Assignment

228

I

Industrial and Labor Relations Review, Vol. 63, No. 2 (Januar y 2010). © by Cornell University. 0019-7939/00/6302 $01.00

THE INFLUENCE OF HUMAN RESOURCE MANAGEMENT

PRACTICES ON EMPLOYEE VOLUNTARY TURNOVER RATES

IN THE CANADIAN NON GOVERNMENTAL SECTOR

VICTOR Y. HAINES, III, PATRICE JALETTE, and KARINE LAROSE*

The authors tested the influence of thirteen human resource management practices on voluntar y turnover rates the following year while controlling for workplace size, the presence of a separate human resources management unit, union density, industr y, and region. Analysis of data from 4,160 workplaces representative of Canadian industries found that employer-provided training is associated with higher turnover, whereas internal labor markets and formal dispute resolution procedures are associated with lower turnover. The findings are consistent with predictions that some human resource management practices reduce workers’ desire to leave and that training may actually make their leaving easier.

*Victor Y. Haines III is Professor of Human Resource Management and Patrice Jalette is Associate Professor of Industrial Relations. Both are at the University of Montreal. Karine Larose graduated from the University of Montreal in Industrial Relations.

The authors thank Richard Long and George A. Slotsve for their helpful comments on an earlier draft of this paper. They also thank Statistics Canada for data access. In particular, they extend their thanks to Denis Gonthier and Lenka Mach, from Statistics Canada, for their help with statistical analyses.

nterest in the labor turnover problem increased with the publication in 1913 of the first comprehensive empirical analysis and measurement of the costs of labor turnover by General Electric employment manager Magnus Alexander (Bruce 2005). Firms that dedicate significant resources to attract, de- velop, and motivate qualified people do not want to see them leave soon after, especially in a tight labor market (Cascio 2000; Glebbeek and Bax 2004). The consequences of such departures are well documented, in terms of disruptions to the normal operations of an organization and lower organizational performance (see Huselid 1995; Kacmar et

al. 2006; Morrow and McElroy 2007). This is why several business leaders have designed re- tention programs that include generous em- ployee benefits, flexible work arrangements, career development opportunities, and a host of other initiatives that are intended to reduce quit rates. We know little, however, of the actual turnover-reducing capacity of such practices. In fact, few organizational- level studies investigated the human resource management determinants of turnover and fewer still focused on voluntary turnover rates (Guthrie 2001; Shaw et al. 1998). Relative to the multitude of turnover studies conducted at the individual level of analysis, the num- ber of organizational-level investigations is noticeably low. Apparently, nearly all of the available research on employee turnover at- tempts to understand the process through which individuals choose to quit or stay rather than to address the question of why some workplaces experience higher turnover rates than others. It seems that in turnover studies, just as in organizational commitment research (Mathieu and Zajac 1990), difficult access to organizational-level measures of aggregate-level constructs is the reason why

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 229

so few studies have addressed organizational variables such as human resource manage- ment practices.

To provide a comprehensive view of how human resource management practices might influence workplace turnover rates, we revisited a broad and highly applicable framework developed by March and Simon (1958), who posited that ease of movement and desirability of movement predict vol- untar y turnover rates. Within this broad framework, we include other theoretical perspectives to support predictions about the influence of specific human resource management practices on voluntary turnover rates. The objective of our study is therefore to provide theoretical and practical clarity to the relationships between human resource management practices and voluntar y turn- over rates investigated at the workplace level of analysis.

Our focus here is on specific human resource management practices, many of which have seldom or have never been investigated in relation to voluntar y turn- over rates. Hence, rather than investigate bundles of practices (Deler y and Doty 1996; MacDuffie 1995; Wright et al. 1999; Yalabik et al. 2008), the present study considers specific human resource management practices designed to (1) develop employee skills (training, internal labor markets); (2) elicit motivation and commitment (internal labor markets, relative pay, benefits, variable pay); and (3) enable employee influence and voice (participation-enhancing work designs, formal dispute resolution) (Wright and Boswell 2002). Although investigations of bundles or systems of human resource management practices may capture syner- gies between coherent sets of practices that generally include selective staffing, training, internal career opportunities, variable pay and self-directed work teams, they provide little information about the specific effects of any single practice on employee turnover rates. In addition, analyzing bundles of hu- man resource management practices may be flawed since some specific practices, like employer-provided training, may actually increase turnover by making employees more attractive to other organizations. By

considering specific practices, we will be able to make further comparisons across studies in which specific practices have been inves- tigated, thereby increasing our confidence in making generalizations about them.

We also note that past research has assessed turnover in a number of ways (Wright and Bonett 1991) and some organizational-level investigations, in fact, have made no dis- tinction between voluntar y and involuntar y turnover (Arthur 1994; Guthrie 2000, 2001; Huselid 1995; Leonard 1987; Richard and Johnson 2001). Potential problems result- ing from a failure to distinguish adequately between voluntar y and involuntar y turnover rates were noted in past research (Gerhart 1990; McEvoy and Cascio 1987; Stumpf and Dawley 1981). Considering that the determi- nants of voluntar y and involuntar y turnover are quite different (Powell et al. 1994; Shaw et al. 1998), findings from studies that do not make this distinction may be flawed. Our study makes that critical distinction and investigates voluntar y turnover (such as resignations) only.

Another feature of our study is our use of a particularly rich source of data from a large multi-industr y, nationally representa- tive sample of workplaces. This stands out in stark contrast to most studies perform- ing similar research for which data were obtained from a small number of compa- nies within a single industr y. In addition, since data were available from workplaces for two consecutive years, we were able to investigate the relationships between year 1 human resource management practices and year 2 voluntar y turnover rates, thus adding a provision for temporality. This marks a clear departure from previous research in this line of inquir y, in which data collected at one point in time were analyzed for a single year, thereby limiting the strength of causal inference, especially regarding employer-provided training (in other words, does training increase turnover or does higher turnover increase training needs as new recruits join the organization?). Finally, this study focuses on the workplace level of analysis rather than on the company level, thereby avoiding the problem of significant variations in human resource management

230 INDUSTRIAL AND LABOR RELATIONS REVIEW

practices between different sites of the same organization associated with company-level studies (see Becker and Gerhart 1996; Ram- say, Scholarios, and Harley 2000).

Theory and Hypothesis

March and Simon (1958) were interested in the decision to participate, which lies at the core of “organizational equilibrium” (Barnard 1938; Simon 1947). They argued that when a change is made in the life of an organization that explicitly alters the inducements offered to any group of partici- pants or the contributions demanded from them, a prediction can be made regarding the effect of the change on participation (p. 88). For instance, a variable pay plan (inducement) made available to employ- ees (participants) could reduce turnover rates (participation). More specifically, the theor y of organizational equilibrium, as formulated by March and Simon (1958), predicts that participation is a function of the desirability of leaving the organiza- tion and the ease of movement from the organization. Desirability of movement reflects participants’ motivation to leave, which relates to their job satisfaction. Ease of movement reflects participants’ assess- ments of how easy it is to find employment in another organization, which is a function of market conditions and competencies.

Although they did not reference the work of March and Simon (1958), Shaw et al. (1998) developed a framework specific to organizational-level turnover prediction from human resource management prac- tices. They categorized those practices as investments and inducements (see pay, benefits, job stability, procedural justice) and expectations enhancing (such as close supervision, low span of supervisory control). As such, the Shaw et al. (1998) categorization scheme only captures factors that affect the perceived desirability of leaving the orga- nization. Expanding their framework and drawing from March and Simon (1958), we argue here that some competency-enhancing human resource management practices may actually increase ease of movement from the organization.

Human Resource Management Practices and Ease of Movement

Ease of movement is simultaneously determined by factors at the market level and human capital. General job availability, evidenced by the unemployment rate in the relevant job market, is assumed to allow or restrict ease of movement. How easily people can move within the job market is also a func- tion of their work experience, diplomas, and task-relevant abilities, or what is sometimes referred to as their “movement capital” (Trevor 2001) or accumulated human capital (Becker 1964).

