Issues in People Management
The Impact of People Management
Practices on Business Performance:
A literature review
Dr Ray Richardson
Prelims.p65 16/06/03, 15:331
Titles in the Issues series:
Employee Motivation and the Psychological Contract
Fairness at Work and the Psychological Contract
How Dissatisfied and Insecure Are British Workers? A survey of surveys
The Impact of People Management Practices on Business Performance
The Impact of People Management Practices on Business Performance:
A literature review
The Impact of People Management Practices on Business Performance:
Practical case studies
Knowledge Management: A literature review
Managing Learning for Added Value
Performance Management through Capability
The State of the Psychological Contract in Employment
Workplace Learning, Culture and Performance
These titles can be ordered from Plymbridge Distributors
Tel: 01752 202301
The Institute of Personnel and Development is the leading publisher
of books and reports for personnel and training professionals, students,
and all those concerned with the effective management and
development of people at work. For full details of all our titles please
contact the Publishing Department:
tel. 020-8263 3387
fax 020-8263 3850
The catalogue of all IPD titles can be viewed on the IPD website:
Prelims.p65 16/06/03, 15:332
Issues in People Management
The Impact of People Management
Practices on Business Performance:
A literature review
INSTITUTE OF PERSONNEL AND DEVELOPMENT
Dr Ray Richardson
London School of Economics
University of Oxford
Prelims.p65 16/06/03, 15:333
© Institute of Personnel and Development 1999
First published 1999
All rights reserved. No part of this publication may be reproduced, stored in an information
storage and retrieval system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, without the written permission of the
Institute of Personnel and Development, IPD House, Camp Road, London SW19 4UX
Typeset by Paperweight
Printed in Great Britain by
Short Run Press, Exeter
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
ISBN 0 85292 832 7
IPD House, Camp Road, London SW19 4UX
Tel: 020-8971 9000 Fax: 020-8263 3333
Registered Charity No. 1038333.
A company limited by guarantee. Registered in England No. 2931892.
Registered office as above.
INSTITUTE OF PERSONNEL
Prelims.p65 16/06/03, 15:334
- Contents -
Executive summary xi
Chapter 1 The background to this report 1
Chapter 2 The theory of human resource strategy 3
Chapter 3 The empirical literature 17
Chapter 4 Practitioner perspectives 25
Chapter 5 Future research 30
Appendix 1 Detailed reviews of key texts 33
Appendix 2 Current and planned research on HR strategies
and business performance in the UK 51
Appendix 3 Analysis of key empirical studies 53
References and further reading 75
Prelims.p65 16/06/03, 15:335
Ray Richardson is Reader in Industrial Relations at the London School of Economics.
He has taught and done extensive research for a number of years in the area of Human
Resource Management. His recent research focus has been on performance related pay
(for example the forthcoming IPD survey report no. 9, UK Performance Pay Trends) and
HR strategies where, apart from the present report, he is working with Marc Thompson
on a survey of HR strategies in the UK.
Marc Thompson is Research Fellow at Templeton College, University of Oxford. His
principal research interest is in strategic human resource management (SHRM). He is
currently conducting a two year programme of work on high performance work systems
in the UK aerospace industry (particularly the diffusion of lean production methods).
Marc is also directing an ESRC research project that looks at the reward practices of
innovative organisations in the knowledge sectors of the economy. He has been working
with the IPD on human resource management and business performance issues for the
last two years.
Prelims.p65 16/06/03, 15:336
Institute of Personnel and Development
There is a new and exciting area of research emerging which marks an important
phase in the development of our understanding of the relationship between people
management and organisational performance. Managers know that the way people
are managed and developed affects the bottom line, but the difficulties of evaluation
have, until now, limited the arguments that can be made in favour of more
sophisticated, systemic approaches to people management. In the last few years
the situation has begun to change, and a body of hard evidence is now emerging
which demonstrates that the way people are managed is a crucial factor in predicting
This report reviews the most significant literature currently available. It analyses
the evidence in an objective, structured way identifying what we can be confident
of, and where we need more data to prove the case. In particular, the researchers
have studied the range of methodologies used and the way in which data have
been analysed and interpreted to identify gaps in the research and hitherto
unexplored sectors and industries.
What we do know is that there is a range of independent studies conducted both
in the UK and overseas which broadly demonstrates that good people management
practices equate to better business performance. Indeed, by 1995 research by the
IPD was able to present a model that indicated that when people management
policies are integrated into a coherent philosophy and where this fits the business
strategy, superior performance results.
But, although the IPD has come further than ever before in proving the link
between people management and development practices and business performance,
the evidence is not yet complete. In particular, we need to develop a much better
understanding of how this relationship works in practice. We need to understand
the impact of context, both economic and geographical. We need to be confident
that the methodology and analysis is robust enough to combat even the most
sceptical, and we need to be able to understand it all within a coherent set of
arguments and theory.
As the professional body representing those concerned with the management of
people, the Institute of Personnel and Development also needs to be able to support
people managers engaging with these issues. We must establish frameworks within
which they will be able to make informed and reliable choices on what to do and
how to do it. We need to demonstrate to managers that there is a body of knowledge
and understanding that can be applied in the real world and that will enable them
to develop a people management strategy that will both fit with and drive their
Prelims.p65 16/06/03, 15:337
Work and research programme
To improve the evidence available and develop our understanding of this area, the
IPD has embarked on an ambitious programme of work. This programme includes
new research, collaborative work and consultation among members and other
interested agencies and individuals. The programme is, therefore, multi-faceted
and aimed at all levels of our profession as well as the wider business and economic
communities. However, there is one overall research hypothesis that is guiding
the entire programme of work: that a coherent framework can be developed within
which organisations can make informed choices about which people management
practices and combinations of practices are most likely to lead to better
The IPD programme of work has its origins in the mid-1990s. From the research
available then, including the extensive work carried out at Cranfield School of
Management, it was apparent that little was known about the causal links between
people management and business performance. The studies that are currently in
progress should add to this understanding, but there is a need to develop a theory
of causality and to identify a form or forms of evaluation to determine how and
why certain practices make a difference.
In 1997 the IPD commissioned the Institute of Work Psychology at Sheffield
University to analyse the individual elements of management activity and measure
their contribution to performance. The results are based on data collected as part
of a ten year longitudinal study (1991-2000) examining market environment,
organisational characteristics and managerial practices in over 100 UK
manufacturing companies. This study forms part of the Sheffield Effectiveness
Programme. The findings are remarkable. They show decisively that people
management practices exert an extremely powerful influence on performance; more
powerful than all the other practices hitherto believed to be the most important,
such as business strategy, technology and research and development.
These results are supported by work carried out in the USA by Mark Huselid and
others, which demonstrates a consistent positive relationship between people
management practice and current and subsequent performance. In addition to the
survey data available, a wealth of best practice evidence is beginning to emerge.
Both David Ulrich and Jeffrey Pfeffer put forward frameworks to analyse the role
that the personnel function needs to adopt to help deliver organisational excellence.
Their evidence has been refined from observations over a number of years in many
different organisations and from the literature that already exists. The Institute is
tapping into this evidence to inform our research programme, and for the 1999
IPD National Conference programme and consultation purposes.
In 1998 we commissioned this report from Marc Thompson and Ray Richardson
to help us in developing our plan of action. We asked them to review the available
evidence from an academic standpoint, identifying the strengths and weaknesses.
We hoped from this process to be able to develop a way forward that would enable
us not only to contribute to and develop the academic debate but to bring the
views of practitioners and academic researchers together in a meaningful and useful
Prelims.p65 16/06/03, 15:338
An action plan for the future
A number of initiatives have already resulted from this strategy A group of senior
people management practitioners, together representing 12 per cent of the UK
workforce, met in February 1999. This group confirmed that while the practitioner
community is aware of the evidence that is becoming available, it needs hard proof
and accessible tools to enable it to apply it in the workplace. It also confirmed that
the research evidence must recognise the impact of market forces on the choices
that organisations are able to make if is to achieve any credibility in the business
community. In particular, the group emphasised the need to develop mechanisms
to integrate people management with strategy, and equip personnel practitioners
with the necessary competencies to manage in a climate of uncertainty.
Feedback from other practitioner groups indicates that practitioners are both able
and willing to take a lead in this area in the further development of our programme.
In particular, there is a strong belief that practitioners must be proactive in the
development of diagnostic tools and engage in meaningful dialogue with academics
if any of the academic efforts are to have real impact and benefit in the day to day
process of managing business. Of course, the IPD is catering for the needs of a
number of different audiences. These range from practitioners who want advice
on the design and choice of practices, to business leaders who need definitive
evidence and explicit understanding to adopt strategies that will both achieve higher
business performance and satisfy the needs of all the stakeholders in their business
– employees, shareholders and customers.
The Institute has therefore developed a plan of action designed to build upon the
achievements to date in this field of research, which we believe represents a major
breakthrough in the understanding of the value of good people management. We
intend to focus out efforts on two major areas:
• extending and developing the data and supporting the development of a theory
to underpin the evidence that exists in this area
• developing understanding of the process or processes by which people
management practices have an impact on business performance and by which
policy intention is turned into action.
In addition it is vital that our programme of work ensures that the debate is carried
out among as wide a group as possible. It is not just the people management
specialists who have a stake in this field, but everyone concerned to develop business
success. We also need to engage the business leaders and those responsible for
shaping the competitiveness agenda. In addition, there is an urgent need to develop
better ways of accessing the considerable body of academic research that already
exists and using it as a basis for developing usable tools for managers. Without
this, much valuable work will be wasted simply because managers have no means
by which to apply it in their own business situation.
The literature search has itself thrown up a number of issues. It is evident that
although there may be some guiding principles, it is unlikely that there will ever
be ‘one way’ to business success that will work in all organisations in all
circumstances. Organisations need to develop strategies, which are rooted in their
particular environment and business position. This report provides a critical focus
Prelims.p65 16/06/03, 15:339
for understanding more about the important relationship between the management
of people and business performance.
We do not intend or expect to achieve all of our aims and objectives alone. We
shall take every opportunity within our work programme to operate in partnership
with other agencies and interested organisations. We also do not want to reinvent
the wheel, and are consulting as widely as possible to be aware of all the initiatives
that are taking place under this broad heading of people management and business
performance. In particular, we intend to look at some of the sectors that, hitherto,
have been untouched by the research that has been carried out.
Despite the multi-faceted approach we are taking, the aim is singular: to
demonstrate unequivocally that better people management makes the critical
difference to organisational performance.
Prelims.p65 16/06/03, 15:3310
This report summarises some of the key themes emerging from an on-going review
of the theoretical and empirical work in the area of people management strategies
and business performance.
There are in the region of 30 empirical studies that have sought to address the
relationship between HR practices and business performance. This research field
has been pioneered in the USA and the bulk of the articles report on studies
conducted in North America. However, there are also a growing number of UK-
based programmes of work.
The published research generally reports positive statistical relationships between
the greater adoption of HR practices and business performance. In a review of
some of the better research in this area, Pfeffer (1998) cites evidence that HR
practices can raise shareholder value (ie stock market value) by between $20,000
and $40,000 per employee. A study by Huselid and Becker (1995) suggested that
‘market value per employee’ was strongly correlated with the sophistication of
the HR practices adopted. In other words, businesses got above average returns
when they used a broader range of practices for more employees and also integrated
these practices within the business. In the UK, researchers at Sheffield University
(Patterson et al, 1997), while unable to put a market-value on the business impact,
found that changes in profitability among a panel of over 60 small to medium
sized single-site manufacturing businesses was significantly correlated with the
adoption of certain HR practices. Thompson (1998) reporting findings from an
establishment-level survey of the aerospace industry found that firms reporting
higher levels of value-added per employee used HR practices for a greater
proportion of their employees and were also more likely to have a board-level
As the work in the area has developed, researchers have begun to delineate three
broad perspectives on the way in which HR practices contribute to business
1 Best practice – one set of HR practices can be identified which when implemented
will raise business performance
2 Contingency – business performance will be improved only when the right fit
between business strategy and HR practices is achieved
3 Bundles – specific combinations of HR practices can be identified which generate
higher business performance but these combinations will vary by organisational
Overall more positive effects have been ascribed to the ‘best practices’ perspective.
However, the claims that a universal best practice HR strategy has been identified
are premature. It is unlikely that adopting a specified set of HR policies is the high
road to organisational success. Even the large amount of empirical work that has
been done has not identified all the general components such a set of policies
would contain. However, this is not to say that the best practice literature is without
Prelims.p65 16/06/03, 15:3311
interest and they give a checklist of policy combinations which organisations should
actively consider adopting.
Virtually all the statistical analysis defines and measures strategy by a statistical
procedure rather than a principled argument. There is a need to start to identify
strategies by their intrinsic theoretical and real world properties.
We need to move away from a concentration on the analysis of two end points and
pay more attention to the intervening steps, for example looking at what mediates
the two variables ‘strategy’ and ‘performance’. We need to understand why
particular combinations of policies might work and what discretion managers might
have within their ability to affect these mediating variables.
How something is done is often more important than what is done, and we need
to pay much more attention to how clusters of HR polices are adopted and
implemented as well as to the specific contexts in which policy innovation is
A good deal of the research work in this area has progressed with very little
engagement with the practitioners who are important ‘customers’ for this research
and whose behaviour is the most likely to be influenced by convincing research
findings. If this situation continues, the studies may not provide enough policy
precision and little guidance for personnel and development managers on the
content and form of strategy. It is clear that there is a need for process-based
research which can also lead to the development of learning tools that can help
practitioners make better decisions about personnel and development investments.
Prelims.p65 16/06/03, 15:3312
The background to this report
In his book, The Human Equation, Jeffrey Pfeffer argues that the way in which
organisations treat their employees is at the heart of their success (Pfeffer, 1998).
Pfeffer is highly prescriptive. He claims that virtually all organisations, commercial
and non-profit organisations alike, will be more successful if they follow a particular
Human Resource strategy, comprising a specified number of people management
policies. He supports this position with a wealth of anecdotes and illustrative cases
but, in contrast to other books in this mode, his message is also firmly grounded
in a growing number of well-thought-out research studies which use heavyweight
For example, he quotes Huselid, who looked at a large sample of firms from across
the whole of the US economy and concluded that the heavy use of a number of
specified HR practices was associated with a level of sales revenue that was on
average $27,000 per year per employee. The corresponding increase in shareholder
value was put at $18,500, while the increase in profits was estimated at nearly
$4,000 per year per employee (Huselid, 1995). Subsequent and more refined studies
by Huselid suggest that the gains are even larger. These are very impressive numbers
but they are not unique. Other studies by a variety of authors, whether in
automobiles, steel, clothing, oil refining or in the service sector, point to similarly
impressive outcomes, usually measured by productivity gains. So, Pfeffer claims,
there is powerful evidence to suggest that deploying certain high performance
management practices makes bottom line sense.
Findings like these excite both the academic and the practitioner, and they promise
to move the personnel or HR function away from being seen as a peripheral
administrative activity and put it at the centre of the organisation’s activities. The
key purpose of this report, therefore, is to look hard and critically at the body of
academic work that underpins Pfeffer’s analysis. Like him, we see these studies as
having very important implications. They stimulate thought, they cannot be lightly
dismissed, and they point to something that deserves sustained attention. Where
we differ from Pfeffer is that we do not see them as being quite so authoritative a
basis for firm and detailed prescription. The studies have their weaknesses and
these should not be underestimated. They are neither absolutely conclusive nor
do they address all the relevant issues. So this report explores both the strengths
and the frailties of the research to see where we stand at present.
As a prologue, and to provide a context for what is to come, we set down a little
more detail from Pfeffer. He concludes that a set of seven people management
policies ‘seem to characterize most if not all of the systems producing profits
through people’ (p.64). He specifies these policies as:
• an emphasis on providing employment security
• putting in a lot of resources to recruit the right people in the first place
• an extensive use of self-managed teams and decentralisation
Chap1.p65 16/06/03, 15:301
• wage levels that are high and are strongly linked to organisational performance
• a high expenditure on training
• attention being paid to reducing status differentials
• a willingness to share information.
Pfeffer’s HR strategy then becomes, at its core, these seven people management
policies, each of which may be embodied in a variety of concrete and detailed
people management practices (eg there are many techniques that might encourage
the sharing of information within an organisation). Pfeffer does not suggest that
the practices have to be the same in every organisation. Even so, however, his
conclusions go some way beyond the academic research.
First, the empirical support for each of the seven policies is of variable strength –
it is clearer for some than for others. Second, there is even less agreement on the
merits of the different practices that might be used to achieve any one policy – for
example, are all methods of linking pay to organisational performance equally
effective? Third, not all of the studies he relies on include all of the seven policies
when they relate HR strategies to business performance. This means that the
empirical support for Pfeffer’s particular combination of policies is less secure
than he implies. Many modern managers will agree that his seven policies are worth
taking seriously, though some may also feel that he has omitted some important
candidates. And Pfeffer is to be applauded for making his case so clearly and with
such enthusiasm. Nevertheless, matters have not been resolved as completely as
his advocacy claims. The management literature has seen too many prescriptions
which seem persuasive, but turn out to be premature or misleading or simplistic.
More caution, and more knowledge about the underlying scientific work, is
appropriate at this stage.
Chapter 2 discusses the meaning of HR strategies and considers the theory
underpinning this area of research. Chapter 3 looks at the available empirical
evidence and reviews the problems and issues with this body of evidence. Chapter
4 reports the key themes from practitioner consultations and their implications
for the future research agenda. Finally, Chapter 5 pulls together the main findings
from the research review, identifies the main research gaps and suggests where
IPD resources might be best allocated.
Chap1.p65 16/06/03, 15:302
The theory of human resource strategy
Strategy in general
This report is concerned with the effectiveness of human resource strategies, so
an obvious first step in a theory section is to discuss what, in general terms, an HR
strategy is. Having done this one can then turn to the theoretical issues raised
when assessing the effectiveness of the human resource strategies.
Defining ‘strategy’ is no easy matter. For the purposes of this report, however, a
strategy, whether an HR strategy or any other kind of management strategy, must
have two key elements: there must be strategic objectives (ie things the strategy is
supposed to achieve), and there must be a plan of action (ie the means by which it
is proposed that the objectives will be met).
