European Human Resource Management: Researching Developments over Time **

This paper uses insights and data gained from 15 years of involvement in a research network studying HRM developments in Europe to explore the notion of “European HRM“ and the meanings of convergence and divergence in HRM Using data from the last decade of the Cranet research, the paper puts forward a more nuanced view of the notions of convergence and divergence, finding evidence of directional convergence, but little evidence of final convergence: whilst there are trends which point in similar directions, national differences remain a key factor in HRM


Introduction
Human Resource Management as a concept was formalised in the USA in the late 1970s and early 1980s, encapsulated in two famous textbooks (Beer et al. 1985;Fombrun et al. 1984). These approaches varied but both differentiated HRM from personnel management and argued that the former involved more integration of personnel policies across functions and with the corporate strategy (with HR being the downstream function); a greater role for line managers; a shift from collective to individual relationships; and an accent on enhancing company performance.
The notion of "European Human Resource Management" was developed largely as a counter to the hegemony of US conceptions of human resource management (HRM). This, in part, reflected developments in the arguments about how we should conceive of the notion of HRM (Kamoche 1996). It was argued (Brewster 1994;Sparrow/Hiltrop 1994) that US assumptions about the nature of HRM were inappropriate in this (and probably other) continents and that Europe needed models of its own. These notions were behind the establishment, towards the end of the 1980s, of a research network based on one university per country, dedicated to identifying trends in HRM in Europe. That network grew from the original five countries to twentyseven in Europe and some dozen others spread across the world. Some fifteen years after the start of the project, and at the point of the publication of the fourth edited book based on the network's outputs (Brewster et al. 2004), this seems to be a good time to review what we have learned.
In particular, the long-term nature of the project allows us to identify trends in the management of human resources in Europe (see Gooderham/Brewster 2003 for a first attempt at doing so using this data). Are the European countries moving towards one another in the way that they manage HRM? If so, are they moving towards or away from a model similar to that operating in the USA? Or are they remaining separate and different?
In presenting these findings, this paper is, therefore, ambitious in scope. First, it conceptualises the notion of "European HRM", setting it in the context of theories of international HRM and convergence and divergence in comparative HRM; it is argued that there are inevitably elements of universality and of national difference that have to be encompassed by such theories and that we need more nuanced approaches to the ideas of convergence and divergence. After briefly exploring the methodological and practical issues of researching large-scale developments in European HRM over an extended period, the paper presents empirical evidence from the research of Cranet, the Cranfield Network of Human Resource Management, as a contribution to the European convergence/divergence debate which, finally, enables us to draw some conclusions about whether there is, indeed, evidence of convergence in European HRM.
also distinguish France as an environment that, while not having a market mentality, is nevertheless deficient in communitarian infrastructures. Others distinguish between, on the one hand, countries such as the UK, Ireland and the Nordic countries, in which the state has a limited role in industrial relations, and the Roman-Germanic countries, such as France, Spain, Germany, Italy, Belgium, Greece and the Netherlands, in which the state functions as an actor with a central role in industrial relations (Due et al. 1991: 90). Arguments have also been made for a "northern European" approach to HRM based around those countries where English is widely spoken and trade unions are stronger (Brewster/Larsen 2000).
One analysis of HRM practices found "three clusters: a Latin cluster [which includes Spain, Italy, France]; a central European cluster . . . and a Nordic cluster' (Filella 1991: 14). The Latin style of HRM is characterized, inter alia, by efforts to modernize HRM, a greater reliance on an oral culture and the presence of subtle 'political' structures which unconsciously nurture docile, dependent attitudes to authority. The Nordic approach to HRM would include the substantial, visible authority of the HR department, extensive written strategies, a widespread collective orientation to management, and extensive consultation. The continental central European model would involve lower authority for HR departments, extensive line management involvement in HR issues and legal support for collaboration with trade unions. Whether there is an 'offshore central European' model is open to question. Filella (1991) argues that the regional groupings may correspond to stages of socio-economic development.
Examining flexible (contingent) work practices, Brewster and Tregaskis (2001) found slightly different groupings in the manufacturing and service sectors. Spain tended to be a category on its own in both cases and, in manufacturing for example, Germany, the Netherlands, Norway and Switzerland were in the "high inclusive" group; France and Ireland in the moderately reactive group; in services the UK, Sweden, Denmark and Belgium in the moderately reactive group. Gooderham and Brewster (2003) found four categories based on a matrix based around communitarianism and autonomy.

Dynamic views: Convergence or divergence?
Whilst valuable in pointing us away from the universalistic prescriptions on HRM there are two major problems with these European or within-Europe approaches. First, whilst they emphasise the importance of culture and institutions, they underemphasise the level at which such differences are most cogent, the national level. Second, they are static: they leave little room for the change that we see all around us.
As far as the level of national differences is concerned, the evidence is widespread. A number of studies show the differences between various aspects of HRM in European countries (e.g. , Pieper 1990;Vickerstaff 1992;Brunstein 1995;Brewster et al. 2004).
However, the evidence concerning convergence or divergence is more equivocal: are the differences between nations being reduced or not? 1 There are at least two variations of the convergence thesis. The first is the market-driven approach, which tends towards arguing that the rest of the world will become increasingly similar to the United States of America, the most powerful market in the world and therefore the exemplar, in the way that organisations are managed, including how they manage human resources. The second explanator is institutional which, although there is a strand of this theory which argues for world-wide convergence, also includes a strand arguing that the institutional power of the European Union and its approaches to employment practices will lead to a convergence towards a specifically European model. The divergence thesis often uses institutional and/or cultural arguments.

