Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 2002 24

EMPLOYEE RESPONSES TO HIGH
PERFORMANCE WORK SYSTEM
PRACTICES: AN ANALYSIS OF THE
AWIRS95 DATA
BILL HARLEY*

S

ince the early 1990s, there has been increasing academic interest in a range of
labour management practices labelled variously as ‘high commitment management’,
‘high involvement management’ and ‘High Performance Work Systems’ (HPWS). Such
sets of practices – which include performance-related pay, various employee communication mechanisms, training and team-based work – when used in combination, are said
to be mutually reinforcing and to generate superior organisational performance. Advocates
of HPWS argue that improvements in organisational performance are generated because
these practices enhance employee discretion which, in turn, flows into improved attitudes
to work. Conversely, critics suggest that HPWS practices lead to work intensification,
with any gains in discretion being marginal. In spite of the volume of literature in this
area, there are few studies that have examined the impact of HPWS on employees. This
article seeks to remedy this gap by reporting results of analysis of the Australian Workplace
Industrial Relations Survey 1995 (AWIRS95) data aimed at assessing the associations
between HPWS practices and employee outcomes. The analysis suggests that there are

few such associations, and those that exist are weak, which calls into question the claims
of both advocates and critics of HPWS.

INTRODUCTION
The present study adds to a growing literature which examines a range of
putatively humanistic and developmental labour management practices variously
conceptualised as ‘high involvement management’, ‘high commitment management’ and ‘High Performance Work Systems’ (HPWS) using large scale
national survey data (see, for example, Ramsay et al. 2000; Wood 1999, Wood &
de Menezes 1998). The particular concern of this paper is to explore employee
responses to such practices (referred to hereafter as ‘HPWS practices’).
The bulk of the existing literature that promotes such practices, and which
derives chiefly from the United States, does so on the basis that the practices
enhance organisational performance (see Becker & Huselid 1998). There is a
consensus among researchers who have demonstrated a link between labour
management systems and organisational performance, that the performance
* Associate Professor, Department of Management, The University of Melbourne, Victoria 3010.
Email: [email protected]

THE JOURNAL OF INDUSTRIAL RELATIONS, VOL. 44, NO. 3, SEPTEMBER 2002, 418–434


HIGH PERFORMANCE WORK SYSTEM PRACTICES

419

outcomes arise chiefly because the practices improve employee orientations to
work, which makes them more productive (see Appelbaum et al. 2000).
Surprisingly, there has been little research which has sought to demonstrate
this causal path. Indeed, most commentators have been happy to assume that it
exists. Similarly, little attention has been paid to the possibility that, rather than
improving employee experiences of work, HPWS practices may contribute to
organisational performance via a process of work intensification, which is likely
to have negative consequences for employees in terms of effort and stress (Ramsay
et al. 2000). As such, in addition to considering outcomes from HPWS in terms
of customary measures of organisational performance, we need to ask whether
HPWS practices are associated primarily with positive or negative employee
responses.
This question is addressed using data from the 1995 Australian Workplace
Industrial Relations Survey (AWIRS95). The AWIRS95 dataset allows construction of indicators of workplace practices which the literature identifies as being
associated with HPWS. These items include factors such as communication
practices, team-based work, employment security, training, and family-friendly

policies. Fortunately, the structure of the AWIRS95 survey makes it is possible
to attach these workplace characteristics to employees and thus to ascertain
whether individual employees work in workplaces with HPWS practices in place.
The AWIRS95 employee survey also provides indicators of a number of
employee ‘orientations to work’ and this study utilises the following such
measures: discretion; commitment; satisfaction; relations with management;
effort; and stress. In addition, the employee survey allows the construction of a
series of control variables which account for characteristics of workplaces
and employees which are likely to intervene in associations between HPWS
practices and employee orientations.
Using these indicators, it is possible to measure associations between individual
HPWS practices and employee orientations to work and thus explore employee
responses. This is done initially by means of bivariate correlation analysis and
then using a series of regression models that integrate control variables. As we
shall see, the findings provide little support for either ‘positive’ accounts, which
suggest that HPWS practices work because employees enjoy positive outcomes,
or ‘negative’ accounts, which highlight work intensification and stress.
The present article is divided into four sections. The first provides an overview
of the literature on HPWS and establishes a research agenda. This is followed
by a discussion of the AWIRS95 data and how these were used to construct

variables. Discussion then turns to analysis and results. The concluding section
considers the implications of the results.

