Corps de l’article

There has been considerable recent debate surrounding the impact of worker participation on performance. For many, firm success is strongly associated with the adoption of participatory work practices designed to involve workers in decision making and the organization of their work (Ichniowski et al., 2000; Handel and Levine, 2004). Others are more sceptical about the robustness and consistency of the performance effects of participation. Hence, a number of studies dispute whether participation on its own leads to superior performance; these studies have shown that the benefits of participation may be dependent upon certain other factors, such as complementary supporting management practices, or the presence of unions (Bryson, Forth and Kirby, 2005; Wood and de Menezes, 1998). Further, controversy exists about whether the oft-reported performance gains of participation stem from enhanced worker commitment or work intensification (Godard, 2004; Harley, Hyman and Thompson, 2005). Relatedly, the participation-performance link may be affected by conditions in the external labour market. There is a longstanding literature that connects the threat of unemployment to work effort and productivity (see for example Marx, 1976, chapter 25). To the extent that this threat effect is operational, it may act to reinforce or undermine sources of worker motivation and consent generated by participation, with consequences for workplace performance.

The following paper seeks to investigate the impact of participation on performance under different labour market conditions. Previous studies that have examined the performance impact of participation have failed to control for differences in labour market conditions and it remains to be established whether any performance effects are dependent upon the threat of job loss. Our empirical study uses data from the British Workplace Employee Relations Survey 1998 (WERS 1998). We look first at the effects of participation on workplace productivity, and gauge the extent to which productivity is sensitive to the combination of participatory practices and appropriate supports. We then investigate whether the performance impact of these practices and supports is influenced by the threat of unemployment. Our results reveal that in establishments which implement participation in a limited way, the observed positive association between participation and productivity depends upon the presence of relatively high unemployment, implying that ‘fear’ and ‘threat’ are important factors in generating high productivity in these establishments.

The paper is organized as follows. In the following section, we consider the interrelationships between participation, performance and labour market conditions. We then describe the WERS 1998 data source, our variables and methods of analysis. This is followed by a presentation and analysis of the results. Finally we draw some conclusions.

Participation, Performance, and External Labour Market Conditions

It has been frequently argued that allowing workers to participate in the management of their work and their firm helps to generate superior workplace performance (Strauss, 1998a; Ichniowski et al., 2000; Handel and Levine, 2004). For Strauss (1998a: 15), participation is a ‘process which allows employees to exert some influence over their work and the conditions under which they work’. Employees may participate indirectly through their representatives or via schemes including joint consultative committees. They can also participate directly through teamworking, quality circles, communication systems, and financial participation, encompassing profit-sharing and employee share ownership schemes (see Marchington, 2005). In both cases, participation is said to give workers an opportunity to contribute to the decision making functions of their firm, or to use their voice in the workplace, leading to improved worker commitment and motivation, and in turn, to higher productivity (Ichniowski et al., 2000). It has been argued that the productivity effect may be stronger where participation is combined with appropriate supporting mechanisms which offer workers the necessary skills and information to involve themselves in the organization and which also help to promote a climate of trust and mutual gains (see Appelbaum et al., 2000; Bryson, Forth and Kirby, 2005).

Some commentators have been more cautious about the effects of participation on performance, however. A number of studies have found little support for a positive participation-performance link, even where ‘systems’ of participation are present (see Godard, 2004, for a review of evidence). In practice, participation may offer workers little real influence over organizational matters (Dundon et al., 2001; Delbridge and Whitfield, 2001). Under these circumstances, participation need not be associated with the higher levels of commitment and motivation that are said by many to generate improvements in organizational performance (see Harley, 2005: 52). Furthermore, adoption rates of participation have been generally low, and only a minority of firms have implemented a coherent set of practices and supports (Godard, 2004). Employers may well be reluctant to embrace participation due to its perceived adverse effects on their ability to exercise control over production and due to its potentially high cost (Levine, 2000: 273–274). These factors may lead them to adopt modest forms of participation that allow workers only limited input into their work, or to reject worker participation altogether in favour of ‘intensification’ approaches that rely mainly on pressurizing workers to work harder.

