Human resource

management and
December 1999
performance
Stephen Wood

The study of human resource management (HRM) has been invigorated by the promise
that there is a best-practice, high-involvement management (HIM) that can guarantee
superior organizational performance. None the less, there remain concerns that
contingency theory still rules, that is, that the fit between the human resource systems
and their context, and particularly the organization's business strategy, is all important
and, thus, that HIM will only outperform other systems in certain circumstances. In the
1990s, there has been a spate of research that has sought to test whether HIM is
indeed universally relevant. This paper reviews these studies. The paper first introduces
the conceptual dimensions of the debate concerning HRM and performance. This
shows that the issues go beyond a simple competition between universalism and
contingency theory. There are more complicated hypotheses linking human resource
practices beneath the surface of the recent literature. The second part of the paper
overviews the studies in the light of these hypotheses, revealing that they present an
uneven picture. Firstly, there are conceptual differences underlying the studies and,
secondly, the results vary between them, and the effects of HIM vary between
performance measures even in particular studies. Though a fair number of the studies
claim to support universalism, their claims are not always unequivocally supported by
their research evidence, and it is premature to conclude in its favour. If anything, there
is more support for the `lean production' argument that stresses the interaction effect
Stephen Wood is from the between HIM and total quality management on performance.
Centre for Economic
Performance, London
School of Economics,
London, UK. following three fits: (1) that between
Introduction: internal versus strategic fit
human resource (HR) practices, (2) that
Synergy, fit and integration are the key between these practices and more
ß Blackwell Publishers Ltd 1999, concepts in modern human resource specifically between HR systems –
108 Cowley Road, Oxford OX4 management (HRM) theory. Attention has coherent sets of HR practices – and other
1JF, UK and 350 Main Street,
Malden, MA 02148, USA been given to one or more of the systems within the organization, (3) that

International Journal of Management Reviews Volume 1 Issue 4 pp. 367–413 367

Human resource between HR systems and the business or Recent discussion has in fact focused on
management and competitive strategy of the organization. the particular set of practices that is
performance There has been in the recent literature little associated with high commitment and
direct consideration of a fourth fit, that involvement. It includes practices such as
between the HR system and the job flexibility, teamworking and minimal
organization’s environment, though this is status differentials that are assumed to
captured indirectly through the concept of reverse past Taylorist methods. These are
the organization’s strategy, this being the assumed to generate high performance
management’s posture towards the through their enhancement of the
environment (Youndt et al. 1996, 87). commitment and skills of employees.
Throughout this paper the first fit will be Variously designated as high-commitment,
called the internal fit of the HR system, high-involvement or simply innovative
the second the organizational fit, the third work practices, the term high-performance
the strategic fit, and the fourth the practices is increasingly being used to
environmental fit. label them, as if their performance-
The bedrock of all recent discussion is enhancing qualities can be taken for
the first, internal fit. It is argued (see, for granted, in the way that a drug might be
example, Baird and Meshoulam 1988, 122) labelled by the disease it is known or
that certain HR practices blend better than purports to cure. None the less, while there
others do, and it is sensible to select is an apparent widespread ‘‘agreement
practices in conjunction with and not in about what constitutes high-performance
isolation from each other. Consequently, practices’’ (Pfeffer 1997, 169), somewhat
when they are found together, the pay-offs paradoxically given the label, there is no
will be greater than the sum of those from consensus about whether they represent the
their individual elements. It is even argued one best way of approaching HRM.
(Barney 1995, 56) that individual practices More specifically, the debate is whether
‘‘have limited ability to generate high-performance systems will universally
competitive advantage in isolation’’ of outperform all other systems or whether
other synergistic practices. The implication the optimal system is relative to the
is that individual HRM practices should be circumstances of the firm (Becker and
viewed as part of a wider system – a HR Gerhart 1996; Becker and Huselid, 1998a;
system – and a key criterion for judging Fenton O’Creevy 1998, 7–13; Youndt et
such systems is whether their individual al. 1996). In discussions on the contingent
components, the HR practices, fit together nature of HR systems, the significant
in a coherent manner. A corollary of this context of the system is generally taken to
is that part of the explanation for the be the organization’s strategic posture
observed differences in organizational (Youndt et al. 1996, 87), the implication
performance lies in the differential usage being that it is the organization’s approach
of these synergistic combinations of HR to its environment rather than this directly
practices. The first issue is, then, whether that is crucial. Thus, the contingency
it is simply sufficient to adopt an theory of HRM implies that, in general,
integrated set of practices or whether HR systems with internally fitted practices
ß Blackwell Publishers Ltd 1999 certain combinations are better than others. that match the strategy of the firm will

368
perform best. In contrast, strong advocates the specific link between high-involvement
(e.g. Pfeffer 1994) of the high-performance systems and total quality management or
management system argue that it lean production. Is it the case that high-
outperforms all other systems in all involvement systems are the personnel side
contexts. Thus, the high-involvement of such manufacturing strategies and,
practices to which they now give the label hence, are only likely to be found when
high-performance indeed have the right to organizationally fitted to these, or is it
be given this status, while, from a simply that, when high-involvement and
contingency perspective, such labelling is total quality systems are used together, the
misleading. This debate is thus reducible performance effects of each will be December 1999
to whether high performance is associated magnified? More recently, attention has
with a specific set of internally consistent been given to whether the HR system is
(high-involvement/performance) practices organizationally fitted in general and not
or rather with the strategic fit of HR simply whether it is fitted to the particular
systems. Accordingly, it can be resolved organizational logic of lean production or
empirically. The burgeoning empirical total quality management (TQM). The
literature on the HRM–performance link argument is that high-performance systems
has indeed focused on testing whether that are integrated with the wider
high-involvement/performance systems are organization will yield stronger results than
universally the best (the best practice those that are not. Third, the results of the
hypothesis) or only in certain empirical studies are not uniform, not least
circumstances (the best-fit hypothesis). because their underlying concepts and
This review will concentrate on this designs differ. Overall then, the terrain of
literature. It will, however, show that issues debate is more open and muddier than is
raised by the literature cannot be reduced often presumed.
simply to a matter of whether the validity The paper is divided into two parts: the
of the high-involvement/performance first part introduces the conceptual
system is universal or contingent. First, discussion surrounding the studies of high-
there are conceptual differences underlying involvement/performance management,
the different terms being used to label the and the second part reviews the studies
specific bundle of involvement practices themselves.
that has been at the forefront of HRM, and
these imply different conceptions of the
Theoretical Developments in the Human
relationship between HR practices and
Resource Management and Performance
performance. Second, the literature
Debate
addresses issues that extend beyond a
simple competition between the The Terminology Surrounding High-
contingency and universalistic perspectives, involvement Management
with its emphasis on testing the
significance of strategic fit. In particular, The common element in all the talk of
there is an issue running through the high-commitment, high-involvement and
literature concerning the importance of high-performance management is the
organizational fit. Initially, this focused on contrast with a Taylorist, control type of ß Blackwell Publishers Ltd 1999

369
Human resource management. The prime purpose of this Walton (1985) (as seemingly do others,
management and section is to assess whether the e.g Wood and Albanese 1995) tends to use
performance proliferation of terms to describe this non- the term ‘high-commitment management’
Taylorism entails differing conceptions, or in the less restrictive way, and his
whether it is simply a matter of semantics. characterization of high-commitment
High-commitment management, following practices draws heavily on observations of
Walton (1985), is generally characterized as leading-edge developments in
entailing (a) a particular orientation on the organizations in the USA that have been
part of employers to their employees, based attempting to abandon their Taylorist
on an underlying conception of them as heritage. Job design theory is, though,
assets to be developed rather than as particularly significant to his thinking.
disposable factors of production, and (b) the Moreover, despite the emphasis on
combined use of certain personnel practices, commitment, Walton’s presentation does
such as job redesign, job flexibility, not rule out the possibility that high-
problem-solving groups, teamworking and commitment practices might impact on
minimal status differences. performance through their having an effect
A literal definition of high-commitment on the competencies of employees.
management would (a) only include in its The term ‘high-involvement
designated combination of practices those management’, as popularized by Lawler
that are known to affect organizational (1986), is equivalent to the less restrictive
commitment, and (b) assume that their interpretation of high-commitment
impact on organizational performance is management. Lawler has certainly drawn
only because of their effect on attention to the importance of enlarging
organizational commitment. A less literal the skills and knowledge of all employees,
version would not limit it to practices with as well as to psychological processes other
verified effects on commitment, but rather than intrinsic motivation such as goal-
would include all practices either setting, and even suggesting a role, at least
hypothesized to impact on commitment in some circumstances, for money
and/or associated with organizations motivation.
known to be forsaking Taylorist methods.
It would also extend the ways in which
High-performance Management
these practices enhance performance
beyond organizational commitment and, in ‘High-performance management’ is the
particular, would emphasize the role of most recent term used to describe the non-
knowledge and skill acquisition. In Taylorist form of organization. Yet,
addition, organizational commitment might paradoxically, in the 1990s, as writers
be treated as an umbrella term for a whole have increasingly gravitated towards the
set of interrelated attitudes and term ‘high-performance work systems’,
orientations, such as having a flexible role different perspectives underlying its usage
orientation, strong group-orientation, a have become more visible.
business and quality consciousness and a First, it may be used as a synonym for
willingness to accept change and high-commitment or -involvement
ß Blackwell Publishers Ltd 1999 contribute to innovation.1 management, particularly in order to

370
highlight its benefits when marketing HR such skills depends on workers being
models to managers who appear increasingly motivated to contribute their discretionary
oriented towards bottom-line results. This effort to resolving problems, organizations
usage may also reflect a growing confidence must ‘‘make a reciprocal investment in
in its economic effects as research began to their well-being’’. Consequently, flexible
demonstrate positive outcomes. production is characterized by a ‘‘bundle’’
Secondly, it is treated as a broader of high-commitment (MacDuffie 1995,
notion, as the unity of this high- 201) or high-involvement (Pil and
involvement management (HIM) and TQM MacDuffie 1996) practices, as well as the
or lean production (Womack et al. 1990). operational methods emphasized by December 1999
Lawler, himself, has begun to use the term Womack et al.
‘high-performance management’ in this The third use of the term ‘high-
way (Lawler et al. 1995). Definitions of performance management systems’ has
TQM and HIM methods typically overlap arisen out of a desire on the part of some
as teamworking, quality circles and the to broaden the focus away from
sharing of performance data are included employees’ attitudes and commitment, so
in both. But Lawler et al. go further and that such factors as skill formation, work
argue that the two sets of practices are structuring, performance management and
‘‘complementary in their impact on pay satisfaction are included in the list of
organizational performance’’. They thus mechanisms through which HR practices
recommend that TQM and employee may impact upon performance. Hence,
involvement ‘‘should be treated as an high-performance management is not just
integrated approach to organizational associated with high-commitment
transformation’’, since ‘‘employee management. Taken to its extreme, this
involvement without TQM practices is less means a management in which high
likely to affect performance positively and performance is targeted directly, not
vice versa’’ (Lawler et al. 1995, 144). indirectly via attitudinal structuring. Its
This second notion of high-performance core would be the use of performance
management is similar to the concept of management methods that are centred on
flexible production systems used by the setting of goals and the linking of pay
MacDuffie (1995) and other industrial to their achievement. It is certainly the
relations specialists (e.g. Kochan) close to case that two of the foremost users of the
the MIT auto project (where the term ‘lean term high-performance management,
production’ originated). To MacDuffie’s Becker and Huselid (1998a, 78), accord
(1995, 198) mind, Womack et al. (1990) ‘‘performance-contingent incentive
defined lean production largely in compensation systems’’ a significant, if not
operational management terms, as they pivotal, role in their characterization of the
focused on the minimization of buffers term. This does set the concept of high-
associated with just-in-time, and thus performance management apart from
neglected ‘‘the expansion of work force earlier concepts of high-commitment
skill and conceptual knowledge required management, since many of its proponents
for problem-solving under this approach’’. (e.g. Beer et al. 1984, 114) – as well as
Since the application and development of advocates of TQM (see e.g. Deming, ß Blackwell Publishers Ltd 1999

371
Human resource 1986) – associate it with the control in which of the three ways it is being
management and strategy and are deeply critical of used, is that it has the potential to
performance performance-related pay. (See Wood outperform all other types of management.
(1996a) for a review of pay systems and The danger is that it promotes a particular
high-commitment management.) model of management before its
None the less, merit pay is, as Foulkes’ performance effects have been adequately
(1980) research showed, used in high- proven. In what follows, the term ‘high-
profile non-union companies such as IBM, performance management’ is used in either
which are often thought to practice high- the second or third of the above three
commitment management. Consequently, ways and only then when referring to
to the extent that the model of high- specific writers who adopt the concept.
commitment/performance management is The term ‘high-involvement management’,
based on observation of leading-edge which does not necessarily assume that its
developments rather than a theory of effects are established and can readily
commitment, it is likely that merit pay will admit of non-commitment routes such as
be included in its elements. Lawler (1991, skill acquisition, will be used to refer to
519) in fact does not reject performance- the less restrictive model of high-
related pay, as he sees money as a commitment management.
potential motivator; none the less his
emphasis is on group-based incentives and
Human Resource Management and
profit sharing (as well as incentives for
Performance
training – skill-based pay) as he treats
individual incentive plans as not very Universalistic vs. contingent best
supportive of the commitment strategy practice. Underlying all three concepts is
(Lawler et al. 1995, 20). the notion of the HR system as a
It has been shown that there are clear synergistic set of personnel practices and
differences in the perspectives underlying the importance of internal fit of
the various terms for the non-Taylorist performance. Moreover, regardless of
organization and, moreover, that the same which concept of the human resource
terms can be used in differing ways. This system authors work with – high-
variety is particularly pronounced in the commitment, -involvement or -performance
case of high-performance management. If – most acknowledge the possibility that it
the intention is to draw attention to the could, at least theoretically, be universally
importance of performance management, or contingently relevant. This is so even
goal-setting and performance-related pay, though the term high-performance
then its usage does reflect a different management would appear to carry an
perspective from that underlying the earlier implication that it is a best-practice model.
conceptions of high-commitment Certainly advocacy of any of the three
management, which stressed commitment models carries this implication. Proponents
as a particularly important route through of univeralism and contingency theorists
which involvement practices impact on differ in terms of the additional criterion to
performance. The implication of the term internal fit by which they judge HR
ß Blackwell Publishers Ltd 1999 ‘high-performance management’, no matter systems. The former, universalists,

