You are on page 1of 19

DOI: 10.1002/hrm.

21990

ORIGINAL ARTICLE

Human resource management systems, employee well-being,


and firm performance from the mutual gains and critical
perspectives: The well-being paradox

Hoang Ho | Bård Kuvaas

BI Norwegian Business School, Oslo, Norway


Abstract
Correspondence In this study, we explored the additive, interactive, and nonlinear relationships among
Hoang Ho, Department of Leadership and
Organizational Behaviour BI Norwegian human resource management (HRM) systems, employee well-being, and firm perfor-
Business School, Oslo N-0484, Norway. mance. Based on a sample of 14,384 employees nested within 1,347 firms, we
Email: hoang.cbs@gmail.com
obtained three main findings. First, HRM systems yield a performance effect that
exceeds the effect of single practice, suggesting positive synergies among HRM prac-
tices. Second, the opportunity bundle has a positive impact on firm performance, but
when integrating it with skills and motivation bundles, the result becomes negative,
indicating dis-synergy of interactions among HRM bundles. Third, at moderate levels
of adoption, HRM practices are positively correlated with employee well-being and
higher levels of commitment, job satisfaction, and management relations, as well as
lower levels of anxiety. However, at high levels, the relationship is less positive and
even turns negative with lower levels of job satisfaction and management relations.
To close, we present research implications and future directions after discussing our
results.

KEYWORDS
employee well-being, firm performance, HRM systems

1 | I N T RO D UC T I O N 2009; Guest, 2017; Van De Voorde, Paauwe, & Van Veldhoven,


2012), as HRM scholars have increasingly recognized that taking care
Concern for employee well-being was first voiced in the early human of employee well-being is important from an ethical perspective
resource management (HRM) literature; for example, Beer, Spector, (Guest, 2017) and that there is empirical evidence suggesting that
Lawrence, Mills and Walton (1984) suggested that employee well- employee well-being may have positive implications for firm perfor-
being should be the long-term consequence considered when design- mance (Daniels & Harris, 2000).
ing an HRM system, and Legge (1998) argued that HRM may result in Two competing views have emerged. The proponents of one view
morally problematic issues in cases where it leads to the exploitation argue that HRM systems benefit both employers and employees
of workers. However, following these early concerns, employee well- (e.g., Kochan & Osterman, 1994; Levine, 1995; Pfeffer, 1998), denoted
being has not become a central research agenda within the field, as by labels such as “shared capitalism” (Kruse, Blasi, & Park, 2010), “high-
HRM scholars have mainly focused on the link between HRM and involvement” (Lawler, 1992), “mutual gains” (Kochan & Osterman,
performance, often known as the HRM–performance paradigm 1994), or “high commitment” (Walton, 1985). Following Kochan and
(Boselie, Dietz, & Boon, 2005). Osterman (1994), we label this view “the mutual gains perspective,” as
More recently, there has been heated debate regarding the impact it conveys the key message that both employers and employees bene-
of HRM systems on employee health or well-being (Boxall & MacKy, fit from HRM systems. Proponents of the other view argue that HRM

Hum Resour Manage. 2019;1–19. wileyonlinelibrary.com/journal/hrm © 2019 Wiley Periodicals, Inc. 1


2 HO AND KUVAAS

systems benefit employers but not employees (Delbridge & Turnbull, 1994; Huselid, 1995; Ichniowski, Shaw, & Prennushi, 1997; Jiang,
1992; Godard, 2001; Legge, 2005). We label this view “the critical Lepak, Hu, & Baer, 2012; Macduffie, 1995; Subramony, 2009; Way,
perspective.” 2002) define HRM systems as an integrated set or cluster of HRM
Unfortunately, the nature of the relationship between HRM sys- practices that have the potential to achieve substantially enhanced
tems and well-being is not accurately demarcated by existing empiri- economic performance. Such HRM practices typically include sophisti-
cal evidence, because such evidence has not conclusively confirmed cated selection procedures, developmental performance appraisals,
or ruled out either the mutual gains perspective or the critical per- substantial investments in training, teamwork, extensive communica-
spective (Harley, Sargent, & Allen, 2010). On the one hand, some tion, motivating job design (e.g., autonomy, empowerment, employee
studies have documented positive associations with employee experi- participation, and flexible work), performance-related pay/promotion,
ence of work (Appelbaum, Bailey, Berg, & Kalleberg, 2000; Butts, Van- harmonization, and employment security. These practices influence
denberg, DeJoy, Schaffer, & Wilson, 2009; Castanheira & Chambel, firm performance by enhancing employees' knowledge, skills, and abili-
2010; Guest, 2002; Harley, Allen, & Sargent, 2007; Macky & Boxall, ties, providing employees the opportunities to use those attributes for
2007). On the other hand, some have revealed negative associations organizational benefit and increasing their motivation to do so (Boselie
(Godard, 2001; Jensen, Patel, & Messersmith, 2011; Kroon, Voorde, & et al., 2005; Combs, Liu, Hall, & Ketchen, 2006; Huselid, 1995).
Veldhoven, 2009; Landsbergis, Cahill, & Schnall, 1999). In addition, In addition, inherent in the systems perspective is the notion of
some studies have indicated a mix of both positive and negative con- complementary resources or synergies; that is, “individual policies or
sequences (e.g., Berg, Appelbraum, Bailey, & Kalleberg, 1996; Ramsay, practices have limited ability to generate competitive advantage in
Scholarios, & Harley, 2000). isolation, but in combination… they can enable a firm to realize its full
We suspect that such conflicting findings, which we refer to as the competitive advantage” (Becker & Gerhart, 1996, p. 784). Conse-
well-being paradox, are attributable to two main reasons. The first rea- quently, to capture the synergies when HRM practices are bundled
son is that past studies have measured well-being as a single dimension together into a system, most studies have created “a unitary index
(Peccei, van de Voorde, & Van Veldhoven, 2013), although there is that contains a set (though not always the same set) of theoretically
wide scientific consensus that well-being has multiple dimensions appropriate HRM practices derived from prior work” (Becker &
(Diener, 1994; Wright, 2014) and that there are often tradeoffs among Huselid, 1998, p. 63).
these dimensions “whereby one aspect of employee well-being In this study, we defined HRM systems in terms of nine HRM
improves but another aspect of employee well-being decreases” (Grant, practices: selective hiring, teamwork, job autonomy, staff training,
Christianson, & Price, 2007, p.51). In the current study, we examine flexible work, participatory decision-making, information sharing, sup-
multiple dimensions of well-being, including the construct's aspects of portive management, and performance-related pay. The selection of
happiness, health, and social relationships. A second reason is that past HRM practices was based on previous studies (Appelbaum et al.,
studies assume a linear relationship among HRM systems, well-being, 2000; Combs et al., 2006; Ichniowski et al., 1997; Way, 2002) and
and performance, whereas logic suggests that the relationship may be particularly on those in the previous analysis of the WERS data
nonlinear. For example, Cappelli and Neumark (2001) argued that HRM (Guest & Conway, 2007; Ogbonnaya et al., 2017; Ramsay et al.,
systems have both value-creating and cost-enhancing effects. That is, 2000). Consistent with the literature, we combined these HRM prac-
at higher levels, the positive effects of HRM systems on well-being and tices and created an HRM system unitary index.
performance may be neutralized or even diminished as the costs associ-
ated with the adoption of these systems substantially increases. Initial
2.2 | Employee well-being
evidence indicates a nonlinear relation between HRM systems and
well-being (Godard, 2001). However, the extant literature has mostly Employee well-being is an elastic concept meaning “any number of
tested the HRM, well-being, and performance relationship in a linear things to various people” (Danna & Griffin, 1999; Wright & Huang,
fashion, thus leading to incorrect inferences. 2012, p. 1188). In a broad fashion, employee well-being “refers to
Accordingly, the overall objective of this study is to contribute to people's evaluations of their lives” (Diener, Suh, Lucas, & Smith, 1999,
HRM research by examining the current well-being paradox. We do p. 213) or “all the things that are important to how we think about
this by (a) examining multiple dimensions of well-being, including hap- and experience our lives” (Rath & Harter, 2010, p. 137). In a narrow
piness, health, and social relationships, and (b) examining the additive, fashion, employee well-being is restricted to one dimension, such as
interactive, and nonlinear relationships among HRM, well-being, and job satisfaction (Cropanzano & Wright, 2001; Grant et al., 2007). At
performance. the operational level, variations in conceptualization are also evident.
Earlier conceptualizations of employee well-being, particularly within
the psychological tradition (e.g., Andrews & Withey, 2012; Bradburn,
2 | D EF I NI NG TH E KE Y C ON CE P T S
1969; Campbell, 1981; Diener, 1984), focused exclusively on pleasant
emotional experiences as a fundamental dimension of employee well-
2.1 | HRM systems
being, often described in academic research as “subjective well-being”
Although there is no universal definition of HRM systems (Datta, Guth- (Diener, 1994) or “psychological well-being” (Wright, Cropanzano, &
rie, & Wright, 2005), most empirical and theoretical works (e.g., Arthur, Bonett, 2007). According to this view, psychological well-being is said
HO AND KUVAAS 3

