Description
Using British workplace data we examine the relationship between human resource management (HRM) and different forms of employee voice. After controlling for observable establishment characteristics, we find voice and HRM are positively correlated, but this positive association is confined to certain voice regimes.
esearch
PSI Research Discussion Paper 27
Employee Voice and Human Resource Management:
An Empirical Analysis using British Data
Alex Bryson, Paul Willman,
Rafael Gomez, Tobias Kretschmer
R
D
P
iscussion
apers
Policy Studies Institute
Policy Studies Institute, 2007
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system or transmitted in any form or by any means, electronic or
otherwise, without the prior permission of the copyright holder.
ISBN: X XXXXXX X
PSI Report No: 905
Policy Studies Institute
For further information contact:
Publications Dept., PSI, 50 Hanson Street, London W1W 6UP
Tel: (020) 7911 7500 Fax: (020) 7911 7501
PSI is a wholly owned subsidiary of the University of Westminster
esearch Discussion Papers R
Policy Studies Institute
Employee Voice and Human Resource Management: An
Empirical Analysis using British Data
Alex Bryson, Paul Willman, Rafael Gomez, Tobias Kretschmer*
*Alex Bryson is at the Policy Studies Institute and the Centre for Economic
Performance, Paul Willman and Rafael Gomez are in the Department of
Management at the London School of Economics and Tobias Kretschmer is
at the University of Munich and the Centre for Economic Performance. Data
and programs are available on request from Alex Bryson
([email protected])
Acknowledgements
We would like to thank Steve Wood and participants at the Voice and Value
Conference 2007 at LSE for valuable comments
1
ABSTRACT
Using British workplace data we examine the relationship between
human resource management (HRM) and different forms of employee
voice. After controlling for observable establishment characteristics, we
find voice and HRM are positively correlated, but this positive association
is confined to certain voice regimes. Previous research has found no
association between HRM and union voice. However, distinguishing
between union-only voice regimes and dual channel (i.e. union and non-
union) voice regimes reveals that union-only regimes have the lowest
incidence and intensity of HRM adoption while dual channel regimes
have the highest HRM incidence and intensity. The implications of these
findings for theory and practice are discussed.
2
The relationship between the adoption of human resource management (HRM)
techniques and the presence of unions has been a topic of considerable debate
on both sides of the Atlantic for a number of years. Early writers such as Guest
(1989:48) emphasized the individual focus of HRM as damaging to collective
organization, while others such as Kochan (1980) saw greater use of HRM
techniques as part of an overt ‘union substitution’ strategy. Kochan and
Osterman (1994), on the other hand, suggested HRM might work better in the
presence of unions. Protagonists in the union versus HRM debate have often
used different definitions of HRM and different measures of union presence, but a
recent and comprehensive analysis of British data by Machin and Wood
concluded that there is “no statistically significantly greater adoption of HRM
practices in non-union workplaces than in unionized ones” (Machin and Wood,
2005:216).
The authors qualify their conclusions in two ways. First, they speculate on the
possibility that their results are “uniquely British” and second they introduce the
possibility that there may be “different reasons for introducing HRM practices in
union and non-union environments” (2005:216). In this paper, we pursue these
arguments further by distinguishing between union and non-union voice,
considering their relationship to HRM separately and in combination. This proves
to be crucial in understanding the relationship between HRM and union voice in
the UK context. We argue that there may be something about the British context
which delivers this idiosyncratic relationship between HRM and voice: we
3
develop this argument below, but in brief it rests on the ability of employers in
Britain to combine union and non-union voice in ways that are not possible in the
US.
The structure is as follows. In the next section, we distinguish voice and HRM
and speculate about their relationship. We then describe some distinctive and
relevant features of the UK context. In the following section, we introduce our
hypotheses about HRM and voice incidence. Next, we describe our data and
methods and test our hypotheses. The final section summarizes and concludes.
Voice and HRM
Voice regimes are governance mechanisms for employment contracts
(Williamson, 1991; Bryson et al 2004). They exist where institutions or processes
are present to generate two-way communication between managers and
employees. Voice regimes can be direct or representative in nature and can be
delivered in a number of ways; via a union, through management led initiatives or
as part of some dual channel where union and non-union voice are both present
(Millward et al., 2000; Bryson, 2004). Our conception of employee voice is not, as
was Freeman and Medoff’s (1984), based solely on unionism. Rather, it is closer
to Hirschman’s (1971) conception, embracing any form of employee voice as
Hirschman embraced any form of consumer voice: it is the institutionalization of
two-way communication between employers and employees designed to reduce
4
transaction and exit costs for both parties. It is thus a contractual governance
mechanism with mutual benefits.
Voice regimes can also be accompanied by HRM practices. The specific
practices that form part of any given HRM system can be quite diverse, but
typically they involve managerial attempts to motivate and manage workers
through a series of workplace practices rather than through strict command and
control structures. HRM is a sophisticated set of control techniques designed to
yield benefits to the firm by use of management–initiated techniques for people
management designed to generate higher performance (Becker et al 2001).
However, the effectiveness of these HRM techniques in improving organizational
performance is contested (Godard 2004).
The HRM practices we consider here are derived from Pfeffer (1995) and Storey
(1992) and are discussed in detail below. They may be related to voice regimes,
but they are logically distinct. Somewhat like mobile phones and cameras, voice
and HRM may be encountered as integrated packages or separately. They may
confer different costs and benefits in different combinations and locations.
5
The UK Context
Britain has historically experienced a ‘voluntarist’ industrial relations system
granting the parties to industrial relations a considerable degree of freedom to
choose their preferred or agreed institutional arrangements (Clegg, 1979). The
two decades prior to the 1998 Workplace Employment Relations Survey (used
here and described in more detail below) broadly coincided with the Thatcher era
during which a further expansion of employer choice was generated by a series
of deregulatory measures which reduced legal support for trade union activity
(Willman and Bryson, 2007). Employers in the UK were therefore virtually
unconstrained in the period up to 1998 in their ability to mix union and non-union
voice at the establishment level. This has made it possible for employers to
supplement union with non-union voice without terminating relationships with
unions (Willman and Bryson 2007). This situation contrasts starkly with the USA
where the legal system proscribes certain non-union voice and HRM practices in
a union environment to preserve unions’ “sole agent” status (LeRoy, 2006).
In spite of this variety in voice regimes, UK studies on the co-existence of HRM
and voice focus almost exclusively on the links between HRM and union voice. It
is usually assumed that HRM may substitute for unions since it offers alternative
solutions to worker problems, thus challenging the solutions offered by unions
and potentially reducing the incentive for union organizing. There is evidence
that HRM policies are associated with higher job satisfaction and a lower
intention to quit (Guest and Conway, 1999) and that they reduce workers’
6
expressed problems at the workplace in both Britain and the US, potentially
limiting workers’ desire for union solutions (Bryson and Freeman, 2006).
However, studies generally find that unionized workplaces differ little from non-
unionized workplaces in terms of the incidence of HRM (Wood, 1996; Wood and
de Menezes, 1998; Machin and Wood, 2005) and – unlike in the USA – HRM
practices do not reduce workers’ desire for unionization (Bryson and Freeman
2006).
1
Furthermore, one study using the same data employed in this paper
found that a positive link between HRM practices and labor productivity was
confined to unionized workplaces(Bryson et al., 2005), suggesting that
unionization and HRM may even be complementary. Unions in the USA may
challenge employer efforts to introduce HRM, aware of the potential for HRM to
undermine worker support for unionization, whether intentionally or
unintentionally (Fiorito, 2001; LeRoy, 2006). These concerns, however, may not
be so pertinent in the UK.
Of particular relevance for our work is that all of the studies above do not take
account of the heterogeneity of voice regimes in the UK. In particular, they do
not distinguish between non-union, union and dual channel (mixed union and
non-union) voice regimes, presenting instead the effects of unionization on HRM
across pure and mixed regimes, usually proxied by a dummy variable identifying
1
Cully et al. (1999: 110-111) note that the higher incidence of HRM practices in unionized
workplaces is largely due to their greater likelihood of being large and being in the public sector.
Machin and Wood (2005) also present bivariate analyses indicating higher HRM incidence in
unionised workplaces but, again, this association disappears when controlling for other factors.
7
workplaces in which at least one union is recognized for pay bargaining. In fact,
non-union forms of voice have become increasingly prevalent.Moreover, where
union voice is present, it often co-exists with non-union voice (Bryson et al.,
2004). Among unionized workplaces with 10 or more employees, six-in-seven
had some form of non-union voice as well (see below).
As in the USA, the union sector has been shrinking in the UK during the past
three decades. By 1998, around half the workplaces in the UK had voice
mechanisms that were exclusively non-union. These include direct and
representative voice channels which are instigated by the employer, sometimes
at the request of employees.
2
To our knowledge, there is no empirical evidence
on the links between these various forms of voice and HRM either in the UK or
the USA, in part because many authors’ measures of HRM subsume voice
elements in their estimation of HRM (e.g. Machin and Wood, 2005).
3
Hypotheses
For our hypotheses, we adopt the following simple notation: Let HRM (X) be the
extent of HRM use given voice regime X, where X can be: A (absence of
2
EU legislation can require the setting up of works councils at the apex of larger organizations,
but this does not affect workplace-level arrangements in the UK. The Information and
Consultation Directive which came into effect in April 2005, may have some effect on voice
arrangements at workplace-level but this was not in place in the late 1990s.
3
There is one exception. Fenton-O’Creevy and Wood (2005) distinguish between union voice and
direct and representative forms of non-union voice. Using data on multinational companies with
headquarters in Britain they find no support for the proposition that direct-voice only regimes have
a higher incidence of high-commitment work practices than other voice regimes.
8
voice), U (use of union voice), and N (non-union voice). U and N can be
present concurrently.
A management control technique like HRM and a contractual governance
mechanism like voice may be complementary or substitutes. If HRM practices
delivered everything voice representation did (and vice versa), the two would be
substitutes. In fact, since HRM techniques emerged much later than unions and
other forms of voice representation, one could argue that HRM is a natural
“successor” and firms will switch from voice to HRM eventually.
On the other hand, if the effectiveness of HRM is enhanced by the simultaneous
presence of voice at a workplace (e.g. because voice reduces costly worker exit
or increases the flow of productivity-enhancing information), the two would be
complements and we would expect them to appear together. Concurrent use of
voice and HRM could also originate from a common factor driving the adoption of
both – for example managerial quality – that enables a firm to cope with
organizational and managerial innovations more easily. In this case, we would
again expect the two practices to appear in conjunction– a workplace that draws
positive net benefits from one will also draw positive net benefits from the other.
We view the complementary argument(s) as more compelling and therefore
hypothesize:
9
H
1
: HRM will be more prevalent in establishments with (any form of) voice than in
those without voice (min[HRM(U), HRM(N), HRM(U,N)] > HRM(A)).