It is not at all clear that in all cases participation operates to the exclusion of coercion, and that it is necessarily incompatible with high levels of work intensity. Evidence suggests that workers feel under greater pressure to work where participation is present, in part, because of the added responsibility of work that is thrust on them, as a consequence of task delegation, and due to the enforcement of high performance norms by peers as well as by managers (Barker, 1993). This suggests that the performance gains secured through participation may be due to increased pressures and expectations resulting from the implementation of these practices. Indeed, Green (2004) finds that increases in participation and changes in work organization requiring greater task flexibility were positively related to effort intensification in Britain over the 1990s. For workers, such pressures may raise levels of stress and anxiety (Ramsey, Scholarios and Harley, 2000; Danford et al., 2004), which in turn may undermine the morale and motivation of workers, and offset any positive performance effects that may arise from greater job autonomy. This may explain why extensive combinations of participatory practices and supports, where these pressures are felt most acutely, are difficult to sustain and seemingly rare in practice (Godard, 2004).

In addition, the payoffs to participation may be expected to vary according to external labour market conditions. Where workers fear losing their jobs, in the presence of high unemployment, employers will face less resistance in intensifying work and these conditions will ensure that the gains from work intensification strategies are high. Higher levels of involuntary effort may be expended because workers fear the consequences of being perceived as underperforming and consequently being dismissed. These effects are likely to be manifest in many types of workplaces; but to what degree and in what direction might the presence of participation influence the effect of the threat of job loss on productivity? Two contrasting scenarios can be envisaged, as we outline below.

On the one hand, the threat of job loss may be anticipated to support or reinforce any coercive effect stemming from the use of participatory practices. Indeed, it may be that, in certain situations, any positive effect on worker commitment from participation is swamped by the effects of a high pressure work environment and the fear of unemployment. To this extent, the positive effect of participation on performance can be attributed mainly to higher involuntary effort and thus may be seen as less sustainable as a result. In support of this view, Drago (1996) finds that participation is present in both ‘transformed’ and what he terms ‘disposable’ workplaces in Australia. In the former, performance gains are delivered through the positive impact of worker participation, underpinned by trust and long-term commitments between the firm and workers. In the latter, by contrast, threats to collective job security are used to motivate worker effort and compliance, with participation practices merely serving to channel this into productivity, quality and cost improvements. This is termed by Drago (1996: 527) ‘employee involvement through fear’.

On the other hand, worker participation may be seen as incompatible with a high threat of job loss, leading to a weak, or even negative, association between participation and performance in the presence of high unemployment. Thus, job security is often seen to be an important supporting element of participation programmes (Kochan and Osterman, 1994: 52; Edwards and Wright, 2001: 570), yet as the discussion above makes clear, the two are not always found together. Indeed, empirical evidence generally finds little or no relationship between participation and job security (Drago, 1996: 526; Godard, 2001). In terms of its effect on performance, the threat induced by unemployment may make workers less willing to use their voice, due to their fear of being perceived by management as awkward or obstructive, and this may impede information flows. Further, employees may fear that any improvements they secure in productivity will endanger their jobs or that any potential financial gains will be lost through job loss (Strauss, 1998b: 207). The threat of unemployment may generate a climate of fear and suspicion that managers will reverse the policy of participation and adopt an intensification approach, thereby undermining any trust and sense of common purpose underpinning participation (Cutcher-Gershenfeld, Kochan and Verma, 1991: 15). Given the reduction in commitment and loyalty this is likely to cause, the effect on performance from participation may be negligible in the face of high unemployment.

To date, there is no direct evidence on the effects of unemployment on the relationship between participation and performance.[1] This paper seeks to fill this lacuna by examining whether the relationship between participation and performance is affected by the threat of job loss. In doing so, the paper seeks answers to the following broad questions. First, is it the case that individual participatory practices are positively associated with productivity, as suggested in much of the literature on participation? Secondly, are combinations of participatory practices and appropriate supporting mechanisms positively associated with performance? Thirdly, are these relationships influenced by the threat of job loss? For the reasons stated above, the impact of unemployment on the productivity-participation link may be either positive or negative. We now turn to the data and methods used to address these questions.