However desirable it may be to have a strategy, it is not an automatic passport to
organisational success. Strategies can be good or bad. A bad strategy would either
specify inappropriate or inconsistent objectives, or propose a plan of action that
was seriously deficient.
A strategy is often taken to imply a long-term perspective. This is often, but not
quite always, true. It might be entirely sensible for an organisation knowingly to
pursue a short-term strategy, or indeed simultaneously to have both a short-term
and a long-term strategy that were not identical.
A further quality a strategy is often thought to have is breadth. In some sense, it
reflects the whole rather than part of the picture. What this means is that, to be a
success, the plan of action should somehow capture all, or nearly all, of the different
elements of the situation.
Organisations do not necessarily have strategies, certainly not in any well-defined
or explicit form; or they may have a strategy for one part of their activities but not
for all of them. They may, for example, have articulated some form of overall
business strategy but not have developed the corresponding strategies for each of
the functional areas of management, such as people management.
Finally, it is probably true that most organisations arrive at their strategies in a
somewhat haphazard way. It is misleading to suggest that senior managers typically
engage in a process of thinking through all their objectives and then calmly,
rationally and comprehensively go through all the policy options before judiciously
choosing the optimal set. On the contrary, most organisations operate on a
piecemeal basis, responding to sudden emergent pressures, and are subject to a
variety of powerful internal political pressures which contribute to inconsistencies
among their policy choices. Most of the literature in the area of human resource
strategies ignores the actual process by which the strategies are formed and
concentrates instead on the chosen policies and practices. This is an important gap.
Chap2.p65 16/06/03, 15:303
The meaning and content of HR strategies
In line with what has already been said, an HR strategy must specify HR objectives
and have an associated plan of HR action. The objectives of an HR strategy could
be expressed in a very general way, perhaps as being precisely the same as the
organisation’s overall objectives. The dominant overall objective for commercial
organisations is usually taken to be the maximisation of profits or ‘shareholder
value’. This is sometimes stated explicitly and sometimes left implicit, in which
case a firm’s mission statement may express its business objectives in such terms
as becoming the leader in its industry, or delivering an innovative and high quality
product range. The presumption is that these other objectives are the concrete
way in which the firm believes that it can maximise shareholder value.
Non-commercial organisations, such as NHS trusts or government agencies, often
have more trouble in specifying their overall objectives in a satisfactory way. They
may specify them as a range of service targets, subject perhaps to staying within a
pre-assigned budget. These service targets can sometimes cause difficulties because
they can be in conflict with one another; they will usually, for example, be
competing for the same limited budget, which implies trade-offs between the
individual targets. But, after some thought, the trade-offs can be prioritised and
overall targets can usually be expressed in a meaningful way.
Important as it is to specify overall business objectives, it is necessary to go further
when fixing the objectives of an HR strategy. Functional management areas like
HR need their own more precise, more specific and more focused objectives. This
is illustrated in approaches such as the Balanced Scorecard (see, for example, Kaplan
and Norton, 1996), which seek to identify the important concrete implications
for the activities and processes of the business that stem from shareholder objectives
or mission statements. These complementary objectives should, of course, be
consistent with overall objectives, but they need to be spelled out separately.
The objectives of HR strategy
To illustrate this point, one may envisage a chain or sequence of connections, see
Figure 1. A commercial firm wanting to increase its profits (its ‘ultimate objective’)
will see that one way to achieve this is to reduce its unit labour costs (a ‘production
objective’); one way to reduce unit labour costs is to increase employee productivity
(a ‘behavioural objective’); one way to increase employee productivity is to make
employees more competent (a ‘generic HR objective’); one way to increase
employee competence is to improve existing training programmes (an HR policy);
one way to improve the training programme is to bring in professional trainers
and make the process more formal and systematic (an HR practice, part of the
firm’s action plan). So we have a possible sequence:
More formal training arrangements
improved labour productivity
lower unit labour costs
The HR objective could be expressed as being any one of higher profits, lower
unit labour costs, higher productivity, or enhanced competence. They are all
objectives but they exist at different levels of generality.
Chap2.p65 16/06/03, 15:304
Table 1 HR objectives at four different levels
Table 1 implies that there are many possible chains or sequences. It suggests that
there may be three generic objectives of human resource policies, all referring to
the organisation’s workforce. All organisations want the right number of
employees, the right mix of employee competencies, and the right set of
motivations from their workforce. Of course, different organisations seek different
numbers of employees, and different bundles of competencies, and different types
of motivation, and an important part of a successful HR strategy is bound to be
the correct identification of these three elements.
The extent to which an organisation achieves its generic objectives will influence
how well it achieves its various ‘behavioural’ objectives. All three generic objectives,
for example, will directly affect productivity levels. But firms are not interested
purely in output per employee; they increasingly want their employees to pay
close attention to quality standards, or to act in an innovative or creative way, or
to co-operate and share valuable information with one another; and they will also
normally want their employees (or, at least, their effective employees) to stay
with the firm rather than leave. Organisations will think that such aspects of
employee behaviour are influenced by employee competence and by the extent to
which employees are motivated to pursue a variety of behaviours. So the extent to
which desired behaviour is achieved will turn heavily on the extent to which an
organisation succeeds in reaching its generic HR objectives.
The extent to which the behavioural objectives are reached will influence the success
the firm has in attaining its ‘production’ objectives. It is no longer adequate to
think of firms being interested solely in unit labour costs, important as these are
for competitive advantage. Real and perceived product quality levels are also
important. In this respect, much of the service sector is in a distinctive situation
because its employees frequently interact directly with customers, so that product
quality inherently becomes a matter of direct employee performance (with little
or no possibility of subsequent ‘rectification’).
Achieving the firm’s ultimate objective, eg profits, will obviously depend on the
degree to which its production objectives are reached. Low unit labour costs lead
directly to profits via higher profit margins; they can also lead indirectly to higher
profits by allowing the firm to cut its product prices and increase its market share.
Similarly, high quality levels may allow prices to be raised, thereby raising profit
Unit labour costs
Chap2.p65 16/06/03, 15:305
margins; equally, they may directly increase market share through customer
perception of better value for money.
So, to summarise, the objectives of an HR strategy can, with equal validity, be
expressed at any one of four different levels of generality, ranging from the ‘generic’
to the ‘ultimate’. If they are expressed at a relatively general level, however, the
accompanying strategic ‘plan of action’ should also reflect the prior sequence of
intermediate objectives, as well as the choice of HR policies. So, for example, if
the HR strategy’s objective is expressed as emphasising quality standards, the
route by which this is to be achieved should be specified from specific HR policies
through generic and behavioural objectives.
The HR strategy’s plan of action
The HR plan of action is the set of things the organisation intends to do in its
attempt to achieve its HR objectives. Most obviously and centrally it includes the
various HR policies and practices the organisation chooses to adopt. But, and this
is of great theoretical importance, it goes significantly wider than that. Policy
intentions are not always translated into action. This means that the way in which
policy is implemented must also be recognised as part of the plan of action (and,
hence, of the strategy). In addition, and lying behind the set of HR policies, there
is some reasoning linking the policies and practices to the objectives; how are they
seen to work, and why do they have the effects they are supposed to have? So the
action plan reflects a lot more than the formally adopted HR policies.
In fact, most researchers in this field measure the firm’s plan of action purely by
its choice of HR practices; indeed, that is typically how they represent the HR
strategy itself (ie different HR objectives are very often ignored). In other words,
researchers tend to focus exclusively on the combination of individual HR practices
that a firm does or does not adopt; it is a particular combination of practices which
assigns the firm to one strategy category or another; the specification of strategic
objectives, the problems of policy implementation, and the nature of the arguments
that might link policies to objectives are usually not explored in any depth.
Huselid, for example, in a very widely quoted study (Huselid, 1995) develops two
complex measures of so-called ‘high performance work practices’ which are, in
essence, his index of the firm’s HR stance, or strategy. As Table 2 shows, the first
measure, ‘Employee skills and organizational structures’, is a conglomeration of
nine HR practices; the second, ‘Employee motivation’, is a set of four such policies,
rules or practices. Huselid has also, however, made some attempt to capture
implementation by asking about coverage or intensity of use (eg in his formulation
‘what proportion of the workforce is covered by…’?). This is not a full measure of
implementation. It should capture how widely a firm’s practices are targeted but
it will certainly not capture such matters as the manner in which they are deployed,
eg whether they are pursued by managers with consistency, equity, vigour and
Chap2.p65 16/06/03, 15:306
Table 2 An attempt to measure HR strategy
Employee skills and organisational structures
01 What is the proportion of the workforce who are included in a formal information
02 What is the proportion of the workforce whose job has been subjected to a formal
03 What proportion of non-entry level jobs have been filled from within in recent
04 What is the proportion of the workforce who are administered attitude surveys on
a regular basis?
05 What is the proportion of the workforce who participate in QWL programs, QCs
and/or labor-management participation teams?
06 What is the proportion of the workforce who have access to profit sharing/gain
07 What is the average number of hours of training received by the typical employee
over the last 12 months?
08 What is the proportion of the workforce who have access to formal grievance
procedures and/or a complaint resolution system?
09 What proportion of the workforce is administered an employment test prior to
10 What is the proportion of the workforce whose performance appraisals are used to
determine their compensation?
11 What proportion of the workforce receives formal performance appraisals?
12 Which of the following promotion rules do you use most often? (a) merit or
performance rating alone; (b) seniority only if merit is equal; (c) seniority among
employees who meet a minimum effort requirement; (d) seniority.
13 For the five positions that your firm hires most frequently, how many qualified
applicants do you have per position (on average)?
Huselid’s approach identifies firms that are more or less heavy users of (some)
high-performance work practices. In this case, therefore, only one HR strategy is
articulated and explored; firms are arranged on a continuum which ranges from
heavy users of high performance practices to those who make no use of them at
all. A different way of specifying HR strategies comes from concentrating on the
links between the set of HR policies and the three generic objectives, ie how and
why HR policies might allow the organisation to achieve its three generic objectives.
Some years ago, Walton (1985) compared two different styles of managing people.
He talked about a control approach and a commitment approach. Both approaches
have active practices in each of the standard HR policy areas, ie all organisations
have to make selection decisions, training decisions, work design decisions, etc.
But, for Walton, the detailed choice of practices within each of these areas were
different for the two HR approaches. What he was essentially distinguishing were
two very different underlying employment relationships (these might also be called
different psychological contracts or different implicit contracts).
Chap2.p65 16/06/03, 15:307
There are more than two possible employment relationships. Chadwick and
Cappelli (1998a) talk about ‘Contract’ and ‘Investment’ relationships. At its core,
their Contract relationship involves the employer first seeking to hire people who
are already more or less fully competent to do the range of tasks they are asked to
perform, so that large training expenditures are not required; the employer then
seeks to motivate them to perform up to their potential by some kind of
performance-related reward system. This particular implicit contract is seen to
operate at ‘arms length’. The relationship is predicated largely on professionalism
and the provision of rewards for performance that both parties find appropriate;
in this scenario there is no premium on the employee generating complex attitudes,
such as loyalty to the organisation.
Chadwick and Cappelli’s ‘Investment’ system, by contrast, is based on an attempt
to foster among employees a sense of common destiny with the firm over the
long term. HR policies are therefore selected in an attempt to boost employee
involvement, enrich jobs and develop new skills.
The second column of Table 3 suggests that employment relationships can vary
on many dimensions. They might, for example, be long term or short term; they
might be predicated on the development of high commitment, or loyalty, or they
might not; they might place a premium on encouraging what is sometimes called
organisation citizenship behaviour (eg encouraging people to go voluntarily beyond
their job descriptions and engage in valuable extra-role behaviour), or not; they
might rest on the belief that the motivation to perform well is merely a matter of
designing the appropriate reward system, or they might think that task is far more
Table 3 The nature of the employment relationship – possible links
between HR policies and objectives
HR policy areas
The nature of the employment
Commitment or loyalty seen to
OCB seen to be important/
Competence hired in, or
developed in house
Trust seen as important/
Correct manning levels
Chap2.p65 16/06/03, 15:308
So an HR strategy can be defined as an attempt to develop a specified employment
relationship through the choice of particular HR practices. A firm wishing to
attack a problem of motivation by basing much of its HR strategy on high levels
of employee commitment to the firm must first address the prior question of
which HR practices are likely to achieve employee commitment. It might therefore
shape its selection policies differently, perhaps by putting more resources into
identifying which applicants are likely to have a disposition to be committed; or it
might rely on an internal labour market in order to encourage a sense that it is
offering greater job security; or it might adjust its reward system towards group
performance-related pay; or it might develop complex communications practices
in order to make employees feel fully part of the firm.
This approach to thinking about strategy puts a premium on exploring the ways
in which particular HR practices are likely to create a relationship which, in turn,
delivers what we have called the generic HR objectives of competence and
motivation. It implies that, to understand an HR strategy, it is necessary to know
not only the particular practices the firm adopts but also something about the
nature of the employment relationship it is trying to generate.
What is crucial here is the surprisingly complex idea of employee motivation.
Employee competence is certainly an important generic HR objective for all
organisations, because without it little can be achieved. But the more ambiguous
or mysterious objective is motivation. For example, changing the England football
manager seems sometimes to have a startling impact on the team’s performance,
even though more or less the same players are involved.
The issue of ‘fit’
A key idea in the strategy literature is ‘fit’. A set of HR policies is often said to
constitute an HR strategy if and only if two forms of ‘fit’ are achieved. First, the
different HR policies must ‘fit’ together sufficiently (have ‘internal fit’); second,
the HR policies must ‘fit’, or be congruent with, the firm’s policy choices outside
the area of HR (have ‘external fit’).
The notion of internal fit emphasises the interdependency between individual HR
policies and practices. Such fit exists if the use of one HR policy enhances the
effectiveness of the others. For example, relying on filling vacancies largely by
internal promotion is unlikely to be effective unless the firm also has
complementary training and development policies that have been well thought
out; the two policies support one another and ‘fit’ together. If a particular policy
works to reduce the effectiveness of the others there will be inconsistency and
negative internal fit. More specifically, effectiveness is to be understood as
enhancing either the competence or motivation levels of the workforce. So synergy
will be achieved (ie internal fit will exist) if the joint impact of a set of HR policies
on competence or motivation exceeds the sum of their individual impacts on
competence or motivation.
What this implies in turn is that the classification of HR strategies, at least as far as
internal fit is concerned, might rest on a theory of what motivates employees. It is
no accident that some modern writing talks about ‘high commitment management’,
ie the HR strategy that rests on the notion that it is employee commitment that
acts as the motivational ‘glue’ that binds many of the different policy elements
Chap2.p65 16/06/03, 15:309
together. We might therefore define one HR strategy as consisting of those HR
policies that are thought to enhance performance by enhancing employee
commitment. Other HR strategies might focus on enhancing motivation through
different transmission mechanisms. So one theoretical basis for specifying different
HR strategies might be management’s view of the relevant motivational ‘glues’.
The notions of fit can also be explored through the contribution of Delery and
Doty (1996) who argue that there are three ‘dominant modes of theorising’ that
characterise current empirical endeavours. They term the modes, the ‘universalistic’,
the ‘contingency’ and the ‘configurational’ and the underlying argument refers to
the importance of different notions of ‘fit’.
Researchers adopting a universalist perspective work from the presumption that a
specified set of HR practices will always produce superior results. This ‘best
practice’ approach is most clearly demonstrated in the works of Pfeffer (1994,
1998) and Osterman (1994). Although Pfeffer has not undertaken empirical work
in this area, his reviews of existing work in the field have led him to conclude that
certain specific practices when adopted will lead to higher business performance,
whatever the accompanying circumstances. Interestingly, his list of practices has
changed over time. In 1994 he suggested that there were 16 practices that would
raise performance, whereas by 1998 he had reduced this list to seven practices
(similar to the number explored by Delery and Doty, 1996, see below). Once
again, this was justified on the grounds of the growing body of empirical evidence.
This approach is basically saying that only internal fit matters, ie that ‘best practices’
work in all contexts.
The contingency perspective builds on the well-established field of contingency
theory, which argues that performance is optimised when there is ‘fit’ between an
organisation’s structure and its strategy. In the field of HR this framework has
been used to argue that an organisation’s set of HR practices must be consistent
with other organisational factors if it is to be effective (i.e. that there should be
external fit). This perspective has often looked at specific areas of HR such as
reward practices (see Gomez-Mejia and Balkin, 1992) and examined how they
vary in relation to different types of business strategy as well as how they influence
a firm’s performance. Perhaps the best example of this genre of studies is the work
by Schuler and Jackson (1987), which developed the Miles and Snow typology of
competitive product strategy to suggest how different market strategies should
shape the nature and range of HR practices that a firm should adopt.
This configurational mode of theorising draws heavily on developments in the
organisational theory and strategic management literatures. In the words of Delery
and Doty (1996) ‘configurational theories are concerned with how the pattern of
multiple independent variables is related to a dependent variable rather than how
individual independent variables are related to the dependent variable’ (their
emphasis). Another way to state this is to claim a strategy’s success turns on
combining ‘vertical’ or external fit and ‘horizontal’ or internal fit. This has
consolidated the potential importance of ‘bundles’ of HR practices, and MacDuffie
(1995), Arthur (1992) and Ichniowski et al (1997) have all explored the extent to
which combinations of practices can be determined and whether organisational
performance is related to the adoption of such ‘bundles’. They extend this work
by suggesting that certain ‘ideal types’ of employment system can be identified
that have characteristics that differentiate them from each other. The two systems
Chap2.p65 16/06/03, 15:3010
they identify are the ‘market-type’ system and the ‘internal system’. Both of these
systems, it is argued, will deliver higher levels of performance but a firm that is
stuck in the middle will by definition have lower levels of performance. Therefore,
a firm that has bundles of HR practices that most closely fit one or other of the
ideal type systems should have a higher level of performance, providing it also
achieves high levels of fit with competitive strategy.
Delery and Doty argue that each of these three theoretical perspectives is valuable
and that empirical work needs to be clear as to which framework is being deployed.
Ideally, they would like to see researchers apply the three frameworks in order to
evaluate their power. In their own empirical work (carried out in the US banking
sector, and looking at one occupational group) they found the strongest support
for the ‘universalistic’ model.
The argument so far is that it is difficult to develop a satisfactory taxonomy or
classification of HR strategies. It is certainly not theoretically adequate simply to
list different combinations of HR practices. This is in part because there is an
astronomical number of possible combinations of practices.