Conditioning on workplaces with voice, HRM, may sit better with some forms of
voice than others. HRM may substitute for union voice where HRM generates
outcomes that reduce employee demand for unionization. The presence of
union-only voice may signal union success in monopolizing worker voice at the
workplace and may imply the potential to block HRM too if the union is not
persuaded of the ‘high performance’ ethos which drives HRM. In addition, union-
only voice may imply a reliance on collective rather than individual forms of
employer-employee engagement that might exclude HRM. Reliant on a narrower
definition of union only voice (rather than union presence) we hypothesize,
amending Machin and Wood (2005), that among voice regimes, HRM will be
lowest in union-only voice regimes.
Non-union voice only exists in the UK at workplace-level where the employer has
chosen to invest in its provision (Bryson et al., 2004).
4
The fact that the employer
has chosen to implement non-union voice implies that the employer may also
invest in HRM to obtain a competitive advantage as described by Pfeffer (1995).
The presence of non-union voice alongside union voice at a workplace is
indicative of one of two scenarios, both of which are conducive to HRM. It may
either signal union weakness which leaves the employer largely unconstrained in
4
It is plausible that direct voice emerges spontaneously in the smallest of workplaces due to
worker proximity. However, these workplaces are absent from our data.
10
mixing voice varieties with HRM, or it may imply a ‘mutual gains’ environment in
which unions are willing to use what Freeman and Medoff (1984) termed their
‘voice face’ to elicit productivity improvements. Therefore, we hypothesize;
H
2
: HRM will be higher in workplaces with non-union voice than in those with
union only voice (min[HRM(N), HRM(U,N)] > HRM(U)).
So far our hypotheses are consistent with the proposition that HRM incidence will
increase in an ordinal fashion from no-voice to union voice to non-union voice
regimes.
Our main departure from previous work in this field is to stress the heterogeneity
of union voice. Specifically, we identify dual voice as empirically more common
than union only voice (Bryson et al 2004). If, as we hypothesize, union only voice
environments restrict HRM and non-union only voice environments are a more
favorable habitat, we must consider how dual voice and HRM will coexist.
Whether a workplace is unionized is, in part, an accident of history: the earlier
cohort of workplaces initially adopted union-voice whereas newer workplaces
adopted non-union only voice (Machin, 2000; Millward et al., 2000; Willman et al.,
2007). Yet very few unionized workplaces have de-recognized their unions in
Britain; instead, they have supplemented union with non-union voice (Millward et
al., 2000; Kersley et al., 2006).
11
Dual voice regimes may have emerged in different ways. For some firms, non-
union voice was added to union only regimes in response to employers
perceiving deficiencies in union-only voice provision. The reverse is also
possible, where employees pressed for unionization in addition to non-union
consultation frameworks. In the UK, it is likely that the former has been more
common.
However, if union voice acts as a drag on the introduction of HRM, employers
may switch to non-union voice. Where there were no constraints on doing so,
adding non-union voice mechanisms to union ones is also a risk-averse strategy
for overall voice improvement (Bryson et al 2004). It is thus likely that in
continuing dual regimes, no such drag operates, and we therefore hypothesize;
.
H
3
: There will be no difference in HRM incidence and intensity between dual
channel and non-union only voice regimes (HRM(U, N) = HRM(N)).
To summarize, our hypotheses on the incidence of voice and HRM can be
expressed in the following ordinal ranking:
HRM(U, N) = HRM(N) > HRM(U) > HRM(A),
12
where the first equality represents H
3
, the first inequality H
2
, and the second
inequality H
1
.
The Data and Context
Our data are the British Workplace Employment Relations Surveys 1998
(WERS98), a large-scale survey of industrial relations in British establishments in
the public and private sectors. The key features of these data are described
elsewhere (Millward et al., 2000, 3-10; 248-55). Our analysis is based on data
collected from human resource managers responsible for workplace industrial
relations which contain the voice-related variables and HRM items needed for our
analysis. All observations are weighted by the inverse of the workplace’s
probability of selection for the survey. With these weights, our analyses provide a
representative portrait of workplaces in Britain with 10 or more employees in
1998. We now define our measures.
Voice is defined as the presence of two-way forms of (representative or direct)
communication between workers and management. In our data, the set of voice
measures is as follows.
1. union recognition
2. union representatives on or off site
3. a joint consultative committee meeting at least once a month
4. non-union representatives on site
13
5. problem solving groups
6. regular meetings between management and employees which allow for
two-way communication
7. team briefings that occur at least once a month and devote time to
employees’ questions/views.
Items 1-2 measure union voice. Items 3-7 measure non-union voice. No-voice
workplaces are defined by the absence of all.
The context and timing of the data collection are relevant to understanding the
voice measures. In 1999 the government enacted legislation which could require
employers to recognize trade unions if the majority of workers so wished, hence
our reliance on the 1998 dataset. We thus have four categories of voice in our
analysis: no voice, union voice only, non-union voice only and a mix of union and
non union voice (termed ‘dual’). The incidence of these regimes in our data was
17, 5, 48 and 30 percent respectively.
The HRM measures are broadly based on Pfeffer (1995). The measure is a count
of practices identified by Pfeffer, supplemented by other aspects of human
resource management identifiable in the literature. These dimensions are as
follows:
14
1. selectivity in recruiting: where manager says skills, qualifications,
experience and motivation are all important factors in recruiting new
employees
2. job security: policy of guaranteed job security for at least some employees
3. incentive pay: profit pay, performance related payments or cash bonuses
4. employee share ownership scheme
5. information sharing: management shares information on investment,
financial position of the organization or staffing
6. ‘empowerment’: core employees have a lot of control over variety in their
work, discretion over how they do their work or control over the pace at
which they do their work
7. self-managed teams: core employees work in teams that are able to
appoint their own leaders, jointly decide how work is done, or have
responsibility for specific products or services
8. on-going training: on-going training is one of the main methods by which
core employees are made aware of their job responsibilities
9. cross-training: at least some core workers are formally trained to do jobs
other than their own
10. symbolic egalitarianism: core employees have standard contracts for all
non-pay terms and conditions of employment
11. promotion from within: internal applicants given preference when filling
vacancies.
15
In addition to these items identified by Pfeffer the score includes two other items:
12. an indicator that the workplace has a formal strategic plan, strategic
planning being a key component of HRM (Storey, 1992)
13. the existence of a widespread appraisal system, that is, where at least
80% of core employees are formally appraised.
A score of 13 denotes affirmative answers to each of these questions. Lacking
any one of these HRM variables would give an establishment a score of 12 and
so on.
5
This summative measure of HRM differs from others in the literature,
including Machin and Wood (2005) who, because they use earlier surveys as
well as the 1998, must operate with a smaller set of HRM measures. In addition,
they include items such as the existence of a joint consultative committee that we
define as voice. In our analysis, measures of HRM and of voice are clearly
discrete sets.
Empirical Findings
Table 1 reports the incidence of voice by workplace characteristics in 1998. The
percentage of establishments with some form of voice ranges from lows of 64
and 67 percent in single-establishments and family owned operations
5
Clearly, this is the simplest way of defining HRM intensity. Assigning different weights to each of
these elements or using clusters of HRM practices does not fundamentally alter our conclusions
(results available upon request).
16
respectively to a high of 100 percent in health establishments. The overall
sample average for all establishments is 83 percent.
[Table 1]
The first indication of a positive correlation between voice and HRM is the higher
incidence of voice in workplaces with high HRM scores (94 per cent among those
with an HRM score of 9+) compared with workplaces with lower HRM scores (80
per cent among those with scores of 8 or under).
6
This relationship, which
supports Hypothesis 1, is illustrated graphically in Figure 1 which shows the HRM
score distribution for no-voice workplaces lies to the left of that for workplaces
with voice.
[Figure 1]
Table 2 reports descriptive statistics on the use of HRM practices, overall and
disaggregated by workplace characteristics. In 1998, establishments used an
average of 7 out of 13 HR practices. This overall figure differs considerably
across workplaces with small, single-establishment, family owned, private sector
and middle-aged establishments using fewer HRM practices than larger, non-
family owned, public sector, and younger or older establishments.
[Table 2]
Table 2 presents HRM scores by workplace characteristics. It confirms that
establishments with voice use more HRM practices. However, HRM intensity
varies significantly across voice regimes. In accordance with Hypothesis 2,
6
Since 1984, non-union only regimes have increased three-fold in the population of workplaces
with 25 or more employees. Union-only regimes have witnessed a similarly impressive decline
over the same period (Willman et al, 2007).
17
union-only voice has the lowest HRM score amongst voice types. HRM scores
are higher in dual channel and non-union only voice regimes.
The distribution of HRM scores by type of voice regime is shown in Figure 2. The
panels are ranked in terms of overall mean in the HRM score. It is clear that
union-only voice is associated with a compressed distribution of HRM scores
(panel A). There are fewer outliers within the union voice type, with most
workplaces located between 5 and 8 practices. In each panel the dotted line
represents the overall frequency of scores.
[Figure 2]
Table 3 shows the incidence of each HRM practice by workplaces with and
without voice. Almost every practice is more likely to be found in a workplace
that also has formal voice. The exception is the use of incentives, perhaps
indicating a more transactional relationships in no-voice regimes.
[Table 3]
Table 4 presents the same set of 13 practices, only now we look at patterns
within the voice sector. Establishments with dual voice have a greater likelihood
than others of adopting almost all HRM practices, except for the compensation
related (extrinsic) components, incentive pay and employee ownership schemes,
where establishments with non-union only voice are more likely to use these two
practices. The biggest disparity between union-only voice regimes and other
voice regimes is in relation to formal appraisal, internal promotion, selective
recruitment and share ownership. The relative absence of these HRM practices,
18
together with the high incidence of information sharing and, to some degree,
participation and empowerment, helps account for the compression of HRM
practices in this sector.
7
[Table 4]
The descriptive relationships presented here appear to confirm the three
hypotheses. However, it is possible that the relationships identified are driven by
observable differences between voice and no-voice workplaces which are also
correlated with the incidence of HRM. For instance, Table 1 showed that the use
of voice increases with establishment size, something that may also influence the
costs and benefits of returns to HRM. We therefore test if these relationships
change once we control for observable workplace characteristics. The results are
presented in Table 5.
[Table 5]
Table 5 confirms many of the bivariate relationships identified in Table 1, such as
the link between voice and establishment size. It also, confirms an independent,
statistically significant relationship between HRM use and the presence of voice
(row 1).
Table 6 reports estimates of the association between varieties of voice and HRM
adoption and intensity. In the first column are the results of a dummy variable,
which uses a cut-off of 9 HRM practices to denote a high HRM score
8
. We then
7
It is not surprising to see the high prevalence of HRM practices such as information sharing in
union workplaces since these are often the object of union bargaining.