The organizational issue addressed in the turnover literature is referred to as retention of talent and not as investment in practices that foster voluntar y departures. Employer- provided training, however, may have that unintended effect by increasing a person’s external marketability and ability to leave the employer. Trevor (2001) noted that this view is consistent with market signaling theor y (Spence 1973) and human capital theor y (Becker 1964; Lippman and McCall 1979). Task-relevant abilities developed through employer-provided training provide the mar- ket with information that indicates productiv- ity (signaling), and they make people more “valuable” (insofar as people are considered human capital). Based on both signaling and human capital theories, a recent individual- level study found that employee participation in a tuition-reimbursement program reduces turnover while employees are still in school but increases turnover when they earn their graduate degree (see Benson, Finegold, and Mohrman 2004). Our study tests such dynamics, but at the workplace rather than at the individual level.

A more nuanced reading of human capital theor y would predict that training in general skills (such as those that are easily transferable to other organizations in the marketplace) would increase voluntar y turnover whereas firm-specific training would diminish vol- untar y turnover. In reality, however, the distinction between general-skills training and firm-specific training is not easy to make considering that “much on-the-job training is neither completely specific nor completely

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 231

general” (Becker 1964: 18), or the training consists of a mix of both firm-specific and general human capital (Baron and Kreps 1999). Further, as Spletzer noted, “one is hard pressed to come up with good examples of training that provides skills that are useful at only one employer” (1997: 730). More than half of the employers sur veyed in recent studies reported, in fact, that the skills their workers possessed are easily transferable to other firms (Barron, Berger, and Black 1999; Fairris 2004). Moreover, employers do not necessarily make clear distinctions between general and specific training (Greer 2000), which means the externalities that may result from employer-provided training generally consist of both general and specific components.

Careful examination of three organization- al-level studies providing unique coefficients for the relationship between employer-pro- vided training and turnover rates reveals no evidence of a turnover reducing effect (Batt et al. 2002; Lincoln and Kalleberg 1996; Shaw et al. 1998). In a study of establishment firms employing at least 20 but fewer than 100 em- ployees, Way (2002) found a non-significant association between training and turnover. In a more recent investigation, Fairris (2004) found a small negative relationship between the percentage of core employees that at- tend off-the-job training programs paid by the company and voluntar y turnover rates. In that same study, however, the number of days devoted to such training was not signifi- cantly associated with turnover rates. Given the samples investigated in the above stud- ies (for example, the telecommunications industr y: Batt et al. 2002; manufacturing plants: Lincoln and Kalleberg 1996; truck- ing organizations: Shaw et al. 1998; small establishments: Way 2002), more research is necessar y to better establish the association between training and turnover. Therefore, in the absence of clear evidence regarding the nature of the association between training and turnover and staying with the theoreti- cally derived arguments above, we formulated the following hypothesis:

Hypothesis 1. More employer-provided training will be positively associated with voluntar y turnover rates.

Human Resource Management Practices and Desirability of Movement

Although a number of motivational forces contribute to the decision to quit (Maertz and Griffeth 2004), March and Simon (1958) have argued that perceived desirability of movement is determined by job satisfac- tion, which is a function of a wide range of job characteristics. For instance, monetar y rewards available in the workplace and par- ticipation in decision-making are likely to influence job satisfaction and thereby the desire to stay or escape a given situation. The presence of an internal labor market, as evidenced by a policy of staffing from within the organization, is likely to reduce desirability of movement through specific skill acquisition or by addressing employee preferences for advancement. In the first case, internal labor markets allow workers to acquire skills that are specific to the firm and thus provide them with an incentive to amortize their investment in these specific skills. In the second case, promoting from within may address employee preferences for advancement within their organization, thus improving job satisfaction and reducing their desire to leave the organization. This predic- tion is consistent with internal labor market theory, which emphasizes investment in train- ing and long-term employment relations (Do- eringer and Piore 1971). This prediction is further supported by an organizational-level study that found that a stronger emphasis on internal mobility and upward progression is associated with lower quit rates (Batt et al. 2002). A recent study, however, found the opposite—that the importance attached to internal promotions is actually associated with higher quite rates (Fairris 2004). Other findings suggest that high performance work systems that emphasize internal promotions reduce turnover rates (Guthrie, 2000, 2001; Huselid 1995). Moreover, according to March and Simon (1958) the perceived possibility of intraorganizational transfer is, along with job satisfaction, a proximal determinant of perceived desirability of movement. Given that an internal labor market allows intraorga- nizational transfer, the following hypothesis is consistent with their framework.

232 INDUSTRIAL AND LABOR RELATIONS REVIEW

Hypothesis 2. Internal labor markets that emphasize staffing from within the orga- nization will be negatively associated with voluntar y turnover rates.

A number of other human resource man- agement practices could reduce desirability of movement. For example, following efficiency wage theor y, more generous compensation packages would make employees unwilling to risk obtaining less money and fewer benefits with a different employer (Campbell 1993; Katz 1986; Stiglitz 1984), thus reducing their desire to move. Consistent with this view, firm-level turnover studies generally find high pay associated with lower turnover rates (Batt et al. 2002; Deler y et al. 2000; Fairris 2004; Guthrie 2001; Leonard 1987; Park et al. 1994; Powell, Montgomer y, and Cosgrove 1994; Shaw et al. 1998). Our study investigates relative pay, which is the establishments’ pay level relative to their industr y pay level. This relationship was investigated with the follow- ing hypotheses:

Hypothesis 3. Higher relative pay will be negatively associated with voluntar y turn- over rates.

Generous employee benefits packages could also be highly effective in bonding employees with their organization. In fact, the prediction that more generous benefits reduce turnover rates was empirically sup- ported by three studies—Bennett, Blum, Long, and Roman 1993; Fairris 2004; and Shaw et al. 1998. Three other studies, how- ever, did not (see Deler y et al. 2000; Park, Ofori-Dankwa, and Bishop 1994; Powell et al. 1994). Efficiency-wage theor y would suggest that more generous benefits offerings reduce desirability of movement and turnover. Our measure of employee benefits, which captures both the availability and funding of seven non-wage benefits, may effectively uncover such an association more so than in past studies. We therefore suggest the following hypotheses:

Hypothesis 4. More generous employee benefits will be negatively associated with voluntar y turnover rates.

The influence of variable pay on turnover rates is not clearly established. Wilson and Peel (1991), for example, found that profit- sharing and share ownership plans reduce

turnover rates. The influence of piece-rate bonuses, however, was not significant in their investigation. In the telecommunications industr y, Batt et al. (2002) found that the percentage of pay of the typical core employee that is variable is positively associated with turnover rates. The researchers in that study explained that performance-based pay fosters financial uncertainty, which increases turn- over. Notwithstanding the Batt et al. (2002) study, the general thrust of the literature links variable pay with high performance work systems that foster employee motiva- tion and commitment which in turn result in lower turnover (Arthur 1994; Guthrie 2000, 2001; Huselid 1995). The same association is supported in the labor economics literature where studies have demonstrated that vari- able pay may reduce turnover because wages are more flexible (Azfar and Danninger 2001). More flexible wages, that is, can re- spond to alternative wage offers and thereby improve retention, according to one study (see Blakemore, Low, and Ormiston 1987).

Although our hypothesis is consistent with these views, we wish to point out that our measure of turnover did not provide us with an opportunity to investigate possible “sorting” effects, whereby variable pay fosters more departures among poor performers than among high performers. There is clear evidence of a negative performance-turnover relationship in individual-level studies (By- cio, Hackett, and Alvares 1990; McEvoy and Cascio 1987). More to the point, however, is evidence of a stronger negative performance- turnover relationship in organizations using performance-contingent rewards (Williams and Livingstone 1994). As high performers receive more rewards, they would ostensibly be more satisfied and less likely to quit. As low performers receive fewer rewards, they would be less satisfied and more likely to quit. Further advancing this view with orga- nizational-level data, Park et al. (1994) found positive relationships between the availability of group incentives and dysfunctional turn- over (that is, more high performers leave) and between the availability of individual incentives and functional turnover (that is, more low performers leave).

This brief review underscores the many

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 233

complexities involved in the associations be- tween individual-collective variable pay plans and employee turnover. For the purpose of hypothesis testing in our study, one regression model will include a composite variable-pay variable, which will provide a test of whether or not more variable pay is better than less. A second regression model will include four variable pay plans. This model will indicate whether some specific forms of variable pay have a potentially greater impact than others.

Hypothesis 5. Greater use of variable pay (individual incentives, gain-sharing, profit- sharing, merit or skill-based pay) will be nega- tively associated with voluntary turnover rates.

