sake of Agile, rather than solving a problem. It is extremely important for these individuals that you use the correct word, rather than achieving an effect". Another manager at HQ made the following self-critical reflection on how the firm's Agile transformation was a reaction to Agile's advance: "Some researchers say: 'Business first, then the organisation'. We have been masters of doing the opposite".
Whilst appreciating standardisation for the cause of internal coordination, HQ shared the opinion of respondents from subsidiaries that one single standardised development practice could not meet the requirements for the whole R&D organisation's heterogeneous operations. One manager at HQ held the opinion that a standardised development practice would do more harm than good: Our R&D is conducted in 30 different sites that work towards different types of customers, with an array of technologies and systems in different life cycle phases. Some develop products in the shape of hardware, software or both and some [develop] services. Imagine that these 30 sites are confronted with the question "What is the problem"? Everybody gets that the answers won't be identicalwhy would the solutions be identical? (HQ Manager) According to the managers at HQ, the heterogeneity of the R&D operations was a crucial reason why they did not opt to standardise and transfer one R&D practice to all the sites. Another HQ manager described the heterogeneity of the R&D operations as follows: "[T]he only thing the product development sites had in common was that they all conducted largescale software development". This meant that large-scale software development was a common denominator of the R&D centres. When sites cooperated with each other and jointly developed software systems, the sites developed modules that were put together into a system, like building construction with Lego building bricks. How each site developed their respective modules didn't matter that much as long as the modules functioned according to specification, this facilitated international collaboration. In addition to the heterogeneous nature of the MNC's R&D operations, the operational nature of the software development practice, involving limited decisions with strategic impact, also facilitated decentralisation.
The egalitarian characteristics of the Agile management idea kept HQ from imposing a standardised full development practice through coercive coordination mechanisms. In contrast to traditional software development practices that have more of a top-down approach, Agile shifts power from management to the development team. According to the Agile idea, the Agile team itself not only should decide what to do and how long it should take but also report directly to the customer, not to a manager. One software designer described the difference: "The whole point [with Agile] is to shorten the path between the one who receives the task from the customer and the one who actually carries out the task. To achieve this, you need to flatten hierarchical pyramids". Indeed, one of the core principles of the Agile idea is that a firm's leaders should serve rather than command its developers. According to Agile, managers should practice servant leadership, serving the team so its members can focus on development (Dierendonck, 2011). I see my role as [that of] a servant leader, in the sense that I perceive myself as a facilitator rather than a manager. I take meetings with different stakeholders and take care of administrative duties, to ensure that the team can focus on the development. (Scrum Master) Whilst interviewees at the subsidiaries often expressed the opinion that HR and financerelated practices were too centralised, none of the interviewees mentioned feeling forced to follow any R&D practice dictated by HQ. Instead, a sense of autonomy was often mentioned concerning their development practice.
When we have tried new ways of working, the HQ has been aware, as we have had very open communication and shared what we tried and whether it went well or not. At the time we started with Agile, HQ wasn't really into it; still, they were quite supportive and I never felt controlled. (Subsidiary Manager A) Figure 1 provides an overview of how the firm is coordinated by principles. Using the wording of Lillrank (1995), number 1 in the figure illustrates that the HQ adopted and translated the principles of the Agile idea but left out the other two components of the management. The HQ's translation of the Agile principles constituted the company's new development framework (number 2 in Figure 1), a framework of principles that intends to align the development practices of the heterogenous subsidiaries. The subsidiaries were asked to apply practices that were aligned with the new R&D framework; within the framework, subsidiaries enjoyed full freedom concerning the other two componentsi.e. tools and organisational structure (Lillrank, 1995).
