Asia Pacific Journal of Management

, Volume 27, Issue 3, pp 423–443

Organisational determinants of employee turnover for multinational companies in Asia


DOI: 10.1007/s10490-009-9159-y

Cite this article as:
Zheng, C. & Lamond, D. Asia Pac J Manag (2010) 27: 423. doi:10.1007/s10490-009-9159-y


High employee turnover rates among multinational companies (MNCs) in Asia have become an organisational issue, which cannot be sufficiently addressed at the individual level. In this paper, we examine the issue of employee turnover at the organisational level. A group of organisational variables (e.g., training, size, age, industry, percentage of expatriate managers and headquarters’ national base) were tested, using a sample of 529 MNCs in six Asian countries. The standard multiple regressions show that training, size, the length of operation in local subsidiary and nature of industry are significantly related to turnover. An effect of the percentage of expatriate managers present in the local subsidiary on employee turnover appears to be moderate. These results fill a research gap by identifying organisational variables (as opposed to individual characteristics) and contribute to better explanation of employee turnover at firm level. Implications to MNCs in the greater Chinese region and Asia are discussed.


Turnover Training Multinational companies (MNCs) Expatriate management Chinese Asia 

The remarkable and resilient growth of the Chinese economy has generated a high rate of employee turnover in the Asia Pacific region. The rapid expansion of international trade and foreign investment associated with this giant economy has heightened the demands for highly skilled managers and professionals particularly among many Asia economies. There have been numerous reports of the continuing trend of high employee turnover in one area or another, as a consequence of the increasing volume of trade and investment as well as a greater degree of freedom in skilled labour migration in the region (see Asia Times, 2006; AsiaNews.IT, 2006; IOM, 2005; Li, 2008; The Straits Times, 1999; Wozniak, 2003). Others suggest that decisions to stay and leave are entirely individual and more or less also culturally bound. An opinion poll from the HR professionals in Hong Kong, Malaysia and Singapore shows an unanimous view that the rampant “job hopping in these Asian countries had, in fact, become a culture” (Khatri, Chong, & Budhwar, 2001: 54). If there is a distinct Asian or ethnic Chinese community culture predominant in the business world in the greater China and Asia region that is related to employee turnover, any multinational company (MNC) doing business in the region will need to know it and master it.

High levels of employee turnover are viewed as an issue as it is costly for organisations to search, select, train and replace leavers (Mobley, 1982; Price, 1977; Staw, 1980) and turnover of high performers, especially those managers and professionals, is negatively related to overall organisational effectiveness and firm success (Holtom, Mitchell, Lee, & Inderrieden, 2005; Mitchell, Holtom, Lee, Sablynski, & Erez, 2001). Nevertheless, the organisational level of analysis on the causes and effects of employee turnover is quite limited, as the large volume of literature in the past tends to focus on explaining employee turnover at the individual level (see reviews by Griffeth, Hom, & Gaertner, 2000; Hom & Griffeth, 1995; Mowday, Porter, & Steers, 1982; Peterson, 2004). The theory of voluntary employee turnover first proposed by March and Simon (1958) argues that employee turnover results from the individual’s perceptions about the perceived desirability of movement and ease of movement. Over time, the perceived desirability of movement has been proxied by job satisfaction (Jackofsky & Peters, 1983). The perceived ease of movement has been equated essentially with the number of perceived job alternatives (Lee & Mitchell, 1994). Since then, most empirical studies have focused on these two constructs in explaining employee turnover. However, a meta-analysis by Steel and Griffeth (1989) concluded that measures of perceived alternatives and turnover are weakly related, as these two variables explain only 13% of the variance in turnover (R2 = 0.13). A longitudinal study conducted by Wong, Hui, and Law (1995) also confirmed that the effects of job satisfaction on the attitudinal antecedents of turnover are negligible. Better work attitudes and job satisfaction were reported to play only a relatively small role overall in employee retention (e.g., Griffeth et al., 2000; Hom & Griffeth, 1995; Hom & Kinicki, 2001; Lee & Rwigema, 2005; Lee, Mitchell, Holtom, McDaniel, & Hill, 1999; Mitchell et al., 2001). Therefore, it might be more enriching to look beyond factors such as job satisfaction and related constructs, as well as job alternatives, and search for other variables that may better explain organisational turnover rates.

Barrick and Zimmerman (2005) searched 1,500 studies in the past 50 years on employee turnover. They found that most of the studies focused on understanding why employees leave. There were few studies at the organisational level of analysis to investigate whether and how employers could reduce turnover. Shaw, Delery, Jenkins, and Gupta (1998) share a similar view, claiming that the research of turnover “leaves a critical gap—the determinants of turnover at the organisational level” (Shaw et al., 1998: 511).

The importance of studying employee turnover at the organisational level was first stressed by Terborg and Lee (1984) who questioned the existence of homology between functional relations found at the individual level of analysis, and those that might exist at the organisational level of analysis (Terborg & Lee, 1984: 794). Terborg and Lee (1984) called for rigorous research to explicitly consider organisational level variables as predictors of turnover. Despite such a call, and 50 years of studying the issue of employee turnover, an organisational model explaining turnover is “yet to be fully developed” (Peterson, 2004: 210).