Team-based work systems and other par- ticipator y work designs may provide partici- pants with enough influence to improve their work environment and reduce their desire to move. Research conducted by Batt et al. (2002) found that the extent of employee participation in problem-solving teams (con- sultative participation) and in self-directed work teams (substantive participation) is associated with lower turnover rates. Wilson and Peel (1991) found that the existence of a joint job evaluation scheme and a work council are associated with lower turnover rates. Their employee-participation and span of control variables did not, however, show significant relationships with turnover rates. In another study, no support was found for the predicted relationship between employee participation in quality circles and turnover rates (Lincoln and Kalleberg 1996). Finally, Deler y et al. (2000) found no relationship between employee participation and turn- over rates in the trucking industr y.

Taken together, these studies provide little evidence that participation-enhancing work designs are associated with lower turnover. In the few studies that are available, we identi- fied four negative associations and four non- significant associations. This underscores the need to investigate further specific forms of employee participation mechanisms. In the present study, drawing from a broad defini- tion of employee participation (Ledford and Lawler 1994), we examine the influence of five participation-enhancing work designs on employee voluntar y turnover. These address consultative and substantive participation

as well as on- and offline participation. As such, they address a diversity of participation- enhancing work designs (Strauss 2006). For the purpose of hypothesis testing, one regres- sion model includes a composite participa- tion variable that tests whether or not more participation-enhancing work designs are better than fewer of them. Another regression model includes five participation-enhancing work designs, indicating whether some spe- cific forms of participation-enhancing work designs have a potentially greater impact than others on employee voluntary turnover.

Hypothesis 6. Greater use of participation- enhancing work designs (employee sug- gestion program, information sharing, problem-solving teams, self-directed work groups, flexible job design) will be negatively associated with voluntar y turnover rates.

The exit-voice framework (Hirschman 1970), when applied to the workplace (Free- man and Medoff 1984), considers voice as any attempt by dissatisfied employees to express their concerns rather than to escape an unsatisfactory situation by quitting. Initial applications of the exit-voice framework to employment relations was based on the as- sumption that unionized employees benefit from institutional protections negotiated by their union and, therefore, are more likely to voice their dissatisfaction than the non- unionized employees who are more prone to quit (Freeman and Medoff 1984). But impartial dispute-resolution procedures, like complaint, appeal, or grievance procedures, whether they are established in a union or non-union setting, allow employees to voice their dissatisfaction within their current workplace rather than to quit. In addition, such procedures provide a voice mechanism through which workers can gain more satisfy- ing conditions of employment. From an orga- nizational justice perspective (see Folger and Cropanzano 1998), dispute resolution proce- dures are expected to improve perceptions of both procedural and distributive justice, which enhance job satisfaction and reduce the desirability of movement as a result.

The current state of research clearly dem- onstrates the need for further investigation of such arguments. In one study, for example, the availability of a formal non-union griev-

234 INDUSTRIAL AND LABOR RELATIONS REVIEW

ance procedure and non-union arbitration were not significantly associated with turnover rates (Batt et al. 2002). Only the presence of peer-review panels was associated with lower turnover rates. In another study, practices reflecting procedural justice were unrelated to turnover rates (Shaw et al. 1998). In still another study, having a formal grievance procedure was not significantly associated with turnover rates in regression equations that controlled for unionization (Deler y et al. 2000). These findings may have resulted from the challenges of measuring dispute resolution procedures in specific contexts. Thus, drawing from the above theoretical frameworks, we tested the due process propo- sition with the following hypothesis:

Hypothesis 7. The presence of a formal dis- pute resolution procedure will be negatively associated with voluntar y turnover rates.

Methods

Data and Sample

The data for this study were collected by Statistics Canada as part of the Workplace and Employee Sur vey (WES) directly from sur vey respondents. Data collection, data capture, preliminar y editing and follow-up of non-respondents are all done in Statistics Canada Regional Offices. In 1999, workplace data were collected in person. Beginning in 2000, computer assisted telephone interviews were conducted. The overall goal of the survey was to examine the way in which employers and their employees respond to the changing competitive and technological environment. The inter views were conducted with general managers of smaller locations and with hu- man resources managers of larger locations. Responding to this sur vey was mandator y.1 For about 20 percent of the sur veyed units (mostly large workplaces), more than one

respondent was required to complete the questionnaire.

The instrument was developed through various consultations with experts, such as those from the Subject Matter Advisor y Group, Human Resources Development Canada, and EKOS Group. Based on their comments and suggestions, appropriate changes to the questionnaires were made. A pre-test of 50 businesses was conducted followed up later by a pilot sur vey.

The sampling frame is stratified by indus- tr y (14), region (7), and size (3), which is defined using estimated employment. The target population for the sur vey is defined as all business locations operating in Canada that have paid employees, with the following exceptions: employers in certain northern locations (Yukon, Nunavut and Northwest Territories), employers operating in crop production and animal production, fishing, hunting and trapping, private households, religious organizations and public admin- istration. The sample comprises the same locations followed over time with rigorous data-collection procedures generating a 1999 sample of 6,322 workplaces with a response rate of 95.2 percent. The year 2000 sample included 6,068 workplaces with a response rate of 90.9 percent. Thus, 96 percent of workplaces that responded to the 1999 sur- vey also responded to the year 2000 sur vey. Such high response rates limit the risk of a non-response bias.

Workplaces with fewer than 10 employees were excluded from the analysis because (a) a different data collection approach was used in those (smaller) workplaces, (b) the questionnaire for smaller workplaces did not include questions about the participation- enhancing practices included in our model, (c) the human resource management prac- tices we investigated are likely to have differ- ent meanings in workplaces of that size. We could have further reduced our sample by including only those workplaces with 100 or more employees, but this would have ham- pered our ability to generalize findings to a meaningful portion of workplaces. With no theoretical or empirical basis to do so, there was no reason to further reduce sample size by setting any other minimum workplace size

1Mandator y participation did not, however, yield a perfect response rate. Some workplaces may have felt over sur veyed by such time-consuming sur veys and refused participation for that reason. Although sanctions could be applied to non-responding workplaces, they are difficult to enforce.

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 235

for inclusion in the analyses. Moreover, em- ployee turnover is a major concern in small as well as in large workplaces. Thus, the final sample was based upon 4,160 workplaces with complete data for both years and that had 10 or more employees in 1999.

These workplaces included a comprehen- sive spectrum of industries (forestr y/min- ing; labor intensive tertiar y manufacturing; primar y product manufacturing; secondar y product manufacturing; capital intensive tertiar y manufacturing; construction; trans- portation/storage/wholesale trade; commu- nication and other utilities; retail trade and commercial ser vices; finance and insurance; real estate, rental, leasing operations; busi- ness ser vices; education and health care; information and cultural industries) and broadly represented Canadian workplaces with more than 10 employees. Weights were calculated using the inverse of the prob- ability of selection of each workplace into the stratified sample, which is a function of industr y, region, and size. The weighted average number of employees per workplace of the final sample was 45.70 employees (SD = 173.32). As such, smaller workplaces were well represented in this study, contrasting with previous turnover studies (see Guthrie 2000, 2001; Huselid 1995). In addition, 17.36 percent (SD = 0.38) of the workplaces in our study were covered by a collective agreement. Other descriptive statistics are presented in the results section.

Measures

In developing our measures, we kept in mind that the unit of analysis for this study is the workplace; that is, we intended all mea- sures to reflect the workplace level of analysis. This would seem most appropriate given that the dependent variable is workplace turnover. The dependent variable is voluntary turnover for year 2000, which was captured as a ratio of the number of employees who had resigned between April 1, 1999 and March 31, 2000, over the number of people employed at the location in the last pay period of March 2000. The weighted average year 2000 voluntar y turnover rate was 21.70 percent.

The sur vey provided detailed information

about several human resource management practices such as employer-provided training, internal labor markets, benefits, variable pay, participation-enhancing work designs, and employee voice. Rather than combine measures into these broad categories or bundles of practices, however, we chose to emphasize specific human resource manage- ment practices. Employer-provided training coverage was a function of the total number of employees who had received training, over the number of people employed at the workplace. The sur vey actually included two questions that addressed this practice. The first asked the respondent, “Please estimate the number of employees who received classroom training between April 1, 1998 and March 31, 1999.” The second question asked the re- spondent, “Please estimate the number of employees who received on-the-job training between April 1, 1998 and March 31, 1999.” In each case, the response was divided by the number of employees at the workplace and the final training measure was computed as the mean of these two ratios (alpha = .82).