The HQ used three main tools to implement the new development principles and align practices (number 3 in Figure 1): layered specification, training of managers and dissemination of knowledge through the intranet. Layered specification meant that each layer of the organisation was welcome to add complementary practices in a bottom-up manner. The HQ preferred to see its R&D principles as the basis of a platform; therefore, based on the principles, each division would add general practices specific to that division. For example, one division added practices they called continuous delivery and release, iterative design and close interaction. The second part of the implementation of the new R&D framework was manager training. The firm had a long tradition of conducting internal courses that included managers from the HQ acting as teachers. An example of these training initiatives was a development leadership course that HQ arranged several times a year, with more than 100 R&D managers from all over the globe participating annually. In courses such as these, the answer to the fundamental question "How should we work"? Was "According to the principles". These courses served several purposes, but from the perspective of the Agile transformation, the goal was that knowledge of the development framework would trickle down in the organisation from the participating managers. Finally, the HQ also diffused its development principles through the firm's intranet, which had played an important role in the firm's internal communication for decades. An internal Wiki contained an abundant amount of material concerning the firm's R&D framework at the group, division and unit levels.

Discussion: antecedents and implications
In this section, we discuss the antecedents resulting in the choice of using the coordination by principles approach and the implications of such an approach in terms of implementation and internalisation of the practice.

Antecedents for coordination by principles
As we discuss in detail below, we identify four antecedents (Table 2) for Coordination by Principles in our case study: the nature of the MNC, the presence of a management idea with a high degree of legitimacy within the industry, HQ's ability to perceive the variety of components of the management practice and the characteristics of the management idea.
5.1.1 The contemporary nature of the multinational corporation. The power dynamics between the HQ and subsidiaries of the MNC rendered a favourable condition for the adoption of management ideas. For example, the HQ subsidiary relationship promoted a sense of autonomy concerning the development practices frequently expressed by respondents at subsidiaries. In the power dynamics between HQ and subsidiaries, we perceive the case firm as an example of the contemporary MNC (Bartlett and Ghoshal, 1989;Hedlund, 1986;Nohria and Ghoshal, 1997) that questions the HQ as the sole generator of business advantages and strategic thinking. The subsidiaries' practice autonomy indicates that they have a relatively high degree of power of process (Hardy, 1996), facilitating both adoptions and internal diffusion of management ideas. Most likely, an MNC that applies a Coordination by Principles approach will show characteristics of a networked firm (Andersson and Holm, 2010). That is, MNCs that comprise integrated and interdependent networks of different but equivalent subunits in which the HQ does not play a dominant role but whose main task is to facilitate knowledge transfer between the subunits. Accordingly, the main challenge for an HQ following coordination by principles approach is to ensure that subsidiaries stay within the framework and do not abuse the freedom.
Whilst HQs hold formal authority (the right to decide) over subsidiaries, HQs may for various reasons not use this authority; this is especially likely for MNCs that give high autonomy to their subsidiaries. HQ's use of a management idea with industry legitimacy to coordinate a management practice amongst its subsidiaries, as depicted in our case, calls to mind the concept of institutional power. Dörrenbächer and Gammelgaard (2011) describe institutional power as a type of power held by subsidiaries that can refer to host country institutional pressures when they are confronted with undesired HQs demands. We suggest that the use of institutional power does not lie exclusively in the hands of subsidiaries, as HQs also can use this type of power to manage their subsidiaries. For example, our findings show that the implementation of HQs new development principles was facilitated by the institutional pressure on the subsidiaries to adopt Agile from both the software development community and from progressive customers and other industry players.
5.1.2 The dominant institutional logic of a management idea. Coordination by principles is greatly facilitated if the principles are based on a well-known management idea that has legitimacy in the industry. For example, the firm's software developers were less aware of HQ's development principles than the principles of the management idea Agile that they were based on. We argue that the HQ had limited choice but to adopt Agile due to the dominance effect (Smith and Meiksins, 1995) and the Agile expectations from the subsidiaries. A couple of years before the MNC's change of development practice, Agile had become a trend within the epistemic community (Knorr-Cetina, 1999) of software developers. The popularity of the Agile management idea can be attributed to the dominance effect as it originated in Silicon Valley in the USA, the dominant nation in contemporary software development. The Table 2.