So the mere counting
of HR practices is not an adequate basis for defining an HR strategy.
A simple list of different combinations of practices is also inadequate because the
specification of interesting strategic choices requires a convincing rationale of how
and why a particular combination of practices generates internal and external fit.
The difficulty here, as so many writers emphasise, is that we are sadly short of
such a rationale. Guest and Hoque, for example, were faced with the problem of
assigning the firms in their sample to different strategic categories based on whether
they had certain combinations of 23 HRM practices; they recognised that some
of these practices might be more important than others, but they concluded that
‘we know of no theoretical basis on which to determine this’, so they treated each
of them as equally important (Guest and Hoque, 1994, p.6). MacDuffie observed
that every HR practice has a complex role and that ‘there is no clear conceptual
basis for separating practices affecting motivation from those affecting skill’
(MacDuffie, 1995, p.203). Even worse, the motivational impact of many individual
HR policies remains to be established.
This last point is very well illustrated in Wright and McMahan’s (1992) discussion
of the role of performance or incentive pay in characterising HR strategies which,
as they note, shows some of the dangers arising from a lack of theoretical care
when designing empirical research. They start by claiming that there is no theory
that can help researchers understand the role of incentive pay in HR strategy and
business performance research – this is perhaps too sweeping. But a lack of
theoretical rigour leads to (some forms of) incentive pay being seen in some studies
as central to ‘high commitment’ or ‘innovative’ HR systems, and in others as
integral to ‘control’ systems (ie low-skill, low-trust systems). An adequate theory
might indicate that a certain form of incentive pay is more likely to form part of a
‘control’ type system, whereas another type of incentive pay scheme would be
part of a ‘commitment’ system. However, it is also possible that incentive pay
schemes may be rewarding different behaviours and outcomes. For example, an
incentive pay scheme tied to innovation might be more appropriate for a business
pursuing a differentiation strategy, whereas a scheme tied to cost savings might
Chap2.p65 16/06/03, 15:3011
work for a business with a cost strategy. Somehow, these possibilities must be
reflected in the taxonomy of the strategies, which can be achieved only if the
theory is spelled out and fully captured in the measure of the strategy.
Researching HR strategies
Empirical and theoretical research into HR strategies might focus on three different
• It might provide a description of HR strategy choices. This involves building
up a taxonomy of possible or actual HR strategies, and then discovering which
are used, and with what frequency. It has already been suggested that this
requires a fair amount of prior theorising on (a) what, in general, is an HR
strategy, and (b) in what sense the different strategies are distinct from one
• It might provide an explanation for HR strategy choices. This addresses the
question of why different firms make the different HR strategy choices they
do (and includes the issue of why so many firms fail to pursue anything that
could reasonably be described as an HR strategy). Any attempt to explain
these choices relies on having solved the problem of properly describing and
enumerating the HR strategies, since explanation presupposes identification.
The explanations implicitly or explicitly have to draw on some general theory
of the firm (ie a general explanation of why firms make the kinds of decisions
they do), and might also distinguish between the strategy-making process and
the concrete form the strategy takes.
• It might provide an evaluation of HR strategies. There are two sub-questions
here. The first relates to whether the various HR strategies are successful, ie
to their impact on various measures of business performance. The second
(which is, in practice, much less frequently considered) seeks to understand
why a strategy is, or is not, successful. As with explaining strategy choices,
evaluating them presumes that different HR strategies have been clearly
identified. An evaluation might be done for an individual firm (eg has firm A’s
HR strategy been a success?), or more generally (eg has a particular HR strategy
typically been a success?). Under this heading one might also consider whether
there is a universally successful HR strategy, or whether success is highly
dependent on the circumstances in which a strategy is used.
The evaluation issue
All three questions are very difficult to answer in a satisfactory way. The main
focus in this paper is the third question, the evaluation of HR strategies. The
simpler of the two issues here is whether the pursuit of a specified strategy enhances
business performance, ie whether it causes or leads to business performance being
better than it would otherwise have been. At a minimum, this requires one first to
characterise or conceptualise the HR strategy, then to measure it, and finally, in
some way or other, to relate its use to a measure of business performance.
Chap2.p65 16/06/03, 15:3012
There are many research approaches one might use to do this, but all of them
require the researcher somehow to take into account the certainty that many things
in addition to the HR strategy will affect business performance. In other words, it
is necessary to isolate the impact of the HR strategy on performance from all the
other determinants. So one automatically has to develop a very complex evaluation
To understand these complexities more clearly, it may be useful to make a
comparison with the relatively simple problem of evaluating a single HR
instrument, for example, performance-related pay policies. Difficult as it is to
carry out authoritative evaluations of a single HR instrument, it is much harder to
evaluate whole strategies.
Developing a taxonomy of policies/strategies
Evaluation rests on being able to specify clearly the thing being evaluated. The
first step when evaluating performance-related pay arrangements is to describe,
or classify or provide a taxonomy of them. This is not a simple matter because
performance-related pay practices may assume a number of distinct forms along a
variety of dimensions (are they group or individual based, do they involve large or
small amounts of money, are the payments made in a lump sum or in smaller, more
continuous amounts, how old are they, etc?). Each form may have its own degree
of success, so one must evaluate them separately. But the number of interesting
dimensions is relatively limited, so workable taxonomies based on them can be
Performance-related pay taxonomies could also be based on other things; one
could, for example, classify performance-related pay arrangements on the basis of
what managers want (or say they want) to get out of them, or on the basis of what
sort of production arrangements firms chose. So there is more than one possible
taxonomic basis. But on any basis, any single policy area has a limited number of
dimensions; and the policy being classified is itself intrinsically easy to identify in
the first place.
As has been shown above, this is not true for HR strategies. There it was suggested
that one would like to know about the underlying employment relationship which
lay at the core of the strategy, as well as the links between chosen policies and that
relationship. Notions of internal and external fit would also be relevant.
The pragmatic alternative of specifying an HR strategy just on the basis of a
somewhat arbitrary collection of individual HR practices looks theoretically
suspect. The problem is, however, that the various employment relationships have
only been sketched, while the links between them and individual practices remain,
at best, somewhat speculative.
Conceptualising business performance and accounting for other
influences on performance
Any evaluation needs evaluation criteria, ie standards or metrics by which outcomes
of the process are to be assessed. As was discussed above, these criteria could be
relatively narrow HR measures, either of employee behaviour (like absenteeism
or quit rates) or of employee attitudes (like trust in management or employee
Chap2.p65 16/06/03, 15:3013
commitment). They could alternatively be broader objective measures of
performance (like labour productivity, profits or stock market prices). The former
could also be seen as the routes by which the latter were achieved (eg productivity
goes up because commitment has been enhanced), see Paauwe and Richardson
The broader objective measures are undoubtedly more impressive and authoritative
to many observers, and some of the US work that claims to establish a close
connection between HR strategies and profits or productivity has undoubtedly
made the whole area very sexy. But these indicators of success are relatively ‘far
away from’ the HR intervention, ie they are all very heavily influenced by a whole
range of non-HR factors. If these latter factors can be taken into explicit account
this may not be a problem, but this is not usually done adequately.
A growing number of studies have complex measures of HR practices. These are
often used in multi-variate analyses, which also incorporate background variables
like capital/labour ratios, firm size, industrial sector, and so on. But they only
rarely include other direct measures of managerial effectiveness. This omission
might mean that all aspects of managerial effectiveness are being represented just
by the HR variables. If so, the results are likely to overstate the contribution of
HR to firm success. One might imagine a parallel exercise to evaluate different
marketing strategies; in the absence of HR variables it might be concluded that a
particular marketing approach paid off handsomely. The reality might be that
successful firms tend to be effective in both HR and marketing – but if only one
policy area is included in the evaluation exercise it will seem that it alone is making
the contribution. What this implies is that even highly sophisticated studies may
be seriously misstating the impact of the policies they are trying to evaluate.
Further thoughts on the theory
In their review of the theoretical underpinnings of HR strategies, McMahan et al
(1999) posed the question as to whether the field was suffering from ‘paradigm
proliferation’ or was enjoying a growing convergence towards one theoretical
framework. It must be concluded that, so far, any convergence that is taking place
is grounded less in the security provided by rigorous theorising and more in
empirical work that is hard to replicate. Guest (1997) in his review of the area
argued strongly for the development of sharper theory in order to guide research
efforts. He felt that there was a need for ‘a theory of HRM, a theory about
performance, and a theory about how they are linked’. We strongly agree. His
own view is that expectancy theory might provide the vehicle to deliver such a
unifying theoretical framework, though others have not yet obviously accepted
One of the difficulties, but also perhaps one of the opportunities, within the field
of HR strategy is the fact that different researchers come from different disciplines.
So there is the prospect of a genuine multi-disciplinary approach developing. This
does not make things any easier, but it could eventually be very fruitful. McMahan
et al (1999) provide a very useful conceptual framework for understanding the
theoretical perspectives influencing the field. They identify a number of broad
theoretical approaches that have influenced empirical work in the area. These they
group into ‘reactive’ theories and ‘proactive’ theories. The proactive perspectives are:
Chap2.p65 16/06/03, 15:3014
• transaction costs theory/agency theory
• cybernetic systems
• resource based view of the firm.
The reactive theories they cite are:
• resource dependence
• institutional theory.
Their review of these various perspectives concludes that the field is still relatively
new; being in an adolescent phase, it does not have a well-established theoretical
or empirical base. They argue that much of the empirical work in the area is ‘data-
driven theory application’ rather than ‘theory-driven’ research.
The first of these is characterised by the rough application of a theoretical
framework to data. For example, it could be posited that bundles of HR practices
lead to higher business performance using the resource-based view of the firm as
a theoretical framework. Most researchers – their example is Huselid and Becker
(1996) – are looking for relationships between these bundles of HR practices and
organisational performance, without considering the intervening conditions that
may also be important in understanding how one set of variables affects others
(eg the importance of changing employee behaviours and attitudes). Thus, much
of the research using this approach (and constrained by specific types of data
collection techniques) fails to take explicit account of important intervening
variables. This failure is almost inevitable given the data demands of so formidable
a model, and it is a common feature of the existing empirical literature in the field.
It does mean, however, that the results from these partial studies have to be treated
with special caution.
One of the features of current empirical work is that researchers are seeking to
refine both the range and the nature of the measures they include in their models.
They are, for example, extending the range of variables in order to control for the
effects of other organisational factors on performance outcomes (eg investment
in technology) as well as trying to improve the precision and reliability of existing
measures (eg as with Huselid, the attempt to gauge policy coverage by asking
what proportion of the workforce is covered by particular policies). While these
pursuits are laudable and will certainly improve future data collection, there is a
risk that as the number of influencing variables grows the only performance that
is really being evaluated is that of the statistical programmes manipulating these
data. In other words, researchers will be forever adding to the contingencies that
could influence performance outcomes. This illustrates the limitations of a research
field without a rigorous theoretical framework.
McMahan et al (1999) advocate a more theory-driven approach to empirical work.
In practice this means giving more thought to the hypotheses that come from an
interesting theoretical framework and developing appropriate methodologies and
measures to test them. Huselid and Becker, on the other hand argue that theoretical
Chap2.p65 16/06/03, 15:3015
development is difficult because of the dearth of empirical research. Thus, there is
an emerging fissure between those who see the dynamic of the field being continued
through deductive methods and those who see it as developing through inductive
An important theoretical criticism made about much of the empirical work in the
area (and also some of the emerging paradigms) is that it fails to take account of
developments over time. Cross-sectional surveys and one-point-in-time case
studies have difficulties in taking proper account of the evolutionary nature of
organisational effectiveness and the relative power and influence of HR and other
managers within the organisation on the development of HR strategies. This points
to the relevance of longitudinal designs (such as the work of West and his colleagues
at Sheffield) and the importance of combining both quantitative and qualitative
However, longitudinal studies, in and of themselves, will not necessarily address
the theory deficit outlined above. Furthermore, longitudinal studies have their
own inherent problems. A longitudinal panel survey such as the one being
conducted by Huselid et al (1997) will not only suffer many of the problems of a
cross-sectional survey but there is also the problem of panel attrition, which reduces
the number of valid observations over time. In addition, measurement error can
be compounded significantly and problems of covariance magnified.
These are all central considerations for developing the IPD strategy in this area
and, for us, they underline the need to understand the competing theoretical models
and their implications for research design.
1 If there were only two HR practices (A and B) there would be four possible
combinations (both A and B, either A or B, and neither A nor B). In this sense
there would be four possible HR strategies. If there were, say, 10 HR practices,
there would be no fewer than 1,024 possible ‘strategies’.
Chap2.p65 16/06/03, 15:3016
The empirical literature
Our review of the empirical literature has thrown up a number of important
problems, some of them to do with the nature of the data that are collected, and
some to do with the way the data are used. Our discussion is divided into seven
Measuring HR practices
HR strategies are everywhere measured by HR policies and practices, although
theory is usually used (frequently in a relatively sketchy way) in the attempt to
justify for which policies data should be collected. Guest and Hoque, (1994), for
example, were interested in a particular HR strategy and thought that its presence
or absence could be captured in the responses to a set of 23 questions on specific
practices. There is, however, only a limited consensus among researchers on the
HR practices that should be included when doing this. The available studies include
different policy areas, in a way that is often confusing. Becker and Gerhart (1996)
looked at the actual practices included by researchers and found a surprisingly
low level of overlap in the practices being explored. Furthermore, the practices
that were common across the studies were often measured in quite different ways.
For example, one study may look at whether a business has performance pay (ie
yes or no) while another measures the proportion of employees covered by such
practices, and another looks at how much is typically paid.
This problem continues to be a serious one. We referred above to a paper by
Chadwick and Cappelli (1998)
which measures a particular HR style or strategy
(the ‘Investment’ relationship) by the following 10 questions:
The ‘Investment’ relationship
01 How many different benefits does the firm offer?
02 Does the firm provide/pay for computer training?
03 Does the firm provide/pay for cross-training?
04 What is the percentage of operatives involved in regularly scheduled meetings to
discuss work-related problems?
05 What is the employee/supervisor ratio?
06 What is the percentage of operatives currently involved in job rotation?
07 Does the firm have pay-for-skill as a part of its compensation package?
08 What is the percentage of operatives currently involved in self-managed teams?
09 Does the firm offer stock options to at least some employees?
10 Does the firm provide/pay for teamwork or problem-solving training?
Chap3.p65 16/06/03, 15:3117
Their second HR strategy (the ‘Contractual’ relationship) is measured by responses
to the following eight items.
The ‘Contractual’ relationship
1 An index of the relative generosity of the firm’s compensation
2 The importance of industry credentials as a hiring criterion
3 The importance of communication skills as hiring criterion for new non-supervisory
4 Does the firm have flexitime?
5 Does the firm allow job sharing?
6 What is the percentage of employees working part time?
7 What is the percentage of total labour costs spent on recruiting?
8 How many different selection methods are used?
Firms scoring high on the ‘Investment Employment Practices’ index are judged
to be more successful than are those using ‘Contractual Employment Practices’.
But the detail contained in each of the two strategies looks odd, and neither
precisely overlaps with that in, say MacDuffie’s excellent study (MacDuffie 1995).
In the latter, the strategy being examined is captured in two indices, each an amalgam
of a set of individual items.
The percentage of the workforce working in teams
The percentage of the workforce in EI or QC groups
The number of suggestions made per employee
The number of suggestions implemented per employee
The frequency of job rotation
The extent to which responsibility for quality control is decentralised
The extent to which the firm’s hiring criteria emphasise openness to learning and
An index measuring the degree to which an individual’s pay is influenced by the output
of the whole workforce
The degree to which status barriers are eliminated
The amount of training of new employees
The amount of training of experienced employees
Chap3.p65 16/06/03, 15:3118
Plants scoring well on both indices are found to have higher labour productivity.
Now, there is certainly some overlap between McDuffie’s specification and that
of Chadwick and Cappelli. But the overlap is not total, which means that the
components of a winning HR strategy have not been identified. The fact is that
different researchers continue to claim great success from rather different HR
. Anyone who is using the various studies to discover ‘best practice’
(whether contingent on specified circumstances or not) still faces a lot of
This is re-enforced by an important related point. Any single HR policy area is
measured in a variety of ways by the different studies. So, if HR managers want to
know whether they are wise to include in their HR strategy, say, some form of
performance-related pay, they might get some very general insight. If they also
wanted to know what precise form of performance-related pay to use (eg group-
based versus individual-based) they would get almost no help. But it cannot be
presumed that the two different forms are equivalent. So what has been captured
in any general finding on performance-related pay? The problem is that the
questions are not posed, and the possibilities are not explored, in sufficient detail.
So, to take a different example, there might be hints in the empirical work on HR
strategies that best practice would involve, say, spending more on selection; but
that does not answer the question – which precise selection techniques to use? In
other words, although there are, rightly, a lot of policy areas captured in each of
the HR strategies, none of the components is measured with as much detail as the
practitioner would like.
Three other general points are illustrated by the lists of items from the studies by
MacDuffie (1995) and Chadwick and Cappelli (1998):
• The precise groupings of the various items into different indices are often
puzzling and seem arbitrary; eg why are communication skills important in
Chadwick and Cappelli’s contractual system but not in their investment
• When constructing the indices to measure the different HR strategies,
researchers usually give each included item the same weight; so, say, eliminating
status barriers is seen to be as important as training arrangements. This would
strike many managers as extremely odd.
• Many indices or surveys simply omit certain HR policy areas; so, for example,
communication policies or interest representation practices are often
downplayed or ignored.
We have already discussed the theoretical difficulties when developing an HR
strategy taxonomy, and have said that the notion of fit is a critical building block.