Internal labor markets exist when internal mobility opportunities are available within the organization. Thus, the sur vey asked how vacant positions are usually staffed. Respondents were instructed to check the most frequently used method: “From within the workplace,” “From another workplace within the same legal company or business enterprise,” or “From outside the company.” To best capture the concept of internal labor markets, the first two response options were coded 1 and the third was coded 0. Three job categories were deemed most relevant to internal promo- tions, namely “managers,” “professionals,” and “clerical/administrative.” Hiring from within is more meaningful in these three job categories because they are less likely to be entr y-level positions. On the basis of the number of employees within each of these three job categories, we computed a ratio of the number of positions targeted by a policy of hiring from within the workplace. If, for example, the workplace included 10 man- agers, 20 professionals, and 70 clerical/ad- ministrative employees and promoting from within was the most frequently used method for professionals only, then we would have

236 INDUSTRIAL AND LABOR RELATIONS REVIEW

a ratio of 20 % (20 professionals divided by the total number of employees within these three job categories in the workplace).

Compensation practices include relative pay, benefits and variable pay. First, the measure of relative pay was computed as the difference between workplace pay and the corresponding industr y pay. Workplace pay was obtained by dividing total gross payroll at the workplace by the number of employ- ees at the workplace. Industr y pay was the average pay of all workplaces within each of the 14 industries. The difference between workplace and industr y pay, or relative pay, indicates to what extent the pay level at each workplace is higher or lower than industr y average pay. The measure of employee ben- efits was a function of both the availability of seven non-wage benefits and how they are funded. Second, the seven non-wage benefits listed in the questionnaire include a pen- sion plan, life and/or disability insurance, supplemental medical, dental care, group registered retirement savings plan (RRSP), stock purchase or other savings plan, and supplements to employment insurance ben- efits (for example, for maternity or lay-off). A value of 0 was assigned when the benefit was not available or available but not funded by the employer, coded 1 if available and funded by both the employee and the employer, and coded 2 if available and fully funded by the employer. The scores were summed across the seven non-wage benefits to obtain an overall score. Third, the sur vey inquired about the presence of individual incentives (bonuses, piece-rates, commissions and stock options), productivity/quality gain-sharing and other group incentives, a profit-sharing plan, and merit- and skill-based pay. Definitions of “productivity gain-sharing,” “profit-sharing plan,” and “merit pay or skill-based pay” were provided in the questionnaire.2 The coding of

these variables reflected the absence (coded 0) or presence (coded 1) of each variable pay scheme within the workplace.

The five participation-enhancing work designs were the presence of an employee suggestion program, information sharing with employees, problem-solving teams, self-directed work groups, and flexible job design. The question was, “For non-managerial employees, which of the following practices exist on a formal basis in your workplace?” Defini- tions were provided3 and the presence of the practice was coded 1 (Yes) whereas the absence was coded 0 (No). The presence of a formal dispute resolution procedure was obtained with the following question: “Does this workplace have a dispute, complaint or grievance system for employees?” A score of 0 was assigned when no such system was in place or when only an informal procedure was available and 1 when a formal dispute resolution system was available.

A number of control variables were includ- ed in this study in order to reduce the likeli- hood of spurious associations. Workplace size was included because of its influence on human resource management practices and employee turnover rates (Huselid 1995; Shaw et al. 1998) and was measured by the natural logarithm of the number of em- ployees. The presence of a separate human resources management unit staffed with more than one person in the workplace was also included as a control variable because of its influence on the adoption of human resource management practices and internal labor

2“Productivity gain-sharing” means benefits to em- ployees for gains realized by increased productivity. Commonly, these benefits can be in the form of money payments in the primary industries. “Profit-sharing plan” means any plan by which employees receive a share of the profits from the workplace. “Merit pay or skill-based pay” means a reward or honor given for superior quali- ties, great abilities or expertise that comes from training, practice, and so on.

3 An employee suggestion program includes employee sur vey feedback. Information sharing with employees includes, for example, with respect to firm’s perfor- mance, colleagues’ wages, technological or organiza- tional changes, and so on. This implies that employees have some feedback on policies. Problem-solving teams are teams whose responsibilities are limited to specific areas, such as quality or work flow (that is, they have a narrower range of responsibilities than those of self- directed work groups). Self-directed work groups are semi-autonomous work groups or mini-enterprise work groups that have a high level of responsibility for a wide range of decisions/issues. Flexible job design includes job rotation, job enrichment/redesign (broadened job definitions), job enrichment (increased skill variety or autonomy of work).

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 237 T

ab le

1 . M

ea n

s, S

ta n

d ar

d D

ev ia

ti o

n s,

a n

d C

o rr

el at

io n

sa

Va

ri ab

le

M ea

n

s. d.

1

2 3

4 5

6 7

8 9

10

11

12

13

14

15

16

17

1.

20 00

t u

rn o

ve r

0. 22

0.

46

2.

T ra

in in

g 0.

37

0. 38

.2

3

3.

In te

rn al

l ab

o r

m ar

ke ts

0.

21

0. 34

–.

05

.1 5

4.

R el

at iv

e p

ay l

ev el

–2

83 1

16 06

0 –.

04

.0 5

.0 1

5.

B

en efi

ts

3. 38

2.

79

–. 12

.1

8 .1

6 .2

6

6.

In

d iv

id u

al i

n ce

n ti

ve s

0. 40

0.

49

.0 4

.0 7

.0 1

.1 8

.2 1

7.

G

ai n

s h

ar in

g (g

ro u

p )

0. 15

0.

36

–. 03

.1

0 .0

3 .1

2 .2

2 .2

9

8.

P

ro fi

t– sh

ar in

g 0.

14

0. 34

.0

1 .0

5 .0

6 .1

1 .2

4 .2

0 .2

2

9.

M er

it &

s ki

ll –b

as ed

p ay

0.

31

0. 46

–.

03

.1 7

.0 3

.1 4

.1 8

.2 9

.3 2

.1 7

10 .

E m

p lo

ye e

su gg

es ti

o n

p ro

gr am

0.

27

0. 45

.0

7 .2

0 .1

0 .0

1 .1

0 .0

8 .0

5 .1

2 .0

4

11 .

In fo

rm at

io n

s h

ar in

g 0.

44

0. 50

–.

02

.2 5

.1 1

.0 7

.1 9

.0 1

.0 8

.1 8

.1 1

.3 9

12 .

P ro

b le

m –s

o lv

in g

te am

s 0.

23

0. 42

–.

00

.2 1

.1 4

.0 2

.1 5

.0 0

.0 8

.1 0

.0 8

.3 7

.4 0

13

. Se

lf –d

ir ec

te d

w o

rk g

ro u

p s

0. 09

0.

29

–. 05

.1

0 .1

4 .0

3 .1

4 .0

7 .1

1 .1

9 .1

2 .1

5 .2

9 .2

6

14

. F

le xi

b le

jo b

d es

ig n

0.

28

0. 45

.0

0 .1

5 .0

9 .0

3 .0

2 .0

1 .0

9 .1

0 .1

2 .3

7 .4

2 .3

6 .2

7

15 .

F o

rm al

d is

p u

te –r

es o

lu ti

o n

0.

24

0. 43

–.

12

.1 3

.1 2

.0 7

.2 8

.0 2

.0 9

.0 6

.0 3

.1 1

.1 7

.1 3

.1 6

.0 3

16 .

Si ze

( lo

g o

f em

p lo

ym en

t) b

45 .7

0 17

3. 32

–.

02

.1 2

.1 6

.1 4

.2 8

.0 8

.0 2

.1 0

.1 0

.1 6

.1 8

.1 8

.1 3

–. 01

.3

3

17 .

U n

io n

d en

si ty

( lo

g) c

14 .1

6 12

0. 93

–.

10

.0 9

.1 1

.1 0

.2 2

–. 10

–.

03

–. 03

–.

02

.0 2

.0 8

.1 2

.0 4

–. 06

.5

4 .4

8 18

. H

u m

an r

es o

u rc

e d

ep ar

tm en

t 0.