Convergence: Market-led
The first variation of the convergence thesis argues that policies of market deregulation and state decontrol are spreading from the US around the world. The power of markets will ensure that those firms that are more productive and where their costs are lower will be successful -others will be driven to copy them to survive. Since the USA is the technological leader, it followed that US management practices represented current "best practice", which other nations would eventually seek to emulate as they sought to adopt US technology. Thus "patterns in other countries were viewed as derivative of, or deviations from, the US model" (Locke et al. 1995: xvi).
There is an institutional version of this theory (sometimes termed the "North-American phenomenological neo-institutionalism, Djelic/Bensedrine 2001) which argues that institutions reflect power relationships and that, therefore, the economic and technical pressures will be reflected institutionally. Thus there will be coercive pressure to ensure that similar structures and practices are adopted throughout the world (such as the de-regulation "strings" typically tied to IMF loans to underdeveloped countries); normative pressures (from professional bodies, international associations and the growing internationalisation of executive education); and cognitive isomorphism (as international organisations attempt to spread their policies and cultures around the world; see DiMaggio/Powell 1983). It has been argued that one effect of this global institutional isomorphism is that the role of nation states becomes less significant (Meyer 2000).
The convergence thesis has also received support from transaction cost economics which also contends that at any one point of time there exists a best solution to organising labour (Williamson 1975;Williamson 1985). "Most transaction cost theorists argue that there is one best organisational form for firms that have similar or identical transaction costs" (Hollingsworth /Boyer 1997b: 34). Likewise, parts of the industrial organisation literature argue that firms tend to seek out and adopt the best solutions to organising labour within their product markets, long-term survival being dependent 1 Our discussion here, for various reasons, including lack of space, is limited to convergence at the national level. Whilst we believe our analysis would have value with reference to convergence or divergence at sectoral, size, interependence and other levels, we leave that for future articles. on their being able to implement them (Chandler Jr. 1962;Chandler Jr. 1977;Chandler Jr./Daems 1980). Thus there is a tendency for firms to converge towards similar structures of organisation.
This argument has been carried through into HRM. It is implicit in many of the universalist texts which simply ignore national differences and assume that findings from the USA are replicable elsewhere, and explicitly argued elsewhere (see, e.g., Locke et al. 1995).
These various convergence perspectives are characterised by their functionalist mode of thought: practices are explained largely by reference to their contribution to technological and economic efficiency (Brewster/Tregaskis 2003). Management, including human resource management, is a dependent variable that evolves in response to technological and economic change, rather than with reference to the sociopolitical context. The effect is that regardless of auspices, effects on management and labour are similar (Kerr 1983).

Convergence: Institutional driven
There is an alternative convergence argument, based on a different version of the institutional analysis.
This argues that since institutions are strong antecedents of difference the ongoing economic and political integration of European Union countries, for example, may create a convergence towards a distinctly European practice -different from the market convergence model. This concept would see regional convergence, but not global convergence, generating a specifically European model of convergence in HRM (Brewster 1995b). In Europe organisations are constrained at a national level, by culture and legislation and at the organisational level by trade union involvement and consultative arrangements. It is clear that, in general, European countries are more heavily unionised than the United States, and indeed most other countries. Trade union membership and influence varies considerably by country, of course, but is always significant. Indeed in many European countries the law requires union recognition for collective bargaining. In most European countries many of the union functions in such areas as pay bargaining, for example, are exercised at industrial or national level, -outside the direct involvement of managers within individual organisations -as well as at establishment level (Traxler et al. 2001;Hegewisch 1991;Morley et al. 1996). Thus in Europe, unlike in the US, firms are likely to deal with well-founded trade union structures. It is worth noting that studies of HRM in the US have tended to take place in the non-union sector (Beaumont 1991b). In fact a constant assumption in research programmes in the US has been the link between HRM practices and nonunionism (see, e.g. Kochan et al. 1984;Kochan et al. 1986). "In the US a number of.... academics have argued that HRM [the concept and the practice] is anti-union and anti-collective bargaining" (Beaumont 1991a: 300).
In HRM, state involvement in Europe is not restricted to the legislative role. Compared to the USA the state in Europe has a higher involvement in underlying social security provision. Equally it has a more directly interventionist role in the economy, provides far more personnel and industrial relations services and is a more sub-stantial employer in its own right by virtue of a more extensive government-owned sector (for an overview about labour markets in Europe see Siebert 1997).
Finally, there are developments at the level of the European Union or the European Economic Area which impact upon all organisations in Europe. In a historically unique experiment, European Union countries have agreed to subordinate national legislative decision-making to European level legislation. These developments have indirect effects upon the way people are managed and direct effects through the EU's adoption of a distinct social sphere of activity. Thus, this strand of the debate would see convergence not on a world-wide basis, but rather towards different regional groupings based on the developing regional institutions. On such an analysis, the European Union, where these institutions are far stronger than they are in any other regional bloc, is a test case.
None of the "convergers" pretend that they do not see the variety of management approaches around the world. However, they argue that, in the long term, any variations in the adoption of management systems at the firm level are ascribable to the industrial sector in which the firm is located, its strategy, its available resources and its degree of exposure to international competition. Moreover, they claim, these factors are of diminishing salience. Indeed, once they have been taken account of, a clear trend toward the adoption of common management systems should be apparent.

