Corps de l’article

Introduction

During the 1970s and 1980s, the western world was in awe of the Japanese economic ‘miracle’, which was arguably predicated upon a complex and integrated system of ‘uniquely’ Japanese management techniques (Dore, 1973; Ouchi, 1981). The strong competitive capabilities of many Japanese corporations coupled with their global expansion partly through foreign direct investment (FDI) were met with significant interest among western management practitioners and scholars. The race was on to unpack the secrets of this economic miracle, particularly ‘Japanese’ human resource management (HRM) and employment relations (ER) practices (Dedoussis, 1995).

Internationally, Japanese management practices were distinct in many ways. Some attention focused on the so-called ‘three pillars’: seniority-based wages; life-time employment; and company-based unionism (Dore, 1973; Sekiguchi et al., 2016). While these were always far from universal—life-time employment, for example, only ever covered a minority of workers, largely those in ‘core’ jobs in large firms—they were a part of a wider set of practices that laid the foundation for the development of ‘lean’ management and just-in-time production, team-based work design and employee involvement to simultaneously maximize quality enhancement and minimize waste and defects (Whittaker, 2013). Scholars of HRM in multinational corporations (MNCs) became interested in the extent to which these practices, or ‘equivalents’, were transferred to the foreign subsidiaries of Japanese MNCs. The findings from these studies included a strong emphasis on personalized forms of management control, a preference for dealing with single unions or avoiding them altogether, and ‘direct’ forms of worker involvement (Benson and Debroux, 1997; Shadur et al., 1995, Purcell et al., 1999, Delios and Bjorkman, 2000; Sekiguchi, 2016).

Recent evidence suggests that Japanese companies at home and abroad are adopting US style HRM practices such as performance-based pay and short-term employment contracts to contain costs and improve labour productivity (Aoki et al., 2014). Furthermore, Pudelko and Harzing argue that “subsidiaries of Japanese MNCs have a very strong tendency to abandon their home-country practices and move toward US practices” (2007: 549). US style HRM and ER practices are underpinned by a unitarist ideology, focused on the individual as a unit of analysis in which HRM practices influenced by the strategic goals of the firm are organized into a mutually reinforcing system of HRM practices (e.g., performance management, training and development, job design, etc.) designed to elicit greater worker discretionary effort (see Kaufman, 2014 for a review of the evolution of American HRM). Over the last two decades, US HRM has evolved to focus on high performance work systems (HPWS) (Copper et al., 2019; Ulrich, 2016). HPWS are systems of HR practices designed to enhance employee skills, commitment, and ultimately performance, and include flexible job assignments, rigorous selection processes, extensive training and development, merit-based performance appraisal and competitive compensation (Cooke et al., 2019). Moreover, American style HRM is often characterized by anti-union sentiment and union avoidance, as well as the dominance of management prerogative (Kaufman, 2014).

Our paper compares Japanese and US MNC subsidiaries on their deployment of HRM and ER practices across four countries. Our study uses convergence-divergence framework to reconcile findings. This framework examines “whether systems of HRM, and the individual practices comprising them, are becoming more similar or dissimilar over time and across nations and regions” (Kaufman, 2016: 338). We draw on data from representative (Japanese and US MNCs N=460), parallel surveys of MNCs operating in Canada, the UK, Spain and Australia and test a set of hypotheses examining similarities and differences between subsidiaries of Japanese and US MNCs in relation to management control across borders, remuneration, representation and worker involvement. These HRM practices are selected based on the literature that suggests that compared to US MNCs, Japanese MNCs rely on centralized control and Japanese managers in subsidiaries (Shiraki, 2007); that employees in Japanese MNCs are engaged on a long-term basis with extensive firm specific knowledge and, as such, short term performance remuneration is less likely (Peltokorpi and Froese, 2016); that union density is likely to be higher in Japanese MNCs but that unions are enterprise-based (Jacoby, 2005); and that workers are more likely to work in teams in manufacturing, linked to lean production systems (Sekiguchi et al., 2016). There are other differences that are noted in the literature around the use of expatriate and third country managers (Pudelko and Tanzer, 2013), as well as distinct arrangements around networking and knowledge transfer (Thite et al., 2012).

The comparators, the US and Japan are the largest investors globally in other countries (UNCTAD, 2018). The four recipient countries selected represent liberal market economies in terms of the varieties of capitalism typology (Hall and Soskice, 2001). Within this broad classification there are national differences linked to regulations, culture and institutions (Edwards et al., 2013). Moreover, the context in which MNCs operate can alter their HRM practices (Edwards et al., 2013). They are all either major sources (the UK) or recipients of foreign direct investment (Australia, Canada and Spain). All four countries have major trade and investment linkages with Japan, and all four countries have been major recipients of US FDI (Edwards et al., 2013). Australia, Spain and Canada are in the top 20 nations in terms of FDI inflows, while the UK is in the top 10 in terms of FDI outflows (UNCTAD, 2018). The four countries share some characteristics but also differ from one another. In terms of shared features, they are all recipients of significant Japanese FDI (Japanese External Trade Organization, 2015) and are countries where there has been significant debate about human capital formation and its relationship to foreign investment (Edwards et al., 2013; McDonnell et al., 2011). These are important characteristics that justify a focus on these countries. In terms of differences, they exhibit quite significant institutional differences on issues such as employee representation and pay determination, and regulations of employment conditions where Spain is the most regulated (and approaching co-ordinated market economy status), the UK the most deregulated, with Canada and Australia occupying intermediate positions (Edwards et al., 2013). There are geographic and trading differences; the UK and Spain are firmly linked to Europe, while Canada is linked to the Americas, especially the USA; and Australia is linked to Asia (especially China and Japan). In terms of economic structure, Canada and Australia are major commodity producers, especially mining, energy and agriculture that makes them important suppliers to Japanese MNCs (Jalette et al., 2019). The four countries selected offer similarities and differences, and as such this variety supports the examination of MNC HRM practices across the four countries.