Data and Methods

The data source used for our empirical investigation is the 1998 Workplace Employee Relations Survey (WERS 1998). This is a nationally representative survey of 2191 British establishments with ten or more employees (see Cully et al., 1999). Our research focuses on the survey’s management questionnaire which provides us with extensive information on a wide range of workplace practices and structures, as well as establishment-level performance indicators.

From this, we focus on a sub-sample of workplaces from the private trading sector providing us with a full set of observations on 1252 establishments. Two reasons lie behind this restriction: some practices, such as gain-sharing, are only meaningful for private firms, whilst inter-establishment performance comparisons may be more easily made by managers within the trading sector. Our dependent variable is the level of reported productivity, taken from the question: ‘compared with other establishments in the same industry, how would you assess your workplace’s labour productivity?’ Managers were asked to rate the productivity of their establishment on a five-point scale ranging from ‘a lot better than average’ through to ‘a lot below average’. As Table 1 shows, 44 per cent of establishments in our sample report average levels of productivity, whilst 39 per cent and 12 per cent report above average and well above average productivity, respectively.

The appropriateness of subjective performance measures has been debated at length. Concern has been raised over two issues in particular. First, to what extent are these subjective measures convergent with more objective measures? Secondly, since subjective performance measures and information on practices are typically gathered from the same respondent, is this likely to result in ‘common-method variance bias’ (see Wall et al., 2004: 99)? If ratings given for performance are biased by the information given on practices, then analysis may identify a relationship between the two which would be absent if objective performance data were being used. Wall et al. (2004) present evidence from two surveys examining the impact of HRM policies and containing both subjective and objective measures of performance. Analysis revealed the subjective performance measures to be closely related to objective measures. Furthermore, results regarding the impact of HRM practices on performance were found to be essentially equivalent whether objective or subjective measures of performance were used (Wall et al., 2004). The authors of this study conclude that there can be some confidence in findings based upon subjective measures of performance, and that where it may be infeasible to collect objective measures (for example, when data are collected at the establishment level), subjective measures may be used successfully.

Using the WERS 1998 management questionnaire, we have derived a set of variables that capture participatory practices and supporting variables within British workplaces. Our choice of variables and their frequencies are reported in Table 1. In constructing this set of variables, our decisions have been informed by the literature. Four participatory practices are represented that have commonly been identified as offering workers input into decision-making and the organization of their work (see for example, Delbridge and Whitfield, 2001; Strauss, 1998a). These four practices are teamworking, problem-solving groups (quality circles), representative participation in the form of a joint consultative committee (JCC), and two-way briefing systems. Where permitted by the data, our measures have been adjusted to capture their extensiveness in the workplace, as it has been argued that the success of participatory initiatives may be dependent upon their breadth and depth within an establishment (Strauss, 1998a). Hence, our measure of teamworking indicates whether 60 per cent or more of the workplace’s largest occupational group work is in formally designated teams. In addition, our measure of quality circles identifies workplaces where 60 per cent or more of the employees in the largest occupational group are participants in such a scheme. We include a measure of briefing systems which meet at least once a month, and in which more than 10 per cent of the time is devoted to employee questions and contributions. This measure of briefing sessions seeks to isolate instances of two-way communication in which workers are able to have some say over how their work is organized.

To test the notion that the effects of participation upon performance may be greater when accompanied by supporting mechanisms (Forth and Millward, 2004), we have attempted to capture the presence of workplace policies and characteristics that may be seen to be supportive to participation. Under ‘individual’ supports, we include three variables that capture the degree to which workers have the skills or capacity to participate effectively in their organization: disclosure of information by management; training in communication, teamworking or problem solving; and high levels of discretion over how work is carried out. Information disclosure helps workers to make a more informed contribution to workplace discussions, whilst human resource training can be seen to facilitate individual involvement in participatory practices. Our measure of discretion over work is used as a proxy for trust relations at work. Where managers report that workers have a lot of discretion over work, we take this to reflect positive levels of trust between employers and employees, which we expect to impact favourably on voluntary effort levels.

