Crossing the streams: HRM in multinational enterprises and comparative HRM

Available online xxxx This paper examines the differences between the stream of international HRM that focuses on comparative HRM and the one that focuses on HRM in multinational enterprises (MNEs). More specifically, we review how the aspects of time, process and context have been treated within the two streams and argue that the streams have largely developed in isolation as opposed to informing one another. Drawing on this analysis we propose a research agenda that illustrate how the crossing of these streams can advance theory and empirical research in both HRM in MNEs and comparative HRM, to the benefit of both. © 2016 Published by Elsevier Inc.


Introduction
One of the first scholarly analyses of human resource management (HRM) by Beer, Spector, Lawrence, Quinn, and Walton (1984), paid considerable attention to the influence of context. In the Beer et al. 'map of the territory' the external environment and in particular the interests of different stakeholders and situational factors were linked through the organisational HRM system to long-term outcomes for the organisation, the employees and the community. However, research since that time has, mostly, been conducted in single countries and much of it has been focused on the activities of HRM specialists within organisations. The International HRM (IHRM) literature has eschewed this narrow focus, especially within its two dominant discourses -HRM in multinational enterprises (MNEs) and Comparative HRM (CHRM).
Much of the content of these two streams of IHRM is shared, in particular the standardisation-differentiation debates, as we indicate below. Both streams also share an interest in how time, process and context affect HRM activities in different parts of the world. However, the three issues are handled rather differently, partly as a result of the two streams having evolved in parallel rather informing each other. After briefly delineating the HRM in MNEs and Comparative HRM streams, we use these differentiating issues to structure our review and contribution.
HRM in MNEs covers a broad range of HRM issues that MNEs face across national borders (Stahl, Björkman, & Morris, 2012), attempting to address the key overarching questions of how people are managed in MNEs and what the outcomes are (Björkman & Welch, 2015). Grounded, to some degree, in the international business and MNE management literature (e.g. Bartlett & Ghoshal, 1987;Prahalad & Doz, 1987), two of the dominant themes in this discourse are headquarters-level HRM practices and their transfer to foreign subsidiaries (Welch & Björkman, 2014). Using contingency-type frameworks, this literature finds that the HRM practices in MNEs and their foreign subsidiaries are contingent upon a complex interaction of external factors relating to the home-and host-country contexts, internal factors relating to the strategy and structure of the global corporation and the subsidiary, and the nature of the headquarters-subsidiary relationship (Edwards & Kuruvilla, 2005).
CHRM, on the other hand, looks at commonalities and differences in HRM between countries (or sometimes regions within countries) or between clusters of countries or geographical regions (Brewster & Mayrhofer, 2012). In this literature, persistent differences between countries have been found not just in the way that HRM is practiced but in the way it is conceived of and evaluated. In other words, the roles of stakeholders and the intended outcomes of HRM are viewed differently in different countries.
The clearest connection between these two different streams lies in their shared interest in the standardisation-differentiation debate. HRM in MNEs approaches the debate predominantly by looking at HRM practices at the organisational level, in particular within and across MNE subsidiaries (e.g. Rosenzweig & Nohria, 1994). Since it is clear that countries are different, MNEs have to consider this and differentiate their HRM practices. However, there are also several reasons why MNEs may want to standardise certain aspects of their HRM practices across their global operations: it is more cost-efficient since 'wheels do not have to be re-invented', it allows for the smoother transfer of best practices around the organisation, it makes transfers of people easier and it is fairer (Evans, Pucik, & Björkman, 2011a, 2011b. This stream of literature generally assumes that MNEs will want to standardise practices wherever they can, but are prevented from doing so by local circumstances. The CHRM literature generally refers to convergence or divergence and notes that MNEs do not standardise practices even when they have the opportunity to do so. In short, CHRM examines the standardisation of HRM debate predominantly by looking at differences across countries. The processes of standardisation and differentiation are also of interest to both streams of literature. HRM in MNEs research has paid considerable attention to the means by which HRM practices are 'transferred' within the organisation across national boundaries (e.g. Gamble, 2010;Smale, Björkman, & Sumelius, 2013). The CHRM literature, on the other hand, has paid more attention to what extent and how certain HRM practices are diffused around the world.
Also linked with the standardisation-differentiation debate is an implicit or explicit interest in understanding developments over time. In both streams, the question of a globally emerging common model of HRM, be it on the national or organisational level, is a key issue. In HRM in MNEs, assessments about whether a common HRM model can be replicated across the organisation has immediate practical consequences for organisations (Morris et al., 2009) and important implications for teaching and consulting. In CHRM this issue relates to the dominance of underlying basic assumptions about HRM in different parts of the world and whether they are converging (Brewster, Mayrhofer, & Cooke, 2015;Festing, 2012;Mayrhofer, Brewster, Morley, & Ledolter, 2011).
This article offers a research agenda that illustrates how the crossing of these streams can advance theory and empirical research related to convergence and divergence in both HRM in MNEs and CHRM, to the benefit of both. In a first step, we briefly review the HRM in MNEs and CHRM streams and present key reasons why the two streams have crossed so seldom in the past. In particular, we examine how the aspects of time, process and context have been treated within these streams. Drawing on the conclusions of this review and using some established theoretical insights, we then develop key points of a research agenda that can advance theory and empirical research on the issue of convergence and divergence.
2. Time, process, and context: a review of the comparative HRM and HRM in MNEs streams Despite widely varying conceptualizations of time (e.g. Zerubavel, 1981), we focus here on the simple view of organising events into a sequence or chronology (similar to Ellis McTaggart's (1908) 'A series of time'). Process is important because it elucidates the 'how and why' behind major issues discussed in both streams and their developments over time. Finally, context is crucial since both streams examine its effects on HRM.