HIGH

PERFORMANCE WORK SYSTEMS

Since the early 1990s there has been a massive growth in academic interest in a
range of labour management practices, which are labelled variously as ‘high
performance’, ‘high involvement’ or ‘high commitment’ approaches. Underlying
these practices, which include measures such as employee involvement, teambased work, ongoing training and development, employment security, and

420

THE JOURNAL

OF

I N D U S T R I A L R E L AT I O N S


September 2002

performance-based pay, is the belief that organisations that treat their employees
well, provide them with a degree of autonomy and develop their potential to
the full, will reap benefits in terms of enhanced performance. The idea that an
emphasis on employee welfare, development and motivation is effective from a
managerial perspective is hardly new. This theme was central to the ‘human
relations’ model (Mayo 1946) and, more recently, has been a centre-piece of
the ‘soft’ Human Resource Management (HRM) approach associated with the
‘Harvard school’ (Beer et al. 1985). To this extent, it can be seen simply as a
continuation of a well-established tradition.
What is distinctive about the recent literature is that it has increasingly
conceptualised such practices as forming coherent systems of mutually reinforcing
HR practices that work together to enhance organisational performance (Barney
1995; Becker & Gerhart 1996; Becker and Huselid 1998). That is, it has been
argued that the performance effect of a system of practices is greater than the
sum of the effects of each of the component practices.
The extent to which practices do actually form coherent and mutuallyreinforcing systems in organisations as opposed to being used on an ad hoc or
opportunistic basis remains the subject of considerable debate (for two divergent
views see: Huselid 1995; Wood & de Menezes 1998: 506). There is also debate

about whether there is ‘one best system’ for all organisations or whether different
systems of practices would be appropriate (Appelbaum et al. 2000). In spite
of this ongoing debate, the ‘one best system’ view appears to have gained the
ascendancy (Ramsay et al. 2000: 503).
There also remain some differences among scholars as to how best to characterise the practices under discussion. The term ‘high involvement’ is used by some
to emphasise the importance of enhanced opportunities for employees to make
decisions and exercise discretion (Lawler 1986). Others use the term ‘high
commitment’ to highlight the role of effective management in enhancing
employee commitment and thereby reducing the need for managerial
monitoring and control (Wood & de Menezes 1998). Nonetheless, there remains
considerable overlap between the practices associated with each approach. In
this study, the term HPWS is used on the basis that it is sufficiently broad to
encompass the range of practices emphasised by different approaches (Ramsay
et al. 2000: 502–7). The specific practices involved are discussed below.
Proponents of the integrated HPWS approach contend that its application
will enhance organisational performance (Becker and Huselid 1998). There is a
consensus among researchers who have sought to demonstrate a link between
HRM systems and organisational performance that the performance outcomes
arise chiefly because the HRM practices improve employee orientations to work,
which in turn makes them more productive (Appelbaum et al. 2000). Edwards

and Wright argue that the causal path assumed by most proponents of HPWS
is as follows:
Systems . . . are established; they influence workplace practice; employee attitudes
change, with increased satisfaction or commitment; there is a consequent effect on
behaviour; and this in turn feeds through to the performance of the work unit and
eventually the company . . . (2001: 570).

HIGH PERFORMANCE WORK SYSTEM PRACTICES

421

In spite of the wide acceptance of this assumed causal chain, most studies have
only measured associations between HPWS and organisational performance
rather than attempting to assess all the links in the chain (Legge 2001: 25; Edwards
& Wright 2001: 570). In particular, there are virtually no studies which have
explored employee responses to HPWS.1
Moreover, theoretical models of HPWS make the assumption that any performance gains are necessarily attributable to a positive impact on employees,
chiefly through increased autonomy, satisfaction and commitment. This overlooks two alternative possibilities: (i) improved organisational performance may
be driven by factors other than HRM practice; and (ii) improved organisational
performance may arise via negative outcomes for workers. One view, informed

by labour process theory and neo-Fordist analyses of developments in work
organisation (for example: Palloix 1976; Aglietta 1979; Prechel 1994), conceptualises HPWS as a managerial strategy to gain increased work effort from
employees. From this perspective, while it is possible that HPWS leads to
increased discretion, this is simply a new form of Friedman’s (1977) ‘responsible
autonomy’. Employees are granted minor increases in discretion as a means
to gain compliance with managerial aims, but these are outweighed by work
intensification, stress and insecurity (Ramsay et al. 2000: 505).
Given the centrality of employee responses in both mainstream and critical
conceptualisations of HPWS and performance, it seems extraordinary that so
little attention has been paid to employee outcomes. As Guest (1999) has observed,
if we are to develop a full understanding of the implications of labour management practices, then it is imperative that we take the views of employees
seriously.
In seeking to advance the debates on HPWS, the present study seeks to assess
the extent to which associations exist between a range of HPWS practices,
on one hand, and a series of employee outcomes on the other. By exploring
these associations, it is possible to move beyond the assumptions about
employee responses which are typically found in the literature. Empirical
inquiry and evidence of this type leaves us better placed to provide a measured
assessment of the competing claims of mainstream and critical accounts of
HPWS. The next section outlines the data which are used to undertake

analysis in an attempt to operationalise and measure HPWS practices and
employee responses.