Divergence: Alternative models
Proponents of the divergence thesis argue, in direct contrast, that personnel management systems, far from being economically or technologically derived, epitomise national contexts that do not respond readily to the imperatives of technology or the market. This may be based upon an institutionalist perspective, in which organisational choice is limited by institutional pressures, including the state, regulatory structures, interest groups, public opinion and norms (DiMaggio/Powell 1983;Meyer/Rowan 1983;Oliver 1991;Hollingsworth/Boyer 1997a). Or they may be based on the notion that cultural differences mean that the management of organisationsand particularly of people -is, and will remain, fundamentally different from country to country. National differences in ownership, structures, educational systems, and laws all have a significant effect on the architecture and the practices of employing organisations. This literature has been synthesised in the work of such authors as Hall and Soskice (Hall/Soskice 2001) who draw a sharp distinction between "co-ordinated market economies" of say Germany and Sweden and "liberal market" (Anglo-American) ones. A more subtle version is propounded by Whitley (Whitley 1999) who sees six different possible varieties of capitalism (Fragmented, Co-ordinated industrial district, Compartmentailsed, State organised, Collaborative, and Highly co-ordinated). He defines business systems as: "...distinctive patterns of economic organization that vary in their degree and mode of authoritative co-ordination of economic activities, and in the organization of, and interconnections between, owners, managers, experts and other employees" (Whitley 1999: 33) -according human resource management a distinctive role in creating the difference between these systems.
These institutionalist writers tend to see cultural differences between nations as an aspects of this analysis. Others (e.g. Hofstede 1980;Trompenaars 1994;House et al. 2002) would see the cultural differences as underlying the institutional differences. Cultural values affect every aspect of work and organisation and are largely unseen by the actors involved.
Divergence theorists argue that national, and in some cases regional, cultures and institutional contexts are slow to change, partly because they derive from deep-seated beliefs and value systems and partly because major re-distributions of power are involved. More importantly, they argue that change is path dependent. In other words, even when change does occur this can only be understood in relation to the specific social context in which it occurs (Maurice et al. 1986;Poole 1986). Performance criteria or goals are thus, at any point in time, socially rather than economically or technologically selected so that they first and foremost reflect the national culture and the idiosyncratic principles of local rationality.
The general management discussion is beginning to be reflected in the specific field of comparative HRM (Boxall 1995;Brewster/Tyson 1991;Hollinshead/Leat 1995;Brewster et al. 2000). In this respect, HRM is catching up with another aspect of the study of employment relationships, the study of industrial relations, which has long recognised the importance of international differences (see, for example, Due et al. 1991;Hyman 1994;Poole 1986;Stephans 1990.;Przeworski/Spague 1986;Visser 1992;Bean/Holden 1992;Locke et al. 1995;Hollinshead/Leat 1995). Human resource management is increasingly acknowledged to be one of the areas where organisations are most likely to maintain a 'national flavour' (Adler 2002;Schuler/Huber 1993;Rosenzweig/Nohria 1994) and is the point at which business and national cultures have the sharpest interface, especially in areas such as forms of control (Harzing/Sorge 2003), work systems (Geppert et al. 2003) and team work (Woywode 2002). Even in companies that try to implement 'world-wide' policies, practice is negotiated or varied at national level (Ferner 1997;Wächter et al. 2003).

Types of convergence
From a theoretical as well as an empirical point of view, the notions of convergence or divergence are complex. Although the general meaning, intuitively, is clear, it becomes more complex at a closer look. We propose to differentiate between two different forms of convergence: directional (type I) and final (type II)convergence (for a more in depth discussion of different forms of convergence 2 see Mayrhofer et al. 2002).
Type 1 -directional convergence: Directional convergence occurs where the development tendency goes into the same direction. Regardless of a starting level in 2 A third type of convergence -majority convergence -occurs if a population of organisations in a country become more homogeneous or heterogeneous, respectively, in the use of a certain management concept or tool. For example, if 50% of the organisations in a country use a specific management tool and 50% do not, one can assume maximum divergence within a country since there is a clear 'split' in the concrete use of this tool. In more general terms: the closer a value in a country has moved to the 50% level, the greater the majority divergence has become. Vice versa, the more the value approaches the 100% or 0% level, the greater the majority convergence has become. However, since within country comparisons are not the focus of this contribution we ignore this form of convergence here. each country, if the variable analysed changes in the same direction in each country there is a convergence in direction at least. Table 1 shows the basic idea. There, both in countries A and B the developments point in the same direction, e.g. the use of a certain management tool in each country increases. Nevertheless, the frequency of use in the two countries is at a different level. Type II -final convergence: Final convergence exists when the developments of a variable in different countries point towards a common end point. In other words, the differences between countries decrease. This development is independent of directional convergence (type 1) as different developments in terms of, for example, frequency of use of a certain management tool, can still result in final convergence. Table  2 shows three country pairs as examples of final convergence. This is the meaning of convergence that is most commonly assumed in the literature, even if rarely stated explicitly and sometimes confused with the other forms. Each of the three examples shows a different case of final convergence for illustration. Our results section discusses some of the empirical findings from the Cranet network on these issues, but first we address more general research issues raised by the process of comparative research.