to occur if a person “experiences frequent positive emotions such as De Winne and Sels (2013, p. 181), has been universally accepted for
joy and happiness and infrequent negative emotions such as sadness the last 30 years. This conventional assumption of “more is better”
and anger” (Bakker & Oerlemans, 2011, p. 179; Diener & Larsen, has led to many HRM scholars to exclusively propose and test
1993). Consequently, employee well-being is operationalized as the hypotheses describing linear relationships between HRM and firm
presence of a dispositional positive effect and the absence of a dispo- performance, often referred to as HRM–performance paradigm
sitional negative effect (Diener et al., 2010). Over the years, additional (Boselie et al., 2005).
dimensions have been added to psychological well-being, such as Although there is still an ongoing debate about the precise mecha-
social relations (Keyes, 1998; Larson, 1996), self-validation (Warr, nisms aligning HRM practices with firm performance, the majority of
2011), and health (Danna & Griffin, 1999). scholars agreed that HRM practices impact firm performance through
In this study, we adopted the definition of Grant et al. (2007), improving workforce skills, providing employees with the opportunity
defining well-being as “the overall quality of an employee's experience to contribute to firm success, and motivating them to perform
and functioning at work” (p. 52), which can conceptually be assessed in (Boselie et al., 2005; Combs et al., 2006; Huselid, 1995; Subramony,
terms of three dimensions: happiness, health, and social relationships 2009). For example, HRM practices, such as selective hiring and train-
(Van De Voorde et al., 2012). Happiness refers to an employee's expe- ing, are intended to enhance employee skills (Jiang, Lepak, Hu, & Baer,
rience and functioning at work (Grant et al., 2007). We focused on 2012). Participatory and nonauthoritarian practices (e.g., teamwork,
commitment and job satisfaction as measures of happiness at work, delegation of decision-making, information sharing, and employee
because they “are frequently used in conceptual models and empirical involvement) are intended to “allow employees to have direct input
studies” (Van De Voorde et al., 2012, p. 394). In organization science, into production process” (Huselid, 1995, p. 647). Finally, motivation
health has been studied both in terms of physical health, such as inju- practices (e.g., contingent compensation) help to reinforce desired
ries and diseases, and mental health, such as job-related anxiety, stress, behaviors.
burnout, and exhaustion (Danna & Griffin, 1999; Grant et al., 2007). Moreover, there is consensus among scholars that a system of
With respect to mental health, Van De Voorde et al. (2012) made a dis- HRM practices is more than the sum of the parts due to complemen-
tinction between stressors and strain. “Stressors (e.g., workload and tarities or synergies among HRM practices (Barney, 1995; Delery,
work intensification) refer to events or situations that give rise to 1998). In other words, combinations of practices into a system or bun-
stress, whereas strain (e.g., stress and burnout) refers to responses to dle rather than individual practices in isolation are the critical determi-
stressors” (Van De Voorde et al., 2012, p. 394). We focused on work nants of firm performance (Ichniowski et al., 1997; Macduffie, 1995).
intensification and anxiety as measures of mental health as they are Likewise, research has shown that HRM systems yield an effect that
“among the most common indicators of employee health and well- exceeds that of a single practice (Combs et al., 2006; Laursen & Foss,
being” in the HRM literature (Ogbonnaya, Daniels, Connolly, & van 2003). Therefore, we propose the following:
Veldhoven, 2017, p. 102–103). Finally, social relationships address the
quality of relationships at work between employees (e.g., cooperation) Hypothesis 1: The positive relationship between
or between employees and their leader or the employment organiza- HRM systems and firm performance is greater for
tion (e.g., trust, social support, reciprocity, leader-member exchange, HRM systems than for individual HRM practices in
and perceived organizational support) (Grant et al., 2007; Guest, 2017). isolation.
We included social relationships, because they have an impact on well-
being (Danna & Griffin, 1999) and are “increasingly being incorporated In regard to the relationship between HRM systems and employee
in conceptual models (Van De Voorde et al., 2012, p. 394). We focused well-being, there is less agreement. On the one hand, some authors
on the relationships between employees and their leaders (here mea- (e.g., Kochan & Osterman, 1994; Levine, 1995) argue that HRM systems
sured through perceptions of good management relations) as a mea- benefit employees and employers (the mutual gains perspective). On
sure of social relationships because they are conceptually relevant the other hand, other scholars (e.g., Delbridge & Turnbull, 1992; Legge,
(Danna & Griffin, 1999; Grant et al., 2007) and have been used in prior 2005) claim that HRM systems benefit employers but not employees
studies (e.g., Ramsay et al., 2000). It should be noted here that our defi- (the critical perspective). The main dispute between the two perspec-
nition of well-being in terms of happiness, health, and social relation- tives is whether HRM systems have a positive or negative effect on
ships is conceptually consistent with the World Health Organization's employee health and well-being. Central to the mutual gains perspective
(WHO) definition of health as “a state of complete physical, mental, is the claim that HRM systems are generally good for employees and
and social well-being” (Larson, 1996, p. 181). their well-being. Such a claim is mainly based on the observations show-
ing that HRM systems are positively associated with various dimensions
of well-being, such as commitment and job satisfaction (i.e., happiness)
3 | T H E O R Y A N D H Y P O T H E S ES
(Takeuchi, Chen, & Lepak, 2009), reduced job-related stress (i.e., health)
(Macky & Boxall, 2008), and social relations (Collins & Smith, 2006). In
3.1 | Traditional wisdom
contrast, based on the assumption of employers' desire to extract
The sentiment that the greater the use of HRM is, the higher firm per- increasing returns from labor (Keenoy, 1990; Willmott, 1993), the criti-
formance is (Kaufman, 2012), or “more is better” in the language of cal perspective predicts that HRM systems lead to work intensification
4 HO AND KUVAAS

(Legge, 2005). This work intensification may increase performance but Drawing on the notions of bundling and economies of scale, we
can have negative implications for workers (Peccei et al., 2013). propose that there is a small positive effect on well-being and firm
Supporting the critical view, research has demonstrated a positive asso- performance at a low level of implementation and that the positive
ciation between HRM systems and higher levels of burnout, stress, and effect increases as more HRM practices are added to the scale of
workload (Jensen et al., 2011; Kroon et al., 2009).Thus, we propose the HRM practices due to labor cost per unit of output decreasing with
following: increasing scale (i.e., economies of scale). However, at extremely high
levels of implementation, we predict that the effect is less positive
Hypothesis 2: HRM systems are positively associated and even turns negative due to the significant magnitude of the addi-
with employee well-being. tional costs associated with overimplementation. Scholars have long
noted that, although HRM systems increase widespread economic
Hypothesis 3: HRM systems are negatively associated benefits, they also increase costs (Bryson et al., 2005; Cappelli & Neu-
with employee well-being. mark, 2001; Godard, 2004). The costs stem from complicated selec-
tion and recruitment practices, continuous investment in training
programs, higher wages, job security guarantees, and other supporting
practices (Chi & Lin, 2011; Godard, 2004). These costs may offset the
3.2 | Nonlinear effects
benefits of HRM systems (Cappelli & Neumark, 2001), indicating that
We challenge the conventional assumption of linearity or more is bet- there is an optimal point at which further investments in HRM sys-
ter. Drawing on Pierce & Aguinis' (2013) meta-theoretical principle of tems are not as valuable (Chadwick, 2007; Chi & Lin, 2011). Econo-
the too-much-of-a-good-thing effect (TMGT effect), we argue that mists label this optimal point “diminishing returns to scale.” Godard
when HRM practices are implemented at extremely high levels, the (2001) provided empirical evidence of diminishing returns to scale of
widely accepted positive effects on well-being and firm performance HRM systems. Godard found that, at moderate levels, alternative
of HRM systems diminish and even turn negative, resulting in a curvi- work practices were associated with an increased sense of belonging,
linear relationship. empowerment, task involvement, and ultimately job satisfaction,
There are at least two compelling reasons to believe that the rela- esteem, commitment, and citizenship behavior. At higher levels of
tionships among HRM, well-being, and firm performance are curvilin- adoption, however, these associations declined in magnitude and even
ear. The first reason is the notion of HRM systems. The basic idea became negative. Similarly, Chi and Lin (2011) found that high-
behind systems is the notion of complementarity or synergy in the performance work systems are positively correlated with firm perfor-
sense that individual HRM policies or practices “have limited ability to mance at low levels of implementation, but that this relationship
generate competitive advantage in isolation, but in combination… they became weaker at higher levels for high-technology firms. Therefore,
can enable a firm to realize its full competitive advantage” (Becker & we hypothesize the following:
Gerhart, 1996, 784). This observation is because HRM practices are
complementary and mutually supportive; if they are bundled together Hypothesis 3: The relationship between HRM sys-
into a cluster or system, they create synergistic effects that are sub- tems and (a) firm performance and (b) well-being is
stantially greater than those of individual HRM practices in insolation curvilinear, with positive effects at low to moderate
(Becker, Huselid, Pickus, & Spratt, 1997; Delery, 1998). According to levels of implementation but negative effects at higher
Godard (2001) and White and Bryson (2011), the idea that bundling levels.
complementary practices together into a coherent system may have
effects exceeding the sum of the parts suggests nonlinear effects that
occur at particular thresholds. In other words, the HRM systems' per-
3.3 | Interaction effects
spective suggests that implementing a wide range of practices, partic-
ularly ones bundled into a coherent HRM system, “will yield higher Our operationalization of an HRM system in terms of a unitary index
returns than an application comprising fewer such practices” (Becker implicitly assumes that relationships among practices constituting
et al., 1997; Bryson et al., 2005, p. 491). HRM systems are additive. However, practices within systems may
The second reason is what economists have called economies of also have interactive relationships (Barney & Felin, 2013; Delery,
scale. The notion of economies of scale suggests that firms benefit 1998) (we would like to thank Motohiro Morishima for drawing our
more from large-scale deployment of an HRM system than small-scale attention to this point). Delery (1998) has identified three forms of
deployment of the same system entailing limited practices, for “the interactive relationships. The first form of interactive relationship is
fixed portions of an HRM practice's administrative expenses are substitutes. Two practices are said to be substitutes when “both prac-
spread out over larger proportions of an organizational work force” tices lead to the identical outcomes” and if one practice is already in
(Chadwick, 2007, p. 502), which explains the marginal value added rel- use, adding the second practice “will add nothing except the expense
ative to the cost of isolated HRM practices. Empirical evidence has associated with its implementation” (Delery, 1998, p. 293). The sec-
shown that implementing a larger scale of HRM practices will yield ond form of interactive relationship is positive synergy. Two practices
higher returns than that of a small scale (Ichniowski et al., 1997). are said to have a positive synergistic effect when “two practices work
HO AND KUVAAS 5