In this way, we examine whether Japanese MNCs deploy practices that are similar or different to US MNCs. Using the convergence-divergence framework, our paper contributes to a greater theoretical understanding of why HRM practices are adapted or not within MNC subsidiary operations. This is an important contribution given that “more than three decades of research into the convergence-divergence debate has failed to lead to a unified understanding of why HRM practices are adapted or not to a new country context” (Mellahi et al., 2013: 2341). We argue that notions of transformation in Japan and the associated disappearance of a home country institutional influence on HRM practices in Japanese MNCs are overstated. On the basis of significant differences in practice in some areas, we argue that institutional influences are far from dead; and if there has been a process of convergence between Japanese and US MNCs, then it is a partial one. We draw theoretical implications for convergence and divergence debates, and the complex development of HRM and ER systems in Japanese and US MNC subsidiaries. We unpack country of origin effects, host country institutional forces, and market competitive forces influencing MNC pressures to adopt ‘universal’ HRM practices to provide greater explanation of the character of HRM and ER in Japanese and US MNC subsidiaries.

Institutions and the Transfer of Practices in MNCs

There is a large body of work that regards national institutions as key influences on employment practice in MNC subsidiaries. Understood as the ‘rules of the game’, institutions create pressures to which MNCs have somehow to adapt (Edwards, Marginson and Ferner, 2013; Edwards et al., 2016). Given that, by definition, MNCs span multiple national environments, subsidiary managers must find a solution to the tension between the desires to obtain legitimacy within the MNC by adopting globally standardized practices favoured by higher levels of management and simultaneously gain legitimacy within their local environment by adapting to contextual norms. This tension is often referred to as ‘institutional duality’ (e.g., Kostova and Zaheer, 1999; Roth and Kostova, 2003). If we are to understand the nature of employment practice in the foreign operations of Japanese and US MNCs, we must locate this within an analysis of the institutions and practices in their home countries.

For US MNCs, there is evidence that their home country institutions affect the forms of management control across borders. The development of strong capabilities in a range of management functions at the centre of large, multi-divisional firms capable of setting the strategic direction and monitoring processes and outcomes across geographically dispersed operations is one of the distinguishing features of American capitalism (Pudelko and Tanzer, 2013). These ‘organizational capabilities’ allowed US MNCs to extend these forms of control to the international level, resulting not only in a relatively centralized form of decision making, but also one that was highly formalized, reliant on rules and procedures (Ferner et al., 2004; Pudelko and Tanzer, 2013), and was able to transmit US style HRM throughout the world (Brewster, Mayrhofer and Farndale, 2018). In this sense, the institutions in the country of origin condition the ways in which US MNCs control their international operations.

Moreover, there are also grounds for anticipating that institutions in the country of origin will affect the practices that MNCs deploy in their foreign operations. The developments in the financial sector in the US have been influential on the system as a whole (Deeg, 2012). The rise of activist institutional investors, the presence of a market for corporate control and the deregulation of many securities markets have created pressures for changes in other spheres, including the labour market; along with other factors, these changes have contributed towards a marked contraction in the role of employee representatives in general and collective bargaining in particular (e.g., growth of unitarist and individualistic HPWS) (Deeg, 2012). Thus, we might expect that hostility to organized labour will be a trait of US MNCs. Corresponding to this hostility is an emphasis on individualism, which is a product of both the formal institutions in the USA, notably the anti-collectivist forms of labour regulation, and the more informal institutions and norms which support and legitimize these practices (e.g., Almond and Ferner, 2006). Consequently, we are likely to see HRM practices in US MNC subsidiaries that emphasize formal systems of control, performance-related pay, non-union management approaches and an emphasis on management prerogative (Brewster, Mayrhofer and Farndale, 2018; Kaufman, 2014).

In Japan, the institutions in the domestic setting are different to those existing in the US in many ways. Notwithstanding recent changes, the financial system has been characterized by stable ownership, with cross-shareholdings between firms within keiretsu being an important distinguishing feature (Aoki and Lennerfors, 2013). These networks of firms, held together through ownership links, provided a platform in which employment practices were also characterized by stability. For instance, many firms developed strong internal labour markets characterized by management positions being filled by promotion from within (Sekiguchi, 2006), leading to managers being seen as insiders with long service. Arguably, we might expect this reliance on management as insiders and personal forms of control to be evident in the way that Japanese firms operate across borders.

The stability in ownership not only affected managerial career paths and forms of control but also laid the foundation for distinct employment practices, many of which seemed to be central to the success of Japanese firms (Sekiguchi et al., 2016). A key element was the mutually reinforcing ‘three pillars’: life-time employment was provided to a cohort of core employees in large industrial organizations to ensure employment security and high commitment from employees (Shiraki, 2007); seniority-based wages complemented life-time employment by offering pay and promotion based on the length of service with the organization (Peltokorpi and Froese, 2016); and enterprise unions were based at the company level and served as the unit for representing and protecting employees’ interests and welfare (Zhu and Warner, 2000). In sectors such as electronics and automotive, these practices were combined with practices designed to involve employees in the organization of work and process improvements, tapping the knowledge of employees and attempting to reduce defects and waste in what became seen as highly successful quality enhancement and cost minimization business strategies (Aoki et al., 2014).

However, things have changed in Japan. One element of this has been the loss of confidence that senior executives have in traditional Japanese management (Pudelko, 2009). It is also the case that institutions and practices in Japan have changed. In some cases, the keiretsu links have been unwound as the confidence of key actors in the profitability of other firms in their network has eroded. Nissan is a prime example of this development (Aoki and Lennerfors, 2013). In the employment sphere, lifetime employment has fallen in its coverage and has in some cases been redefined as ‘lifetime career support’ in which firms commit to helping their employees stay in employment but not necessarily with them (Aoki et al., 2008). Seniority-based pay has also diminished in importance as some firms have experimented with merit- and performance-related pay schemes and greater use of a peripheral workforce of temporary employees (Aoki, 2008). Aoki et al., (2014: 2569) argue that, “it is evident that pressures to reduce costs, particularly since 2008, have contributed to the increasing influence of a market-logic price system or the Anglo-American model”. More generally, the Japanese model has become more open to outside influence. As Vogel (2006: 220) states, “Japanese corporations have more foreign owners, managers and business partners than ever before, and these foreign actors bring with them different practices and norms”. This may also be explained by the ‘global’ export and proliferation of American-style HRM and HPWS as ‘best practice’ approaches to the management of human resources (Kaufman, 2016).