The role of time
Time has played an intriguingly minor role in International HRM research, given its significance. This partly arises due to the difficulty of researching complex issues over time. Given the required investment of resources and the difficulty of collecting and maintaining data, the quantitative data tends to be varied in its scope and its detail.
The issue of time, and its relationship to standardisation and differentiation, has received little empirical attention within the HRM in MNEs stream. Most quantitative studies of HRM in MNEs have been based on cross-sectional data and, predominantly, have used subjective and varying measures or classifications of HRM practices, rendering it almost impossible to make inferences about patterns of standardisation or differentiation over time. There have been some notable exceptions. For example, Björkman, Smale, Sumelius, Suutari, and Lu (2008) examine the HRM practices of European subsidiaries in China in 1996 versus 2006. Although they were not matched samples, they found that the HRM practices in Chinese subsidiaries had come to more closely resemble MNE headquarters' HRM practices as well as the HRM practices of local firms. Evidence of HRM convergence over this ten-year period, not only between local Chinese firms and foreign subsidiaries, but also between European MNEs, was explained as being due to a combination of MNE and host-country coercive, mimetic and normative institutional pressures. Other research comparing HRM in subsidiaries has argued that there is a 'dominance' of (US-style) 'best practice' that is creating an inexorable move towards convergence of practice across foreign subsidiaries (Pudelko & Harzing, 2007).
There has been some insightful, longitudinal-type, qualitative work, mostly in the form of case studies. These have revealed the important role that power relations and micro-political processes play in determining the use and effectiveness of different HRM control mechanisms over time (Martin & Beaumont, 1999;Ferner, 2000). They have contributed a more dynamic view of HRM standardisation by demonstrating that standardisation, centralisation in particular, should be seen as being subjected to continual negotiation between parent and subsidiary, and is thus better viewed as comprising contested processes of 'oscillation' between greater global integration and greater local responsiveness Ferner, Almond, & Colling, 2005;Sippola & Smale, 2007). However, whilst case studies can be informative about mechanisms (see the next section about process), they are a poor way of judging overall trends or showing representative change over time.
Despite these exceptions, time generally not being integrated explicitly into research designs, the HRM in MNEs stream of research provides a fairly weak empirical basis on which to contribute to the debate about the globalisation of HRM over time, either at MNE or country levels. At the level of the MNE this is regrettable given the propensity for MNEs to go through 'pendulum swings' in terms of their focus on HRM standardisation versus differentiationas noted in the ABB case (Bélanger, Berggren, Björkman, & Kähler, 2000;Evans et al., 2011aEvans et al., , 2011b. The likelihood is that approaches to the standardisation-differentiation of subsidiary HRM practices will change over time as part of the MNE's internationalisation and 'tortuous evolution' (Bartlett & Ghoshal, 1989;Perlmutter, 1969). Indeed, Farndale et al. (2010) adopt a contingency approach to illustrate how changes in the environment, strategy and structure of leading MNEs are affecting their HRM function configurations in different ways.
By contrast, there is a much heavier emphasis on understanding changes over time in the CHRM stream. Part of the reason for this stems from the heightened interest of CHRM in how HRM in different countries develops over time and whether one can see a common model evolving. The globalisation, or convergence, debate is a lively issue in CHRM. Beyond cross-sectional analyses providing static snapshots of commonalities and differences, addressing this issue requires longitudinal analyses. Such analyses have been greatly helped through the data collection efforts of Cranet (www.cranet.org) which has provided the CHRM field with 25 years of broadly representative country CHRM data at roughly five-year time intervals. This has enabled CHRM to address core questions about developments in HRM over time, thus providing insights into the convergence/divergence debate. The convergence thesis is driven largely by part of the economic and general management literature which argues that the forces of globalisation will create management practices that become increasingly alike as technology and communications spread around the world (for early voices of this see e.g. Kerr, Dunlop, Harbison, andMyers (1960 (Orig. 1953)); Galbraith (1967)). This is reflected in the CHRM literature, where 'dominance effects' have been identified (Smith & Meiksins, 1995) whereby what are seen as the most successful models have exemplary power on other communities, setting standards for the rest of the world. Another strand of the convergence debate reflects the international business literature (Rugman & Verbeke, 2004) and argues that regions are more important. The influence of the European Union on HRM has been profound, spreading legislation on a range of employment contract issue. Rather than global convergence, this strand of comparative HRM research suggests that over time we will witness regional convergence, with the regions becoming increasingly differentiated. Others within CHRM suggest that we should not expect to find any convergence since both the cultural literature (Hofstede, 1980;House, Hanges, Javidan, Dorfman, & Gupta, 2004a, 2004bSchwartz, 2008) and the institutional literature (e.g. Drori, Meyer, & Hwang, 2006;Krücken & Drori, 2009) argue that countries are distinctive, change only slowly and that any change over time is path-dependent.
Given the centrality of time in the CHRM stream, clarity has been sought over what convergence actually means (Mayrhofer, Müller-Camen, Ledolter, Strunk, & Erten, 2002). Directional convergence points towards similar trends. Of course, even if the same trend is apparent in many countries, different starting points and different rates of change may mean that countries remain as far apart as ever, or even diverge further. For final convergence, they would have to become more alike in the way that they manage their HRM over time. Overall, the empirical evidence in the CHRM stream shows little evidence of final convergence. Whether based on cultural or institutional analyses, or on a combination of the two, the research has tended to reveal continuing differentiation. Whilst we see instances of directional convergence/similarity (e.g. Mayrhofer et al., 2011), no convincing case for final convergence has been made.