together to enhance each other's effectiveness” (Delery, 1998, p. 294) their decision-making is constrained by organizational structure.
or the sum is greater than its parts (Ichniowski et al., 1997; Macduffie, Macduffie (1995) even goes so far as to argue that to maximize returns
1995). Becker et al. (1997) have called this situation “powerful con- from investments in HRM systems, firms must adopt all three bundles.
nection.” Finally, the third form of interactive relationship is negative Supporting his view, Macduffie (1995) found that interactions among
synergy. Two practices are said to have a negative synergistic effect HRM bundle, production systems bundle, and use of buffers bundle
when they work against one another; if combining together, “the two have significantly larger effects on assembly plant productivity and qual-
practices will have poor performance” than if they are employed alone ity than the effects of individual bundles. Following this line of reason-
(Delery, 1998, p. 294) or the sum is less than its parts. Becker et al. ing, we argue interactions among the skills, opportunity, and motivation
(1997) have called this situation “deadly combination.” bundles will yield a combined positive effect that exceeds that of a single
To capture these different types of interactive effects, we decompose bundle due to the positive synergies among them. Delery (1998) has
HRM systems into skills, opportunity, and motivation bundles based on called this situation positive synergy and Becker et al. (1997) powerful
the AMO model. According to Boselie et al. (2005), the literature has connection. Thus, we propose the following:
shown a preference for using the AMO model to categorize HRM prac-
tices into subdimensions. The skills bundle aims to enhance workforce Hypothesis 4: Interactions among skills, opportunity,
skills (Subramony, 2009). Firms can enhance the overall skill level of the and motivation bundles will be associated with higher
workforce by selecting the best potential qualified employees and by pro- firm performance than each bundle taken in isolation,
viding current employees with training and development activities reflecting the notion of positive synergy.
(Delaney & Huselid, 1996; Delery, 1998). Indeed, research has shown that
selective recruitment and training are positively related to employee skills Hypothesis 5: Interactions among skills, opportunity,
(Jiang, Lepak, Hu, et al., 2012; Takeuchi, Lepak, Wang, & Takeuchi, 2007). and motivation bundles will be associated with higher
As a result, we combined selective hiring and training to create the skills well-being than each bundle taken in isolation,
bundle. The opportunity bundle is to provide employees the opportunity reflecting the notion of positive synergy.
to have a direct voice over the production process (Huselid, 1995; Sub-
ramony, 2009). We used five empowerment and involvement HRM prac- On the other hand, interactions among the skills, opportunity, and
tices (job autonomy, teamwork, supportive management, information motivation bundles may also yield a combined effect that is significantly
sharing, and participatory decision-making) to create the opportunity bun- smaller than the effect of a single bundle or even a negative effect. The
dle. These five HRM practices together provide mechanisms for rationale is that, although investment in HRM systems yields economic
employees and managers to communicate on work-related issues and benefits, it also entails costs to firms (Bryson et al., 2005; Godard,
were selected based on the review of prior work (e.g., Huselid, 1995; 2004; Guthrie, 2001; Marchington & Grugulis, 2000). This means that
Jiang et al., 2012; Subramony, 2009). Finally, the motivation bundle aims there is a point beyond which adding more HRM practices will yield
to motivate employees to perform their jobs (Subramony, 2009). Firms lower incremental per-unit returns or even negative returns if the cost
can adopt various HRM practices to influence employee motivation associated with adding more HRM practices exceeds the benefits.
(e.g., performance-related pay and flexible work). However, performance- Economists have labeled this situation “the law of diminishing returns.”
related pay and flexible work enhance employee motivation differently. Indeed, a dozens of studies indicated that HRM systems are associated
The former motivates employees extrinsically, linking their work efforts to with higher labor costs (Cappelli & Neumark, 2001; Way, 2002). In
external rewards (Jiang, Lepak, Hu, et al., 2012). The later motivates addition, to the extent that the skills, opportunity, and motivation bun-
employees intrinsically, allowing employees to have control over their dles are substitutes or work against one another, combining them
work and ultimately results in a more motivated, satisfied, and productive together will generate a net return that is significantly smaller than the
workforce (Appiah-Mfodwa, Horwitz, Kieswetter, King, & Solai, 2000; Bal- return of individual bundles in isolation (Delery, 1998). Becker et al.
tes, Briggs, Huff, Wright, & Neuman, 1999; Hobson, Delunas, & Kesic, (1997) called this situation a deadly combination and Delery (1998)
2001). Consequently, we combined performance-related pay and flexible substitutes or negative synergy. Thus, we propose the following:
work to create a motivation bundle (for the link between HRM practices
and the three HRM bundles, see Table A1). Hypothesis 6: Interactions among skills, opportunity,
According to the AMO model, the skills, opportunity, and motivation and motivation bundles will be associated with lower
bundles are mutually reinforced (Appelbaum et al., 2000; Blumberg & firm performance than each bundle taken in isolation,
Pringle, 1982; Boxall & Purcell, 2003). In other words, in isolation, the reflecting the notions of substitutes and negative
skills, opportunity, and motivation bundles yield a limited effect but their synergy.
integration creates a net effect that is significantly greater than that of a
single bundle. For example, the effectiveness of skilled workers is limited Hypothesis 7: Interactions among skills, opportunity,
if they are not motivated to use their skills and knowledge toward orga- and motivation bundles will be associated with lower
nization success. Similarly, the effectiveness of motivated workers is lim- well-being than each bundle taken in isolation,
ited if they do not have skills and knowledge to perform their jobs. reflecting the notions of substitutes and negative
Finally, the effectiveness of skilled and motivated workers is limited if synergy.
6 HO AND KUVAAS

4 | METHODS 4.3 | Employee well-being


We measured employee well-being in three dimensions: happiness,
4.1 | Data
mental health, and social relationships. Two variables, organizational
The data used to test the relation among HRM systems, employee commitment and job satisfaction, were used as proxies for happiness.
well-being, and firm performance were provided by the Workplace Two variables, work intensification and anxiety, were used as proxies
Employment Relations Survey 2011 (WERS, 2011). WERS 2011 is the for mental health. One variable, management relations, was used as a
sixth survey in a series that aims to provide a nationally representative proxy for social relationships (for justifications, review the well-being
account of the state of employment relations and working life in Brit- definition section). Next, we conducted three sets of confirmatory
ish workplaces. The fieldwork for WERS 2011 occurred from March factor analyses to confirm the factor structures of well-being. The ini-
2011 to June 2012. tial measurement model had three latent factors (i.e., happiness, men-
WERS comprises two datasets. The first dataset is a management tal health, and social relationships) (Grant et al., 2007). We estimated
survey collected from a total of 2,680 face-to-face interviews with a model with three latent variables, a one-factor model and a five-
primarily senior managers who are responsible for employment rela- factor model. To determine the goodness of fit of the measurement
tions, human resources, or personnel at the sampled workplace, yield- model, we relied on the root mean square error of approximation
ing a response rate of 46.3%. The second dataset is employee surveys (RMSEA, Hu & Bentler, 1999), the comparative fit index (CFI, Bentler,
in both small and large organizations. A total of 21,981 employees 1990), and the nonnormed fit index (TLI, Hu & Bentler, 1999).
completed the survey, yielding a response rate of 54.3%. According to Hu and Bentler (1999), RMSEA ≤ .06 or CFI ≥ .95 or TLI
WERS 2011 covers both private and public firms, but our sample ≥ .95 indicates a good fit.
includes only private firms. After filtering to include only workplaces The three-factor model fit the data poorly (χ2 [227] = 26,882.635;
and individual employees for which there was sufficient information RMSEA = .09; CFI = 0.865; TLI = 0.850; SRMR = 0.091; AIC = 1,132,000).
on all study variables and control variables, our sample included The one-factor model was clearly insufficient (χ2 [230] = 116,094.383;
14,384 employees nested within 1,347 firms. The median number of RMSEA = .187; CFI = 0.414; TLI = 0.355; SRMR = 0.209; AIC = 1,221,000).
respondents was 15 with a range from 3 to 25. The five-factor model fit the data well (χ2 [220] = 9,729.184; RMSEA = .05;
CFI = 0.952; TLI = 0.945; SRMR = 0.043; AIC = 1,115,000); therefore, we
decided to examine the five indicators of well-being separately. Subse-

4.2 | Measures quently, we provide a detailed description of the five variables; we also
conducted a pilot study to verify the construct validity (convergent and
HRM systems were measured with nine HRM practices: two practices discriminant validity) of the measures of organizational commitment and
from the management survey (selective hiring and teamwork) and work intensification. The data in the pilot study provided evidence for the
seven practices from the employee survey (job autonomy, staff train- construct validity of both commitment and work intensification by dem-
ing, flexible work, participatory decision-making, information sharing, onstrating both convergent and discriminant validity (results available
supportive management, and performance-related pay). All HRM prac- upon request).
tices were measured using multiple-item scales, except staff training. Organizational commitment (alpha = .77) was measured by means
Furthermore, four HRM variables (teamwork, performance-related of three items adapted from the well-established scale by Meyer and
pay, flexible work, and selective hiring) were measured using binary Allen (1997): “employees share the organization's values,” “employees
items. The use of binary items is a common practice in the field feel loyal to the organization,” and “employees are proud to tell people
(e.g., Bryson & White, 2008; Guest & Conway, 2007). In line with about the organization.” The measure uses a five-point rating scale
Ogbonnaya et al.'s (2017) recommendation, we aggregated the seven (1 = strongly disagree; 5 = strongly agree).
HRM variables from the employee survey (i.e., job autonomy, staff Job satisfaction (alpha = .87) was measured by seven items,
training, flexible working, participatory decision-making, information reflecting seven facets of work, which have been used in a number of
sharing, supportive management, and performance-related pay) as the WERS-based studies (e.g., Bryson, Cappellari, & Lucifora, 2010):
mean scores that proxy organizational-level HRM practices. The “employee satisfaction with influence,” “employee satisfaction with
remaining two variables were derived from the management survey, achievement from work,” “employee satisfaction with using initiative,”
so they serve as proxies for organizational-level measures. Prior to “employee satisfaction with job security,” “employee satisfaction with
aggregation, we computed the intraclass correlation (ICC)—both ICC1 the training received,” “employee satisfaction with involvement in
and ICC2—to examine interrater reliability. ICC1 values ranged from decision-making,” and “employee satisfaction with the work itself”
.10 to .36 and ICC2 values from .56 to .86, which are similar to those (1 = very dissatisfied; 5 = very satisfied).
reported in Ogbonnaya et al. (2017), which included observed values Work intensification (alpha = .70) was assessed by means of two
ranging from .07 to .20 for ICC1 and from .53 to 0.78 for ICC2. In items, measuring the perception of work pressure based on Karasek's
other words, these results offered sufficient justification for data (1979) definition of job demands: “the extent to which employees felt
aggregation (see Table A2 for a full description). that their jobs required them to work very hard” and “whether they
HO AND KUVAAS 7

felt they had enough time to get their work done” (1 = strongly dis- effects of them all together in a unitary index. Furthermore, consistent
agree; 5 = strongly agree). with Way's (2002) suggestion, the estimates of the effects of each
Anxiety (alpha = .94) was assessed by means of three items mea- HRM practice on firm performance were controlled for the impact of
suring negative emotional states based on Warr's (1990) measure: other HRM practices in the set. We used the following indicators to
“thinking of the past few weeks, how much of the time has your job evaluate the existence of synergistic effects: variable's statistical sig-
made you feel tense,” “thinking of the past few weeks, how much of nificance, magnitudes of the coefficients, and change in R-square.
the time has your job made you feel uneasy,” and “thinking of the past As shown in Table 2, only job autonomy and performance-related
few weeks, how much of the time has your job made you feel wor- pay had a significant and positive correlation with firm performance,
ried” (1 = never; 5 = all of the time). with each accounting for 10% of the variance in firm performance
Management relations (alpha = .92) were assessed by means of (b = .10, p < .05; b = .10, p < .01, respectively). The other seven prac-
four items, measuring employees' perception of their relations with tices were nonsignificant. However, the HRM systems, composing of
managers in terms of trust, understanding, honesty, and fairness all nine practices, had a significant and positive correlation with firm
(1 = strongly disagree; 5 = strongly agree). performance and accounted for 17% of the variance in firm perfor-
mance (b = .17, p < .001). In other words, HRM systems have signifi-
cantly larger effects on firm performance than the effects from
4.4 | Performance adopting each practice individually. Hypothesis 1 was thus supported.
We measured organizational performance using three items based on Hypothesis 2a predicted a positive relationship between HRM
the information provided by managers who had primary responsibility systems and employee well-being, whereas Hypothesis 2b predicted a
for employment relations. These managers were asked to rate the per- negative relationship. To evaluate Hypotheses 2a and 2b, we first ran
formance of their own workplaces compared with that of other work- a base model with control variables and added HRM systems in a sec-
places in the industry in terms of financial performance, labor ond model. As shown in Table 3, HRM systems have significant asso-
productivity, and quality of product service. Ratings were based on a ciations with all well-being measures except work intensification.
five-point scale (far above average to far below average). Available evi- Moreover, the signs were positive with organizational commitment,
dence indicates that managerial assessments correspond closely to job satisfaction, and management relations but negative with anxiety.