The scale of these changes appears to have created a hybrid model, rather than a complete transformation. Whittaker (2013: 259) notes that Japanese employment relations are now perceived as located “somewhere between the organizational pole of the ‘classic model’ and the ‘market pole’ of, for example, the USA”. Any shift away from traditional Japanese management practices should therefore be understood within a context of the importance of complex group relations and enduring values of security and stability (Carr and Pudelko, 2006; Pudelko, 2005). Vogel (2006) argue that the interests of those in Japan have been in preserving a degree of continuity with the past in a process of selective change that he terms ‘patterned innovation’.

This raises an important question relating to the HRM and ER practices of Japanese MNCs’ subsidiaries in relation to their US counterparts operating in the same countries: Are there significant differences in practice between the two groups as a reading of institutional effects would suggest? Or, alternatively, have Japanese MNCs gone further than firms domestically in abandoning a distinct Japanese flavour to their practices such that there are no significant differences between Japanese and US MNCs, as Pudelko and Harzing (2007) claim? In the next section, we briefly outline the convergence-divergence framework, then follow the logic of institutional differences in developing hypotheses concerning the extent of differences between US and Japanese MNCs.

Convergence-divergence framework

The examination of convergence-divergence of HRM practices among comparative and international HRM scholars has grown in recent years (Mellahi et al., 2013; Brewster, Mayrhofer and Cooke, 2015). In the convergence-divergence literature, scholars have focused on how HRM practices across different countries are changing and the extent to which they are becoming similar or dissimilar (Kaufman, 2016). Brewster, Sparrow and Vernon (2007) develop a useful four-fold typology to explain the complexity of HRM convergence-divergence. First, directional convergence occurs “when the trend is in the same direction”; second, final convergence refers to “when the trend is not only similar but toward a common end point”; third, stasis refers to “no change”; and fourth, divergence occurs “when the trend is in different directions” (p. 68-69). Kaufman (2016) adds some other permutations including majority convergence (organizations in countries becoming more similar) and cross convergence (conflicting pressures from cultural versus economic practices). Within most of the convergence-divergence HRM literature, the US model of HRM is often the benchmark given that the dominant paradigm of HRM/HPWS was developed in the US and that US-style HRM has been shown to consistently lead to higher levels of individual and organizational performance. Moreover, US-style HRM/HPWS practices are universally regarded by many HRM scholars as leading to higher levels of individual and organizational performance in other national contexts, irrespective of cultural and institutional differences (e.g., China, see Choi, 2019; Ma, Shu and Zhong, 2019; Liu and Wang, 2019). Consequently, given global competitive forces, Mellahi et al. (2013: 2343) state, “there is an expectation that US models will gain a hegemonic position in the global business context and that all HRM practices will ultimately converge on this US model”. However, scholars using the varieties of capitalism framework argue that institutional forces are powerful and that there is little evidence of final convergence of HRM practices given the ‘countervailing effects’ of local cultures (e.g., social preferences) and institutions (e.g., legal regimes, religions, history and political and business systems, see Brewster, Mayrhofer and Cooke, 2015; Kaufman, 2016). Moreover, Kaufman (2016) argues that given the need for organizations to adapt to global competition, HRM practices will be influenced by the international division of labour, as well as options of competitive advantage through the differentiated use of HRM practices (as influenced the production of low-cost versus differentiated goods and services) which will invariably lead to cross-national and within-country HRM divergence.

Hypotheses development

One of the key characteristics of how MNCs manage their international workforces concerns their approach to controlling their operations across borders. To ensure the ongoing success of subsidiary operations, MNC headquarters exercise control through the development of coordination mechanisms to manage the relationship. With appropriate control, MNCs can achieve strategic goals and minimize deviations from standards which may ultimately lead to an alignment between the MNCs’ subsidiaries function and the headquarters’ strategic goals and management policy and practice (Paik and Sohn, 2004). Control can be exercised in different ways (Ferner, 2000). Personal control is characterized by close personal supervision of the operating units by headquarters’ staff, as well as direct and frequent contact between senior headquarters’ managers and the subsidiaries (Ferner, 2000). As such, this may involve assigning personnel from headquarters to key positions in the subsidiary to oversee operations (Pudelko and Tanzer, 2013). In contrast, bureaucratic control attempts to restrict subsidiary management’s role and authority by enforcing formal rules, regulations and procedures to regulate the operation of the subsidiary. This form of control is likely to be exercised through formal policy-making bodies and other management structures that monitor subsidiaries.

US companies grew domestically through the development of structures capable of exerting bureaucratic or procedural control. Many US MNCs extended the centralized and standardized forms of decision making that they used within the US to the international level, resulting in forms of cross-border control that emphasized formal rules and procedures (Ferner et al., 2004). Ferner and Almond (2013) claimed that US firms are more inclined to use formalized and proceduralized forms of control than personal control.

In contrast, the Japanese business system is characterized by strong links between firms within keiretsu and strong personal links between individuals within and across firms; in other words, it relies on networks. The tendency of Japanese firms to develop strong internal labour markets is a part of this emphasis on networks, with management positions being filled largely by insiders. There are grounds for anticipating that this will be extended to the international operations of Japanese firms. Accordingly, studies of the management of subsidiary operations in Japanese MNCs over a long period (summarized in Ando, 2014) have demonstrated that they have a greater tendency to adopt an ethnocentric staffing policy when compared with MNCs of other nationalities which could be “partially attributed to their managerial practices, including consensus building and shared decision making, which require people-dependant and communication-intensive management processes” (2014: 2009). Other research has indicated that Japanese MNCs generally appoint parent country nationals to managerial positions in their overseas operations (Delios and Bjorkman, 2000; Shiraki, 2006; Pudelko and Tanzer, 2013). These appointments represent a way to assure consistency in managerial practices across foreign operations which might be required given that the practices associated with Japanese firms, such as lean production and worker involvement in work organization, involve processes that are dependent on people’s tacit knowledge and intensive communication. Japanese MNCs are found to utilize close and frequent personal contact and extensive supervision of subsidiary operations by expatriate managers as a means of control (Ferner, 2000; Legewie, 2002; Sekiguchi, 2006). We propose that Japanese MNCs are more likely than those from the US to use personal forms of control and are less likely to use procedural forms of control in their foreign subsidiaries.