07

0. 26

–.

03

.1 1

.1 5

.1 6

.2 7

.0 8

.0 5

.0 7

.1 0

.0 6

.1 1

.1 0

.0 8

–. 01

.2

4 .4

2 .3

1 a

F o

r al

l co

rr el

at io

n s

gr ea

te r

th an

.0 6,

p <

.0 01

; f o

r al

l co

rr el

at io

n s

gr ea

te r

th an

.0 4,

p <

.0 1;

f o

r al

l co

rr el

at io

n s

gr ea

te r

th an

.0 3,

p <

.0 5.

b , c

T h

e d

es cr

ip ti

ve s

ta ti

st ic

s fo

r th

es e

va ri

ab le

s ex

p re

ss t

h ei

r va

lu es

b ef

o re

n at

u ra

l lo

ga ri

th m

s w

er e

co m

p u

te d

. T h

e co

rr el

at io

n c

o ef

fi ci

en ts

w er

e o

b ta

in ed

a ft

er

n at

u ra

l lo

ga ri

th m

s w

er e

co m

p u

te d

.

238 INDUSTRIAL AND LABOR RELATIONS REVIEW

market structures (Pfeffer and Cohen 1984). In previous studies, the presence of a union was shown to influence employee turnover (see Freeman and Medoff 1984) and human resource management practices such as pay (see Fang and Verma 2002) and benefits (see Akyeampong 2002). We included union density as a control variable in this study. It was computed as the total number of em- ployees covered by a collective agreement in the workplace over total employment in the workplace. The natural logarithm of union density was used in the analyses. Responding workplaces were assigned 1 of 14 industr y stratum codes and dummy variables were created for each industr y. Finally, region was included to control for market condi- tions that may influence ease of movement. Dummy variables were created for each of the following regions: Atlantic, Quebec, Ontario, Manitoba, Saskatchewan, Alberta, and British Columbia.

Analysis and Weights

The aim of our analyses was to to capture the influence of 1999 human resource man- agement practices on year 2000 voluntar y turnover rates. We used a tobit procedure because turnover rates are left-censored. A censored variable includes a large fraction of obser vations at the minimum or maximum. Voluntar y turnover rates have a large frac- tion of obser vations at the 0 minimum. In our sample, 33.7 percent of workplaces had 0 turnover; because of this, OLS estimates of the mean and variance of the turnover variable would be biased. tobit models cor- rect for biasing coefficient estimates of the predictors due to the truncation of the de- pendent variable. The tobit technique uses all obser vations, both those at the limit and those above it, to estimate the regression line. This technique is preferable, in general, to techniques that estimate a line only with the obser vations above the limit (McDonald and Moffitt 1980) and was applied in recent organizational-level analyses of employee turnover rates (see Batt 2002; Batt et al. 2002).

The sample of workplaces is based on a stratified sample design that incorporates information on employment size, industr y,

and region. For the sample data to correctly represent the population from which the stratified sample of workplaces was drawn, proportional weights were assigned in the analyses. Thus, notwithstanding the above exceptions, the results of this study general- ize to all Canadian workplaces of more than 10 employees in the targeted industries. Weighted data were used not only to account for the sampling design, but also because Statistics Canada will only release weighted outputs as a data confidentiality precaution. The weights were calculated using the inverse of the probability of selection of each work- place into the stratified sample, which is a function of industr y, region, and size. Tobit estimation with Stata, however, does not allow the use of the proportional weights. Thus, we used intreg, a generalization of tobit, that allows proportional weights and robust variance estimates (Stata 2003). With the use of proportional weights in the intreg com- mand and the inclusion of employment size, industry, and region as control variables in the regression models, the robust standard error estimates account for most of the complexi- ties of the design used for the WES sur veys.

Results

Table 1 presents the means, standard devia- tions, and correlations for all variables except the 14 dummy-coded industr y variables and the seven dummy-coded region variables (full correlation results available upon request). The descriptive statistics demonstrate that some human resource practices such as in- dividual incentive systems and information sharing were more frequent than others such as profit-sharing and self-directed work groups. There are generally weak correlations among the human resource practices (mean r = 0.14), suggesting that the results are not likely to suffer from multicollinearity. The correlations further show that ten of the four- teen human resource management practices are significantly associated with lower volun- tar y turnover rates the following year. Three practices—training, individual incentives, and an employee suggestion program—are significantly associated with higher voluntary turnover rates the following year. Amongst

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 239

the control variables, union density and the presence of a human resource department are associated with lower turnover. Regard- ing industr y and region, we found a pattern of significant results that is consistent with other sources.

Table 2 presents the results from the multivariate analysis in which the year 2000 turnover rates are regressed on the control variables and the human resource manage- ment practices.

The results reveal associations between human resource practices and and voluntar y turnover rates while controlling for workplace size, presence of a separate human resource department, union density, industr y, and region. What should be noted first is the positive association between workplace size and turnover as well as the negative associa- tion between union density and turnover in Model 1. Our analysis of data from 4,160

workplaces in a wide array of industries found that employer-provided training is associated with higher turnover whereas internal labor markets, or a greater emphasis on staffing from within, is associated with lower turnover. The presence of a formal dispute resolution procedure is related to lower turnover rates the following year. As such, hypotheses 1, 2, and 7 received support in regression models that controlled for workplace size, the pres- ence of a human resource department, union density, industr y, region, and other human resource management practices.

Using the procedure outlined by Batt (2002) to estimate the effect sizes of coeffi- cients in tobit models,4 a one-standard-devi-

Table 2. Voluntar y Turnover as a Function of HR Practicesa

Model 1 Model 2 Model 3

Variable z p z p z p

Control variables Size (log of employment) 2.68 0.007 2.74 0.006 2.60 0.009 Union density (log) –3.52 0.000 –1.70 0.089 –1.43 0.152 Human resource department 0.43 0.670 0.67 0.503 0.80 0.424 Industr yb — — — — — — Regionb — — — — — — HR practices Training 2.58 0.010 2.70 0.007 Internal labor markets –2.14 0.032 –2.29 0.022 Relative pay level –1.27 0.205 –1.42 0.157 Benefits –1.75 0.079 –1.87 0.062 Variable pay 0.02 0.986 Individual incentives 1.55 0.122 Gain sharing (group) –1.26 0.207 Profit–sharing 1.49 0.137 Merit or skill-based pay –1.71 0.088 Participation –1.12 0.263 Employee suggestion program 1.15 0.252 Information sharing –1.66 0.096 Problem-solving teams 0.32 0.752 Self-directed work groups –1.04 0.297 Flexible job design –0.75 0.450 Formal dispute–resolution –2.05 0.040 –2.10 0.036

Sample 4160 4160 4160 Wald c2 50.76 66.23 71.75 Probability c2 0.001 0.000 0.000

a Unstandardized tobit-intreg estimates are reported. b Not included but available upon request.

4Indeed, according to the procedure described by Batt (2002: 592) in a similar study, “to estimate the ef- fect sizes of coefficients in tobit models, the coefficients must be decomposed into changes in the probability

240 INDUSTRIAL AND LABOR RELATIONS REVIEW

ation increase in employer-provided training is associated with a voluntar y turnover rate that is 1.38 (0.51 x 2.70) percentage points higher; a one-standard-deviation increase in hiring from within is associated with a turnover rate that is 1.17 percentage points lower; and a one-standard-deviation increase in the presence of a formal dispute resolu- tion procedure is associated with a turnover rate that is 1.07 (0.51 x – 2.10) percentage points lower.

Associations between other human re- source management practices and turnover are either marginally significant (p < .10) or not significant at all. Relative pay is not significantly associated with turnover (Hypothesis 3). The negative association between more generous employee benefits and turnover is only marginally significant (Hypothesis 4). Variable pay, as measured by a composite variable (Model 2), is not significantly related to turnover (Hypothesis 5). The negative relationship between merit and skill-based pay and turnover (Model 3), however, is marginally significant. Participa- tion-enhancing work designs, as measured by a composite variable (Model 2), are not significantly associated with turnover (Hypothesis 6). Of the five participation- enhancing work designs (Model 3), only the negative association between information sharing and turnover was significant, but marginally.