AWIRS95

DATASET AND

HPWS

VARIABLES

The second Australian Workplace Industrial Relations Survey (AWIRS95) was
conducted in 1995.2 It replicated many of the items in the first AWIRS survey
(AWIRS90), but also included a survey of approximately 20 000 employees, drawn
from workplaces which were included in the main workplace survey, and sampled
so as to allow population estimates to be derived. The employee survey represents a significant advance on AWIRS90 in that it allows the systematic analysis
of the impact of organisational practices on employees. Data from the employee
survey can be linked to items from the workplace surveys, thereby providing a
means to explore relationships between workplace characteristics on one hand


422

THE JOURNAL

OF

I N D U S T R I A L R E L AT I O N S

September 2002

and employee experience on the other.3 The AWIRS95 data therefore lend
themselves very well to analysis of the links between HPWS practices and
employee outcomes.4
There is little agreement in the literature about precisely what practices should
be included in a measure of HPWS (Legge 2001: 25); that is, there is no
standard set of criteria for identifying practices as being consistent with
HPWS. There is, nonetheless, considerable commonality in the range of
practices included in significant studies of HPWS (Appelbaum et al. 2000; Ramsay
et al. 2000; Huselid 1995; MacDuffie 1995; Wood & Albanese 1995). The
approach adopted in the present study has been to be as inclusive as possible by

identifying, from the AWIRS95 dataset, as many indicators as possible of
workplace mechanisms which emphasise employer–employee communication,
employee involvement, training and development, performance incentives, job
security, ‘family-friendliness’, and participative forms of work organisation. This
process yielded 25 individual items, which are represented in Table 1. Some individual items are combined into scales, giving a total of 14 ‘high performance’
variables used in the analysis. All of the selected variables are consistent with
measures used in other studies. While it was not possible to capture all of the
phenomena included in other studies – a perennial problem with secondary data
analysis – the range of indicators covers the majority of practices analysed in
earlier studies.
An exploratory factor analysis of the 25 items indicated that very few of them
loaded on more than one component; that is, there were few underlying factors.5
In the few cases where items did form groups they were combined into composite scales (subject to reliability analysis indicating a sufficiently high Chronbach
Alpha). This lack of grouping of variables is of considerable substantive interest
in itself, since it suggests that, at least as measured by these items, HPW systems
do not seem to be empirically evident in Australian workplaces. This point
will not be pursued here, since it is outside the main focus of the paper, but it
suggests that the practices are being used in an opportunistic or ad hoc fashion,
rather than in a systematic way.6 In methodological terms, this result means it is
not possible to test associations between a system or systems of practices, on one
hand, and employee outcomes, on the other. Rather, the associations to be
explored are between a series of individual practices which are associated with
the HPWS approach to labour management. Accordingly, the term ‘HPWS
practices’ is used in the paper.
The AWIRS95 data set provides measures of employee attitudes to various
workplace issues, which can be used as indicators of responses to HPWS
practices. While they do not capture all of the outcomes of interest, there is
a sufficient range of indicators of both positive and negative outcomes to allow
the elucidation of key aspects of the debates about HPWS. These items appear
in Table 2.
A series of workplace and employee control variables was also drawn from
the dataset, on the basis that the outcome variables were likely to vary
depending on employee and workplace characteristics. These are set out in
Table 3.

Table 1

High Performance Work System practice variables*

Aggregated variables and sub-items
Use of appraisals for non-managerial employees (ca19) (0/1)

Mean (SD)
.68 (.47)
2.14
.91
.79
.49

(.92)
(.29)
(.40)
(.50)

Upward communication practices used (summed score)
Employee attitude surveys (cd1d) (0/1)
Regular meetings between employees and supervisors/managers (cd1f) (0/1)
Existence of suggestion schemes (cd1b) (0/1)
Existence of regular newsletters (cd1c) (0/1)

1.90
.23
.82
.29
.56

(1.11)
(.42)
(.38)
(.46)
(.50)

Equal opportunities practice
(0 = no written policy; 1 = written policy, no monitoring;
2 = written policy and monitoring)
Existence of formal written EEO policy (cl1) (0/1)
Monitoring of workforce composition (cl3c) (0/1)
Incentive schemes score
(0 = no PRP and no share ownership; 1 = PRP scheme which is not profit sharing
or share ownership scheme; 2 = PRP which is profit sharing or share ownership
scheme; 3 = PRP which is profit sharing and share ownership)
Existence of performance-related payment schemes (cm1) (0/1)
Performance pay based on profit-sharing scheme (cm2d) (0/1)
Existence of employee share-ownership scheme (0/1)

.90 (.75)