The purpose of this paper is two-fold. First, we intend to fill the research gap by using Peterson’s (2004) human resource development (HRD) model and investigating some organisational variables (as opposed to individual employee characteristics), which may potentially contribute to developing an organisational model of employee turnover. Second, the results of this study may better explain the phenomenon of rampant employee turnover in Asia. To achieve our aims, we use a sample of 529 MNCs operating in six Asian countries, namely, Indonesia, Malaysia, Philippines, Singapore, Taiwan and Thailand, for this study. We believe that these study outcomes have a significant implication for Asia Pacific management research, in addition to their particular relevance to those MNCs operating in the greater China region and Asian nations whereby businesses are predominantly controlled by ethnic Chinese communities.

Literature review and development of hypotheses

Earlier employee turnover models have overwhelmingly focused on the individual rather than institutional characteristics that address the role of the organisation in contributing to the individual employee’s decision to stay or leave. The key merit of these past models is that they provide us with “incremental knowledge” (Lee et al., 1999: 450) on multidimensional causes and consequences of turnover over time (for a comprehensive review of the past models on employee turnover at the individual level, see Peterson, 2004). Some of this knowledge has now led to more promising research, examining the role of the organisation in reducing the turnover rate (Peterson, 2004; Shaw et al., 1998). For the purpose of this paper, four groups of models relating to managing employee turnover at the organisational level are assessed.

First is the process model grouping, which emphasises the sequential steps of the process leading to employee quitting (see Dougherty, Bluedorn, & Keon, 1985; Hom & Griffeth, 1995; Mobley, 1977; Tett & Meyer, 1993). The model tends to explain the antecedents of job satisfaction and organisational commitment as “job scope, role stress, group cohesion” and “procedural justice, employment security, training and job investments” (Peterson, 2004: 213). Rhoades, Eisenberger, and Armeli (2001) and Allen, Shore, and Griffeth (2003) further develop these antecedents to include the organisational level of human resource (HR) practices, such as organisational rewards, employee participation in decision-making, training and career development opportunities, and overall organisational supports. These are particularly relevant to the organisational level of analysis on turnover in this study.

The second group comprises the socialisation models developed by Felderman (1976), Fisher (1986) and Allen and Meyer (1990). These models tend to associate “individual characteristics with the organisational process of socialisation” (Peterson, 2004: 214) and emphasise organisational culture as a predictor of how well employees can adjust to and function in the organisational environment. The key argument is that if employees fit in with the organisational culture and function in a satisfactory manner, they are likely to stay (Sheridan, 1992; Taormina, 1999). Our main concern of the model is that it does not extend to incorporate the interaction between the national culture and corporate culture and how this may affect employees. MNCs tend to operate in different nations, hence developing quite distinctive corporate cultures based on their adjustment to different national cultures and the degree of localisation. It is the interaction between the national and corporate culture that may be experienced by expatriates and local employees of MNCs, which could have had an underlying effect on employee turnover (Lueke & Svyantek, 2000).

The third grouping is that of expanded models, which are essentially process models, but extended to cover multidimensional organisational factors (Peterson, 2004). Even though these models still discuss the development of individual employees’ desires to stay with the organisation, they focus more explicitly on the organisational perspective (Mowday et al., 1982: 203). These models stress some confounding variables such as industry nature, size, employee types, cost control and time, which could determine the strongest linkage, and ultimately affect the turnover decision. The expanded models are relevant to the current study as “the role of the organisation appears to be more prominent than in most other models” (Peterson, 2004: 216).

The last employee turnover model grouping is defined by Holtom et al. (2005) as a “shocks” model. This model was developed by Lee and Mitchell (1994) and further refined by Lee et al. (1999) to look beyond job dissatisfaction as a common precipitation for individuals to resign, but to examine some unexpected events as explanations for employee turnover. “Shocks,” or unexpected events, are those to do with alternative job offers, job transfers, acquisition/merger of a company, and changes in marital status or spouse’s work. Use of “shocks” to explain turnover is relevant to the current study, as “shocks” are believed to be potentially manageable at the organisational level (Holtom et al., 2005). In addition, taking unexpected events such as family changes and job transfers as reasons to leave the organisation is common, and regarded as culturally acceptable in the Asia context (Khatri et al., 2001). Hence, the examination of the shocks, in particular inter-company transfer of staff and its effects on employee turnover is meaningful in this aspect.

The above-discussed model groupings tend to stop at organisational commitment and do not carry the process through to employee turnover. Nor do they address what employers can do to control turnover. A turnover model recently developed by Peterson (2004) deals more directly with the turnover issue by emphasising the role of organisational HRD. The model is relevant as the current study also focuses on the extent of organisational level of commitment to employee training and how it would affect employee turnover among the MNCs examined.

The focus of the Peterson (2004) model is on internal organisational factors over which the organisation has substantial control, and in which HRD can play a critical role. The argument for focussing on HRD in controlling turnover has its own merits. First, as noted in the earlier discussion, it is essential to go beyond individual characteristics and examine organisational factors which may better explain turnover. Peterson’s (2004) conceptual model is not yet fully tested, but it may provide a framework to clarify the understanding about the importance of the organisation in creating a learning environment with employees in mind, where employees perceive they are valued and, in return, are more committed to staying.

Second, a number of empirical studies suggest that training and employee skill development, together with other HR practices, lead to higher employee commitment and lower labour turnover rates (e.g., Arthur, 1994; Carley, 1992; Glance, Hogg, & Huberman, 1997; Guthrie, 2001; Huselid, 1995; Shaw et al., 1998). It is logical, therefore, to test from this HRD perspective to see whether it explains a greater percentage of the variance in turnover.