internal objective performance indicators (Dess & Robinson, 1984; In other words, the results indicate that HRM systems have a positive

Wall et al., 2004) and external secondary data (Venkatraman & effect on well-being. Hypothesis 2a was therefore supported, but

Ramanujam, 1987). To reduce the complexity of the analysis, we com- hypothesis 2b was not supported.

bined the three variables into a single global construct (alpha = .79). In Hypothesis 3, we posited that the relationship between HRM
systems and (a) firm performance and (b) well-being is curvilinear, with
positive effects at low to moderate levels of implementation but nega-
4.5 | Control variables tive effects at higher levels. To test for nonlinear effects, we followed

In the analysis, we included control variables at the workplace and Chadwick's (2007) method. We first created an HRM index to present

employee levels. At the employee level, the five control variables were HRM systems and then converted this index into its quadratic and
cubic terms. If nonlinear effects are present, the quadric or cubic
sex (1 = female; 0 = male), age (eight dummies), type of job (1 = perma-
terms should be significant.
nent; 2 = temporary or fixed), hours per week, and union membership
With respect to firm performance, Table 4 shows that the linear
(1 = yes; 2 = no). At the workplace level, the three control variables
effect of HRM systems was positive and significant association
were industry (12 dummies), age of firm (the number of years of oper-
(b = .13, p < .05), but the quadratic and cubic terms were nonsignifi-
ation), and firm size (1 = fewer than 50 employees; 12 = 100,000
cant (b = .01, p > .05). Hypothesis 3a was thus not supported.
employees or more).
With respect to well-being, Table 4 shows that there is evidence
of nonlinear relations for job satisfaction, management relations, and
5 | RESULTS anxiety (as indicated by significance of the quadratic or cubic terms)
but not for organizational commitment and work intensification
Table 1 presents the means, ESs, and correlations. Hypothesis 1 con- (as indicated by nonsignificance of the quadratic or cubic terms). For
cerns synergy among HRM practices when combined into a system, job satisfaction, the quadratic effect was significant and negative
that is, whether HRM systems have larger effects on firm performance (b = −.02, p < .05) (Table 4), indicating that satisfaction first increased
than the effects from adopting each practice individually. Based on and then declined as more intensive implementation is reached. A plot
the literature review, Chadwick (2010) argued that an ideal approach of the relationship shows that the relationship was initially positive
to testing the existence of synergy is “to explicitly compare the perfor- but declined as the number of HRM practices increased (Figure 1). For
mance effects of a set of HRM practices with a contextually and theo- management relations, the quadratic effect sign was significant and
retically appropriate aggregation of those same practices” (p. 89). In negative (b = −.04, p < .001) (Table 4), indicating that management
this study, we followed Chadwick's (2010) recommendation to com- relations first increased and then declined as more
pare the effects of nine individual practices with the performance intensive implementation was reached (i.e., decreasing returns to
8

TABLE 1 Means, SEs, and correlations of study variables

Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12 13
1. Subj. performance 3.75 .59 1
2. HRM system 14.9 1 0.15*** 1
3. Org. commitment 3.82 .81 .08*** .31*** 1
4. Job satisfaction 3.65 .74 .09*** .29*** .62*** 1
5. Anxiety 2 .86 −.04*** −.15*** −.37*** −.51*** 1
6. Work intensification 3.65 .77 −.00 .01 .03** −.05*** .34*** 1
7. Management relations 3.04 .99 .04*** .11*** .27*** .26*** −.18*** −.04*** 1
8. Firm size 9.16 3.28 −.05*** −.18*** −.11*** −.10*** .06*** −.01 −.03** 1
9. Firm age 1.31 .45 −.02* −.05*** .01 −.03** .01 .00 −.03** .08*** 1
10. Female .49 .49 .03** .13*** .11*** .09*** −.06*** .04*** .08*** −.12*** −.05*** 1
11. Age 5.66 1.47 −.03** −.06*** .03** .02** −.01 .07*** −.01 .04*** .11*** −.07*** 1
12. Hours/week 36.03 12.81 .00 −.01 −.01 −.04*** .16*** .18*** −.06*** −.01 .03** −.30*** .06*** 1
13. Permanent job .92 .25 −.03** −.02 −.01 .00 .06*** .09*** −.03** .01 −.00 −.01 .13*** .12*** 1
14. Member of union .24 .43 −.08*** −.16*** −.10*** −.12*** .09*** .05*** −.09*** .30*** .11*** −.15*** .18*** .08*** .07***

Note: Firm size coded 1 = less than 50 (reference), 2 = 50–99, 3 = 100–149, 4 = 150–249, 5 = 250–499, 6 = 500–999, 7 = 1,000–1,999, 8 = 2,000–4,999, 9 = 5,000–9,999, 10 = 10,000–49,000,
11 = 50,000–99,999, and 12 = 100,000 or more. Gender coded 1 = female, 0 = male (reference). Age coded 1 = 16–17 (reference), 2 = 18–19, 3 = 20–21, 4 = 22–29, 5 = 30–39, 6 = 40–49, 7 = 50–59,
8 = 60–64, and 9 = 65 or more. Member of trade union/permanent job coded 1 = yes, 0 = no (reference). For ease of presentation, we omitted industry and sector. Please write to the first author for a complete
correlation matrix.
Abbreviation: HRM, human resource management.
*p < .05; **p < .01; ***p < .001.
HO AND KUVAAS
HO AND KUVAAS 9

T A B L E 2 The impact of individual HRM practices and HRM reflecting the notion of positive synergy. In contrast, Hypothesis
systems on firm performance 5 predicted that interactions among skills, opportunity, and motivation
Overall subjective bundles would be associated with (a) lower firm performance and
Variable firm performance ΔR2 (b) lower well-being than each bundle taken in isolation, reflecting the
Job autonomy .10(04)* 0.006* notions of substitutes and negative synergy.
Teamwork .01(04) 0.000 To test the interactions, we followed Macduffie's (1995) recom-

Training −.02(05) 0.000 mendation. First, a base model was run, which contained all the con-
trol variables and the tree bundles. A second model was then run that
Performance-related pay .10(04)** 0.009**
added the two-way interactions among the three bundles (Skills ×
Flexible work −.03(04) 0.000
Opportunity; Skills × Motivation; Opportunity × Motivation). Finally, a
Selective hiring −.02(04) 0.000
third model added the three-way interaction (Skills × Opportunity ×
Participatory decision-making .09(05) 0.004
Motivation). Because multilevel modeling is used given the nested
Information sharing .10(05) 0.005
structure of the data (employees within firms), instead of the change
Supportive management .03(05) 0.000
in R-square, the likelihood ratio test will be used to determine
HRM systems .17(04)*** 0.022*** whether adding interaction terms imply a significant gain in the
Note: Cell entries are the estimated coefficients with the SE of estimate in model's explained variance.
brackets. All analyses include all controls as described in the text and the The influence of interactions among the three bundles on firm
new controls for each HRM practice (the index of the other eight practices
performance is presented in Table 5B. The results show that, in isola-
of the HRM systems).
tion, the opportunity bundle had a significant and positive association
Abbreviation: HRM, human resource management.
*p < .05; **p < .01; ***p < .001. with firm performance (b = .05, p < .001) (Model 1), but it had a signif-
icant and negative association when interacting with skills and motiva-
tion bundles (−.04, p < .05) (Model 3). In other words, the three-way
scale). A plot of the relationship showed that the relationship initially
interactions among the bundles do not yield a net effect that exceeds
increased but declined as the number of HRM practices increased
the effect of single bundle. Hypothesis 5a was thus supported, but
(Figure 2). Finally, for anxiety, the cubic effect was significant and neg-
Hypothesis 4a was not supported.
ative (b = −.01, p < .05), but the magnitude of coefficient was reduced
Supplemental analyses (Table 5B and Table 6) reveal that, in isola-
from −.14 to −.01 (Table 4), indicating that anxiety increased as HRM
tion, the opportunity bundle had a statistically significant positive
implementation moves from lower to higher levels. A plot of the rela- association with labor productivity (.08, p < .05) (Model 1), but it
tionship shows that for systems with very few practices, adding an became nonsignificant when interacting with skills and motivation
additional practice has a great impact in reducing anxiety. This impact, bundles (−.03, p > .05) (Model 3). In other words, the insignificant
however, diminishes for firms with average number and high number result is attributed to the rise labor costs associated with the integra-
of practices where the curve flattens (Figure 3). On the whole, three tion of the three bundles, as reflected by the nonsignificance between
out of five indicators of well-being show nonlinear paths with HRM the two- and three-way interactions between the bundles and labor
systems. Hypothesis 3b was therefore supported. productivity. These results indicate that for firm performance, the
Hypotheses 4 and 5 concern the interactive effects. Hypothesis interrelationships of skills, opportunity, and motivation bundle are
4 predicted that interactions among the skills, opportunity, and moti- substitutes for negative synergy but not for positive synergy.
vation bundles would be associated with (a) higher firm performance The influence of the interactions among the three bundles on
and (b) higher well-being than each bundle taken in isolation, well-being is presented in Table 6. The results were generally

TABLE 3 The impact of HRM systems on well-being

Well-being dimensions

Happiness Mental health Social relationships

Variable Organizational commitment Job satisfaction Anxiety Work intensification Management relations
Model 1
Controls Yes Yes Yes Yes Yes
Model 2
HRM systems .32(01)*** .30(01)*** −.18(02)*** −.01(02) .10(01)***
LRT 362.11*** 377.20*** 122.70*** 0.46 56.35***