In sum, although the correlations provide almost full support for all of the hypotheses, the regression models that controlled for third factors (such as workplace size, the presence of a human resource depart- ment, union density, industr y, and region) provided full support for hypotheses 1, 2, and 7, and little or no support for the other hypotheses. Nonetheless, it is quite clear from these analyses that some specific human resource management practices are significantly associated with voluntar y turnover rates.

Discussion

Drawing from an extensive nationally representative sample of workplaces, our aim has been to develop a comprehensive view of how various human resource management practices influence voluntar y turnover rates. While controlling for critical contextual variables, our findings provide support for predictions derived from the March and Simon (1958) framework. A significant posi- tive association between employer-provided training coverage and voluntar y turnover rates suggests that greater investments in human capital may provide greater ease of movement, resulting in employee departures. Conversely, significant negative associations between internal labor markets, employee benefits, dispute-resolution procedures, merit- and skill-based pay, information shar- ing and lower turnover rates are consistent with desirability of movement predictions.

In the context of specific human resource management practices, our findings clearly indicate that employer-provided training may actually increase voluntar y turnover rates (Batt et al. 2002; Lincoln and Kalleberg 1996; Shaw et al. 1998). Using a measure that included estimates of the number of employees involved in on-the-job and classroom training, the relationships we obser ved suggest that employer-provided training is associated with higher turnover the year following the training. Thus, the ease of movement explanation (March and Simon 1958) seems more applicable than the organizational commitment argument (Lincoln and Kalleberg 1996) to account for the influence of employer-provided training on voluntary turnover. However, in Canada, as in the United States, employee training may consist more of portable technical skills than of socialization in a particular corporate cul- ture, such as in Japanese firms (Lincoln and Kalleberg 1996), and our study did not distin- guish initial training from ongoing training. Our findings, therefore, may not generalize to organizations that make training an integral part of their socialization process. Thus, the question remains: All things being equal, why invest in human capital if that human capital can be poached by competitors? (see Forrier

of obser ving an outcome above the left limit, the latter of which provides an interpretation equivalent to OLS estimates (McDonald and Moffitt 1980). In this case, the effect sizes of the tobit coefficients are equal to 0.51 of the OLS coefficients.”

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 241

and Sels 2003). The alternatives that involve reductions in training investments or hiring talent from competitors may have worse con- sequences still. For instance, the workplace may stagnate or experience retaliation from competitors (Gardner 2005). Clearly, more research is needed to understand the specific association between employer-provided train- ing and turnover.

In both theor y and practice, there exists a strong positive association between employer- provided training and voluntar y turnover rates, suggesting that it may be appropriate to think about the turnover-reducing functions of human resource management practices in terms of the March and Simon (1958) framework. In practice, this implies that more attention must be paid to how differ- ent human resource management practices may reduce turnover. Investing in some practices, such as formal dispute-resolution procedures, may reduce turnover and thereby cancel out the potential losses associated with significant investments in training. Knowing the specific effects of such practices may guide organizations that cannot other wise dedicate sufficient resources to implement comprehensive systems of human resource management practices.

Our study found strong evidence that internal labor markets are associated with lower turnover rates in a way that is consistent with internal labor market theory (Doeringer and Piore 1971). Such evidence is also con- sistent with reports that high performance work systems that include an emphasis on internal promotions reduce turnover rates (Guthrie 2000, 2001; Huselid 1995). It is es- pecially consistent with the finding that more internal mobility opportunities are associ- ated with lower quit rates (Batt et al. 2002). Meta-analyses of research in organizational behavior also support internal labor market predictions (see Cotton and Tuttle 1986; Hom and Griffeth 1995). The implications for practice are straightfor ward: Organiza- tions wishing to control or reduce voluntar y turnover would be well advised to hire from within and offer vertical (promotions) and lateral (transfers) opportunities. Staffing policies that emphasize hiring from within provide participants with the opportunity to

move into the firm’s more desirable jobs and may help build a strong corporate culture that is conducive to employee retention.

It is generally expected that higher wages reduce the probability that employees accept job offers from other firms. Although a nega- tive correlation between relative pay and turn- over supports this view, the association was not significant in the regressions that included control variables. Thus, our results provide little support for efficiency wage theory, which predicts that more generous wages make employees unwilling to risk obtaining less money with a different employer (Katz 1986; Stiglitz 1984). The efficiency wage theor y prediction of a positive relationship between relative wage and productivity may not apply as well to organizational-level models of vol- untar y turnover. The prediction that more generous employee benefits reduce turnover rates, however, was marginally supported by our data (p < .10). Consequently, with three studies showing negative associations between benefits and turnover (Bennett, Blum, Long, and Roman 1993; Fairris 2004; Shaw et al. 1998) and three studies showing no significant associations between those variables (Deler y et al. 2000; Park et al. 1994; Powell et al. 1994), our findings marginally support the turnover-reducing effect. More generous employee benefits packages, those that include more benefits financed more by the employer than by the employee, appear to be effective in bonding employees with their organization.

The data on variable pay suggests that, other than the negative and marginally sig- nificant (p <.10) association between merit and skill-based pay and voluntar y turnover rates, performance-based pay plans generally had no significant effect on turnover. Taken together, these findings indicate that other forces are at work shaping the associations between variable pay and turnover. As Batt et al. (2002) suggested, more attention should be given to the financial uncertainty involved in some variable pay plans. Moreover, the distinction between functional and dysfunc- tional turnover may need to be made in future research, for variable pay may foster more departures among poor performers, an effect that we were unable to verify in this study.

242 INDUSTRIAL AND LABOR RELATIONS REVIEW

Apparently, there is more at work here than mere conventional wisdom would suggest, pointing to the need to investigate financial insecurity and the functional-dysfunctional aspects of voluntar y turnover.

Among the participation-enhancing work designs, information sharing was found to be marginally (p < .10) associated with lower turnover rates the following year. Such a practice is likely to reflect a work environ- ment that allows participants to express their concerns rather than to want to exit the workplace. Other practices included in the participation-enhancing work designs, however, that apparently do not allow em- ployees to voice their concerns to the same extent—employee suggestion programs, problem-solving teams, self-directed work groups, and flexible job design—are more task-focused than information sharing. Over- all, the results did not reveal a clear pattern justifying meaningful distinctions between consultative and substantive participation or online and offline participation.

Like information sharing, formal dispute- resolution procedures also allow employees to voice their dissatisfaction in the workplace, and our findings clearly show that formal dispute-resolution procedures are associated with lower turnover rates the following year. This finding is consistent with both the exit- voice framework (Hirschman 1970) and the March and Simon (1958) framework. The presence of a formal procedure allows par- ticipants to voice their concerns and thereby reduce their dissatisfaction and desirability of movement rather than to quit. Interest- ingly, although Deler y et al. (2000) found that the presence of a formal grievance pro- cedure was not significantly associated with turnover rates in regression equations that controlled for unionization, our finding is robust even when union density is included as a control variable. As such, the presence of such a workplace procedure offers voice advantages beyond those offered by union representation.

These results need to be interpreted in light of the limitations of this study. Its de- sign predicted associations between human resource management practices and turnover rates. According to meta-analytic evidence,

the best predictors of turnover are proximal precursors in the withdrawal process, such as job satisfaction and intentions to leave (Griffeth, Hom, and Gaertner 2000) whereas human resource management practices are more distal predictors of turnover. They may, as such, influence turnover indirectly through attitudes and perceptions of organizational justice or other more proximal constructs. For instance, individual incentives may reduce employee turnover to the extent that they improve job satisfaction. Therefore, to better understand the organizational determinants of turnover, future researchers might investi- gate multi-level associations between human resource management practices and more proximal predictors of turnover.

The measurement of human resource man- agement practices could also be improved in future investigations. The measurement of training, for instance, might distinguish between initial and ongoing training. With regards to the dependent variable, it would be of great value to find a measurement approach that could distinguish functional from dysfunctional voluntar y turnover. Al- though much theorizing draws from such a distinction, actual measures of functional or dysfunctional turnover at the organizational level of analysis are lacking.