372
concentrate on judging personnel cost-minimization, and between HIM and
management by the extent to which it his quality-oriented strategy. The
fosters employee involvement and the availability of such generic fits provides a
human capital of the organization, on the user-friendly way to incorporate the human
assumption that a synergistic set of factor into strategy formulation and ensure
practices that reverses key features of the a role for personnel managers in this
past ‘command’ Taylorist organization is process, as they have the expertise to
the best way of managing employment. advise on the most appropriate HR system
This general argument can be formulated for a particular situation. This approach
thus: accords with a central tenet of all HRM December 1999
that ‘‘human resources must . . . become an
High-involvement/performance HR systems integral component of the strategic
will have positive effects on organizational planning process’’ (Guest 1987, 512).
performance (H1). What it does not do, however, is guarantee
that the HR will be accorded a high status
By contrast, for contingency theorists, in the competitive strategy or that it will
strategic fit – the integration of HR be treated as a major asset to be
systems with the overall strategy of the developed. If the firm is adopting the cost-
organization – is the additional criterion. It minimization strategy under this alignment
is aligned HR systems that will perform approach, a Taylorist low-involvement
best. The principle that HR systems should approach is appropriate. The second,
be designed in the light of the contingency, thesis can be stated thus:
organization’s strategy means that they
would be derived from the corporate Strategic factors will moderate the
strategy. Human resource planning would relationship between high-involvement HR
be downstream from the corporate systems and organizational performance
strategy, yet incorporated into the (H2).
mainstream of business planning. Porter’s
distinction between two generic strategies, The dispute within the area between those
cost minimization and innovative/quality who think the non-Taylorist organization –
strategies, is often used as the basis on regardless of how constituted – is a best-
which to differentiate contexts (e.g. Hoque practice model, and those who view its
1999; Schuler and Jackson 1987), and it is relevance as dependent on the
argued that the Taylorist system fits the organization’s strategy and context, is
former strategy, while HIM is more fundamental. The main axis of debate in
appropriate for the latter. Human resource recent literature has indeed become the
theory mirrored the development of two competing hypotheses (H1 and H2).
generic strategies by specialists in business
strategy and sought to provide HR
The High-involvement-TQM/Lean
strategies that fitted each of these. Schuler
Production Link
and Jackson (1987) provide the best
example of this, as they articulate the link Yet the issue cannot be laid to rest here.
between the control approach and Porter’s First, underlying the second definition of ß Blackwell Publishers Ltd 1999

373
Human resource high-performance management is the concern, the elevation of TQM or lean
management and implication that high-involvement and total production to the status of a strategy and
performance quality or lean production methods ought not just a tactic of operational
to go together and that, when they do, the management. This is certainly the
pay-off from each is increased. It is in fact approach taken by Beaumont (1995, 40)
based on a particular perspective on TQM, and is consistent with many of Kochan’s
which assumes that it ‘‘cannot be (e.g. Kochan 1992; Kochan and Osterman
successful without employee involvement’’ 1995, 50, 79–89) writings. In Beaumont’s
(Lawler et al. 1995, 45). At a minimum, (1995, 40) terms, an organization with a
employees will receive a substantial strong commitment to HRM ‘‘competes
amount of information on quality and on the basis of product quality and
training in TQM methods but, more differentiation as well as price’’ and
significantly, they will be involved in adopts personnel polices that include a
monitoring the quality of their own work high investment in training, employment
as well as in the design of ‘‘work methods security guarantees and functional
and work procedures that influence flexibility. This overall approach has
quality’’ (Lawler et al. 1998, 7). Such sometimes been labelled the high road
involvement was a key feature of the (Fenton O’Creevy 1998, 10; Osterman
pioneering of quality methods in Japanese 1994) to sustained economic performance,
manufacturing. Womack et al.’s (1990, and is often put forward as the only
91–93) documentation of the Japanese viable option for modern organizations, at
successful pursuit of this ‘‘lean least in OECD countries. Its significance
production’’ suggests that the pursuit of is derived partly from the inability of such
quality and cost minimization are not countries to compete with the wage rates
mutually exclusive, and that its of less developed countries, but more
distinctiveness is precisely that it can importantly because the Japanese adoption
achieve both simultaneously.2 of the lean production approach has
The implication is that Porter’s (1985) shown that quality and cost minimization
dichotomy between cost-minimization and are not antithetical.
quality-oriented competitive strategies is This approach thus sees HIM as a
no longer valid and, hence, it takes us necessary but not sufficient basis for high
beyond the contingency argument that performance. Organizational fit is the
high-involvement is only of relevance to additional element and, more specifically,
organizations with a quality-enhancement the fit between HIM and TQM. It can be
or innovative strategy. All organizations formally stated thus:
have to compete on both price and quality
and, by tending to quality, they should be
able to reduce costs. The implication is Organizations adopting high-involvement
management in alignment with total
that high-performance management
quality management should outperform all
defined as the combination of HIM and other organizations and there is an
total quality methods is universally interaction effect between the adoption of
relevant. High-involvement management high-involvement management and total
ß Blackwell Publishers Ltd 1999 thus gets defined as part of a wider quality management on performance (H3).

374
The Importance of a Strategically Fitted HIM by default rather than by design –
High-involvement Management perhaps ‘‘through outside guidance, . . .
emulation, or . . . following fads’’ (Guest
Guest, though he initially wrote (1987) and Hoque 1994, 17) – are labelled
before the lean production vogue, ‘‘lucky’’. The lack of an underlying
effectively goes one step beyond this strategic orientation in these lucky
argument, proposing that all elements of organizations is then assumed to reduce or
the organization’s strategy must be even undermine the potency of the high-
integrated. Thus, though HIM is for him involvement model. We would then expect
superior to other methods, its effects will HIM to have a significant impact on December 1999
be moderated by whether it is – in Guest’s performance in all circumstances, but that
terms – strategically integrated into the the extent of this will depend on the
business strategy. At first sight, this degree of strategic integration. Taking this
appears to be melding the impossible, i.e. to be Guest’s line of argument, it can be
the external fit of contingency theory with formally stated thus:
a universalistic treatment of HIM as best
practice. None the less, Guest marries the High-involvement management has a
two by implying that HIM is compatible positive effect on organizational
with any strategy; it is the fact that there is performance, but in addition there is an
a strategic approach to personnel in the interaction effect between high-involvement
organization that is important, not that the management and strategic integration on
HIM is linked to a particular strategy. He performance (H4).
thus differentiates a strategically aligned
HIM from one without such a strategic Nowhere in his writing does Guest
underpinning. He defines this as HRM, really spell out why there is the added
thus treating both HIM and strategic fit as effect from strategically integrating high-
core components of it, the latter being involvement systems. He seems to have in
defined in a way that implies that the mind the commonly observed phenomena
elevation of HR to the strategic level and of the powerless personnel department,
the strategic integration of personnel which is disconnected from mainline
management go hand-in-hand (Guest business concerns. Strategic integration is,
1987). Guest’s point of departure is not, then, the institutional integration of
though, to assume that HIM and strategic personnel management. This may lead to
fit will coexist. human resources being given more weight
If HRM is defined as strategic HIM, in the organization (though it may be a
then all HRM entails their coexistence, but cause of this) and the tailoring of high-
not all HIM is HRM. High-involvement involvement practices to the organization’s
management is best practice but, where it requirements. Business strategy thus enters
is strategically integrated, its benefits will the HR system through its operation, not
be greater than where this underpinning through its design. We can illustrate how
does not exist. In Guest’s later writings, he this might work with the case of the
labels the former ‘‘good’’ HRM, while appraisal process; in general it may be
those managements that have arrived at aimed at involvement and identification, ß Blackwell Publishers Ltd 1999

375
Human resource but it achieves the strategic integration of the Porterian developments in the strategy
management and the individual through the alignment of literature, so Huselid and Becker mirror
performance personal goals with the strategic goals of developments in business strategy and
the organization, an outcome achieved particularly the advent of the resource-
within the appraisal process. Such based theory of the firm. In this, the
adaptations of HR practices ensure the strategy of the firm is no longer viewed in
strategic integration at the individual level, terms of businesses and products, but as
so that employees have a clear concept of founded on core competencies. Since these
the organization’s goals and how their own are both reflected in and created by the
tasks fit into their achievement (Guest skills and talents of the individuals in the
1987). This kind of argument mirrors that organization, HR becomes a major input to
of Pfeffer (1994, 65), in which HIM is the organization’s strategic initiatives.
best practice but the way in which it is Moreover, it is argued that competitive
implemented may be affected by advantage depends on inimitable resources
contingent factors.3 Following this rational and since the ‘‘traditional sources of
reconstruction of Guest along these lines, competitive advantage (quality, technology,
it should be said that there is an overlap economies of scale . . .) have become easier
between his approach and the lean to imitate’’ (Becker and Huselid 1998a,
production-based argument. In his initial 54), a premium is now placed on the skills,
framework paper, he gives some weight to knowledge and aptitudes of an
the importance of the goals that underpin organization’s workforce as a source of
management’s design of the HR system inimitable assets. The implications of this
and, in particular, associates HRM with are, firstly, that managers must focus on
the pursuit of quality and flexibility. ‘‘identifying and solving the human capital
elements of important business problems’’,
defined as ‘‘problems [that] are likely to
Human Resource Management as the
impede growth, lower profitability, and
Driver of Competitive Advantage
diminish shareholder value’’ (Becker et al.
Becker and Huselid, in several papers in 1997, 39), and, secondly, that top
the 1990s (Becker and Gerhart 1996; management treats the HR system as a
Becker and Huselid 1998a,b; Huselid 1995; potential strategic lever for the
Huselid and Becker 1996) take the organization, an essential element of the
argument another stage further. Their firm’s ‘‘strategic infrastructure’’ (Becker et
emphasis is on the elevation of the HR to a al. 1997, 39–40).
central role in the organization’s strategy. First, the HR system plays a strategic
The HR strategy is integral to the corporate role in ensuring the effective
strategy, and, it is no longer downstream of implementation of the organization’s
business strategy as in contingency theory, strategy by delivering behaviours that are
or even, as in Guest’s work, simply necessary for its achievement. Secondly,
integrated with it. The logic of their the HR system facilitates the development
argument is that HR matters are upstream of the human capital that defines the core
of the product market strategy. Much as competencies that are required to give the
ß Blackwell Publishers Ltd 1999 Schuler and Jackson in the 1980s reflected organization a sustained competitive

376
advantage based on unique assets. To fulfil organization’’ and produces the tacit
its role successfully, the HR system must, knowledge that constitutes a major
according to Becker and Huselid (1998a), invisible asset of the firm that is difficult
be internally coherent, externally aligned, to imitate (Becker and Huselid 1998a, 58).
and effectively implemented and it too None the less, Becker and Huselid’s
must not be easily imitated. The HR definition appears consistent with the
systems should be highly idiosyncratic and universalism of high-performance
tailored to each firm’s individual situation. management, but this clashes with the
Moreover, if they are successful at principle that the high-performance system
maintaining and producing inimitable must be fitted to the strategy, or at least December 1999
human assets, the logic is that business the strategic intent, of the organization, in
strategies will be increasingly distinctive. order for it to foster the distinctive
Idiosyncratic HR systems will thus be competencies on which strategies are
coexisting with a range of distinct founded. Becker et al. (1997, 41) are,
strategies and themselves creating this furthermore, wary of the notion of best
increasing diversity. So, rather than seeing practice, as it is not consistent with the
high-performance management as the emphasis on the inimitability of assets in
model fit for the quality-enhancing or the resource-based theory. Were there a
product-differentiation strategy à la best practice, it could be easily copied
contingency theory, the differentiation and through benchmarking. They are also
distinctive quality of the organization’s concerned that the contingency theory
offerings are seen as a consequence of (with its model of strategic integration as
high-performance management. external fit) only produces a ‘‘limited
Becker and Huselid work with the third range of strategy-HRM matches’’ so that
concept of high-performance management, following it will not produce any
including as it does a strong role in substantial competitive advantage (Becker
performance-related pay, writing that it is and Huselid 1998a, 58).
‘‘generally thought to include rigorous Becker and Huselid in fact argue that a
recruitment and selection procedures, reconciliation of the universalistic and
performance-contingent incentive contingency theses is possible, and,
compensation systems and management somewhat confusingly, conclude that high-
development and training activities linked performance management both is and is not
to the needs of the business’’ (Becker et a ‘‘best-practice’’ model (Becker and
al. 1987, 40). High-performance Huselid 1998a, 59). There are, however,
management is presented as if it were the two different arguments discernible in the
only internally consistent approach that growing body of their work. First, Huselid
can have a strategic impact. The specific (1995, 644) argues, along the lines of
set of high-performance practices Becker Guest, that high-performance practices
and Huselid identify is, by implication, should lead to positive outcomes for all
unique in being able to link the types of firms but contingent, firm-specific
employees’ behaviour to the strategy of factors enter through their implementation.
the firm. In particular, it sends ‘‘consistent Successful tailoring of them so as to elicit
signals regarding valued behaviors in the behaviours that are in line with the two ß Blackwell Publishers Ltd 1999

377
Human resource strategic requirements of the organization Borrowing MacDuffie’s concept of
management and should ‘‘at the margin’’ (Hueslid 1995, 644) organizational logic, they argue that a
performance achieve extra performance. This supportive organizational logic is
reconciliation first implies that high- ‘‘required to leverage the HRM system’’
performance management is the main effect (Becker and Huselid 1998b, 2). This is
and that there is a weak interaction effect operationalized as meaning that it is the
between it and strategic integration (H4 ‘‘complementarities between the HRM
above). system and other organizational policies’’
Secondly, Becker, first writing with that enable the high-performance system to
Gerhart (Becker and Gerhart 1996) and have a strategic impact. In the terms of
then with Huselid (Becker and Huselid this paper, it is the organizational fit of the
1998a,b), goes further and argues that HRM system that is crucial, the hypothesis
contingencies enter the determination of is that an organizationally fitted high-
the practices themselves. Distinguishing performance management system should
practices from principles (which they call outperform all others.
the architecture of a HR system),4 Becker The first part of this review has exposed
and Gerhart (1996, 786) argue that the differing concepts underlying recent
principles are generalizable and universal, discussion of HRM, and in so doing has
but contingencies and firm-specific identified differing concepts of the HRM–
aspirations enter into the design of performance link: (a) it may be treated as
practices (and policies), and any idea that a general relationship (H1), though there is
there are ‘‘best-practice’’ practices is still the issue of whether it is mediated
mistaken, since they must be ‘‘very firm principally by its effects on employees’
specific and idiosyncratic’’.5 For example, commitment and sense of involvement or
linking rewards to performance is a high- their skills; (b) it may be treated as
performance principle. According to this contingent, i.e. moderated by a third
line of argument, universalism is factor, namely corporate strategy (H2); (c)
unequivocally (and perhaps only) it is affected by whether the HRM is
applicable to the higher levels of the HR embedded in lean production (H3); or (d) it
activity. This second attempt at a is moderated by the integration of the
reconciliation of universalism and HRM within the overall organization at
contingency theory implies that the link either the institutional or the cognitive
between high-performance practices and level (H4).
performance is moderated by strategic
integration and the interaction effect
Empirical Study of the Human Resource
dominates. None the less, in Becker and
Management--Performance Relationship
Huselid’s (1998a, 87) latest work, the
HPM system remains the main effect, this Ideally, research that aims to assess the
and strategic alignment (or what they call competing hypotheses identified in the first
the broader ‘‘supporting organizational part should include data on the following
logic’’) are treated as two separate variables: high-involvement practices, total
elements in a high-performance quality practices, the extent of strategic
ß Blackwell Publishers Ltd 1999 organization, rather than as ‘‘substitutes’’. integration, the extent to which the HR is