In short, time is still an understudied dimension within both streams, but especially in the HRM in MNEs stream. Despite a shared interest in the standardisation/differentiation debate, this has made it difficult for the two streams to 'talk to' each other about the dynamics behind standardisation and convergence on the one hand, and differentiation and divergence on the other. There is much to be learned, particularly given their broadly distinct findings and the two streams have much to offer one another, as we will try to show in the final section presenting suggestions for a future research agenda.
A greater focus on aspects of time will almost inevitably raise issues of process and the mechanisms that are at play behind patterns, and changes to the patterns, of standardisation and differentiationa subject to which we now turn.

The role of process
The role of process is implicit in the two streams, and yet it has only recently started to attract substantial theoretical and empirical attention. For explaining the mechanisms at work when analysing convergence and divergence in HRM, good theories are indispensable. Sound theoretical foundations allow us to address the 'how and why' behind observed empirical developments and can guide future research to uncover the importance and scope of suggested theoretical relationships.
It has been argued that a lack of focus on the processes through which HRM practices are created and developed is a shortcoming in the general HRM literature (Lengnick-Hall, Lengnick-Hall, Andrade, & Drake, 2009;Paauwe, 2004). The 'HRM process' school of thought in particular is gradually infiltrating the HRM in MNEs stream, conceiving of HRM systems as comprising different process features such as, e.g. validity, visibility, consensus, justice, that send signals that are interpreted by individuals, which in turn, go towards creating strong/weak organisational climates (Bowen & Ostroff, 2004). This approach to process, which is positioned as complementing the HRM 'content' school (Katou, Budhwar, & Patel, 2014), emphasises the role of actors (Chung, Sparrow, & Bozkurt, 2014) and highlights the role of psychological processes through which employees attach meaning to HRM (Sanders, Shipton, & Gomes, 2014).
Significantly, the process approach emphasises the need for HRM research that crosses levels of analysis, and incorporates the perspectives of other stakeholders than owners and shareholders (Beer, Boselie, & Brewster, 2015;Nishii, Lepak, & Schneider, 2008;Sanders & Yang, in press;Sanders, Dorenbosch, & de Reuver, 2008;. There have also been calls for actor-centric research into who the key actors are and what their roles are in HRM development processes (Rupidara & McGraw, 2011;Welch & Welch, 2012). It is suggested that a focus on how HRM practices, practitioners and praxis interact will advance our understanding of how people-related decisions within organisations are made, implemented and enacted .
Taken together, the dominant message therefore seems to be that in order to better understand HRM and its influence at the firm level, we should search for explanations in processes located at and between the lower levels. Interestingly, arguments supporting this form of micro-foundations approach to HRM have not been accompanied with equally strong arguments to search for explanations at the higher levels, leading some to adopt a more critical stance towards the further 'psychologisation' of HRM (Godard, 2014). The general shift in the HRM in MNEs stream towards incorporating increasingly subjective measures of HRM practices and micro-level (individual) theories, when compared to CHRM's greater emphasis on capturing objective measures of HRM and macro-level (contextual) theories, represents another potential challenge for the crossing of the two streams.
In the HRM in MNEs stream, although 'standardisation' and 'adaptation' imply that certain processes are at play, empirical studies have been slow to analyse the processes involved (Rupidara & McGraw, 2011;Smale, 2008). This is particularly noteworthy in the branch of this stream that looks at the transfer of HRM practices, since the conceptualisation of standardisation as being the result of transfertypically from headquarters to subsidiarysuggests that our understanding of patterns of HRM practices within MNEs requires an understanding of what happens, or does not happen, during the transfer process (Gamble & Huang, 2009;Chang, Smale, & Tsang, 2013).
Possible explanations for this relative lack of focus on issues of process in the HRM in MNEs stream are the disproportionately high number of quantitative studies (for a recent review see Welch and Björkman (2014)) that are less well suited to investigating process, and the dominance of contingency models. A significant body of HRM in MNEs research has been built around contingency-type integrative frameworks that relate a number of key MNE-external, MNE-internal and HRM-specific factors to MNE outcomes (e.g. DeCieri & Dowling, 2012;Schuler, Dowling, & Cieri, 1993;Taylor, Beechler, & Napier, 1996). Whilst these have been useful in highlighting a multitude of factors arising from different sources that influence in complex ways the transfer of HRM, they provide a somewhat static account. They do not adequately acknowledge the more dynamic nature of transfer and adaptation processes that help us to understand the relationships between these different sets of factors, or the key actors involved.
Most of the scholarly work that has been conducted on the mechanisms of HRM transfer within MNEs has been limited to case studies. Collectively, these studies have shown that MNEs often utilise a range of direct and indirect mechanisms in combination and at varying strengths (Sippola & Smale, 2007;Smale et al., 2013), even across subsidiaries within the same MNE (Edwards, Ferner, & Sisson, 1996). They have also shown that expatriates, typically operating outside of HRM function, are pivotal in determining the degree of standardisation and adaptation due to their varying competence in HRM and their boundary-spanning roles that involve the two-way communication of headquarters' expectations and hostsubsidiary needs Gamble, 2003Gamble, , 2010Hetrick, 2002). A major conclusion from these studies, which notably tend to adopt the subsidiary perspective, is that processes of HRM standardisation and adaptation, and HRM practice enactment more broadly, are fraught with difficulty. More specifically, drawing on power relations and the micro-political perspective, HRM centralisation and formalisation are highly contested and political processes, characterised by the use of power resources by both MNE headquarters and subsidiary management (Almond et al., 2005;Ferner, 2000;Ferner et al., 2005). Subsidiary actors, by leveraging their role as 'interpreters' of the local HRM environment, are seen as influential, capable of shaping the patterns of HRM standardisation and adaptation that unfold . In short, it is suggested that the micro-political perspective has arisen out of a need to incorporate explanations based on agentic behavior that have not been adequately accounted for in institutional theorising about HRM configuration processes within MNEs (Rupidara & McGraw, 2011).