Note: Cell entries are the estimated coefficients with the SE of estimate in brackets. All analyses include all controls as described in the text.
Abbreviations: HRM, human resource management; LRT, likelihood ratio test.
*p < .05; **p < .01; ***p < .001.
10 HO AND KUVAAS

TABLE 4 Nonlinear effect results

Well-being dimensions

Happiness Mental health Social relationships

Overall subjective Organizational


firm performance commitment Job satisfaction Anxiety Work intensification Management relations
Model 1
Controls Yes Yes Yes Yes Yes Yes
Model 2
Controls Yes Yes Yes Yes Yes Yes
HRM system (linear) .17(.04)*** .32(02)*** .30(.01)*** −.18(.02)** −.01(.02) .10(.01)***
Model 3
Controls Yes Yes Yes Yes Yes Yes
HRM system (linear) 13(05)* 33(02)*** 30(02)*** −.13(02)*** 02(02) 11(02)***
HRM system (quadratic) .01(.03) −.00(.01) −.02(01)* .02(.01) −.00(.01) −.04(.01)***
HRM system (cubic) .01(.01) −.00(.01) .00(.00) −.01(.01)* −.01(.01) −.01(.00)

Note: Cell entries are the estimated coefficients with the SE of estimate in brackets. All analyses include all controls as described in the text.
Abbreviation: HRM, human resource management.
*p < .05; **p < .01; ***p < .001.

nonsignificant except for the interaction between motivation and linearity demonstrating that the relationship is nonlinear. Specifically,
skills bundles for organizational commitment. More precisely, in isola- we found that, at moderate levels of implementation, HRM practices
tion, the motivation bundle had a significant and positive correlation are positively correlated with employee well-being as well as higher
with organizational commitment (b = .08, p < .05) (Model 1). When levels of commitment, job satisfaction, management relations, and
interacting with skills bundle, the relationship became negative (−.10, lower levels of anxiety. However, at high levels, the relationship is less
p < .05), indicating the combined effect of interaction between skills positive and even turns negative with lower levels of job satisfaction
and motive bundles does not exceed the effect of single bundle. How- and management relations. This curvilinear finding bridges the divide
ever, on the whole, we found little evidence of the influence of inter- between the mutual gains and the critical scholars regarding the
actions between the three bundles on well-being. Hypotheses 4b and impact of HRM systems on well-being, suggesting that the conflicting
5b were therefore not supported. findings of the relationship between HRM systems and well-being
may be attributable to the fact that prior research is constrained by its

6 | DISCUSSION linearity obsession and hence ignores a possible nonlinear association


between HRM systems and well-being. In other words, from a theo-
This study attempts to shed light on the well-being paradox; HRM retical viewpoint, our study extends the research on well-being specif-
practices have at times positive effects on well-being and at other ically and the HRM literature generally by arguing that the nonlinear
times negative effects. Our study has both research and practical
implications.

6.1 | Research implications


First, our study contributes to the well-being debate. As argued,
researchers have recently engaged in active debate about the nature
of the relationship between HRM systems and well-being. Some
scholars have argued that HRM systems are good for employees and
employers (the mutual gains perspective) (Kochan & Osterman, 1994).
Other scholars have argued that HRM systems are bad for employees
but good for employers (the critical perspective) (Legge, 2005). The
accumulating body of evidence supports both views, which we have
called the well-being paradox in this study. Furthermore, with some
exceptions (White & Bryson, 2013), researchers have been gravitating
toward the assumption that the relationship between HRM practices FIGURE 1 Curvilinear relationship between HR system and job
and well-being is linear. We falsified this conventional assumption of satisfaction
HO AND KUVAAS 11

TABLE 5A Interaction effects between bundles and firm


performance

Overall subjective firm


performance

Bundles β Δ2
Model 1
Controls Yes
Skills −.02(04)
Opportunity .05(01)***
Motivation .12(08)
Model 2
Controls Yes 0.001
Skills −.03(04)
F I G U R E 2 Curvilinear relationship between HR system and Opportunity .05(01)***
management relations
Motivation .12(09)
Skills × Opportunity −.00(01)
model is better to describe the nature of the relationship between Skills × Motivation −.03(09)
HRM systems and well-being than the linear model, opening a new Opportunity × Motivation .02(02)
door for potential groundbreaking research and theory development. Model 3
Second, our study contributes to research on HRM-performance
Controls Yes 0.006*
paradigm. Most research within the HRM-performance paradigm
Skills −.02(04)
explicitly or implicitly assumes that there are additive or positive syn-
Opportunity .05(01)***
ergistic effects among the HRM practices constituting HRM systems
Motivation .18(09)
or “more is better” in the language of De Winne and Sels (2013). Our
Skills × Opportunity −.00(01)
results challenged the conventional wisdom of more is better. We
Skills × Motivation −.07(09)
found that when an opportunity bundle was implemented in isolation,
Opportunity × Motivation .02(02)
it had a positive effect on firm performance. However, when an
Skills × Opportunity × Motivation −.04(02)*
opportunity bundle interacted with skills and motivation bundles, it
LRT
had a negative effect. Supplemental analyses revealed that the nega-
tive result was due to the rise labor costs associated with the integrat- Note: Cell entries are the estimated coefficients with the SE of estimate in
brackets. All analyses include all controls as described in the text.
ing the three bundles (i.e., the three-way interaction among the
Abbreviation: LRT, likelihood ratio test.
bundles was not associated with labor productivity). These findings *p < .05; **p < .01; ***p < .001.
indicate that more is not necessarily better, as implementing a large
scale of HRM practices (i.e., integrating all the three bundles together) leads to negative outcomes due to the increase in labor costs. In other
words, contrary to the common assumption that the interrelationship
among skills, opportunity, and motivation bundles exhibits positive
synergistic effects (Macduffie, 1995), our study showed that the rela-
tionship is substitutable and negative synergistic effects, suggesting
that our “more is better” assumption may be based on faith rather
than on hard evidence. It should be noted here that the finding that
investments in HRM practices increase labor costs supports the argu-
ment that HRM systems have both value-creating and cost-enhancing
effects (Godard, 2004; Guthrie, 2001; Marchington & Grugulis, 2000);
an argument that is, according to our best knowledge, rarely tested.
Third, our study contributes to research on fit or synergy.
Although fit or synergy has been a central concept in the HRM litera-
ture for decades, our knowledge of it remains limited (Chadwick,
2010). Recent major reviews have revealed that not only is there few
research on fit (Jiang, Lepak, Han, et al., 2012) but also that we are,
FIGURE 3 Curvilinear relationship between HR system and given the current accumulated body of knowledge, neither able to
anxiety demonstrate the existence of synergy nor refute it (Chadwick, 2010;
12 HO AND KUVAAS

TABLE 5B Supplemental analyses: Interaction effects between bundles and individual firm performance outcomes

Financial performance Labor productivity Quality of product

Bundles β ΔR2 Β ΔR2 β ΔR2


Mode 1
Controls Yes Yes Yes
Skills −.05(04) −.01(04) .02(04)
Opportunity .12(04)** .08(04)* .13(04)**
Motivation .08(04) .07(04) −.02(04)
Model 2
Controls Yes 0.005 Yes 0.003 Yes 0.002
Skills −06(04) −.03(04) .03(04)
Opportunity .12(04)** .08(04) .13(04)**
Motivation .06(04) .08(04) −.02(04)
Skills × Opportunity −.03(03) −.02(03) .03(03)
Skills × Motivation .00(04) −.04(04) .00(04)
Opportunity × Motivation .06(03) .02(03) −.02(03)
Model 3
Controls Yes 0.004 Yes 0.002 Yes 0.004
Skills −.06(04) −.03(04) .04(04)
Opportunity .13(04)** .08(04) .14(04)**
Motivation .09(04)* .09(04)* −.00(04)
Skills × Opportunity −.04(03) −.03(03) .01(03)
Skills × Motivation −.01(04) −.05(04) −.01(04)
Opportunity × Motivation .05(03) .02(03) −.02(03)
Skills × Opportunity × Motivation −.04(02) −.03(03) −.04(02)

Note: Cell entries are the estimated coefficients with the SE of estimate in brackets. All analyses include all controls as described in the text.
*p < .05; **p < .01; ***p < .001.

Gerhart, 2007; Wall & Wood, 2005). This inability is because most of system of practices has more influence on firm performance than the
research on fit or synergy is “driven by convenience rather than by individual practices, specifying different relationships within an HRM
theory,” with little specification for the forms of synergy within an system (e.g., additive, substitutable, positive, and negative synergistic
HRM system or explanations for how synergies within an HR system relationships) and explaining how these influence organizational out-
operate (Boon, Den Hartog, & Lepak, 2019; Chadwick, 2010, p. 86; comes. For example, drawing on the AMO model, we argued the skills,
Delery, 1998; Kepes & Delery, 2007). Our study overcomes these lim- opportunity, and motivation bundles are complementary. As a result,
itations. In our study, we went beyond the general statement that a we predicted that if these bundles were integrated together, they

TABLE 6 Interaction effects between bundles and well-being

Well-being dimensions

Happiness

Bundles Organizational commitment Job satisfaction


1 2 3 1 2 3
Skills .02(01) .03(02) .03(02)* .01(01) .01(01) .01(01)
Opportunity .10(00)*** .10(00)*** .10(00)*** .11(00)*** .11(00)*** .11(00)***
Motivation .08(03)* .11(03)** .12(03)*** −.07(03)* −.06(03) −.06(03)
Skills × Opportunity .01(00) .01(00) −.00(00) −.00(00)
Skills × Motivation −.10(03)* −.11(03)* −.03(03) −.03(03)
Opportunity × Motivation .01(01) .01(01) .01(01) .01(01)
Skills × Opportunity × Motivation −.01(01) .00(00)
LRT 10.92* 1.6 2.21 0.00
HO AND KUVAAS 13

TABLE 6 (Continued)

Well-being dimensions

Mental health Social relationships

Bundles Anxiety Work intensification Management relations


1 2 3 1 2 3 1 2 3
Skills .00(02) .01(02) .01(02) −.00 −.00 .00(01) −.01(01) −.02(01) −.02(01)
(02) (02)
Opportunity −.06 −.06 −.06 −.01 −.01 −.01(01) .05(00)*** .05 .05
(00)*** (00)*** (00)*** (01) (01) (00)*** (00)***
Motivation .02(03) .00(03) .00(04) .03(03) .02(03) .04(04) −.11 −.09 −.08
(02)*** (03)** (03)**
Skills × opportunity .00(00) .00(00) −.00 −.00(00) −.01(00)* −.01(00)*
(00)
Skills × motivation .01(04) .01(04) .02(04) .01(04) .00(03) .00(03)
Opportunity × motivation .00(01) .00(01) .01(01) .01(01) −.01(01) −.01(01)
Skills × opportunity × .00(01) −.03 −.01(01)
motivation (01)**
LRT 2.56 0.02 1.26 9.01** 11.12** 3.71