Table 1

Productivity Performance and Incidence of Participatory Policies in British Workplaces, 1998

 

Percentage of workplaces

Productivity relative to comparable workplaces

 

A lot better than average

12

Better than average

39

About average for industry

44

Below average

5

A lot below average

1

Workplace policies

Participatory practices

 

Representative forum: workplace has a joint consultative committee, works council or some other representative forum

21

Briefing: workplace has workforce briefings, meeting at least monthly which allow 10 per cent or more time for worker questions

57

Teamworking: at least 60% of employees in largest occupational group work in formally designated teams that depend on members working together

54

Quality circles: workplace has groups to solve specific problems or discuss issues of performance or quality, in which 60 per cent or more of non-managerial employees have participated in the last year.

34

Individual supports

 

Information disclosure: management regularly provide workers with information about workplace’s financial position, investment plans and staffing

36

Human relations training: members of the largest occupational group have received off-the-job training in teamworking, improving communication or problem solving in last 12 months

41

Job discretion: workers in the largest occupational group have discretion over how they do their work

28

Organizational supports

 

Internal recruitment: managers either only recruit internally, or give preference to internal candidates all else equal

29

Financial participation: at least 60 per cent of non-managerial employees have participated in an employee-share ownership scheme, profit-related or performance-related pay in the last 12 months

31

Consult: managers say that they consult with workers in making decisions

69

Combinations of policies

 

High participation practice score (at least 3 participatory practices)

25

High individual support score (at least 2 individual supports)

29

High organizational support score (at least 2 organizational supports)

38

Classifications of workplaces

 

Low participatory workplace (low score on all 3 sets of policy combinations)

34

Mixed participatory workplaces (high on one or two scores with policies from at least two sets)

61

High participatory workplaces (high score on all three sets of policies)

5

Unweighted base

1252

Weighted base

1284

Note: Base is trading establishments with 10 or more employees in Britain

Source: Workplace Employee Relations Survey, 1998

-> Voir la liste des tableaux

Our variables capturing ‘organizational’ supports consist of measures of internal recruitment, broad-based financial participation and the climate of consultation. Organizational supports are seen by advocates as providing the cultural conditions under which worker participation can succeed in raising productivity (Bryson, Forth and Kirby, 2005; Forth and Millward, 2004). Both internal promotion prospects and the possibility of sharing in the financial success of the firm can be seen as mechanisms for generating incentives for effective performance and long-term commitment. Our consultation measure attempts to capture the commitment of management to consult with workers over decisions taken, and may be seen as a further indicator of whether the organization takes participation seriously. In contrast to some other studies which examine the linkages between participatory practices, supports and performance (see Forth and Millward, 2004), we do not include a measure of job security policies as an organizational support, since previous research has indicated that there is no straightforward association between participatory initiatives and the existence of job security guarantees (Drago, 1996: 527). This decision, as we shall see below, is vindicated by our sample.

Looking at the distribution of policies (table 1), briefings and teamworking are the most common participatory practices (present in over half of workplaces). Individual supports are similarly widespread, each of our measures being present in at least a quarter of workplaces. In terms of organizational supports, while a majority of establishments (69 per cent) claim to consult with workers over workplace decisions, the incidence of financial participation and internal recruitment is somewhat lower at about 30 per cent. Overall then, most of these policies are not uncommon in British workplaces, at least when looked at individually.

However, in our empirical investigation, we are not simply concerned with the independent relationships between individual policies and reported productivity at the workplace level. We are also interested in the effectiveness of different combinations of the participation practices and supporting mechanisms since the existing literature indicates that when implemented together, performance is greater. Drawing on Forth and Millward (2004), we have created variables indicating whether a workplace scores highly on each set of participatory practices, individual and organizational supports. A high participation score is awarded if the workplace has at least three out of our four participatory practices (25 per cent of workplaces). Turning to individual and organizational supports, here workplaces score highly if they have at least two out of the three support policies (29 per cent and 38 per cent, respectively). Where workplaces score below these thresholds, they are deemed to be ‘low’ on that set of policies.