In comparative HRM, the debate draws on a number of theoretical sources, depending on whether convergence or divergence is the underlying assumption. Much of the convergence argument comes from two sources. The first is the rational actor model of the firm (Coleman, 1990) that assumes that firms act under conditions of bounded rationality and implement practices that contribute to economic goals. Global capitalism and some of its central tenets such as rationality or cost effectiveness support the emergence of reasonably similar organisational structures and processes. The second is the world-polity approach (Meyer, 2000;Meyer & Rowan, 1977) which also favours convergence. It argues that, especially since 1945, Western cultural patterns and institutions dominate global developments so that core individual and collective actors, including organisations and nation states, are subject to isomorphic pressures to follow the Western model of rationalization (Drori et al., 2006). Although practices traveling across the globe are edited and customized to specific contextual settings (see, e.g., the work of Boxenbaum, 2006;Czarniawska & Sevón, 2005;Sahlin-Andersson & Engwall, 2002), this reduces the importance of nation states (Ohmae, 1995) and leads to a world-system (Frank & Gills, 1993) or a world-society (Krücken & Drori, 2009).
The alternative theories, leaning towards divergence or, most often, the maintenance of differences, are cultural and institutional arguments. Culture theorists (e.g. Hofstede, 1980;House et al., 2004aHouse et al., , 2004b point out that cultures reflect deep-rooted values and mindsets which are different across the globe and changeable only in the long run. Therefore, it is reasonable to expect sustained differences at the individual and organisational level. Institutionalist arguments head in the same direction. Arguments have been made that differences in such factors as legal systems (Botero, Djankov, La Porta, Lopez-de-Salanes, & Shleifer, 2004;La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 1999, politics (Roe, 2003), or political systems (Pagano & Volpin, 2005) will impact the power of owners of businesses, creating differences between nations. Synthesizing these ideas and adding other institutional and economic differences, the comparative capitalisms literature (e.g., Amable, 2003;Hall & Soskice, 2001;Jackson & Deeg, 2008;Whitley, 1999) points towards differences in institutional arrangements in the economic, social and legal realm at the national level (Hollingsworth & Boyer, 1997) and their relative inertia. Hence, globalisation clearly has its limits (Guillén, 2001) and institutional change is slow (Djelic & Quack, 2003). HRM, it has been argued, is one of the areas of management which most likely reflects local circumstances (Rosenzweig & Nohria, 1994) and, via the 'societal effect' (Maurice, Sellier, & Silvestre, 1986), a country's unique institutional arrangements. This raises significant doubts concerning universalistic best-practice HRM (Sparrow & Hiltrop, 1997).
Some of the main criticisms of institutional theory and particularly the comparative capitalisms approaches that have been used in the CHRM stream are their propensity to provide rather static accounts, and their downplaying of the role of agency in responding to institutional pressures. The comparative capitalisms literature has tended to imply that economic categories are broadly static although the reality is that in any society there are institutional disjunctions and a lack of complementarityand societies change (Thelen, 2014). Furthermore, especially in societies where institutions are particularly new or weak, MNEs may have considerable power to influence at least the institutions controlled by the state in the direction that they wish. They may, therefore, have room to apply standardised policies despite local pressures. Whilst the evidence on the dominance effect is somewhat inconclusive, few studies (if any) have empirically examined the processes underlying this effect: including how the spread of best practices takes place, who are the key actors, and what is the role of inter-organisational and inter-personal networks.
On balance, despite the greater emphasis on time in the CHRM stream, one could argue that more empirical attention has been paid to issues of process in the standardisation-differentiation debate within the HRM in MNEs stream. Nevertheless, it is argued that "the literature has presented only limited accounts of what actually happens in the process of constructing and negotiating HR systems in the context of so many potentially conflicting institutional rationales" (Rupidara & McGraw, 2011: 175). Given the dearth of empirical research on the processes involved in the national and global context influencing MNEs' HRM practices, on the one hand, and MNEs influencing national and global systems of HRM on the other, there is clearly plenty of scope for the two streams to cross in a way that could help to shed light on the mechanisms of standardisation/convergence and differentiation/divergence.

The role of context
How the two streams have incorporated context, in both theory and research design, and the kinds of conclusions they draw about the importance of context, also reveal some clues as to why the streams have not tended to cross in the past. However, at least in terms of the diversity of contexts, there are welcome signs of change (Budhwar & Debrah, 2001;Budhwar & Mellahi, 2006;Dessler & Tan, 2009;Dowling & Donnelly, 2013;Elvira & Davila, 2005;Kamoche, Debrah, Horwitz, & Muuka, 2004). Most international HRM research has been conducted in the WEIRDwestern, educated, industrialized, rich, democraticcountries (Henrich, Heine, & Norenzayan, 2010). This applies equally to research in both streams. What we know from extant research on HRM mainly covers a minority of the world's countries, with limited data from countries outside of the Global North in particular.