378
central to the firm’s competitive strategy, they are best discussed separately. They
and organizational performance measures. will be grouped according to the
It might also include data on whether the hypothesis(es) that they seek to test.
firm is pursuing one or other of Porter’s The review only covers studies that
generic strategies of cost minimization or measure HR systems by a broad range of
high quality. In the cases where it is practices and have organizational
following a hybrid, further information performance measures. It thus omits
would be required to gauge whether this research that (a) has a limited number of
reflects a hesitant or ambiguous strategy or HR practices (e.g. Dunlop and Weil 1996),
is an attempt to combine high quality and (b) assesses whether high-involvement December 1999
cost minimization in the manner of the practices tend to coexist as one would
lean production model. Finally, the latest expect if they are synergistic, but does not
terminology of Becker and Huselid implies go on to examine their performance effects
a need to differentiate between high- (e.g. Wood 1999; Wood 1996b and
performance principles and practices, and Albanese 1995), or (c) examines the effects
to test their impact on performance of HIM on individual attitudes (e.g. Godard
separately so that data on at least the 1999; Guest and Conway 1999; Parker et
principles, if not philosophy of al. 1997). It describes the studies largely in
management, would be required in their own terms. An element of rational
addition to those on HR practices. reconstruction is, however, inevitable, but
No one data set includes all such this will be largely a matter of putting the
information. The studies that have studies in the context of the debate and
attempted to assess the impact of HRM terminology set out in the first part.
systems on performance have concentrated
on Hypothesis 1 (universalism) either
Studies that Test Hypothesis 1, the
alone or in conjunction with one or other
Universal Hypothesis
of Hypothesis 2 (contingency theory) or an
investigation of lean production Betcherman, McMullen, Leckie and
(Hypothesis 3) or strategic integration Caron’s study of the Canadian workplace
(Hypothesis 4). They have focused on in transition. Betcherman et al. (1994)
collecting data on the use of HR practices report the results of a questionnaire survey
in order to construct a measure of the HR of 714 establishments (a response rate of
system and to assess whether there is an 42.4%) in four sectors of the Canadian
association between this and certain economy – wood products (representing a
performance measures. Underlying all the resource industry), fabricated metal
studies is a concern for practices that products (a traditional manufacturing
develop either employee involvement or industry), electrical and electronic products
skills, i.e participation and human capital. (a high value-added manufacturing
Beyond this though, because authors have industry), and selected business services (a
worked with differing terminology and dynamic service sector). Information was
underlying concepts, there is no consistent acquired on 12 HR practices, which
relationship between the terms and the include formal job design, formal
bundles of practices used. Consequently, employee-participation programmes, ß Blackwell Publishers Ltd 1999

379
Human resource vacancies filled by internal promotion, profits – are not statistically different
management and variable pay and family-car benefits. The between the clusters. So, the high-
performance list also included a measure of what we involvement model as measured by the
have been referring to as the strategic participation cluster is found to be superior
integration of human resources, namely to traditional management on some
‘‘formal human-resource planning’’ that is measures, most of which are most directly
‘‘integrated into overall business planning’’ related to the HR.6
(Betcherman et al. 1994, 50). Betcherman
et al. find a strong correlation between Arthur’s steel study in the USA. Arthur’s
many of the 12 practices and so use (1994) study concentrates on the US steel
cluster analysis to identify three HR industry, the sample being 30 mini-mills.
systems: traditional, which is characterized Using cluster analysis, he identified six
by the absence of any of the 12 practices, types of HR systems (or what he calls in
though there is a medium use of three one paper (1992) industrial relations
practices, internal promotion, merit systems), defined by their differential use
promotion and high wages; compensation, of 10 practices: pure type cost reducers,
in which there is high use of 10 practices conflictors, inducers, collective bargainers,
but no use of job design or participation involvers and pure commitment
programmes; participation, in which there maximizers (this is reported in depth in
is a high use of many of the practices and Arthur 1994). He then classified these in
particularly job design and participation terms of the control–commitment
but a low use of high benefits, high wages, dichotomy, the former – control strategy –
flexible scheduling and merit promotion. being, in his terms, oriented towards
In some respects, the compensation cluster reducing direct labour costs and
is even less traditional than is the compliance with specified rules and
participation cluster, though it lacks two of procedures, the latter – the commitment
the core non-Taylorist practices associated strategy – towards ‘‘forging psychological
with HIM. Betcherman et al. (1994, 66– links between organizational and employee
67) assessed the relative performance of goals’’ (Arthur 1994, 672). Pure-cost
the three clusters by the reported reducers, conflictors and inducers are
improvements over a 5-year period forms of control systems, and collective
(between 1988 and 1993) on 11 measures. bargainers, involvers and pure-commitment
The participative system is superior to the maximizers are commitment-maximizing
two other systems on two labour outcomes systems. Mills using the control model
(lay-offs and accidents), while both it and have significantly lower levels of
the compensation cluster are superior to productivity (measured by the labour hours
the traditional one on quits. The to produce the finished product) and
improvement in unit costs is the only other higher scrap rates (the number of tons of
performance measure on which both raw steel that had to be melted to produce
clusters outperform the traditional. The 1 ton of finished product) than were the
other performance measures – labour commitment maximizers. The statistical
productivity, customer complaints, product/ significance level is, however, only
ß Blackwell Publishers Ltd 1999 service quality, sales, market share and between 5 and 10%. Labour turnover is

380
twice as high in organizations using the universalistic conclusions. What we cannot
control model. Moreover, the expected be sure about, however, is whether the
negative relationship between labour combination of a commitment HR system
turnover and the performance indicators, and a differentiated business strategy
labour productivity and scrap rate, is performs better than the use of such a
stronger in commitment HR systems than system under a low-cost business strategy
in control systems. (since there is only one case). If it does, and
The study is limited by the sample size the former combination also implies the use
and the results may be specific to the steel of TQM, then such a result would support
industry, or manufacturing situations akin to Hypothesis 3, according to which high December 1999
it. Moreover, no attempt was made to performance reflects the use of a total lean
establish the nature of the relationship strategy and not just the HR system.
between the commitment strategy and
turnover when other factors are controlled Ichniowski, Shaw and Prennushi study of
for, and the control variables are limited steel finishing lines in the USA. This study
where multivariate analysis is used. Arthur used a variety of observational methods to
notes that, because of the small sample size, collect data on a range of HR practices,
he was not able to test for the performance technical aspects and performance measures
effects of the fit between business strategy for 36 steel production lines across 17 US
and HR strategy. None the less, his companies. The articulated starting point is
equations linking productivity and scrap ‘‘recent incentive contract theories’’, which
rates to HR systems include a measure of Ichniowski et al. (1997) view as arguing
business strategy, which is based on that ‘‘complementarities often exist among
classifying the mills by their competitive a firm’s employment practices’’ and that
strategy, defined in terms of Porter’s the adoption of these clusters of practices
dichotomy of cost leadership versus ought to lead to superior organizational
differentiation. This variable is not performance. The implication of this is that
significantly related to the performance it is the internal fit between practices that is
measures, thus the commitment model has significant and that marginal changes in any
positive pay-offs regardless of the business one employment practice will have little
context in which it is implanted. We are, benefit. Ichniowski et al. then couch their
however, unable to say whether this is the argument in terms of a specific
case for labour turnover. In addition, the ‘‘complement of practices’’, which is made
results of Arthur’s research (1992, 504) up of high-involvement/performance
linking the two HR systems to business practices, these being incentive
strategy show that, while only one plant compensation plans, extensive selection
with a cost-leadership business strategy used procedures, work teams, skills training and
the commitment model, 40% of those with a labour–management communication. They
strategy of differentiation did not have the do not acknowledge the high-involvement
commitment model. This suggests that, were language or literature, referring simply to
a test of contingency theory conducted, at articles from the economics of the firm (e.g.
least in Porterian terms, it would not be Milgrom and Roberts 1995); but they do
supported, so verifying Arthur’s conceive these practices as being ß Blackwell Publishers Ltd 1999

381
Human resource innovative, as distinct from the ‘traditional’ results are found for quality, measured by
management and steel system, which operated on strict work the percentage of total production that met
performance rule and narrow job responsibilities. They the standards for designation as ‘‘prime’’
differentiated four types of HRM systems finished steel (prime-yield), as for
on the basis of the seven types of HR productivity. Ichniowski et al. thus show
practices, though for some of these more that HIM can produce simultaneously high
than one practice was included in its productivity and high quality, as those who
measurement, e.g. teamworking was associate it with lean production prophesy.
measured by whether there were formal They did not attempt to establish whether
work teams, whether there were problem- the workplaces that are using high-
solving groups, and, if they existed, involvement practices are also using the
whether the majority of operators were production procedures associated with lean
involved in one or other of these. production or TQM, such as just-in-time
The first cluster is the traditional system, stock control. Instead, they argue that
in which none of the practices was present. greenfield sites are more likely to adopt
The second is where some information the innovative HR and that there are non-
sharing and communication is done, and pecuniary costs that inhibit traditionalists
teams have been introduced, but the extent from adopting it, though these are not spelt
of participation throughout the line is low; out in any great detail. This research
the third, is similar to the second, but the clearly supports the universalistic
level of participation in the teams is high, perspective on HIM but, especially given
and there is also extensive skills training; its positive effects on both productivity
while the fourth system is characterized by and quality, we cannot rule out the
the usage of all practices. The classes possibility that the results are consistent
represent, in the terms of this paper, a with the hypothesis that links performance
progression from low- to high-involvement to both HR and quality/production
management. Ichniowski et al. acquired concepts (Hypothesis 3).
data on productivity on a monthly basis,
and thus they were able to conduct Kalleberg, Knoke, Marsden and Spaeth’s
econometric analysis on a panel data set of national study of organizations in the USA.
2190 monthly observations of productivity. Within a national study of around 650
The results show a hierarchical pattern establishments in the USA, Kalleberg et
in the productivity differentials of HRM al. (1996) collected data on what they treat
systems, with the highest productivity (p. 115) as the ‘‘four basic dimensions’’ of
corresponding to the fourth, high- a ‘‘model of a ‘high performance work
involvement system, and further tests organization’’’. The main elements they
controlling for a range of technical and included in their measure of high-
managerial factors confirmed that HIM is performance systems are the existence of
associated with high levels of productivity. decentralized decision-making, investment
Moreover, the contribution of individual in training, performance-based
practices in isolation of the others is small, compensation and internal labour markets.
relative to the effect of using them in Kalleberg et al. found that, on these four
ß Blackwell Publishers Ltd 1999 conjunction with each other. Similar items, workplaces clustered into three

382
groups: the first they labelled high grouping. They conclude that the practices
performing on the basis of their having all associated with high-performance
four elements of the model; the second, management do in fact enhance
trainers that invested in training but had organizational performance, thus in line
none of the other three; and the third, they with the universalistic hypothesis. Leaving
call (somewhat misleadingly) low- aside questions about the adequacy of the
performance work organizations, which performance measures, the main weakness
neither invested in training nor had of the study is that there are no controls
internal labour markets, but did have included in the analysis of the relationship
higher scores on both the compensation between high-performance practices and December 1999
and decentralization scales than did the performance. Consequently, the results
trainers, though not as high as the high- may have been consistent with hypotheses
performing class. other than universalism, had other
Analysis was confined simply to variables been included in the analysis.7
assessing whether the mean scores were
significantly different between the three Wood and de Menezes’s analysis of UK
groups on a range of performance government survey data. Wood and de
measures. Performance was measured by Menezes (1998) assessed the effects of a
asking the respondent to compare (on a range of practices associated with HIM,
four-point scale ranging from much better using the 1990 British Workplace
to worse) their organization’s performance Employee Relations Survey and its sister
over the past 3 years with that of other survey, the Employers’ Manpower and
organizations that do the same kind of Skills Practice Survey. They both covered
work. The performance indicators used a representative sample of workplaces
were quality of new products, services or across the whole economy, private and
programmes; development of new public sector. Together, they form a rich
products, services or programmes; ability data set that contains information on a
to attract essential employees; ability to range of high-involvement practices,
retain essential employees; satisfaction of including direct communication practices
customers or clients; relations between such as quality circles and teambriefing,
management and other employees; skill formation methods such as training
relations between employees in general; needs analysis, appraisal, no clocking-on
marketing; growth in sales; profitablility; or -off, monthly pay, cashless pay and
and market share. There may be problems information sharing. The study first
of accuracy with some of these measures, assessed whether these tended to be used
especially those that require knowledge of in conjunction with each other, that is
the inside workings of competitors, such as whether they form a unity as the theory of
the relations amongst employees. high-involvement systems assumes. This
Kalleberg et al., none the less, show that was done using an appropriate latent
the high-performing group scored highest variable model, which, since the practices
on all dimensions of performance except were measured dichotomously, meant
customer satisfaction, where they are in either a latent trait (where the underlying
fact second to the low-performance latent variable is continuous) or a latent ß Blackwell Publishers Ltd 1999

383
Human resource class (the latent variable is categorical) productivity, labour turnover and employee
management and model. A four-class latent class model fits relations climate do not significantly vary
performance the data, direct communication and between the four classes. The lack of
information disclosure is the main source variation on the employee relations climate
of discrimination at the bottom end of the reflects the fact that the bulk of the
scale and appraisal the main one at the variation to be explained is at the top end
top. There is an ordered progression from of the measures, between the scores 4 and
the first to the fourth; the first class can be 5 on a five-point scale. The results are
characterized as low-involvement thus mixed, showing that practicising HIM
management (using present terminology, in its total form does have some effects on
though in the paper it is referred to as some outcome variables but that these
‘‘low HCM’’) and the fourth as HIM. The effects are not unique to those
second and third (the two medium) groups organizations that do this.
are distinguishable first by the latter’s Caution is necessary about concluding
higher probability of having all the high- that HIM does not work on the strength of
involvement practices and, secondly, by its this analysis. First, the finding that the two
use of monthly and cashless pay, which extremes, high- and low-involvement
can be taken as indicators of salaried classes, do better than the two medium
status or white collarization, to use classes on the overall financial
Koike’s (1987) terminology. performance measure and employment
The identification of the four classes of growth may support the principle of
HIM suggests that there is some logic to internal fit, assuming that the low category
the use of high-involvement practices. The represents a coherent Taylorist or
researchers then went on to assess whether authoritarian management. Second, the
they are associated with different levels of data sets do not contain data on contingent
performance and, more particularly, factors, so if contingency theory were
whether the high-involvement group right, and all the high performers have
consistently outperforms the others. The successfully fitted their HRM to their
results show that this group does not context, the results could reflect the fact
perform better than all others on any of that some of the high performers are
the eight performance criteria in the data operating in a stable environment. Finally,
set. Its level of employment growth and the coverage of high-involvement practices
overall financial performance is, however, was not complete, the most notable
superior to that of the two medium- omissions involving elements of job design
involvement classes but not greater than – teamworking, functional flexibility and
the low-involvement group. The medium- conscious job design to enrich jobs.
involvement management groups do,
however, have significantly lower levels of Patterson et al.’s study of single-site
absenteeism than did the HIM class. The companies in UK manufacturing. A study
medium-to-high (third) group has, on by Patterson et al. (1997) has been widely
average, a significantly higher level of reported in the UK as demonstrating the
labour productivity than all other groups, value of HRM (see, for example, Patterson
ß Blackwell Publishers Ltd 1999 while the rate of change of labour et al. 1997, Preface by M. Emmott, vii). In