  • H1a: Japanese MNCs are more likely to use personal forms of control in their foreign subsidiaries relative to US MNCs.

  • H1b: Japanese MNCs are less likely to use procedural forms of control in their foreign subsidiaries relative to US MNCs.

A second area of exploration is the extent to which an individual’s pay is based wholly or partly on a regular and systematic assessment of job performance, a practice that is well established in the US (Chiang and Birtch, 2007). By international standards, pay in the US is more commonly determined by assessments of individual performance than is the case in other counties (Jacoby, 2005). US style performance-related pay is short-term oriented and performance can be appraised through management by objectives (Keizer, 2011). There is some evidence that US MNCs are distinguished from those of other nationalities by their tendency to deploy performance-related pay schemes in their foreign operations (Faulkner, Pitkethly and Child, 2002; Ferner and Almond, 2013).

Traditionally, the pay of workers in Japan was shaped more by seniority than performance. However, the push for more efficient decision making as a result of intense global competition and the post-1990 technological revolution have forced some Japanese firms to modify their HRM systems to enhance their performance (Miyamoto and Higuchi, 2007). The introduction of performance-related pay systems from the 1990s onwards was part of an attempt to develop a stronger relationship between individual performance and remuneration, representing a significant innovation in Japanese compensation practices and broader HRM practices (Keizer, 2011; Miyamoto and Higuchi, 2007; Aoki et al., 2014). Compared with the US, such developments would seem to be in their infancy and have only achieved partial coverage, being pursued vigorously by those firms that were most affected by foreign influences but much less evident in other firms (Aoki et al., 2008; Aoki et al., 2014). Moreover, while performance-related pay is often used as a control mechanism for managers in Anglophone countries (Ferner and Almond, 2013; Bartram et al., 2015), evidence suggests that Japanese companies are more likely to use more personalized forms of control for managers rather than performance-based controls. Therefore, we expect that Japanese MNCs are less likely to use performance-related pay practices in their foreign subsidiaries relative to US MNCs.

  • H2: Japanese MNCs are less likely to use performance-related-pay practices in their foreign subsidiaries relative to US MNCs.

Jacoby (2005: 129) argues that non-unionism is one of the things that “distinguishes US corporations from their Japanese counterparts” and that the employees of US firms are “weak stakeholders at best”. We know that there is evidence that the approach of US MNCs is shaped in part by their approach at home (e.g., Almond and Ferner, 2006).

The situation in Japan is notably different. Many Japanese firms have a long-established practice of recognizing trade unions. A key pillar of Japanese HR practice is enterprise unionism. This was adopted in post-1945 Japan and differs significantly from union structures and operations developed in Western countries, which are usually categorized by industry or sectoral-based unions (Vo and Rowley, 2010). Japanese unions are enterprise or workplace-based and represent workers solely within the company (Benson and Gospel, 2008), although they are usually affiliated with industry and national confederations allowing a degree of coordination across the enterprise unions (Benson and Gospel, 2008). There are grounds for anticipating that Japanese MNCs will not find it easy to ‘export’ such enterprise unionism to their foreign subsidiaries, particularly where unions are organized along different lines and do not identify with the goals of a particular company. Accordingly, the evidence suggests that while some Japanese MNCs adapt to various types of union structures when developing and managing their subsidiary operations abroad, many others tend to avoid trade unions and place a high emphasis on a union-free workplace (Vo and Rowley, 2010). Despite this clear difference in practice within Japan compared to that within the US, therefore, we do not expect to see significant differences between Japanese and US MNCs in other countries.

Where we might expect to see a difference is in relation to alternative forms of employee representation that do not involve unions as key actors. Japanese culture places great emphasis on group harmony with the business enterprise and, consequently, corporate decisions in Japanese firms are often made with one eye on achieving group consensus (Sekiguchi et al., 2016). Decision making in which groups have influence is seen in Japan as promoting morale, developing team spirit and gaining employee commitment (Sekiguchi et al., 2016). The traditional process for corporate decision making is known as the ‘ringi’ system which involves ideas, suggestions and requests being submitted to senior-level managers by lower-ranking managers. In addition, prior to the submission of the ‘ringi’, informal discussions known as nemawashi take place several times among employees to ensure that consensus is reached before a request for a decision is sent to the superiors (Gill and Wong, 1998). Moreover, the ringi sho is a formal document that requires all parties involved to sign prior to the final decision, forming a central part of the traditional decision-making process in Japanese firms (Black and Mendenhall, 1993). In some cases, these traditions find formal expression in joint employee-management committees, which are committees that are developed throughout the organization including the level of the company, department and production facilities, and may typically involve staff across varying hierarchical levels of the organization discussing a wide range of issues which vary from complex organizational goals, production and business challenges to working conditions (Kato and Morishima, 2002). While these processes will be very difficult to recreate in other countries with different formal and informal institutions, it may well be that Japanese firms seek to operate a representative structure that fulfils an equivalent role to these processes. Indeed, some available evidence indicates that Japanese firms operating abroad often have an employee representative structure such as a consultative committee or board, which is independent of, and in some cases an alternative to, unions (Vo and Rowley, 2010). Therefore, we anticipate that Japanese MNCs are more likely to use representative structures that do not entail a formal role for unions.

  • H3: Japanese MNCs are more likely to use non-unionized representative structures such as consultative committees relative to US MNCs.