Methodology
Researching large-scale developments in European HRM over an extended period of time raises not only conceptual, but also methodological and research practical problems (for basic views on international and/or cross-cultural research see, e.g., Adler 1983;Kochan et al. 1992;Cavusgil/Das 1997). When doing the kind of longitudinal and comparative research that is addressed here, at least three major issues arise. First, basic decisions about methodology have to be made. Here, as in every other research effort, researchers have to position themselves within the available spectrum of epistemological and methodological options. Second, doing research across national and cultural boundaries adds specific elements to the 'traditional' problems encountered in empirical research, e.g., language barriers, different national traditions of doing research or varying practical circumstances of, say, data gathering. Third, such research efforts cannot be taken on alone. Working together in various forms of research networks seems to be a conditio sine qua non. Hence, the problems of working together in a culturally mixed, geographically dispersed group over a longer period of time and considerable network dynamics have to be taken into account. Consequently, this chapter touches methodological issues at three levels. First, some basic problems of comparative research have to be addressed. Second, there is a need to discuss the consequences of doing research in a small team that is not anchored in a more or less homogeneous national background and, consequently, in a common understanding and, most often, sharing of joint standards of science. Third, the 'usual' issues of the concrete research project have to be mentioned.

Basic problems of cross-cultural and comparative research Choosing a methodological angle
Based on Cartesian dualism which distinguishes between the physical as external reality and thinking as internal world, two basic paradigms of scientific thought and methods have emerged: the objective, deductive and often called quantitative and the subjective, interpretative and frequently labelled qualitative paradigm (see, e.g., Lamnek 1988;Lueger 2000).
From a subjective, interpretative point of view the world is not simply given as an objective reality. Rather, it is subjectively constituted and socially pre-interpreted, formed by the observation schemes of the individual actors. In this process, objective and subjective meaning can be differentiated. Subjectively, the actors themselves attribute meaning to their own actions. On the other side, action can be linked with meaning without referring to the psyche of the actor through the observation of observers (Soeffner 1989;Schütz 1981). Given this background, the methods used within this paradigm usually have to meet specific criteria like openness, communicativity, contextuality or search for meaning (e.g., Lamnek 1988).
From an objective, deductive perspective the focus is on the world as 'given' entity that can be looked at and analysed without referring to subjective interpretation.
Archetypically, this approach is reflected in the approach of natural science which strives for universal laws and testing of hypotheses via quantitative, experimental methods. Critical rationalism presupposes an objective reality, i.e. truth. Through a collective effort called science this truth can be approached more and more. Critique becomes crucial in this approach as it is essential for the core elements like intersubjectively checking results and methods (Popper 1972;Scholtz 1991). The methods used in this paradigm have to meet criteria like connection with theory, objective research process, operational definition and isolation of relevant measures, rational explanation, primacy of falsification (Friedrichs 1973).
Studies covering developments in European HRM are implicitly or explicitly rooted in the objective, deductive paradigm. Although using a variety of research methods including 'typical' elements of 'qualitative' research like single company cases studies or personal in-depth interviews, the basic assumption is quite clear: objective reality -in this case: developments in HRM in Europe -can be captured by using appropriate methods and operationalised through theory-based constructs.

Classic cross-cultural issues
In cross-cultural research a distinction between emic and etic approach has been made (see, e.g., Ronen 1986; Thomas 1993;Holzmüller 1995). "Whereas emics apply in only a particular culture, etics represent universality -they apply to all cultures in the world" (Ronen 1986: 47). If different cultures are compared on the basis of predefined categories (so called cultural dimensions), the researcher's position is outside the observed cultural system because cultures are compared on the basis of dimensions developed in the culture of the researcher. Hofstede's or Trompenaars' crosscultural analyses are examples of etic research approaches. The main disadvantage and critique is the fact that an "ethnocentric bias" cannot be avoided, because one particular dimension (category) may have different meanings in different cultures or one cultural dimension may not exist in every culture. As far as the emic approach is concerned researchers are within the investigated cultural system, they try "to see how the natives conceptualise the world" (Ronen 1986). The emic approach can provide a culture specific level of understanding that gives the researcher an in-depth feel for the nation (Gannon 2001: 52). Though this can be characterised as a culture-adapted, non-ethnocentric approach, categorisation is under strong influence of the researcher's way of perception and the investigator's blind spot may cause distortions. In practical research, the emic/etic differentiation leads, among others, to the question of inclusion or exclusion of country representatives with local know-how and the effects of (not) doing so.
Specific problems in comparative research -noble and not so noble... Different scientific traditions in methodological and epistemological terms exist across countries. Especially in continental Europe, there is a long tradition of interpretative research. Only in recent times, in spite of the domination of the 'objective-deductive perspective', this angle has become more acceptable also in respected top journals (see, e.g., the increasing number of interpretative contributions in top management journals like the Academy of Management Journal). These different traditions, com-bined with a varying amount of methodological training researchers receive during their career and the problem of speaking a common language -even if English is the lingua franca in HR research, the command of English especially in its subtle nuances varies significantly -provide a background where it is not easy to agree on a common approach in international research networks consisting of members from a great number of countries.
Struggling with the issue of 'likeness' and equivalence is one of the big topics in international comparative research (Cavusgil/Das 1997). Although they are not unique to that type of research but a 'universal' issue in research, it becomes especially salient at the international level. The same empirical phenomena can be labelled differently in different countries and, vice versa, different things can carry the same labels. Likewise, the same data gathering procedures can yield quite different results. Therefore, comparative research "is concerned with attempting to compare like with like. In international settings this is not an easy task." (Tregaskis et al. 2004: 440) In addition, practical circumstances of doing research in each country vary considerably. For example, different traditions of how to approach companies -via letter, personal interviews, the web etc. -or the financial resources available for an 'average' researcher make a joint approach even more difficult.