The role of context within the HRM in MNEs stream is most frequently framed as a constraint to HRM standardisation. Indeed, there appears to be an implicit assumption in much of this research that a key objective for MNEs is to standardise their HRM policies and practices and that they are only prevented from doing so by local contextual constraints that require them to compromise (Edwards & Kuruvilla, 2005). The debate is sometimes couched in terms of affecting the degree of headquarters centralization versus subsidiary autonomy or 'discretion' over HRM decisions (Ferner et al., 2011) and sometimes in terms of the external and internal variables that explain why MNEs may want to standardise/transfer HRM ('desirability'), or why this is not possible ('feasibility') (Dickmann, 2003). The reverse case, i.e. how the local context for HRM is influenced by MNEs and MNE subsidiaries, is very seldom researched, if at all.
Historically, context within this stream of international HRM is reflected in an over-representation of studies looking at Western, developed-country MNEs transferring HRM practices to non-Western, developing, host-country settings. Studies on developing-country MNEs are under-represented, although they have been increasing (Chang, Mellahi, & Wilkinson, 2009;Sidani & Al Ariss, 2014). Furthering our understanding about the diffusion of HRM practices from the perspective of emerging country MNEs "without assuming the prevailing Western ethnocentric orthodoxy" is viewed as being especially important in this regard (Thite, Wilkinson, & Shah, 2012: 251). From the subsidiary perspective, the contextual antecedents of reverse HRM transfer have also attracted growing attention, looking at conditions under which foreign subsidiaries, albeit quite often from institutionally strong host-country settings, are capable of transferring HRM practices to headquarters Edwards & Tempel, 2010;Thory, 2008), as well as laterally to other MNE subsidiaries (Edwards, Sanchez-Mangas, Bélanger, & McDonnell, 2015).
In a review of the HRM in MNEs stream, Welch and Björkman (2014) found only two articles dedicated to MNE context (macro, exogenous factors), both of them non-empirical. They conclude that "there has been relatively little IHRM scholarly interest in investigating global and home-country factors influencing HRM policies and practices in MNCs […] and even less in examining how MNCs through their HRM practices and systems influence the societies in which they operate. It is perhaps this neglect that has given rise to the perception that IHRM is disconnected from the wider IB community" (Welch & Björkman, 2014: 12-13).
Relatively little of the HRM in MNEs research looks at contextual effects on a regional level, either externally in terms of regional exogenous effects or internally in terms of regional MNE structures (e.g. regional HQ). This is particularly surprising given the regional nature of international business (Rugman & Oh, 2013;Rugman & Verbeke, 2004) and the tendency for larger MNEs to establish regional headquarters and to regionalise their operations (Nguyen, 2014;Sparrow, Brewster, & Harris, 2004). Neither does it tend to acknowledge within-country differences that might explain differences in HRM between subsidiaries located in the same country (e.g. in studies of Chinese subsidiaries).
Largely due to convenience sampling and difficulties involved in attaining a sufficient number of responses, there are two main variations in the studies that there have been about how geographical context is incorporated into research designs in the HRM in MNEs stream: (i) studies on foreign subsidiaries in one host country belonging to MNEs from one or more countries of origin (e.g. Bjorkman & Budhwar, 2007;Farley, Hoenig, & Yang, 2004;Rosenzweig & Nohria, 1994), and (ii) studies on subsidiaries in one or more host countries belonging to MNEs from one country of origin (e.g. Dickmann, 2003;Ferner, Quintanilla, & Varul, 2001). Designs that combine both MNEs from multiple countries of origin and subsidiaries from multiple host countries are rare. Combined with a paucity of longitudinal designs, the absence of contextual diversity at MNE and subsidiary levels has meant that only a few studies within the HRM in MNEs stream have been able to cross with the CHRM stream and contribute meaningfully to the convergence-divergence debate.
Again, there have been some notable exceptions. Gunnigle, Murphy, Cleveland, Heraty, and Morley (2002) compare how subsidiaries of European and US MNEs in six European countries (United Kingdom, Ireland, France, Denmark, Germany and Sweden) adapt their HRM practices to account for the host environment. They find that whilst both European and US MNEs locally adapt their subsidiary HRM practices due to host-country institutional constraints, the degree of adaptation within the subsidiaries of US MNEs is lower. Comparing MNEs and their foreign subsidiaries in three different countries (USA, Germany and Japan), Pudelko and Harzing (2007) present evidence that they suggest indicates convergence towards a worldwide best-practices model. Smith and Meiksins (1995) referred to this as the 'dominance model', arguing that the global hegemony of the USA means that HRM in that country, and in MNEs headquartered in that country, are seen as exemplary. Björkman, Fey, and Park's (2007) comparative study of the institutional factors influencing HRM practices in US, Japanese and European MNE subsidiaries located in Russia, Finland and the USA, allowed them to combine both home-country and MNE-specific explanations with hostcountry and subsidiary-specific explanations for the subsidiaries' selection of HRM practices. Given the general lack of these kinds of research designs and the promise that they hold, there have been calls for more multilevel studies that include the headquarters of MNEs from different countries of origin and their foreign subsidiaries located in various host countries (McDonnell, Lavelle, & Gunnigle, 2014).