Coordination of practices by principles
Coordination of practices by principles Characteristic -HQ needs to strive for control AND local adaptation. To balance these confronting needs, HQ sets a framework consisting of the principles of the management practice to which subsidiaries need to adhere. Within the framework (i.e. related to choices of tools and organisational structure), subsidiaries enjoy full freedom Antecedents -The contemporary nature of the MNC -The dominant institutional logic of a management idea -HQ's ability to perceive the variety of components of the management practice -The idiosyncrasy of the management idea Coordination mechanisms used -Both formal (setting the framework) and informal (allowing for freedom regarding organisational structures and tools) Advantages -Increased chances of local internalisation and implementation -Low risk of not-invented-here syndrome/resistance -Essential componentprinciplesis aligned in the global organisation Disadvantages -HQ needs to ensure that subsidiaries stay within the framework and do not abuse the freedom; control is difficult and resource-demanding -Less integration compared to fully standardised practices dominance effect shaped the system of meaning (Smith and Meiksins, 1995) amongst subsidiary managers and increased their receptiveness to adopt Agile. In addition to the external dominance effect, there was an internal expectation for Agile on HQ as several subsidiaries had adopted the management idea on their own initiative. Hence, if HQ had not based the new principles on the Agile idea there would have been competing institutional logics (Thornton and Ocasio, 2008) on how software ought to be developed, as Agile was the prevailing institutional logic amongst software developers in the subsidiaries. Thus, from a legitimacy point of view, HQ might not have had a choice of whether to adopt the management idea. We perceive the HQ as an actor with limited ability to exert full control over the MNC's network of subsidiaries, a perception in line with the critical perspective on power and politics within MNCs (Dörrenbächer and Geppert, 2017;Edwards and Bélanger, 2009;Morgan and Kristensen, 2009), a perspective which emphasise the role of the micro-politics in the diffusion of practices within MNCs (Elger and Smith, 2005;Morgan and Kristensen, 2006). From the perspective of MNCs as "contested terrains" (Edwards and Bélanger, 2009), we argue that there would have been more power struggles if the principles by HQ had not been in line with the Agile management idea.
5.1.3 Headquarter's ability to perceive the variety of components of the management practice. Our findings show the importance of acknowledging that management practices consist of several components. Lillrank (1995) suggests that management practices consist of three components of which one is called principles and consists of the normative values of management practice. One HQ Manager framed this understanding as follows: Neither the principles of Agile nor the principles of lean software development tell you "how" to do thingsthey state "what" is important. Then we thought, "Perhaps, we should define our own specific principles"? Which we did and [this] resulted in the product development Principles.
Using this perspective, the HQ formulated a general message to the subsidiaries: Subsidiaries may shape their development practices as long as they follow our practice principles, which are based on the principles of Agile. Hence, whilst providing the subsidiaries with a high degree of freedom, the HQ used the principles of Agile as a uniting common denominator.
5.1.4 The idiosyncrasy of the management idea. Most management ideas have unique characteristics, including how power between actors ought to be distributed. Agile, for example, shifts power from those at the top and to those closest to the action, a difficult shift for executives at established companies (Birkinshaw, 2017). Hence, if the HQ in our case had imposed fully standardised practices onto its subsidiaries, this would have conflicted with the egalitarian aspect of the Agile management idea that prevailed amongst developers. The idiosyncrasies of Agile also conflict with formal coordination mechanisms such as standardisation, formalisation and centralisation, as these mechanisms are best suited for static problems and rarely suited for highly dynamic tasks in complex environments (Morschett et al., 2015). Whilst Agile aims for flexibility and new innovative solutions, formal coordination mechanisms lead to standardised solutions that might constitute a barrier to these aims.
In summary, antecedent one concerns the power dynamics between HQ and the subsidiaries, particularly a high degree of freedom of subsidiaries. The second antecedent concerns the presence of a management idea with a high degree of legitimacy within the industry. The third antecedent refers to the importance of perceiving a management practice consisting of a variety of components by HQ. Finally, antecedent four concerns the characteristics of management ideas, e.g. how power ought to be distributed between actors according to the logic of management ideas. Whilst related, these are separate antecedents, as they concern four different aspects.