8
Other cut-offs were tried and results were not significantly different; results available upon
request.
19
explore the robustness of these results to a change in dependent variable, where
the HRM intensity score, as defined previously, is used. In column (2) the mean
score was 6.89 and as our results are qualitatively the same in both columns, we
will confine our interpretation of results to the more intuitive HRM score results.
[Table 6]
There is strong support for our hypotheses regarding links between voice and
HRM (Table 6, row 1). Treated as a single entity, voice and HRM were positively
correlated (Table 5), supporting our first hypothesis. However, across voice
regimes, HRM is least prevalent in union-only regimes (Table 6), confirming
Hypothesis 2. The difference between union-only regimes and those containing
non-union voice is statistically significant. Indeed, controlling for observable
differences across workplaces, HRM is no more prevalent in union-only regimes
than it is in no voice regimes. Thus, in accordance with Hypothesis 3, it is the
voice regimes with some non-union voice present – either in isolation or in
combination with union voice – where HRM is most apparent.
Descriptive analyses indicated that dual channel voice regimes had a greater
HRM incidence than non-union only voice regimes, However, although the HRM
adoption coefficients are largest in dual channel regimes, these are not
significantly higher than the coefficients in non-union only regimes in line with
hypothesis 3.
9
This suggests that the differences in the descriptive analysis were
9
The 0.07 difference between non-union only and dual channel coefficients in the HRM score
model has a t-statistic of 0.29.
20
partly attributable to observable differences between establishments with these
regimes that are controlled for in the regression.
For our other controls we find results in line with much of the associated literature
on HRM adoption. Specifically, we find that:
• Workplace size has a positive and significant effect on the number of HRM
practices adopted;
• Workplaces that are part of some multi-establishment network also adopt
more HRM practices, with the number of those practices increasing with
network size;
• Age of establishment displays an inverted-U shape with workplaces aged
3 to 4 years being the most intensive users of HRM, while those aged
under 3 and more than 21 years have lower scores;
• Organizational affiliations generally increase the use of HRM up to a point,
as the highest category (4 affiliations) seems to make little difference.
Workplaces with three organizational affiliations have the highest use of
HRM.
Conclusions
In the last two decades of the 20
th
Century unionization was in decline in Britian.
Voice regimes, on the other hand, were not in decline since employers were
substituting union voice with non-union voice (Millward et al., 2000; Bryson et al.,
21
2004). At the same time, HRM practices were on the increase, although they
remained far from ubiquitous. In their important contribution to the literature on
union voice and HRM, Machin and Wood (2005) argued that there was no causal
link between declining unionization and rising HRM. However, their time-series
data means their study relies on a partial measure of HRM which conflates HRM
and voice. Furthermore, in keeping with the rest of the literature, they take no
account of heterogeneity in voice regimes. This is the first paper to draw attention
to the heterogeneity of union and non-union voice regimes and, in particular, the
importance of dual channel voice regimes in understanding the links between
HRM and unionization. We find that, although union-only voice regimes and
HRM do not easily co-habit, dual channel voice and HRM do. This may go some
way to explaining why despite union membership decline, large scale union de-
recognition has not occurred in Britain; dual channel employers simply did not
need to get rid of unions in order to introduce HRM. There is also no evidence of
a switch to no-voice with HRM. Most of the decline in individual unionization was
brought about by new workers entering new workplaces that had simply by-
passed the union voice solution.
By using a richer conception of HRM than the previous literature and by
analyzing both the incidence and intensity of HRM, we have shown that the
intensity of HRM, its distribution and the relative frequency of its components
differ across the union and non-union sectors and within the union sector. In
particular, financial incentives are less prevalent in the union sector whereas self-
22
managed teams and information sharing are more prevalent in the union sector.
Furthermore, dual channel regimes have more intensive HRM than union-only
regimes and there is some compression in the distribution of HRM in the union-
only sector compared with the dual channel sector. More ‘permissive’ legal and
public policy arrangements obtain in the UK than the USA, making employer and
worker choices more salient in determining voice and HRM outcomes. Although
we can not directly answer Wood and Machin’s (2005: 216) question regarding
the motivations for the use of HRM in the union and non-union sectors, we argue
that these patterns are consistent with explanations which emphasize the
heterogeneity of union voice regimes.
The greater part of the union voice sector in the UK is also to a material degree
non-union and this, we argue, explains much of the compatibility between unions
and HRM. The clearest complementarity is between non union voice and HRM.
Union-only voice regimes are more similar to no-voice regimes in their
relationship to HRM. In this sense, the UK case may not be a good basis for
generalization. Where employers face a straight choice between union or non-
union voice (as in the US), the implications for HRM adoption and use may be
very different.
23
References
Becker, Brian, Mark Huselid and Dave Ulrich. 2001. The HR Scorecard; Linking
People, Strategy and Performance, Boston: Harvard Business School Press,.
Bryson, Alex. 2004. ‘Managerial responsiveness to union and non-union worker
voice in Britain’, Industrial Relations: A Journal of Economy and Society, Vol. 43,
No. 1, pp. 213-241
Bryson, Alex, John Forth and Simon Kirby. 2005. ‘High-performance practices,
trade union representation and workplace performance in Britain’, Scottish
Journal of Political Economy, 53, 3: 451-491
Bryson, Alex and Richard Freeman (2006) Worker Needs and Voice in the US
and the UK, NBER Working Paper No. 12310, Cambridge, Mass.
Bryson, Alex, Paul Willman and Rafael Gomez. 2004. “The End of the Affair? The
Decline in Employers' Propensity to Unionize.” In John Kelly and Paul Willman
(eds.), Union Organization and Activity, Routledge, London, pp.129-149.
Clegg, Hugh A. 1979 The System of Industrial Relations in Great Britain, Oxford:
Blackwell.
Felstead, Alan and David Ashton. 2000. “Tracing the link: organisational
structures and skill demands”. Human Resource Management Journal, 10,3, 5-
21.
Fenton-O’Creevy, M. and Wood, S. (2005) ‘Direct Involvement, Representation
and Employee Voice in UK Multinationals in Europe’, European Journal of
Industrial Relations, Vol. 11, No. 1, 27-50
Fiorito, Jack. 2001. “Human Resource Management Practices and Worker
Desires for Union Representation”, Journal of Labor Research, Vol. 22, No. 2, pp.
335-54
Freeman, Richard and James Medoff. 1984. What Do Unions Do? New York:
Basic Books:
Godard, John. 2004. “A Critical Assessment of the High-Performance Paradigm.”
British Journal of Industrial Relations, Vol. 42, No. 2, pp. 349-378.
Guest, David E. 1989. “Human Resource Management; Its Implications for
Industrial Relations and Trade Unions.” In John Storey (ed) New Perspectives on
Human Resource Management. London; Routledge pp. 41-55.
24
Guest, David and Neil Conway. 1999. “Peering into the Black Hole: The
Downside of the New Employment Relations in the UK”. British Journal of
Industrial Relations, Vol. 37, No. 3, pp. 367-90
Hirschman, Albert. 1970. Exit, Voice and Loyalty. Cambridge, Harvard University
Press.
Kochan, Thomas A. 1980. Collective Bargaining and Industrial Relations .
Homewood, Ill: Richard D. Irwin.
Kochan, Thomas A., and Paul Osterman. The Mutual Gains Enterprise. Boston:
Harvard Business School Press.
LeRoy, Michael. H. 2006 ‘The power to create or obstruct employee voice: does
US public policy skew employer preference for “no voice” workplaces?’ Socio-
Economic Review, 4, pp. 311-319
Machin, Stephen. 2000. ‘Union decline in Britain’, British Journal of Industrial
Relations, Vol. 38 No. 4: pp.631–45.
Machin, Stephen, and Stephen Wood. 2005. “Human Resource Management as
a Substitute For Trade Unions in British Workplaces” Industrial and Labor
Relations Review Vol. 58, No. 2 pp.201-18.
Millward, Neil, Alex Bryson and John Forth. 2000. All Change at Work, London:
Routledge..
Pfeffer, Jeffrey. 1995. “Producing sustainable competitive advantage through the
effective management of people,” Academy of Management Executive, Vol.9,
No. 1, pp. 55-72
Storey, John. 1992. Developments in the Management of Human Resources.
Oxford: Blackwell.
Williamson, Oliver. E. 1991. ‘Comparative Economic Organization: the Analysis
of Discrete Structural Alternatives’, Administrative Science Quarterly, Vol. 36,
Issue 2, pp. 269-298.
Willman, Paul and Alex Bryson (2007) ‘Union Organization in Great Britain’,
Journal of Labor Research, Vol. 28, 1: pp93-115
Willman, Paul, Alex Bryson and Rafael Gomez. 2006 ‘The Sound of Silence:
Which Employers Choose ‘no voice’ and why?’, Socio-Economic Review, 4: 283-
299
25
Willman, Paul, Alex Bryson and Rafael Gomez. 2007. “The Long Goodbye: The
Fall of Union Voice In Britain” International Journal of Human Resource
Management forthcoming
Wood, Stephen. 1996. “High Commitment Management and Unionization in the
UK”. International Journal of Human Resource Management, Vol. 7, No. 1, pp.
41-58
Wood, Stephen, and Maria Albanese. 1995. “Can we speak of high commitment
management on the shop floor?” Journal of Management Studies, Vol 32 No. 2
pp215-47
Wood, Stephen and Lilian De Menezes. 1998. “High Commitment Management
in the UK: Evidence from the Workplace Industrial Relations Survey and
Employers’ Manpower and Skills Practices Survey”. Human Relations, Vol. 51,
No. 4, pp. 485-515
26
Table 1: Incidence of Voice (%) by Selected Workplace Characteristics, 1998
1. All Workplaces 83.1
2. By Sector
Public 98.9
Private services 80.8
Private manufacturing 65.2
3. By Establishment size (employees)
25-49 79.5
50-99 82.7
100-199 89.9
200-499 93.5
500-999 95.1
1000 plus 96.5
4. By Ownership
Foreign 72.5
Domestic 84.2
5. By Establishment
Single 70.7
Multi-establishment 87.7
6. By Size of Multi-Establishment Network*
Single 71.8
2-10 83.9
11-50 83.3
50+ 92.5
7. By Set-up date
Pre 1980 88.5
Post 1980 78.5
8. By Decade of Set-up date
Pre 1980 --
1980s --
1990s --
9. By Establishment Age*
<3 years 80.0
3-19 years 82.1
20+ years 86.1
10. By Industry
Manufacturing 68.5
Electricity, gas and water 99.9
Construction 70.5
Wholesale and retail 83.3
Transport + Communication 86.0
Financial services 76.8
Other business services 93.6
Public administration 99.9
Education 98.7
Health 99.9
Other community services 98.1
11. By HRM Score*
High [9-13] 93.6
Low [0-8] 80.0
12. By Ownership Structure*
Family owned/controlled 66.5
Other 88.8
13. By Employer Association Status
Yes 87.9
No 82.9
14. By Number of Organizational Affiliations*
None 82.9
One 82.1
Two 85.3
27
Three 87.9
Four 99.4
15. By Franchise Status*
Franchisee 91.7
Non-franchisee 83.1
Number of Observations
1954
28
Table 2: Average HRM Score (score 1 to 13) actual and normalized by selected
workplace characteristics, 1998.