The comparative HRM stream, by contrast and almost by definition, is strongly linked to the concept of context. The general conclusion is that context still matters. There are cases of convergence, but these are outweighed by the continuing distinctions between countries, market economies and regions, with some cases pointing towards hybridisation. Patterns and models of HRM vary between countries. There have been many such studies and the following are just a few examples. Some studies have emphasised the difference between a particular country (e.g. Germany) and the pattern of HRM typically represented in the international journals and by consultancies (Festing, 2012;Giardini, Kabst, & Müller-Camen, 2005;Müller, 1999aMüller, , 1999b. Others identify differences between countries such as the UK and Japan (see e.g. Lam, 1994;Storey, Okasaki-Ward, Edwards, Gow, & Sisson, 1991), the UK and China (Easterby- Smith, Malina, & Yuan, 1995), China and the Netherlands (Verburg, Drenth, Koopman, Muijen, & Wang, 1999), Britain and India (Budhwar & Khatri, 2001), and the People's Republic of China and Taiwan (Warner & Zhu, 2002). And between groups of countries (Budhwar, 2000;Cheng & Brown, 1998;Tregaskis & Brewster, 2006;Wood, Brewster, & Brookes, 2014); and between the countries in Europe transitioning from communism to capitalism (Cooke, Wood, Psychogios, & Szamosi, 2011;Morley, Heraty, & Michailova, 2009;Poor, Karoliny, Alas, & Vatchkova, 2011;Sahadev & Demirbag, 2012). A smaller number of empirical studies takes a broad view of HRM and includes a wide range of countries (Clark & Pugh, 1999Geringer, Frayne, & Milliman, 2002;Teagarden et al., 1995).
Data from the on-going Cranet surveys initially provided data on European countries and then expanded to include a wider range (Brewster, Mayrhofer, & Morley, 2000, 2004Parry, Stavrou, & Lazarova, 2013). Apart from showing clear distinctions between countries on HRM, studies from this project, and from other sources, have identified clear differences between countries: in responses to the ageing workforce in the UK and Germany (Müller-Camen, Croucher, Flynn & Schröder, 2011); in the professionalism of the HRM function between the UK and other European Countries (Farndale, 2005), in compensation practices in ten countries across three continents (Lowe, Milliman, De Cieri, & Dowling, 2002), related differences in financial participation (Pendeleton, Poutsma, Ligthart, & Brewster, 2013), differences in training (Goergen, Brewster, Wood, & Wilkinson, 2012), management development systems across Europe (Mabey & Ramirez, 2004), employee turnover in the USA and nine European countries (Croucher, Wood, Brewster, & Brookes, 2012), and downsizing in 16 different countries (Goergen, Brewster, & Wood, 2013).
Theorising about the role of context within the CHRM stream has also been somewhat eclectic, but has gravitated towards explanations based on the institutional literatureboth the traditional theory and its variant new institutional theoryand the comparative capitalisms literature (Delbridge, Hauptmeier, & Sengupta, 2011;Wood et al., 2014). Within both streams, questions have been raised about explanations based on cultural theory. 'Cultural' arguments for HRM practice differentiation have been criticised for over-essentialising culture in a way that views it as static rather than "a shifting and changeable repertoire with diverse strands" (Gamble, 2003: 384), and for being unclear regarding the specific origins of global and local effects (Edwards & Kuruvilla, 2005). Proposed categorisations of culture overestimate the measures they use (Weller & Gerhart, 2012) and are inconsistent (Avloniti & Filippaios, 2014). Further, since any culture contains a more or less normal distribution of people who fit fully into the relevant stereotype and those who do not, organisations can select in (or out) those who fit the profile they would prefer rather than those who fit the local culture (Vaiman & Brewster, 2015).
Others are calling for theoretical pluralism within single studies rather than across them. Farndale and Paauwe (2007) have considered some of the macro-and meso-level factors in HRM in MNEs. Gamble, based on his analysis of processes leading to HRM practice hybridisation (Gamble, 2003: 705), concludes that "approaches based upon culturalist, national business systems, industry sector, international division of labour, and agency perspectives are shown to be inadequate, individually, to account for the complex patterns of transfer, local adoption, and adaptation" and calls instead for "conceptual bricolage" and multilevel analysis. Similarly, Edwards et al. (2015) stress the need for multilevel analyses since their study revealed that national, corporate and functional HRM contexts all matter.
To summarise the discussion so far, differences between these two streams of international HRM in terms of their approach towards context and the kinds of conclusions that they draw, in addition to time and process, could also be seen as obstacles that have prevented the crossing of the streams. The streams obviously have different but not incompatible objectives and perspectives. It seems that the HRM in MNEs stream is focused on standardisation and practice transfer whilst the CHRM stream accentuates the persistent differences between countries in their HRM, partly in response to the 'globalisation' claims implicit in much of the HRM in MNEs literature. Basic underlying divisions about the importance of context might thus be seen as making HRM in MNEs and CHRM uncomfortable bedfellows and perhaps provide a reason for the streams not to cross at all.
However, we believe that the two streams also share some important similarities, certainly in terms of content, and, building on their relative strengths and interests, have a lot to offer one another. For instance, whilst their occupation with different levels of analysis -CHRM with the macro level, and HRM in the MNEs with organisations, subsidiary and increasingly the individual levelhave tended to keep the streams apart, calls for more multilevel research and research that incorporates the MNE context would seem to suggest a clear need for crossing the streams. Moreover, since neither of the two streams has seriously begun to address empirically the influence of MNEs on national systems of HRM and the societies in which they are embedded, crossing the streams would shed light on this complex and multi-faceted question. Next, we attempt to spell out what these complementarities could be in the form of a research agenda.