Third, it is widely supported that human capital contributes to high productivity, quality performance and national economic growth; and that the most important investments in human capital are education and training (Booth & Snower, 1996; Lucas, 1988; Schultz, 1961). Effective employee training provides an organisation with a unique and differentiating position to improve its standard and quality of service to customers, resulting in continual innovation and increased productivity and profitability (Matthew, 2002; Taylor & Davies, 2004). It is, therefore, in the organisation’s interest to train employees.

Various discussions presented in the empirical studies suggest a positive relationship between organisational training and employee retention (proxied by low employee turnover rate), notwithstanding that synergistic HR functions also influence the organisational level of turnover. For example, Carley (1992) used organisational learning to predict firm level of personnel turnover over time. Arthur’s (1994) work with steel mill managers concluded that a “commitment” driven HR system (e.g., participative decision-making, training and skill development, social activities, benefits and bonuses) tends to result in a lower employee turnover rate and higher productivity. Huselid (1995) claimed that employee turnover, together with other organisational performance indicators, are significantly influenced by employee skills and organisational structure; he argued that HRM practices do not and cannot directly influence organisational performance, rather, they can indirectly influence the quality and behaviour of employees. Glance et al. (1997) examined the dynamic interplay of four organisational variables—the extent of training, turnover rate, firm size and productivity. The results show that those organisations that continuously train employees improve productivity over time. Shaw et al. (1998) studied a single American trucking industry and found that organisational variables (such as size, age of establishment, union status and employees’ characteristics) accounted for 27% of the variance in explaining the voluntary turnover rate, but only three HR constructs (pay, selection and training) were significantly related to turnover.

The positive relationship between training and turnover is further supported by other reports, which indicate that the workers with less training tend to change jobs more often than those with training (e.g., Hequet, 1993; OECD, 1993). A recent study by Benson (2006) confirms the direct link between employee development and low turnover rate; on-the-job training in particular is positively related to organisational commitment and negatively related to intention to leave. Hence, our key hypothesis in this study is:

Hypothesis 1 Training is negatively related to employee turnover. The more training organisations provide the lower will be the employee turnover rate.

Price (1977) posited that turnover rates are likely to be higher in organisations that are centralised and lower in organisations that encourage frequent communication and participation. In addition, supervisory style, such as more or less consideration, or authoritarianism, could affect turnover rate (Mobley, 1982). In the context of MNCs, we have witnessed in the past decades the increasing use of expatriates as supervisors in local subsidiaries to control, coordinate and contribute to achieving MNCs’ global centralised strategy (Harzing, 2001a). From the turnover models discussed earlier, it is likely that more control from headquarters would reduce the sense of attachment felt by local employees (see for example the process models discussed by Allen et al., 2003; Rhoades et al., 2001). Hence, it is likely that the more dominant role played by expatriates to control local subsidiaries, the more likely threat on local employees (Harzing, 2001b). Consequently there might be a higher level of intention to quit by local employees, as they perceive less opportunities for their career advancement (Reiche, 2007), often with expatriates occupying the top local managerial positions. However, the effects of expatriates on turnover can be varied, depending on the different roles expatriates exercised in the local subsidiaries. Harzing (2001b) suggested that there has been a tendency for MNCs to exercise more direct expatriate control, rather than indirect types of control through socialisation and informal communication (Harzing, 2001b: 369); and that MNCs’ ability to socialise and build informal network tends to enhance with years of overseas operation (Harzing, 2001b: 372). Given a presumptuously higher level of local responsiveness in the context of MNCs operating in Asia, expatriate slack in socialisation and building informal networks may result in the so-called “Penrose effect,” leading to inferior expatriate performance (Goerzen & Beamish, 2007), which could have effect on local employee turnover. Hence we argue that excess expatriates may be linearly or inversely related to local employee turnover, subject to the MNC’s international experience. In addition, Harzing (2001b) found that socialisation and informal communication appeared quite important roles for expatriates even among very young subsidiaries. This suggests a likely U-shaped relationship between expatriates and turnover in different stages of MNC development in local subsidiary.1 Therefore we hypothesise:

Hypothesis 2 The proportion of expatriate workforce present in the local subsidiary is associated with the turnover rate for local employees. The direction of association depends on multinational company’s years of operation in local subsidiary.

Intra-company staff transfers have often been regarded as methods for developing global mindsets and helping employees achieve successful “boundaryless careers” (Dowling & Welch, 2004; Evans, Pucik, & Barsoux, 2002). Less often was career shift from border to border being assessed based on personal preferences (Arthur, Khapova, & Wilderom, 2005). Arthur et al. (2005) postulated to take into account a person’s ability to accommodate work and family or other issues of life work balance in measuring the success of a borderless career. This implies that good organisational intention of intra-company staff transfer may result in bad drop-out of individual employees who may see the opportunity as a threat to balancing their work and family life. In recent time, Holtom et al. (2005) also proposed “shocks” to explain potential employee turnover. The “shocks” or unexpected events may cover alternative job offers or job transfer. There appears a degree of association between intra-company job transfer and employee turnover despite the potential career prospects (Holtom et al., 2005; Welch, 2003). In addition, inability for intra-company transferred managers to successfully manage their careers from location to location may have an indirect effect on their local employees’ turnover intention. Accordingly, we hypothesise:

Hypothesis 3 The higher proportion of intra-company transferred managers in the local subsidiary, the higher will be the employee turnover rate.

Similar to the previous research on examining employee turnover at the organisational level (Khatri et al., 2001; Mowday et al., 1982; Shaw et al., 1998), we also include some moderating factors such as size and industry type in analysis. Relating to MNCs operating in Asia, their experience in the host country and source of headquarter would have impact on both expatriate and local employees’ turnover intention (Reiche, 2007).