Note: Cell entries are the estimated coefficients with the SE of estimate in brackets. All analyses include all controls as described in the text. Model
1 contains all the control variables and the tree bundles. Model 2 adds the two-way interactions between the three. Model 3 adds the three-way
interaction.
Abbreviation: LRT, likelihood ratio test.
*p < .05; **p < .01; ***p < .001.

would yield a combined positive synergistic effect on firm perfor- concept of well-being. Grant et al. (2007) have long observed that
mance that exceeds the effect of single bundle. From a theoretical HRM practices “frequently create tradeoffs between different dimen-
viewpoint, our theory-driven definitions of different forms of synergi- sions of employee well-being, whereby one aspect of employee well-
es/relationships and our specific discussions of how their interactions being improves but another aspect of employee well-being decreases”
effect on firm performance not only have a conceptual contribution (p. 51); our results confirm their observation.
to research on fit/synergy but also advance our still incomplete under-
standing of the black box connecting HRM to firm performance.
6.2 | Practical implications
Fourth, in the well-being literature, it is generally accepted that
“psychologically well people are more prone to experience positive Perhaps the most important practical implication of our study is that
emotions and less prone to experience negative emotions” (Wright & failing to recognize the existence of a nonlinear relationship may lead
Cropanzano, 2000, p. 84). That is, well-being scholars tend to treat managers to make wrong decisions about the practical utility of HRM
well-being as a bipolar construct, assuming that as positive well-being practices. To illustrate our point, let us consider a hypothesized case
increases negative well-being decreases (Diener, 1984; Wright, 2014). that managers want to increase the well-being of their workforce,
Our result questions this traditional wisdom, showing that at higher because they believe that happy employees are more productive than
levels of implementation, HRM systems have favorable effects on unhappy employees. If mangers followed the traditional wisdom of
employee well-being by reducing anxiety, but they also have debilitat- more is better (i.e., linearity), they would increase their investments in
ing effects on employee well-being by lowering job satisfaction and HRM by adopting a large scale of HRM practices with the hope of
management relations. This observation suggests that some dimen- maximizing employee well-being. Because the HRM/well-being rela-
sions of well-being are independent, unipolar constructs, which can tionship is curvilinear, as documented by our study, their investments
potentially arise simultaneously. In other words, some dimensions of in HRM would initially result in increased well-being. Continuing
well-being are too distinct from one another to be merged into one investments in HRM would, however, ultimately lead to decreased
composite score; if bundled together, we obscure the different foci of well-being. If managers had known about our study and, therefore,
the individual dimensions of well-being, leading to inaccurate interpre- had understood that investments in a large scale of HRM practices
tations of theoretical and practical consequences. We believe that this could lead to decreased instead of increased well-being, they would
result might also explain the previously noted inconsistencies in stud- have limited the scale of their investments and focused on a small,
ies examining the relationship between HRM systems and well-being; economic scale of practices, avoiding waste energy and resources. In
that is, the inconsistencies may be attributable to the fact that prior other words, knowing the existence of a nonlinear relationship can
research has not paid sufficient attention to the complexities of the prevent missteps in the application of HRM practices.
14 HO AND KUVAAS

Our study also highlights the importance of how to bundle HRM cautioned when interpreting the subjective firm performance results
practices. Although our study suggested that while combining HRM reported in our study. That said past studies have shown that subjec-
practices into a system or bundle has a positive performance impact, tive measures are positively associated with objective measures
not all the combinations yield positive returns. For example, we found (Dess & Robinson, 1984; Wall et al., 2004) and that the problem of
that when adopted in isolation, the opportunity bundle enhances firm single source is not as serious as normally assumed (Crampton & Wag-
performance, but when integrating it with motivation or skill bundles, ner, 1994; Spector, 2006). Nevertheless, we acknowledge that using
firm performance declined due to the increase in labor costs. For prac- objective financial data would boost the robustness of our results.
tical managers, this observation means that they should pay attention Fourth, we rely on existing survey data and are therefore unable to use
to or avoid this kind of “deadly combination,” to use the language of more established, validated measures of our constructs, most notably the
Becker et al. (1997), when designing their HRM systems. measures of organizational commitment and work intensification. Although
the data in the pilot study provided evidence for the construct validity (both
in terms of convergent and discriminant validity), this use of measurement
6.3 | Limitations and future research
clearly limits the comparability of our findings to previous research.
We acknowledge that our research has several limitations. First,
although our conceptual model implies causality, with HRM systems
preceding well-being and firm performance, the cross-sectional nature
7 | C O N CL U S I O N S
of our data precludes such direct causal conclusions. For example, the
Researchers have long been divided regarding the impact of HRM sys-
reverse order of the relation may be possible (Cappelli & Neumark,
tems on employee well-being and such systems' implications for firm
2001) such that high-performing firms may have the financial resources
performance. At one end of the spectrum, researchers have argued
to invest in HRM systems, resulting in higher firm performance.
that both employers and employees benefit from HRM systems
Although our results on the effects of HRM systems on firm perfor-
(i.e., the mutual gains perspective). On the other end of the spectrum,
mance are consistent with prior work that explicitly corrected for simul-
researchers have argued that employers, but not employees, benefit
taneity bias (Huselid, 1995), a recent longitudinal study that tested the
causal associations between HRM systems and performance using a from HRM systems (i.e., the critical perspective). Both approaches

large longitudinal dataset with three time points revealed that the asso- have been successful with respect to accumulating evidence, yet they

ciation is probably reciprocal; past HRM systems have positively con- seem fundamentally incompatible. We considered this paradox in the

tributed to later productivity, and the reverse has also been found current study, and our findings suggest that at moderate levels of

(Shin & Konrad, 2017). In addition, as nonlinear effects emerge over adoption, HRM systems enhance well-being but reduce well-being at

time, longitudinal designs are more suitable for testing such effects higher levels. In other words, our results suggest that the con-

(Pierce & Aguinis, 2013). We therefore encourage future research to tradicting findings in the literature may be attributable to the extant

retest our nonlinear hypotheses with longitudinal designs. literature's failure to account for the possible nonlinear associations
Second, we did not examine whether moderating conditions affect between HRM systems and well-being and to pay sufficient attention
the studied relationships. Previous research has suggested that the to the tradeoffs between different components of well-being. We
relationship between HRM systems and well-being may be moderated hope that our results stimulate further debate and research on this
by a range of individual, organizational, and institutional factors. For important topic.
example, broader institutional and legislative contexts set limitations
on the agency of managers with regard to how they address human
ACKNOWLEDG MENTS
resource policies and practices as well as employment relations
(Paauwe, 2004). These limitations, in turn, may affect well-being. For insightful feedback, we thank Dr. Motohiro Morishima and the
Another example can be found in Jensen et al. (2011), who deter- three synonymous reviewers.
mined that HRM systems lead to work intensification. However, this
relationship was contingent on employees' perceptions of job control;
OR CID
that is, employees with low levels of perceived job control reported
higher work intensification than those with higher levels of perceived Hoang Ho https://orcid.org/0000-0002-7242-6022
job control. Therefore, we encourage future research to systematically Bård Kuvaas https://orcid.org/0000-0002-9516-5259
examine these possible moderator effects to develop a more complete
understanding of the relationships among HRM systems, well-being,
RE FE RE NCE S
and firm performance.
Third, we rely on subjective measures of firm performance. Andrews, F. M., & Withey, S. B. (2012). Social indicators of well-being: Amer-
Although there is some merit in relying on perceived measures of per- icans' perceptions of life quality. New York, NY: Springer Science &
Business Media.
formance due to the difficulty of gathering “hard” financial data (Way,
Appelbaum, E., Bailey, T., Berg, P., & Kalleberg, A. (2000). Manufacturing
2002), using subjective measures may make our results biased advantage: Why high-performance work systems pay off. Ithaca, NY:
(Becker & Gerhart, 1996). With this in mind, the reader should be Cornell University Press.
HO AND KUVAAS 15