384
order to get round the problem that better job design factors, while 18% of the
financial performance data are publicly productivity change is predicted by these.
available at the company level but Patterson et al. go one stage further
collecting data on HR practices is likely to than other writers and attempt to assess
be more reliable at the establishment level, what they call other managerial practices
the researchers concentrated on to see whether they affect performance.
manufacturing firms with single sites. They concentrate on four such practices,
Sixty-seven manufacturing companies were which they term strategy, emphasis on
included in the study of HR practices and quality, use of advanced manufacturing
performance. Two main performance technology and research and development December 1999
indicators were used: the rate of change in (R&D) investment. The first of these,
profitability and in productivity. The HR which fits uneasily under the umbrella of a
practices of the firm were measured by managerial practice, was measured in
three groups of variables: (a) the Porterian terms by whether the firm adopts
comprehensiveness of the selection, a cost leadership or a differentiation
induction, training and appraisal systems; business strategy. We are not, however,
(b) the extent of skill flexibility, job told what the expected relationship is
responsibility, job variety and formal between this variable and performance. As
teamworking; and (c) the use of quality it happens, neither it, nor the quality or
improvement teams, direct communication technology variables, is significantly
methods, harmonized terms and conditions, related to either performance measure.
incentive compensation systems, and high R&D investment, in contrast, explains 6%
relative pay. Factor analysis revealed that of the variation in productivity change and
the first and second sets of variables form 8% in that of profitability change.
two independent factors (this being similar Patterson et al. thus highlight that ‘‘HRM
to Huselid’s (1995) study, which is practices are far more powerful predictors
reviewed later). The researchers initially of change in company performance’’ than
treated the first group as measuring the other areas of management (Patterson et
firm’s concern for what they call al. 1997, 19). Given the centrality of skill
‘‘acquisition and development of employee acquisition and job design to HIM, the
skills’’ (hereafter skill), the second as a research lends support to the universalistic
concern for ‘‘job design’’ (Patterson et al. theory – though the sample is limited to
1997, 15). The practices in the third group firms in one area of the economy and are
are treated as independent (though we are of one type, publicly quoted single-site
not told the extent to which any of them co- companies. Patterson et al. do not,
vary with each other or with the skill or job however, attempt to test the full HIM
design factors). Both factors are theory as they could have done through
significantly related to each measure of seeing whether there is an interaction
performance, while none of the independent effect between the two dimensions. There
practices is. Nineteen per cent of the is also the problem that other dimensions
variability in profitability change is of HIM – harmonization, quality-
accounted for by HR practices, the bulk of improvement teams and direct
which is the combination of the skill and communication – are not linked to the skill ß Blackwell Publishers Ltd 1999

385
Human resource or job design practices nor do they affect officer. They acquired information on the
management and either performance indicator. Yet, insofar HR practices and business strategy from
performance as incentive compensation systems are the bank’s senior HR manager and further
neither connected to the other practices nor ‘‘general information’’ (Delery and Doty
affecting performance, this offers support 1996, 814) about the bank and its strategy
for high-involvement rather than high- from the bank’s president.
performance concepts. Moreover, the lack Delery and Doty tested the universal and
of a strong performance effect from contingency thesis firstly by examining the
quality practices offers little support for effects of individual practices on
the theory underlying the second performance. The finding is that results-
conception of high performance, though oriented appraisal, profit sharing and
the researchers do not assess whether there employment security have positive effects
is any interaction between quality and their on the return on average assets, regardless
HR factors. of the strategy, while two of these, results-
oriented appraisal and profit sharing, have
an effect on the return on equity; the effect
Studies that Test Hypotheses 1 and 2,
of the results-oriented appraisal on both
Universalism versus Contingency
measures of performance is subsequently
Delery and Doty’s study of US banks. found to vary, depending on the strategy,
Delery and Doty (1996) differentiate two though it still has a pay-off in all
ideal types of employment systems, an circumstances. In addition, the effects of
internal system and a market-type system. internal career opportunities and
The first type approximates to the high- participation/voice are contingent on the
involvement model, as it includes internal strategy of the organization, though
recruitment, extensive formal training, participation only affected the return on
feedback for development purposes, average assets.
performance measured by behaviour- In order to test the contingency theory,
oriented measures, few incentive systems, Delery and Doty relied on Miles and
employment security, and participation in Snows’ (1984) trichotomy between
decisions (although jobs are very tightly prospectors, analysers and defenders to
defined in their characterization). The categorize ideal type strategies and
second, the market type, is contrasted with measured these by 13 items, including the
the internal type above all else by its rate of product innovation, technological
reliance on the external labour market for progress, active marketing and a proactive
recruitment, but it has the key non- management style. The market-type
Taylorist practice of loose job definitions. employment system was deemed
It also includes profit sharing and appropriate to the prospector strategy and
appraisals geared to assessing results in the internal type to the defender. In this
quantifiable terms, which, as we shall see, way, the internal system is seen more as a
are included in Huselid’s high-performance conservative rather than innovative style,
model. In a study of 1050 US banks, which contrasts with most authors, and
Delery and Doty concentrated on the particularly those who associate HIM with
ß Blackwell Publishers Ltd 1999 management of one occupation, the loan lean production and continuous

386
improvement. The employment system Thirdly, they assess the extent to which
deemed appropriate for the analysers is organizations have the appropriate degree
labelled the middle-of-the road approach. of each HR practice for their particular
The ideal types are scored first by location on the prospector–defender
specifying the middle-of-the-road strategy continuum (the contingency hybrid types
as the mean value of the seven HR fit). The second and third tests involve
practices, then the score for the market- strategic fit. None of the three tests
system was set at plus one standard supports contingency theory for the return
deviation from the mean of each variable on either average assets or equity.
and that for the internal system at minus Delery and Doty are cautious about December 1999
one standard deviation. inferring that the market ideal type is
Delery and Doty assess universalism by absolutely superior since, when they put
measuring the extent to which the firm’s the individual HR practices together in
employment practices are similar to each regressions on performance, they
of the ideal types and assessing the collectively explain more of the variance
association between this measure and in performance than did the market-type
performance. (It is not specified whether ideal system. They conclude that this
the ideal types are simply the two extremes means that the market-type system is best
of the market-based and internal strategies treated as the starting point for
or included the middle-of-the-road constructing ideal types rather than as a
strategy.) It is found that the more closely model of best practice. Yet, the superior
an organization’s employment system explanatory power of the equation
resembles the market-based type, the better containing the individual practices may
its return on average assets and equity. In simply reflect the fact that the ideal type is
contrast, the greater the proximity to the measured by a fixed point from the mean
internal ideal type, the lower the of the seven variables. Thus, if the market-
performance on these two measures. based system is treated as a variable and
Contingency theory is assessed in three not as an absolute concept, the research
ways. First, Delery and Doty analyse seems to suggest it is superior. The
whether organizations that conform to one problem, though, is how the concept
of the three ideal type are more effective relates to the high-involvement or
relative to those (hybrids) banks that do not performance terminology, since it contains
conform to one of these (in their terms this the non-Taylorist job design but relies on
is the ideal-type fit hypothesis). In our an external labour market and is
terms, this assesses the importance of characterized by low degrees of training
internal fit, assuming that the ideal-types and participation. Furthermore, there is no
posited are the internally fitted sets of attempt to assess whether the practices co-
practices and exhaust the possible internally vary or the clusters form natural clusters.
fitted sets. Secondly, they examine whether
those organizations that have the Youndt, Snell, Dean and Lepak’s study
appropriate ideal type for their strategy of US metal-working plants. Youndt et al.
outperform others (in their language, the (1996) set out to test the universal versus
contingency ideal-types fit hypothesis). contingency theory. They start from the ß Blackwell Publishers Ltd 1999

387
Human resource premise, however, that they are not effect or, rather, is a function of the
management and necessarily mutually exclusive, since it is performance enhancement obtained when
performance possible that there may be a best-practice it is linked to a quality manufacturing
model but that its impact is ‘‘further strategy. In addition, they test whether
enhanced when practices are matched with those firms with a cost strategy that have
the competitive requirements inherent in a an administrative HR system do perform
firm’s strategic posture’’, an argument that better than those without it.
is consistent with Hypothesis 4 (Youndt et The research was conducted in 97
al. 1996, 853). They treat what has been manufacturing plants in the metal-working
termed here the high-involvement strategy industry, drawn from a sample of 512
as the contender for best practice, across the USA. Initially, a mail
preferring the label human capital- questionnaire was sent to the general
enhancing HR system and characterize it manager of each plant. This dealt with the
as consisting of two types of practices: HR system and plant performance. A
those aimed at enhancing the skill base of second questionnaire, concentrating on the
employees and those concerned to promote HR practices, was sent to managers in four
empowerment, problem-solving and functional areas, operational, quality,
teamwork. They focus on manufacturing production and human resources. Finally,
strategy as the contingent factor since this 18 months later, another questionnaire was
has a ‘‘more direct relationship to what sent to all those who had responded in the
managers and first-line employees do first round; this asked about the
daily’’ (Youndt et al. 1996, 861). This is, manufacturing strategy and current
none the less, connected to the performance. Cases were only included in
classifications of competitive or business the final sample if at least two managers
strategies that have derived from Porter, as had responded to the questionnaires. Mean
they identify theoretically three main types values across each respondent per plant
of manufacturing strategy – cost, quality were used in the analysis.
and flexibility. Youndt et al. associate a Performance was measured by a range
human-capital-enhancement strategy with of indicators, which factor-analysed into
both the quality and flexibility strategies. three factors: machine efficiency (measures
They ally the cost strategy to what they of this included equipment utilization and
call an administrative HR system, based on scrap minimization), customer alignment
notions of command and control (Youndt (e.g. product quality, on-time delivery),
et al. 1996, 843), which involves the use and what Youndt et al. call employee
of selection oriented towards immediate productivity (which included employee
needs, training procedures that are geared morale as well as employee productivity).
to physical skills and general information, The measure of performance used in the
results-based appraisal, and compensation analysis was based on an average of the
systems based on individual incentives and performance scores from the two
treating manual workers as hourly paid. questionnaires. Thirty-one items were used
The issue Youndt et al. test is whether any to measure manufacturing strategy, and the
impact that the human-capital-enhancement factor analysis revealed four stable factors:
ß Blackwell Publishers Ltd 1999 system has on performance is a main the quality and cost items form one factor,

388
but the flexibility items loaded on two, the The first result supports the contingency
first, labelled delivery flexibility by argument of Youndt et al., but the second
Youndt et al., included items such as is counter to what they expect, as they
releasing new products and making hypothesize that flexibility strategies
deliveries on time, and the second, termed should be linked to a human-capital-
scope flexibility, contained items relating enhancing HR system. They conclude that
to the adjustment of product mix and the ‘‘the moderation results provide strong
handling of non-standard items. The HR evidence that manufacturing strategy
practices were grouped into two indexes. influences the HR–performance
The first measured the human-capital- relationship’’ (Youndt et al. 1996, 855). December 1999
enhancing HR system, and consisted of This is largely based on the fact that the
selective staffing, selection for technical interaction between quality strategy and
and problem-solving skills, developmental human-capital-enhancing system is related
and behaviour-based performance to all the performance measures. Thus, to
appraisal, external equity, group incentives, Youndt et al., the research supports a
skill-based pay, and salaried compensation. contingency theory of HIM, but the lack of
The second measured the administrative strong effects for the other interactions
system by the following items: selection means that it does not support wholesale a
for manual and physical skills, policies and contingency theory of HR systems. This
procedures training, results-based may, though, reflect conceptual and
performance appraisal, individual equity, measurement problems, particularly
individual incentives and hourly paid surrounding the notion of an administrative
status. HR system, and perhaps the two flexibility
The results support the contingency manufacturing strategies. Taken at face
argument. The human-capital-enhancing value, the results in the terms of this paper
HR system had a strong main effect on are supportive of the third hypothesis, that
employee productivity and a weaker, but combining high-involvement and quality
still significant, effect on customer management may be universally successful.
alignment and equipment utilization.
When, however, the interaction between it Becker and Huselid’s company-level
and the quality manufacturing strategy is surveys in the USA. Becker and Huselid,
included in the analysis, this is no longer working at the firm level, base their
the case for all performance measures, but analysis on a measure of high-performance
the interaction term is significant, thus the management developed by Huselid, which
main effects are predominately a reflection entails a set of practices including:
of the external fit between the human- information sharing; formal job analysis;
capital-enhancing HR system and the internal recruitment; attitude surveys;
quality strategy. Moreover, the interaction quality of work life measures or quality
between a cost strategy and the circles; incentive, profit-sharing and gain-
administrative HR system significantly sharing schemes; grievance procedures;
predicts equipment efficiency and delivery- employment tests; performance appraisal;
flexible strategy, and the administrative HR basing pay on performance appraisal; and
system is related to customer alignment. merit (as opposed to seniority) as a ß Blackwell Publishers Ltd 1999

389
Human resource promotion criterion. Two other measures the precise meaning of each of the two
management and are included in the set: (1) the average factors, is far from clear. The incentive,
performance number of hours of training received by a profit-sharing and gain-sharing scheme
typical employee over the last 12 months measure falls in the skills and organization
(this would appear to mean formal off-the- factor, while the pay linked to appraisal
job training, though this is not specified); falls in the other.
and (b) the number of qualified applicants Secondly, there is found to be a link
on average for the five positions that the between high-performance management
firm hires most frequently. The precise and three performance measures. Labour
nature and coherence of Huselid’s list of turnover is found to be negatively related
practices is questionable, since it includes a to the skills and organization factor but not
number of items that could legitimately be the motivation factor. The explanation
associated with either traditional industrial given for the lack of association in this
relations (grievance procedures) or case is weak, as we are told (Huselid
professional personnel management (formal 1995, 656) that ‘‘the use of incentive
job analysis and selection tests). He did not compensation systems may actually
simply ask whether the organization uses encourage employees who are performing
the practice, but asked for the proportion of poorly to leave a firm’’. The incentive
the workforce to whom the usage is system here presumably refers to the
extended, e.g. in the case of quality circles, linking of pay to appraisal, but it begs the
the percentage who participate in them. question of why the incentive pay measure
This procedure is required because of the in the first, skill and organization, factor
company level at which Huselid is does not have the same effect.
operating. As it results in a set of 13 Productivity, measured by sales per
continuous variables, he is able to use employee, in contrast, is related to the
factor analysis, the best-known latent employee motivation rather than the skill
variable model, to test whether they tend to and organizational factor. Corporate
co-vary. financial performance, measured by
In Huselid’s first study (1995), a Tobin’s q, a measure of value added that
questionnaire constructed from Compact is calculated by dividing the market value
Disclosure was mailed to a sample of 3452 of the firm by the replacement cost of its
US publicly owned companies with over assets, is significantly related to both
100 employees; 968 firms (28%) factors, while that measured by the gross
responded. The respondent was the senior rate of return on assets is only related to
HR professional. First, the factor analysis the skills and organization factor. Thirdly,
reveals that the 13 items fall into two it is shown that the effect of high-
factors – Huselid labels the first performance practices on financial
‘‘motivational’’ and the second ‘‘employee performance is mediated by their effect on
skills and organizational structures’’, since both labour turnover and productivity. In
they appear to him to be practices that other words, a significant proportion of the
affect the skills of people and the way impact on high-performance practices on
they are able to do their jobs. The extent financial performance is attributable to
ß Blackwell Publishers Ltd 1999 of the coherence of either list, and hence these more direct HR outcomes.