Based on the literature, organisational size has been investigated as a predictor of voluntary turnover rates (Mobley, 1982; Shaw et al., 1998; Terborg & Lee, 1984). Mobley states that conceptually, large organisations might have low turnover rates “because of greater internal mobility opportunities, more sophisticated human resource management processes, more competitive compensation systems, and more activities devoted to the management of turnover” (Mobley, 1982: 92). But Terborg and Lee argue that large organisations can experience high turnover rates because of “problems with communication, poor group cohesion, impersonalisation, and bureaucratisation” (Terborg & Lee, 1984: 795). Hence, size is related to turnover rate, but the direction of the relation cannot be determined in advance.

In the process of industrialisation across most Asian countries, there has been a general shift from developing the labour-intensive manufacturing sector to the capital-intensive service sector (ADB, 2005). It is possible that a higher unemployment rate would occur in the manufacturing sector than in the service sector where labour is likely to be in high demand. The service sector could have experienced a high labour turnover rate as a result of high demand. In addition, generic differences between service and manufacturing firms are usually referred to in terms of their intangibility, inseparability, heterogeneity and perishability (Bradley, 2002). Those working in the service sector, which has a high index of intangibility, inseparability and heterogeneity, tend to also possess more transferable skills, compared to those working in the manufacturing sector, who are often required to have specific technical skills. It is anticipated that employees in the service industry can be more mobile, hence having a greater effect on turnover rate than those in the manufacturing sector.

With regard to expatriates, it is believed that the morale of local employees would drop greatly when expatriates fail to meet performance expectations (Dowling & Welch, 2004). Often the earlier return of expatriates from assignments leads to the dispersion of local employees. Expatriates tend to stay and perform better in their assignments if they have strong social support and business networks within the local subsidiaries. However, strong social support and extensive business networks can only gradually be built, while multinational firms establish their roots deeper and longer in local markets. As argued by Cavusgil, Zou, and Naidu (1993), firms with longer international experience tend to understand the subtle characteristics of the local markets and are able to use and share cumulative knowledge to improve organisational performance, which contributes to higher performance by expatriates and local employees. And if we use Mowday et al.’s (1982) conceptualisation of “the organisational level of turnover,” it is all about the employee–organisation linkage (Mowday et al., 1982). The linkage tends to be strong when employees feel attached to their organisations, and when they perceive more of their individual opportunities to grow as their organisations grow and expand. In the context of MNCs operating in Asia and worldwide, employees, both expatriates and local employees, would be more likely to stay voluntarily if they see their firms sustain their operation in local subsidiaries for a long period of time (Reiche, 2007).

The HR philosophy and practices by MNCs whose headquarters are located in the Western economies could be different from those developed in Asia and particularly those large family businesses controlled by ethnic Chinese in the Asian economies (Budhwar, 2004; Nankervis, Chatterjee, & Coffey, 2006). Consequently, the impact of these HR policies and practices on turnover could be dissimilar as well. It is likely that those MNCs developed by Asians with dominant Confucian influences would have a different set of values and HR practices, compared to those MNCs from a non-Asia and non-Confucian-influenced world, e.g., Europe, America and Oceania countries. Prior research tends to support the notion that Japanese, Taiwanese and domestically Chinese-owned businesses, some of them largely developed by closed family members, are more likely to emphasise employee commitment (e.g., Kuwahara, 1993; Lin, 1998; Zheng, Morrison, & O’Neill, 2006). Hence, employee turnover rates among the Asian-owned MNCs might be expected to be more stable than those non-Asian- or non-Chinese-owned companies.

Research methods


The data in this paper were sourced from a multinational research project under the auspices of the Asia Pacific Economic Cooperation (APEC; see Stahl & Zheng, 2002). Researchers from six Asian countries participated in the research project. A five-page mailed survey questionnaire was sent to the MNCs as identified by each of the country researchers. A total of 529 companies returned the surveys, which were completed by CEOs, financial controllers and HR managers in the local subsidiaries of MNCs.

The profile of the firms responding to the survey are shown in Table 1. About 21% of the sample were firms operating in Singapore, 19% in Taiwan and Malaysia, 15% in the Philippines, 14% in Thailand, and 12% located in Indonesia. Firms were more or less evenly split between the manufacturing industry (47%) and the service industry (53%). A large proportion of the Asian-bred MNCs were represented in the sample (52%). These Asian companies were predominantly owned by Japan (139 companies, 51%), seconded by Singapore (45 companies, 16%), followed by Hong Kong (24 companies), Taiwan (17) and South Korea (8) and two from Mainland China. Twenty-four MNCs were bred in the Philippines, five in Thailand, and three in Indonesia and Malaysia respectively. Also although our data did not distinguish whether these companies were owned by indigenous Filipinos, Thai or Malays, we speculate that the MNCs developed in these South East countries were controlled largely by ethnic Chinese. UNCTAD (2005) listed the top 50 transnational companies from developing South East nations, 95% of which were owned by ethnic Chinese (see also Yeung, 2006).
Table 1

Survey profile—MNCs operating in Asia.