One form of fit is external fit, or alignment with the wider business strategy, and
this in turn has at least two important dimensions. The first is the fit with the
firm’s product market strategy, eg does the firm emphasise quality or low prices
(see, eg Schuler and Jackson, 1987); the second is the fit with the firm’s choice of
production arrangements, eg is there an emphasis on getting quality right first
Chap3.p65 16/06/03, 15:3119
time or is there a wide use of rectification (see, eg MacDuffie 1995). The interest
of the IPD, and for that matter of most of the researchers in the area, is probably
the link to business performance and not strategy. However, an important
determinant of the HR practices being adopted may well be the influence of
competitive strategy and the ability of firms to develop linkages between strategy
and HR systems. It should also be noted that many of the models of business
strategy that have been used in the research up to now are thought to be outdated
and have their utility questioned within the wider field of strategic management
This suggests that research designs need to take account of developments in
thinking about competitive strategy. Here, notions of ‘dynamic capabilities’ and
resource-based view of the firm may be important to incorporate. This points to a
greater need for single-industry studies, with more time spent on developing
relevant measures of competitive strategy. In an attempt to respond to some of
these problems, Thompson (1998a) in his study of the aerospace sector developed
competitive strategy measures through a process of interviewing MDs and senior
managers in the sector about their business objectives. However, the ability to
capture strategic management objectives through cross-sectional questionnaires
typically completed by one respondent (often a personnel manager) does question
the reliability of such data. Although the Thompson survey was completed mostly
by managing directors or Board-level respondents, there still remain issues about
the extent to which strategic objectives can be measured. Indeed, the notion of
corporate strategy as ‘emergent’ and ‘a pattern in a stream of decisions’ does pose
problems for cross-sectional survey designs.
The reliable and accurate measurement of HR practices is one of the challenges
facing researchers in this area. The other is how to measure performance. We have
already raised the issue of how business performance is to be conceptualised, and
made the point that financial measures of performance are often quite distant
from many of the HR practices being explored.
Unit of analysis
One of the other problems that has dogged the area is the unit of analysis adopted
in survey design. A wide number of approaches have been used, and the following
breakdown illustrates some of the possibilities as well as examples of existing work
that fall into each category:
• Cross-sectional surveys (postal and telephone based)
multi-industry (eg Huselid, 1995; Osterman, 1994)
single-industry (eg Delery and Doty, 1996; Thompson, 1998; Guest and
Chap3.p65 16/06/03, 15:3120
• Case studies
multi-firm, single-industry (eg Arthur, 1992; MacDuffie, 1995; Ichniowski
et al, 1997)
• Mixed method
multi-industry cross-sectional survey and case studies
single-industry survey and case studies
longitudinal survey and case studies (eg West, 1996; Huselid et al, 1997).
Our review of the research in this area found that there was a broad range of
designs, techniques and approaches deployed. For example, Huselid (1995) wanted
to use financial databases to derive financial performance indicators as his outcome
measures. This meant that he needed to collect data on HR practices at the
corporate level (to ensure that the outcome and practices data both refer to the
same units). A widely recognised problem with this is that the data on HR practices
is not necessarily reliable because responses will often come from corporate head
offices and fail to recognise the diversity in practices across establishments within
multi-establishment firms. In the UK, the work by West et al at Sheffield has
concentrated on single-site companies in the manufacturing sector in order to
control for variation, and the work by Guest and Peccei (1994) in the NHS and
Thompson (1998b) in the UK aerospace industry both use an establishment
approach as well as gathering establishment-level performance measures.
Establishment-level surveys raise their own problems. Such surveys (eg Huselid,
Guest and Peccei, Thompson) tend to ask about employees in a very general way
(eg ‘please respond for your non-management employees’) and tend to obscure
the variations in jobs, skills and occupations that go to make up an employment
site. For example, the category ‘non-management employee’ could cover technical
and professional staff as well as semi-skilled workers. Firms may well be operating
different types of HR practice for these employee groups. To give an example,
engineers in the aerospace sector may be subject to performance pay as would
skilled shop-floor workers but the performance measures (ie competency for
engineers and output measures for shop-floor workers) may differ significantly
in their character and how they fit with the rest of the HR system. In the case of
engineers, competency-linked pay might be part of a new career development
structure (ie a dual career ladder for technical and managerial staff) whereas for
shop-floor employees the output measures might be part of a very flat job structure.
The pursuit of perfect measures can in itself be a problem, as any measure can be
open to criticism. In the case of ‘level of analysis’, there is a strong preference for
cross-sectional survey work to work at the lowest possible level (ie the
establishment) where more accurate data on HR practices can be captured. The
business unit rather than the company should be where data gathering takes place.
Chap3.p65 16/06/03, 15:3121
The need to understand rationale and process
One of the limitations of the cross-sectional survey technique that tends to
dominate work in this area is that it cannot tell us very much about the processes
that organisations use to develop HR strategies and systems. The survey technique
allows us to say what organisations are doing (though this too is open to criticism)
rather than how they achieved positive or negative outcomes. The issue of process
is becoming more central to the strategic management agenda as the ‘competences’
of firms in areas such as ‘change management’ are increasingly being seen as
differentiating organisational performance. From this scenario it is perfectly
plausible that a survey could identify two firms in the same industry with the
same HR practices but with different performance levels. The explanation for the
success of one organisation over the other may be more to do with its ability
successfully to design and implement the appropriate practices. In other words, it
is the firm’s management capabilities that are the source of competitive advantage,
not the HR practices it introduces. This perspective has important implications
for the HR function. It suggests that not only do HR managers need to display
strategic capabilities in order to work through the implications of new strategic
management decisions for HR but also that they need to have technical capabilities
in the design and implementation of new practices.
This area has been partly addressed by Huselid, Jackson and Schuler (1997) in
their study on the technical and strategic capabilities within the HR function.
They found that firms that had high levels of strategic capability in the HR function
tended to have higher levels of financial performance.
The need to understand process and rationale is a point taken up by Marchington
and Sparrow (1998). They argue that cross-sectional work can arrive at spurious
conclusions about what leads to organisational effectiveness, particularly if there
is lack of theory informing the research. They argue for more attention to be
given to process and suggest that ‘we need to map out the process through which
the HRM contribution links to organisational performance’. This leads us to the
The research field is currently dominated by positivistic research designs that only
partly succeed in capturing managerial views and attitudes about the adoption of
new HR practices and their relationship to improved business performance. These
designs can help us understand the correlations between HR practices and
performance outcomes (close or distant) as well as estimating the size effects of
these relationships. They do not help us understand the factors that have influenced
firms to adopt these practices (or even whether they are introduced as ‘bundles’)
or indeed, the other organisational changes that have been introduced alongside
these new HR practices. Similarly, the emphasis in the literature on demonstrating
that the greater use of ‘high-performance’ practices does indeed lead to higher
business performance often prohibits an examination of examples of where the
introduction of new HR practices has failed. Understanding the factors that are
associated with such failures is probably as important as understanding stories of
Chap3.p65 16/06/03, 15:3122
Indeed, cross-sectional and random survey designs may lead to an
overrepresentation of successful cases and an underrepresentation of failures
(Becker and Gerhart 1996). This imbalance can often lead to a misunderstanding
of the factors that may be influencing the successful adoption of HR practices
and their influence on business performance. In other words, the reason why
organisations introduce (or fail to introduce) HR practices that academic research
points to as beneficial to business performance, and that employer bodies (eg CBI
and IPD) are increasingly advocating, tends to get overlooked. Furthermore, these
research designs are not good at picking up the process dimensions of introducing
HR practices which may themselves be an important determinant of successful
To illustrate the importance of design and implementation issues to business
performance we go back to the issue of performance pay. In an important
longitudinal case study of incentive pay schemes in UK manufacturing firms in
the late 1970s and early 1980s, Bowey and Thorpe (1986) found that the success
of schemes could be attributed more to the processes associated with the design
and implementation of incentive pay arrangements than the presence of a scheme
itself. In other words, higher business performance (as measured by employee
productivity) was not the result of adopting a particular incentive scheme, but the
way in which it was adopted and implemented – it’s not what you do, it’s the way
that you do it that counts.
To extend this example, a cross-sectional survey would pick up firm A and firm B,
both of which have adopted the same incentive scheme. However, firm A has low
relative organisational performance despite using, on the surface, the same incentive
scheme. We may be able to explain this by other ‘contingency’ factors covered in
a survey, but it may also be because of much softer process issues that the survey
approach cannot grasp.
And herein lies one of the great risks of the research endeavour on HR practices
and business performance. Policy and employer bodies may advocate the adoption
of so called ‘best practices’ in HR without understanding the complementary
circumstances and factors that are needed for the optimal performance outcomes
for the business. Managers acting upon such guidance may find themselves
introducing new HR practices without fully understanding the risks involved or
even if the practices are appropriate, given their organisational circumstances.
This problem was recognised by Becker and Gerhart (1996) in their review of the
research in the field when they concluded that ‘there appears to be no best practice
magic bullet short of organising a firm’s HR system from a strategic perspective’.
What we now need to understand is ‘how’ firms achieve this strategic approach to
the management of the HR system and what internal process factors account for
success or failure. This suggests that a wider set of research techniques may need
to be adopted and that future studies should consider more interdisciplinary
research designs and methods. For example, Marchington and Sparrow (1998)
allude to various psychological methods that can be used to explore managerial
decision-making (eg cognitive mapping, scripts, schema) which may help
researchers understand the causal linkages and paths that explain why organisations
adopt, or fail to adopt new HR practices.
Chap3.p65 16/06/03, 15:3123
Through using such methods (in combination with other econometric methods)
researchers might be able to address more confidently the extent to which managers
consciously and deliberately introduce special ‘bundles’ of HR practices or whether
they think that organisational performance is going to be improved by such
approaches. At this point in time, we have less than satisfactory answers to these
Intermediate measures of performance outcomes:
employee motivation and commitment
The point was made earlier in the report that much of the empirical work
(particularly the pioneering studies in the US) has tended to use company-level
financial performance data and mix them with establishment-level measures of
HR practices. Such an approach raises important questions about causality and
the extent to which other organisational factors influence these performance
outcome measures (eg currency hedging, acquisitions, technology). These measures
are too distant from the ‘scene of the action’.
Although there have been some detailed case studies of industries and companies
that have sought to limit these methodological problems, little if any of the current
literature looks at both strategic fit ideas as well as performance outcomes in terms
of employee motivation and commitment. The only study (that we are currently
aware of) that has attempted to combine both measures of organisational
performance, measures of strategy and measures of impact on motivation and
commitment is the work of Patterson et al (1997) at Sheffield. Their work
concentrates on SMEs in the manufacturing sector and found that 18 per cent of
the variation between companies in productivity could be accounted for by certain
HR practices (namely, job design and skill acquisition and development). The
Sheffield study, although under way for several years, has not reported any
subsequent findings and there is little detailed information available on
methodology and measures used.
1 Our reviews of this paper and the MacDuffie paper which form the basis of
the discussion in this section are given in the appendix. A number of other
papers are also reviewed.
2 The extent to which such findings give support to the emerging ‘bundles’
orthodoxy is also open to question given the observed lack of theory in the
Chap3.p65 16/06/03, 15:3124
An important part of the current review involved consulting with practitioners
and involving them in the development of the next stage of the IPD research
programme. A number of events have taken place, including a one-day event hosted
by the IPD in London in February, 1999, which brought together senior HR
managers and academics. This chapter aims to bring together the main themes and
issues arising from this meeting, as well as separate discussions held with
practitioners across the UK co-ordinated by the IPD. The focus is on practitioner
views that are relevant to HR strategies and business performance, and not
necessarily on the wide range of views expressed. In that sense the task has been
to narrow down the contributions to reflect the on-going interest in the relationship
between HR strategies and business performance, but not to the extent of excluding
relevant and pertinent research questions – a difficult balance.
It was clear from the discussions that practitioners are not generally aware of the
growing body of evidence on HR strategies and business performance. While HR
directors in larger international businesses tend to have a more informed view, the
bulk of the HR profession could be said to be reasonably uninformed. This is, in
large part, due to the highly quantitative nature of the existing work (which makes
severe demands on the reader and also limits its wider appeal) and the fact that
most published research is of US origin (and both more difficult to access and
relate to). In meetings with practitioners in the IPD regions it was clear that few
had read or heard of even the more accessible work of writers such as Pfeffer
(either Competitive Advantage through People or The Human Equation). Not many
were aware of the Sheffield study or articles published in People Management on
This is not to say that practitioners are unaware of the importance of these issues,
just that they seem rarely to read research-based evidence. Furthermore, as the
discussions with practitioners revealed, even if they were aware, many are sceptical
of such research and remain to be convinced of its merits. This raises a number of
questions about how knowledge and learning takes place among senior managers
and how research findings can be effectively diffused. It has been suggested that
social learning processes such as forums, seminars and short programmes where
reading is minimal may be of more relevance
. These are clearly important concerns
for the IPD and third-level institutions providing CPD; however, in the context
of this research review they are of secondary importance. We are interested in
understanding the practitioners’ agenda – what do they feel they need to know in
order to make HR strategies deliver better bottom line performance. From this
perspective, practitioner interest tends to be more heavily weighted in favour of
process-based research. In other words, understanding how organisations develop
HR strategies, implement them and how this works right the way through to
Chap4.p65 16/06/03, 15:3125
business performance, rather than on establishing correlations between measures
of practices and measures of organisational performance.
To them, the argument that HR makes an impact on the bottom line is not in
dispute, their interest is in knowing the best means to make an impact. These
interests have clear ramifications for preferred research designs and suggest that a
greater emphasis be placed on longitudinal case studies that explore the processes
of HR strategies, organisational change and performance over time.
A common theme from the practitioner discussions was that the process of
developing HR strategies differs considerably, depending on a number of specific
contextual factors. Indeed, although the London meeting did not seek to develop
data that could be content analysed as a step towards understanding the relative
weight of these various factors, the issues that did emerge are important in shaping
our thinking about some of the terrain that process-based research should encompass.
Political coalitions and HR capabilities
An important dimension in the development and implementation of HR strategies
that can improve business performance was the significance of relationships with
other management functions or groups. A number of participants stressed the
importance of developing coalitions of common interest across management areas
in order to provide the bedrock on which HR strategies could be developed. In
some cases this was explicit, as with Siemens – where HR managers are building
cross-functional management capabilities because the business recognises that old
structures do not deliver system-wide solutions. In other contexts, it was recognised
that the lack of coalitions, and in particular the ability of HR to be a player in such
networks, could determine the success or failure of HR strategies.
The importance of building partnerships with business leaders was also identified
as a key attribute, and this opened up a range of issues around the skills,
competencies and behaviours that are required of senior HR managers to shape
and develop effective relationships at this level of the business. These sets of issues
link very strongly to the work of Ulrich (1998) and others on the changing nature
of HR and the new skills and abilities now being demanded. Although Ulrich’s
work can be criticised on a number of grounds, particularly its strong prescriptive
tenor and the lack of rigorous empirical evidence to support his models of the
new HR role, it provides one of the few insights into these areas. Clearly, there is
a demand for research that explores these dimensions in the UK context.
‘Good’ v ‘bad’ processes
Another important theme was the importance in understanding what processes
were most important for ensuring the effective development and implementation
of HR strategies. Managers recognised that not only did they need some way of
determining which HR strategies to prioritise and sequence, but they also needed
to understand what design and implementation processes were the best ones to
adopt. These concerns were at the level of system-wide HR strategies as well as at
the level of single practices (such as payment system design).
Chap4.p65 16/06/03, 15:3126
Business performance models
A strong and recurrent theme in discussions was the importance of wider business
performance metrics and methodologies. Two in particular were mentioned. First,
a growing number of organisations are using the Business Excellence Model (BEM)
developed by the European Foundation for Quality Management (EFQM) to
understand how a wide range of business processes impact on performance. This
model is built explicitly on the assumption that people processes account for 12
per cent of total business performance processes.
The model (and the audit tool that accompanies it) is helpful in encouraging
organisations to develop a broad range of metrics to understand the impact of
various business processes. However, the criticisms that practitioners levelled at
the extant research on HR strategies and business performance also hold true for
this model. It does not help to resolve practitioner concerns about which HR
strategies and processes should be adopted in which contexts and why.
A second framework that is gaining currency, and also includes an element of
people measurement, is the Balanced Scorecard (Kaplan and Norton 1996). This
management control process requires the development of a number of performance
indices across various dimensions of business performance. The idea is that the
senior management team is forced to manage the organisation using a much broader
set of performance indicators such as customer and employee satisfaction rather
than the traditional narrow set of primarily financial metrics. Given the growing
numbers of organisations adopting this methodology, practitioners felt that the
Balanced Scorecard could be developed to include a richer set of HR metrics.
Practitioners, particularly in larger private and public sector organisations, are
increasingly exposed to these business models (as well as others such as Economic
Value Added, which is being widely deployed in the manufacturing sector) and
this raises issues for them in terms of what HR metrics they should be developing
to link with these models.
Two major constraints face senior HR managers — time and resources. These
featured prominently in debates, and practitioners wanted methods or criteria for
making decisions on the allocation of time and resources to various HR strategies.
Their issues were not so much which practices they might want to adopt (ie.
Pfeffer’s seven practices) but understanding which ones were best in their own
particular organisational context. The discussions revealed that practitioners
generally worked very much from a contingency or configurational perspective.
Furthermore, they also wanted to know if there was a sequence that should be
followed — did they need to introduce a certain set of changes prior to putting in
place a new HR strategy, or was there a pattern to the introduction of HR strategies?
There is a risk, as became apparent in some of the group discussions at the London
seminar, that practitioners may still be thinking in terms of discrete practices (such
as performance pay) rather than in terms of HR systems. Understanding what an
HR system consists of and how it can be developed to meet business needs, is a
Chap4.p65 16/06/03, 15:3127
There was considerable interest in the bundles approach, which had considerable
intuitive appeal to practitioners, and raised questions about what particular bundles
of HR practices are relevant in different organisational settings. Bundles pointed
to the possibility of HR systems with distinctive properties, which some of the
research evidence that we have reviewed is beginning to indicate. This raised further
issues as to whether industry-level bundles could be differentiated, pointing to
the need for the possible development of sector-based research designs, such as
the work by Thompson in Aerospace and that of Guest and Peccei in health. In
other words, can we talk of broad based-industry HR practices (which might appear
contrary to the notion of SHRM) rather than organisational-specific bundles of
Practitioner interests pointed to the need to develop heuristic devices or diagnostic
tools that could help them make resource allocation decisions at key times in the
Time-based element of change
While the need to make fast and informed decisions in the context of rapidly
changing business environments is recognised as one of the key challenges,
practitioners also recognised that there is a time-based element to the development,
introduction and acceptance of new HR strategies in an organisation. For example,
Peugeot reckoned it took nearly five years for their communication processes to
become embedded and start to pay returns (based on employee attitude survey
This time lag between policy deployment and impact becomes an important tension
in businesses that are looking for ‘quick wins’, and practitioners felt they needed
some rules of thumb to guide them on the time-period for HR investments to
start making a payback.
