An Employment Systems Approach to Turnover: Human Resources Practices, Quits, Dismissals, and Performance

TL;DR: The authors examined the relationship between alternative approaches to employment systems and quits, dismissals, and customer service, using cross-sectional, longitudinal data from nationally repre- ected data.

Abstract: This study examines the relationship between alternative approaches to employment systems and quits, dismissals, and customer service, using cross-sectional, longitudinal data from nationally repre...

Summary

INTRODUCTION

• Turnover is among the most important employment relations outcomes for both employees and organizations.
• The costs of operational disruption and replacement of human capital versus the benefits of lower-cost labor, or fresh skills to compete on innovation.
• Here, quits and dismissals are viewed as related phenomena with similar antecedents deriving from the type of employment system that employers choose.
• They also employ over 4 million workers in interactive voice services (3 percent of the US workforce), compared to less than 400,000 comparable employees in India and the Philippines (Batt, Holman, & Holtgrewe, 2009: 465) .

THEORY AND HYPOTHESES

• The theory that quits and dismissals have distinct antecedents draws on the early work of Dalton and others, who distinguished between 'functional' turnover ( where poor performers leave) and 'dysfunctional' turnover (where high performers leave) (Dalton, Krackhardt, & Porter 1981; Abelson & Baysinger, 1984) .
• Donaghue and Castle (2006) found several differences between quits and dismissals, but noted that "…most of the organizational characteristics held similar associations to voluntary and involuntary turnover…. " (2006:467) .
• While some of their predictions were confirmed, the paper raised other questions.
• By contrast, a high turnover model may be acceptable or even desirable for cost-focused firms if lower operational costs outweigh higher turnover costs.
• In the sections below, the authors examine specific HR practices, and their theoretical relationship to quits, dismissals, and operational performance.

Work Organization

• Variation in work organization has grown in recent decades as employers have used different approaches to compete on price, quality, innovation, or time-to-market.
• These literatures also assume that problem-solving is best achieved through group collaboration: Under TQM, quality circles that discuss problems 'off-line' (away from work stations); and under STS, work organized into 'online' teams, with considerable operational autonomy.
• While theory and evidence provide reasonable support that high involvement work design reduces quit rates (Arthur, 1994; Batt, 2002; Huselid, 1995) , the predictions for dismissals are less clear.
• Dismissals may be higher than in Taylorized operations if those employees who have fewer skills or abilities are given more discretion, and they stand out as unable to perform; or if peer-monitoring exposes free-riders, leading to their discipline or dismissal (Barker, 1999) .
• A less generous argument yields a similar hypothesis.

Long-term Incentives: Investment and Inducement Strategies

• The concept of investments and inducements draws on internal labor market theory which posits that higher relative pay, benefits, internal promotion opportunities, and procedures for employment security provide long-term incentives that lead employees to stay with the organization (Doeringer & Piore, 1971) .
• Firms that invest more in the workforce may be reluctant to dismiss workers with this accumulated human capital (Klaas, Brown & Heneman, 1998) , unless the employee misbehavior is egregious.
• Finally, employers who attempt to compete by building a culture of trust and the synergies of social capital (Leana & Van Buren III, 1999 ) may try to limit overall turnover as much as possible because it disrupts the social fabric of work.
• Empirical support for these arguments is mixed.
• In sum, the bulk of existing theory and evidence suggest that inducements and incentives should have the same relationship to quits and dismissals: Hypothesis 2: Employment practices emphasizing inducements and investments will be associated with lower quit rates and lower dismissal rates.

Short-term Incentives: Performance-enhancing Expectations

• The concept of performance-enhancing expectations (Shaw et al., 1998; Tsui et al., 1997) has advanced the literature by identifying distinct parts of the HR system that may differentially affect employee performance.
• They may include behavior-oriented measures (electronic monitoring, performance appraisals) and outcome-oriented measures, such as pay linked to employee performance.
• Employees typically view ongoing electronic monitoring as an invasion of privacy and an indication of lack of trust, leading even 'good performers' to be dissatisfied and quit.
• Thus, the impact of commission pay on quit rates is likely to depend on how risk averse workers are.
• Nadler and Wiswall (2009) , for example, found that teacher support for a policy shift from scheduled, fixed-pay increases to merit pay was higher in school districts that offered higher base salaries and had a higher proportion of teachers with masters' degrees (as opposed to bachelors' degrees).