Manufacture MNCs (%)

Service MNCs (%)

Asian-owned MNCs (%)

Non-Asian- owned MNCs (%)

Total %


26 (43%)

35 (57%)

26 (43%)

35 (57%)

61 (12%)


34 (34%)

66 (66%)

23 (23%)

77 (77%)

100 (19%)


50 (62%)

31 (38%)

51 (63%)

30 (37%)

81 (15%)


44 (40%)

67 (60%)

77 (69%)

34 (31%)

111 (21%)


45 (45%)

55 (55%)

54 (54%)

46 (46%)

100 (19%)


49 (64%)

27 (36%)

43 (57%)

33 (43%)

76 (14%)


248 (47%)

281 (53%)

274 (52%)

255 (48%)

529 (100%)

Measurement of dependent and independent variables

Employee turnover

Several methods were proposed by Price (1977) to measure employee turnover, including accession, separation, stability, survival and wastage rates. On the whole, however, it seems that the most prevalent approach in organisations to turnover measurement is the separation rate (Bennett, Blum, Long, & Roman, 1993; Mobley, 1982; Mowday et al., 1982; Steel, Shane, & Griffeth, 1990). The separation rate (or attrition index) is expressed as the number of members who left during the period divided by the average number of members during the period (Price, 1977). This measure reflects a general tendency to leave the organisation regardless of employment status. As such, it perhaps provides the broadest indication of a turnover problem (Mowday et al., 1982: 110).

Only voluntary employee turnover was used in this study, based on two reasons. First is that involuntary turnover due to retirement and death is essentially uncontrollable, and is therefore meaningless to evaluate at the organisational level. Second, where involuntary turnover is the result of the employer’s decision to terminate the employment relationship, for example to replace poor performers (Mobley, 1982; Shaw et al., 1998), involuntary turnover will not be a concern to organisations. It is voluntary turnover that has been a “salient managerial issue” employers need to address at the organisational level (Lee & Mitchell, 1994: 51; see also Barrick & Zimmerman, 2005; Breukelen, Vlist, & Steensma, 2004; Mitchell et al., 2001 for similar arguments).

To measure voluntary turnover, we adopted Steel et al.’s (1990) approach and asked respondents to indicate “how many of the following categories of employees have left voluntarily from your local company in the past 12 months?” We categorised employees according to the International Labour Organisation’s (ILO, 1998/2000) international standard classification of occupation (ISCO). The employee types included managers, professional staff, clerical, service, sales and production in the local subsidiary. The total number of employee categories in the local subsidiary who had left voluntarily in the previous 12 months was summed and divided by the average number of staff, as expressed in the formula below:
$${\text{Total}}_{{{\text{employee turnover}}}} {\text{ = }}{{\text{Total}}_{{{\text{all employees left}}}} {\text{ }}} \mathord{\left/{\vphantom {{{\text{Total}}_{{{\text{all employees left}}}} {\text{ }}} {{\text{ }}{\left[ {{{\left( {{\text{Total all employees left + }}{\sum {{\text{Total employees}}} }} \right)}} \mathord{\left/{\vphantom {{{\left( {{\text{Total all employees left + }}{\sum {{\text{Total employees}}} }} \right)}} {\text{2}}}} \right.\kern-\nulldelimiterspace} {\text{2}}} \right]}}}} \right.\kern-\nulldelimiterspace} {{\text{ }}{\left[ {{{\left( {{\text{Total all employees left + }}{\sum {{\text{Total employees}}} }} \right)}} \mathord{\left/{\vphantom {{{\left( {{\text{Total all employees left + }}{\sum {{\text{Total employees}}} }} \right)}} {\text{2}}}} \right.\kern-\nulldelimiterspace} {\text{2}}} \right]}}$$

The number was used in the subsequent analysis as a dependent variable.


Previous studies have asked for the respondents’ general perception of training in their firms (e.g., Arthur, 1994) or for respondents to indicate the number of hours spent on training (e.g., Huselid, 1995; MacDuffie, 1995; Shaw et al., 1998). As such, the investment of training specifically in dollar terms was not quantified and valued. Recently, a number of authors (e.g., Park, Gardner, & Wright 2004; Ulrich & Brockbank, 2005) called for developing HR capabilities in quantifying and evaluating each of the HR functions, including training. A similar approach was used by Ng and Siu (2004) in testing the impact of training on firm performance in China. In addition, employee value propositions (EVPs) could affect the intention of turnover (Kochaniski & Redlard, 2001; Lawler, 2005; Ulrich & Brockbank, 2005); that is the degree of investment in employees’ personal and professional development which may determine the level of commitment.2 Therefore, in this study, we used dollars spent to measure the extent of training. The question, “What is the total amount (in US$) your company spent on all employee training in the past 12 months?” was asked to obtain the scale of training investment. To avoid the confounding effect of firm size on the training budget (the larger organisations may have a higher training budget because there are more people to be trained), we used the average training expenditure per person as an independent variable to measure the relationship between training and turnover.

Representation of expatriates in local subsidiary

The respondents were asked to answer the questions “how many expatriates do you have in your local firm?” and “how many staff were transferred internally during the past 12 months?” Expatriates are those staff, especially managerial and professional staff, sent from firms’ headquarters, and intra-company transferred staff from location to location. Again, to overcome the confounding effect of the firm size, we used standardised percentages of expatriates and intra-company transferred managers in the subsequent analysis.

Control variables

Organisational size is measured by a total number of employees and age measured by the years of MNC establishment in local subsidiary, similar to the measurement used by Shaw et al. (1998). Most of the MNCs operating in Asia tend to either manufacture goods with cheap labour and low costs, or provide IT, finance and other business services. Manufacturing and service industries were coded 1 and 0 respectively.

From the survey, it is apparent that there were a large number of Asian MNCs from some more advanced Asian economies such as Japan, Korea, Singapore and Taiwan. Compared to those companies headquartered in the non-Asian world, it is likely that Asian-owned MNCs would have different set of values and HR practices, which could impact turnover. Hence we categorised the MNCs surveyed into Asian and non-Asian parent source. Asian MNCs were coded as 1 and non-Asian MNCs were coded as 0.