Average
HRM Score
Normalised
Z-Score
1. All Workplaces 6.9 0.09
*
2. By Sector
Public 7.45 6.22
Private 6.71 -2.00
3. By Establishment size (employees)
10-24 6.6 -3.22
25-49 6.94 0.56
50-99 7.19 3.33
100-199 7.59 7.78
200-499 7.93 11.56
500+ 8.05 12.89
4. By Ownership
Foreign 7.15 2.89
Domestic 6.89 0.00
Joint Venture 5.81 -12.00
5. By Establishment
Single 5.81 -12.00
Multi-establishment 7.39 5.56
6. By Size of Multi-Establishment Network
Single 5.83 -11.78
2-10 7.03 1.56
11-50 6.87 -0.22
51+ 7.98 12.11
7. By Organization Size
Small [<50] 6.60 -3.22
Large [51+] 7.92 11.44
8. By Establishment Age
<3 years 6.44 -5.00
3-19 years 7.20 3.44
20+ years 6.64 -2.78
9. By Industry
Manufacturing 6.01 -9.78
Electricity, gas and water 9.19 25.56
Construction 5.11 -19.78
Wholesale and retail 7.31 4.67
Hotels and restaurants 6.56 -3.67
Transport + Communication 7.10 2.33
Financial services 8.65 19.56
Other business services 7.14 2.78
Public administration 7.17 3.11
Education 7.42 5.89
Health 7.15 2.89
Other community services 5.56 -14.78
10. By Ownership
Family owned/controlled 5.93 -10.67
Other 7.22 3.67
11. By Employer Association Status
Yes 6.80 -1.00
No 6.88 -0.11
12. By Number of Organisational Affiliations
0 6.88 -0.11
1 6.67 -2.44
2 6.96 0.78
3 7.08 2.11
4 7.87 10.89
13. By Franchise Status
29
Franchise 6.77 -1.33
Non-franchise 6.90 0.11
14. By Type of Voice I
No Voice 5.71 -13.11
Union 6.41 -5.33
Non-Union Only 6.86 -0.33
Dual Channel 7.51 6.89
15. By Type of Voice II
No Voice 5.71 -13.11
Representative Only 6.28 -6.78
Representative and Direct 7.63 8.22
Direct Only 6.75 -1.56
Number of Observations 1929
Source: Data are for Britain using WERS data 1998. *The figure 0.09 is the standard deviation
for the sample.
30
Table 3: Incidence of HRM practices by workplaces with and without formal voice (%),
1998.
By Presence of Voice at workplace
HRM Practices
No
Yes
All Workplaces
1. Presence of Formal Strategic Plan
47.1
78.9
73.7
2. Guaranteed Job Security
7.2 10.9 10.3
3. Selective Recruitment
46.4 54.8 53.4
4. Employee Ownership Scheme
11.2 15.4 14.7
5. Presence of Incentive Pay
53.3 52.7 52.8
6. Ongoing Training
57.5 73.2 70.6
7. Internal “Symbolic” Equity
20.0 46.6 42.2
8. Internal Promotion
24.2 25.8 25.6
9. Formal Appraisal System
37.4 56.7 53.5
10. Information Sharing
57.5 73.2 79.9
11. Self-Managed Teams
62.6 77.5 75.0
12. Job Enrichment
61.2 70.1 68.6
13. Participation and Empowerment
69.9 59.5 61.1
Number of Observations 346
(0.17)
1742
(0.83)
2088
(1.00)
Notes: Source: Data are for Britain using WERS data 1998. Numbers in parentheses refer to
sample proportions.
31
Table 4: Incidence of HRM practices by type of workplace voice (%), 1998.
Type of Formal voice
HRM Practice
Non-Union
Only
Dual
Channel
Union
Only
1. Presence of Formal Strategic Plan
74.4
87.7
72.7
2. Guaranteed Job Security
6.9 18.4 4.0
3. Selective Recruitment
48.5 67.2 41.1
4. Employee Ownership Scheme
16.8 14.4 8.3
5. Presence of Incentive Pay
61.3 41.2 39.4
6. Ongoing Training
72.1 75.0 67.7
7. Internal “Symbolic” Equity
41.9 53.7 49.7
8. Internal Promotion
26.5 26.5 16.1
9. Formal Appraisal System
59.9 53.8 42.5
10. Information Sharing
79.9 88.0 88.9
11. Self-Managed Teams
71.8 85.6 82.1
12. Job Enrichment
66.2 76.4 68.9
13. Participation and Empowerment
59.3 59.8 60.2
Number of Observations 1005
(0.58)
634
(0.37)
103
(0.06)
Notes: Source: Data are for Britain using WERS data 1998. Numbers in parentheses refer to
sample proportions
32
Table 5: The Determinants of Voice at the Workplace, WERS 1998.
Dependent Variable Mean
0.84
1. HRM Score
0.03
(2.41)
2. Public Sector [Private]
0.10
(2.18)
3. Foreign Owned [Domestic]
-0.07
(-1.11)
Joint-venture 0.21
(3.11)
4. Family Owned or controlled [Other]
-0.12
(-2.08)
5. Franchise [Non-Franchise]
0.16
(2.28)
6. Workplace Size [10-24 employees]
25-49
0.07
(1.52)
50-99 0.07
(1.78)
100-199 0.13
(3.43)
200-499 0.10
(2.55)
500+ 0.10
(2.15)
7. Size of Establishment Network [Single]
2-10 0.04
(0.76)
11-50 0.03
(0.63)
51+ 0.05
(1.14)
8. Number of Organisational Affiliations [None]
One 0.03
(0.71)
Two 0.12
(2.24)
Three 0.10
(1.06)
Four 0.24
(3.98)
9. Age of Establishment [21+ yrs]
10-20 -0.05
(1.00)
5-9 -0.05
(-0.81)
3-4 0.01
(0.26)
<3
-0.07
(-0.94)
10. Industry [Wholesale and Retail]
Manufacturing -0.07
(-0.81)
Electricity, gas and water 0.08
(1.44)
33
Construction -0.12
(-1.11)
Hotels and restaurants -0.01
(-0.12)
Transport and communication -0.07
(-0.72)
Financial services -0.06
(1.08)
Other business services 0.07
(1.08)
Public administration
0.08
(1.03)
Education
0.03
(0.40)
Health
0.02
(0.29)
Other community services 0.02
(0.18)
11. Intercept 0.66
(6.13)
Observations 1583
R-squared 0.17
Notes: Items in [ ] refer to omitted reference category. Linear estimation of the dependent
variable voice (0,1). The coefficients can be interpreted as the percentage point change in voice
adoption based on falling into one of our dependent categorical variable classifications.
Similar results are obtained with a logit model available on request.
34
Table 6: The Determinants of HRM Intensity at the Workplace, WERS 1998.
Dependent variable:
High HRM Score Dummy
[Probit Estimates]
Dependent variable:
HRM Score
[OLS]
Dep. Var. Mean 0.43 6.89
Coefficient t-stat Coefficient t-stat
1. Type of Voice [No Voice]
Union only -0.44 -1.25 -0.14 -0.39
Non-Union only 0.52 1.97 0.59 1.93
Dual Channel 0.58 2.41 0.67 2.56
2. Public Sector [Private]
-0.17 -0.78 -0.02 -0.05
3. Foreign Owned [Domestic]
0.32 1.60 0.09 0.27
Joint-Venture -0.36 -3.45 -1.91 -3.48
4. Family Owned or controlled [Other]
-0.60 -3.22 -0.56 -2.34
5. Franchise [Non-Franchise]
-0.04 -0.12 -0.21 -0.43
5. Workplace Size [10-24 employees]
25-49 0.06 0.39 0.08 0.44
50-99 0.15 1.00 0.36 1.76
100-199 0.42 2.82 0.81 3.77
200-499 0.57 3.70 0.94 4.17
500+ 0.60 3.29 1.13 4.28
6. Size of Establishment Network [Single]
2-10 0.46 2.50 1.46 5.39
11-50 -0.03 -0.98 0.40 1.47
51+ 0.57 3.38 0.88 3.90
8. Age of Establishment [21+ yrs]
10-20 0.24 1.49 0.51 2.31
5-9 0.22 1.20 0.53 2.30
3-4 0.51 2.13 0.94 2.96
<3 0.22 0.93 0.22 0.70
9. Number of Organizational Affiliations [None]
One 0.02 0.13 0.03 0.21
Two 0.23 1.29 0.42 1.68
Three 0.52 1.94 0.75 1.88
Four -0.15 -0.46 0.98 1.71
10. Industry [Wholesale and retail]
Manufacturing -0.65 -2.72 -0.94 -2.75
Electricity, gas and water 0.72 2.16 1.11 2.38
Construction -0.88 -3.08 -1.52 -3.52
Hotels and restaurants -0.36 -1.20 -1.07 -2.66
Transport and communication -0.11 -0.32 -0.37 -1.05
Financial services -0.05 -0.19 0.59 1.65
Other business services -0.01 -0.04 -0.05 -0.13
Public administration -0.01 -0.01 -0.54 -0.97
Education -0.00 -0.01 0.15 0.44
Health 0.33 1.23 0.29 0.80
Other community services -0.32 -1.13 -1.49 -2.99
11. Intercept -1.00 -3.54 6.97 18.68
Observations 1583 1583
F-stat/R-squared 5.62 0.32
Notes: Probability estimates refer to marginal probabilities. Items in [ ] refer to omitted reference
category.
35
Figure 1: Frequency of HRM Scores by Voice For British Workplaces
0
5
10
15
20
25
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
Non-Voice [5.7]
Voice [7.3]
36
Figure 2: Frequency of HRM Scores by Voice Type For British Workplaces (the
dotted line showing the distribution of HRM for all workplaces)
Panel A: Workplaces With Union Voice Only [6.4]
Panel B: Workplaces With Non-Union Voice Only [6.8]
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
37
Panel C: Workplaces With Dual-Channel (Union + Non-Union) Voice [7.5]
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
doc_310772471.pdf
Using British workplace data we examine the relationship between human resource management (HRM) and different forms of employee voice. After controlling for observable establishment characteristics, we find voice and HRM are positively correlated, but this positive association is confined to certain voice regimes.
esearch
PSI Research Discussion Paper 27
Employee Voice and Human Resource Management:
An Empirical Analysis using British Data
Alex Bryson, Paul Willman,
Rafael Gomez, Tobias Kretschmer
R
D
P
iscussion
apers
Policy Studies Institute
Policy Studies Institute, 2007
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system or transmitted in any form or by any means, electronic or
otherwise, without the prior permission of the copyright holder.