The US ‘mass production’ model was based on the realization of economies of scale through standardized products in volume. Employees performed highly standardized tasks that did not require advanced skills, making them relatively easy to dispose of, while their performance was closely monitored by supervisors. Improvements in process were achieved not by employees but rather by experts, such as designers and engineers, and consequently there was little emphasis on involving employees in the production process. In contrast, the Japanese ‘lean’ production model was predicated on a principle of continuous improvement and waste minimization. The employment relationship was a long-term one, with workers receiving ongoing training and some workers enjoying explicit guarantees of life-time employment, as we have seen above. The crucial characteristic for our purposes is that process improvement and innovation “were not seen as the domain of experts but rather as part of the ordinary job responsibility of all workers” (Jürgens, 2003: 222).

There is evidence that firms in Japan focus strongly on practices such as direct information sharing with employees and instilling teamwork (Swierczek and Onishi, 2003; Shiraki, 2007), while the extensive use of participatory employment practices in Japanese firms are widely seen as contributing to superior performance by strengthening employee identification with the company and promoting high commitment to production practices that emphasize high quality outputs and have become one of the hallmarks of Japanese management (Kato, 2006; Aoki et al., 2014). Specifically, the purpose of problem-solving teams is to ensure continuous improvement, with such teams identifying a problem area from various categories including cost reduction, productivity and efficacy improvements, quality, workplace safety, and environmental improvement (Ichniowski and Shaw, 1999). Related to this, quality circles are made up of small groups of individuals who voluntarily meet on a regular basis to discuss concerns about product or service quality and operations, and are responsible for providing suggestions to senior management (Gill and Wong, 1998; Pudelko et al., 2013).

On this basis we propose that:

  • H4: Japanese MNCs are more likely to use direct employee involvement practices in their foreign subsidiaries relative to US MNCs.

We now turn to the research methodology that we deployed.

Research methodology

The paper draws on comparative survey data on employment relations practices within multinational companies in four countries (UK, Canada, Spain and Australia). Each survey covered foreign-owned MNCs above a certain size (namely those with at least 500 employees worldwide and at least 100 in the host country). The claims to representativeness are based on extensive cross-checking of multiple sources in establishing population listings and careful checks between the profile of the responding firms and the population in which all the national teams engaged. The surveys contained some innovations, notably in the use of questions relating to multiple occupational groups; where appropriate, we asked questions about practice relating to managers, the largest occupational group (LOG), and a ‘key’ group which we defined as a category of employees who are crucial to the generation and maintenance of a company’s competitive advantage.

The surveys were nationally representative and contained MNCs of a range of nationalities; in this paper we select only those that are Japanese- or American-owned. In total, there were 460 MNCs included in this analysis (see Table 1).

Table 1

Breakdown of Country of Ownership by Country of Survey

Breakdown of Country of Ownership by Country of Survey

-> Voir la liste des tableaux

The surveys were similar but not identical. The mode of delivery of the survey to respondents was different in Canada, where it was carried out through a combination of online, mail and telephone modes, differing from the face-to-face mode used elsewhere. This meant that some questions had to be omitted from the Canadian survey, including those relating to the key group. A further difference was that the surveys were not all in the field at precisely the same time; however, our analysis of this issue indicates that the variables used in this paper were not time sensitive[1].

For each of the four sets of hypotheses, suitable indicators are devised as the dependent variables.

These are:

  • Personal control: The number of key positions in foreign subsidiaries which are filled by nationals of the country of ownership. The question asks how many of the top five management positions are held by nationals from the country of ownership. This question was not asked in the Canadian survey, so comparisons are based on data from the UK, Spanish and Australian surveys. Given the skewed nature of the data (many of the responses are of zero or one, relatively few are of four or five), linear regression models would not fit these data well (ordered models have problems relating to empty cells), so the data are transformed into a binary dependent variable for logistic regression analysis in which any of the top 5 management positions held by parent country national equals 1, and none of the top 5 management positions held by parent country national equals 0.

  • Procedural control: Two variables are used separately to investigate this; first, the presence of an international HR committee; and, second, the presence of an HR information system, both of which are binary variables.

  • Variable pay: Pay and remuneration practices are investigated through a range of dependent variables. The first set of variables were taken from questions which asked separately whether share ownership schemes, profit sharing and share options were offered to managers, the key group or the LOG (i.e., nine questions in total). These questions relating to share ownership, share options and profit sharing were not asked in the Canadian survey so the analysis presented in Table 2c uses data from the Australian, UK and Spanish surveys. The second set of indicators pertain to a question concerning whether the subsidiary operates a system of performance-related pay for each of the three occupational groups. The question of whether the system of performance-related pay was applied to the key group was asked only in the Spanish and Australian surveys.

  • Employee representation: One question in the survey asked whether the subsidiary had a trade union presence. In those firms which did not have a recognized trade union, we asked whether there was an alternative employee representative structure that did not involve unions. The question was not asked in the Spanish survey because of the distinctiveness of the institutional context; specifically, the meaning of being a non-union firm is quite different. So the data are taken from the Australian, UK and Canadian surveys.

  • Direct employee involvement: Two variables were selected to investigate the use of direct employee involvement practices: the presence of problem solving or continuous improvement groups; and the presence of formally designated work teams responsible for organizing work and carrying out tasks. Both of these are binary variables.

The dependent variables defined above are all variables, taking the value of either one or zero, meaning that binary logistic regression is an appropriate technique for analysis. We regressed the dependent variables onto a common set of independent variables (IVs)—the dummy variable for Japanese ownership and controls for sector, size and survey country. Our main variable of interest, the dummy for national ownership, took a value of 1 if the firm was Japanese and 0 if it was US-owned. Controls were selected on the basis of factors that are likely to shape the existence of the different HRM and employment relations practices that we examine. First, the broad industrial sector is used as a binary variable, distinguishing between production (largely manufacturing) and services. The reference category is production. Many studies of employment practice in MNCs (e.g., Purcell et al., 1999) have found strong sectoral effects. Our own measure is a broad one as a high proportion of the MNCs in our surveys operated in either production or services, but within each were diversified into multiple sub-sectors, making a narrower sectoral variable problematic. Second, the size of the subsidiary operations is used, in which four size bands by employment are constructed (100-499, 500-999, 1000-4999 and 5000+), with the smallest as the reference category. This is a standard control in the area of employment relations given the relationship between size and many of the practices we explore, particularly representative structures. Third, we include the survey country as a control given the differences noted above in the institutional contexts of employment relations across the four countries. Australia forms the reference category.