Validation of data

We sought to determine the face validity and discriminant validity of the data before running a standard multiple regression analysis to examine the relationship between turnover and selected organisational variables, in order to avoid the issue of measurement error (Gravetter & Wallnau, 2000; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Face validity refers to the extent to which an empirical measure adequately reflects the meaning of the construct under consideration (Churchill, 1979). The first step in the measurement development process is to define the constructs under investigation in this study, and to specify their domain. The relevant literature was reviewed to gain insight in terms of the definition and the domain of the constructs. In addition, the questionnaire was shown to six researchers across member economies to obtain further feedback. Modifications to the items were made on the basis of the comments from the researchers to enhance the face validity.

Discriminant validity shows that the measure is measuring the right construct. It assesses the degree to which a construct is unique and not a manifestation of another measure (Churchill, 1979). Discriminant validity is achieved when the intercorrelations between the constructs are not too high. It is also acknowledged that, since all of the countries in this study were in Asia, there remains the issue of differential scale response bias (Fischer, 2004). To overcome this issue, we extracted the descriptive statistics of all variables (see Table 2) and saved the standardised variables, using Z scores for the subsequent regression analysis. We then analysed the interrelations between the standardised constructs, using bivariate correlation (Fischer, 2004). Table 3 shows that none of the correlations have a value of more than 0.4, indicating that no pair of constructs is so intercorrelated that they could reflect the same construct. Consequently, discriminant validity of all constructs is supported. The highest coefficient is between the percentage of expatriate and intra-company transferred managers (coefficient = 0.355, p < 0.001). This is understandable because companies with more expatriates are more likely to encourage the transfer of managers from one location to another for both job enhancement and enlargement purposes, especially in relation to international assignments (Dowling & Welch, 2004).
Table 2

Descriptive statistics.











Std. Error


Dependent variable

Employee turnover rate







Independent variables

Size of company







Percent of expatriate managers







Percent of intra-company transferred managers







Years in local subsidiary







Industry (manufacturing vs. service)







Average training expenditure per employee







Company parent source—Asia







Valid N (listwise)


Table 3

Correlation using Z score.


Employee turnover







Years in local subsidiary



Size of company




% of expatriate managers





% of intra-company transferred managers






Average training expenditure







Asian-based headquarter
















aCorrelation is significant at the 0.05 level.

bCorrelation is significant at the 0.01 level.


Table 4 shows that, in general, the organisational predictors, such as training, percentage of expatriate and intra-company transferred staff, firm size, age of operation in local subsidiary, industry type and parent source, accounted for a significant amount of the variance in turnover (R = 0.627, R2adj = 0.322, F = 7.280, df = 12, 517, p < 0.001). This is much higher than the prior studies using individual job satisfaction (R2adj = 0.13; Steel & Griffeth, 1989). The results demonstrate the importance of using organisational predictors in explaining employee turnover, at least in the context of MNCs operating in the Asian economies as surveyed in the current study.
Table 4

Results of multiple regression analysis—organisational variables determining employee turnover.


Standardised beta value




Employee turnover

Average training expenditure




Not supported

% of expatriate managers





% of intra-company transferred managers




Not supported

Asian-owned companies





Years in local subsidiary





Size of company










Country comparison




























Overall model fit

R = 0.627, R2adj = 0.322, F = 7.280, df = 12, 515, p < 0.001

Hypothesis 1 stating that training is negatively related to employee turnover is not supported. In fact, companies’ average training expenditure is shown to be positively related to employee turnover (beta = 0.086, p < 0.05). In other words, organisations providing more training experience a higher level of employee turnover. This is the opposite of the findings from the earlier studies (see Ahmad & Bakar, 2003; Arthur, 1994; Huselid, 1995; Shaw et al., 1998), but may be explained in terms of the developmental shifts taking place in the Asian economies within which the companies are located—additional training makes the employees more employable and so more attractive to other employers.

Hypothesis 2 which suggests the association between expatriate proportion and local employee turnover rate is supported. The result shows that the percentage of expatriate managers explains about 8% of variance relating to local employee turnover (beta = 0.084, p < 0.1). In contrast, the result does not support Hypothesis 3; the level of intra-company staff transfers appears to have no significant impact on local employee turnover rate.

Control variables (e.g., size, years of operation, industry and parent source) may have confounded the relationship between company training and turnover, and between expatriation and local employee turnover. In the context of MNCs operating in the six Asian countries, our results show that there is a reverse relationship between turnover and years in local subsidiary (beta = −0.113, p < 0.01). It seems that the longer the companies have experienced in the local subsidiary, the better they managed the retention of their staff (Cavusgil et al., 1993; Harzing, 2001b; Reiche, 2007). Size also matters. Bigger firm size leads to lower employee turnover rate (beta = 0.091, p < 0.05). The results support Mobley’s (1982) argument for the higher capability of larger firms to control their employee turnover through more competitive compensation and other HRM instruments.

Relative to the manufacturing sector, those MNCs in the service sector from our current study have experienced a relatively higher level of employee turnover (beta = −0.107, p < 0.05). This is understandable as workers in the service industry possess more intangible and transferable skills and, hence, tend to be more mobile. It is evident that “shock” effects on intention to quit may have also been stronger in the service sector such as many banking, accounting and insurance firms included in the current study; a similar outcome was concluded in Holtom et al.’s (2005) study. People gave up current jobs likely because of receiving better job offers, starting their own consulting businesses or moving to another location because of spouse’ change of job. This type of turnover is common in the context of Asia (Khatri et al., 2001).