.67 (.47)
.35 (.48)
.53 (.72)

HIGH PERFORMANCE WORK SYSTEM PRACTICES

Downward communication practices used (summed score)
Communication with staff about workplace performance (cd24c) (0/1)
Communication with staff about staffing plans (cd24a) (0/1)
Communication with staff about investment plans (cd24b) (0/1)

.33 (.47)
.15 (.36)
.16 (.37)

423

424

Table 1

Continued

Aggregated variables and sub-items

77 (26)
0 (.70)
.68 (.47)
2.26 (1.05)
4.40 (.85)
3.37 (1.81)

(.52)
(.48)
(.17)
(.45)
(.34)
(.50)

I N D U S T R I A L R E L AT I O N S

.3751
.35
0
.29
.13
.47

OF

2.75 (1.15)

THE JOURNAL

Indicator of employment security (% of core workers; f12 subtracted from 100)
Training provision composite (mean of standardised [z] scores)
Formal off-the-job training for largest occupational group (LOG) in last
12 months (cc4) (0/1)
Proportion of supervisors with training in people management skills (ca9)
(1 = none; . . . 4 = all)
Preference for consulting directly with employees rather than unions (be1b)
(coded to scale of agreement 1 = strongly disagree . . . 5 = strongly agree)
Involvement of union delegates in change (bf7e)
(coded to scale 1 = not involved; . . . 5 = decided)
Involvement of employees in change
(coded to scale 1 = not involved; . . . 5 = decided)
Family friendly policies (summed score)
Non-managerial employees entitled to parental leave (ck4 and ck5) (0/1)
Workplace-based childcare (ck6a) (0/1)
Existence of autonomous work groups (ca17c) (0/1)
Existence of quality circles (ca17f) (0/1)
Existence of team building (ca17g) (0/1)

Mean (SD)

* Data weighted to effective sample size (rempwt1). n ranges from 382 to 1177 workplaces. PRP, performance related payment system.
September 2002

425

HIGH PERFORMANCE WORK SYSTEM PRACTICES

The variables outlined above provide the means to measure associations
between HPWS and a series of employee outcomes and, thus, provide an
empirical basis on which to advance our understanding of these issues.

EMPLOYEE

RESPONSES TO HPWS PRACTICES: ANALYSIS AND RESULTS

In order to elucidate associations between HPWS practices and employee outcomes, analysis proceeded in two stages. First, bivariate correlation analysis was
undertaken. The results of this analysis are presented in Table 4. The analysis
shows that all of the practices are statistically significantly associated with at least
some of the outcome variables. In particular, almost all are positively associated
with employee discretion, which is consistent with the claims made in most of
the HPWS literature. It should be noted, however, that the correlation coefficients
are very small indeed in all cases, suggesting very weak associations. Nonetheless,
Table 2

Employee outcome variables*

Outcome variable
Composite discretion (alpha = 0.825)
E28A: ‘In general, how much influence do you have over . . .
The type of work you do’ †
E28B ‘ . . . How you do your work’
E28C ‘ . . . When you start and finish work’
E28D ‘ . . . The pace at which you do your job’
E28E ‘ . . . The way the workplace is managed or organised’
E28F ‘ . . . Decisions which affect you at this workplace’
Insecurity
E29C ‘I feel insecure about my future here’ ‡
Effort
E29D ‘I put a lot of effort into my job’ §
Attitude to management
E30E ‘[are you satisfied with] The way management treat
you and others here’¶
Think about leaving job
E29F ‘I often think about leaving this job’ **
Stress
E29D ‘My job is very stressful’ ††
Job satisfaction
E30H ‘[are you satisfied with] Your job overall’ ‡‡

Mean (SD)

1.68 (1.06)
2.18
1.42
1.93
1.01
1.28

(.94)
(1.20)
(1.06)
(1.05)
(1.00)

.87 (.83)
1.87 (.38)
1.17 (.84)

.85 (.85)
1.15 (.79)
1.51 (.70)

* Data weighted to effective sample size. n = 14 067–14 503 (employees). † Each item has a
four–point response scale which I coded as follows: None (0); A little (1); Some (2); and A lot (3). The
items are summed to form a scale with a range 0–18. ‡ Agree (2); Neither Agree nor Disagree (1); Disagree
(0). § Agree (2); Neither Agree nor Disagree (1); Disagree (0). ¶ Satisfied (2); Neither Satisfied nor
Dissatisfied (1); Dissatisfied (0). ** Agree (2); Neither Agree nor Disagree (1); Disagree (0). †† Agree (2);
Neither Agree nor Disagree (1); Disagree (0). ‡‡ Satisfied (2); Neither Satisfied nor Dissatisfied (1);
Dissatisfied (0).