Establishing and maintaining research networks
Ambitious international comparative research efforts cannot be done 'alone' or with only a small group of people. Its scope in terms of content and methods, the geographical spread, the financial involvement and the time investment requires a larger research group with dedicated actors -in other words: an international research networks. Establishing and sustaining such research networks has pitfalls of its' own beyond the typical problems of multicultural work teams (see, e.g., Davison 1996;Ely/Thomas 2001;Erez/Somech 1996).
First, the question of who is starting the network comes up. This has an individual as well as an institutional component. At the individual level, a well known figure in the field may use his or her reputation and contacts to bring together individuals often not (very well) known to each other. This can lead to a problematic communication structure at the beginning as much of the communication and integration is focused on the central actor. In addition, such a 'father/mother figure' has, for a long time, a special place in the social structure of the network. This can have positive as well as negative consequences for the development of such networks. At the institutional level, the initiator's institutional affiliation is crucial. Its reputation as well as the country location plays a role that should not be underestimated in terms of the prestige and credibility ascribed to the network.
Second, the question of leadership, internal power distribution and decision mechanisms constitute critical areas. The scientific communities are -at least at a superficial level -characterised by formal equality and a great amount of individualistic freedom, the latter being for some members of the scientific communities one of the main motivational drivers. Nevertheless, in international research networks a number of subtle (and less subtle) differences resulting in different power bases exist, e.g., qualifications, reputation, seniority, available social capital, country of origin, com-mandment of the lingua franca of the network etc. In addition, larger research networks require some kind of decision mechanisms beyond basic-democratic 'everybody is involved in and decides everything'-style of decisions. Thus, hierarchical differences return through the 'backdoor'. Handling the tension between the equality/individualistic principle of the larger context and the concrete requirements of the research network is a crucial point.
Third, managing international research networks has its peculiarities. Due to the loose institutional bonds, the cultural diversity and the geographic spread such networks have to rely on a number of mechanisms for co-ordination and control (for different types of such mechanisms see, e.g., Mayrhofer 1998;Turati et al. 1998).
Fourth, if networks exist over a longer period of time, inevitably some internal network dynamics occur. Fluctuations in the membership, changes in the formal and/or informal status of network members based on developments in individual careers and professional development and, consequently, modified social relations, conflicts about the future course of the network, struggles about the use of especially promising findings and, linked with that, the distribution of reputation are just a few of the examples of issues arising. Processes become even more complicated if the research network consists of members from different reference systems, e.g., from the scientific area and from the area of consulting. In addition to the issues just mentioned, new problems arise. The ultimate, often implicit understanding of the goal of one's effort -roughly characterised by knowledge creation/insight vs. practical applicability -is usually different. Likewise, different time frames, i.e. more long term vs. more short term, exist.
As can be seen from these few examples, the use of large international research networks -though essential for specific types of international comparative researchis fraught with dangers. Some of those networks are successfully managed and yield excellent output (see, e.g., the Globe-project in the area of comparative leadership research, House et al. 1999). However, other research consortia (it would be invidious to name them here) either never take off or clearly lag behind their own ambitions.