We do not find that there is a statistically significant relationship between parent source of MNCs and turnover. It seems that under the rapid industrialisation process, regardless of where the MNC is sourced from, whether they are Asian based or non-Asian based, the challenge of retaining high skilled labour is unanimously faced by all MNCs surveyed, and this cannot be overcome by way of increased training. Comparing country differences in employee turnover rates, we find that the less developed economies, Indonesia, Malaysia, Philippines and Thailand, experienced a significantly lower level of employee turnover rate than the more advanced economies, such as Singapore and Taiwan. We suspect that the high proportion of service sector in the advanced economies and relatively tight labour market conditions with high demand for skilled labour in Singapore and Taiwan may have heightened the issue of turnover stress for many MNCs operating in these two countries.


This is one of the only handful of studies that examines employee turnover from the organisational and institutional perspective. The results show that use of organisational predictors accounts for a much higher amount of the variance in explaining employee turnover (over 32%), than use of individual perceptions of job satisfaction and commitment within the existing turnover models.

Even though accumulating evidence suggests that training enhances employee commitment and reduces turnover rate, the current study indicates the opposite result. In fact, more training (proxied by average training expenditure in US$) provided by MNCs in Asia leads to a higher level of employee turnover. A number of reasons could explain this phenomenon. First, “the expense of cheap labour” syndrome used by Dowling, Welch, and Schuler (1999: 166) could serve as one way to explain why more training provided by a MNC leads to more employees leaving the firm. As explained by Dowling et al. (1999), most MNCs, even those Asian-owned businesses, have realised the benefits of training to enhance firm performance, so they have invested heavily in training local employees, as also shown in many cases in this study. However, trained and skilled employees may well become very attractive to the competitors, who simply offer better compensation packages to lure them away. “Job hopping” by employees and “poaching” by competitors were repeatedly reported as a key challenge to HR managers in Asia (Dowling et al., 1999; Khatri et al., 2001).

Second, an overinvestment employee–organisation-relationship approach suggested by Tsui, Pearce, Porter, and Tripoli (1997) may have been used by the MNCs surveyed in Asia. The use of this approach appears to be much influenced by “trade unions contracts and some government bureaucracies” (Tsui et al., 1997: 1093). Many MNCs operating in the selected Asian economies in this study may not be bound by trade unions, as would be the case in Tsui et al.’s (1997) study. However, their operation and HR policies were largely controlled by respective local governments. Training tends to be a legislative requirement in some Asia Pacific countries. For example, at the early stage of economic reform in China, foreign partners, mostly MNCs, were required to train unskilled and lower skilled Chinese workers as a part of the joint venture deal (Tsang, 1994). Similarly, in Mexico, companies must train local managers in order to get a licence for operation or be accepted into the local business community (Sargent & Matthews, 1998). MNCs in Singapore are specifically required to invest in training and people development (Wan, 1996). Therefore, it seems that training expenditures could be more a legal and/or host country requirement, rather than an effective investment by MNCs to address the issue of employee turnover.

Third, factors influencing retention are quite complex, including the combined effect of remuneration, career development opportunities, general employment conditions, and employees’ satisfaction with work and work culture. Prior studies have indicated that training alone may not lead to enhancing employee commitment (Huselid, 1995; Zheng et al., 2006). Instead, it is the combination of training with other HR practices that creates a synergy where organisations develop a higher level of organisational commitment in employees, and a lower level of turnover rate is the result (Becker & Huselid, 1999; Beer, Spector, Lawrence, Mills, & Walton, 1984; Zheng et al., 2006). So there is a need to integrate training with the other HR practices to confirm its effect on controlling employee turnover.

While the results regarding the relationship between training and employee turnover were unexpected, taken together, the results highlight the importance of moving the focus of theory development relating to employee turnover from an individual perspective to a combined individual/organisational level of analysis. We have seen in recent years some very promising research being carried out in this direction. For example, both Rhoades et al.’s (2001) and Allen et al.’s (2003) research have directed attention towards examining the role of organisational support in controlling turnover. Peterson (2004) has developed a conceptual HRD model to explain employee attrition rate. Lee and Rwigema (2005) have recognized the dynamic process of individual decisions and complex effects of changing and using a number of organisational HR practices over time as strong predictors for employees to stay or leave organisations. Certainly, many strategic HRM theories developed in recent times (e.g., Martin-Alcazar, Romero-Fernandes, & Sanchez-Gardey, 2005; Nankervis, Compton, & Baird, 2005; Schuler & Jackson, 2000) seek to address attraction and retention of employees within the complex organisational internal and external environment. It seems that new theory building on employee turnover should focus on combining both organisational behaviour theories and HRM theories in order to capture both individual aspects of intention to leave and organisational factors which could influence individual decisions to stay or leave.

Contrary to the explanations from earlier studies, employee turnover rates among the MNCs in Asia studied here are not related to a lack of training. Clearly there is a need to consider the training aspect in the context of a broader strategic approach to HRM practices in these companies. We know that turnover is related to firm size, industry characteristics and years of experience companies have in the local subsidiary, so building the wider extent of operation and broadening expatriate experiences of managing local employees may lead to good reputation about the companies in the local subsidiary and be able to retain local staff more effectively.