The importance of intermediate measures
A fair amount of scepticism was apparent concerning studies that sought to relate
HR practices to company financial performance outcomes. Practitioners recognised
that many other factors complicated this relationship, such as product strategy
and technology. Many felt that it made more sense to look specifically at those
intermediate measures that are most meaningful in terms of employee performance
(motivation, commitment, absence, turnover, etc).
However, it was also recognised that these intermediate measure needed to be
taken seriously at board level, and some work needed to be done on developing
metrics that had ‘legs’ at that level.
If the use of more ‘innovative’ HR strategies is regarded as important in explaining
variations in business performance between firms, why are more organisations
not adopting these practices? It was felt that we needed to understand why
organisations failed to adopt high performance work practices and HR systems.
Chap4.p65 16/06/03, 15:3128
By identifying the barriers to the diffusion and uptake of practices we should be
in a better position to make HR practitioners aware of the nature of the hurdles
that they may face in trying to introduce new HR approaches.
This review of the range of practitioner interests and needs underlines the need
for a new departure in the research that currently exists on HR strategies and
business performance. The demand is clearly for process-based research and detailed
case studies that track the process of change in organisations (perhaps something
akin to the study by Pettigrew on ICI) but which can also lead to the development
of learning tools that can help practitioners to make better decisions about HR
investments and also be more aware of the problems and issues likely to arise.
1 Management academics and HR professionals at a meeting of the steering
group expressed these views.
Chap4.p65 16/06/03, 15:3129
At this stage we should like to flag up a number of points for further consideration.
In settling on these we are concerned with the interests not of the academic
community but of the practitioner community keen to draw guidance from the
frontier work being done by academic researchers. Such people, either directly or
via their professional associations like the IPD, are trying to learn what they can
do to improve the performance of their organisations. In this respect, the IPD
faces a major task in disseminating the existing research findings in a form that
can be properly appreciated by its members. Our impression was that the
practitioners in the seminars were keen to be involved.
From what we have already said in this report we would highlight the following
1 The claims that a universal best practice HR strategy has been identified are
premature. Managers should not en masse be discarding their present
arrangements. This is not to say that the studies on which such claims are
based are without interest. Quite the contrary, and they give a checklist of
policy combinations which organisations should very actively consider
adopting. But there are so many difficult problems still to be resolved that
they should not be considered to be a sufficient basis for policy decisions.
2 It is unlikely that merely adopting a specified set of HR policies is the high
road to organisational success. From what has already been said, even the large
amount of empirical work that exists at the moment has not identified all the
general components that such a set would contain; still less have the precise
policy variants been identified. We are not clear, for example, what the optimal
training arrangements, or selection arrangements, or reward systems, or
teamworking arrangements look like. Still less do we know what the optimal
combinations of these and other things are.
3 Even if we did have a clear idea of the optimal combinations, we should still
have to resolve the formidable problem of whether they were the same for all
organisations (as the universalist approach claims) or differed according to
different underlying motivational links or wider business strategies. It would
be rash to conclude, just because the weight of current evidence seems to
support a universalist interpretation, that policy advice should move heavily
in that direction too.
4 Virtually all the present statistical analysis, no matter how sophisticated it
becomes, operationally identifies HR strategies by adding up a mixture of
items from a somewhat arbitrary list of HR policies and practices. This might
Chap5.p65 16/06/03, 15:3230
be done in a very rough and ready fashion (eg Guest and Hoque’s approach of
counting as HRM the use of any majority combination of 23 practices) or in a
technically very advanced way (eg Huselid’s use of factor analysis, or Delery
and Doty’s use of cluster analysis). At some point, therefore, a strategy is
defined and measured by a statistical procedure rather than by a principled
argument grounded in awkward reality. We need to go beyond this and start to
rely on identifying strategies by their intrinsic theoretical and real world
5 We need to turn from so heavy a concentration on analysing the two end points
of evaluation, ie from associating ‘strategy’ with ‘performance’. We should
pay more attention to the intervening steps, looking at what mediates these
two variables. We should look more closely at why a particular combination of
policies might work, and how much discretion managers might have in their
ability to affect these mediating variables.
6 We should pay much more attention to how clusters of HR policies are adopted
and implemented, as well as to the specific contexts in which policy innovation
is attempted. We have all heard simple recipes for salvation, whether commercial
or personal. We have all witnessed policy fads. Surely we should have learned
that how something is done is often more important that what is done – existing
empirical studies concentrate almost entirely on what is done.
Implications for future research
Our review of the literature and discussions with academics and practitioners point
to a number of important research gaps that may attract IPD funding. An
interesting feature of the academic and practitioner perspectives is the growing
convergence between both. The research agenda being developed by researchers
is beginning to reflect many of the themes raised by practitioners. This is
particularly evident in the concentration on process and, by implication, the
preference for more detailed, process-based case study designs. However, it should
be stressed that the field as a whole suffers from a lack of theoretical rigour that
can help shape effective research designs and also serve to underpin the transfer of
knowledge to practitioners. Without a sound theoretical basis, it will be difficult
to translate research into learning materials and processes of benefit to IPD
members. This is a crucial area of research and one that is central to the future
Recognising the importance of theory as a first step, the potential areas for future
research are (and this list is neither exhaustive or in sufficient detail):
• understanding the HR strategy formulation process and the methods that
organisations use to prioritise and schedule HR investments
• understanding the perceptions of non-HR managers in this strategy-forming
process and how this influences outcomes
• mapping out the linkages in the chain from HR strategies through to business
performance (and the factors influencing these key linkages)
Chap5.p65 16/06/03, 15:3231
• developing intermediate-level performance metrics that are valid measures of
HR strategies impact and which are also taken seriously by senior business
• understanding the extent to which bundles of HR practices work at an industry/
sector level and the extent to which they are determined by firm-specific
• connected to this, to what extent are bundles mediated by occupational category
(ie is the bundle for an IT worker different from that of a customer service
worker in the same organisation)
• how important is time lag in terms of the design and introduction of new HR
strategies as well as their impact on organisational effectiveness
• and, connected to this point, how important are HR capabilities in designing
and implementing HR strategies (that may be common across businesses) in
ensuring that they have maximum impact on employee and organisational
These are just some of the potential areas to which we feel resources could be
directed to meet both practitioner and academic needs. These, of course, raise a
number of linked methodological issues and here, too, we find considerable
convergence of the practitioner and academic agendas. The emphasis on
understanding process and the ‘how’ of effectiveness clearly lends itself to more
case study designs. Furthermore, the need to broaden out the range of factors that
may shape and influence successful HR strategies suggests a need for greater
interdisciplinary designs, and lastly, the recognition of the time-based element
between strategy implementation and impact forces us to consider more
longitudinal studies (such as the Sheffield programme).
Given the growing convergence of practitioner and academic agendas, it may be
important for the IPD to consider how it can encourage greater opportunities for
debate and discussion on these important areas. This may be through the greater
involvement of practitioners not only in the development of research programmes
but also through participation on project steering groups.
While research is important, it also needs to be recognised that the majority of the
IPD practitioner base are mostly unaware of its existence and relevance. This
suggests that considerable attention needs to be given to the diffusion of knowledge
and research and the ways in which senior HR managers, in particular, learn.
Chap5.p65 16/06/03, 15:3232
Detailed Reviews of Key texts
MACDUFFIE J. (1995) ‘Human resource bundles and manufacturing
performance: organizational logic and flexible production systems in the world
auto industry’. Industrial and Labor Relations Review. Vol. 48, No. 2. pp197–
Broad theoretical background and basic hypotheses
MacDuffie argues that ‘innovative’ HR practices (ie those usually associated with
models of HRM) are likely to work most powerfully when all of the following
conditions are met:
• employees possess knowledge and skills that managers lack
• employees are motivated to apply their knowledge and skills through
discretionary effort, and
• the firm’s production strategy can be achieved only when such discretionary
effort is forthcoming.
Another way to express these propositions is to say that innovative HR practices
are the way to secure employee commitment to the organisation. Discretionary
effort is seen to be a consequence of commitment, and harnessing the otherwise
hidden skills and knowledge possessed by employees is seen to be most easily
accomplished when employees are committed. In this sense, an important focus
for MacDuffie is commitment-focused HR policies.
In addition to that, however, he wishes to investigate explicitly whether there is a
synergy among HR practices, ie whether the performance consequences of a
carefully specified set or ‘bundle’ of practices is greater than the sum of their
individual consequences. So his basic purpose is to estimate the effect of a particular
bundle of innovative HR practices on two measures of organisational performance.
That bundle is said to exhibit both external fit (here, in the sense that it is coherent
with the production system) and internal fit.
MacDuffie can therefore be interpreted as investigating the merits of a coherent
HR strategy fundamentally focused on securing both high employee commitment
to the organisation and high levels of employee competence.
Context and sample details
The propositions are tested in the particular context of the auto industry.
Production there can be organised either under a ‘mass production’ system or
‘flexibly’. A flexible approach is defined, in comparison with the traditional mass
production system, as arrangements which reduce inventories and other ‘buffers’
(like rectification areas), which increase interdependencies within the production
App1.p65 16/06/03, 15:2833
process, and which highlight production problems and therefore stimulate the
search for improvements. In line with his core argument, MacDuffie suggests that
this is precisely the context within which innovative HR practices can work most
effectively. His hypothesis is that innovative HR has a large potential in flexible
systems (and, presumably, little or no payoff in traditional mass production
systems). The data come from a survey of, and field visits to, 70 plants in the
world’s auto industry.
Basic analytical procedure
MacDuffie’s basic analytical procedure is to estimate a production function;
specifically this means regressing a set of general explanatory variables on two
measures of firm performance (ie labour productivity and product quality), and
then adding three more variables which are designed to capture the hypotheses of
special interest for the particular hypotheses of the study. These three variables are:
• ‘buffers’, an index of the degree to which a plant adopts the flexible production
• ‘work system’, an index of work arrangements said to be complementary to
the flexible approach
• ‘HRM policies’, an index of HR practices said to be complementary to the
Of particular interest for the synergy argument is whether, in combination, these
three variables have a larger effect on performance than when they are entered as
Our interest is mostly on the two indices ‘work system’ and ‘HRM policies’. The
components of these are shown in Table A1. In spite of their labels, both variables
are essentially HR variables. Thus, all of the six items which make up ‘work system’
would normally be thought of as reflecting HR considerations (cf the items in
Guest and Hoque). This is reinforced by MacDuffie’s claim that the first four will
affect employee motivation and commitment (p.203); why the other two might
not have the same effect is not made clear.
MacDuffie does not make completely clear his basis for choosing the five ‘HRM
policy’ items except that in his field work these were seen to differentiate the two
main types of production systems. This index is therefore inductively generated
and omits important areas of HR policy. There is, for example, no mention of job
security or communication policies. There is also arbitrariness in the precise way
in which the HR policies are measured and calibrated (see the notes to Table A1).
Also, the five items are given equal weight in the overall index, eg in their effect on
performance, policies on training and pay are seen to be no more important than
that on status differentials, so the former are seen to have no stronger an impact
on motivation or commitment or capabilities.
App1.p65 16/06/03, 15:2834
The two indices are a product not of factor analysis (the usual statistical technique
in this area) but of cluster analysis. MacDuffie’s use of factor analysis yielded
factors that were combinations of items from both ‘work system’ and ‘HRM
policies’. He does not provide full details of the cluster analysis, so it is hard to
know how reliable his two variables are.
Table A1 Mean values for 11 items, broken down by production
% work force in teams
% work force in EI or QC groups
Suggestions made per employee
% suggestions implemented
Decentralisation of quality control
Training of new employees
Training of experienced employees
Notes to Table A 1
Job rotation: 0 = none; 1 = infrequent rotation within teams; 2 = frequent rotation
within teams; 3 = frequent rotation within and across teams in same department; 4 =
frequent rotation within and across teams and across departments.
Decentralisation of quality control: 0 = functional specialists responsible for quality; 1,
2, 3, and 4 depending on how many of inspection of incoming parts, work-in-process,
finished products, gathering SPC data, production workers responsible for.
Hiring criteria: low scores for criteria which emphasise fit between applicant’s existing
skills and the job requirements (eg previous experience in a similar job), and high scores
for criteria which emphasise openness to learning and interpersonal skills. The metric
used is not specified.
Contingent pay: 0 = none; 1 = pay contingent on corporate performance; 2 = pay
contingent on plant performance, for managers only; 3 = pay contingent on plant
performance or skills acquired, for production employees only; 4 = pay contingent on
plant performance, all employees.
App1.p65 16/06/03, 15:2835
Status differentiation: 0 = no attempt to break down status barriers; 1, 2, 3 or 4
depending on how many of common uniform, common cafeteria, common parking,
and no ties are implemented.
Training of new employees (ie in first six months): 0 = less than one week; 1 = one to
two weeks; 2 = two to four weeks; 3 = more than four weeks.
Training of experienced employees: 0 = average of 0-20 hours per year; 1 = 21-40
hours; 2 = 41-80 hours; 3 = over 80 hours.
Table A1 shows that Flexibility is associated with higher values for all of the items
in ‘work system’ and ‘HRM policies’ than is mass production, and that this seems
to be especially true for ‘work system’. This is not surprising because these items
have been chosen precisely because they differentiate between the two systems.
MacDuffie also reports that the average values of both productivity and quality
are much higher for the flexible plants.
The formal multiple regression analysis supports these simple findings. First,
‘buffer’, ‘work system’ and ‘HRM policies’ each works to increase both
productivity and quality; second, synergy seems to be present, in that an overall
index combining all three elements has a more powerful effect on both dependent
On the face of it, the results suggest that, other things being equal, the buffered
system, ‘work system’ and ‘HRM policies’ all tend to raise productivity and quality
levels when used in isolation. This means that (eg) ‘HRM policies’ are associated
with higher productivity in both mass production and flexible systems (and there
is no evidence that the marginal effect would be different between the two systems).
In that sense, one of MacDuffie’s propositions, that innovative HR practices are
useful only in certain circumstances, seems to be rejected. This seems to be a
universalistic conclusion, ie the policies have some beneficial effect whatever the
But the ‘fit’ argument seems to be strongly supported, in that the performance
consequences of using ‘buffers’ together with ‘work system’ and ‘HRM policies’
seem to be especially notable. So using all three components together does seem
to have a distinctively powerful effect.
Doubts and queries
MacDuffie does not attempt to measure levels of either employee commitment or
employee competence in his plants. He does not, therefore, know whether they
are higher in Flexible plants, and so cannot be sure that he has identified the correct
channels through which the HR strategy might work.
MacDuffie does not report collecting data on a large number of HR variables, or
on different ways of trying to measure any one of them, and then of finding that
only some have a powerful effect. So it is by no means clear that the 11 items he
focuses on are indeed necessary and sufficient for performance to be enhanced.
He does not report, for example, the results of dropping any of his 11 items to see
App1.p65 16/06/03, 15:2836
whether the impact on productivity or quality would be materially different. It is
even less clear that any of the 11 are best captured in the way he does, eg his
contingency pay measure includes nothing on pay that depends on collective
performance at a more disaggregated level than that of the plant, even though
many firms pay according to, say, work group. So the precise policy implications
for firms remain in some doubt.
The conceptual basis for having the two variables, ‘work system’ and ‘HRM
policies’, is still fragile. MacDuffie rejects the argument that these primarily reflect
‘skill/knowledge’ and ‘motivation/commitment’ respectively, saying that many
items in either variable will have an effect on both of these influences. He dismisses
the appropriateness of factor analysis to discover meaningful bundles when he
says that that technique ‘is best suited to identifying the interrelationships among
a set of items in a scale, all designed to measure the same construct’ (p.204); he is
therefore saying that there are no such constructs in play here.
His preferred approach is to use cluster analysis, which he says is appropriate
when constructing ‘an index from interrelated items, each of which represents a
different construct’ (p.204). So MacDuffie implies that there is no underlying
construct embedded in either ‘work systems’ or ‘HRM policies’; rather, in each
case, there is just an index of interrelated items. This is surely an odd position.
Most would suggest that these variables would have an underlying theoretical unity
that went beyond mere statistical interrelatedness.
There is another problem, and it is very important. All that MacDuffie’s cluster
analysis achieves is to distinguish between different production systems by the
degree to which they use his 11 items (see Table A 1). The cluster is therefore
defined on the production system, not on the work arrangements and HR policies.
By itself, it does not allow him to bundle the 11 items into a smaller number of
variables. The conceptual basis on which this reduction was done is not explained.
So how MacDuffie moves from 11 items to two variables is quite unclear. His
factor analysis evidently gives a different assignment from the one presented in
the published study. So we are some way short of a convincing specification for
these influences. The fact that he subsequently carries out reliability analysis is
not terribly useful here. Many different combinations of the 11 items might well
have acceptable reliability scores; if this were the case, we should still have the
problem of deciding which bundles were best, and why.
USELID M. (1995) ‘The impact of HRM practices on turnover, productivity
and corporate performance’. Academy of Management Journal. Vol.38, No 3.
This study draws upon the resource-based view of competitive advantage. This
theory posits that firm-specific competitive advantage arises when human resources
are defined by the following criteria:
App1.p65 16/06/03, 15:2837
• must add value
• skills must be rare
• skills cannot easily be imitated
• skills cannot easily be substituted by technology.
HR practices can influence each of these areas. Therefore, in order to assess the
impact of innovative HR practices any study would need to develop appropriate
measures in these domains.
Huselid contends that the approach in his article has three novel dimensions:
• He looks at the ‘system’ of HR in the firm rather than individual practices
(seen as a more strategic perspective).
• The outcome measures used include both ‘intermediary measures’ (ie absence
and turnover) as well as financial performance outcomes. He argues that the
intermediary measures provide a check against criticisms of reverse causality
(ie successful companies are able to introduce innovative practices because
they are successful).
• The data permit an assessment of both ‘internal’ and ‘external’ fit in HR
practices and competitive strategy.
This is seen as limited because of the range of HR practices covered (partial and
not consistent), lack of consistency in the dependent variables used and the
industries explored (some are multi-industry and others single industry studies).