Crossing the streams: Towards an enriched research agenda for the convergence-divergence discourse
We take the title of our article from the Ghostbusters films, where one of the heroes often warns the others about the dangers of 'crossing the streams' of their ghost-catching guns. As in that case, the contents of the streams are a little unclear, why they should work at all is not obvious and mixing them may have unforeseen consequences: but if it helps to exorcise a few ghosts in the convergence/divergence debate, then we should try it. Our proposed agenda builds on the two stream's traditions and developments across the issues of time, process and context and suggests ways in which they can mutually benefit each other. This is done by focusing on the processes through which various actors and activities within MNEs and within national (institutional) systems of HRM interact over time, and the mutual, bi-directional, effects these have on both HRM practices within MNEs and the broader societal context for HRM. More specifically, with regards to the issue of time we emphasise the more empirical elements of a future research agenda; with regards to issues of process we emphasise the role of better theorisation; and with regards to context we emphasise the kinds of products that could emerge out of a deeper cross-fertilisation across the two streams.

Time
The first set of suggestions speaks to the issue of time and belongs to the usual repertoire of a future research agenda: do what you have done up until now, but do it more and better. Indeed, the question of time in HRM convergence/divergence should be more central to empirical research efforts within both streamswithin CHRM due to convergence/divergence being at its core; and within HRM in MNEs due to its conspicuous absence (Hippler, Brewster, & Haslberger, 2015). The contributions time can make to understanding changes in MNE corporate and subsidiary HRM practices can, in turn, help to shed light on the convergence/divergence debate.
This almost inevitably means more longitudinal research. Besides serving as a sounder basis for inferences about the direction of causalityan issue that afflicts a substantial proportion of the HRM in MNEs literatureit also allows studies to capture how change occurs, the speed of change and whether the rate of change is constant. This is an easy call, but in practice, a long, winding and steep road. For instance, this calls for the more targeted use of and improvements to existing data. Beyond the continued use of HRM-specific databases such as Cranet, we also see a lot of potential in using other available data for the purpose of empirically grasping the issue of time in convergence/divergence research. One example may suffice. In the European Working Condition Survey (EWCS; see https://eurofound.europa.eu/european-working-conditions-surveys-ewcs), two waves of data from 2005 and 2010 are available with a new round coming up in 2015. In this data, which basically covers EU-member states plus selected other countries, a number of questions are designed to function as trend questions over the years, with obvious relevance for country-specific HRM policies and practices (questions about individual's employment contracts, for instance). This data could be examined from an HRM point of view to broaden our insight about developments at the country level, particularly related to the micro-aspects of HRM.
Most, if not all, current theoretical and empirical work on convergence/divergence adopts a naturalistic, simplifying view of time as a divisible construct where each part has the same significance, choosing, for example, certain time spans and relating the findings of one to the other. Whilst in practical terms this makes a lot of sense, a closer look reveals that this assumption is questionable, to say the least. When investigating convergence/divergence, time spans should also be related to the context they denote. Three years in a comparatively stable period, e.g. 1982-1985, do not equal three years in a time of dramatic global change, e.g. 2007-2010. Hence, assigning qualitative elements to time spans, taking into account sequential effects, e.g. three years of rapid change followed by years of relative stasis or vice versa, could bring more insight into the dynamics of convergence or divergence.
Given the centrality of time in achieving a nuanced understanding of HRM convergence/divergence we also advocate the use of more advanced research designs and the pursuit of better data quality. In this respect, there is scope to complement existing 'trend' studies with empirical studies that rely on matched samples. Due to the additional challenges presented by participant recruitment, tracking and retention, these studies may have to make certain concessions in terms of smaller samples, fewer countries and shorter time spans. However, the benefits in terms of a deeper understanding of how, when and why convergence/divergence occurs may outweigh the costs. More ambitious longitudinal research designs would also allow for the application of more advanced modelling and data analysis, such as change score models, autoregressive and cross-lagged panel models, and latent growth curve models (Bednall, 2014). Such models would more readily allow for investigations into the respective roles of time-varying (expected to change over the time span of the study) and time-invariant factors (not expected to change much, if at all) and the longevity of their effects on convergence/divergence.

Process
In terms of process, we emphasise the need to strengthen the use of existing theoretical approaches for analysing the dynamics behind convergence/divergence. For this, both streams of research should open themselves up for more in-depth discussions about their respective theoretical discourses. Again, one example, neoinstitutionalist thinking, can illustrate what is meant by this. Whilst this school of thought is already established in both streams, we argue that the more refined developments within this theoretical discourseoften developing into theoretical streams of their own under the broader umbrella of the overall theoryhave neither fully found their way into the HRM discourse, nor been used to explain convergence and divergence nor, for that matter, standardisation and differentiation. For example, the neoinstitutionalist discourse has produced a number of theoretical insights which point more towards potential mechanisms of standardisation/convergence such as interlocks (Shipilov, Greve, & Rowley, 2010), proximity (Knoben & Oerlemans, 2006), exemplars (Jones & Massa, 2013), or carriers (Scott, 2003); and towards mechanisms of differentiation/divergence, in particular translation and diffusion (Czarniawska & Sevón, 2005), and intra-organisational dynamics, visibility/status, and core-periphery location. For example, the literature on different types of proximity allows a more differentiated understanding of the role of various types of proximity, geographical, technological and organisational, for analysing the HRM developments over time at both the national level and at the level of MNEs. In a similar manner, carriers such as symbolic systems, relational systems, routines, and artifacts help institutional elements, e.g. HRM practices, to transfer. They support the spread and affect the transmitted elements (Scott, 2003). Of course, not all of these ideas can immediately be translated into empirical research settings within CHRM or HRM in MNEs. Yet, we do suggest that a more systematic incorporation of these theoretical insights can open up new avenues for making sense of the processes underlying HRM convergence and divergence and standardisation/differentiation (Festing & Eidems, 2011).