Employee turnover was slightly higher when a high proportion of expatriate managers occurred in the local subsidiary, even though the relationship between the extent of intra-company manager transfers and local employee turnover was not statistically significant. We speculate that lack of expatriate and local employees’ social adjustment may be continuously prevalent among the MNCs (e.g., Goodall & Roberts, 2003; Neupert, Baughn, & Dao, 2005; Sargent & Matthews, 1998). This may have created group division, instead of the cohesion and communication that most Asian employees from the surveyed MNCs would be expected to prefer (Hofstede, 1980; Hofstede & Bond, 1988). The lack of socialisation in large MNCs may have led to a higher level of employee turnover among the selected Asian economies in this study.

It appears that selection and recruitment of competent expatriates and associated issues of managing international workforce is still a challenge to many MNCs operating in Asia (Budhwar, 2004; Dowling & Welch, 2004; Nankervis et al., 2006). MNCs must realise that, in addition to selecting qualified expatriates and designing quality expatriate training programs, policy adjustments have to be made to keep a happy and motivated local workforce (Li & Kleiner, 2001: 55). To address the overall issue of attraction and retention of quality local employees in Asia, MNCs are required to have a set of HR policies and procedures at the organisational level. However, these policies and procedures should be relevant to meeting the career aspiration needs of local employees, not just merely focusing on spending more on training and sending more expatriates.

An earlier study among 212 Singapore-based companies identifies that three key factors contributed to high employee turnover in Asia, they were organisational commitment, procedural justice and a job hopping attitude (Khatri et al., 2001). Often the level of organisational commitment among MNCs is low because local employees do not see corporate policies and procedures actually working favourably for their career development. Instead, in the process of international operation, MNCs tend to treat their expatriates far more favourably than local staff especially in terms of rewards and performance evaluation. This has created not only procedural injustice but also distributive injustice, which causes resentment among local employees and creates disloyalty to parent firms (Khatri et al., 2001; Stahl & Zheng, 2002). “Job hopping” has been regarded as an “Asian employee attitude” (Khatri et al., 2001), and it has become a culture which is hard to change, but we wonder whether an increased use of local managers could be more effective in reducing resentment among local employees and retaining competent staff.

Last but not least, it is found that there is no difference between MNCs headquartered in the West and those owned by Asian economies. We expect that the companies owned by those with a strong Confucian influence would emphasise more loyalty and commitment, hence they would have achieved low levels of employee turnover rates, as were suggested in earlier studies (Kuwahara, 1993; Lin, 1998; Zheng et al., 2006). Statistically insignificant results between parent source and turnover presented in the current study seem against the conventional thinking that culture is an additional explanatory factor to the effect of HRM practices on controlling employee turnover. Extra care should be taken into interpreting these results. However, the results are consistent with the conclusions drawn from some recent work (e.g., Chang & Tam, 2004; Yeung, 2006), which appear to suggest that as the large ethnic Chinese and Asian businesses grow globally, they are likely to face similar organisational issues as those Western Anglo-American companies. We may need to reevaluate the extensive theoretical discussion and empirical studies in identifying different HRM practices cross border, focusing more on “organisational imperative” instead of “culture imperative” (Chang & Tam, 2004). It is believed that as the world is moving towards becoming like a global village, the HR philosophy and practices of the Confucian-dominated businesses may also gradually converge with those of mainstream Western practices. At least, from our current study, it is seen that the challenge to retain skilled labour is not dissimilar among all MNCs surveyed. We recommend that companies should look at the issue of employee turnover from the organisational perspective, especially for those operating in transition economies, rather than from the individual and cultural perspective. It is argued that institutional transitions could unravel some hidden organisational factors that might have affected firm performance (Su, Xie, & Li, 2009). Similarly, focus on explaining turnover at the organisational level may help us better identify key variables that have effective control over employee turnover whilst cultural hypothesis and individual perspective only direct us to look at core of the problem. Therefore, it would be potentially more enriching to conduct further research with aims to search for the organisational variables in explaining employee turnover, in addition to examining culture and country differences and evaluating how these differences affect the divergent practices and outcomes. It is anticipated that the contingent patterns of ethnic Chinese, Asian and non-Asian companies will “reflect more of divergent organisational imperatives rather than convergent cultural imperatives” (Chang & Tam, 2004: 25) as MNCs interact with and learn from each other in the globalised world.


The current study examined a set of organisational variables as potential predictors of employee turnover in the context of MNCs in Asia, and showed that use of those variables significantly increased the proportion of variance in employee turnover that can be explained, well beyond what was explained by individual factors in the earlier studies. In our study, however, there remains about 70% of the variance in employee turnover which is unexplained. This might be due to the use of training as a proxy to the overall organisational HRM which can be a limitation, even though training is an important component of HRM functions. Recognising the synergistic effects of various HRM practices on employee attraction and retention, future studies should include the examination of other areas of HR practices in developing turnover models. Some context specific nature of Asian HRM functions should also be taken into consideration in future studies (Budhwar & Debrah, 2009). More importantly, the analysis of antecedents and consequences of employee turnover will be much enriched with the incorporation of both individual and organisational variables. In particular, organisational culture, structure, technology, strategy, managerial philosophy and contextual environment should be included in further studies.


We appreciate Professor Johngseok Bae’s comment on this point, which has helped substantially improve the development of Hypothesis 2.


We again thank Professor Johngseok Bae for this important aspect on EVPs.


Copyright information

© Springer Science+Business Media, LLC 2009

Authors and Affiliations

  1. 1.School of Management, College of BusinessRMIT UniversityMelbourneAustralia
  2. 2.Nottingham Business SchoolNottingham Trent UniversityNottinghamUK

Personalised recommendations