ISBN: X XXXXXX X
PSI Report No: 905
Policy Studies Institute
For further information contact:
Publications Dept., PSI, 50 Hanson Street, London W1W 6UP
Tel: (020) 7911 7500 Fax: (020) 7911 7501
PSI is a wholly owned subsidiary of the University of Westminster
esearch Discussion Papers R
Policy Studies Institute
Employee Voice and Human Resource Management: An
Empirical Analysis using British Data
Alex Bryson, Paul Willman, Rafael Gomez, Tobias Kretschmer*
*Alex Bryson is at the Policy Studies Institute and the Centre for Economic
Performance, Paul Willman and Rafael Gomez are in the Department of
Management at the London School of Economics and Tobias Kretschmer is
at the University of Munich and the Centre for Economic Performance. Data
and programs are available on request from Alex Bryson
([email protected])
Acknowledgements
We would like to thank Steve Wood and participants at the Voice and Value
Conference 2007 at LSE for valuable comments
1
ABSTRACT
Using British workplace data we examine the relationship between
human resource management (HRM) and different forms of employee
voice. After controlling for observable establishment characteristics, we
find voice and HRM are positively correlated, but this positive association
is confined to certain voice regimes. Previous research has found no
association between HRM and union voice. However, distinguishing
between union-only voice regimes and dual channel (i.e. union and non-
union) voice regimes reveals that union-only regimes have the lowest
incidence and intensity of HRM adoption while dual channel regimes
have the highest HRM incidence and intensity. The implications of these
findings for theory and practice are discussed.
2
The relationship between the adoption of human resource management (HRM)
techniques and the presence of unions has been a topic of considerable debate
on both sides of the Atlantic for a number of years. Early writers such as Guest
(1989:48) emphasized the individual focus of HRM as damaging to collective
organization, while others such as Kochan (1980) saw greater use of HRM
techniques as part of an overt ‘union substitution’ strategy. Kochan and
Osterman (1994), on the other hand, suggested HRM might work better in the
presence of unions. Protagonists in the union versus HRM debate have often
used different definitions of HRM and different measures of union presence, but a
recent and comprehensive analysis of British data by Machin and Wood
concluded that there is “no statistically significantly greater adoption of HRM
practices in non-union workplaces than in unionized ones” (Machin and Wood,
2005:216).
The authors qualify their conclusions in two ways. First, they speculate on the
possibility that their results are “uniquely British” and second they introduce the
possibility that there may be “different reasons for introducing HRM practices in
union and non-union environments” (2005:216). In this paper, we pursue these
arguments further by distinguishing between union and non-union voice,
considering their relationship to HRM separately and in combination. This proves
to be crucial in understanding the relationship between HRM and union voice in
the UK context. We argue that there may be something about the British context
which delivers this idiosyncratic relationship between HRM and voice: we
3
develop this argument below, but in brief it rests on the ability of employers in
Britain to combine union and non-union voice in ways that are not possible in the
US.
The structure is as follows. In the next section, we distinguish voice and HRM
and speculate about their relationship. We then describe some distinctive and
relevant features of the UK context. In the following section, we introduce our
hypotheses about HRM and voice incidence. Next, we describe our data and
methods and test our hypotheses. The final section summarizes and concludes.
Voice and HRM
Voice regimes are governance mechanisms for employment contracts
(Williamson, 1991; Bryson et al 2004). They exist where institutions or processes
are present to generate two-way communication between managers and
employees. Voice regimes can be direct or representative in nature and can be
delivered in a number of ways; via a union, through management led initiatives or
as part of some dual channel where union and non-union voice are both present
(Millward et al., 2000; Bryson, 2004). Our conception of employee voice is not, as
was Freeman and Medoff’s (1984), based solely on unionism. Rather, it is closer
to Hirschman’s (1971) conception, embracing any form of employee voice as
Hirschman embraced any form of consumer voice: it is the institutionalization of
two-way communication between employers and employees designed to reduce
4
transaction and exit costs for both parties. It is thus a contractual governance
mechanism with mutual benefits.
Voice regimes can also be accompanied by HRM practices. The specific
practices that form part of any given HRM system can be quite diverse, but
typically they involve managerial attempts to motivate and manage workers
through a series of workplace practices rather than through strict command and
control structures. HRM is a sophisticated set of control techniques designed to
yield benefits to the firm by use of management–initiated techniques for people
management designed to generate higher performance (Becker et al 2001).
However, the effectiveness of these HRM techniques in improving organizational
performance is contested (Godard 2004).
The HRM practices we consider here are derived from Pfeffer (1995) and Storey
(1992) and are discussed in detail below. They may be related to voice regimes,
but they are logically distinct. Somewhat like mobile phones and cameras, voice
and HRM may be encountered as integrated packages or separately. They may
confer different costs and benefits in different combinations and locations.
5
The UK Context
Britain has historically experienced a ‘voluntarist’ industrial relations system
granting the parties to industrial relations a considerable degree of freedom to
choose their preferred or agreed institutional arrangements (Clegg, 1979). The
two decades prior to the 1998 Workplace Employment Relations Survey (used
here and described in more detail below) broadly coincided with the Thatcher era
during which a further expansion of employer choice was generated by a series
of deregulatory measures which reduced legal support for trade union activity
(Willman and Bryson, 2007). Employers in the UK were therefore virtually
unconstrained in the period up to 1998 in their ability to mix union and non-union
voice at the establishment level. This has made it possible for employers to
supplement union with non-union voice without terminating relationships with
unions (Willman and Bryson 2007). This situation contrasts starkly with the USA
where the legal system proscribes certain non-union voice and HRM practices in
a union environment to preserve unions’ “sole agent” status (LeRoy, 2006).
In spite of this variety in voice regimes, UK studies on the co-existence of HRM
and voice focus almost exclusively on the links between HRM and union voice. It
is usually assumed that HRM may substitute for unions since it offers alternative
solutions to worker problems, thus challenging the solutions offered by unions
and potentially reducing the incentive for union organizing. There is evidence
that HRM policies are associated with higher job satisfaction and a lower
intention to quit (Guest and Conway, 1999) and that they reduce workers’
6
expressed problems at the workplace in both Britain and the US, potentially
limiting workers’ desire for union solutions (Bryson and Freeman, 2006).
However, studies generally find that unionized workplaces differ little from non-
unionized workplaces in terms of the incidence of HRM (Wood, 1996; Wood and
de Menezes, 1998; Machin and Wood, 2005) and – unlike in the USA – HRM
practices do not reduce workers’ desire for unionization (Bryson and Freeman
2006).
1
Furthermore, one study using the same data employed in this paper
found that a positive link between HRM practices and labor productivity was
confined to unionized workplaces(Bryson et al., 2005), suggesting that
unionization and HRM may even be complementary. Unions in the USA may
challenge employer efforts to introduce HRM, aware of the potential for HRM to
undermine worker support for unionization, whether intentionally or
unintentionally (Fiorito, 2001; LeRoy, 2006). These concerns, however, may not
be so pertinent in the UK.
Of particular relevance for our work is that all of the studies above do not take
account of the heterogeneity of voice regimes in the UK. In particular, they do
not distinguish between non-union, union and dual channel (mixed union and
non-union) voice regimes, presenting instead the effects of unionization on HRM
across pure and mixed regimes, usually proxied by a dummy variable identifying
1
Cully et al. (1999: 110-111) note that the higher incidence of HRM practices in unionized
workplaces is largely due to their greater likelihood of being large and being in the public sector.
Machin and Wood (2005) also present bivariate analyses indicating higher HRM incidence in
unionised workplaces but, again, this association disappears when controlling for other factors.
7
workplaces in which at least one union is recognized for pay bargaining. In fact,
non-union forms of voice have become increasingly prevalent.Moreover, where
union voice is present, it often co-exists with non-union voice (Bryson et al.,
2004). Among unionized workplaces with 10 or more employees, six-in-seven
had some form of non-union voice as well (see below).
As in the USA, the union sector has been shrinking in the UK during the past
three decades. By 1998, around half the workplaces in the UK had voice
mechanisms that were exclusively non-union. These include direct and
representative voice channels which are instigated by the employer, sometimes
at the request of employees.
2
To our knowledge, there is no empirical evidence
on the links between these various forms of voice and HRM either in the UK or
the USA, in part because many authors’ measures of HRM subsume voice
elements in their estimation of HRM (e.g. Machin and Wood, 2005).
3
Hypotheses
For our hypotheses, we adopt the following simple notation: Let HRM (X) be the
extent of HRM use given voice regime X, where X can be: A (absence of
2
EU legislation can require the setting up of works councils at the apex of larger organizations,
but this does not affect workplace-level arrangements in the UK. The Information and
Consultation Directive which came into effect in April 2005, may have some effect on voice
arrangements at workplace-level but this was not in place in the late 1990s.
3
There is one exception. Fenton-O’Creevy and Wood (2005) distinguish between union voice and
direct and representative forms of non-union voice. Using data on multinational companies with
headquarters in Britain they find no support for the proposition that direct-voice only regimes have
a higher incidence of high-commitment work practices than other voice regimes.
8
voice), U (use of union voice), and N (non-union voice). U and N can be
present concurrently.
A management control technique like HRM and a contractual governance
mechanism like voice may be complementary or substitutes. If HRM practices
delivered everything voice representation did (and vice versa), the two would be
substitutes. In fact, since HRM techniques emerged much later than unions and
other forms of voice representation, one could argue that HRM is a natural
“successor” and firms will switch from voice to HRM eventually.
On the other hand, if the effectiveness of HRM is enhanced by the simultaneous
presence of voice at a workplace (e.g. because voice reduces costly worker exit
or increases the flow of productivity-enhancing information), the two would be
complements and we would expect them to appear together. Concurrent use of
voice and HRM could also originate from a common factor driving the adoption of
both – for example managerial quality – that enables a firm to cope with
organizational and managerial innovations more easily. In this case, we would
again expect the two practices to appear in conjunction– a workplace that draws
positive net benefits from one will also draw positive net benefits from the other.
We view the complementary argument(s) as more compelling and therefore
hypothesize:
9
H
1
: HRM will be more prevalent in establishments with (any form of) voice than in
those without voice (min[HRM(U), HRM(N), HRM(U,N)] > HRM(A)).