426
Table 3

THE JOURNAL

OF

I N D U S T R I A L R E L AT I O N S

September 2002

Workplace and employee characteristics variables*

Characteristic
Workplaces
Establishment size (no. employees)
Establishment age (years)
Establishment union density
(% total employees)
Australian ownership (≥51% )
Non-Australian ownership (≥50% )
Private sector
Production sectors †
Total sample (workplaces)
Employees (Gender)
Male
Female
Employees (Age)
50
Union members
Occupational group
Manager
Professional
Para-professional
Clerical
Trades
Sales and personal service
Plant and machine
Labourer
Total sample (employees)

Number

Mean %

1177
1112
1041

98
26
43

1015
158
841
253
1177

87
13
72
22


8101
6634

55
45

977
1459
1956
4059
3886
2350

7
10
13
28
26
16

1424
12 556

10
90

1839
7777
4460
480
7209

13
53
31
3
50

987
2491
1422
2768
1204
2101
1288
2232
14 735

7
17
10
19
8
14
9
15


* All analysis weighted to effective sample size. † Manufacturing, construction, electricity, water and
gas.

Table 4

Correlations between employee outcomes and ‘High Performance Work System’ (HPWS) practice variables†

Variable

2

3

4

5

6

7

1
–.199**
.119**
.304**
–.217**
–.010
.325**
.047**
.043**
.017*
.029**
.014
.072**
.020**
–.010
–.042**
.044**
.058**
.056**
.015
.039**


1
–.072**
–.294**
.289**
.152**
–.289*
.007
.016
.010
.044**
–.041**
.057**
.020*
–.046**
–.056**
–.005
.025**
.032**
.023**
.004



1
.102**
–.175**
.113**
.156**
.028**
–.017
.014
–.011
–.014
–.005
–.009
.014
.031**
.019*
.006
.010
.006
.007




1
–.416**
–.244**
.525**
.003
–.006
–.041**
–.067**
.031**
–.083**
–.033**
.058**
–.024**
–.009
–.050**
–.002
.000
.015





1
.274**
–.549**
.003
–.009
.014
.025**
–.012
.041**
–.006
–.013
–.010
.018*
.037**
.001
–.014
–.005






1

.035**
.030**
.046**
.044**
–.044**
.112**
.004
–.063
.045**
.013
.064**
.016
.015
–.010







1
.019*
.002
–.020*
–.016
.016
–.066**
.005
.039**
–.013
–.014
–.033**
–.003
.022**
.012

427

* P < 0.05, ** P < 0.01.
† Weighted to effective sample size [rempwt1]; n ranges from 5742 to 14 733.

HIGH PERFORMANCE WORK SYSTEM PRACTICES

Composite discretion
Insecurity
Effort
Attitude to management
Think about leaving job
Stress
Job satisfaction
Use of appraisals
Downward communications
Upward communications
Equal opportunities practice
Incentives
Employment security
Training
Direct employee consultation
Involve union in change
Involve employees in change
Family friendly policies
Autonomous work groups
Quality circles
Team building

1

Regressions of employee outcome variables on ‘High Performance Work System’ (HWPS) practice variables and control variables†

Independent variables

–.055**
.012
.013
.011
.001
.044*
.049**
.002
–.055**
–.108**
.038*
–.009
–.033
–.034
.001
–.047*
–.006

–.002
–.027
–.025
–.029
–.008
–.045*
–.137**
–.054**
.039*
.066**
–.007
.017
–.022
.003
–.016
–.011
–.020

.006
–.002
.017
–.023

.010
.022
–.026
.028

–.027
.030
–.012
–.009

.011
–.015
.005
–.005

.007
–.006
–.017
–.003
.018
–.044*
.039*
.031*
.031
.090**
–.032
–.023
–.013
.003
–.074**
.012
–.002

–.021
.010
.002
.004
–.004
.112**
.093**
–.041*
.026
–.046*
.047*
.039*
.012
.003
.039
–.005
.016

.000
–.025
–.006
–.021
–.028
.017
.053**
–.080**
.148**
.040*
.102**
.078**
.074**
.032
.061**
–.007
.013

.007
.003
–.006
.009

.024
–.026
–.009
–.006

–.008
.006
–.009
–.007

September 2002

–.048**
.022
.017
.021
–.002
.022
.066**
.011
–.072**
–.106**
.061**
.033
.032
.042*
.065**
.009
.036

Stress

I N D U S T R I A L R E L AT I O N S

–.030*
–.003
.033
–.009
.037*
–.023
–.021
–.081**
.085**
–.103**
.276**
.195**
.111**
.128**
.103**
.038**
.008

Effort

OF

Job satisf

THE JOURNAL

Step 1: Control variables
Establishment size
Establishment age
Foreign ownership
Union density
Production sectors
Female
Age
Casual status
Hours worked/week
Union member
Manager
Professional
Para-professional
Clerical
Sales and personal services
Trades
Plant and machine operator
Step 2: HPWS items
Use of appraisals
Downward communication
Upward communication
EEO