390
 Finally, tests for the importance of work (Huselid and Becker 1996, 413) on
strategic fit were conducted. It was the assumption that, because they have the
measured in two ways: (a) using Porter’s same effects on performance, they are
categories of the extent to which the firm identical. In one paper (Becker et al. 1997,
is using a differentiation or focus strategy 42), this combined index is treated as a
relative to a cost-minimization one; and measure of what he calls sophisticated HR
(b) behaviourally by asking the firm about architecture, even though it appears to be
the degree of emphasis it places on based on a set of practices and, as we have
‘‘aligning its human resource management seen, was viewed in this way when first
practices and competitive strategy’’. The introduced. This measure is consistently December 1999
results are reported for only the financial shown to be related to the key
performance measures. They reveal only performance measures used in Becker and
one interaction effect: the product of the Huselid’s studies.
behavioural strategy measure and the Becker et al. (1997) observe in some of
employee motivation factor is significantly their studies that the relationship between
related to the return on assets measure. A high-performance management and
test of the interaction effect between the financial performance of the firm –
two high-performance factors on the measured by market value per employee –
financial measures is negative. is non-linear. They argue that this is
In drawing his conclusions from the indicative of the ‘‘linkage between best
results, Huselid (1995, 667) skates over the practices and strategic alignments’’, that is,
unevenness of the results – ‘‘my results that the high-performance practices are
were consistent across diverse measures of ‘‘directly aligned with business priorities
financial performance’’ – and argues that and operating initiatives [that are] most
high-performance practices appear to have likely to create economic value’’ (Becker
a universal effect. The one result et al. 1997, 42). This implies that HIM is
supporting some contingent effect is left the main determinant but that, at higher
aside. Huselid also neglects the levels of it, an interaction effect comes
significance of the differences between the into play, although the effect of strategic
results for the two factors, in so doing integration might be treated as independent
almost denying the significance of the of this. This is never directly tested. This
factor analysis. He, in fact, does not non-linear relationship could simply be a
differentiate between them when making reflection of the synergistic effect of the
his overall conclusion that ‘‘a one practices, and indeed an exponential
standard-deviation increase in such (high- relationship between a measure based on a
performance) practices is associated with a coherent bundle of practices is precisely
relative 7.05 per cent decrease in turnover what we would expect.
and, on a per employee basis, $27,044 In their latest work, Becker and Huselid
more in sales and £18, 641 and $3,814 (1998b, 2) do, however, tackle head-on the
more in market value and profits, issue of the strategic fit of the high-
respectively’’. performance system and test their claim
Significantly, Huselid ends up that it is the ‘‘complementarities between
combining the two factors in his later the HRM system and other organizational ß Blackwell Publishers Ltd 1999

391
Human resource policies’’ that enable the high-performance performance principle, e.g. ‘‘to what extent
management and system to have a strategic impact, that is, does your firm effectively hire qualified
performance regardless of the level of high-performance employees?’’ or ‘‘to what extent does your
management. The sampling frame firm’s performance management and
consisted of all US public companies with appraisal system effectively reward
more than 100 employees and sales of $5 employee behaviours that are consistent
million or more, the total being 3840. with the firm’s competitive strategy?’’
Seven hundred and two firms responded to Adaptability (strategic fit) was measured
a mail survey, a response rate of 18%. We by two main questions: ‘‘to what extent
are not told who within the firm completed does your firm make an explicit effort to
the questionnaire, but the implication is align business and HR strategies’’ and ‘‘to
that it was a single person. Missing data what extent is the HR department involved
reduced the sample to 533. The HRM in the firm’s strategic planning process’’.
system was measured by a similar set of The measures of strategic insight include
practices to the earlier studies, though the ‘‘to what extent is HR (i.e. the people side
number was increased to 21. of the business) seen primarily by senior
Organizational fit was measured not by the management as a source of value creation
interaction of the HRM system with versus cost to be minimized throughout the
measures of other organizational policies, organization?’’ and ‘‘to what extent does
as in Lawler et al.’s studies discussed your firm have a clear strategic mission
below, but rather by a direct measure of that is well communicated and understood
what Becker and Huselid (1998b, 6) call at every level throughout the firm?’’ The
implementation alignment. It aims to items load on one factor ( = 0.90) and
measure the extent to which the thus implementation alignment is the mean
organization is oriented towards treating of the standardized values for 16 questions.
the HR as a strategic asset and supporting Though the proficiency measures
high-performance management concentrate on how effective the
(organizational fit) as well as strategic fit. organization is at high-performance
In practice, Becker and Huselid focus on: management, and the measures of the HR
(a) the functional proficiency of the HR system are concerned with the extent to
system; (b) the fit of the HR strategy to the which high-performance practices are used,
larger corporate strategy of the firm (what there may have been some overlap between
Becker and Huselid call adaptability); and the two types of measures, at least in the
(c) ‘‘the role of the senior management in minds of the respondents. No assessment of
recognizing the intrinsic value of whether they are independent is reported.
organizational resources’’ (Becker and The scores on the overall implementation
Huselid 1998b, 6) – in their terms, alignment index are correlated with the HR
‘‘strategic insight’’. They developed a system measure (0.54).
measure of implementation alignment Becker and Huselid (1998b, 8) used
based on multiple measures of each of cluster analysis to identify firms according
these. Functional proficiency was measured to both their HR system and their
by questions about how effectively the implementation alignment. Four clusters
ß Blackwell Publishers Ltd 1999 organization pursues a particular high- are identified: the two extremes of low on

392
both dimensions (termed personnel cluster, words, ‘‘the only piece in the puzzle’’.
i.e. non-strategic HRM) and high on both There is additional leverage from high-
dimensions (the high-performance cluster); performance management if it is
a low on HR systems and medium on organizationally aligned. Moreover, those
alignment (termed alignment); and an organizations that are organizationally
above average on HR systems and well aligned but do not practice high-
below average on alignment (called the performance management will get some
compensation cluster, because the score on performance benefits from this. Those that
HR systems reflects the high usage of are highly aligned but are not practising
high-performance compensation elements). high-performance management will, December 1999
In a regression on the market value of the however, perform no better relative to
firm, when the two continuous measures of those that strongly practice it and are
HR system and implementation alignment weakly aligned, as one would expect from
are included, both are significant, though the analysis of the interaction effects. In
the impact of the former is greater. The addition, part of the poorer performance
results remain the same when the associated with membership of the
interaction between the two is included, compensation cluster is derived from its
this being not significant. This confirms lack of complete alignment. Becker and
that high-performance management Huselid (1998b, 14) tentatively conclude
systems have an effect independently of that the compensation cluster represents
whether they are internally fitted to the ‘‘an evolution in the firm’s capability to
organization. It is, none the less, the case compete based on their human capital’’,
in this study that, as organizations develop perhaps because strengthening the pay–
these systems, they tend to become more performance link is the easiest initial
aligned (which may be an artefact of the change an organization can make.8
measurement system in this case, and may Overall Huselid and Becker’s research,
bias the equations that include the therefore, supports a universalistic thesis of
interaction term). An analysis comparing high-performance management, rather than
the effects of cluster membership allows a contingency theory that, in this research,
for this, and reveals that the high- would have meant that implementation
performance cluster significantly alignment was the dominant predictor of
outperforms all others. The compensation performance. None the less, it shows that
and alignment clusters also perform there are additional benefits if the high-
significantly better than the personnel performance system is organizationally and
cluster, as one would expect from the strategically fitted; organizational fit
regression results using the continuous complements the HRM system (Becker
measures. The compensation measure has and Huselid 1998b, 15), rather than
stronger effects on performance than does moderates its effects on performance. As
the alignment cluster. The ranking of the such, there is no support for Hypothesis 4.
effect that each cluster has on performance Regardless of whether the relationship
confirms that high-performance between high-performance management
management is crucial, though it is not, to and performance is linear or non-linear (or
use Becker and Huselid’s (1998b, 13) whether its power is affected by ß Blackwell Publishers Ltd 1999

393
Human resource organizational fit), it is difficult to square namely labour productivity, quality and
management and such a strong relationship with the authors’ financial performance, which were all
performance overall emphasis on the importance of judged relative to the perceived average in
both inimitability and contingent the hotel industry. Information was also
adaptation. If there is such a need to gained on the business strategy of the
develop unique configurations of practices hotels so that they could be classified in
(or architectures), we would not expect to terms of whether their competitive strategy
find a general association between a set of was, in Porter’s terms, based on cost
practices and performance (though in reduction/price competition or quality
Becker and Huselid’s terms we would enhancement. A third ‘‘other’’ category of
expect this to be the case for high- those hotels with an ambiguous strategy
performance principles). Indeed, this could was also included.
only be the case if the set were a best- Hoque tests the contingency argument
practice model and all organizations that by dividing the sample into three groups
follow it design their practices so they fit on the basis of the hotel’s business
both with high-performance principles and strategy and then examining the
with their strategic aspirations. There does, relationship between HIM (in his terms
however, appear to be a discrepancy HRM) and performance for each subset,
between Becker and Huselid’s research rather than by including the cross-product
results and their underlying theoretical of HIM and business strategy or dummies
arguments. for each type of strategy in a single
equation. Controlling for union presence,
Hoque’s study of UK hotels. Hoque size, UK ownership, price of a standard
(1999) researched UK hotels along very room, and the age of the hotel, he finds
similar lines to his earlier work with that, amongst the quality enhancers,
Guest. Through a questionnaire completed commitment, job satisfaction, quality of
by a single managerial respondent from a work, quality of service and financial
sample of 209 hotels with 25 or more performance, as perceived by the
employees, he collected data on the use of respondent, are all strongly related to the
22 high-involvement (called HRM by use of HRM practices (measured as a
Hoque) practices and a range of outcome simple aggregation of the 22 practices).
variables. The majority of these practices Both job mobility and the flexibility of
corresponded to those in his earlier staff are related to HIM, though the
greenfield-site study with Guest (Guest and relationship was not very strong in either
Hoque 1994), which is discussed below case. The quality of staff and labour
and itself drew on those used by Wood productivity are not related to HIM. In
and Albanese (1995). The outcome contrast, amongst the cost minimizers,
measures were of two types: ‘‘human HIM affects only commitment and job
resource outcomes’’, which all pertained to satisfaction of the HR outcomes and has
lower grade staff – the commitment, job no effect on the three performance
satisfaction, flexibility of staff, job variables. Within the ambiguous (termed
mobility, quality of work and quality of other) group, HIM is related to the same
ß Blackwell Publishers Ltd 1999 staff – and ‘‘performance outcomes’’, outcomes as in the subsample of quality

394
enhancers. It is, however, also related to all circumstances. Moreover, the effect of
all other variables, the quality of staff and HIM on commitment and job satisfaction
labour productivity. The strength of the is universal; there is an effect in all three
relationship between HIM and staff types of hotels, though the strength of the
flexibility is stronger for the ‘other’ relationship is slightly weaker in the cost
category than it is for the quality reducers than in the other two categories.
enhancers, though that between HIM and In Hoque’s terms, it is the effects of the
quality of service is considerably weaker commitment and satisfaction generated by
within the former than the latter. Hoque is HIM that are seemingly not universal,
inclined to treat the ‘‘other’’ category as since in cost-minimizing contexts it does December 1999
similar to the quality enhancers, arguing not seem to be having much effect on
that from their effects on outcomes their organizational performance or, to be more
emphasis is on quality and not cost accurate, the commitment and job
minimization. satisfaction it generates does not seem to
Hoque (1999, 437) concludes that his translate into organizational performance.
results support the contingency theory. There is no attempt to assess the link
Having done this, he then addresses the between these two HR outcomes and the
question of whether it is indeed the case performance measures, but perhaps the
that organizations that both have a quality reason that commitment and satisfaction
enhancement strategy and practise HIM have no effect is that total quality systems
outperform all others in the sample. He are not in place and/or individual
concludes that they do, on the basis that, discretion is low. While the implication of
across the range of performance measures, the results for the HIM hotels is that the
they perform better than any other single quality enhancers are getting better
type (based on classifying the firms by financial performance, not through superior
their business strategy and degree (low, productivity but through their better
medium, or high) of HIM). For example, quality (price having been controlled for),
they outperform others on five of the other the ‘‘others’’ are getting it through a
eight categories on commitment and job combination of labour productivity and
satisfaction, and they perform significantly quality.
better than all categories of hotels on at It is safe to conclude that a strategy
least one of the three measures of based on cost reduction or price
performance, but it should be noted that competition leads to suboptimal
the high-HIM ‘‘other’’ category performs performance within the industry, as Hoque
as well as the high-HIM hotels on all concludes (1999, 439), but it may be that
indicators except for the quality of work the other category is superior, as might be
(which, if they truly are aspiring ‘‘lean the case if Hypothesis 3 is correct. It is not
producers’’, may reflect the greater entirely clear what is included in the
standards they have). This conclusion, category – it could include those with an
rather than confirming the contingency ambiguous strategy or those that are
theory, supports Hypothesis 3, which is genuinely oriented towards the
that the high (quality and involvement) simultaneous pursuit of quality and cost
route should result in high performance in minimization, or even those that have no ß Blackwell Publishers Ltd 1999

395
Human resource real strategy (ad hocism). Hoque’s starting Studies that Test Hypothesis 3, the
management and point seems to be Porter’s association of Combined Effect of High-involvement
performance hybrid strategies with prevarication and, Management and Total Quality
hence, ultimate failure and, on finding that Management
the ‘‘others’’ are doing as well as the
quality enhancers, he treats them as Lawler et al.’s surveys of high-
equivalent by putting primacy on quality. performance practices. Lawler et al.
This may be misleading for, following the (1995), like Huselid, conducted a
lean production notion, a strategy that company-level survey. The sample frame
combines quality and cost minimization was the Fortune 1000 largest
could be interpreted as the most manufacturing and service companies, and
sophisticated. Given that the strength of the 279 usable responses were received (a
relationship between HIM and productivity response rate of 28% as 985 companies
in the other category is greater than that received questionnaires). Given their
between HIM and quality, they may be conception of high-performance
giving more or at least as much weight to management as the combination of high-
costs as to quality. The results, then, may involvement and TQM, they acquired
well be consistent with the argument that information on both employee involvement
lean production is universally successful, (EI) and total quality practices. Total
that is, an argument that places emphasis quality practices were grouped into two
on the importance of methods aimed at the main categories: core and production
simultaneous pursuit of quality and cost. oriented. Core practices were quality-
The superior financial performance of the improvement teams, quality councils,
‘‘others’’ and the quality enhancers does cross-functional planning, process re-
not appear to rest on their being able to engineering, work simplification, customer
charge a price premium, since price is satisfaction monitoring, direct employee
controlled for in the analysis. We would exposure to customers; while production-
need more information on the extent to oriented practices consisted of self-
which the HIM practices are being used in inspection, statistical control method used
association with other practices. If they are by front-line employees, just-in-time
not, the results could still be interpreted as deliveries, and work or manufacturing
supporting an argument that combines cells. Information was collected for four
contingency and lean production notions: types of EI practices, as in Huselid’s
HIM is most successful in those firms that studies, by asking for the percentage of
treat leanness as their strategy, while it can employees covered by a practice. The first
successfully achieve the objectives of those type, information sharing, asked for the
managers that emphasize quality percentage of employees who received
enhancement, and reduce the ‘alienation’ information on each of such items as the
of those working under cost-minimization corporate operating results, business plans/
regimes without this translating to higher goals, and competitors’ performance; the
performance, as one would expect if there second type, training and skills, asked
were little attempt to design jobs with about the extent of training in areas such as
ß Blackwell Publishers Ltd 1999 discretion. group decision-making/problem-solving