The study is also limited in that causality cannot be inferred on the basis of a two-year panel design. For instance, our results sug- gest that training increases turnover rates, but high turnover rates may also increase the need for initial training of the employees hired to replace those who have left. Although a study design that assesses training and then turnover rates the following year would seem to rule out this possibility, the occur- rence of employee turnover may be quite stable across years in the same workplace and high turnover this year may indicate high turnover the previous year within the same workplace. Thus, the significant posi- tive relationship between 1999 training and year 2000 turnover rates may in fact reflect the relationship between 1999 training and 1999 or even 1998 turnover rates. Therefore, to provide an additional causality check, we conducted another regression analysis with 1999 training and year 2000 turnover rates

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 243

with 1999 turnover rates included as a control variable. The results again provided a positive significant relationship between 1999 train- ing and year 2000 turnover rates (coefficient = 2.20; p = 0.028). This study, however, did not test the possible influence of interactions between training and other human resource management practices on turnover rates. Future research, we suggest, should probe the moderating influence of other human resource management practices or “self- reinforcing strategies” on the relationship between training and turnover (Dearden, Machin, Reed, and Wilkinson 1997).

Future research should also consider industr y dynamics. Our study found, for example, that industr y was a significant con- trol variable. It was also a necessar y control variable given our multi-industr y sample. It could be argued, however, considering the centrality of industr y in competitive analysis (Porter 1980), that intra-industr y dynamics are just as meaningful. The policies and practices that are associated with turnover in the retail sector, for instance, might be quite different from those that have an impact in manufacturing. An industr y-by-industr y analysis would reveal more intra-industr y specificities.

This study provides a detailed theor y- based assessment of the influence of human

resource management practices on voluntary turnover rates. Our investigation tested the associations between specific of human resource management practices and volun- tar y turnover at the organizational-level of analysis and found a pattern of relationships that support the March and Simon (1958) framework. For people involved in human resource management decisions, our findings bring to the forefront the element of risk involved in human resource investments (Greer 2000). This study also brings to light what may be the most significant predictors of voluntar y turnover rates among a host of human resource management practices. Specifically, our findings suggest that orga- nizations need to focus upon developing policies and instituting practices that allow employees to voice their concerns and that provide opportunities for advancement within the organization. In addition, busi- ness leaders need to consider how they invest in employee training because the often-mentioned fear of losing those you invest in may not be that unreasonable. In sum, the findings we obtained from a large database of workplaces suggest a complex pattern of relationships between human resource management practices and vol- untar y turnover rates, one that deser ves further empirical scrutiny.

244 INDUSTRIAL AND LABOR RELATIONS REVIEW

REFERENCES

Akyeampong, Ernest B. 2002. “Unionization and Fringe Benefits.” Perspective on Labour and Income, Vol. 14, No. 3, pp. 45–50.

Arthur, Jeffrey B. 1994. “Effects of Human Resource Systems on Manufacturing Performance and Turn- over.” Academy of Management Journal, Vol. 37, No. 3, pp. 670–87.

Azfar, Omar, and Stephan Danninger. 2001. “Profit- Sharing, Employment Stability, and Wage Growth.” Industrial and Labor Relations Review, Vol. 54, No. 3, pp. 619–630.

Barnard, Chester I. 1938. The Functions of the Executive. Cambridge, MA: Har vard University Press.

Baron, James N. and David M. Kreps. 1999. Strategic Human Resources: Frameworks for General Managers. New York: John Wiley.

Barron, John M., Mark C. Berger, and Dan A. Black. 1999. “Do Workers Pay for on-the-Job Training?” Journal of Human Resources, Vol. 34, No. 2, pp. 235–52.

Batt, Rosemar y. 2002. “Managing Customer Ser vices: Human Resource Practices, Quit Rates, and Sales Growth.” Academy of Management Journal, Vol. 45, No. 3, pp. 587–97.

____, Alexander J.S. Colvin, and Jeffrey Keefe. 2002. “Employee Voice, Human Resource Practices, and Quit Rates: Evidence from the Telecommunications Industr y.” Industrial and Labor Relations Review, Vol. 55, No. 4, pp. 573–94.

Becker, Brian E. and Barr y Gerhart. 1996. “The Impact of Human Resource Management on Organizational Performance: Progress and Prospects.” Academy of Management Journal, Vol. 39, No. 4, pp. 779–801.

Becker, Gar y S. 1964. Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. New York: National Bureau of Economic Research; distributed by Columbia University Press.

Bennett, Nathan, Terr y C. Blum, Rebecca G. Long, and Paul M. Roman. 1993. “A Firm-Level Analysis of Employee Attrition.” Group and Organization Manage- ment, Vol. 18, No. 4, pp. 482–99.

Benson, George S., David Finegold, and Susan, A. Mohrman. 2004. “You Paid for the Skills, Now Keep Them: Tuition-Reimbursement and Voluntar y Turn- over.” Academy of Management Journal, Vol. 47, No. 3, pp. 315–31.

Blakemore, Arthur. E., Stuart A. Low, and Michael B. Ormiston. 1987. Employment Bonuses and Labor Turnover. Journal of Labor Economics, Vol. 5, No. 4, S124–S135.

Bruce, Kyle. 2005. “Magnus Alexander, the Economists and the Issue of Labour Turnover.” Business Histor y, Vol. 47, No. 4, pp. 493–510.

Bycio, Peter, Rick D. Hackett, and Kenneth M. Alvares. 1990. “Job Performance and Turnover: A Review and Meta-Analysis.” Applied Psychology: An International Review, Vol. 39, No. 1, pp. 47–76.

Campbell III, Carl M. 1993. “Do Firms Pay Efficiency Wages? Evidence with Data at the Firm Level.” Journal of Labour Economics, Vol. 11, No. 3, pp. 442–70.

Cascio, Wayne F. 2000. Costing Human Resources: The Financial Impact of Behavior in Organizations. Cincinnati:

South-Western. Cotton, John L., and Jeffrey M. Tuttle. 1986. “Employee

Turnover: A Meta-Analysis and Review with Implica- tions for Research.” Academy of Management Review, Vol. 11, No. 1, pp. 55–70.

Dearden, Lorraine, Stephen Machin, Howard Reed, and David Wilkinson, D. 1997. Labour Turnover and Work-Related Training. London: The Institute for Fis- cal Studies.

Deler y, John E., and D. Harold Doty. 1996. “Modes of Theorizing in Strategic Human Resource Management: Tests of Universalistic, Contingency, and Configura- tional Performance Predictions.” Academy of Manage- ment Journal, Vol. 39, No. 4, pp. 802–35.

____, Nina Gupta, Jason D. Shaw, G. Douglas Jenkins Jr., and Margot L. Ganster. 2000. “Unionization, Com- pensation, and Voice Effects on Quits and Retention.” Industrial Relations, Vol. 39, No. 4, pp. 625–46.

Doeringer, Peter B., and Michael J. Piore. 1971. Internal Labour Markets and Manpower Analysis. Lexington, MA: Lexington Books.

Fairris, David. 2004. “Internal Labour Markets and Worker Quits.” Industrial Relations, Vol. 43, No. 3, pp. 573–94.

Fang, Tony, and Anil Verma. 2002. “Union Wage Pre- mium.” Perspective on Labour and Income, Vol. 14, No. 4, pp. 17–23.

Folger, Robert, and Russell Cropanzano. 1998. Organiza- tional Justice and Human Resource Management. Thousand Oaks, CA: Sage Publications.

Forrier, Anneleen, and Luc Sels. 2003. “Flexibility, Turn- over and Training.” International Journal of Manpower, Vol. 24, No. 2, pp. 148–68.

Freeman, Ricihard B., and James L. Medoff. 1984. What Do Unions Do? New York: Basic Books.

Gardner, Timothy M. 2005. “Interfirm Competition for Human Resources: Evidence from the Software Industr y.” Academy of Management Journal, Vol. 48, No. 2, pp. 237–56.

Gerhart, Barr y. 1990. “Voluntar y Turnover and Alterna- tive Job Opportunities.” Journal of Applied Psychology, Vol. 75, No. 5, pp. 467–76.

Glebbeek, Arie C., and Erik H. Bax. 2004. “Is High Employee Turnover Really Harmful? An Empirical Test Using Company Records.” Academy of Management Journal, Vol. 47, No. 2, pp. 277–86.

Greer, Charles R. 2000. Strategic Human Resource Man- agement: A General Managerial Approach. 2nd edition, Upper Saddle River, NJ: Prentice-Hall.