To examine the importance of participatory practices when accompanied by supporting mechanisms, we use these score variables to classify workplaces. Where workplaces score highly on all three sets of practices and supports we classify them as ‘high participatory workplaces’ (5 per cent of sample). Those workplaces which score highly on two sets of policies, or which score highly on one set and have policies from at least one other, we label ‘mixed participatory workplaces’. Also included in this group are workplaces which have two core participation practices and at least one support policy. This classification identifies workplaces with strength in some areas but not across all sets of policies. Of the workplaces in our sample, 61 per cent fall into the ‘mixed’ category. Where workplaces score low on all three sets, or score highly on one of the sets but have no policies from the other two, they are classified as ‘low participatory workplaces’ (34 per cent). Thus, we derive a classification which captures variations in both the extent of participatory practices and the degree of worker and organizational supporting mechanisms.

In addition to these measures of participation, we are able to derive a large set of control variables from WERS 1998. We control for a range of establishment characteristics (size, sector, part of a larger organization, establishment age), workforce characteristics (wages, occupational structure, training, part-time employment, union recognition, industrial action) and market factors (sales performance, number of competitors), all of which are expected to have an impact on workplace productivity. We also include a variable capturing internal job security, based on the existence of an expectation of long-term employment in the organization for the largest occupational group and the presence of a policy of no compulsory redundancies, with six per cent of establishments meeting both these criteria. This variable is entered as a control variable rather than as one of our supporting participation policies, in line with our earlier discussion of the interconnections between participation, labour market conditions and performance, with previous research suggesting that internal job security is weakly associated with participatory practices (Drago, 1996). Analysis of our sample supports this view, with the incidence of internal job security among high, mixed and low participation workplaces being eight, nine and one per cent, respectively.

Finally, and most importantly for the discussion that motivates this paper, we examine the impact of unemployment on productivity and whether the relationship between participation and productivity operates differently according to the external labour market environment. To do this, we include information on the unemployment rate in the ‘travel-to-work area’ in which the establishment is located.[2] Data on local unemployment are given in five bands, from which we create two variables indicating whether workplaces are operating in areas of relatively low unemployment (below 3 per cent) or relatively high unemployment (at or above 7 per cent).[3] At the time of the survey, the average unemployment rate in Britain stood at 4.6 per cent. These measures allow us to examine whether relatively tight or relatively slack local labour markets affect the operation of participation mechanisms through the absence or presence of the potential threat of job loss. In our sample, 25 per cent of workplaces are found in the areas of lowest unemployment compared with 11 per cent in the highest. Next we turn to our empirical results.

Results

Given the categorical nature of the dependent variable, ordered probit analysis was used for the estimation. We apply weights to the data in line with WERS recommendations (Purdon and Pickering, 2001). The main results are presented in table 2. Firstly, we examine the results for the individual and joint effects of our participation variables. Secondly, we investigate interactions between participation and the threat of job loss.

Workplace Participation and Productivity

In column 1 of table 2, we examine the independent effects of our four measures of participatory practices, three individual supports and three organizational supports on productivity, along with controls. Numerous studies of the participation-performance link have raised the question of the direction of causation in any relationship found between these variables. In other words, does participation lead to higher productivity, or does high performance provide establishments with the resources to invest in participation programmes? Existing longitudinal empirical studies suggest that the direction of causation runs from participation to performance (Handel and Levine, 2004), and we proceed on the basis that participation can be a potential cause of higher productivity.