Appiah-Mfodwa, A., Horwitz, F., Kieswetter, G., King, D., & Solai, L. (2000). Bryson, A., & White, M. (2008). Organizational commitment: Do workplace
Flexible work practices, productivity improvement and employment. practices matter? (No. CEPDP0881). Centre for Economic Performance,
Society in Transition, 31(2), 95–110. London School of Economics and Political Science.
Arthur, J. B. (1994). Effects of human resource systems on manufacturing Butts, M. M., Vandenberg, R. J., DeJoy, D. M., Schaffer, B. S., &
performance and turnover. Academy of Management Journal, 37(3), Wilson, M. G. (2009). Individual reactions to high involvement work
670–687. processes: Investigating the role of empowerment and perceived orga-
Bakker, A. B., & Oerlemans, W. G. M. (2011). Subjective well-being in orga- nizational support. Journal of Occupational Health Psychology, 14(2),
nizations. In K. Cameron & G. Spreitzer (Eds.), The Oxford handbook of 122–136. https://doi.org/10.1037/a0014114
positive organizational scholarship (pp. 178–189). New York, NY: Campbell, A. (1981). The sense of well-being in America: Recent patterns and
Oxford University Press. https://doi.org/10.13140/2.1.1145.4723 trends. New York: MacGraw-Hill.
Cappelli, P., & Neumark, D. (2001). Do “high-performance” work practices
Baltes, B. B., Briggs, T. E., Huff, J. W., Wright, J. A., & Neuman, G. A.
improve establishment-level outcomes? Industrial and Labor Relations
(1999). Flexible and compressed workweek schedules: A meta-analysis
Review, 54(4), 737–775. https://doi.org/10.1525/sp.2007.54.1.23
of their effects on work-related criteria. Journal of Applied Psychology,
Castanheira, F., & Chambel, M. J. (2010). Reducing burnout in call centers
84(4), 496–513.
through HR practices. Human Resource Management, 49(6),
Barney, J. (1995). Looking inside for competitive advantage. Academy of
1047–1065. https://doi.org/10.1002/hrm.20393
Management Executive, 9(4), 49–61. https://doi.org/10.5465/AME.
Chadwick, C. (2007). Examining non-linear relationships between human
1995.9512032192
resource practices and manufacturing performance. Industrial and
Barney, J. A. Y., & Felin, T. (2013). What are microfoundations? Academy Labor Relations Review, 60(4), 499–521. https://doi.org/10.2307/
of Management Perspectives, 27(2), 138–155. 25249107
Becker, B. E., & Huselid, M. A. (1998). High performance work systems Chadwick, C. (2010). Theoretic insights on the nature of performance syn-
and firm performance: A synthesis of research and managerial implica- ergies in human resource systems: Toward greater precision. Human
tions. In G. R. Ferris (Ed.), Research in personnel and human resource Resource Management Review, 20(2), 85–101.
management (Greenwich). Citeseer. Chi, N. W., & Lin, C. Y. Y. (2011). Beyond the high-performance paradigm:
Becker, B. E., Huselid, M. A., Pickus, P. S., & Spratt, M. F. (1997). HR as a Exploring the curvilinear relationship between high-performance work
source of shareholder value: Research and recommendations. Human systems and organizational performance in Taiwanese manufacturing
Resource Management, 36(1), 39–47. https://doi.org/10.1002/(SICI) firms. British Journal of Industrial Relations, 49(3), 486–514. https://doi.
1099-050X(199721)36:1<39::AID-HRM8>3.0.CO;2-X org/10.1111/j.1467-8543.2010.00778.x
Becker, B., & Gerhart, B. (1996). The impact of human resource manage- Collins, C. J., & Smith, K. G. (2006). Knowledge exchange and combination:
ment on organizational performance: Progress and prospects. Academy The role of human resource practices in the performance of high-
of Management Journal, 39(4), 779–801. technology firms. Academy of Management Journal, 49(3), 544–560.
Beer, M., Spector, B., Lawrence, P. R., Quinn Mills, D., & Walton, R. E. Combs, J., Liu, Y., Hall, A., & Ketchen, D. (2006). How much do high-
(1984). Human resource management. New York, NY: Free Press. performance work practices matter? A meta-analysis of their effects
Bentler, P.M. (1990). Comparative Fit Indexes in Structural Models. Psy- on organizational performance. Personnel Psychology, 59(3), 501–528.
chological Bulletin, 107(2), 238–246. Crampton, S. M., & Wagner, J. A., III. (1994). Percept-percept inflation in
Berg, P., Appelbraum, E., Bailey, T., & Kalleberg, A. L. (1996). The perfor- microorganizational research: An investigation of prevalence and
mance effects of modular production in the apparel industry. Industrial effect. Journal of Applied Psychology, 79(1), 67–76.
Relations, 35(3), 356–373. https://doi.org/10.1111/j.1468-232X. Cropanzano, R., & Wright, T. A. (2001). When a “happy” worker is really a
1996.tb00411.x “productive” worker: A review and further refinement of the happy-
Blumberg, M., & Pringle, C. D. (1982). The missing opportunity in organiza- productive worker thesis. Consulting Psychology Journal: Practice and
tional research: Some implications for a theory of work performance. Research, 53, 182–199.
Academy of Management Review, 7(4), 560–569. Daniels, K., & Harris, C. (2000). Work, psychological well-being and perfor-
Boon, C., Den Hartog, D. N., & Lepak, D. P. (2019). A systematic review of mance. Occupational Medicine, 50(5), 304–309.
human resource management systems and their measurement. Journal Danna, K., & Griffin, R. W. (1999). Health and well-being in the workplace:
A review and synthesis of the literature. Journal of Management, 25(3),
of Management, 45(6), 2498–2537.
357–384. https://doi.org/10.1177/014920639902500305
Boselie, P., Dietz, G., & Boon, C. (2005). Commonalities and condradictions
Datta, D. K., Guthrie, J. P., & Wright, P. M. (2005). Human resource man-
in HRM and performance research. Human Resource Management Jour-
agement and labor productivity: Does industry matter? Academy of
nal, 15(3), 67–94. https://doi.org/10.1111/j.1748-8583.2005.
Management Journal, 48(1), 135–145.
tb00154.x
De Winne, S., & Sels, L. (2013). Progress and prospects for HRM-
Boxall, P., & MacKy, K. (2009). Research and theory on high-performance
performance research in small and medium-sized businesses. In
work systems: Progressing the high-involvement stream. Human
J. Paauwe, D. Guest, & M. Patrick (Eds.), HRM & performance: Achieve-
Resource Management Journal, 19(1), 3–23. https://doi.org/10.1111/j.
ments & challenges (pp. 173–196). West Sussex (UK): John Wiley &
1748-8583.2008.00082.x
Sons Ltd.
Boxall, P., & Purcell, J. (2003). Strategy and human resource management.
Delaney, J. T., & Huselid, M. A. (1996). The impact of human resource
Basingstoke: Palgrave Macmillan.
management practices on perceptions of organizational performance.
Bradburn, N. M. (1969). The structure of psychological well-being. Chicago:
Academy of Management Journal, 39(4), 949–969.
Aldine. Delbridge, R., & Turnbull, P. (1992). Human resource maximisation. In
Bryson, A., Cappellari, L., & Lucifora, C. (2010). Why so unhappy? The P. T. P. Blyton (Ed.), Reassessing human resource management. Newbury
effects of unionization on job satisfaction. Oxford Bulletin of Economics Park, CA: Sage.
and Statistics, 72(3), 357–380. Delery, J. E. (1998). Issues of fit in strategic human resource management:
Bryson, A., Forth, J., & Kirby, S. (2005). High-involvement management Implications for research. Human Resource Management Review, 8(3),
practices, trade union representation and workplace performance in 289–309. https://doi.org/10.1016/S1053-4822(98)90006-7
Britain. Scottish Journal of Political Economy, 52(3), 451–491. https:// Dess, G. G., & Robinson, R. B. (1984). Measuring organizational perfor-
doi.org/10.1111/j.0036-9292.2005.00352.x mance in the absence of objective measures: The case of the
16 HO AND KUVAAS

privately-held firm and conglomerate business unit. Strategic Manage- Jensen, J. M., Patel, P. C., & Messersmith, J. G. (2011). High-performance
ment Journal, 5(3), 265–273. https://doi.org/10.1002/smj. work systems and job control: Consequences for anxiety, role over-
4250050306 load, and turnover intentions. Journal of Management, 39(6),
Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95(3), 1699–1724. https://doi.org/10.1177/0149206311419663
542–575. Jiang, K., Lepak, D. P., Han, K., Hong, Y., Kim, A., & Winkler, A.-L. (2012).
Diener, E. (Ed.). (1994). Assessing subjective well-being: Progress and Clarifying the construct of human resource systems: Relating human
opportunities. Social Indicators Research, 31(2), 103–157. resource management to employee performance. Human Resource
Diener, E., & Larsen, R. J. (1993). The experience of emotional well-being. Management Review, 22(2), 73–85.
In M. Lewis & J. M. Haviland (Eds.), Handbook of emotions. New York, Jiang, K., Lepak, D. P., Hu, J., & Baer, J. C. (2012). How does human
NY: Guilford Press. resource management influence organizational outcomes? A meta-
Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well- analytic investigation of mediating mechanisms. Academy of Manage-
being: Three decades of progress. Psychological Bulletin, 125(2), ment Journal, 55(6), 1264–1294.
276–302. Karasek, R. A. (1979). Job demands, job decision latitude, and mental strain: Impli-
Diener, E., Wirtz, D., Tov, W., Kim-Prieto, C., Choi, D., Oishi, S., & Biswas- cations for job redesign. Administrative Science Quarterly, 24, 285–308.
Diener, R. (2010). New well-being measures: Short scales to assess Kaufman, B. E. (2012). Strategic human resource management research in
flourishing and positive and negative feelings. Social Indicators the United States: A failing grade after 30 years? The Academy of Man-
Research, 97(2), 143–156. agement Perspectives, 26(2), 12–36.
Gerhart, B. (2007). Horizontal and vertical fit in human resource systems. Keenoy, T. (1990). HRM: A case of the wolf in sheep's clothing ? Personnel
In C. Ostroff & T. A. Judge (Eds.), Perspectives on organizational fit. Review, 19(2), 3–9. https://doi.org/10.1108/00483489010004306
New York, NY: Psychology Press. Kepes, S., & Delery, J. E. (2007). HRM systems and the problem of internal
Godard, J. (2001). High performance and the transformation of work? fit. In J. P. P. Boxall & P. Wright (Eds.), Oxford handbook of human
Industrial and Labor Relations Review, 54(4), 776–805. https://doi.org/ resource management (p. 385). New York, NY: Oxford University Press.
10.2307/2696112 Keyes, C. L. M. (1998). Social well-being. Social Psychology Quarterly, 61(2),
Godard, J. (2004). A critical assessment of the high-performance paradigm. 121. https://doi.org/10.2307/2787065
British Journal of Industrial Relations, 42(2), 349–378. https://doi.org/ Kochan, T., & Osterman, P. (1994). The mutual gains enterprise: Forging a
10.1111/j.1467-8543.2004.00318.x winning partnership among labor, management and government. Boston,
Grant, A. M., Christianson, M. K., & Price, R. H. (2007). Happiness, health, MA: Harvard Business School Press.
or relationships? Managerial practices and employee well-being Kroon, B., Van De Voorde, K., & Van Veldhoven, M. (2009). Cross-level
tradeoffs. Academy of Management Perspectives, 21(3), 51–63. https:// effects of high-performance work practices on burnout: Two coun-
doi.org/10.5465/AMP.2007.26421238 teracting mediating mechanisms compared. Personnel Review, 38(5),
Guest, D. E. (2017). Human resource management and employee well- 509–525. https://doi.org/10.1108/00483480910978027
being: Towards a new analytic framework. Human Resource Manage- Kruse, D. L., Blasi, J. R., & Park, R. (2010). Shared capitalism in the US econ-
ment Journal, 27(1), 22–38. omy: Prevalence, characteristics, and employee views of financial partici-
Guest, D, & Conway, N. (2007). Human resource management, employee pation in enterprises. In J. R. B. D. J. Kruse & R. B. Freeman (Eds.), Shared
attitudes and workplace performance: An examination of the linkages capitalism at work: Employee ownership, profit and gain sharing, and broad-
using the 2004 Workplace Employment Relations Survey. Report pre- based stock options (pp. 41–75). Chicago: University of Chicago Press.
pared by the Department for Business, Enterprise & Regulatory Landsbergis, P. A., Cahill, J., & Schnall, P. (1999). The impact of lean pro-
Reform (BERR). duction and related new systems of work organization on worker
Guest, D. (2002). Human resource management, corporate performance health. Journal of Occupational Health Psychology, 4(2), 108–130.
and employee wellbeing: Building the worker into HRM. The Journal of https://doi.org/10.1037/1076-8998.4.2.108
Industrial Relations, 44(3), 335–358. https://doi.org/10.1111/1472- Larson, J. S. (1996). The World Health Organization's definition of health:
9296.00053 Social versus spiritual health. Social Indicators Research, 38, 181–192.
Guthrie, J. P. (2001). High-involvement work practices, turnover, and pro- Laursen, K., & Foss, N. J. (2003). New human resource management prac-
ductivity: Evidence from New Zealand. Academy of Management Jour- tices, complementarities and the impact on innovation performance.
nal, 44(1), 180–190. Cambridge Journal of Economics, 27(2), 243–263.
Harley, B., Allen, B. C., & Sargent, L. D. (2007). High performance work Lawler, E. E., III. (1992). The ultimate advantage: Creating the high-
systems and employee experience of work in the service sector: The involvement organization. San Francisco, CA: Jossey-Bass Pubs.
case of aged care. British Journal of Industrial Relations, 45(3), 607–633. Legge, K. (2005). Human Resource Management: Rhetorics and Realities
https://doi.org/10.1111/j.1467-8543.2007.00630.x (10th anniv). Basingstoke, England: Palgrave Macmillan.
Harley, B., Sargent, L., & Allen, B. (2010). Employee responses to ‘high per- Legge, K. (1998). The morality of HRM. In T. A. C. C. Mabey & D. Skinner
formance work system’ practices: An empirical test of the disciplined (Eds.), Experiencing human resource management (pp. 14–32). London,
worker thesis. Work, Employment and Society, 24(4), 740–760. England: Sage.
Hobson, C. J., Delunas, L., & Kesic, D. (2001). Compelling evidence of the Levine, D. I. (1995). Reinventing the workplace: How business and employees
need for corporate work/life balance initiatives: Results from a can both win. Washington, DC: Brookings Institution Press.
national survey of stressful life-events. Journal of Employment Counsel- Macduffie, J. P. (1995). Human resource bundles and manufacturing per-
ing, 38(1), 38–44. formance: Organizational logic and flexible production systems in the
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance World auto industry. Industrial and Labor Relations Review, 48(2), 197.
structure analysis: Conventional criteria versus new alternatives. Struc- https://doi.org/10.2307/2524483
tural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. Macky, K., & Boxall, P. (2007). The relationship between “high-performance
Huselid, M. A. (1995). The impact of human resource management prac- work practices” and employee attitudes: An investigation of additive and
tices on turnover, productivity, and corporate financial performance. interaction effects. The International Journal of Human Resource Manage-
Academy of Management Journal, 38(3), 635–672. ment, 18(4), 537–567. https://doi.org/10.1080/09585190601178745
Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects of human Macky, K., & Boxall, P. (2008). High-involvement work processes, work intensi-
resource management practices on productivity: A study of steel fication and employee well-being: A study of New Zealand worker experi-
finishing lines. The American Economic Review, 87(3), 291–313. ences. Asia Pacific Journal of Human Resources, 46(1), 38–55.
HO AND KUVAAS 17