Griffeth, Rodger W., Peter W. Hom, and Stefan Gaertner. 2000. “A Meta-Analysis of Antecedents and Correlates of Employee Turnover: Update, Moderator Tests, and Research Implications for the Next Millennium.” Jour- nal of Management, Vol. 26, No. 3, pp.463–88.

Guthrie, James P. 2000. “Alternative Pay Practices and Employee Turnover: An Organization Economics Perspective.” Group and Organization Management, Vol. 25, No. 4, pp 419–39.

____. 2001. “High-Involvement Work Practices, Turnover, and Productivity: Evidence from New Zealand.” Acad- emy of Management Journal, Vol. 44, No.1, pp. 180–90.

HRM PRACTICES AND EMPLOYEE VOLUNTARY TURNOVER RATES 245

Hirschman, Albert O. 1970. Exit, Voice, and Loyalty. Cambridge, MA: Har vard University Press.

Hom, Peter W., and Rodger W. Griffeth. 1995. Em- ployee Turnover. Cincinnati: South-Western College Publishing.

Huselid, Mark A. 1995. “The Impact of Human Resource Management Practices on Turnover, Productivity, and Corporate Financial Performance.” Academy of Manage- ment Journal, Vol. 38, No. 3, pp. 635–72.

Kacmar, K. Michele, Martha C. Andrews, David L. Van Rooy, R. Chris Steilberg, and Stephen Cerrone. 2006. “Sure Ever yone Can be Replaced…But at What Cost? Turnover as a Predictor of Unit-Level Performance.” Academy of Management Journal, Vol. 49, No. 1, pp. 133–144.

Katz, Lawrence F. 1986. Efficiency Wage Theories: A Partial Evaluation. National Bureau of Economic Research, Working Paper No. 1906.

Ledford, Gerald E. Jr., and Edward E. Lawler III. 1994. “Research on Employee Participation: Beating a Dead Horse?” Academy of Management Review, Vol. 19, No. 4, pp. 633–36.

Leonard, Jonathan S. 1987. “Carrots and Sticks: Pay, Super vision, and Turnover.” Journal of Labor Economics, Vol. 5, No. 4, pp. S136–52.

Lincoln, James R., and Arne L. Kalleberg. 1996. “Com-1996. “Com- mitment, Quits, and Work Organization in Japanese and U.S. Plants.” Industrial and Labor Relations Review, Vol. 50, No. 1, pp. 39–59.

Lippman, Steven A., and John J. McCall. 1979. Studies in the Economics of Search. Amsterdam and New York: North Holland.

MacDuffie, John P. 1995. “Human Resource Bundles and Manufacturing Performance: Organizational Logic and Flexible Production Systems in the World Auto Industr y.” Industrial and Labor Relations Review, Vol. 48, No. 2, pp. 197–221.

Maertz, Carl. P., Jr., and Griffeth, Rodger W. 2004. “Eight Motivational Forces and Voluntar y Turnover: A Theoretical Synthesis with Implications for Research.” Journal of Management, Vol. 30, No. 5, pp. 667–83.

March, James G. and Herbert A. Simon. 1958. Organiza- tions. New York: John Wiley.

Mathieu, John E., and Dennis M. Zajac. 1990. “A Review and Meta-Analysis of the Antecedents, Correlates, and Consequences of Organizational Commitment.” Psychological Bulletin, Vol. 108, No. 2, pp. 171–194.

McDonald, John F., and Robert A. Moffitt. 1980. “The Uses of Tobit Analysis.” The Review of Economics and Statistics, Vol. 62, No. 2, pp. 318–21.

McEvoy, Glenn M., and Wayne F. Cascio. 1987. “Do Good or Poor Performers Leave? A meta-Analysis of the Relationship Between Performance and Turn- over.” Academy of Management Journal, Vol. 30, No. 4, pp. 744–62.

Morrow, Paula, and James McElroy. 2007. “Efficiency as a Mediator in Turnover-Organizational Performance Relations.” Human Relations, Vol. 60, No. 6, pp. 827–849.

Park, Hong Y., Joseph Ofori-Dankwa, and Deborah R. Bishop. 1994. “Organizational and Environmental De- terminants of Functional and Dysfunctional Turnover: Practice and Research Implications.” Human Relations, Vol. 47, No. 3, pp. 353–66.

Pfeffer, Jeffrey, and Yinon Cohen. 1984. “Determinants of Internal Labour Markets in Organizations. Admin- istrative Science Quarterly.” Vol. 29, No. 4, pp. 550–72.

Porter, Michael, E. 1980. Competitive Strategy: Tech- niques for Analyzing Industries and Competitors. New York: Free Press.

Powell, Irene, Mark Montgomer y, and James Cosgrove. 1994. “Compensation Structure and Establishment Quit and Fire Rates.” Industrial Relations, Vol. 33, No. 2, pp. 229–48.

Ramsay, Har vie, Dora Scholarios, and Bill Harley. 2000. “Employees and High-Performance Work Systems: Test- ing Inside the Black Box.” British Journal of Industrial Relations, Vol. 38, No. 4, pp. 501–31.

Richard, Orlando C., and Nancy B. Johnson. 2001. “Strategic Human Resource Management Effective- ness and Firm Performance.” International Journal of Human Resource Management, Vol. 12, No. 2, pp. 299–310.

Shaw, Jason D., John E. Delery, G. Douglas Jenkins Jr., and Nina Gupta. 1998. “An Organizational-Level Analysis of Voluntar y and Involuntar y Turnover.” Academy of Management Journal, Vol. 41, No. 5, pp. 511–25.

Simon, Herbert A. 1947. Administrative Behavior. New York: Macmillan.

Spletzer, James R. 1997. “Reexamining the Added Worker Effect.” Economic Inquir y, Vol. 35, No. 2, pp. 417–427.

Spence, Michael. 1973. “Job Market Signaling.” Quarterly Journal of Economics, Vol. 87, No. 3, pp. 355–74.

Stata. 2003. Stata Base Reference Manual. Release 8, Volume 4, S-Z, College Station, Texas: Stata Press.

Stiglitz, Joseph. 1984. Theories of Wage Rigidity. Na- tional Bureau of Economic Research, Working Paper No. 1442.

Strauss, George. 2006. “Worker Participation—Some Under-Considered Issues.” Industrial Relations, Vol. 45, No. 4, pp. 778–803.

Stumpf, Stephen. A., and Patricia, K. Dawley. 1981. “Predicting Voluntary and Involuntary Turnover Using Absenteeism and Performance Indices.” Academy of Management Journal, Vol. 24, No. 1, pp. 148–63.

Trevor, Charlie O. 2001. “Interactions among Actual Ease-of-Movement Determinants and Job Satisfaction in the Prediction of Voluntar y Turnover.” Academy of Management Journal, Vol. 44, No. 4, pp. 621–38.

Way, Sean A. 2002. “High Performance Work Systems and Intermediate Indicators of Firm Performance Within the US Small Business Sector.” Journal of Management, Vol. 28, No. 6, pp. 765–85.

Williams, Charles R., and Linda P. Livingstone. 1994. “Another Look at the Relationship between Perfor- mance and Voluntary Turnover.” Academy of Management Journal, Vol. 37, No. 2 , pp. 269–98.

Wilson, Nicholas, and Michael J. Peel. 1991. “The Impact on Absenteeism and Quits of Profit-Sharing and Other Forms of Employee Participation.” Industrial and Labor Relations Review, Vol. 44, No. 3, pp. 454–68.

Wright, Patrick M., Blaine McCormick, W. Scott Sher- man, and Gary C. McMahan. 1999. “The Role of Human Resource Practices in Petro-Chemical Performance.” The International Journal of Human Resource Management, Vol. 10, No. 4, pp. 551–71.

Wright, Thomas A., and Douglas G. Bonett. 1991.

246 INDUSTRIAL AND LABOR RELATIONS REVIEW

“Growth Coping, Work Satisfaction and Turnover: A Longitudinal Study.” Journal of Business and Psychology, Vol. 6, No. 1, pp. 133–45.

Yalabik, Zeynep, Y., Shyh-Jer Chen, John Lawler, and

Kwanghyun Kim. 2008. “High-Performance Work System and Organizational Turnover in East and Southeast Asian Countries.” Industrial Relations, Vol. 47, No. 1, pp. 145–52

Copyright of Industrial & Labor Relations Review is the property of Cornell University and its content may not

be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.