Conditioning on workplaces with voice, HRM, may sit better with some forms of
voice than others. HRM may substitute for union voice where HRM generates
outcomes that reduce employee demand for unionization. The presence of
union-only voice may signal union success in monopolizing worker voice at the
workplace and may imply the potential to block HRM too if the union is not
persuaded of the ‘high performance’ ethos which drives HRM. In addition, union-
only voice may imply a reliance on collective rather than individual forms of
employer-employee engagement that might exclude HRM. Reliant on a narrower
definition of union only voice (rather than union presence) we hypothesize,
amending Machin and Wood (2005), that among voice regimes, HRM will be
lowest in union-only voice regimes.
Non-union voice only exists in the UK at workplace-level where the employer has
chosen to invest in its provision (Bryson et al., 2004).
4
The fact that the employer
has chosen to implement non-union voice implies that the employer may also
invest in HRM to obtain a competitive advantage as described by Pfeffer (1995).
The presence of non-union voice alongside union voice at a workplace is
indicative of one of two scenarios, both of which are conducive to HRM. It may
either signal union weakness which leaves the employer largely unconstrained in
4
It is plausible that direct voice emerges spontaneously in the smallest of workplaces due to
worker proximity. However, these workplaces are absent from our data.
10
mixing voice varieties with HRM, or it may imply a ‘mutual gains’ environment in
which unions are willing to use what Freeman and Medoff (1984) termed their
‘voice face’ to elicit productivity improvements. Therefore, we hypothesize;
H
2
: HRM will be higher in workplaces with non-union voice than in those with
union only voice (min[HRM(N), HRM(U,N)] > HRM(U)).
So far our hypotheses are consistent with the proposition that HRM incidence will
increase in an ordinal fashion from no-voice to union voice to non-union voice
regimes.
Our main departure from previous work in this field is to stress the heterogeneity
of union voice. Specifically, we identify dual voice as empirically more common
than union only voice (Bryson et al 2004). If, as we hypothesize, union only voice
environments restrict HRM and non-union only voice environments are a more
favorable habitat, we must consider how dual voice and HRM will coexist.
Whether a workplace is unionized is, in part, an accident of history: the earlier
cohort of workplaces initially adopted union-voice whereas newer workplaces
adopted non-union only voice (Machin, 2000; Millward et al., 2000; Willman et al.,
2007). Yet very few unionized workplaces have de-recognized their unions in
Britain; instead, they have supplemented union with non-union voice (Millward et
al., 2000; Kersley et al., 2006).
11
Dual voice regimes may have emerged in different ways. For some firms, non-
union voice was added to union only regimes in response to employers
perceiving deficiencies in union-only voice provision. The reverse is also
possible, where employees pressed for unionization in addition to non-union
consultation frameworks. In the UK, it is likely that the former has been more
common.
However, if union voice acts as a drag on the introduction of HRM, employers
may switch to non-union voice. Where there were no constraints on doing so,
adding non-union voice mechanisms to union ones is also a risk-averse strategy
for overall voice improvement (Bryson et al 2004). It is thus likely that in
continuing dual regimes, no such drag operates, and we therefore hypothesize;
.
H
3
: There will be no difference in HRM incidence and intensity between dual
channel and non-union only voice regimes (HRM(U, N) = HRM(N)).
To summarize, our hypotheses on the incidence of voice and HRM can be
expressed in the following ordinal ranking:
HRM(U, N) = HRM(N) > HRM(U) > HRM(A),
12
where the first equality represents H
3
, the first inequality H
2
, and the second
inequality H
1
.
The Data and Context
Our data are the British Workplace Employment Relations Surveys 1998
(WERS98), a large-scale survey of industrial relations in British establishments in
the public and private sectors. The key features of these data are described
elsewhere (Millward et al., 2000, 3-10; 248-55). Our analysis is based on data
collected from human resource managers responsible for workplace industrial
relations which contain the voice-related variables and HRM items needed for our
analysis. All observations are weighted by the inverse of the workplace’s
probability of selection for the survey. With these weights, our analyses provide a
representative portrait of workplaces in Britain with 10 or more employees in
1998. We now define our measures.
Voice is defined as the presence of two-way forms of (representative or direct)
communication between workers and management. In our data, the set of voice
measures is as follows.
1. union recognition
2. union representatives on or off site
3. a joint consultative committee meeting at least once a month
4. non-union representatives on site
13
5. problem solving groups
6. regular meetings between management and employees which allow for
two-way communication
7. team briefings that occur at least once a month and devote time to
employees’ questions/views.
Items 1-2 measure union voice. Items 3-7 measure non-union voice. No-voice
workplaces are defined by the absence of all.
The context and timing of the data collection are relevant to understanding the
voice measures. In 1999 the government enacted legislation which could require
employers to recognize trade unions if the majority of workers so wished, hence
our reliance on the 1998 dataset. We thus have four categories of voice in our
analysis: no voice, union voice only, non-union voice only and a mix of union and
non union voice (termed ‘dual’). The incidence of these regimes in our data was
17, 5, 48 and 30 percent respectively.
The HRM measures are broadly based on Pfeffer (1995). The measure is a count
of practices identified by Pfeffer, supplemented by other aspects of human
resource management identifiable in the literature. These dimensions are as
follows:
14
1. selectivity in recruiting: where manager says skills, qualifications,
experience and motivation are all important factors in recruiting new
employees
2. job security: policy of guaranteed job security for at least some employees
3. incentive pay: profit pay, performance related payments or cash bonuses
4. employee share ownership scheme
5. information sharing: management shares information on investment,
financial position of the organization or staffing
6. ‘empowerment’: core employees have a lot of control over variety in their
work, discretion over how they do their work or control over the pace at
which they do their work
7. self-managed teams: core employees work in teams that are able to
appoint their own leaders, jointly decide how work is done, or have
responsibility for specific products or services
8. on-going training: on-going training is one of the main methods by which
core employees are made aware of their job responsibilities
9. cross-training: at least some core workers are formally trained to do jobs
other than their own
10. symbolic egalitarianism: core employees have standard contracts for all
non-pay terms and conditions of employment
11. promotion from within: internal applicants given preference when filling
vacancies.
15
In addition to these items identified by Pfeffer the score includes two other items:
12. an indicator that the workplace has a formal strategic plan, strategic
planning being a key component of HRM (Storey, 1992)
13. the existence of a widespread appraisal system, that is, where at least
80% of core employees are formally appraised.
A score of 13 denotes affirmative answers to each of these questions. Lacking
any one of these HRM variables would give an establishment a score of 12 and
so on.
5
This summative measure of HRM differs from others in the literature,
including Machin and Wood (2005) who, because they use earlier surveys as
well as the 1998, must operate with a smaller set of HRM measures. In addition,
they include items such as the existence of a joint consultative committee that we
define as voice. In our analysis, measures of HRM and of voice are clearly
discrete sets.
Empirical Findings
Table 1 reports the incidence of voice by workplace characteristics in 1998. The
percentage of establishments with some form of voice ranges from lows of 64
and 67 percent in single-establishments and family owned operations
5
Clearly, this is the simplest way of defining HRM intensity. Assigning different weights to each of
these elements or using clusters of HRM practices does not fundamentally alter our conclusions
(results available upon request).
16
respectively to a high of 100 percent in health establishments. The overall
sample average for all establishments is 83 percent.
[Table 1]
The first indication of a positive correlation between voice and HRM is the higher
incidence of voice in workplaces with high HRM scores (94 per cent among those
with an HRM score of 9+) compared with workplaces with lower HRM scores (80
per cent among those with scores of 8 or under).
6
This relationship, which
supports Hypothesis 1, is illustrated graphically in Figure 1 which shows the HRM
score distribution for no-voice workplaces lies to the left of that for workplaces
with voice.
[Figure 1]
Table 2 reports descriptive statistics on the use of HRM practices, overall and
disaggregated by workplace characteristics. In 1998, establishments used an
average of 7 out of 13 HR practices. This overall figure differs considerably
across workplaces with small, single-establishment, family owned, private sector
and middle-aged establishments using fewer HRM practices than larger, non-
family owned, public sector, and younger or older establishments.
[Table 2]
Table 2 presents HRM scores by workplace characteristics. It confirms that
establishments with voice use more HRM practices. However, HRM intensity
varies significantly across voice regimes. In accordance with Hypothesis 2,
6
Since 1984, non-union only regimes have increased three-fold in the population of workplaces
with 25 or more employees. Union-only regimes have witnessed a similarly impressive decline
over the same period (Willman et al, 2007).
17
union-only voice has the lowest HRM score amongst voice types. HRM scores
are higher in dual channel and non-union only voice regimes.
The distribution of HRM scores by type of voice regime is shown in Figure 2. The
panels are ranked in terms of overall mean in the HRM score. It is clear that
union-only voice is associated with a compressed distribution of HRM scores
(panel A). There are fewer outliers within the union voice type, with most
workplaces located between 5 and 8 practices. In each panel the dotted line
represents the overall frequency of scores.
[Figure 2]
Table 3 shows the incidence of each HRM practice by workplaces with and
without voice. Almost every practice is more likely to be found in a workplace
that also has formal voice. The exception is the use of incentives, perhaps
indicating a more transactional relationships in no-voice regimes.
[Table 3]
Table 4 presents the same set of 13 practices, only now we look at patterns
within the voice sector. Establishments with dual voice have a greater likelihood
than others of adopting almost all HRM practices, except for the compensation
related (extrinsic) components, incentive pay and employee ownership schemes,
where establishments with non-union only voice are more likely to use these two
practices. The biggest disparity between union-only voice regimes and other
voice regimes is in relation to formal appraisal, internal promotion, selective
recruitment and share ownership. The relative absence of these HRM practices,
18
together with the high incidence of information sharing and, to some degree,
participation and empowerment, helps account for the compression of HRM
practices in this sector.
7
[Table 4]
The descriptive relationships presented here appear to confirm the three
hypotheses. However, it is possible that the relationships identified are driven by
observable differences between voice and no-voice workplaces which are also
correlated with the incidence of HRM. For instance, Table 1 showed that the use
of voice increases with establishment size, something that may also influence the
costs and benefits of returns to HRM. We therefore test if these relationships
change once we control for observable workplace characteristics. The results are
presented in Table 5.
[Table 5]
Table 5 confirms many of the bivariate relationships identified in Table 1, such as
the link between voice and establishment size. It also, confirms an independent,
statistically significant relationship between HRM use and the presence of voice
(row 1).
Table 6 reports estimates of the association between varieties of voice and HRM
adoption and intensity. In the first column are the results of a dummy variable,
which uses a cut-off of 9 HRM practices to denote a high HRM score
8
. We then
7
It is not surprising to see the high prevalence of HRM practices such as information sharing in
union workplaces since these are often the object of union bargaining.