HR Practices, Quits, Dismissals, and Performance

• Theory and evidence on the relationship between HR practices and organizational performance has expanded considerably in the last two decades, although questions remain unanswered.
• Commitment-enhancing incentives motivate employees to use their skills and discretionary effort to advance the interests of the organization.
• The concept of performance-enhancing expectations has not been part of the high involvement literature.
• High involvement work organization, investment and inducement practices, and performance-enhancing expectations will each individually be associated with higher levels of operational performance, also known as Hypothesis 4a.
• Theory also predicts that the synergies from these three types of HR practices should be multiplicative, but some studies have found evidence of interactions (Huselid, 1995; Shaw et al., 2009) , while others have not (e.g. Cappelli and Neumark 2004) .

Turnover and Performance

• The theory linking turnover and organizational performance is more equivocal (Bluedorn 1982; Osterman, 1987) .
• The phenomenon also may be context specific.
• Regardless of whether an employee quits or is fired, recruitment and training costs are the same.
• The preponderance of evidence is that total turnover and voluntary turnover are both negatively related to operational performance, and specifically to service quality and customer satisfaction (Kacmar, Andrews, Van Rooy, Steilberg, & Cerrone, 2006; McElroy et al., 2001) .

Sample

• The data for this study come from a 1998 stratified nationally-random survey of call centers in the telecommunications industry, based on the Dun and Bradstreet listing of establishments, and a 2003 nationally-random survey of call centers across all industries.
• Limiting the sample to organizations providing the same types of services enhances comparability and reduces extraneous sources of variation and measurement error (Becker & Gerhart, 1996: 792; Osterman, 1994) .
• The longitudinal dataset includes telecommunications industry establishments surveyed in both 1998 and 2003; but due in part to the dot.com bust in the intervening period, only 40 percent of the 1998 sample was able to be identified in 2003.
• Identifying the 2003 population of call centers across industries was difficult because there is no official data on call centers, and many are located within larger organizations.

Measures 1

• The four dependent variables provide indicators of quits, dismissals, total turnover, and customer satisfaction.
• The authors measured high involvement work organization by three variables: Investment and inducement practices are a slightly modified version of Shaw et al.'s measure (1998; 2009) , which captures the use of internal labor market practices (internal mobility opportunities, relative pay, pensions, and employment security).
• This definition of pay measures the relative attractiveness of the current job compared to local alternatives, which is likely to influence turnover decisions.
• How often core employees receive statistics on performance; how often their calls are listened to by supervisors; and how often they get feedback on phone technique (frequency scale of 1-8, where 1 = never and 8 = daily) (Alpha = .66).

Control variables.

• The authors included several variables to control for market, organizational, and workforce characteristics.
• The local unemployment rate controls for variation in labor market conditions:.
• The authors control for industry (telecommunications versus others) as telecommunications is over-represented in the sample.
• The authors also control for the age of the establishment, as older establishments may have more stable employment systems; and whether it had recent layoffs (in the last year), as research has shown that downsizing may undermine employee commitment, leading to higher quit rates (Trevor & Nyberg, 2008) .
• The percentage of the workforce that is female (percent female); the average number of years of education of the typical worker at the site (average education) (Bretz, Boudreau, & Judge, 1994; Trevor 2001 ); and the number of weeks of initial training and on-the-job experience for workers to become fully qualified (initial training).

RESULTS

• Table 1 presents the means, standard deviations, and bi-variate correlations for the variables.
• The average annual quit rate in this sample is 14.9 percent, the dismissal rate is 9.8 percent, and average total turnover is 24.7 percent (with a range of close to zero to over 100 percent).
• Table 2 presents the estimation equations for the first three dependent variables (quits, dismissals, and total turnover), using the cross-sectional sample.
• To further investigate these relationships, the authors now turn to their multivariate models.

Multivariate Analyses

• Estimation of multivariate models for their dependent variables requires accounting for the functional form of these variables, which are measured as rates, and therefore have a lower bound of zero (negative values are impossible).
• The authors also found strong support for hypothesis 2: The investments and inducements index is significantly associated with lower quits (p<.05, model 2a, table 2), dismissals (p<.05, model 2b, table 2), and total turnover (p<.01, model 2c, table 2).
• The results of the longitudinal analysis mirror those found in Table 2 , but as expected, the significance levels are lower given the much smaller sample size of 93 establishments.
• When both the high involvement index and total turnover are included in the model, total turnover is a significant predictor of customer satisfaction (step3, model 6, table 4 ).