For the regressions reported, a range of tests were carried out to check for statistical significance and goodness of fit. Model and step chi square statistics are reported to indicate the statistical significance of the overall model and the effect of introducing the control variables. Goodness of fit was checked using the Hosmer and Lemeshow test. The standard correlation matrix shows the association between the independent variables and indicates that the independent variables are not highly correlated (see Table 2a). Additional checks were made for collinearity problems using conditions indices and VIF/tolerance tests. The results confirmed that none of the variables violated any of the diagnostic checks for multicollinearity, all of the regressions were found to achieve statistical significance at the 5% level or better, and the Hosmer and Lemeshow tests indicated that regressions fit the data satisfactorily.

Table 2a

Descriptive Statistics for the Independent Variables

Descriptive Statistics for the Independent Variables

N=460

-> Voir la liste des tableaux

The regression tables report both beta coefficients and odds ratios. Odds ratios represent the magnitude of the effect of a change in the independent variable from the reference category. The beta coefficient is the log of the odds ratio. Thus, for the key independent variable—the Japanese dummy variable—odds ratios greater than 1 (or positive beta coefficients) indicate that the practice or feature is more likely in Japanese MNCs.

Findings

Table 2a gives the descriptive statistics for the right-hand side variables; means, standard deviations and correlations. Due to differences in survey design and also differential missing values on particular items, the Ns for the regressions presented below vary from 194 to 452. We present descriptive statistics for the full sample of Japanese and US MNCs (N=460). In addition to the standard correlation matrix showing the association between the independent variables (to check for issues of multicollinearity) we present a correlation matrix for the dependent variables (Table 2b). The rationale is that we suggest that the practices and features of MNCs may be a suite of practices. Our hypotheses are set up to capture our expectations that some of the practices will be more common in Japanese MNCs, and others in US MNCs. The correlation matrix is therefore a way of showing the association between the practices. Because of the differential survey design, these correlations use the pairwise deletion function for missing values.

Table 2b

Descriptive Statistics for the Dependent Variables

Descriptive Statistics for the Dependent Variables

-> Voir la liste des tableaux

The correlations indicate limited support for the suggestion that Japanese and US MNCs operate different bundles of practices, though in practice the distinction is far from clear cut. The correlation matrix shows few significant correlations, indicating that the associations between the variables are not strong. We would expect variables which are associated with Japanese practices (such as personal control) to be negatively correlated with variables associated with US firms (such as international HR committees) if the data were split along ownership lines. However, while the significant correlations are generally in the expected direction, there are many more insignificant correlations indicating that the patterns are not strong. In so far as the count data were able to shed light on the matter, it was found that among Japanese MNCs, few had all or none of the practices, and most had some.

The frequencies of dependent variables are shown in Table 2c. The figures refer to the percentage of Japanese and US MNCs which use a particular practice or feature. For example, the personal control row shows that 71% of Japanese subsidiaries had a senior manager from Japan whereas 39% of US subsidiaries had a US national in a senior management position. Chi square tests were performed to test for significant differences between Japanese and US MNCs and these are reported in the conventional way.

Table 2c

Comparison between Japanese and US MNCs

Comparison between Japanese and US MNCs

-> Voir la liste des tableaux

Control

Our first hypothesis, H1a, proposed that Japanese MNCs are more likely to use personal forms of control in their foreign subsidiaries relative to US MNCs. The bivariate analysis shows that Japanese MNCs have a significantly higher average score on this measure than US MNCs (Table 2c); 71% of Japanese MNCs use personal control compared with 39% of US MNCs. The regression results for personal control are given in Table 3. The significant positive beta coefficient on the nationality dummy variable indicates that Japanese MNCs are more likely to use personal control compared with US MNCs. The odds of a Japanese MNC using personal control are more than 3.5 times greater than for a US MNC. Overall, this model provides strong support for the expectation that Japanese MNCs are more likely than US MNCs to use personal forms of control, so we accept H1a.

Table 3

Personal Control

Personal Control

-> Voir la liste des tableaux

We then tested whether Japanese MNCs are less likely to use procedural forms of control in their foreign subsidiaries relative to US MNCs (H1b). The bivariate analysis shows that both of the measures of procedural control—the presence of an international HR committee (73% versus 35%) and of an HR information system (71% versus 16%) were significantly more common in US than in Japanese MNCs (Table 2c). Regression results are shown in Tables 4a and 4b. For both regressions, significant negative beta coefficients confirm the prediction that Japanese firms are less likely to use procedural forms of control compared with US MNCs. In the case of international committees, the odds of a US firm are over 5.5 times greater than for a Japanese MNC. For HR information systems, the magnitude of the effect is larger still, with the odds being over 16 times greater. The results also indicate some host country effect, with procedural forms of control less common in the UK, compared with the reference category of Australia—significant at the 1% level for HR committees and significant at the 10% level for HR information systems. In sum, H1b is supported.

Table 4a

Procedural Control A

Procedural Control A

-> Voir la liste des tableaux

Table 4b

Procedural Control B

Procedural Control B

-> Voir la liste des tableaux

Variable pay

The next hypothesis (H2) contended that Japanese MNCs are less likely to use performance-related pay practices in their foreign subsidiaries relative to US MNCs. There are two sets of measures of this phenomenon—the different types of remuneration which may be given to employees as well as a measure of performance-related pay. Cross-tabulating the remuneration variables—share ownership schemes, profit sharing and share options—against country of ownership revealed some strong patterns of association (Table 2c). In all cases, the Japanese subsidiaries exhibited a significantly lower take-up than their US counterparts, while in some cases, none of the Japanese subsidiaries in a country used the practice. While this lower incidence is in keeping with our hypothesis, the absence of the practice in any Japanese companies for some of the variables meant there was no variation in the dependent variable and, consequently, regression analysis was problematic. Regressions were run separately for each of the features for each of the employment groups—i.e., nine separate regressions. Due to the lack of variation in the dependent variable for these nine regressions, the models either did not find a final solution or were insignificant at the standard significance level. Therefore, we do not report these models.