8
Other cut-offs were tried and results were not significantly different; results available upon
request.
19
explore the robustness of these results to a change in dependent variable, where
the HRM intensity score, as defined previously, is used. In column (2) the mean
score was 6.89 and as our results are qualitatively the same in both columns, we
will confine our interpretation of results to the more intuitive HRM score results.
[Table 6]
There is strong support for our hypotheses regarding links between voice and
HRM (Table 6, row 1). Treated as a single entity, voice and HRM were positively
correlated (Table 5), supporting our first hypothesis. However, across voice
regimes, HRM is least prevalent in union-only regimes (Table 6), confirming
Hypothesis 2. The difference between union-only regimes and those containing
non-union voice is statistically significant. Indeed, controlling for observable
differences across workplaces, HRM is no more prevalent in union-only regimes
than it is in no voice regimes. Thus, in accordance with Hypothesis 3, it is the
voice regimes with some non-union voice present – either in isolation or in
combination with union voice – where HRM is most apparent.
Descriptive analyses indicated that dual channel voice regimes had a greater
HRM incidence than non-union only voice regimes, However, although the HRM
adoption coefficients are largest in dual channel regimes, these are not
significantly higher than the coefficients in non-union only regimes in line with
hypothesis 3.
9
This suggests that the differences in the descriptive analysis were
9
The 0.07 difference between non-union only and dual channel coefficients in the HRM score
model has a t-statistic of 0.29.
20
partly attributable to observable differences between establishments with these
regimes that are controlled for in the regression.
For our other controls we find results in line with much of the associated literature
on HRM adoption. Specifically, we find that:
• Workplace size has a positive and significant effect on the number of HRM
practices adopted;
• Workplaces that are part of some multi-establishment network also adopt
more HRM practices, with the number of those practices increasing with
network size;
• Age of establishment displays an inverted-U shape with workplaces aged
3 to 4 years being the most intensive users of HRM, while those aged
under 3 and more than 21 years have lower scores;
• Organizational affiliations generally increase the use of HRM up to a point,
as the highest category (4 affiliations) seems to make little difference.
Workplaces with three organizational affiliations have the highest use of
HRM.
Conclusions
In the last two decades of the 20
th
Century unionization was in decline in Britian.
Voice regimes, on the other hand, were not in decline since employers were
substituting union voice with non-union voice (Millward et al., 2000; Bryson et al.,
21
2004). At the same time, HRM practices were on the increase, although they
remained far from ubiquitous. In their important contribution to the literature on
union voice and HRM, Machin and Wood (2005) argued that there was no causal
link between declining unionization and rising HRM. However, their time-series
data means their study relies on a partial measure of HRM which conflates HRM
and voice. Furthermore, in keeping with the rest of the literature, they take no
account of heterogeneity in voice regimes. This is the first paper to draw attention
to the heterogeneity of union and non-union voice regimes and, in particular, the
importance of dual channel voice regimes in understanding the links between
HRM and unionization. We find that, although union-only voice regimes and
HRM do not easily co-habit, dual channel voice and HRM do. This may go some
way to explaining why despite union membership decline, large scale union de-
recognition has not occurred in Britain; dual channel employers simply did not
need to get rid of unions in order to introduce HRM. There is also no evidence of
a switch to no-voice with HRM. Most of the decline in individual unionization was
brought about by new workers entering new workplaces that had simply by-
passed the union voice solution.
By using a richer conception of HRM than the previous literature and by
analyzing both the incidence and intensity of HRM, we have shown that the
intensity of HRM, its distribution and the relative frequency of its components
differ across the union and non-union sectors and within the union sector. In
particular, financial incentives are less prevalent in the union sector whereas self-
22
managed teams and information sharing are more prevalent in the union sector.
Furthermore, dual channel regimes have more intensive HRM than union-only
regimes and there is some compression in the distribution of HRM in the union-
only sector compared with the dual channel sector. More ‘permissive’ legal and
public policy arrangements obtain in the UK than the USA, making employer and
worker choices more salient in determining voice and HRM outcomes. Although
we can not directly answer Wood and Machin’s (2005: 216) question regarding
the motivations for the use of HRM in the union and non-union sectors, we argue
that these patterns are consistent with explanations which emphasize the
heterogeneity of union voice regimes.
The greater part of the union voice sector in the UK is also to a material degree
non-union and this, we argue, explains much of the compatibility between unions
and HRM. The clearest complementarity is between non union voice and HRM.
Union-only voice regimes are more similar to no-voice regimes in their
relationship to HRM. In this sense, the UK case may not be a good basis for
generalization. Where employers face a straight choice between union or non-
union voice (as in the US), the implications for HRM adoption and use may be
very different.
23
References
Becker, Brian, Mark Huselid and Dave Ulrich. 2001. The HR Scorecard; Linking
People, Strategy and Performance, Boston: Harvard Business School Press,.
Bryson, Alex. 2004. ‘Managerial responsiveness to union and non-union worker
voice in Britain’, Industrial Relations: A Journal of Economy and Society, Vol. 43,
No. 1, pp. 213-241
Bryson, Alex, John Forth and Simon Kirby. 2005. ‘High-performance practices,
trade union representation and workplace performance in Britain’, Scottish
Journal of Political Economy, 53, 3: 451-491
Bryson, Alex and Richard Freeman (2006) Worker Needs and Voice in the US
and the UK, NBER Working Paper No. 12310, Cambridge, Mass.
Bryson, Alex, Paul Willman and Rafael Gomez. 2004. “The End of the Affair? The
Decline in Employers' Propensity to Unionize.” In John Kelly and Paul Willman
(eds.), Union Organization and Activity, Routledge, London, pp.129-149.
Clegg, Hugh A. 1979 The System of Industrial Relations in Great Britain, Oxford:
Blackwell.
Felstead, Alan and David Ashton. 2000. “Tracing the link: organisational
structures and skill demands”. Human Resource Management Journal, 10,3, 5-
21.
Fenton-O’Creevy, M. and Wood, S. (2005) ‘Direct Involvement, Representation
and Employee Voice in UK Multinationals in Europe’, European Journal of
Industrial Relations, Vol. 11, No. 1, 27-50
Fiorito, Jack. 2001. “Human Resource Management Practices and Worker
Desires for Union Representation”, Journal of Labor Research, Vol. 22, No. 2, pp.
335-54
Freeman, Richard and James Medoff. 1984. What Do Unions Do? New York:
Basic Books:
Godard, John. 2004. “A Critical Assessment of the High-Performance Paradigm.”
British Journal of Industrial Relations, Vol. 42, No. 2, pp. 349-378.
Guest, David E. 1989. “Human Resource Management; Its Implications for
Industrial Relations and Trade Unions.” In John Storey (ed) New Perspectives on
Human Resource Management. London; Routledge pp. 41-55.
24
Guest, David and Neil Conway. 1999. “Peering into the Black Hole: The
Downside of the New Employment Relations in the UK”. British Journal of
Industrial Relations, Vol. 37, No. 3, pp. 367-90
Hirschman, Albert. 1970. Exit, Voice and Loyalty. Cambridge, Harvard University
Press.
Kochan, Thomas A. 1980. Collective Bargaining and Industrial Relations .
Homewood, Ill: Richard D. Irwin.
Kochan, Thomas A., and Paul Osterman. The Mutual Gains Enterprise. Boston:
Harvard Business School Press.
LeRoy, Michael. H. 2006 ‘The power to create or obstruct employee voice: does
US public policy skew employer preference for “no voice” workplaces?’ Socio-
Economic Review, 4, pp. 311-319
Machin, Stephen. 2000. ‘Union decline in Britain’, British Journal of Industrial
Relations, Vol. 38 No. 4: pp.631–45.
Machin, Stephen, and Stephen Wood. 2005. “Human Resource Management as
a Substitute For Trade Unions in British Workplaces” Industrial and Labor
Relations Review Vol. 58, No. 2 pp.201-18.
Millward, Neil, Alex Bryson and John Forth. 2000. All Change at Work, London:
Routledge..
Pfeffer, Jeffrey. 1995. “Producing sustainable competitive advantage through the
effective management of people,” Academy of Management Executive, Vol.9,
No. 1, pp. 55-72
Storey, John. 1992. Developments in the Management of Human Resources.
Oxford: Blackwell.
Williamson, Oliver. E. 1991. ‘Comparative Economic Organization: the Analysis
of Discrete Structural Alternatives’, Administrative Science Quarterly, Vol. 36,
Issue 2, pp. 269-298.
Willman, Paul and Alex Bryson (2007) ‘Union Organization in Great Britain’,
Journal of Labor Research, Vol. 28, 1: pp93-115
Willman, Paul, Alex Bryson and Rafael Gomez. 2006 ‘The Sound of Silence:
Which Employers Choose ‘no voice’ and why?’, Socio-Economic Review, 4: 283-
299
25
Willman, Paul, Alex Bryson and Rafael Gomez. 2007. “The Long Goodbye: The
Fall of Union Voice In Britain” International Journal of Human Resource
Management forthcoming
Wood, Stephen. 1996. “High Commitment Management and Unionization in the
UK”. International Journal of Human Resource Management, Vol. 7, No. 1, pp.
41-58
Wood, Stephen, and Maria Albanese. 1995. “Can we speak of high commitment
management on the shop floor?” Journal of Management Studies, Vol 32 No. 2
pp215-47
Wood, Stephen and Lilian De Menezes. 1998. “High Commitment Management
in the UK: Evidence from the Workplace Industrial Relations Survey and
Employers’ Manpower and Skills Practices Survey”. Human Relations, Vol. 51,
No. 4, pp. 485-515
26
Table 1: Incidence of Voice (%) by Selected Workplace Characteristics, 1998
1. All Workplaces 83.1
2. By Sector
Public 98.9
Private services 80.8
Private manufacturing 65.2
3. By Establishment size (employees)
25-49 79.5
50-99 82.7
100-199 89.9
200-499 93.5
500-999 95.1
1000 plus 96.5
4. By Ownership
Foreign 72.5
Domestic 84.2
5. By Establishment
Single 70.7
Multi-establishment 87.7
6. By Size of Multi-Establishment Network*
Single 71.8
2-10 83.9
11-50 83.3
50+ 92.5
7. By Set-up date
Pre 1980 88.5
Post 1980 78.5
8. By Decade of Set-up date
Pre 1980 --
1980s --
1990s --
9. By Establishment Age*
<3 years 80.0
3-19 years 82.1
20+ years 86.1
10. By Industry
Manufacturing 68.5
Electricity, gas and water 99.9
Construction 70.5
Wholesale and retail 83.3
Transport + Communication 86.0
Financial services 76.8
Other business services 93.6
Public administration 99.9
Education 98.7
Health 99.9
Other community services 98.1
11. By HRM Score*
High [9-13] 93.6
Low [0-8] 80.0
12. By Ownership Structure*
Family owned/controlled 66.5
Other 88.8
13. By Employer Association Status
Yes 87.9
No 82.9
14. By Number of Organizational Affiliations*
None 82.9
One 82.1
Two 85.3
27
Three 87.9
Four 99.4
15. By Franchise Status*
Franchisee 91.7
Non-franchisee 83.1
Number of Observations
1954
28
Table 2: Average HRM Score (score 1 to 13) actual and normalized by selected
workplace characteristics, 1998.