The concrete research project
International HRM has recently experienced a steady growth in research efforts and publications. European researchers in particular have made a number of significant contributions to theoretical, empirical and methodological advances in the field (e.g., Pieper 1990;Poole 1990;Brewster 1995b;Brewster et al. 1996;Gooderham et al. 1999;Brewster et al. 2000;Brewster et al. 2004). Cranet has been part of these efforts, dedicated to analysing developments in the area of HRM in public and private organisations with more than 200 employees in a national, cross-national and quasilongitudinal way (see Brewster/Hegewisch 1994;Brewster et al. 2000;Tregaskis et al. 2004). At the outset, Cranet had two major goals: First, to research whether a pattern of 'Europeanisation', i.e. convergence to a common pattern could be found over time and second, to identify whether changes in personnel policies towards a more strategic human resource management approach have occurred .
The Cranet survey is the largest and most representative independent survey of HRM policies and practices in the world. It includes 37 countries, 27 of them in Europe. Six major survey rounds have been conducted since 1990. Overall, data from roughly 30,000 respondents -public and private organisations -are now available and the number continues to increase. The survey concentrates on 'hard data' like numbers, percentages, ratio etc. and avoids, as far as possible, attitudinal information. To reduce respondent and cross-country bias very few open-ended questions are included. In addition, the translation-retranslation technique (Brislin et al. 1973;Brislin 1976) is used for every country in every survey round.
Inevitably, Cranet has to cope with the basic options and problems of comparative research in HRM highlighted above. From a methodological point of view, it is rooted in the objective and etic paradigm. The assumption made is that across cultures and countries data can be obtained focusing on 'hard' evidence which is not very likely to be biased by cultural assumptions. Nevertheless, a number of variables are inevitably subject to culturally propelled interpretations. For example, concepts like performance evaluation are likely to be interpreted in a different way depending on one's cultural perspective. Although translation-retranslation techniques can help, they cannot solve the basic problems linked with doing objective, etic research in this area.
Cranet handles -not solves -the issues linked to the establishment and maintenance of international research networks by relying on a number of guiding principles. A basic decision has been made to adhere to a set of (network) universal standards. For example, a standard operating procedure has been set up for data collection and input, integration of new questions etc. However, some local adaptations -for example, in the area of how to approach companies in the data collection process -have been made and are accepted within the network. In addition, the methodological discussion is kept going within the research network. This propels mutual learning processes and improves the solutions every network partner chooses for the respective country. In doing this, one has to manage skilfully the trade-off balances between practical necessities and methodological rigorousness. Likewise, there always has been a local partner in each country where Cranet collects data. Thus, the 'ethnocentric bias/temptation' is reduced. In terms of network dynamics, Cranet has gone through various phases. In the initial start-up phase with a lot of growth in member countries, the emphasis was on data collection. The key actors remained quite stable. In the scientific community, the network started to get international attention. In a second phase, the growth rate declined. The focus switched from data collection to data analysis. At the same time, the fluctuation in the network increased as established members were leaving the network or re-positioning themselves within the network. In terms of reputation, the activities of the network gained visibility and respectability. In a third phase, many activities and individual members of the network have become well known and an accepted part of international comparative research in HRM. Some high profile publications in top journals (e.g., Gooderham et al. 1999) indicate this. Still, the size of the network and the differentiated career paths its members pursue means that it remains a fragile unit.
In the following section, we will try to pull together insight from different pieces of the empirical work of Cranet over the past 15 years. As these come from various sources using different sub-sets of our own data as well as other authors' work, the necessary information about the used part of our sample is presented at the respective place in the text. Of course, for our own as well as for other authors' work, further information and details can be obtained by following the given references.

Results
By identifying major trends in the development of European HRM, the focus is 'zoomed out': We do not want to point at specific countries or industrial sectors. Rather than commenting on such details, we would like to identify major, constituting characteristics of the overall picture. To be sure, this overall picture is by no means complete. Despite the efforts of Cranet and many other most valuable contributions, empirically and theoretically, we are far from understanding completely what goes on in European HRM. Nevertheless, parts of a picture are there and -with all the caution required -we can see at least four major trends constituting important characteristics of the evolving picture or HRM in Europe. The first two of them -'European HRM is different' and 'Great variety of HR practices in European countries' -refer to a more static view. The latter two -'Frequent stasis' and 'Some evidence of convergence' -include a dynamic component.

European HRM is different
As we have seen, there have been claims to identify a unique European model of HRM. The conceptual arguments have been outlined in section 2 of this paper (see Brewster 1995b). Empirically, however, it is not easy to demonstrate convincingly that HR practices in Europe and the US (or, indeed, elsewhere) are clearly different, the main reason being a lack of adequate comparable data. Nevertheless, considerable evidence points in the direction of a distinctive European approach.
First, the legal environment relevant for HRM differs considerably between the US and Europe. Undeniably, even within Europe large differences exist between countries with a lot of labour related regulations like the German speaking area and comparatively less regulated countries like, e.g., Ireland or the UK. Yet, clearly, the density of labour regulations is higher in Europe than in the US (Grubb/Wells 1993).
Second, the main actors in industrial relations have a different role in Europe and the US. Trade unions and employers' associations have more members and more influence in the former. The role of collective bargaining and collective agreements, the influence of trade unions in the political system and their importance for management decisions are just three of the important factors.
Third, a number of studies examining specific aspects of HRM point towards important differences between Europe and the US. These would include aspects like skill-level and available types of qualifications in the work force (Mason/Finegold 1997), the role of human resource development professionals (Nijhof/de Rijk 1997) and managerial attitude towards employees' participation in decision making (McFarlin et al. 1992), Fourth, and maybe especially relevant, is a situation where US and European views of HRM, encounter each other directly: in the case of subsidiaries of US multinationals in Europe. Multinationals and their subsidiaries seem to play a special role in the diffusion of practices and know-how in HRM (see, e.g., Müller 1998;Ferner 1997;Gooderham et al. 1998). The subsidiaries of US multinationals are different from in-digenous organisations or multinationals from other countries which, by and large, adapt more to the local environment. US multinationals not only seem to have a rather ethnocentric approach to international HRM, with little re-transfer of best practices from their overseas operations, but are also particularly proactive in searching to by-pass local conditions that they see as constraints (Ferner 1997;Gooderham et al 1999). These are indicators for differences between how HRM is understood and implemented in the US and in Europe.

Great variety of HR practices in European countries
When 'zooming in' towards HR practices in specific European countries, it becomes clear that in all major functional areas there are significant differences between European countries. Practices from four core functional areas of HR -organisational role, recruitment and selection, staffing, compensation -can serve as examples: • Formal representation of the HR function at the highest board of the organisation and the stage at which HR is involved in the development of corporate strategy -as indicators of its role and importance. • The use of internal mechanisms for filling managerial vacancies -as an indicator of recruitment approaches for crucial parts of the work force. • The proportion of workforce on part-time contracts -as an indicator of the configuration of the work force. • The proportion of the annual salaries and wage bill spent on training and development -as an indicator of the importance of well qualified labour. When looking at these issues, the following picture emerges (see Table 3): Without going into the detail -and the problematic -of such comparisons, the table illustrates a remarkable degree of difference: what is widespread or standard practice in one country plays little or no role in others.