Turning to the second set of variables concerning variable pay—whether the subsidiary operates a system of linking an individual’s pay to performance (performance-related pay) for each of the three occupational groups—the descriptive statistics can be found in Table 2c. This shows that the use of variable pay systems is lower in Japanese MNCs compared with US MNCs for managers (80% versus 93%) for the LOG (61% versus 69%), but for the key group variable pay is more commonly used in Japanese MNCs (94% versus 81%).

Regression analysis is reported for the LOG and managerial groups (Table 5). The overall regression model where performance-related pay for the key group was the dependent variable was non-significant and is therefore not included. It is important to note that the question about whether performance-related pay was available to the key group was asked in the Spanish and Australian questionnaires, but not in the UK or Canadian questionnaires. The regression for managers shows a significant negative beta coefficient on the Japanese dummy variable, at the 1% level. For the LOG, the coefficient is also negative, though at the 10% level—it attains marginal significance in other words. The odds for US MNCs using variable pay systems are over 1.5 times greater for the LOG and over 3 times greater for managers.

Table 5

Performance-Related Pay for Managers and LOG

Performance-Related Pay for Managers and LOG

-> Voir la liste des tableaux

Overall, the evidence suggests that Japanese MNCs are less likely to use performance-related pay for the managerial group. For other groups, the story is less clear cut. The descriptive statistics for the different reward systems indicate that Japanese MNCs are less likely to use these forms of reward compared with US MNCs. In sum, the evidence concerning variable pay provides partial support for H2.

Employee representation

The next hypothesis (H3) we tested concerned Japanese MNCs being more likely to use non-unionized representative structures such as consultative committees relative to US MNCs. We began our analysis of this by addressing the presence or absence of trade unions. Overall the incidence of trade unions was shown to be 51% in US MNCs and 53% in Japanese MNC; the difference being non-significant according to the chi square statistic. While we did not hypothesize differences in the presence of union arrangements, we ran a regression anyway on this variable given its importance to the issue more generally. This analysis (not reported in detail here) confirmed the lack of significant difference between Japanese and US MNCs.

To test hypothesis H3, we used the presence of employee representative structures in non-union firms as the dependent variable. For these MNCs without trade union presence, bivariate analysis indicates significantly higher incidence of non-union forms of employee representation among Japanese MNCs (54%) than in US MNCs (34%). Japanese MNCs are significantly more likely than US MNCs to use non-union representative structures, an effect which holds when the controls are introduced to the regression model (see Table 6). The odds of a Japanese MNC having non-union representation are more than 2.5 times that of a US MNC. We can also note that MNCs operating in the UK are also found to be more likely than those operating in Australia to have non-union structures with an odds ratio of just over 2.5. These results lead us to accept H3.

Table 6

Non-Union Representative Structures

Non-Union Representative Structures

-> Voir la liste des tableaux

Direct involvement

The final hypothesis contends that Japanese MNCs are more likely to use direct employee involvement practices in their foreign subsidiaries relative to US MNCs (H4). Two variables were selected to investigate the use of direct employee involvement practices: the presence of problem solving or continuous improvement groups; and the presence of formally designated work teams responsible for organizing work and carrying out tasks. The bivariate associations (Table 2c) indicate non-significant differences on both of these measures. 81% of US MNCs had problem-solving groups compared with 80% of Japanese MNCS; for work teams the respective figures were 69% and 63%.

Regression analyses confirmed the lack of significant difference between Japanese and US MNCs on these measures at standard levels of significance, though for the regression for work teams the Japanese dummy coefficient was found to be marginally significant—i.e., significant at the 10% level (See Tables 7a and 7b). These results also show that problem solving groups are more commonly found in production, while MNCs operating in Spain, the UK, and Canada are all less likely than the reference category of Australia to use formally designated work teams.

Table 7a

Problem-Solving Groups

Problem-Solving Groups

-> Voir la liste des tableaux

Table 7b

Formally designated Teams

Formally designated Teams

-> Voir la liste des tableaux

The findings for the two variables lead to the conclusion that Japanese MNCs are no more likely to use direct employee involvement practices in their foreign subsidiaries relative to US MNCs, so H4 is not supported.

Additional analyses

Looking at the controls across the models on the range of issues, there are some cases where there were significant differences between the host countries. In relation to non-union structures, for instance, those MNCs in Australia were significantly less likely to have such practices compared with those in the UK, while those in Australia were significantly more likely than all of the other national groups to have formally designated work teams. Given these findings, we explored the role of the host country in two ways. First, we split the sample and repeated the analysis by individual country to see if the patterns that we found in the pooled analysis held at country level. The regression models were then applied with three independent variables rather than four since the host country was omitted. The results are strikingly similar to the pooled regressions, with the only change being that in a small number of cases the effect of the ownership dummy fell short of significance in the individual country regressions, something that is probably caused by a smaller N. In other words, the aggregate findings are not masking quite different patterns on a country-by-country basis. Second, in the pooled analysis we constructed interaction terms between the ownership dummy and the country of operation dummies to test whether the impact of Japanese ownership varies systematically across the countries. Their inclusion did not significantly improve the fit of any of the models, and therefore the regressions are reported without the interaction terms. Taken together, these further checks suggest that the effect of Japanese ownership compared with American ownership for the variables we have investigated is broadly similar across host countries.

Discussion and Conclusion

The findings provide support for the central contention that the forms of control and nature of employment practices in the foreign subsidiaries of Japanese MNCs are shaped by the nature of institutions and practices in Japan. More specifically, we have shown that Japanese and US MNCs behave differently in terms of the forms of control that they exercise, with Japanese firms exhibiting a greater tendency to use personal forms of control in their foreign subsidiaries and a lower tendency to use procedural forms of control. In terms of HRM practices, Japanese MNCs are distinctive in relation to pay systems and practices—they are less likely to operate a range of forms of variable or performance-related forms of pay—and are more likely than their US counterparts to adopt non-union representative structures in their non-union subsidiaries. Overall, despite the pressures of globalization in general, and despite the changes in Japan in particular, we can say that there is a clear country of origin effect from Japan over Japanese MNCs, and from the US over American MNCs.