Marchington, M., & Grugulis, I. (2000). “Best practice” human resource Warr, P. (1990). The measurement of well-being and other aspects of
management: Perfect opportunity or dangerous illusion? The Interna- mental health. Journal of Occupational and Organizational Psychology,
tional Journal of Human Resource Management, 11(6), 1104–1124. 63(3), 193–210.
https://doi.org/10.1080/09585190050177184 Way, S. A. (2002). High performance work systems and intermediate indi-
Meyer, J. P., & Allen, N. J. (1997). Commitment in the workplace: Theory, cators of firm performance within the US small business sector. Journal
research, and application. Thousand Oaks, CA: Sage. of Management, 28(6), 765–785.
Ogbonnaya, C., Daniels, K., Connolly, S., & van Veldhoven, M. (2017). Inte- White, M., & Bryson, A. (2011). HRM and workplace motivation: Incre-
grated and isolated impact of high-performance work practices on mental and threshold effects. CEP DISCUSSION Paper no. 1097.
employee health and well-being: A comparative study. Journal of Occu- White, M., & Bryson, A. (2013). Positive employee attitudes: How much
pational Health Psychology, 22(1), 98–114. human resource management do you need? Human Relations, 66(3),
Paauwe, J. (2004). HRM and performance: Achieving long-term viability. 385–406.
Oxford: Oxford University Press on Demand. Willmott, H. (1993). Strength is ignorance; slavery is freedom: Managing
Peccei, R. E., van de Voorde, F. C., & Van Veldhoven, M. (2013). HRM, culture in modern organizations. Journal of Management Studies, 30(4),
well-being and performance: A theoretical and empirical review. In 515–552.
D. Paauwe, E. Guest, & P. M. Wright (Eds.), HRM & performance: Wright, T. A. (2014). Putting your best “face” forward: The role of
Achievements & challenges (pp. 15–46). Chichester, UK: Wiley. emotion-based well-being in organizational research. Journal of Organi-
Pfeffer, J. (1998). The human equation: Building profits by putting people zational Behavior, 35(8), 1153–1168.
first. Boston: Harvard Business Press. Wright, T. A., & Cropanzano, R. (2000). Psychological well-being and job
Pierce, J. R., & Aguinis, H. (2013). The too-much-of-a-good-thing effect in satisfaction as predictors of job performance. Journal of Occupational
management. Journal of Management, 39(2), 313–338. Health Psychology, 5(1), 84–94.
Ramsay, H., Scholarios, D., & Harley, B. (2000). Employees and high- Wright, T. A., Cropanzano, R., & Bonett, D. G. (2007). The moderating role
performance work systems: Testing inside the black box. British Journal of employee positive well being on the relation between job satisfac-
of Industrial Relations, 38(4), 501–531. tion and job performance. Journal of Occupational Health Psychology,
Rath, T., & Harter, J. K. (2010). Wellbeing: The five essential elements. 12(2), 93–104.
New York, NY: Simon and Schuster. Wright, T. A., & Huang, C. (2012). The many benefits of employee well-
Shin, D., & Konrad, A. M. (2017). Causality between high-performance being in organizational research. Journal of Organizational Behavior, 33
work systems and organizational performance. Journal of Management, (8), 1188–1192.
43(4), 973–997.
Spector, P. E. (2006). Method variance in organizational research: Truth or
urban legend? Organizational Research Methods, 9(2), 221–232.
AUTHOR BIOGRAPHIES
Subramony, M. (2009). A meta-analytic investigation of the relationship
between HRM bundles and firm performance. Human Resource Man-
agement, 48(5), 745–768.
Takeuchi, R., Chen, G., & Lepak, D. P. (2009). Through the Looking glass of Hoang Ho received his Ph.D. from BI Norwegian Business School.
a social system: Cross-level effects of high-performance work systems He studies strategic human resource management, organizational
on employees' attitudes. Personnel Psychology, 62(1), 1–29. performance, high-performance work systems implementation,
Takeuchi, R., Lepak, D. P., Wang, H., & Takeuchi, K. (2007). An empirical
organizational commitment, workforce diversity, affect, emotions,
examination of the mechanisms mediating between high-performance
work systems and the performance of Japanese organizations. Journal turnover, work stress, health, and employee well-being.
of Applied Psychology, 92(4), 1069–1083.
Van De Voorde, K., Paauwe, J., & Van Veldhoven, M. (2012). Employee Bård Kuvaas is a Professor of Organizational Psychology at BI
well-being and the HRM–organizational performance relationship: A Norwegian Business School in Oslo. He studies framing and
review of quantitative studies. International Journal of Management
mood, motivation, social exchange, pay systems, performance
Reviews, 14(4), 391–407.
Venkatraman, N., & Ramanujam, V. (1987). Measurement of business eco- appraisal, training, and human resource management systems.
nomic performance: An examination of method convergence. Journal
of Management, 13(1), 109–122.
Wall, T. D., Michie, J., Patterson, M., Wood, S. J., Sheehan, M.,
Clegg, C. W., & West, M. (2004). On the validity of subjective mea-
sures of company performance. Personnel Psychology, 57(1), 95–118. How to cite this article: Ho H, Kuvaas B. Human resource
Wall, T. D., & Wood, S. J. (2005). The romance of human resource manage-
management systems, employee well-being, and firm
ment and business performance, and the case for big science. Human
Relations, 58(4), 429–462. performance from the mutual gains and critical perspectives:
Walton, R. E. (1985). From control to commitment in the workplace. Har- The well-being paradox. Hum Resour Manage. 2019;1–19.
vard Business Review, 63(2), 76–84. https://doi.org/10.1002/hrm.21990
Warr, P. (2011). Work, Happiness, and Unhappiness. New York: Psychology
Press.
18 HO AND KUVAAS

APPENDIX

TABLE A1 HRM practices and their links to the AMO model bundles

HRM practices Skills bundle Motivation bundle Opportunity bundle


Job autonomy x
Teamwork x
Staff training X
Performance-related pay x
Flexible work x
Selective hiring X
Supportive management x
Information sharing x
Participatory decision-making x

TABLE A2

Items in the Human Resource Management System Index

Variables Observed items Response scale


Job autonomy How much influence do you have over 1 = none to 4 = a lot
what tasks you do?
How much influence do you have over the
pace at which you work?
How much influence do you have over how
you do your work?
How much influence do you have over the
order in which you carry out tasks?
Teamwork Team members depend on each other's 0 = no to 1 = yes
work to be able to do their jobs.
Tasks or roles rotate among the members of
the team.
Staff training Training either paid for or organized by your 0 = have had no training to 1 = have had
employer during the last 12 months? some training
Performance-related pay Payments based on the overall performance 0 = otherwise to 1 = yes
of the group or team?
Payments based on the overall performance 0 = otherwise to 1 = yes
of your workplace or organization?
Flexible work Are flexi-time arrangements available to 0 = no to 1 = yes
you?
Are job sharing arrangements available to
you?
Are chances to reduce your working hours
available to you?
Are working at or from home in normal
working hours available to you?
Selective hiring Individuals' skills are important when 0 = not selected to 1 = selected
recruiting new employees.
Individuals' qualifications are important
when recruiting new employees.
Individuals' experience is important when
recruiting new employees.
(Continues)
HO AND KUVAAS 19

Continued

Variables Observed items Response scale


Supportive management Managers understand employee 1 = strongly disagree to 5 = strongly agreed
responsibilities outside of work.
Managers encourage staff to develop their
skills.
Information sharing Managers keep employees informed about 1 = very poor to 5 = very good
the way job is done.
Managers keep employees informed about
changes in staffing.

Managers keep employees informed about


financial matters.
Participatory decision-making Managers seek employee views. 1 = very poor to 5 = very good
Managers respond to employee
suggestions.
Employees influence final decisions.

You might also like