Dependent variables
Attn to mngmt Cont. com.
Insecurity

Discretion

428

Table 5

Table 5

Continued

Independent variables
Discretion

–.009
.001
.013
–.019
.003
.016
–.010
.015
.011
.035*
.155
.001
4229

–.012
–.016
–.002
–.017
.027
.032
–.014
–.006
.042**
.012
.024
.003
4229

Dependent variables
Attn to mngmt Cont. com.
Insecurity

–.002
–.039*
–.005
–.010
.018
.031
–.002
.025
.041
.033
.024
.004
4229

.004
.020
–.023
.020
–.020
–.033*
.003
–.001
–.037*
–.028
.023
.003
4229

–.010
.049*
.005
–.010
–.058**
.012
.028
.023
.005
–.017
.023
.003
4229

* P < .05, ** P < .01, *** P < .00.
† Regression coefficients are standardised; occupation dummy variables are referenced to labourers and related workers.
Job satisf., job satisfaction; Attn to mngmt, attention to management; Cont. com., continuance commitment.

Effort

Stress

–.006
–.016
–.005
.026
.040*
–.011
.000
–.005
.008
.016
.024
.000
4229

–.006
–.020
–.023
–.031
–.014
–.001
–.007
.010
.018
–.015
.061
.000
4229

HIGH PERFORMANCE WORK SYSTEM PRACTICES

Step 2 Continued
Incentives
Employment security
Training
Direct employee consultation
Involve unions in change
Involve employees in change
Family friendly policies
AWGs
Quality circles
Team building
Step 1 Adjusted r2
Step 2 Change in adjusted r2
n

Job satisf

429

430

THE JOURNAL

OF

I N D U S T R I A L R E L AT I O N S

September 2002

there are sufficient statistically significant associations to justify taking the
analysis to the next stage.
The second stage of analysis involved the use of ordinary least squares
regression. Earlier research has shown that employee outcome variables in
the AWIRS95 dataset vary systematically with a range of employee and
workplace variables (Harley 1999). Thus it is necessary to control for
these additional factors before drawing conclusions about HPWS and
employee outcomes. The regression analysis was conducted for each of the
dependent variables in two steps. The first step utilised the control variables
while the second step also included HPWS items in the model. This
approach allows an assessment of the extent to which the addition of the
HPWS variables increased the r2 value for each model. The results appear
in Table 5.
The regression analysis suggests, firstly, that very few of the HPWS practices
are statistically significantly associated with outcome variables when employee
and workplace characteristics are included in the model. Secondly, there is a
considerable number of statistically significant associations involving the control
variables and the outcome variables, suggesting that they are more important than
the HPWS practices in determining outcomes. This point is reinforced by the
fact that the addition of the HPWS variables to the models makes virtually no
difference to the r2 values, indicating that they account for little of the variation
in the independent variables. It should be made clear, however, that the r2
values for the models are small in any case, which reinforces the point that
employee outcomes are likely to be influenced by a wide and complex range of
factors.
Overall, the results suggest that HPWS practices make little difference to
employees. Of course, given the range of factors likely to impinge on employee
orientations to work, as well as the fact that the AWIRS95 data only allow us to
identify whether employees work in workplaces with the practices present, rather
than whether they are directly subject to them, we should not be surprised at the
results.
If we accept that associations between workplace practices and employee
responses are only ever likely to be modest, however, and thus treat any
statistically significant association as meaningful regardless of its magnitude,
what do the results suggest? In terms of positive employee outcomes, there were
a number of significant associations. Firstly, employee discretion is positively
associated with team building, which is consistent with claims that forms of
work organisation such as teams enhance employee discretion. Secondly, job
satisfaction is associated with quality circles, again consistent with claims about
the positive outcomes of such forms of work organisation. Thirdly, employee
attitudes to management were negatively associated with employment security,
suggesting that the greater the proportion of employees employed on a permanent
full-time basis, the more negative the employee perception of management. While
this finding is difficult to explain, it runs counter to the view that offering employment security will lead to more positive orientations for employees. Finally, continuance commitment was negatively associated with employee involvement in