Huselid also questions the linkages between sets of dependent variables – for
example, the link between increased productivity and profits is not a ‘given’ nor is
the link between profitability and productivity a ‘given’. Other independent factors
can explain variations in these measures. In order to get round the partial nature
of measures used in prior work, Huselid seeks to develop a fuller range of measures
including intermediate and financial performance
He deploys Delaney, Lewin and Ichniowski’s (1989) 10 measures and adds three
more. These measures are organised into two broad categories: employee skills
and organisational structures, and employee motivation. The questions used are
in the tint box below:
App1.p65 16/06/03, 15:2838
Employee skills and organisational structures
What is the proportion of the workforce:
• who are included in a formal information sharing programme (eg a
• whose job has been subjected to a formal job analysis
• who are administered attitude surveys on a regular basis
• who participate in a QWL programmme, QCs and/or labour-management
• who have access to company incentive schemes, profit-sharing and/or gain-
• who have access to formal grievance procedure and/or complaint resolution
• administered an employment test before hiring?
What proportion of non-entry level jobs have been filled from within in recent
What is the average number of hours training received by a typical employee
over the last 12 months?
What is the proportion of the workforce:
• whose performance appraisal is used to determine their compensation
• that receives formal performance appraisals?
Which of the following decision rules do you use most often: (a) merit or
performance rating alone, (b) seniority only if merit is equal, (c) seniority among
employees who meet a minimum merit requirement, (d) seniority?
For the five positions your firm hires most frequently, how many qualified
applicants do you have per position (on average)?
These practices were measured for both managerial and non-managerial employees
and the responses weighted by the proportion of employees in each category. From
these measures, two scales of HR practices were created and then tested for
convergent validity by testing against two independent external measures:
• ratio of HR staff to total staff
• references in the annual report to the importance of people.
App1.p65 16/06/03, 15:2839
A number of financial measures were also derived from the annual reports: Tobin’s
q (market value of firm divided by book value of assets), and GRATE (gross rate
of return on capital).
The questionnaire was sent to the senior HR manager in firms with more than 100
employees. Firms were excluded if they were foreign owned, holding companies
or publicly held divisions or business units of larger firms. His response rate was
28 per cent (ie 986 firms).
His review of the literature provided several research hypotheses:
• Systems of high performance work practices will diminish employee turnover
and increase productivity and corporate financial performance.
• Employee turnover and productivity will mediate the relationship between
systems of high performance work practices and corporate financial
• Complementarities or synergies among HPWPs will diminish employee
turnover and increase productivity and corporate financial performance.
• Alignment of a firm’s system of HPWPs with its competitive strategy will
diminish employee turnover and increase productivity and corporate financial
Main finding is that 1 standard deviation increase in HR practices index is associated
with 7.05 per cent decrease in labour turnover and $27k more in sales, $18.6k
more in market value and $3.8k more in profit.
In terms of the ‘internal’ and ‘external’ fit models, the evidence was not convincing.
There was no support for internal fit and modest support for external fit.
The unit of analysis is the company level, and data on both HR practices and
financial performance are reported at this level. A problem with this level of
response data is that they encompass a wide number of multi-establishment
enterprises. We know that both performance and HR practices can vary
considerably between establishments in a company. This heterogeneity may be
diffusing the results and weakening both internal and external fit models through
the reporting of ‘average’ scores across establishments with diversity in HR
practices and performance outcomes. Therefore, it is possible that ‘fit’ may exist
in some of the establishments and not in others, and performance may be better
or worse in some establishments than others. The averaging out of responses that
happens at the company level in multi-establishment enterprises does not help us
understand the cases in which fit and performance are occurring. The study would
App1.p65 16/06/03, 15:2840
need to have controlled for centralisation of HR policy as well as single/multi-
For example, in the aerospace study one of the authors (MT) received replies
from several BAe sites as well as the company HO. There is considerable diversity
in the reporting of both HR practices and in evaluations of firm performance.
Some establishments conform to the high-performance work practices/business
success model, whereas other sites do not.
This may suggest that there is an element of strategic choice in the employment
models being used, and this dimension is not considered in the Huselid article.
Overall, a dense and sophisticated article that excels in the art of measurement.
RTHUR J. B. (1994) ‘Effects of human resource systems on manufacturing
performance and turnover’ Academy of Management Journal. Vol. 37, No. 3.
Strategic choice from the organisational theory field (revisionist contingency theory
arguing for a mediation between actor-centred theories of organisational change
and structural determinism).
Strategic HRM: drawing on typologies developed by Dyer and Reeves (1995),
Miles and Snow (1984), Schuler and Jackson (1987), etc).
‘Commitment’ and ‘Control’ typology developed in HR/IR literature.
Study expects to find different HR practices associated with these broad approaches
and also different performance outcomes. Hypothesised that commitment model
will lead to higher performance than control model.
Cross-sectional survey of 54 steel mini-mills employing 460 people on average
(sd 318). Establishment-based responses, completed by HR manager and top line
manager (for business strategy information); 29 usable replies from both
HR system – 10 indicators (developed from literature):
App1.p65 16/06/03, 15:2841
• Decentralisation: level of discretion of non-management employees over a
range of production decisions (eg quality, work flow,
investment in kit, product development)
• Participation: percentage of non-management employees involved in
problem-solving groups, trained in such activities or
involved in joint union-management committees
• General training: proportion of training given to areas that are not task/job
• Skill: proportion of maintenance and craft workers
• Supervision: number of production workers per supervisor
• Social interaction: number of social events organised
• Due process: percentage of grievances dealt with formally
• Wages: average total employment cost per production worker
• Benefits: percentage of average hourly compensation cost accounted
for by benefits
• Bonus: percentage of total average employment cost accounted for
by bonus/incentive payments.
Measure of firm performance used:
• Labour efficiency: number of labour hours required to produce one tonne of
• Scrap rate: tonnes of steel melted
• Labour turnover: voluntary and involuntary.
Cluster analysis used to differentiate HR systems. Six clusters identified but
reduced to two because of degrees of freedom problems (ie small number of cases)
in subsequent analysis.
Hypotheses tested using regression analysis. Unionisation the main control variable
– a wider set (thus enabling more complex contingency tests) could not be used
because of small sample size.
Mills that conformed to the ‘commitment model’ demonstrated higher
productivity, lower levels of employee turnover and lower scrap rates.
App1.p65 16/06/03, 15:2842
The study also found that control and commitment HR systems moderated the
relationship between employee turnover and firm performance.
The cross-sectional survey does not permit understanding of causality in
development of HR systems and firms’ better performance. It also fails to take
account of process issues and managers as strategic actors.
No idea of employee or managerial impact of these models – strong normative
underpinning to ‘commitment’ model.
The measures developed and used my not be comprehensive enough to understand
the full impact of HR systems on performance.
ONYON M. (1998) Human Resource Management Systems and the Productivity
of British Firms. (mimeo)
This is a very welcome addition to the UK literature on HR strategies. Like all
such papers, however, it has its more debatable points.
Essentially, Conyon seeks to estimate the association between firms’ sales levels
and the degree to which they use a variety of so-called ‘innovative’ HR policies.
He deploys a standard production function approach and concludes that, after
taking certain other conditions into account, a more intense use of innovative HR
policies is associated with a significantly higher level of sales; given that he controls
for the size of the labour force, this means that innovative HR is also associated
positively with labour productivity (ie output per employee). He suggests that a
‘unit increase’ in HR innovation may raise productivity by 8 per cent.
He uses data from the 139 respondents to a survey sent to the HR director of the
largest 1,000 UK firms with a stock market quotation (excluding investment
trusts). He is centrally concerned with the issue of complementarities among HR
practices, and his review of the theoretical literature is relatively wide ranging in
this respect. However, his way of operationalising complementarities is perhaps
not so well developed. This is done in the following way.
First he has data on six HR areas:
• incentive pay systems
• work teams
• employment security
App1.p65 16/06/03, 15:2843
• job flexibility
• skills and training.
For each of these areas he has more than one indicator, ranging from two indicators
for employment security to nine for incentive pay systems (not eight as is stated
in the text, page 18). There are 32 indicators in total. Each indicator is measured as
a yes/no dummy; ie do you make any use of policy X? For each of the six areas a
firm is said to make intensive use of innovative practices if it answers yes to at
least 75 per cent of the indicators (this presumably implies that for a firm to be an
intensive user in those areas with only a few indicators, it has to answer yes to all
the questions). Conyon then sums up the number of intensive HR areas to form
his measure of the HR system; (on page 19 he says that this can range from a value
of 1 (least innovative) to 6 (most innovative); in fact, the lowest possible score is
surely 0; see also page 23). High values on this scale are said to imply greater HR
complementarity, ie the practices are not seen to be possible substitutes for one
This procedure and the findings raise a number of issues:
• Within any of the six HR areas, all indicators are given the same weight. So,
for example, when measuring the intensity of incentive pay, the use of executive
share options is deemed to have the same importance as, say, individual pay
being linked to skills. Further, all six areas are given the same weight; so that a
firm which is an intensive innovator in incentive pay, work teams and skills
and training gets the same HR system score as one stressing communications,
employment security and job flexibility. Are all these indicators/areas really of
equal importance for productivity? Do communications policies have as strong
an impact on productivity as pay systems? There is no theorising here.
• If very different combinations/bundles of the six have the same score, is
Conyon saying that it doesn’t matter which bundle a firm goes for? This seems
odd; it certainly sits oddly with notions of internal and external fit.
• There are potentially important problems in measuring HR policies (a) at
corporate level (recall that these are all large firms, often with complex divisional
structures), and (b) by dummy variables (which does not allow one to know
how extensively any practice is used).
• Conyon estimates the regressions for a set of years, 1989 to 1995. But the HR
measures mostly refer to arrangements in a single year, 1996. If HR
arrangements change significantly over time (as eg Huselid claims), it’s not
quite clear how one interprets Conyon’s associations, which are between past
productivity and current practices.
• Conyon wishes to explore complementarities further by seeing whether there
is more explanatory power if the individual HR practices are added to the
equation (which means that he runs an additional 32 equations). When doing
this he finds that most of the individual practices are not significant (but no
fewer than eight – one quarter – are!). This looks like an odd procedure, and it
is not clear what it shows. First, the individual practice is already represented
in the overall HR index (though its impact may be pretty dilute); to have it a
App1.p65 16/06/03, 15:2844
second time in the same equation at least raises the possibility of multi-
collinearity and, therefore, possibly bogus non-significance. Second, the same
problem might well come through if there are high correlations among the
components of each of the six sub-indices. In all, this might be a weak test of
• The incidence of the indicators looks to be pretty high. Nearly one-third of
the 32 practices are used by more than 90 per cent of the sample, 19 of the 32
are used by more than 80 per cent of the sample, and all but one is used by a
majority of the sample. I would guess that the wider population values were
• In Table 3 Conyon reports that there are 991 observations; given that he has
data for seven separate years on 139 companies, it’s not easy to see why there
are quite as many as 991 observations.
• Has Conyon investigated whether the relationship between HR and
productivity is linear? Is there an 8 per cent rise in productivity associated
with each unit increase in the HR index, or is it the case, as eg Huselid suggests,
that there is a complex pattern of increasing and diminishing returns to HR
UEST D. and HOQUE K. (1994) ‘The good, the bad and the ugly: employment
relations in new non-union workplaces’. Human Resource Management Journal.
Vol. 5, No. 1. pp.1-14.
Objectives of the study
The authors claim that they have three objectives:
• to present a new, four-fold classification of HR arrangements, and relate these
to ‘background characteristics’
• to assess a variety of consequences of using each of the four types, and
• to provide a test of the impact of a ‘high utilisation model of HRM’.
Sample characteristics and research method
The usable data relate to a sample of 119 ‘new’ non-union establishments in the
UK employing more than 50 people; the great majority of these were in
manufacturing industry and were contacted in 1993. ‘New’ means first that
establishments which were set up only after 1979 were included, and second that
establishments were either greenfield sites or had recently experienced a major
change of business and ownership. Managers at all these establishments were
therefore said to be ‘unconstrained by any local history or tradition in the
App1.p65 16/06/03, 15:2845
Data were collected by postal questionnaire. The parent study collected data on
347 establishments (from 1000 originally contacted), but this study is confined to
the non-unionised establishments.
A fourfold classification of HR arrangements
Guest and Hoque divided the establishments by (a) whether they had an HR
strategy, and (b) the content of their HR practices.
Was there a strategy?
In deciding whether there was an HR strategy Guest and Hoque relied on the
answers to a single survey question ‘Do you have a human resource strategy
formally endorsed and actively supported by the top management team at the
establishment?’; 64 said yes, 55 said no.
The content of the HR practices
In assessing this, Guest and Hoque seem to have used the yes/no answers to all of
23 questions, each of which had the form of ‘Do you have…?’ A list of these is
shown in the appendix. Two points may be made. First, the last two of these 23
items do not refer to practices as such but to policy coherence (ie external and
internal fit); they might have been better used to measure the existence of strategy.
Second, the other 21 items all refer, in some sense, to ‘modern’ HR practices; ie
they are often linked to a philosophy of HRM. So a high score on the practices
measure is taken to mean that the establishment is closer to practising ‘HRM’.
These data were used in a somewhat crude way. Guest and Hoque say that ‘It
might be argued that some practices are more important than others, but we know
of no theoretical basis on which to determine this’ (page 6). If this is true it might
also be true that there is no clear theoretical basis for specifying the original list –
certainly, other research specifies somewhat different lists. The decision was taken
to count ‘high’ HRM practice where establishments used more than half the
practices, whatever precise combination was employed; low HRM practice
establishments used a minority of the practices.
We therefore have the following fourfold classification, using Guest and Hoque’s
somewhat tendentious labels:
Table A2 Four HR classifications
HRM Policy and Practice
App1.p65 16/06/03, 15:2846
Of the 64 establishments judged as having a strategy, all but eight used more than
half the practices; of the 55 judged to be without a strategy, roughly half used
more than half the practices. It should be noted that the cell sizes are not always
big enough for some of the subsequent analysis.
HR type and performance
Guest and Hoque have three sets of outcome variables. They have eight so-called
HR outcomes (eg employee commitments levels, staff flexibility, etc), three so-
called IR outcomes (industrial dispute incidence, labour turnover rates and
absenteeism), and six organisational outcomes (eg quality and productivity
measures, and economic health). All but the IR variables and one of the
organisational variables come from questions that ask for respondent judgements
on five-point scales.
Guest and Hoque follow two different procedures. First, they look to see whether
the average score on each of these 17 measures differs as between the four HR
types; second, they carry out a limited multivariate analysis to see whether, certain
other things held constant, outcomes vary by HR type. Their own conclusion is
that ‘The good…consistently report the best results for…[organizational]
performance outcomes. In contrast, the bad…consistently report the poorest
outcomes’ (page 11). This seems to be an over-statement, certainly if one wishes
to employ tests of statistical significance. We concentrate here on the six
The results of the simple comparisons is shown in Table A3. There is little to
distinguish the four types as far as their organizational outcome is concerned.
‘Good’ firms tend to have slightly higher scores, but the differences are not statistically
significant except (weakly) in the case of quality benchmarked against the UK.
The multivariate analysis suffers from the limited number of interesting variables
that could be controlled for (establishment size, type, parent organisation, region
and national ownership). ‘Good’ certainly does not do worse than the others but
there is no clear tendency for it to do better than each of the other three types, or
for ‘Lucky’ and ‘Ugly’ to do better than ‘Bad’, which is what one version of the
theory might predict.
Table A3 Organisational performance outcomes by HR type
Percentage of quality targets attained
How well was the recession weathered
Productivity benchmarked against UK
Quality benchmarked against UK
Productivity benchmarked against world
Quality benchmarked against world
App1.p65 16/06/03, 15:2847
Conclusions and discussion points
• There is at best a weak and inconsistent link between HR type and
• The measure of whether an establishment has an HR strategy is rather weak,
but at least the question is explicitly posed.
• The measure of practices is weakened by the use of dichotomous measures
(Do you/don’t you have…?) rather than more continuous measures.
• The use of the practices data, simply adding up any combination of 23 practices
see Table A4 , regardless of whether they fit together, is questionable.
• Can we do better than just bundle together any possible combination of
practices? Specifically, can we develop some theory which says these must be
present for a certain strategy to exist, while these others are optional extras
(and get a lower weight)?
Table A4 The incidence of HR practices
Harmonised terms and conditions
Internal promotion the norm
No compulsory redundancy
Trainability as a major selection criterion
Use of psychological tests for all selection
Realistic job previews
System to communicate values to new staff
Deliberate development of a learning organisation
Minimum annual training requirements for all
Flexible job descriptions
Jobs designed to make full use of skills
Teamworking for majority of staff
Staff involved in setting performance targets
Staff responsible for own quality
Majority involved in QCs
Regular use of attitude surveys
Team briefing/information cascades
Information of market position/company performance
Merit pay for all
Formal appraisal at least annually
HR policy integrated with business strategy
HR policies integrated with each other
App1.p65 16/06/03, 15:2848
HUNTER LARRY W. (1998) What Makes for a High-Performance Workplace?
Evidence from retail bank branches. Unpublished paper, Wharton Business
School, University of Pennsylvania.
Adopts high-performance/high-involvement workplace model to inform research
design. Also draws upon manufacturing strategy and notions of ‘underlying
production logics’. Piore’s crisis of regulation idea is also alluded to.
Customer service ‘logics’ are seen as highly relevant in understanding HR
approaches. A distinction is made between ‘encounter-based’ and ‘relationship-
based’ customer service activities. All banks provide a mix of both, but one may
be more dominant than the other in how work is organised.
To understand whether flexibility (through task cross-training) and autonomy
(devolution of decision-making) are related to the business performance of retail
banks. The paper hypothesises that both will lead to superior performance
outcomes and moreover that there will be a ‘synergistic’ relationship between
autonomy and flexibility (ie the sum is greater than the whole).
Survey of technology, work practices and performance of 135 large US retail banks.
Questionnaire surveys were sent to branch manager. These were combined with
independent measures of customer behaviour and census data on local
demographics. 241 branches were included in analysis from 101 banks.