Implications on implementation and internalisation
We derive three related implications of coordination by principles from our study. Firstly, barriers related to the institutional context are reduced. Secondly, internalisation of new practices is facilitated as a result of the autonomy of subsidiaries to enact principles. Thirdly, coordination by principles inevitably decreases global integration compared to coercively imposing a standardised management practice. These three implications are elaborated on below.
Firstly, we suggest that coordination by principles has the potential to reduce barriers to the new practice related to the local institutional context. When management ideas travel between institutional contexts, unexpected consequences have shown to be common when they are implemented due to barriers of various nature (Mantalay and Chakpitak, 2014). Barriers related to the institutional context of the host country are called institutional barriers by Zimmermann and Bollbach (2015) and can consist of formal institutions (such as governmental and educational institutions), as well as informal institutions such as the cognitive and normative dispositions that guide the behaviour of the members of the institutional context. These institutional barriers can be mitigated using coordination by principles. Drawing on Lillrank (1995), we conclude that the HQ only standardised the principles component of the practice, a move that provided the subsidiaries autonomy for the other two components, organisational structures and tools. We see organisational structures and tools as actionable enactments of the principles and argue that these two components are not ends in themselves, so these two components should vary to suit the institutional context, particularly the organisational structure component.
Our second implication suggests that coordination by principles facilitates internalisation of practice compared to traditional transfer of a standardised full practice, as this strategy increases subsidiary managers' sense of local ownership. Similarly, Kostova and Roth (2002) found that imposed practices risk being ceremonial (Meyer and Rowan, 1977) as internalisation is more challenging than the implementation of practices. As our case shows, internalisation of the practice principles amongst the MNC's software developers was not an issue, as they were based on the Agile idea, which already held legitimacy within the global software development community. This implication is also in line with findings from Ansari et al. (2014): subsidiaries have given greater latitude to adopt a standardised management practice exhibited a greater sense of local ownership. Ansari et al., argue that having a sense of local ownership is necessary for building organisational and individual commitment to implement a management practice throughout the MNC. Our argument is also in line with Yu and Zaheer (2010): adaptation of standardised management practice can further increase the effectiveness of the practice as it helps people accept the new practice. In our case, the HQ of the MNC only provided the product development principles as a platform and allowed each vertical organisational layer to specify the new development practice. As presented in the findings, the practice became increasingly specified further down in the organisational hierarchy. This joint practice specification gave subsidiaries the mandate to design the components of Organisational Structure and Tools themselves, which facilitated internalisation by decreasing the risk of the not-invented-here syndrome (Birkinshaw and Ridderstråle, 1999).
A third apparent implication of only transferring the principles component of management practice is a decreased degree of global integration. With this implication in mind, the literature does not advocate that all business functions of the MNC should be globally integrated into the same manner or degree. In their study of how integration modes of MNC business functions contribute to financial performance, Kim et al. (2003) conclude that for the global integration of R&D, coordination through people and information systems is more effective than centralised decision-making and standardisation of work procedures, rules, policies and manuals. Thus, MNCs should not coordinate their global R&D operations through standardisation of practices but coordinate through a shared vision, values, norms and trust. The necessity of an equal degree of global integration amongst all functions is also absent in the contemporary view of MNC (Faulconbridge, 2010;Hedlund, 1986). Nonetheless, our case illustrates that there are methods HQ can apply to strengthen global integration. The HQ of the firm applied two methods to strengthen the alignment of the practice on a global level: global training of managers with HQ representatives as teachers and dissemination of practice knowledge through the intranet. Without these two informal coordination methods, it is our belief that the development practices may have sprawled beyond all control.