396
skills, leadership skills, teambuilding skills of the practices with between only 1 and
and cross-training; the third referred to 20% of its employees. In the absence of
reward practices associated with EI and multivariate analysis, it is not in fact
included the whole range of variable possible to conclude that TQM and EI
systems such as knowledge/skilled based practices tend to coexist.9
pay, profit-sharing, individual incentives, Lawler et al.’s analysis of the effects of
work-group/team incentives, as well as all- TQM and EI on performance was also
salaried pay systems, and employment based on correlational analysis. Lawler et
security and open pay information; finally, al. (1995, 87–92) analyse the effects of EI
the fourth, labelled power sharing, included and TQM on measures of economic and December 1999
suggestion system, survey feedback, job financial performance, obtained from
redesign, quality circles, employee Compact Disclosure, for companies in
participation groups, union-management their sample. The measures were total
QWL committees, minibusiness units, self- factor productivity, sales per employee,
managing work teams and employees return on sales, return on assets, return on
policy/strategy committees. investment and total return to investors.
Lawler et al. used simple pairwise This time, they conducted multiple
correlations to examine the relationship regression analysis of the effects of EI and
between the individual TQM measures and TQM (again it is not clear how these are
the four indices of EI, as well as an measured) controlling for industrial sector
overall index based on the average scores and capital intensity. We are told that
across the indices. The correlations on a these EI and TQM variables are most
high proportion of all pairs involving the strongly related to return on equity and
information-sharing skills, and power- return on assets, while all of the other
sharing indices are all significantly above outcome measures are significantly, but
zero, though the rewards index is only weakly, related to their usage, with the
related to one of the TQM practices, self- exception of the total return to investors.
inspection, and it is only weakly (0.20) The percentage of corporate performance
correlated with it. The correlations on the variance that is accounted for by EI and
three other indices range from a maximum TQM practices is, though, relatively small;
of 0.47 to 0.08. Lawler et al. (1995, 58) none the less, because of the wide range of
conclude that ‘‘most companies have both performance, a small movement in these
EI and TQM initiatives’’ and that they practices could translate into a relatively
‘‘are most frequently co-ordinated or large effect on performance. A one
managed as one integrated program’’. The standard deviation increase in EI and TQM
size of the correlations between the use of practices would, Lawler et al. (1995, 92)
particular TQM practices and the EI estimate, mean that an additional 30% of
indices is, however, not consistently high employees within a company would be
enough to suggest that the dominant covered by them, and this would have
pattern is a fully integrated TQM and EI. quite significant effects on five of the six
Nor is the frequency of use of many of performance indicators.
them: while most firms use at least some In addition, Lawler et al. examine
of the practices, the typical firm uses most whether the extent of the use of the various ß Blackwell Publishers Ltd 1999

397
Human resource practices is related to management’s In 1996, Lawler et al. replicated their
management and perceptions of the impact of EI and TQM 1993 survey, using the same conceptions
performance on performance. The majority of the EI and measures of EI and TQM, though they
practices are found to be related to three also added what they see as a third type of
measures of performance – an index of practice concerned with process re-
direct performance outcomes (made up of engineering. The sample frame was again
productivity, customer satisfaction, quality the Fortune 1000, but this time the
and speed of response); profitability and response rate was lower (22%) yielding
competitiveness; employee satisfaction and 212 usable data points. The results, which
quality of working life – and a measure of are reported in Lawler et al. (1998), are
management’s satisfaction with EI overall. almost the same as those for the 1993
The correlations again are not consistently study. First the pattern of simple pairwise
strong, ranging from 0.13 to 0.31. None of correlations between the TQM measures
the power-sharing or reward practices, bar and the four indices of EI and the overall
employment security, is related to summary index are almost identical. The
employee satisfaction, though the overall main difference is that the rewards index,
power-sharing index is. Similarly the which was only correlated with self-
majority of TQM practices are related to inspection in 1993, is highly related to
the perceived impact of TQM on direct work simplification in 1996, and weakly
performance outcomes, profitability and related to cross-functional planning. The
competitiveness, and satisfaction with power-sharing index in 1993 is strongly
TQM. The majority of the practices are, related to the use of quality councils but,
however, not significantly related to by 1996, it is only weakly related. Again,
employee satisfaction, the exceptions all the size of the correlations is not
being core, rather than production-oriented, consistently high enough to conclude that
practices. Further analysis shows that some TQM and EI coexist, though a majority (26
of the effect of EI is a reflection of the out of 49) of the significant correlations (in
TQM processes that are being used in 1993 or 1996) had increased with three
conjunction with them. The skills, power- remaining the same. The use of EI and to a
sharing and reward dimensions of EI have lesser extent TQM practices has increased
no significant impact on the performance slightly, but the typical company still uses
measures (with the exception of skills, the practices to a very limited extent.
which impacts on profitability and For the 1996 data, Lawler et al. (1998,
competitiveness) in non-TQM companies 142–153) do not report the results of a
(though it is unclear how this is measured). regression analysis on the effects of the
Similarly, in the case of the production- combined use of EI and TQM on financial
oriented practices, these have no impact performance as they did in 1995. It is
without EI (again it is not clear how this is shown to be the case, however, that the
measured). The impact of the core high users of both EI and TQM do in fact
practices is considerably enhanced when perform better on return on sales, return on
accompanied by EI. These results, Lawler assets, return on investment and return on
et al. (1995, 85) conclude, show the equity. A regression analysis of EI usage
ß Blackwell Publishers Ltd 1999 intertwined nature of TQM and EI. on its own shows that it is related to sales

398
per employee and return on assets, as it redesign and such cost reduction methods
was in 1993. Additionally, it is related to as doing the same work with less
return on investment. However, it is not, supervision, were analysed in a similar
as was the case in 1993, related to return way to the EI and TQM data. Their usage
on sales and return on equity. Total quality is shown to be correlated with EI and
management usage, when assesssed in TQM practices. The exceptions are the
isolation of EI, is related to return on lack of association between any of them
sales, return on assets and return on equity, and the reward practices and also the cost-
which was not the case for 1993. The reduction methods and information sharing
strength of the overall conclusion of the (Lawler et al. 1998, 72–73). The use of re- December 1999
studies, that financial performance is engineering practices also helps both EI
affected by the use of EI and TQM, is and TQM to be more effective when
enhanced by time-lagged analysis, which measured by the perceptual measures, but
shows that the use of practices in 1993 is does not contribute to their impact on
related to financial performance in 1996, perceived employee satisfaction (Lawler et
though we are not given information to al. 1998, 130–131), though no such tests
show that financial performance in 1993 are conducted that use the non-perceptual
was unrelated to the use of practices in financial measures. A greater use of re-
1996. engineering, though, is shown to impact
The results of the effects of TQM and positively on all the financial returns
EI on the perceptual measures of measures except the sales per employee
performance are again similar to those for (Lawler et al. 1998, 148).
1993. Employee involvement practices are So, overall, Lawler’s research, within the
related to all these measures of limits of its methodology, offers some
performance, although again few of the support for Hypothesis 3. Lawler et al.
power-sharing or reward practices are (1998, 138–140) none the less introduce
correlated with these. Total quality into the 1996 survey some questions on
management practices remain correlated the nature of the organization’s change
with the perceived TQM outcomes, but strategy and find that this does moderate
they have little impact on employee the impact of the practices on performance
satisfaction, though in 1993 most of the as measured perceptually. Change
core practices did. There is no attempt to strategies that emphasize integrated,
assess the impact of EI practices clearly articulated reasons for the direction
controlling for the impact of the effect of of change and are led by top management,
TQM or vice versa as there was in 1993. mean EI, TQM and re-engineering
None the less, Lawler et al. present practices will have more effect. This may
evidence that shows that the impact of be consistent with Hypothesis 4.
TQM programmes will be greater if they
include EI practices. MacDuffie’s multi-plant automotive
The additional data on the re- study. Like Arthur, MacDuffie (1995)
engineering practices, which include conducted a single-industry study, this
process simplification, creation of cross- time based on the 62 final assembly plants
functional teams, major information system in the major car-producing countries, using ß Blackwell Publishers Ltd 1999

399
Human resource data from the MIT Future of the Auto producing standardized cars with
management and Industry project, the birthplace of the lean standardized operating procedures (see
performance production concept. He differentiated two Wood 1989, 1993). The third, intermediate
types of what he calls innovative HR group uses buffers and innovative HR
practices, work system practices and HRM practices to an extent that is halfway
practices/policies. MacDuffie measured the between the two other systems, while their
former by the existence of work teams, usage of innovative work systems is at a
problem-solving groups, job rotation, similar low level to the traditional ‘‘mass’’
decentralization of quality-related tasks plant.
and an effective system for employee MacDuffie finds that both productivity
suggestions. The HRM policies included and quality are superior in the lean
recruitment and hiring criteria geared production plants, while the intermediate
towards openness to learning, interpersonal plants also perform better than the
and teamworking skills, contingent pay traditional ones on both these measures,
systems, minimizing status differentials, though their quality levels are far closer to
training for new employees, and training the traditional than they are to the lean
of experienced employees. In addition, plants. All the three elements of the lean
MacDuffie measured the extent to which production system, the non-use of buffers,
the production regime was lean, or the work system and HR, are related to
bufferless, by the percentage of total productivity and, moreover, there is a
assembly area space dedicated to final strong interaction effect between them.
assembly repair, the average number of The results for quality are less clear cut, as
vehicles in the work-in-process buffer the low use of buffers is not associated
between the paint and assembly areas (as a with high quality. Nevertheless, the plants
percentage of one shift’s production), and that have low buffers and the work and
the average level of inventory stocks for a HR practices associated with lean
sample of eight key parts (weighted by the production do, again, appear to fare better
cost of each part). Through cluster than others. There is no measure of
analysis, he identified three discrete types business strategy along the Porterian lines
of plants. At one extreme are lean plants of Arthur’s study. But insofar as the three
with few buffers, and the characteristics of groupings, or at least the two extreme
both innovative work systems and HR groups, may be seen as having work
systems; he labelled these flexible systems and HR systems that match their
production systems. At the other extreme production regimes (what MacDuffie
are traditional buffered plants, which make (1995, 198) calls their organizational
little use of innovative work practices and logics), it provides one test of the strategic
hire on the basis of a simple match to the fit hypothesis with the results clearly
job requirements and train very little. lending no support to it. Taking the
Under the influence of Piore and Sabel innovative HR and work system practices
(1984), MacDuffie labels these mass as together constituting HIM, the results
production, though there is no evidence could be taken as support for the
here or elsewhere to suggest that the lean universalistic theory. None the less, the
ß Blackwell Publishers Ltd 1999 producers are not mass producers, still strong interaction effect between these and

400
the bufferless system on the performance what we have been treating as strategic
measures confirms the broader hypothesis integration, namely the integration of HR
that links lean production and high- policy with business strategy and the
involvement (Hypothesis 3 above). There integration of HR policies with each other.
would appear to be insufficient plants in The measure of strategic integration used
the study that have high-involvement by Guest and Hoque is based on a single
practices with high use of buffers for the item constructed on the basis of asking the
sample to have provided us with a test of respondent whether there is ‘‘a human
whether HIM works within any regime. resource strategy formally endorsed and
actively supported by the top management December 1999
team at the establishment’’. Having
The Study that Tests Hypothesis 4, the
measured HIM by aggregating 23 items
Moderating Effect of Strategic
(21 practices and two strategy measures),
Integration on High-involvement
Guest and Hoque divide the organizations
Management
into their four groupings – good, lucky,
Guest and Hoque’s UK greenfield-site bad and ugly – on the basis of two criteria,
survey. Guest’s arguments imply that his a division between plants with 12 or more
‘good’ (strategic HIM) organizations will and those with 11 or fewer of these
perform better than the ‘lucky’ (non- practices, and whether there is a HR
strategic HIM) ones, and both should strategy.
perform better than the bad and ugly. If Their conclusions are that the results,
the principle of strategic integration taken overall, are supportive of the HRM
extends to the control organization, then argument: ‘‘the results demonstrate that
we would expect the ‘ugly’ to perform strategic HRM does pay off’’ (p. 11). None
better than the ‘bad’. Guest tests this the less, according to their analysis, the
argument in a direct way in his research good perform only better than all other
with Hoque (Guest and Hoque 1994), organizations on three of their measures –
which involved a sample of predominantly labour turnover, disputes and quality. In
greenfield sites. The research instrument some other cases – the commitment and
was a postal questionnaire completed by quality of staff – the good and the lucky
the manager responsible for personnel at fare better than the other two, the bad and
the workplace. Guest and Hoque (1994) ugly variants of the control model; in some
measure HIM (in their terms, high- others (mobility of labour and productivity),
commitment management) by aggregating the good and ugly are superior to the two
the use of what they call HRM practices. non-strategic variants. Finally, absenteeism
These include 21 practices commonly is on average lower in all three of good,
associated with HIM, which both replicates lucky and bad organizations than it is in the
the set developed by Wood and Albanese ugly. The results are clearly not consistent
(1995), e.g. quality circles, no compulsory across measures, with some suggesting that
redundancy, and extends it to include such HIM alone is significant, some that
items as job previews and selection tests. strategic integration is significant, and yet
Guest and Hoque also incorporate into others that the interaction between the two
their measure of HIM two indicators of is significant. ß Blackwell Publishers Ltd 1999

401
Human resource Moreover, Guest and Hoque’s inclusion contingency argument was the interaction
management and of two items relating to strategic of these measures. The results offer
performance integration in their measure of HIM raises stronger support for Guest and Hoque’s
questions about its meaning. It also means own conclusion than did their own
that their measure of strategic integration analysis, since HIM is significantly related
ends up being based on a single item. In an to 11 of the 14 key performance measures
act of collegiality and openness, for which included in the study. Strategic integration
I am very grateful, Guest and Hoque gave is, however, only related to five of these,
me access to their data set to conduct commitment, and the quality and
further analysis. First, I reclassified the productivity measures, and in all these
strategy items that they included in their cases, less strongly than was HIM. In
HRM practice measure as measures of contrast, the interaction term is not related
strategic integration. A latent trait analysis to any of them. These results suggest that
of these plus the third measure of whether Guest and Hoque’s lucky organizations
a HR strategy exists revealed that they do may perform as well as good organizations,
form a unity. I conducted a latent trait questioning whether having a strategic
analysis of the high-involvement practices, underpinning is in fact important.
limiting these to the core elements that
corresponded to those in my earlier work
Discussion
with Albanese (Wood and Albanese 1995),
which revealed that they do tend to coexist. The studies differ on a range of
There is, though, an especially strong dimensions. Firstly, they differ according
relationship between merit pay and to whether they focus solely on assessing
appraisal, which are both included in Guest whether the HR system upon which they
and Hoque’s set. In statistical terms, they are focusing outperforms all others in all
are more correlated than the latent variable circumstances (Hypothesis 1), or whether
model predicts, and this association is they examine other HR systems as well.
explained by a common factor over and Secondly, they vary in their underlying
beyond that which explains the relationship conceptions and in particular whether they
between the whole list of items. The are built around the notion of high-
implication of this is that only one of these involvement or high-performance
practices should be included in any management with the added complication
measure, and that, if there is a fair number of the internal versus external labour
of workplaces on the border line between market contrast, as used by Delery and
high and low in Guest and Hoque’s Doty. Thirdly, they differ in the way that
division of this that have both of these, the practices are measured – some being
their results may have been misleading. measured continuously, others
In a regression analysis of the outcome dichotomously. Fourthly, studies vary
variables, the measure of strategic according to whether or not they attempt
integration and the measure of HIM (which to assess the relationship between the
excluded merit pay as well as the two practices before they develop their overall
strategic integration items) were both measure of HR systems. Researchers who
ß Blackwell Publishers Ltd 1999 included. Also included in order to test the have not simply aggregated the number of