Work organisation: eight items used to measure ‘branch flexibility’. Seventeen
items used to assess autonomy (data derived from bank manager reports about
Performance of branches: uses operationally based rather than financially based
measures because of comparability difficulties across banks. Productivity measured
by mix of employment and output indices at branch level. Effectiveness measured
by ‘cross-sell’ rate of other financial products to same customers. Quality measured
by level of mistakes – in this case ‘accounts out of balance’.
Control variables: local demographics to control for customer base (income,
population size, etc); level of technology by the number of terminals and software
programme complexity. HR practices – education levels of employees; use of
variable pay; share of part-time workers; de-skilled tellers.
App1.p65 16/06/03, 15:2849
OLS regression model. Variance in performance measures (productivity, cross-
sell rates and branch errors) used to understand contribution of work organisation
and HR practices. Use of other independent data sources to assist in building
reliable performance measures.
The analysis found that flexibility and autonomy when used separately are
correlated with higher productivity and sales. However, when they are used together
they are negatively associated with performance. These results go against the
findings of manufacturing-based studies that found complementarity and
synergistic results when broad jobs and greater decentralisation of decision-making
were combined. This underlines some of the difficulty in studying these issues in
service sector firms.
Hunter draws attention to the finding that autonomy and flexibility are positively
correlated and interprets this as showing that managers are making choices about
what work organisation systems they want to use – managers who want to minimise
loss might chose a different work system than those who want to increase sales. A
strategic choice model of HRM may therefore be supported by the data.
Measures of HR practices appear to be both inconsistent and weak. Few of measures
used in manufacturing studies were deployed and the author resorts to human
capital measures instead (ie education levels).
The study is looking more at work organisation than HR practices in terms of
explaining performance. The poor measures of HR practices do not help us
understand the relative contribution of these factors to performance.
App1.p65 16/06/03, 15:2850
Current and planned research on HR
strategies and business performance in
A number of studies are being co-ordinated by either David Guest or Jonathan
Michie. Guest has received top-up funding from the IPD in order to adopt a
telephone survey methodology for a wider study being funded by the ESRC. Michie
is also running a telephone survey looking at innovative work practices, which is
being funded by Leverhulme. Both studies are measuring HR practices as well as
gathering data on business performance (both subjective and objective).
Guest is also working with Sheffield (Patterson) and Kings College (Peccei) on
further analysis of the 10-year longitudinal data set held at Sheffield.
It is likely that Birkbeck will also be using WERS to explore HR and business
Thompson has a number of studies that are of relevance. First, there is an
establishment-based survey of the aerospace sector, which captures HR practices
as well as value-added per employee performance data. This survey, first conducted
in 1997, is being repeated in 1999 and will contain a large panel element, which
allows for comparability over time. This was a mono-respondent survey but
completed by senior managers (MDs, etc). Second, a number of case studies of
aerospace establishments are being conducted that look at HR strategies in practice,
particularly their impact on employees. The psychological contract model is being
used to compare HR strategies in different organisational contexts. Third, under
the ESRC innovation programme, a mixed method design is being adopted to
look at HR and reward practices and one dimension of business performance
(innovation). This survey is of the ‘knowledge’ economy, and deals mostly with
high technology and pharmaceutical companies. Five years of business performance
data have also been captured. A number of case studies are being conducted. Fourth,
Thompson is working with Richardson and Peccei on cross-sectoral survey data
collected in 1997 that explores HR practices and business performance issues.
Thompson is also working with SKOPE at Oxford to develop work in the aerospace
sector that looks at skills, training and economic performance. A number of detailed
case studies are being developed, and the intention is to look at comparative research
designs in other European aerospace establishments. SKOPE will be focusing on
high performance work systems as one of their core areas and will shortly be
commissioning a critical review of research and theory.
App1.p65 16/06/03, 15:2851
Conyon with funding from PriceWaterhouse Coopers is developing a programme
of work in this area. To date a small-scale pilot survey has been conducted of
around 120, mostly large, private sector organisations. There are plans to extend
and develop this work.
John Purcell is co-ordinating work on the engineering sector that, among other
issues, is looking at HR practices and business performance. This is concentrating
specifically on SMEs, which is a neglected area in the current literature, and as
well as survey material it also involves detailed case studies.
Workplace Employee Relations Survey (WERS)
A number of Academics are working on the recently released WERS data. These
data contain a sample of establishments with Census of Production data, which
give measures of firm performance. These findings will be coming out over the
course of 1999 and 2000. Key universities working on the data are Cardiff, LSE,
Warwick, Strathclyde and Birkbeck.
App1.p65 16/06/03, 15:2952
Analysis of key empirical studies
Prepared by Donna Buttigieig, Research Associate,
Templeton College, Oxford
The data were collected at 30 US steel mini-mills. The study is cross-sectional and
industry specific. OLS regression used.
• HR Systems (control or commitment) categorised according to levels of
decentralisation, participation, general training, skill, supervisor, social, due
process, wages, benefits, bonus (incentive payments).
• Total turnover.
• Control variables include age, size, union status and business strategy.
• Labour efficiency (average number of labour hours to produce one tonne of
• Scrap rate (number of tonnes of raw steel melted to produce one tonne of
• A commitment HR system is a positive and significant determinant of both
labour hours and scrap rates.
Effectiveness of commitment-type HR systems on performance.
App3.p65 16/06/03, 15:2953
GUEST D. and HOQUE K. (1994)
Cross-sectional data collected in 119 ‘new’ non-union establishments (largely in
manufacturing) in the UK. Ordered Probit, OLS, or LOGIT analysis used, where
• Establishments are categorised as ‘good’ (clear HR strategy and good HR
practices full utilisation, high involvement HR model), ‘ugly’ (efficiency driven
model with HR strategy but little use of HR practices), ‘lucky’ (no clear
strategy but use of HR practices) and ‘bad’ (no HRM strategy and low uptake
of HR practices). These categorisations are made on the basis of a question
about the existence of HR strategy and the use of 23 practices.
• Controls include size, national ownership, sector, regional location and degree
• Percentage of quality targets attained.
• How well the recession was weathered.
• Productivity benchmarked against UK.
• Quality benchmarked against UK.
• Productivity benchmarked against the world.
• Quality benchmarked against the world.
• The ‘good’ claim to have weathered the recession significantly better than the
• The ‘bad’ claim to be doing the best in terms of attainment of quality targets
• The ‘good’ organisations claim to perform better on quality when benchmarked
against other organisations in the UK.
App3.p65 16/06/03, 15:2954
Cross-sectional data collected from 62 automotive plants representing 16 countries.
Hierarchical regression analysis using Cobb-Douglas production function.
• Use of buffers
• Work systems (index incorporating measures such as per cent in work systems,
employee involvement, production-related suggestions, job rotation and quality
• HR policies (index incorporating hiring criteria, compensation scheme, status
barriers and training)
• Total automation
• Production scale
• Model mix complexity
• Parts complexity
• Production design age.
• Labour productivity (hours of actual working effort to build single vehicle
adjusted for comparability across plants)
• Quality (consumer perceived quality, defined as defects per 100 vehicles).
• The three production organisation indices (leaner buffers, more multi-skilled
work system and practices) are associated with productivity and quality (except
use of buffers for quality)
• A combination of all three elements in a single index has a significant effect on
productivity and quality.
App3.p65 16/06/03, 15:2955
• Assembly plants using flexible production systems, which bundle HR practices
into a systems integrated with production/business strategy) outperform plants
using traditional mass production systems in productivity and quality.
HUSELID M. (1995)
Cross-sectional, multi-industry study of 968 US-owned firms with over 100
employees. Regression Analysis used.
• Employee skills and organisational structures index of practices that enhance
employees’ knowledge, skills and abilities, ie formal job design programme,
providing training, QWL programmes, quality circles, labour-management
teams, information-sharing programmes, formal grievance procedures, profit
and gain-sharing plans.
• Employee motivation index of high performance work practices including
formal appraisals and links with pay, merit in promotion.
• Controls include total employment, K intensity, union coverage, industry union
coverage, concentration ratio, sales growth, R&D sales, systematic risk and
• Productivity (log of sales per employee)
• Corporate financial performance (Tobin’s q & GRATE).
• Turnover determined by employee skills and organisational structures (+).
• Productivity determined by employee motivation
App3.p65 16/06/03, 15:2956
• Tobin’s Q determined by employee skills and motivation (+).
• GRATE determined by employee skills and organisational structures (+).
• One standard deviation increase in each practice reduces turnover by 1.30 raw
• Productivity – one standard deviation increase raises sales an average of $27,044
• Per employee effect of increasing practices one standard deviation was $18,641
DELANEY J. T. and HUSELID M. A. (1996)
Cross-sectional data were collected in 727 organisations (multi-industry). Multiple
• Staffing selectivity index
• Training index
• Incentive compensation
• Grievance procedure
• Decentralized decision-making
• Internal labour market index
• Vertical hierarchy
• Control variables include non-profit organisations, subsidiary, total
employment, firm age, market competition, product, service, union pressure
and percent managers.
App3.p65 16/06/03, 15:2957
• Perceived organisational performance constructed from seven items assessing
performance of organisation over the last three years relative to rest of industry.
• Perceived market performance constructed from four items examining
perceptions of firm performance over three years relative to product market
• When HR practices are entered simultaneously training (+), incentive
compensation (+) and vertical hierarchy (+) are significant determinants of
perceived organisational performance.
• Perceived market performance is determined by staffing selectivity (+) and
incentive compensation (+).
• Tests for complementarities of HR practices through interaction tests were
• Some HR practices have a positive impact on perceived performance.
• There is no evidence that complentarities in HR practices enhance performance.
YOUNDT M. A., SNELL S. A., DEAN Jr J. W. and LEPAK D. P. (1996)
Longitudinal study conducted in 97 manufacturing plants. Regression analysis used.
• Administrative HR system index (ie selection for manual and physical skills,
training, results based performance appraisal, individual equity, individual
incentives and hourly pay)
• Human capital enhancing HR system index (ie selective staffing, selection for
problem-solving and technical skills, development and behaviour-based PA,
external equity, group incentives, skill-based pay and salaried compensation)
App3.p65 16/06/03, 15:2958
• Manufacturing strategy (quality)
• Manufacturing strategy (delivery flexibility)
• Manufacturing strategy (Scope flexibility)
• Manufacturing strategy (Cost)
• Controls: size and industry environment (three factors: munificence,
• Machine efficiency (self-report scale consisting of items such as equipment
utilisation, scrap minimisation)
• Customer alignment (product quality, on time delivery)
• Employee productivity (employee morale, employee productivity).
• Models that included interaction terms with HR strategies show that HR
systems were all significant determinates (main effects) to all independent
• Interaction terms that were significant
Equipment efficiency determined by interaction between cost and
An interaction between quality and HK-enhancing HR system predicts
customer alignment, employee productivity and equipment efficiency.
An interaction between delivery flexibility and administrative HR system
predicts customer alignment.
• Analysis exploring clusters of manufacturing strategies were not significant.
• Regressions exploring each HR system on the four manufacturing strategies
found that, aside from those companies employing a quality strategy, firms
not making a consistent connection between strategy and HR systems.
• Broad support for the contingency perspective:
Human capital-enhancing HR system important to firms competing on
App3.p65 16/06/03, 15:2959
Administrative HR systems still appropriate in context where need to
Negative interaction between scope flexibility strategy and administrative
HR explained, as this strategy requires employees with high discretion.
KOCH M. A. and GUNTHER MCGRATH R. (1996)
Data collected in 319 establishments. Multi-industry. Regression used.
• Planning investment (two item index dichotomously coded)
• Hiring investment (index combining number of interviews per placement,
recruiting sources and tests used)
• Investment in employee development (index created by weighting number of
formal training programmes and extent to which internal promotions used)
• Overall HR sophistication (index created by adding the scores on HR planning,
hiring and development)
• Control variables include business unit size, unionisation, industry, technology,
and capital intensity.
• Labour productivity (net sales divided by number of employees).
• Labour productivity is determined by HR planning (+) and hiring (+).
• Labour productivity model including interaction and main effects show HR
planning (+), Planning* capital (+), Hiring (+), Hiring* capital (+),
development (+) and development* capital (+) (where capital indicates capital
intensive industries) are significant.
• Labour productivity is determined by HR sophistication (+) and an interaction
with capital-intensive industries.
App3.p65 16/06/03, 15:2960
• Labour productivity is related to firms that formally plan their HR requirements
and where employers evaluate their recruitment and selection policies. Further,
this is of more importance in capital-intensive industries.
DELERY J. E. and DOTY D. H. (1996)
Cross-sectional study in the banking industry (114 banks). Information collected
from various sources. Analysis limited to core jobs only. Hierarchical regression
• Internal career opportunities
• Profit sharing
• Employment security
• Employee participation
• Clearly defined jobs
• Strategy (scale of six items measuring product/market innovation)
• Control variables include bank size, age, holding company and Federal Reserve
• Return on average assets
• Return on equity.
App3.p65 16/06/03, 15:2961
• ROA determined by results-oriented appraisals, profit sharing and employment
security. HR practices significantly explained 12.55 per cent of variance above
the control variables.
• ROE determined by results-oriented appraisals and profit sharing. HR practices
significantly explained 9 per cent of variance in ROE above the control variables.
• Model exploring interactions between strategy and each of the practice variables
had insignificant results. Post hoc tests revealed some support for the
contingency view-links between performance and the practices of performance
appraisal, participation and internal career opportunities are contingent on
• A model that classifies each bank according to the ideal employment system it
closely resembles showed the closer the employment system resembled the
market type system the higher the ROA and ROE.
• Results provided strong support for the universalistic perspective.
• Some support for the contingency view – banks able to align HR practices
with strategy are estimated to have a 50 per cent higher ROA and ROE than
those banks whose HR practices were one standard deviation of alignment.
• The configurational results suggest that some configurations of HR practices
are better than others in terms of performance ie the closer the bank’s system
resembles the market-type system the higher its performance.
HUSELID M. A. and BECKER B. E. (1996)
Cross-sectional, multi-industry data collected from 548 firms in 38 industries.
OLS estimation technique used.
• HRM system index constructed from mean of 24 questions focusing on
intensity with which high performance policies and practices have been adopted
by the firm.
App3.p65 16/06/03, 15:2962
• Effectiveness and alignment index constructed from 16 items reflecting
alignment of organisational context with principle of high performance work
• Controls include prior firm growth in sales, tangible assets, no. of employees,
investment in R&D, unionisation, firm systematic risk and industry codes.
• Variation of Tobin’s q
• Both factors are significant determinants of Tobin’s q.
HUSELID M. A., JACKSON, S. E. and SCHULER R. S. (1997)
Multi-industry, cross-sectional design. Sample of 293 publicly held firms.
• Strategic HRM (scale of eight items including teamwork, communications,
involvement, enhancing quality and developing talent to serve business in
• Technical HRM (scale of recruitment, selection, training, PAs and
• Professional HRM capabilities (scale of 11 items that describe expertise and
skill relevant to performing excellently within a traditional HR department)
• Business related capabilities (scale of three items describing amount of business
experience HR staff have had outside of HR function)
• Control variables include capital intensity, union coverage, firm size, industry
concentration, sales growth, stock variability, industry and R&D sales. Some
models contained control for a previous year’s performance.
App3.p65 16/06/03, 15:2963
• Employee productivity (net sales per employee)
• Employee productivity (GRATE)
• Employee productivity (Tobin’s q).
• Labour productivity (model including strategic and technical HRM only)
predicted by strategic HRM (+).
• GRATE (model including strategic and technical HRM only) predicted by
strategic HRM (+)
• Tobin’s q (model including strategic and technical HRM only) predicted by
technical HRM (+)
• Technical HRM effectiveness is determined by professional HR capabilities (+)
• Strategic HRM effectiveness is determined by professional HR capabilities
(+) and business-related capabilities (+).
• On a per employee present value basis, a one standard deviation increase in
overall HRM effectiveness corresponds to an increase in sales per employee
of 5.2 per cent, valued at $44,380.
• The impact of a one standard deviation increase in HRM effectiveness on profits
yielded an estimated increase in cash flow of 13.3 per cent, valued at $9673 per
• One standard deviation increase in HRM effectiveness yielded an estimated
increase in market value of 6 per cent, valued at $8882 per employee.
ICHNIOWSKI C., SHAW K. and PRENNUSHI G. (1997)
Data collected on 36 finishing lines in 17 US companies matched with 2190 monthly
observations of productivity data. OLS regression models.
App3.p65 16/06/03, 15:2964
• Traditional HRM (system 4) (firms identified as having no innovative practices
with close supervision, strict work rules, narrow job responsibilities, incentive
pay based on quantity, no work teams, no information sharing and no training)
• System 3 (similar to system 4 except introduced innovative practices through
involvement of teams and increased communication)
• System 2 (similar to three but also includes extensive training and high
involvement in teams)
• System 1 (incorporates innovative practices in all areas)
• Note: systems are classified on the basis of profit sharing, line incentives, high
screening, recruitment, high participation, multiple teams, formal team practice,
employment security, job rotation, high training, low training, information
sharing, managers meet workers regularly, meet with union, unionised, low
• Twenty-five controls that affect uptime (ie capital vintage, learning curve effects,
• Productivity (production-line uptime, which is the percentage of scheduled
operating time that the line actually runs)
• Quality (the percentage of total production that met the standards as prime
• Lines with HRM system 1 have the highest productivity, followed by systems
2, 3 and 4.
• The more innovative the HR practices the higher the quality of steel produced
on the lines.
• A model incorporating individual HR practices as well as the four HR systems
indicate that the individual practices have no additional impact on productivity.
• If one line changed form HRM system 4 to system 2 and maintained these
changes over 10 years, it would increase its operating profits by over $10 million
strictly as a result of the HRM changes.
App3.p65 16/06/03, 15:2965
• Evidence also shows that systems of HRM practices determine productivity
and quality. Marginal changes in individual work practices have little effect.
ICHNIOWSKI C. and SHAW K. (1997)
Comparative (Japan and US) study exploring steel finishing lines in 36 US
production lines and five Japanese production lines. The productivity analysis is
based on a panel sample of up to 2,594 monthly observations. The data collected
are of both a qualitative and quantitative nature. GLS models.
• Japanese model:
Japanese HRM system
• American model:
Innovative HRM practices
High teamwork and training US system
Low teamwork and communication US system
Traditional US system
• Categories are determined by scores on seven HRM policy areas: incentive
pay, recruiting, teamwork, job security, job flexibility, training and labour
• Twenty-five control variables are included. For example, age of the line,
technology and quality of supplied steel.