Average
HRM Score
Normalised
Z-Score
1. All Workplaces 6.9 0.09
*
2. By Sector
Public 7.45 6.22
Private 6.71 -2.00
3. By Establishment size (employees)
10-24 6.6 -3.22
25-49 6.94 0.56
50-99 7.19 3.33
100-199 7.59 7.78
200-499 7.93 11.56
500+ 8.05 12.89
4. By Ownership
Foreign 7.15 2.89
Domestic 6.89 0.00
Joint Venture 5.81 -12.00
5. By Establishment
Single 5.81 -12.00
Multi-establishment 7.39 5.56
6. By Size of Multi-Establishment Network
Single 5.83 -11.78
2-10 7.03 1.56
11-50 6.87 -0.22
51+ 7.98 12.11
7. By Organization Size
Small [<50] 6.60 -3.22
Large [51+] 7.92 11.44
8. By Establishment Age
<3 years 6.44 -5.00
3-19 years 7.20 3.44
20+ years 6.64 -2.78
9. By Industry
Manufacturing 6.01 -9.78
Electricity, gas and water 9.19 25.56
Construction 5.11 -19.78
Wholesale and retail 7.31 4.67
Hotels and restaurants 6.56 -3.67
Transport + Communication 7.10 2.33
Financial services 8.65 19.56
Other business services 7.14 2.78
Public administration 7.17 3.11
Education 7.42 5.89
Health 7.15 2.89
Other community services 5.56 -14.78
10. By Ownership
Family owned/controlled 5.93 -10.67
Other 7.22 3.67
11. By Employer Association Status
Yes 6.80 -1.00
No 6.88 -0.11
12. By Number of Organisational Affiliations
0 6.88 -0.11
1 6.67 -2.44
2 6.96 0.78
3 7.08 2.11
4 7.87 10.89
13. By Franchise Status
29
Franchise 6.77 -1.33
Non-franchise 6.90 0.11
14. By Type of Voice I
No Voice 5.71 -13.11
Union 6.41 -5.33
Non-Union Only 6.86 -0.33
Dual Channel 7.51 6.89
15. By Type of Voice II
No Voice 5.71 -13.11
Representative Only 6.28 -6.78
Representative and Direct 7.63 8.22
Direct Only 6.75 -1.56
Number of Observations 1929
Source: Data are for Britain using WERS data 1998. *The figure 0.09 is the standard deviation
for the sample.
30
Table 3: Incidence of HRM practices by workplaces with and without formal voice (%),
1998.
By Presence of Voice at workplace
HRM Practices
No
Yes
All Workplaces
1. Presence of Formal Strategic Plan
47.1
78.9
73.7
2. Guaranteed Job Security
7.2 10.9 10.3
3. Selective Recruitment
46.4 54.8 53.4
4. Employee Ownership Scheme
11.2 15.4 14.7
5. Presence of Incentive Pay
53.3 52.7 52.8
6. Ongoing Training
57.5 73.2 70.6
7. Internal “Symbolic” Equity
20.0 46.6 42.2
8. Internal Promotion
24.2 25.8 25.6
9. Formal Appraisal System
37.4 56.7 53.5
10. Information Sharing
57.5 73.2 79.9
11. Self-Managed Teams
62.6 77.5 75.0
12. Job Enrichment
61.2 70.1 68.6
13. Participation and Empowerment
69.9 59.5 61.1
Number of Observations 346
(0.17)
1742
(0.83)
2088
(1.00)
Notes: Source: Data are for Britain using WERS data 1998. Numbers in parentheses refer to
sample proportions.
31
Table 4: Incidence of HRM practices by type of workplace voice (%), 1998.
Type of Formal voice
HRM Practice
Non-Union
Only
Dual
Channel
Union
Only
1. Presence of Formal Strategic Plan
74.4
87.7
72.7
2. Guaranteed Job Security
6.9 18.4 4.0
3. Selective Recruitment
48.5 67.2 41.1
4. Employee Ownership Scheme
16.8 14.4 8.3
5. Presence of Incentive Pay
61.3 41.2 39.4
6. Ongoing Training
72.1 75.0 67.7
7. Internal “Symbolic” Equity
41.9 53.7 49.7
8. Internal Promotion
26.5 26.5 16.1
9. Formal Appraisal System
59.9 53.8 42.5
10. Information Sharing
79.9 88.0 88.9
11. Self-Managed Teams
71.8 85.6 82.1
12. Job Enrichment
66.2 76.4 68.9
13. Participation and Empowerment
59.3 59.8 60.2
Number of Observations 1005
(0.58)
634
(0.37)
103
(0.06)
Notes: Source: Data are for Britain using WERS data 1998. Numbers in parentheses refer to
sample proportions
32
Table 5: The Determinants of Voice at the Workplace, WERS 1998.
Dependent Variable Mean
0.84
1. HRM Score
0.03
(2.41)
2. Public Sector [Private]
0.10
(2.18)
3. Foreign Owned [Domestic]
-0.07
(-1.11)
Joint-venture 0.21
(3.11)
4. Family Owned or controlled [Other]
-0.12
(-2.08)
5. Franchise [Non-Franchise]
0.16
(2.28)
6. Workplace Size [10-24 employees]
25-49
0.07
(1.52)
50-99 0.07
(1.78)
100-199 0.13
(3.43)
200-499 0.10
(2.55)
500+ 0.10
(2.15)
7. Size of Establishment Network [Single]
2-10 0.04
(0.76)
11-50 0.03
(0.63)
51+ 0.05
(1.14)
8. Number of Organisational Affiliations [None]
One 0.03
(0.71)
Two 0.12
(2.24)
Three 0.10
(1.06)
Four 0.24
(3.98)
9. Age of Establishment [21+ yrs]
10-20 -0.05
(1.00)
5-9 -0.05
(-0.81)
3-4 0.01
(0.26)
<3
-0.07
(-0.94)
10. Industry [Wholesale and Retail]
Manufacturing -0.07
(-0.81)
Electricity, gas and water 0.08
(1.44)
33
Construction -0.12
(-1.11)
Hotels and restaurants -0.01
(-0.12)
Transport and communication -0.07
(-0.72)
Financial services -0.06
(1.08)
Other business services 0.07
(1.08)
Public administration
0.08
(1.03)
Education
0.03
(0.40)
Health
0.02
(0.29)
Other community services 0.02
(0.18)
11. Intercept 0.66
(6.13)
Observations 1583
R-squared 0.17
Notes: Items in [ ] refer to omitted reference category. Linear estimation of the dependent
variable voice (0,1). The coefficients can be interpreted as the percentage point change in voice
adoption based on falling into one of our dependent categorical variable classifications.
Similar results are obtained with a logit model available on request.
34
Table 6: The Determinants of HRM Intensity at the Workplace, WERS 1998.
Dependent variable:
High HRM Score Dummy
[Probit Estimates]
Dependent variable:
HRM Score
[OLS]
Dep. Var. Mean 0.43 6.89
Coefficient t-stat Coefficient t-stat
1. Type of Voice [No Voice]
Union only -0.44 -1.25 -0.14 -0.39
Non-Union only 0.52 1.97 0.59 1.93
Dual Channel 0.58 2.41 0.67 2.56
2. Public Sector [Private]
-0.17 -0.78 -0.02 -0.05
3. Foreign Owned [Domestic]
0.32 1.60 0.09 0.27
Joint-Venture -0.36 -3.45 -1.91 -3.48
4. Family Owned or controlled [Other]
-0.60 -3.22 -0.56 -2.34
5. Franchise [Non-Franchise]
-0.04 -0.12 -0.21 -0.43
5. Workplace Size [10-24 employees]
25-49 0.06 0.39 0.08 0.44
50-99 0.15 1.00 0.36 1.76
100-199 0.42 2.82 0.81 3.77
200-499 0.57 3.70 0.94 4.17
500+ 0.60 3.29 1.13 4.28
6. Size of Establishment Network [Single]
2-10 0.46 2.50 1.46 5.39
11-50 -0.03 -0.98 0.40 1.47
51+ 0.57 3.38 0.88 3.90
8. Age of Establishment [21+ yrs]
10-20 0.24 1.49 0.51 2.31
5-9 0.22 1.20 0.53 2.30
3-4 0.51 2.13 0.94 2.96
<3 0.22 0.93 0.22 0.70
9. Number of Organizational Affiliations [None]
One 0.02 0.13 0.03 0.21
Two 0.23 1.29 0.42 1.68
Three 0.52 1.94 0.75 1.88
Four -0.15 -0.46 0.98 1.71
10. Industry [Wholesale and retail]
Manufacturing -0.65 -2.72 -0.94 -2.75
Electricity, gas and water 0.72 2.16 1.11 2.38
Construction -0.88 -3.08 -1.52 -3.52
Hotels and restaurants -0.36 -1.20 -1.07 -2.66
Transport and communication -0.11 -0.32 -0.37 -1.05
Financial services -0.05 -0.19 0.59 1.65
Other business services -0.01 -0.04 -0.05 -0.13
Public administration -0.01 -0.01 -0.54 -0.97
Education -0.00 -0.01 0.15 0.44
Health 0.33 1.23 0.29 0.80
Other community services -0.32 -1.13 -1.49 -2.99
11. Intercept -1.00 -3.54 6.97 18.68
Observations 1583 1583
F-stat/R-squared 5.62 0.32
Notes: Probability estimates refer to marginal probabilities. Items in [ ] refer to omitted reference
category.
35
Figure 1: Frequency of HRM Scores by Voice For British Workplaces
0
5
10
15
20
25
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
Non-Voice [5.7]
Voice [7.3]
36
Figure 2: Frequency of HRM Scores by Voice Type For British Workplaces (the
dotted line showing the distribution of HRM for all workplaces)
Panel A: Workplaces With Union Voice Only [6.4]
Panel B: Workplaces With Non-Union Voice Only [6.8]
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
37
Panel C: Workplaces With Dual-Channel (Union + Non-Union) Voice [7.5]
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8 9 10 11 12 13
HRM Score (1 t o 13)
F
r
e
q
u
e
n
c
y
i
n
p
e
r
c
e
n
t
doc_310772471.pdf