402
practices to develop their measure have company-level studies do not reveal a
used one of two methods to analyse the consistent set of results while
relationship between practices: some form establishment-level ones show another set.
of latent variable analysis (the precise Moreover, it is not possible to prioritize
form depending on the level of the results of some studies over others,
measurement of the practices) or cluster since no one method is demonstrably more
analysis. Fifthly, the unit of analysis valid than another, even though some
differs between studies, and, in particular, studies may be more sophisticated than
whether it is conducted at the company or others and have fewer weaknesses than
workplace/establishment level. Finally, the others. Secondly, what can be concluded December 1999
type of performance measures used in the from some of the studies is uncertain, for
studies varies, with some concentrating on within several studies the effects of HR
productivity or other labour-related systems vary between performance
variables, while others include financial indicators and/or the results might be
performance data. There is also a consistent with more than one hypothesis.
difference between the type of Thirdly, there is some uncertainty about
measurement of these indices, as some what the various indexes of HR systems
studies rely on published company data measure and, more significantly, the nature
and others on the assessment of relative of the underlying concepts. For example,
performance by a representative of the in the case of Delery and Doty (1996), the
organization. Table 1 summarizes the main index that is positively related to
studies in terms of these dimensions. performance is more a measure of a
The studies vary so markedly between market employment relation than the
each other that there is not even a pair of organizational-based relation implied by
studies that differ simply on one or two the concept of high-involvement, but in
dimensions. There is no systematic this case there is the added problem that
relationship between the dimensions of the non-Taylorist job flexibility is included in
studies. So, for example, it is not the case the market index rather than the internal
that only those studies that use the concept one. Some of Huselid and Becker’s
of high-performance systems include some measures, and particularly their emphasis
form of contingent pay in their bundle of on the pay–performance link, could be
HR practices. It is the case, though, that construed as indictors of a market-based
company-level studies tend to use model, a contractualization, not an
continuous measures of their HR practices, internalization, of the employment relation.
while establishment-level ones use Earlier reviews of the field (Ichniowski
dichotomous measures. It is thus difficult et al. 1996, 322; Guest 1997) have none
to group the studies. the less tended to conclude that high-
A consistent picture does not emerge involvement or performance management
from the studies. First, it is not the case does have positive performance effects.
that the marked differences between the This partly reflects the fact that some of
results of the studies reflects in any the less conclusive studies that are
systematic way the underlying concepts or included in this survey were published
designs of the studies. So, for example, the after the earlier reviews were written. But, ß Blackwell Publishers Ltd 1999

403
Human resource it also reflects a glossing over the are suitable. In the case of cluster analysis,
management and unevenness in the results. The focus of the however, one cannot be sure about what
performance reviewers’ attention has been on whether the cluster represents, since there is no test
the effects of high-involvement/ for an underlying latent variable that
performance systems are universal or explains the association between the
contingent, and finding little or no support practices (as there is in Wood and de
for contingency theory, they have felt Menezes 1998). Moreover, it is not an
justified in proclaiming that ‘‘innovative appropriate method for testing synergy.10
human resource management practices can Past reviews have also made light of any
improve business productivity’’ conceptual differences, yet the overview in
(Ichniowski et al. 1996, 322). This the first part of this paper highlighted that
neglects the conclusions of Delery and the debate was founded on a range of
Doty in favour of the contingency theory. potentially conflicting concepts and
It also overlooks two possibilities: first, extended beyond a straight competition
that had those studies that demonstrated between the universal or contingent nature
universal effects but did not measure of high-involvement/performance systems.
contingent factors been able to do this, it A key issue is the extent to which high-
may have turned out that the effects were performance systems accentuate pay
not as strong in certain circumstances as in systems, and whether, if this is the case, it
others; and secondly, that the lack of a is to the neglect of the fundamentals of
universal effect in some studies (e.g. Walton’s high-commitment management –
Wood and de Menezes 1998) may be job redesign, minimal status differentials
because of the importance of contingent and job security guarantees. That is, it is a
factors that were not included in the data truly distinctive concept, even its opposite,
set. Furthermore, the tests of the universal the market-relations as in Delery and
effects of high-involvement/performance Doty’s terms. Or is Becker and Huselid’s
management have been limited. Proof of argument more about the importance of
their synergistic effects would ideally high-involvement systems (as recognized
require (a) analysis showing that they do by Walton and others) that are oriented
tend to coexist and (b) that the relationship towards the performance requirements of
between the measure of high-involvement/ the organization, the key issue then
performance management and performance becoming more about whether they are a
is exponential. Huselid’s study is the one necessary but not sufficient condition for
that gets closest to this ideal, but there are organizational effectiveness, the additional
questions about his measures and the element being organizational fit. The
interpretation of his factor analysis, and review of the empirical studies shows that,
many of his performance tests assume as yet, there has been no serious attention
linear relations. Similarly, in my study to the sensitivity of the results to what is
with de Menezes, the high-involvement in the bundle of practices (though if they
measures are limited. The tests based on are highly inter-correlated this may be
identifying a group of high-involvement/ difficult to detect). Insufficient attention
performance organizations and then has also been given to Hypotheses 3 and
ß Blackwell Publishers Ltd 1999 assessing whether they outperform others 4, which partly reflects limitations on data

404
 Table 1. Overview of Studies of the HRM – Performance Link

Study: Betcherman et al. Arthur Ichniowski et al. Kalleberg et al. Wood and de Menezes Patterson et al. Delery and Doty
(1994) (1994) (1997) (1996) (1998) (1997) (1996)

Hypothesis(es) tested Universalism Universalism Universalism Universalism Universalism Universalism Universalism and
contingency
Underlying Participative/ Control/commitment Contract theory High-performance Control/commitment HRM? Market type vs. internal
conception traditional dichotomy work organizations dichotomy employment system
Unit of analysis Establishments Plant Workplace Establishment Establishment Company Company
Sample 714 Canadian 30 US steel plants 36 steel finishing sites 629 establishments in 1693 sample of 67 UK firms 1050 US banks
establishments with in 17 US steel firms the USA representation sites Manufacturing firms
more than 40 across UK with single plants
employees in 4
sectors
Measures of HR 11 practices plus HR 10 practices 13 innovative 4 main features of 7 key commitment 2 main factors: 7 HR practices
practices in business plan practices falling high-performance practices. Job design
under 7 categories work organizations: Latent class model Skill acquisition
corresponding to decentralization, job used to identify 4 Other individual
high-involvement training, PRP, internal classes. practices
practices labour markets
Pay system in HR Variable pay High overall wages Profit-sharing, Bonuses for Salaried payment High pay Performance-based
system Non-incentive incentives with performance, profit style Incentive pay as compensation
schemes quality bonus sharing or stock Pay not included individual item
options
Measure of HR 3 clusters: traditional, Clusters of practices 4-fold classification of 3-fold classification 4 progressive latent 2 scales based on 3 ideal types – market,
system compensation and systems from based on cluster classes factors middle of road,
participation traditional to analysis internal
innovative
Outcome/ Labour quits + Labour efficiency + Productivity + Performance on Financial
Profitability + Return on assetsa
performance Lay-offs + Scrap rate + Quality + range of indicators performance
Performance + Return on equitya
variables Accidents Labour turnover + over last 3 years: Job creation
Grievances + including Employee relations
Unit costs + Quality+ Climate
Labour productivity Profitability + Labour turnover
Customer complaints Relations between Absenteeism
Product/service managers and
quality employees+
Sales Customer satisfaction
Market share and
profits
Measures of strategic None Porter's cost- None None None None Miles and Snow's
integration/business minimization and prospector, analyser
strategy quality dichotomy and defender
trichotomy
Moderators None None None None None None Fit tested
Supported H1 on some H1 but maybe H3 H1 but maybe H3 H1 on all but 1 No unique effect for H1 for some practices H2
hypothesis(es) outcomes outcome high-involvement
management
405
406

Table 1 (continued)

Study: Youndt et al. Huselid Becker and Huselid Hoque MacDuffie Lawler et al. Lawler et al. Guest and Hoque
(1996) (1995) (1998a,b) (1999) (1995) (1995) (1998) (1994)

Hypothesis(es) Universalism and Universalism and Universalism and Universalism/ Universalism plus Universalism plus Universalism plus Universalism and (mild)
tested contingency contingency contingency element of interaction with interaction with interaction with TQM contingency theory
contingency quality practices TQM
Underlying Human-capital- High-performance High-performance HRM as Lean production High-performance High-performance HRM as commitment
conception enhancing HR management management commitment and work systems work systems and strategic
system (HCS) strategic integration
integration
Unit of analysis Workplace Company Company Workplace Plant Company Company Workplace
Sample 97 manufacturing Publicly quoted US 702 publicly quoted US 209 hotels in the 62 car assembly 279 US companies 212 US companies in 390 UK greenfield sites
plants firms firms UK with 25 or more plants in major car- in Fortune 1000 Fortune 1000 mainly manufacturing
employees producing
countries
Measures of 9 HR practices for 13 HRM practices – 21 HRM practices 22 high- High-commitment Range of practices Range of practices 21 practices – 19 high-
HR practices HCS factor analysis commitment HR practices plus classified into four classified into four commitment, two
6 for alternative revealed 2 factors – practices building flexible work groups: groups: information concerned with
administrative HR skills and on Wood and system practices information sharing, training and strategic integration
system motivation Albanese and Guest and bufferless sharing, training skills, reward, and
and Hoque (lean) systems and skills, reward, power sharing
and power sharing
Pay system in HR Group incentives, 1 item covering Merit or incentive pay Contingent pay Various forms of Various forms of Merit pay
system Skill-based pay, incentive profit- based on performance system non-day rate or non-day rate or
External equity, sharing and gain- appraisal salaried pay salaried pay
Salaried sharing plans Firm's target rate for
compensation total cash
(all in HCS) compensation,
Cash incentive, profit-
sharing or gain-sharing
plans,
Deferred incentive,
profit-sharing or gain-
sharing plans
 Measure of HR Aggregate-index of 2 scales from factor Aggregate index of 21 3-fold classification Clustering of 4 measures 4 measures 4 groups on basis of
system HCS based on 9 analysis items based on of HRM – based on practices corresponding to corresponding to median of aggregation
practices standardized means practices types of practices types of practices of 21 practices and
measure of whether
have HR strategy
Outcome/ Machine efficiencya Labour turnover Firm's market value + Commitment + Productivity + Total factor Sales per employee + Labour turnovera
performance Customer Productivity Job satisfaction + Quality + productivity + Return on sales + Disputesa
variables alignmenta Corporate financial Flexibility of staffa Sales per Return on assets + Qualitya
Employee performance Job mobilitya employee + Return on investment Commitment +
productivitya Tobin's q + Quality of worka Return on sales + + Quality of staff +
Quality of staffa Return on assets Total return to Labour mobility
Labour ++ investors Productivity
productivitya Return on Absenteeism
Quality of servicea investment ++
Financial Total return to
performancea investors
Measures of Manufacturing Strategic Organizational and Quality enhancers None but extent of None None 3 measures – 2
strategic strategy – quality, integration strategic fit measured Cost reduction/ quality practices included in measure of
integration/ delivery flexibility, measure used to within index of price competition measured high commitment
business strategy scope flexibility, validate scale implementation management, other
and cost alignment separately
Moderators Interactions Interactions tested Interaction between Business strategy Interaction Interaction with Interaction with Interaction effect
between HP system and between HR, work quality practices quality practices and important on three
manufacturing implementation systems and tested ++ re-engineering measures
strategy and HR alignment tested quality/stock practices tested ++
systems practices tested ++
Supported H2 claimed, maybe H1 for some H1 and strategic fit H1 on some H3 H3 H3 H1 on some outcomes,
hypothesis(es) H3 outcomes shown to have an outcomes, H2 on interaction effect with
independent effect some, may support strategic integration
H3 on some outcomes,
strategic integration
alone has effect on
some others

Note: + High-involvement management significantly related to outcome.

a
Significant contingent relationship between high-involvement management and outcome.
407
Human resource collection but also the way these TQM (an organizational logic of quality) or
management and hypotheses have emerged in the debate as the significance given to the HR (an
performance it has progressed. This suggests that the organizational logic of HRM or high-
issues are far from fixed. performance in Becker and Huselid’s
terms). Both conceptually and empirically,
this aspect of the area is underdeveloped. It
Conclusions
is at this point that it particularly needs to
This paper has concentrated on connect with the critical literature on
overviewing the recent studies on HR HRM. More specifically, a central claim is
systems and performance largely in their that lean production and even high-
authors’ own terms. Thinking in terms of involvement management, at least in
the onion metaphor, I have concentrated practice, leave intact the basic Taylorist
on taking off the top few layers of skin. design of the majority of jobs, which is
The review has shown that beneath the top assumed to deskill workers. A corollary of
layer of universalism vs. contingency are a this is that if they have any effect on
number of more complicated hypotheses performance it is through work
linking HR practices to performance. The intensification, perhaps in conjunction with
debate is firstly a matter of differing routes the wage system if this is matched by pay
between these and performance, and how increases. In contrast, Becker and Huselid’s
one constitutes one’s HR system depends drive to show that HRM has a direct
on which of these one deems as most ‘‘bottom line’’ effect is based on an
significant. The empirical work, none the assumption that a skilful contribution is
less, has concentrated on assessing the link required of all members of the
between practices and performance, with organization. Once technological factors
an increasing disregard for the mechanisms and product differentiation dominated
linking them. This has meant that there has competition, but now human capital
been no systematic examination of the link dominates, thus creating the space for
between different performance measures personnel management to prove its worth
and particularly between HR outcomes and and for all to be involved in the
performance. Moreover, there has been an organization. As a narrative, it works as
increasing neglect of the psychological long as there are no Taylorist jobs. But the
processes that mediate or moderate the evidence critics draw on is that Japanese-
link between HR practices and style high-involvement and quality
performance. This may be a reflection of management in manufacturing settings is
the ascendancy of the strategy over social more about procedural changes in the way
psychology literature, in which so much of jobs are designed and less about their
the early foundations of HRM were based. substance (Babson 1995; Berggren 1993,
The issue has been treated as an 22–55; Wood 1993). In the extreme, the
organizational-level matter, when it is in worker is involved in perfecting the design
fact a multi-level question.11 of his/her routinized work (thus breaking
Secondly, the debate is about the the Taylorist divide between conception
importance of organizational fit and, in and execution but making his/her execution
ß Blackwell Publishers Ltd 1999 turn, about the relative importance of either even more Taylorist). Becker and Huselid,