HIGH PERFORMANCE WORK SYSTEM PRACTICES

431

change and with quality circles. This, again, is difficult to explain, but certainly
runs counter to the conventional wisdom.
Turning to negative employee outcomes, perceived insecurity is associated positively with security as measured by the percentage of employees on full-time
permanent employment contracts. This result seems counter-intuitive, but a
possible explanation is that where few employees have job security they do not
expect it, do not value it, and thus do not feel insecure. Insecurity is associated
negatively with the involvement of unions in organisational change, perhaps
reflecting the fact that union representation provides employees with protection
during periods of uncertainty. Secondly, work effort is positively associated with
the involvement of unions in change. This finding is surprising, in the sense that
we might expect union involvement to provide employees with protection from
work intensification, but perhaps it can be explained by the fact that employees
in workplaces with active unions are more aware of such issues. Turning to the
final negative outcome variable, stress, there are no statistically significant
associations.
Even if we opt for a liberal interpretation of the data and accept that any
statistically significant associations are reflective of genuine associations, the
results provide little evidence to support either mainstream or critical accounts
of HPWS. The evidence suggests that the practices tend not to be used as
‘bundles’ or systems. Moreover, it seems that few of the individual practices are
associated with employee outcomes.

DISCUSSION

AND CONCLUSION

What do these results tell us about the competing claims of mainstream and
critical accounts of HPWS? Like other recent studies of new approaches to
work organisation based on large-scale national level surveys (Harley 1999;
Harley 2001), the results provide little support for either account of HPWS.
To put it bluntly, HPWS practices do not appear to make much difference to
employees either way. Nor, on the basis of this analysis, do the practices seem
to form the kinds of coherent systems discussed in the literature.
Why is it that HPWS practices seem not to matter? There are a number of
possible explanations that relate to the AWIRS95 data itself. Firstly, as noted
earlier, the HPWS items used in the analysis tell us only whether the
employees worked in workplaces where the practices were present. Since
employees in the AWIRS95 survey were not necessarily representative of
all employees in their workplace, it is possible that some of the employees included
in the analysis were not subject to HPWS practices even if the practices
operated in their workplace. Secondly, it may be that the range of practices
and/or the range of employee responses used in the analysis were insufficiently
comprehensive to test the HPWS models. Finally, large-scale survey data of
this kind may be too blunt to apprehend the complex and subtle ways in which
HPWS influence employee outcomes.
While these limitations certainly cannot be discounted, the AWIRS95 data
are drawn from a very reliable survey and the items are very much consistent
with those used in other studies of HPWS (for example, Appelbaum et al. 2000).

432

THE JOURNAL

OF

I N D U S T R I A L R E L AT I O N S

September 2002

If we accept that the lack of associations cannot be explained entirely by
problems with the data, what then are we to make of the results?
Firstly, it may be that, like many managerial fads, HPWS practices are of
little practical significance to how employees do their work. That is, the
practices do not represent a major divergence from established work practices.
This point is reinforced by the relative importance of other factors. The variables most strongly associated with employee outcomes included occupation, age,
gender, employment status, working hours and union membership, a fact that
has been illustrated in other studies (Harley 1999; Harley 2001).
It is hardly surprising that these controls were associated with the outcome
variables. In particular, the marked association between occupation and discretion
is unremarkable, given that most organisations retain hierarchical structures and
that occupation is a proxy for employees’ positions in hierarchies (Boreham 1992).
Similarly, it is not surprising that longer hours are associated with higher stress,
that older workers are less likely to consider leaving their workplaces or that
those who are less positively oriented to their work are more likely to be union
members. The key point that the continued salience of these factors suggests is
that HPWS practices do not appear to represent a major determinant of the way
employees work and, hence, they are of little importance in determining employee
outcomes.
A second explanation for the results is that, at the time of the survey, the
practices did not appear to be used in a systematic fashion. That is, as suggested
above, the individual practices are relatively minor factors in the experience of
work and few workplaces use them as a coherent system which might have
the potential to transform work significantly. If, however, they were used in a
coherent fashion (Huselid 1995), they might indeed have a mutually reinforcing
effect, leading to changes in employees’ experiences of work. This possible explanation is supported by the results of Ramsay et al. (2000) which found evidence
of British employers using systems of practices as well as evidence of associations
between such systems and employee outcomes. If this explanation is correct, it
suggests that there remains a possibility that HPWS may have an impact
on employees, although the nature of that impact and its role in determining
organisational performance remains uncertain.
The findings of this study do not exclude the possibility that HPWS can have
an impact – positive, negative or mixed – on employee experiences of work.
Nonetheless, the results suggest that, on the basis of the most recent and
comprehensive survey of Australian workplaces, HPWS practices have little
impact on employees. This suggests that both advocates and critics of this
approach may well have made too much of the importance of this currently
fashionable approach to the management of labour.

ENDNOTES
1. For a recent exception, see Ramsay et al. (2000).
2. I acknowledge the assistance of the Department of Employment, Workplace Relations and Small
Business and the Social Science Data Archive in making the AWIRS95 data available.