Frequent stasis
We live in a world of full of change. The accepted wisdom is that individuals and organisations have -ideally: as quickly and as smoothly as possible -to adapt to ever changing contextual conditions. Massive change drivers like globalisation and the diminishing importance of national barriers, growing competitive pressures and virtualisation trigger new forms of organisations, work and careers. Indeed, prima facie plausibility and singular, eclectic evidence both seem to confirm the dictum of ubiquitous change.
However, Cranet's empirical results seem to be in striking contrast to such a dictum. In the area of European HRM, stability and little change seems to be not the exception but rather the rule. To be sure, this does not mean that everything stays exactly the same. Yet, very often even in areas where one might expect change (even the academics are seemingly caught up in the "change-frenzy" often generated by consultants or idealised cases), the data remain unclear or even point in the opposite direction. Let us look at three examples.
The question of evaluation of HR departments has been widely discussed in recent years. Due to an increasing pressure on all units not directly adding value, HR departments and their work are under close scrutiny, the acid test for HR being: is it worth it? The need to prove a contribution to the overall organisational performance leads to a growth in the formal evaluation of HR activities. Given this background, one would expect a sharp increase in the number of HR departments that are regularly and formally evaluated.
The empirical reality challenges these expectations. Our analyses use 18 European countries, (i.e. all EU before the latest enlargement in 2004 except Luxemburg, with Germany split into the old and the new federal provinces, and in addition Turkey, Switzerland and the Czech Republic). Overall, for 20,610 private sector organisations with more than 200 employees appropriate data were available for the decade between 1989 and 2000. 4 For the longitudinal analyses only variables were included where identical or nearly identical questions were available for all survey rounds.
Across the surveyed countries there is no clear trend when looking at core variables for which a longitudinal analysis is possible. Several examples can illustrate this.
A first example is the degree to which HR departments are formally evaluated. When looking at the developments in this respect, the following picture emerges (see table 4). Obviously some change occurs, but these changes do not all reflect any clear trend or a significant development into one direction or the other. Of course, this does not say anything about changes at the level of the individual country or organisation. Analysing aggregate data at the overall European level averages out such changes. But it provides a "bird's-eye" view of the overall situation across Europe. A second example is the importance of training and development measures. Given the widely claimed emergence of the knowledge society, the changing demands of work processes and the increasing significance of a well-educated work force for handling organisational transformation processes, one would expect an ever-increasing amount of training in organisations.
However, if one looks at a major indicator for training and development activities -the number of training days per year for various groups of employees-no such trend whatsoever is discernable (see table 5). Again, a certain amount of variation is visible. However, nothing points in the expected direction. Practices seem to be quite stable. A final example refers to the role of the HR function and its relationship to line management. The notion of a general decentralisation of HRM might leads us to expect that parts of the responsibility as well as the operational tasks will shared with line management. Hence, the percentage of European companies devolving HRM responsibility away from specialists towards line management should have been increasing through the decade.
However, again, the data does not support such a hypothesis. The mean value of a composite index indicating the relative distribution of responsibility between HR specialists and line management reveals little momentum over time. Indeed, there is a slight tendency towards centralisation -on average 0.15 points per year between 1989/90 and 1999/2000, on a scale ranging from 5 to 20. In only two countries are there statistically significant changes, in all the other countries the changes are not sig-nificant and are as likely to go towards greater centralisation as greater decentralisation (for a more in depth discussion of these findings see Mayrhofer et al. 2002;Larsen/Brewster 2003;Mayrhofer et al. 2004).
These three examples were deliberately chosen from areas where one would, relying on "received wisdom", expect obvious changes. But even in such areas stability seems more common than change. Hence, one would not be surprised to find little variation in other areas. In fact, many of the areas researched by Cranet are remarkable stable in their development. Overall, the absence of change is an important part of the European picture.
Of course, in exploring these three examples a number of difficulties like the operationalisation of the variables or the problem of multi-level phenomena have to be kept in mind. Nevertheless, we would argue that if organisational practices had changed dramatically, the data would -notwithstanding research imperfections -reflect these changes. Put simply, they do not.
However, there is some evidence of change and the issue of convergence or divergence of HR practices in Europe still is open for debate

Some evidence of convergence
There are, perhaps, at least some first hints about areas of convergence in some aspects of the Cranet data. Our analyses focus on practices in various functional areas of HR that do seem to show signs of such changes. We want to analyse the effects of major change drivers -whether cultural, institutional or the market: Do they lead to converging or diverging developments? For the analyses, the sample briefly described above from the 18 European countries was used. 5

Directional convergence
Developments towards directional convergence in Europe can be found in four areas (more details and tables in Mayrhofer et al. 2004): • Decreases in the size of the HR department relative to the overall work force, • slight rises in the percentage of the annual salaries and wage bill spent on training (which, of course, may reflect a growing disparity between wage growth and the growing costs of training rather than a 'real' growth in the amount of training), • increases in the amount of information being given to employees about company strategy and financial performance, and • a more frequent use of performance related compensation systems. In all these cases, the change in the average developments over all countries is statistically significant, all statistically significant changes at the country level point in the same direction and are compatible with the average development, and the majority of the non-significant changes at the country level are in the same direction as the significant changes overall and at the country level. 5 For a more detailed discussion about the methodology applied and the detailed results see Mayrhofer et al. 2002;Müller et al. 2001;Mayrhofer et al. 2004 In all other patterns of practices analysed -the use of flexible work arrangements, the level at which the HR policy is determined and the responsibility shift from HR departments to line managers -the evidence is mixed or rather weak. Although there are some indications of converging developments, the empirical evidence is by no means clear and, therefore, we choose to be cautious in our claims.