408
in arguing that all employees can have a strategy, will be included in the analysis.
strategic impact, either romanticize the role Going beyond the existing terms of the
that workers can have or genuinely believe debate particularly requires more thought
that high-performance is only possible with being given to the implications of the
well-designed jobs. The latter is though a variety of TQM systems that we find in
psychological, not a business strategy practice, and particularly to two distinctions
argument. This takes us back to the origins that have emerged in the literature. First is
of high-commitment management in job the differentiation between TQM systems
design theory, the implication being that that are focused on controlling processes
any further onion-peeling will lead to the and add little involvement and those December 1999
long-standing debate about the conditions oriented towards organizational learning
in which job design is required or will (Sitken et al. 1994). The second distinction
work. The HRM message about the is between those that rely on heavy
importance of internal fit is then first and supervision and those based on self-
foremost about the practices that either managed teams (Appelbaum and Batt 1994;
make up or complement good job design. Wood 1990, 181). Attending to such
If one’s arm were twisted to make an nuances will itself require that
‘‘overall’’ conclusion on the balance of the consideration is given to the neglected
evidence so far, one in favour of fourth, environmental fit, and hence back to
Hypothesis 3 or even 4 would be just as the original basis of the contingency theory
justified as the universal, first hypothesis. of organizations, variation in uncertainty,
The thrust of this review is, though, that but it may equally lead us to the issue of
any such conclusion would be premature values, and ultimately away from best-
because of conflicting research results but, practice thinking.
more importantly, because the debate is still
in its infancy. Taking the terrain within its Acknowledgements
own terms, there are four priorities. The
first should be to acquire more I would like to thank Emmanuelle
comprehensive data sets. The second is to Callerot, Lilian de Menezes, John Godard,
assess more the relationships amongst both Mick Marchington, Ana Lasaosa and Toby
the HR practices and the various Wall for their comments on any earlier
performance measures, which should draft of this paper and, in addition, to
include the overall health and stress of Mick Marchington for his editorial help.
employees and measures of rates of change,
not least because of TQM’s emphasis on Notes
continuous improvement. Thirdly, more 1 See Parker et al. (1997) for a discussion of
attention should be given to the what they term the key strategic and role
psychological mediators and moderators. orientations assumed to be central to new
Fourthly, the issue of environmental fit production methods.
should be addressed directly and not 2 Womack et al. (1990, 91–93) show that there
reduced to strategic fit. Thus, both are car plants in Japan, which through the
use of lean production methods were able to
contextual variables and measures of
have high quality and low unit costs.
strategy, or better still, dimensions of ß Blackwell Publishers Ltd 1999

409
Human resource 3 Kanter (1989, 12), in a similar vein, Organizations Survey, which confirms the
management and reconciled her advocacy of the benefits of a association between high-performance
performance new post-entrepreneurial model with her practices and performance.
long-standing adherence to contingency 8 My research in UK manufacturing plants
theory by arguing that the former referred to (Wood 1996a) suggests that as managements
‘‘broad strategies that hold across many adopt HIM pay systems tend to lag HIM
industries’’, but that ‘‘the specific ways the practice, both conceptually in the minds of
strategies are interpreted and used clearly managers when planning change and
depend on each company’s history, temporarily in the implementation process.
environment, and current situation’’. 9 My research (Wood 1999), which uses a data
4 Becker and Huselid (1998a, 59), in the set for US workplaces, suggests that quality
fullest statement of their developing and high-involvement practices may well
framework, follow Schuler (1992) and break tend to coexist.
down what they call strategic HRM into five 10 It could be claimed that the assumption that
elements: HRM philosophy, HR policies, HR the practices have synergistic effects is not
programs, HR practices and HR processes. In necessarily being made in the studies that
these terms, the philosophy drives the design aggregate practices. That they form a family
of the other four elements, and there is a (i.e. there is some internal fit), is, however,
‘best practice’ philosophy: it is one that assumed; the main assumption then made is
approaches the design of HR systems that each practice has the same effect on
according to the principles of internal and performance, so firms on the lower and
external fit. Strategic contingencies enter at middle range of the index may use a variety
the practices stage. of practices. Becker and Huselid (1998a, 64)
5 Becker and Huselid (1998a, 59) do not present this as a virtue. The underlying
entirely rule out ‘‘best practice’’ practices project of all writers, though, does appear to
since, once ‘‘the requirements of a particular assume that the whole is greater than the
firm’s policy in that area have been sum of the parts, diversity between writers
developed based on strategic considerations (if it exists) being more about the extent to
it may well be that a ‘best practice’ for such which the practices individually or, when
a policy exists’’. used sparsely, have any effect at all. Most
6 In addition to the measure of HR in business authors do not though write sufficiently
planning, Betcherman et al. (1994, 21) also about this for one to make a definitive
included a measure of the business strategy judgement.
of the firm: differentiating between those 11 The increasing neglect of the individual level
that concentrate on process, cost or product means that the results of commitment
improvements. The process strategy increases research at this level have been insufficiently
the probability of adopting the participation taken on board. In addition, the importance
model and, to a lesser extent, the of employee perceptions of justice and
compensation one, while decreasing the management’s motivations for adopting
probability of having a traditional approach practices (highlighted by Koys’ 1988 study)
(Betcherman et al. 1994, 58). Betcherman et have been largely ignored. A concern for
al. do not, however, look at whether the justice would provide space for the re-entry
strategy moderates the effects of the HR of industrial relations institutions and a fuller
system or whether those firms that have the treatment of the added value of having
participative HR system and adopt a process partnership relations with either unions or
strategy outperform all others. works council when following HIM that is
7 Delaney and Huselid (1996) did an assumed by Kochan and other industrial
ß Blackwell Publishers Ltd 1999
alternative analysis of the National relations specialists.

410
References and Walton, R. (1984). Managing Human
Assets. New York: Free Press.
Appelbaum, E. and Batt, R. (1994). The New Berggren, C. (1993). The Volvo Experience.
American Workplace: Transforming Work London: Macmillan.
Systems in the United States. Ithaca, NY: Betcherman, G., McMullen, K., Leckie, N. and
Cornell IR Press. Caron, C. (1994). The Canadian Workplace in
Arthur, J.B. (1992). The link between business Transition. Kingston, Ontario: IRC Press.
strategy and industrial relations systems in Delaney, J.T. and Huselid, M.A. (1996). The
American steel minimills. Industrial and Labor impact of human resource management
Relations Review, 45(3), 488–506. practices on perceptions of organizational
Arthur, J.B. (1994). Effects of human resource performance. Academy of Management December 1999
systems on manufacturing performance and Journal, 39(4), 949–969.
turnover. Academy of Management Journal, Delery, J.E. and Doty, D.H. (1996). Modes of
37(4), 670–687. theorizing in strategic human resource
Babson, S. (1995). Lean Production and Labor: management: tests of universalistic,
Empowerment or Exploitation. Detroit: Wayne contingency, and configurational performance
State University Press. predictions. Academy of Management Journal,
Baird, L. and Meshoulam, I. (1988). Managing 39(4), 802–835.
two fits of strategic human resource Deming, W.E. (1986). Out of Crisis: Quality,
management. Academy of Management Review, Productivity and Competitive Position.
13(1), 116–128. Cambridge: Cambridge University Press.
Barney, J. (1995). Looking inside for competitive Dunlop, J.T. and Weil, D. (1996). Diffusion and
advantage. Academy of Management Executive, performance of modular production in the
9(4), 49–61. apparel industry. Industrial Relations, 35(3),
Beaumont, P. (1995). The Future of Employment 356–373.
Relations. London and Thousand Oaks: Sage. Fenton O’Creevy, M. (1998). Evaluating Human
Becker, B.E. and Gerhart, B. (1996). The impact Resource Initiatives. Milton Keynes: Open
of human resource management on University Press.
organizational performance: progress and Foulkes, F. (1980). Personnel Policies in Large
prospects. Academy of Management Journal, Non-union Companies. Englewood Cliffs, NJ:
39(4), 779–801. Prentice Hall.
Becker, B.E. and Huselid, M.A. (1998a). High Godard, J. (1999). The transformation of work
performance work systems and firm and high performance? Alternative work
performance: a synthesis of research and practices, the quality of working life, and
managerial implications. In Ferris, G.R. (ed.), performance-related behavior. Mimeo, Faculty
Research in Personnel and Human Resources, of Management, Winnipeg.
Vol. 16. Stamford, CT: JAI Press, pp. 53–101. Guest, D. (1987). Human resource management
Becker, B.E. and Huselid, M.A. (1998b). Human and industrial relations. Journal of
resources strategies, complementarities, and Management Studies, 24(5), 503–521.
firm performance. Mimeo. School of Guest, D. (1995). Trade unions and industrial
Management, State University of New York at relations. In Storey, J. (ed.), Human Resource
Buffalo. Management: a Critical Text. London:
Becker, B.E., Huselid, M.A., Pickus, P.S. and Routledge, pp. 110–141.
Spratt, M.F. (1997). HR as a source of Guest, D. (1997). Human resource management
shareholder value: research and and performance: a review and research
recommendations. Human Resource agenda. International Human Resource
Management, 36(1), 34–47. Management, 8(3), 263–276.
Beer, M., Spector, B., Lawrence, P., Mills, D. Guest, D. and Conway, N. (1999). Peering into ß Blackwell Publishers Ltd 1999

411
Human resource the black hole: the downside of the new organizational commitment. Employee
management and employment relations. British Journal of Responsibilities and Rights Journal, 1(1), 57–
performance Industrial Relations, 37(3), 367–389. 68.
Guest, D. and Hoque, K. (1994). The good, the Lawler, E. (1986). High-involvement
bad and the ugly: employee relations in new Management. San Francisco: Jossey-Bass.
non-union workplaces. Human Resource Lawler, E. (1991). The design of effective reward
Management Journal, 5(1), 1–14. systems. In Steers, R.M. and Porter, L.W. (eds),
Hoque, K. (1999). Human resource management Motivation and Work Behaviour, 5th edition.
and performance in the UK hotel industry. New York: McGraw-Hill, pp. 507–531.
British Journal of Industrial Relations, 37(3), Lawler, E.E., Mohrman, S.A. and Ledford, G.E.
419–443. Jr (1995). Creating High Performance
Huselid, M.A. (1995). The impact of human Organizations. San Francisco: Jossey-Bass.
resource management practices on turnover, Lawler, E.E., Mohrman, S.A. and Ledford, G.E.
productivity, and corporate financial Jr (1998). Strategies for High Performance
performance. Academy of Management Organizations. San Francisco: Jossey-Bass.
Journal, 38(3), 635–672. MacDuffie, J.P. (1995). Human resource bundles
Huselid, M.A. and Becker, B.E. (1996). and manufacturing performance: organizational
Methodological issues in cross-sectional and logic and flexible production systems in the
panel estimates of the human resource–firm world auto industry. Industrial and Labor
performance link. Industrial Relations, 35(3), Relations Review, 48(2), 197–221.
400–422. Miles, R.E. and Snow, C.C. (1984). Designing
Ichniowski, C., Kochan, T., Levine, D., Olson, C. strategic human resource systems.
and Strauss, G. (1996). What works at work: Organizational Dynamics, 13(1), 36–52.
overview and assessment. Industrial Relations, Milgrom, P. and Roberts, J. (1995).
35(3), 299–333. Complementarities and fit: strategy, structure,
Ichniowski, C., Shaw, K. and Prennushi, G. and organizational change in manufacturing.
(1997). The effects of human resource Journal of Accounting and Economics, 19(2/3),
management practices on productivity. 179–208.
American Economic Review, 87(3), 291–313. Osterman, P. (1994). How common is workplace
Kalleberg, A., Knoke, D., Marsden, P. and transformation and who adopts it? Industrial
Spaeth, J. (1996). Organizations in America. Relations and Labor Relations Review, 47(2),
Thousand Oaks and London: Sage. 173–188.
Kanter, R.M. (1989). When Giants Learn to Parker, S., Wall, T. and Jackson, P. (1997).
Dance. London: Simon & Schuster. ‘‘That’s not my job’’: developing flexible
Kochan, T. (1992). Principles for a post-New employee work orientations. Academy of
Deal employment policy. Sloan School of Management Journal, 40(4), 899–929.
Management, MIT, Working Paper 5. Patterson, M., West, M., Lawthorn, R. and
Cambridge, MA: MIT. Nickell, S. (1997). Impact of people
Kochan, T. and Osterman, P. (1995). Mutual management practices on business
Gains. Boston, MA: Harvard Business School. performance. Issues in People Management,
Koike, K. (1987). Human resource development No. 22. London: Institute of Personnel and
and labour-management relations. In Development.
Yamamura, K. and Yasuba, Y. (eds), The Pfeffer, J. (1994). Competitive Advantage
Political Economy of Japan, Vol. 1, The Through People: Unleashing the Power of the
Domestic Transformation. Stanford, CA: Workforce. Boston: Harvard University Press.
Stanford University Press, pp. 289–330. Pfeffer, J. (1997). New Directions for Organization
Koys, D.J. (1988). Human resource management Theory. Oxford: Oxford University Press.
ß Blackwell Publishers Ltd 1999
and a culture of respect: effects on employees’ Pil, F.K. and MacDuffie, J.P. (1996). The

412
 adoption of high-involvement work practices. management model and new technology.
Industrial Relations, 35(3), 423–455. Applied Psychology, 39(2), 169–190.
Piore, M. and Sabel, C.F. (1984). The Second Wood, S. (1993). The Japanization of Fordism.
Industrial Divide. New York: Basic Books. Economic and Industrial Democracy, 14(4),
Porter, M. (1985). The Competitive Advantage. 538–555.
New York: Free Press. Wood, S. (1996a). High commitment
Schuler, R.S. (1992). Strategic human resource management and payment dystems. Journal of
management: linking people with with the Management Studies, 33(1), 53–77.
need of the business. Organizational Wood, S. (1996b). How different are human
Dynamics, 21 (summer), 18–32. resource practices in Japanese ‘‘Transplants’’ in
Schuler, R.S. and Jackson, S. (1987). Linking the UK? Industrial Relations, 35(4), 511–525. December 1999
competitive strategies with human resource Wood, S. (1999). Getting the measure of the
management practices. Academy of transformed high-performance organization.
Management Executive, 1(3), 207–219. British Journal of Industrial Relations, 37(3),
Sitken, S.B., Sutcliffe, K.M. and Schroeder, R.G. 391–417.
(1994). Distinguishing control from learning in Wood, S. and Albanese, M. (1995). Can you
total quality management: a contingency speak of a high commitment management on
perspective. Academy of Management Journal, the shop floor? Journal of Management
19(3), 537–564. Studies, 32(2), 215–247.
Walton, R. (1985). From ‘‘control’’ to Wood, S. and de Menezes, L. (1998). High
‘‘commitment’’ in the workplace. Harvard commitment management in the UK: evidence
Business Review, 63(2), 77–84. from the workplace industrial relations survey,
Womack, J., Jones, D.T. and Roos, D. (1990). and employers’ manpower and skills practices
The Machine that Changed the World. New survey. Human Relations, 51(4), 485–515.
York: Rawson Associates. Youndt, M.A., Snell, S.A., Dean, J.E. and Lepak,
Wood, S. (1989). The transformation of work. In D.P. (1996). Human resource management,
Wood, S. (ed.), The Transformation of Work. manufacturing strategy, and firm performance.
London: Allen & Unwin, pp. 1–43. Academy of Management Journal, 39(4), 836–
Wood, S. (1990). Tacit skills, the Japanese 866.

ß Blackwell Publishers Ltd 1999

413