• Productivity (percentage of scheduled operating time that the line actually
• Quality (prime yield rate – the percentage of total production in a month that
meets specific quality standards for designation as prime finished steel).
App3.p65 16/06/03, 15:2966
• Regression estimating the average uptime differential between Japanese and
all US lines, independent of effects of controls, indicate that the Japanese lines
are significantly more productive than the US lines.
• Regression exploring the four different HR categories in the USA indicate
that compared to the productivity of the traditional US system, the low
teamwork and communication system achieves higher productivity (1.5 per
cent). Similarly, the high teamwork and training system and the innovative
system are higher in productivity (43.6 per cent and 7.3 per cent, respectively).
• Regressions estimating the relative effects of the US and Japanese HRM
systems on the quality measure indicate that, on average, Japanese lines have
significantly higher quality.
• The regression that explores relative differences between the types of US lines
shows that quality is substantially higher in the Japanese and innovative US
lines compared with the high teamwork or the low teamwork system.
• Japanese production lines are more productive that the average US lines (by 6
per cent), however where the US lines have introduced innovative work systems
they perform equally as well as the Japanese lines.
• The lines having innovative HR practices in Japan and the US are on average 7
per cent more productive than the US employing traditional HR practices.
• Japanese production lines produce better quality products, even in comparison
to the innovative US HR system.
• Qualitative data indicate that the different work practices comprising the
Japanese HR system are mutually reinforcing. HR systems that support
employee participation enable workers to fine-tune production process with
PATTERSON M. G., WEST M. A., LAWTHORN R. and NICKELL S. (1997)
Longitudinal study of single-site, single-product UK manufacturing firms with
less than 1,000 employees. Data collected relate to 67 firms. Data collection is
both qualitative and quantitative and multi-level.
App3.p65 16/06/03, 15:2967
• Acquisition and development of employee skills (selection, induction, training
and use of appraisals).
• Job design (skill flexibility, job responsibility, job variety and use of formal
• Quality improvement teams
• Comparative pay
• Incentive compensation systems.
• Labour productivity (ratio of sales over employment in firm, divided by the
ratio of sales over employment in the industry).
• Real profits per employee (profits before tax, deflated by the producer price
index of the industry and controlling for size of firm).
• HRM practices account for 19 per cent of variation between companies in
change in profitability. Acquisition and development of skills and job design
are significant determinants of change in profitability.
• HRM practices together account for 18 per cent of variation between
companies in the change of productivity. Acquisition and development of skills
and job design are significant determinants of change in productivity.
CHADWICK C. and CAPPELLI P. (1998)
Cross-sectional, multi-industry (manufacturing and retail) data collected in 3167
private sector US establishments. Structural equation modeling is used.
App3.p65 16/06/03, 15:2968
• Investment HR system (number of benefits offered, computer training, cross-
training, employee involvement meetings, employee:supervisor ratio, job
rotation, pay for skill, self-managed teams, stock options teamwork training)
• Contractual HR system (average compensation compared to market,
importance of industry credentials in hiring, flextime, job sharing, part-time
workers, recruiting costs and selection techniques).
• Cost differentiation business strategy.
• Differentiation business strategy
• Focus business strategy
• TQM programme
• Capital intensity
• Control variables include multi-establishment, size and R&D.
• Total sales of the establishment (natural log of sales)
• Value added of the establishment (logged)
• Investment HR systems are universally more effective than contractual systems
in increasing sales and value added and decreasing turnover.
• Interaction tests show that low cost strategies reduce sales and value added
significantly and increase turnover when combined with investment HR
systems (by 4 per cent, 4 per cent and 1 per cent respectively)
• Contractual HR systems combined with cost reduction strategy reduces sales
by 2 per cent and turnover by 1 per cent.
• Interaction tests between differention strategies and contractual HR systems
reduce sales by 3 per cent and decrease turnover by 1 per cent.
• Interaction tests between focus strategies and contractual HR systems show
that sales are reduced by 2 per cent and turnover is decreased by 1 per cent.
App3.p65 16/06/03, 15:2969
• Investment systems are associated with a 34 per cent increase in sales compared
with contractual systems that decrease sales by 5 per cent
• Investment systems increase value added by 31 per cent
• Investment systems decrease turnover by 5 per cent (compared with contractual
systems that decrease turnover by 2 per cent)
• Conclusion – investment HR systems are more likely to improve performance
than contractual HR systems.
• Traditional definitions of business strategy may be among the least important
contingencies in determining effects of HR systems. Also, that the costs of
mismatch between a generic strategy and HR system may be more powerful
than the positive effects of good matches.
• Investment HR systems are associated with better performance in unionised
establishments, where there are TQM programmes and establishments that
are more capital intensive.
SHEPPECK M. A. (1998)
Cross-sectional, multi-industry study of 106 organisations. Hierarchical regression
• Competitive strategy
• Employee skills and work policies (ie staffing, training, work design and
employee relations policies that foster skill acquisition)
• Supportive environment (ie diversity, employee assistance, employee voice and
• Performance measurement and reinforcement (performance appraisal and
• Market organisation (work design and compensation practices utilising a
App3.p65 16/06/03, 15:2970
• Management support for HRM.
• Perceived organisational effectiveness (self-report measure consisting of six
items comparing the rater’s organisation to competitors in the industry).
• Hierarchical regression results show that HR practices contribute and
additional 13.2 per cent to the prediction of organisation effectiveness above
the variance accounted for by the control variables, environment and
competitive strategy along (although no single practice in itself is significant).
GOMEZ-MEJIA L. R. (1988)
Longitudinal study of 388 manufacturing firms. Export performance data collected
30 months after predictor data. OLS regression used.
• International HRM strategy (scale incorporating extent to which HR strategies
of status and influence, financial rewards, hiring, promotion, training and
development and performance appraisal)
• Controls include age of firm, size, R&D intensity, stage in life cycle, foreign
exposure, risk avoidance, international orientation of firm’s future plans,
profitability expectations, attitude towards growth, profit and market
development, systematic exploration of marketing opportunities and
commitment to foreign advertising.
• Export performance (index created by standardising and adding three measures:
changes in international market share; changes in export intensity; and export
intensity relative to industry).
App3.p65 16/06/03, 15:2971
• International HR strategy is a significant determinant (+) of export
performance and increases explained variance by 0.09 per cent (after controlling
for other factors).
CONYON M. J. and READ L. E. (1999)
Cross-sectional company-level data collected in 135 UK stockmarket companies.
OLS regressions employed.
• Companies allocated into four categories of HR system based on index
containing items related to incentive pay systems, communications, work
teams, employment security, job security, skills and training.
• HRM system 1 (company uses no HR practices)
• HRM system 2 (company uses 1-2 HR practices)
• HRM system 3 (company uses 3-4 HR practices)
• HRM system 4 (company uses 5-6 HR practices)
• Controls include log of employment, capital use, union, labor relations and
• Productivity (log of real company sales).
• All systems of HRM are significant determinants of productivity in the OLS
• Most variance is explained by the HR system as opposed to individual practices.
App3.p65 16/06/03, 15:2972
• Measures of HR systems are positively correlated with a measure of company
• Random effects model indicates that a movement from using no practices to
using 1 or 2 yield 28 per cent increase in firm productivity.
HUNTER L. W. (1999)
Cross-sectional, single industry study (banking) of 241 US bank branches from
101 retail banks. Survey data were matched with customer behaviour data and
census data. Analysis used is OLS regression.
• Branch flexibility (8 item scale)
• Worker autonomy (17 item scale)
• Use of high skilled employees
• Share of part-time work
• Existence of limited function tellers
• Controls include local market conditions, demographics, and technology.
• Branch productivity
• Sales effectiveness (cross-sell rate)
• Quality (extent to which banks are out of balance).
App3.p65 16/06/03, 15:2973
• HR practices have no impact on branch productivity. Repeated analysis with
interaction term between autonomy and cross-training is negatively associated
• HR practices have no impact on the cross sell rate. Repeated analysis with
interaction term between autonomy and cross-training is negatively associated
with the cross-sell rate.
• Quality is determined by the interaction of cross-training and autonomy (+).
• The study highlights the difficulties in identifying and establishing ‘high
performance work organisations’ in customer service organisations
• For productivity and sales effectiveness, autonomy and job breadth appear to
contradict one another.
However, these results should be treated with some caution as autonomy and
flexibility are positively correlated.
App3.p65 16/06/03, 15:2974
References and further reading
ARTHUR, JEFFREY B (1992) ‘The link between business strategy and industrial
relations systems in American steel minimills’ Industrial and Labor Relations
Review Vol 45.
RTHUR, JEFFREY B (1994) ‘The effects of human resource systems on
manufacturing performance and turnover’ Academy of Management Journal
Vol 37 No 3.
ECKER, BRIAN and GERHART, BARRY (1996) ‘The impact of human resource
management on organisational performance: progress and prospects’ Academy
of Management Journal Vol 39 No 4.
OWEY, A and THORPE, R (1986) Payment Systems and Productivity. Macmillan
OXALL, PETER (1996) ‘The strategic HRM debate and the resource-based view of
the firm’ Human Resource Management Journal Vol 6 No 3.
HADWICK, CLINT and CAPPELLI, PETER (1998a) ‘Investments or contracts?: The
performance effects of human resource systems under different contingencies’
HADWICK, CLINT and CAPPELLI, PETER (1998b) ‘Alternatives to generic strategy
typologies in strategic human resource management’ in P Wright, L Dyer, J
Boudreau and G Milkovich (eds) Research in Personnel and Human Resource
Management Greenwich CT: JAI Press.
ONYON, M and READ, L (1999) ‘Corporate productivity, complementarities and
human resource management systems’. (unpublished paper, Warwick Business
ELANEY, JOHN T and HUSELID, MARK A (1996) ‘The impact of human resource
management practices on perceptions of organisational performance’ Vol 39
ELERY, JOHN E and DOTY, HAROLD D (1996) ‘Modes of theorising in strategic
human resource management: tests of universalistic, contingency and
configurational performance predictions’ Academy of Management Journal Vol
39 No 4 pp 802–35.
YER, L and REEVES, T (1995) ‘HR strategies and firm performance; what do we
know and where do we need to go?’ International Journal of Human Resource
Management, Vol 6 pp 656–70.
ERHART, BARRY, WRIGHT, PATRICK M, MCMAHAN, GARY C and SNELL, SCOTT A
(1998) ‘Measurement error in research on human resource decisions and firm
performance: how much error is there and how does it influence effect size
estimates?’ (paper presented at Academy of Management conference, San
Refs.p65 16/06/03, 15:3475
GOMEZ-MEJIA, L R (1988) ‘The role of human resources strategy in export
performance: a longitudinal study’ Strategic Management Journal, 9
OMEZ-MEJIA, L R and BALKIN, D B (1992) Compensation, Organisational Strategy
and Firm Performance. Cincinnati: Southwestern.
UEST, DAVID E (1997) ‘Human resource management and performance: a review
and research agenda’ The International Journal of Human Resource Management
Vol 8 No3 pp 263–76.
UEST, DAVID E and HOQUE, KIM (1994) ‘The good, the bad and the ugly:
employment relations in new non-union workplaces’ Human Resource
Management Journal Vol 5 No 1 pp 1–14.
UEST, D and PECCEI, R (1994) ‘The nature and causes of effective human resource
management’. British Journal of Industrial Relations 32 pp 219–242.
UEST, DAVID E and PECCEI RICCARDO (1997) ‘HR and Organsational Effectiveness
in the NHS’ (mimeo)
UNTER, LARRY W (1998) ‘What makes a high-performance workplace? Evidence
from retail bank branches’ (paper presented to the IRRA World Congress,
UNTER, L W (1999) ‘What makes a high-performance workplace? Evidence from
retail bank branches’ (unpublished paper, Wharton School, University of
USELID, MARK A (1995) ‘The impact of human resource management practices
on turnover, productivity and corporate financial performance’ Academy of
Management Journal Vol 38 No 3 pp 635–72.
USELID, MARK A and BECKER, BRIAN E (1996) ‘Methodological issues in cross-
sectional and panel estimates of the human resource-firm performance link’
Industrial Relations Vol 35 No 3 pp 400–422.
USELID, MARK A, JACKSON, SUSAN E and SCHULER, RANDALL S (1997) ‘Technical
and strategic human resource management effectiveness as determinants of
firm performance’ Academy of Management Journal Vol 40 No 1.
CHNIOWSKI, CASEY, and SHAW, KATHRYN (1997) ‘The effects of human resource
management systems on economic performance: an international comparison
of US and Japanese plants’ (NBER research paper)
CHNIOWSKI, CASEY, SHAW, KATHRYN and PRENNUSHI, GIOVANNA (1997) ‘The effects
of human resource management practices on productivity: a study of steel
finishing lines’ The American Economic Review June.
ACKSON, SUSAN E and SCHULER, RANDALL S (1987) ‘Linking competitive strategies
with human resource management practices’ Human Resource Management
Vol 23 No 3 pp 241–55.
Refs.p65 16/06/03, 15:3476
JACKSON, SUSAN E and SCHULER, RANDALL S (1995) ‘Understanding human resource
management in the context of or,ganisations and their environments’ Annual
Review of Psychology Vol 46 pp 237–64.
APLAN, R S and NORTON, D P (1996) The Balanced Scorecard. HBR Press.
OCH, MARIANNE J and MCGRATH GUNTHER, RITA (1996) ‘Improving labor
productivity: human resource management policies do matter’ Strategic
Management Journal Vol 17 pp 335–54.
OSSEK, ELLEN ERNST and YAKURA,ELAIN K (1998) ‘Opening the black box of fit
in strategic human resource management’ (paper presented at Academy of
Management conference, San Diego 1998)
ENGNICK-HALL, C A, and LENGNICK-HALL, M L (1988) ‘Strategic human resource
management: a review of the literature and a proposed typology’ Academy of
Management Review Vol 13 pp 454–70.
ACDUFFIE, JOHN PAUL (1995) ‘Human resource bundles and manufacturing
performance: organisational logic and flexible production systems in the world
auto industry’ Industrial and Labor Relations Review Vol 48 No 2 pp 197–221.
ARCHINGTON, M and SPARROW, P (1998) ‘Re-engaging the HRM function:
rebuilding work, trust and voice’, in Human Resource Management: The new
agenda edited by Sparrow, P and Marchington, M, FT:Pitman.
CMAHAN, GARY C, VIRICK, MEGHANA and WRIGHT, PATRICK (1999) ‘Alternative
theoretical perspectives for strategic human resource management revisited:
progress, problems and prospects’ in P Wright, L Dyer, J Boudreau and G
Milkovich (eds) Research in Personnel and Human Resource Management
Greenwich CT: JAI Press (forthcoming).
ILES, R and SNOW, C (1984) ‘Designing strategic human resources systems’
Organizational Dynamics Summer: pp 36–52.
STERMAN, P (1994) ‘How common is workplace transformation and who adopts
it?’ Industrial and Labour Relations Review 47 pp 173–88.
AAUWE, J and RICHARDSON, J (1997) ‘Introduction to the special issue on HRM
and Performance’ The International Journal of Human Resource Management
Vol 3 No 6.
ATTERSON, M, WEST, M, LAWTHORN, R and NICKELL, S (1997) The Impact of People
Management Practices on Business Performance. London, IPD.
FEFFER, J (1994) Competitive Advantage through People. Boston, HBS Press.
FEFFER, J (1998) The Human Equation. Boston, HBS Press.
OGERS, EDWARD W and WRIGHT, PATRICK M (1998) ‘Measuring organisational
performance in strategic human resource management: problems and prospects’
(working paper, School of Industrial and Labor Relations, Cornell University)
Refs.p65 16/06/03, 15:3477
SCHULER, R S and JACKSON, S E (1987) ‘Linking competitive strategies with human
resource management practices’ Academy of Management Executive Vol 1 No
3 pp 207–19.
HEPPECK, M A (1998) ‘SHRM configurations and perceived organizational
performance’ Academy of Management Proceedings San Diego.
HOMPSON, M (1998a) ‘Jet setters’ People Management April.
HOMPSON, M (1998b) ‘What works at work in UK Aerospace?’ Society of British
Aerospace Companies and Department of Trade Industry, Briefing Paper.
HOMPSON, M (1998c) ‘Skills, training and business performance in UK Aerospace’
Society of British Aerospace Companies and Department of Trade Industry,
RUSS, C and GRATTON, L (1994) ‘Strategic human resource management: a
conceptual approach’ International Journal of Human Resource Management
Vol 5 No 3 pp 663–86.
LRICH, D (1998) Human Resource Champions. Boston, HBS Press.
ENKARAMAN, N (1989) ‘The concept of fit in strategy research:toward verbal and
statistical correspondence’ Academy of Management Review Vol 14 pp 423–44.
ALTON, R (1985) ‘From control to commitment in the workplace’ Harvard
Business Review 63 (2) pp 77–84.
EST M (1996) ‘Reflexivity and work group effectiveness: a conceptual integration’
in West M (ed.) Handbook of Work Group Psychology Chichester, Wiley. pp
EST, MICHAEL and PATTERSON, MALCOLM (1997) ‘The impact of people
management practices on business performance’ IPD Research Paper No 22.
RIGHT, PATRICK M and MCMAHAN, GARY C (1992) Theoretical perspectives for
strategic human resource management’ Journal of Management Studies Vol 18
No 2 pp 295–320.
RIGHT, PATRICK M and SCOTT SHERMAN, W (1999) ‘Failing to find fit in strategic
human resource management: theoretical and empirical problems’ in P Wright,
L Dyer, J Boudreau and G Milkovich (eds) Research in Personnel and Human
Resource Management Greenwich CT, JAI Press (forthcoming).
RIGHT, PATRICK M and SNELL, SCOTT A (1999) ‘Toward a unifying theory for
exploring fit and flexibility in strategic human resource management’ Academy
of Management Review (forthcoming).
OUNDT, M, SNELL, S, DEAN, J and LEPAK, D (1996) ‘Human resource management,
manufacturing strategy, and firm performance’ Academy of Management Journal
Vol 39 No 4 pp 836–866.
Refs.p65 16/06/03, 15:3478