By extensionand in line with a renewed interest in actor-centric viewpoints in literatures about strategy (Eisenhardt, Furr, & Bingham, 2010) and institutional thinking (Powell & Colyvas, 2010)we also call for research that incorporates the roles of key actors in these processes. More detailed accounts of the roles and activities of key actors at various levels of the MNE will help to shed light on the dynamic interactions between MNE actors and their context in shaping HRM practices (Rupidara & McGraw, 2011). This will also serve to complement theoretical insights from neoinstitutionalism with those based around agency, power and politics in revealing the complex dynamics of the contested institutional terrain (Ferner, 2009). Like the suggested 'HRM-as-Practice' approach that builds upon the strategy-as-practice literature we advocate a broader conceptualisation of 'practitioners' to include key individual and collective actors both internal and external to the MNE . This is an area where the Comparative HRM literature would be enhanced by the addition of an actor perspective. In addition to the 'usual suspects', external key actors could include parties such as management consultancies, business schools, HRM professional networks and associations, accreditation bodies and certain pressure groups. Empirical studies could shed light on, for example, how different actors are involved in the kinds of institutional processes of convergence/divergence. In short, if our understanding of HRM convergence and divergence processes are to progress beyond comparing similarities and differences between practices we need more theoretically-grounded research on issues of process.

Context
Our suggestions concerning context emerge even more directly from crossing the two streams and they are, in a sense, cross-fertilisation products. Context is the topic of several discourses across a number of disciplines, reaching far beyond management or organisation studies. Against this backdrop, it seems especially fruitful to exploit these discourses in the narrower realms of international HRM. Cross-fertilisation can principally be realised through empirical examinations of the processes taking place at the intersections between MNEs and the national/global systems of HRM in which they are embedded. As previously noted, it is rare that studies in HRM in MNEs place equal emphasis on MNE headquarters/home-country issues and subsidiary/host-country issues (Ferner, 2009;McDonnell et al., 2014). The field would benefit from more parallel examinations of home-and host-country effects (see Almond, 2011 for a critical review of country-of-origin effects). The added challenge of data access notwithstanding, such studies are much better placed to draw conclusions about HRM convergence/divergence in ways that can combine and leverage the explanations offered by each of the two streams.
Research within HRM in MNEs has exhibited a preoccupation with studying the effects of external contextual determinants on MNE headquarters' and subsidiary HRM practices. The reverse case, how the HRM practices of MNEsfor example, through the deployment of various power and political resourcesinfluence the institutional and social contexts in which they operate, needs to be the centre of more research attention. MNEs are sometimes regarded as a crucial impulse and motor of development, in particular in the so-called less developed economies. However, to the best of our knowledge, there is no analysis of whether different degrees of MNE presence in various countries have systematic effects on developments within these countries, or whether these effects are similar in other countries with a comparable situation. Specifically, we know little about the effect of foreign MNEs on the institutional environment governing HRM in a country and, of particular importance with regard to the convergence/divergence debate, the long-term effect on HRM in these countries, e.g. the emergence of a standardised national model. Combining the insights from research in HRM in MNEs and CHRM, could deepen our understanding of convergence/divergence by taking into account the mutual effects between the national and organisational levels. Given the historically greater importance attached to the role of MNEs in this regard, Welch and Björkman (2014) suggest looking to the international business literature for examples of how this could be done.
Lastly, context is a highly diverse concept used, among others, by linguists, anthropologists, and political scientists. Especially in countries such as Nepal, Pakistan, Angola, or Libya with 'difficult' or 'extraordinary' contextual situations in terms of geography, language diversity, political stability, intra-national relationships between different groups or tribes, we know very little about how HRM unfolds and develops in such settings. For instance, we know next to nothing about whether MNEs choose similar HRM strategies to cope with the situation in such countries and whether these strategies change or stay stable over time. Our understanding of what happens with regard to convergence/divergence would deepen were we to apply a more differentiated view of what context means and look at areas currently 'off the map' for the most part.
Collectively, we recognise that pursuing the kind of agenda we have set out above also implies a call, also made by others (Edwards & Kuruvilla, 2005), for renewed efforts to establish global consortia of scholars dedicated specifically to pursuing the convergence/divergence issue in HRM. Such consortia could both combine existing databases and set out to establish new efforts. Scholars operating within the CHRM stream have been more active in this respect than their counterparts within the HRM in MNEs stream, but we need both.

Limitations
Such a review inevitably has limitations. We have restricted our focus to HRM, and largely ignored the closely related fields of talent management and industrial relations. A recent Special Issue of the Journal of World Business (2014, 49 (2) tackled talent management in various national/international contexts. Our review and agenda have also been limited by our use of the time, process, and context lenses. In these increasingly researched topics it has inevitably been impossible to cite all the relevant authors or studies and to deal with all relevant topics (the other contributions to this special issue are useful here). The time, process and context framework helps us to structure the topics that we have covered but, with space constraints, have also restricted the kinds of issues we have raised.

Implications for practitioners
Our focus here has been on the research domain, but we also believe that our analysis has some implications for practitioners. The HRM specialists' main interests will be in the HRM in MNEs area. Here, a greater understanding of when and why global policies may not only be difficult to implement but may be inappropriate in many contexts would be of great use, and help expatriate managers to make more informed decisions about complex people issues. Local managers who understand the variations within the meaning and import of HRM across national boundaries may be better able to ensure that relevant policies are introduced in their area. This, in turn, will benefit employees. Nationally, the more that the IHRM debates in both streams are brought into the policy arena, the more likely it is that policy-makers will be able to develop approaches that benefit their communities.