Looking at column 1, work-based teams and quality-circles exhibit a positive coefficient, but neither has a statistically significant impact upon establishment-level productivity. Previous research into the effects of teamworking has suggested that it may, in some cases, be associated with increased work intensity, via the elimination of ‘slack-time’ (Barker, 1993). Others have argued that teams constitute a relatively ‘narrow’ form of participation, in so far as they provide workers with a limited amount of influence beyond their immediate work environment (Harley, 2001: 738), and the lack of a significant impact from teamworking in our study is in line with such an interpretation (Harley, 2001: 738). This ‘narrow’ description has also been applied by some to quality-circles and other problem-solving schemes (Delbridge and Whitfield, 2001) and may explain why no significant effect on productivity is found in our sample from this practice. Our workplace briefing variable has a positive and insignificant coefficient. A number of previous studies (Addison and Belfield, 2001; Fernie and Metcalf, 1995) have found a significant positive relationship between briefings and workplace productivity using earlier WERS data, but our study suggests that, independently at least, there is no significant productivity effect from workplace briefings. Our fourth participatory practice, the existence of a representative forum, is found to have a negative and significant relationship with productivity. This is in line with previous studies using WERS 1998 (see, for example, Perotin and Robinson, 2000). One possible explanation for the negative result is that these forums are ineffective in communicating employee voice (Dundon et al., 2001). This might be the case where management dictates the agenda, where the scope of such committees is restricted to relatively minor issues, or where employee opinions are ignored. In these circumstances, the presence of a representative forum may ultimately be counter-productive, in so far as workers come to resent the ‘charade’ of participation.

Table 2

Workplace Participation, External Labour Market Conditions and Productivity (WERS 98)*

 

1

2

3

4

5

6

Teamworking

0.102

(0.119)

 

 

0.102

(0.118)

 

 

Quality circles

0.174

(0.115)

 

 

0.180

(0.115)

 

 

Representative forum

–0.317**

(0.125)

 

 

–0.319**

(0.125)

 

 

Briefing

0.087

(0.116)

 

 

0.094

(0.116)

 

 

Information disclosure

–0.215*

(0.114)

 

 

–0.218*

(0.114)

 

 

Human relations training

0.440***

(0.120)

 

 

0.437***

(0.120)

 

 

Work discretion

0.087

(0.131)

 

 

0.084

(0.130)

 

 

Internal recruitment

–0.049

(0.124)

 

 

–0.031

(0.124)

 

 

Financial incentive

0.264*

(0.140)

 

 

0.291**

(0.142)

 

 

Consultation

0.408***

(0.122)

 

 

0.419***

(0.122)

 

 

High participatory practices

 

0.044

(0.128)

 

 

 

 

High individual supports

 

0.042

(0.129)

 

 

 

 

High organizational support

 

0.265**

(0.131)

 

 

 

 

Low participatory workplace

 

 

reference

 

 

 

Mixed participatory workplace

 

 

0.234*

(0.137)

 

 

 

High participatory workplace

 

 

0.602***

(0.214)

 

 

 

High unemployment

 

 

 

0.028

(0.123)

–0.185

(0.171)

–0.293

(0.201)

Low unemployment

 

 

 

–0.189

(0.125)

–0.156

(0.171)

–0.224

(0.201)

High participatory practices

 

 

 

 

–0.082

(0.154)

 

High individual supports

 

 

 

 

0.241

(0.161)

 

High organizational supports

 

 

 

 

0.158

(0.173)

 

High participatory practices*High unemployment

 

 

 

 

0.045

(0.275)

 

High participatory practices*Low unemployment

 

 

 

 

0.383

(0.289)

 

High individual supports*High unemployment

 

 

 

 

0.090

(0.265)

 

High individual supports*Low unemployment

 

 

 

 

–0.733***

(0.266)

 

High organizational supports*High unemployment

 

 

 

 

0.424

(0.268)

 

High organizational supports*Low unemployment

 

 

 

 

0.301

(0.268)

 

Low participatory workplace

 

 

 

 

 

reference

Mixed participatory workplace

 

 

 

 

 

0.153

(0.175)

High participatory workplace

 

 

 

 

 

0.637**

(0.298)

Mixed participatory workplace*High unemployment

 

 

 

 

 

0.518**

(0.257)

Mixed participatory workplace*Low unemployment

 

 

 

 

 

0.158

(0.265)

High participatory workplace*High unemployment

 

 

 

 

 

0.492

(0.459)

High participatory workplace*Low unemployment

 

 

 

 

 

–0.237

(0.415)

N

1252

1252

1252

1252

1252

1252

Pseudo-R2

0.106

0.074

0.075

0.109

0.085

0.079

Wald Chi2

135.62

79.80

82.53

146.82

108.45

109.35

* The dependent variable is the manager’s subjective assessment of workplace labour productivity (compared to other establishments in the same industry). All data apply to the trading sector. For details on variables used, see Table 1 and section entitled “Data and Methods” of the text. Controls included establishment size (number of employees), size squared, whether part of a larger organization, public sector, age of establishment, 12 industry dummies, 5 wage dummies, 9 occupational dummies, proportion of part-time workers (3 dummies), union recognition, industrial action, workplace training, whether sales had increased in last 5 years, market competition (3 dummies), and job security. Estimation is by ordered probit. Robust standard errors are shown in brackets.