Additional Analyses

• As a check on the robustness of their findings the authors also estimated models using the individual practices that make up these indexes.
• This provides some additional support for their argument that these two practices have similar effects and can be linked together.
• Relative to no pay at risk (the largest group), each of the 3 groups representing the highest levels of pay at risk (10-25%; 26-50%; 51-100%) had significantly higher quits, dismissals, and total turnover and the magnitude of the coefficients for the groups increased as the level of pay at risk increased.
• The research record is mixed on whether the turnover-performance relationship is linear, curvilinear, or contingent on other factors (Alexander et al., 1994; Kacmar et al., 2006; Shaw, Gupta, & Delery, 2005; Siebert & Zubanov, 2009) .
• Among the three interaction terms, one reached a modest level of significance (p<.10): the interaction of turnover and the high involvement index was negatively related to customer satisfaction.

DISCUSSION

• In this study the authors took an employment systems approach to understanding how human resource practices affect turnover and performance.
• The authors found that quits and dismissals have similar antecedents in HR practices.
• While call centers have a widespread reputation of being 'electronic sweatshops', employing low-skilled workers with little attachment to the labor force, in fact the authors observed a considerable range of variation in business and HR strategies and turnover in the centers they visited.
• Finally, a general limitation of this study is that the data are based on reports from the general manager at each establishment, so the authors are unable to check for inter-rater reliability.

CONCLUSION

• The authors study advances the past literature in two main areas.
• First, contrary to past research in this area, the authors found strong similarities in the antecedents of both quits and dismissals, suggesting that they are influenced by common processes relating to the type of employment system that exists in each workplace.
• And clearly, the dismissal of individual employees for poor performance may improve organizational performance if they are replaced by more capable individuals.
• This study presents a plausible theory and suggestive evidence that the relationship between HR practices, quits, and dismissals may function in similar ways in some circumstances.

Full text

An Employment Systems Approach to Turnover:
HR
Practices,
Quits,
Dismissals,
and Performance
May 2011
ROSEMARY BATT
rb41@cornell.edu
and
ALEXANDER J. S. COLVIN
ajc22@cornell.edu*
*The authors are listed alphabetically to reflect equal contribution to authorship. Rosemary Batt is
the Alice H. Cook Professor of Women and Work, and Alexander J.S. Colvin is associate professor, both
at Cornell University. The research is based on a multi-year study of call center operations conducted in
collaboration with Harry C. Katz and Jeffrey Keefe and generously funded by the Alfred P. Sloan
Foundation.
Copies of the computer programs used to generate the results in the paper may be obtained from the
authors at the Industrial and Labor Relations School at Cornell University, Ithaca, NY
14853-3901,
607-
254-8206 (ph), ajc22@cornell.edu.

1
AN EMPLOYMENT SYSTEMS APPROACH TO TURNOVER:
HR PRACTICES, QUITS, DISMISSALS, AND PERFORMANCE
Abstract
This study examines the relationship between alternative approaches to employment
systems and quits, dismissals and customer service, based on cross-sectional and longitudinal
data from nationally representative surveys of call center establishments. Contrary to prior
literature, the antecedents and consequences of quits and dismissals are quite similar. Comparing
three dimensions of employment systems, we find that high involvement work organization and
long-term investments and inducements are associated with significantly lower quit and dismissal
rates,
while short term performance-enhancing expectations are related to significantly higher
quit and dismissal rates. Establishments with higher quit and dismissal rates have significantly
lower customer service, as reported by managers.