Conclusions
In contrast to traditional practice transfer in the context of the MNC (Jensen and Szulanski, 2004;Kostova, 1999;Kostova and Roth, 2002), this study demonstrates how an MNC coordinates its management practices through principles. In practice, this is achieved by enforcing coercive pressure of only one component of the management practice, the principles. This study adds knowledge to the scarce, although important, literature that questions the assumption that HQs are solely motivated to transfer practices to subsidiary units that will improve subsidiary efficiency and performance but that practices are transferred to prevent longer-term economic and social/legitimacy losses (Collings and Dick, 2011). As shown through our case, the HQ simply felt the pressure to introduce the Agile management idea in the global organisation due to both internal and external industry expectations.
Referring to our research questions, we demonstrate that there are four main antecedents to why MNCs coordinate practices through principles: the contemporary nature of the MNC; the dominant institutional logic of a management idea; HQ's ability to perceive the variety of components of the management practice; and the idiosyncrasy of the management idea.
We further theorise on three related implications of coordination by principles. Firstly, we suggest that barriers related to the institutional context can be reduced compared to the traditional transfer of a standardised full practice. Secondly, internalisation of new practices is facilitated as a result of the autonomy of subsidiaries to enact the principles. Thirdly, compared to a coercively imposed standardised management practice, the approach may result in a decreased sense of global integration. Coordination by principles may provide a feasible compromise between context adaptation and the traditional standardised practice transfer. In contrast to the formal coordination mechanism seen in the literature on practice transfer, these insights contribute to an alternative informal approach to how the MNC coordinates its practices. As an informal coordination mechanism, Coordination by principles is in line with Martinez and Jarillo's (1989) observation that MNCs are increasing their use of informal coordination mechanisms. We recognise that the generalisability of this study is an important limitation. As Johnston and Menguc (2007) inform us, it is well known in the international management literature that MNCs from different countries have varying predilections with regard to subsidiary autonomy and that Swedish MNCs have a tendency to decentralise decisionmaking more than others. As this study draws on a single case study, generalisations may be limited. However, even a single case can be considered a methodologically fully acceptable approach to establish patterns as long as it meets the established objective. To assess the study's generalisability, future studies should explore how common coordination by principles is amongst MNCs with HQs based outside Sweden.
In this study, coordination by principles is presented as an internal coordination approach in the MNC context although the coordination approach can also be observed in how some global lead firms coordinate their global production network (GPN) (Coe and Yeung, 2015). For example, Nadvi et al. (2011) show how Adidas and Nike supported local suppliers in Pakistan to reconfigure their production organisation to follow the principles of Lean production. As power relations between actors is one of the focus areas of the GPN concept, future studies should focus on the impact on these relations from management ideas to further develop the GPN concept. Management ideas do not only affect power relations between actors within MNCs but also MNCs' external relations with other actors in their GPNs.
We provide three recommendations for managers. Firstly, coordination by principles may be particularly suitable for MNCs whose operations are highly heterogeneous. Secondly, internalisation of practices can be facilitated by aligning them with the principles of a management idea that has legitimacy in the firm's industry. Thirdly, in MNCs where subsidiaries have the authority to decide their practices, HQ managers should be aware that as the legitimacy of a management idea increases in the industry, the likelihood that it will be adopted by subsidiaries will also increase. The more subsidiaries adopt a management idea, the harder it will become for the HQ to reject it. Note 1. Extant literature on how practices are managed within the MNC use terms such as transfer, implementation and coordination. Thus, it is challenging to avoid mixing these terms whilst at the same time staying true to the references. We share the definitions of transfer and implementation with Kostova (1999), who views transfer as an intentional process by a transferring actor (e.g. the HQ). The process consists partly of the diffusion of a formal set of rules and partly of the effort to internalise the rules at the recipient unit level. Implementation is defined as "the degree to which the recipient unit follows the formal rules implied by the practice" (Kostova, 1999, p. 311). For the definition of coordination, we follow Martinez and Jarillo (1991), who view coordination as the process of integrating activities that remain dispersed across subsidiaries.