HIGH PERFORMANCE WORK SYSTEM PRACTICES

433

3 It should be noted that the data were collected from employees in workplaces with a workforce of 20 or more, thereby being representative of the majority of Australian employees but
not the majority of workplaces (Morehead et al. 1997: 26).
4 For full details of AWIRS95, see Morehead et al. (1997).
5 Factor analysis was conducted on the variables at a workplace level, rather than employee level,
since the aim was to see if they grouped together as workplace practices.
6 The indicators of HPWS only tell us whether the practices were present in a given workplace,
rather than anything about which employees were involved. Thus, when they are linked to
employee data for the purposes of analysis they tell us whether a given employee worked in a
workplace with HPWS practices in place, rather than whether the employee was involved
in such practices. This limitation should be borne in mind when interpreting the results. For
example, the only commitment measure is of continuance commitment and there are no
measures available of normative or affective commitment (see Meyer & Allen 1997).

REFERENCES
Aglietta M (1979). A Theory of Capitalist Regulation. London: New Left Books.
Appelbaum E, Bailey T, Berg P, Kalleberg AL (2000) Manufacturing Advantage: Why High
Performance Work Systems Pay Off. New York: Cornell University Press.
Barney JB (1995) Looking inside for competitive advantage. Academy of Management Executive
9, 49–61.
Becker BE, Gerhart B (1996) The impact of human resource management on organizational
performance: progress and prospects. Academy of Management Journal 39, 779–801.
Becker BE, Huselid MA (1998) High performance work systems and work performance: a
synthesis of research and managerial implications. In: Ferris G, ed., Research in Personnel and
Human Resources Management, 16, pp. 53–101. Stamford, Conn.: JAI Press.
Beer MP, Spector P, Lawrence, D, Mills Q, Walton R (1985) Human Resource Management: A General
Manager’s Perspective. Glencoe: Free Press.
Boreham P (1992) Class and control: the labour process and the politics of production. In: Baxter
J, Emmison M, Western J, eds, Class Analysis and Contemporary Australia. Melbourne: Macmillan
Australia.
Edwards P, Wright M (2001) High-involvement work systems and performance outcomes: the
strength of variable, contingent and context-bound relationships. International Journal of Human
Resource Management 12 (4), 565–85.
Friedman A (1997) Industry and Labour: Class Struggle at Work in Monopoly Capitalism. London:
Macmillan.
Guest D (1999) Human resource management–the workers’ verdict. Human Resource Management
Journal 9 (3), 5–25.
Harley B (1999) The myth of empowerment: work organisation, hierarchy and employee
autonomy in contemporary Australian workplaces. Work Employment and Society 13 (1), 41–66.
Harley B (2001) Team membership and the experience of work in Britain: an analysis of the
WERS98 data. Work, Employment and Society 15 (4), 721–42.
Huselid MA (1995) The impact of human resource management practices on turnover, productivity,
and corporate financial performance. Academy of Management Journal 38, 635–72.
Lawler E (1986) High Involvement Management. San Francisco: Jossey-Bass.
Legge K (2001) Silver bullet or spent round? Assessing the meaning of the ‘high commitment
management’/performance relationship. In: Storey J, ed., Human Resource Management: A Critical
Text, Second edition. London: Thomson Learning.
MacDuffie JP (1995) Human resource bundles and manufacturing performance: organizational
logic and flexible production systems in the world auto industry. Industrial and Labor Relations
Review 48, 197–221.
Mayo E (1946) Human Problems of an Industrial Civilisation. New York: Macmillan.
Meyer JP, Allen NJ (1997) Commitment in the workplace: Theory, Research and Application. London:
Sage.
Morehead A, Steele M, Alexander M, Stephen K, Duffin, L (1997) Changes at Work: The 1995
Australian Workplace Industrial Relations Survey. Melbourne: Addison Wesley Longman.
Palloix C (1976) The labour process: from Fordism to neo-Fordism. In: Conference of Socialist
Economists, Pamphlet no. 1, The Labour Process and Class Strategies, pp. 46–67. London:
Stage 1.

434

THE JOURNAL

OF

I N D U S T R I A L R E L AT I O N S

September 2002

Prechel H (1994) Economic crisis and the centralisation of control over the managerial process:
corporate restructuring and neo-Fordist decision-making. American Sociological Review
59, 723–45.
Ramsay H, Scholarios D, Harley B (2000) Employees and high performance work systems:
testing inside the black box. British Journal of Industrial Relations 38 (4), 501–31.
Wood SJ (1999) Getting the measure of the transformed high-performance organization. British
Journal of Industrial Relations 37 (3) 391–418.
Wood SJ, Albanese M (1995) Can you speak of high commitment management on the shop floor?
Journal of Management Studies 32, 215–47.
Wood SJ, de Menezes L (1998) High commitment management in the UK: evidence from the
Workplace Industrial Relations Survey, and Employers’ Manpower and Skills Practices Survey.
Human Relations 51 (4), 485–515.