2
INTRODUCTION
Turnover is among the most important employment relations outcomes for both
employees and organizations. Employees face the uncertainties of job search and potential
unemployment. Organizations confront uncertain costs and benefits: The costs of operational
disruption and replacement of human capital versus the benefits of lower-cost labor, or fresh
skills to compete on innovation. These uncertainties raise the question of which management
practices increase turnover; how much turnover is harmful or beneficial, and under what
conditions and why? These questions have spawned a small, but growing set of organization
level studies of the antecedents and consequences of turnover. This research complements the
large micro-organizational literature on how and why individuals choose to quit their jobs
(Griffeth, Hom, & Gaertner, 2000; Hom & Griffeth, 1995; Maertz & Campion 2004).
However, research has largely focused on the issue of voluntary turnover, or total turnover
more generally, neglecting involuntary turnover in the form of dismissals. Here, we define
dismissals as employer decisions to fire individual employees, rather than decisions to cut costs
through mass layoffs, downsizing, early retirement buyouts, or organizational restructuring. The
few existing studies of dismissals have conceptualized them as distinct from quits, with different
antecedents and consequences (Dalton & Todor 1979; Donaghue & Castle, 2006; Knight &
Latreille, 2000; Shaw et al. 1998). While dismissals are viewed as the result of hiring errors,
which employers may correct by terminating poor performers, quits are seen as a relatively
rational process in which employees weigh their current job against alternative prospects. The
literature often assumes that quits undermine performance because better performers are more
likely to quit than poor performers, but theory and evidence are equivocal (McEvoy & Cascio,
1987;
Osterman, 1987; Trevor, Gerhart, & Boudreau, 1997).

3
This paper draws on an alternative perspective, internal labor market theory (Doeringer &
Piore
1971;
Osterman, 1987), which focuses on how employers establish sets of internal
administrative rules (e.g., HR practices or employment systems) to advance their goals. Here,
quits and dismissals are viewed as related phenomena with similar antecedents deriving from the
type of employment system that employers choose. Employers may manage turnover levels via
their choice of HR practices; and the optimal turnover level may vary by market, industry
conditions (Abelson & Baysinger, 1984), business strategy, or values (Baron, Burton, & Hannan,
1996).
Cost-focused firms, for example, may use high turnover to minimize labor costs. Jack
Welch at GE, by contrast, built a ‘performance culture’ through an explicit policy of dismissing
10 percent of employees each year (Welch, 2005: 37-52).
To investigate these alternative theories, we examine the relationship between quits and
dismissals and three dimensions of the employment system: work organization, commitment-
inducing long term incentives, and performance-enhancing short-term incentives. We also
explore their relationship to customer satisfaction, an increasingly important outcome for firm
competitiveness (Fornell, Van Amburg, Morgeson, & Bryant, 2005; Lovelock & Wirtz, 2005).
We draw on a 2003 national survey of US call centers, as well as longitudinal data for a subset of
these establishments. Call centers are an appropriate context for this study because high rates of
turnover are a major source of high costs and poor service quality. Despite off-shoring trends,
they also employ over 4 million workers in interactive voice services (3 percent of the US
workforce), compared to less than 400,000 comparable employees in India and the Philippines
(Batt, Holman, & Holtgrewe, 2009: 465).

4
THEORY AND HYPOTHESES
The theory that quits and dismissals have distinct antecedents draws on the early work of
Dalton and others, who distinguished between ‘functional’ turnover (where poor performers
leave) and ‘dysfunctional’ turnover (where high performers leave) (Dalton, Krackhardt, & Porter
1981;
Abelson & Baysinger, 1984). Dismissals are equated with functional turnover while quits
are equated with dysfunctional turnover (Jacofsky, 1984; Sturman, Trevor, Boudreau, & Gerhart,
2003).
McElroy and colleagues, for example, argued that, “Dismissals are a function of poor
individual performance or of insubordination….” (2001:1294). Shaw and colleagues noted that,
“In an organization with high quit rates, for various reasons employees find it more attractive to
leave than to stay. In an organization with high discharge rates, however, presumably incorrect
hiring decisions are remedied through termination” (1998:512). Of course, poor performance
also may be due to other factors than hiring errors – new production systems may render
employee skills obsolete, employee motivation or commitment may change.
The assumption that quits and dismissals have distinct antecedents is theoretically
plausible, but has not been systematically tested, and existing studies show mixed results.
Wanous,
Stumpf,
and Bedrosian (1979) found that job attitudes and performance were
significant predictors of both types of turnover, but job attitudes explained more variance for
quits,
and performance more variance for dismissals. Organizational factors (job training and
pay) accounted for the most variance for both types of turnover. Stumpf and Dawley (1981)
found that absenteeism and job performance significantly explained the incidence of both quits
and dismissals, but absenteeism explained a higher proportion of variance in dismissals.
Recent organization-level research also provides mixed results based on industry-specific
studies of trucking (Shaw et al. 1998) and nursing homes (Donaghue & Castle, 2006). Donaghue