Final convergence
Overall, the evidence for all the HR practices analysed is clear: there is no unequivocal trend towards final convergence. On the contrary, developments across European countries diverged during the 1990s, having a maximum point of convergence mostly around the middle of the decade rather than at the end (see Table 6 and, for a more detailed picture of the results, see Mayrhofer et al. 2004). 'Pure' final convergence would include a 'narrowing down' of differences between European countries indicating a more common practice in the various countries. For the first part of the '90s, this is the case in some areas such as the use of variable and performance related elements of compensation or the use of flexible working practices. However, between the mid-'90s and the end of the '90s, the heterogeneity of HR practices in Europe increased again. Instead of a decrease of variety, (in other words, final convergence, i.e., a movement towards a common or more similar 'endpoint'), the data shows increased divergence.

Concluding remarks and open questions
Overall, how are we to understand the evidence in the light of the conceptual and empirical considerations? Neither the emergence of a European model of HRM nor the great variety of HR practices in European organisations come as a big surprise. Conceptual as well other empirical work have indicated this before. The empirical work reported here corroborates previous insight and puts new nuances to it. The empirical data show that in very 'traditional' areas of HR country differences do continue to exist. Given the current situation within Europe this is again not very surprising. Despite the increasing common elements in the legislative framework of the EU countries, the access countries and even countries such as Switzerland that for a number of reasons make efforts to harmonise their legal system with essential legislative rules of the EU, country differences in institutional terms still matter. The role of trade unions and works councils, the level of safety and health regulations or the amount of regulation of the labour markets are just a few prominent examples of these institutional differences. Clearly more exciting from our point of view are the findings that deal with the more dynamic element of developments over time. In some circumstances, the absence of change is remarkable. The development or, more precisely, the non-development of various areas of HR are such a remarkable instance. As shown above, the data imply a relative constant picture during the 1990s. This is a clear antidote to the 'change frenzy' that has infiltrated much of scientific and practitioner oriented writing about the situation in Europe. Combined with the previous point about the heterogeneity of the stable picture, this is a further attack on the messages of 'ultimate solutions', "best-practices" that lead to organisational success and related models (Marchington/Grugulis 2000). The frequent stasis does not rule out change and even convergent developments. As was shown, there are some areas where at least directional convergence occurs. A closer look shows that directional convergence, i.e. the movement of a variable into the same direction, for example: more frequent use of certain instruments, occurs in areas where there is a clear and 'overwhelming' pressure coming from various sources. If economic necessities, institutional requirements and/or management folklore point in the same direction, then a more or less consistent trend can be seen. The reduction in the comparative size of the HR department or the increasing amount of money spent for training and development measures can be mentioned as examples of this. Here, at least the mentioned sources seem to be supporting factors for the observable developments. One might speculate that only the joint appearance of a variety of factors pointing in the same direction has a measurable impact on the European landscape of HR. In all other cases, not much changes besides the rhetoric.
To be sure, this is not to say that HR stays the same at the level of the single organisation. However, without the combined effects of various sources we seem to experience replacement effects. It is likely that as some companies move in one direction, others move in the opposite one. The combined effect would be an observation of little change at the aggregate level of HR practices across countries.
In general, it seems clear that we need a more nuanced of convergence in HRM policies and practices than has been apparent hitherto. Clearly, there are differences between European countries and, in turn, overall they differ from the general picture of HRM presented in the US literature. However, things appear to change slowly in HRM and perhaps the decade-long data presented here examines too short a period. From a directional convergence point of view, there seems to be at least some positive indication of convergence in some areas. However, looking at final convergence, things become more blurred. None of the HR practices converge at the end of the decade. Rather, the maximum point of convergence is reached in the middle of the decade with signs of divergence after that.
These broad conclusions leave us with a number of open questions: • How are we to handle unresolved methodological questions, such as whether to measure HRM at local, national or regional levels? which aspects of HRM to measure? What techniques to employ? and so on. • What determines whether stasis, convergence or divergence occurs?
• What is a sufficient and appropriate time segment over which to measure convergence or divergence? To answer these questions or at least have more insight, more evidence and more analysis, as so often, is clearly needed. Cranet members are committed to continuing the work they have started and a new round of the survey is currently underway. In addition, the Cranet network continues to expand, taking in more and more new countries. Other networks, such as the GLOBE network (House et al 2004), taking different approaches to these and related topics, will also add to our understanding. Support or challenge to these findings will also continue to be forthcoming from the groups applying more case study based and qualitative methods. There is much scope for expanding our understanding of comparative differences in HRM and how they are developing. It is a difficult, but worthwhile, task.