However, the hypotheses were not universally supported. In particular, the hypothesized differences in relation to direct employee involvement were not significant. One explanation for this could be to do with the changes in Japan. We know from other sources that as the economy faltered and became more subject to the influence of foreign actors, some firms moved away from the hitherto taken-for-granted practices of keiretsu linkages and lifetime employment, while ‘pure’ forms of lean production were also amended in certain respects (Jürgens, 2003; Shimizu, 2009; Aoki and Lennerfors, 2013). While this was not a complete transformation in Japan, it did involve a notable number of organizations, many of which were multinational, and as such the commitment to using direct employee involvement in their operations might have weakened. A different explanation for the lack of significant differences between Japanese and US MNCs may be not that we over-estimated the strength of forces towards using direct employee involvement practices in the Japanese firms but rather that we under-estimated this force in US firms. Arguably, the motivation of managers in US MNCs to engage in direct involvement practices lies not in the complementary nature of such practices to a model of production but in their potential role in union avoidance (which is a major characteristic of American-style HRM, see Kaufman, 2014). Indeed, it is quite possible that some of the US firms in our sample have used such practices as a way of trying to persuade employees that their interests are being served by management and there is no need to join a union (e.g., Jacoby, 2005; Almond and Ferner, 2006). If this is the case, then similarity in practice does not mean there are not institutional effects; rather, different institutional pressures can lead to a similar prevalence of practices. Moreover, this could also be explained by American MNCs adopting ‘traditional’ Japanese management practices such as quality control groups and work teams to compete in the marketplace. We discuss these points in more detail below.

Our argument concerning the existence of some institutional influences on the subsidiaries of Japanese MNCs contrasts with Pudelko and Harzing’s (2007) interpretation of a comparative study of MNCs headquartered in the US, Japan and Germany, which claimed “strong convergence” towards the US model of HRM practice in general and that Japanese MNCs have a “very strong tendency to abandon their traditional practices” (2007: 549-550). We contend that the difference between our findings and theirs is due to subtle differences in the methodology. Pudelko and Harzing (2007) sought information concerning the entire workforce, which may lack precision as respondents provide information that is an ‘average’ of the whole workforce, masking differences between groups, or the respondent may focus on a particular group which is not representative of the workforce. Asking about particular occupational groups, as our study did, minimizes these problems. In addition, Pudelko and Harzing (2007) asked about broad characteristics to which practices tend (e.g., the difference between “primarily material incentives” versus “a mix of material and immaterial incentives”), which contains the danger of capturing what senior HR staff aspire to rather than what is actually in operation and experienced by employees, as the authors themselves accept (2007: 552). We sought data on the existence of particular practices in operation, a more concrete, less perceptual measure. Our findings are more consistent with the picture of ‘patterned innovation’ and selective adoption of American practices of firms within Japan (Vogel, 2006).

At different points we have noted the changes affecting the Japanese business system in recent times and the greater diversity in firm strategies that is one consequence of this. It is logical to expect that this will lead to greater diversity in the HRM practices of Japanese MNCs and possible approaches to HRM that combine different types of practices within each firm. Our own analysis indicates that very few of the Japanese firms in our sample used none of the practices that we explored; equally none used all of them; the vast majority used some but not others. This suggests that, while a small number of firms may be changing markedly in the US direction, and a small number are sticking closely to traditional Japanese practices, the majority are engaging in US practices rather selectively. A full analysis of the diversity of strategies of Japanese MNCs and the variety of ways in which this selective change may be implemented is left to future research.

This raises some important theoretical implications for the HRM convergence-divergence debate and the complexity of HRM systems. First, it is clear that Brewster, Sparrow and Vernon’s (2007) convergence-divergence typology, albeit a useful frame, does not take into account the complexity of the simultaneous forces shaping the adoption of MNC HRM practices—that is—the interaction between country of origin institutions and traditions of HRM, local cultural and institutional practices, global competitive market forces and other isomorphic pressures to adopt universal best practice HRM. Our findings demonstrate that the majority of Japanese organizations in our sample use a mixture of traditional Japanese HRM practices (possibly the least culturally specific and easily transferable), as well as some American HRM practices. We suggest that there is paradoxically some evidence of directional convergence (i.e., continuous improvement groups and work teams) and divergence of HRM practices (i.e., managerial control practices) of Japanese and US MNCs. Interestingly, US MNCs are adopting what have arguably been popularized Japanese management practices such as continuous improvement groups and work teams. Second, there is some evidence of the US model of HRM being adapted by Japanese firms—that is—incorporating the universal aspects of HRM with their own uniquely Japanese management approaches. We may be witnessing the development of hybrid HRM systems which incorporate American-style HRM practices, albeit the most universally accepted practices in Japanese management systems. The necessity for Japanese MNCs to adapt to global competition has driven their adoption of some US HRM practices (Pudelko and Tanzer, 2013; Kaufman, 2016). In line with Kaufman (2016), this points to the notion of ‘converging divergences’ in that Japanese MNCs are adopting a mixture of traditional Japanese and US HRM practices. Moreover, competitive forces have also led US MNCs to adopt some traditionally Japanese management practices. Both Japanese and US MNCs are adopting the most universal aspects of each other’s management practices and integrating them into their own unique systems of management in response to global market forces.

One of the limitations of the paper—and the reason why we cannot analyse such variation and selective change further ourselves—is that the sample of US MNCs is much greater than the sample of Japanese MNCs, restricting the scope for additional analysis in which the data could be interrogated in further ways. Despite this, however, the hypothesized differences are largely supported. A further limitation is that the countries cover a certain type of developed market economy. While there are notable institutional differences between the four countries, we do not cover a northern European country with labour market institutions that have extensive coverage, such as Germany or the Nordic countries. Equally, we do not cover emerging or developing economies. This is an area for future research.