Significance levels: *** = 1%, ** = 5%, * = 10%.

-> Voir la liste des tableaux

Mixed results are found when we examine the independent impact of supporting mechanisms on workplace productivity. Firstly, of the three individual supports, we find a significant positive relationship between human relations training and productivity. Our inclusion in the model of a variable to control for the provision of training in general suggests that there is some independent effect on productivity from specialized training that seeks to support particular participatory practices. We also find a positive and insignificant relationship between high levels of employee discretion and productivity. In contrast, information disclosure has a negative and significant relation to productivity. This may reflect the negative effects of positive information on worker motivation: here feelings of job security may act to undermine discretionary effort. Equally, it could be that the dissemination of negative news on firm performance undermines worker morale, to the detriment of productivity. In common with our findings, Peccei et al. (2005) report a negative relation between information disclosure and workplace productivity, but only for workplaces in which levels of worker commitment are high. They argue that this may reflect the increased willingness of management to report ‘bad news’ where workers exhibit high levels of commitment to the establishment. Clearly, this result presents scope for further research, perhaps using linked worker-management responses, or qualitative data.

Secondly, of the three organizational supports, two are significant. A positive relationship is found for establishments with financial participation, suggesting that, all else equal, worker motivation is enhanced by the sharing of rewards. This confirms the findings of a host of previous studies (see, for example, Pendleton, Wilson and Wright, 1998). A positive effect is also observed from our variable capturing the extent to which management are willing to consult with employees in making decisions about the workplace. This suggests that this engagement with employees can have a beneficial impact upon productivity.

Of the control variables, three are statistically significant at conventional levels. The overall state of business (increased sales) has a positive and significant relation to the level of productivity. This may reflect security of employment within establishments experiencing rising sales, although the result could also be explained by economies of scale effects. There is also some evidence that workplaces with very high wages (over £29,000 per year for the largest occupational group [around CAN$59,000]) are associated with higher levels of productivity (coefficients not reported) but there is no evidence of any significant effect at lower wage levels. On the other hand, it could be that high productivity establishments pay high wages, with the direction of causation running in reverse. Finally, we find that establishments employing up to 30 per cent part-time staff have significantly lower levels of productivity as compared with those employing no part-time employees. This suggests potential organizational problems in establishments using some part-time workers, compared to those with none or a high proportion.[4]

Column 2 investigates the possibility that high concentrations of particular participatory policies, or high concentrations of supports, are positively associated with productivity. Our results provide no evidence of a significant positive relationship between workplaces with a high score on the four participatory practices (at least 3 out of 4) and workplace productivity. This reveals that a simple ‘more is better’ approach to participatory work initiatives does not necessarily lead to higher levels of productivity. Turning to the independent effects of groupings of supporting policies, while we do not find a significant relationship between concentrations of individual supports and productivity, a high level of organizational supports is positively and significantly related to productivity. This suggests that these organizational supports are themselves complementary.

However, advocates of participation argue that success is dependent upon the co-existence of practices and supports in workplaces. We examine this argument in column 3. Here we use our classification of workplaces to examine whether complementarities between participatory practices, individual and organizational supports result in higher levels of reported productivity. Compared to the reference category of ‘low participatory’ workplaces, we find that ‘mixed’ and ‘high’ participatory workplaces are both significantly associated with higher levels of productivity. In the latter case, there is a strong, significant relationship with the coefficients suggesting that such combinations exert the greatest influence on productivity. This provides support for the notion that extensive supports are necessary for any performance-enhancing effect from participatory initiatives. These supports—some of which are individually associated with higher productivity—may facilitate participation by providing workers with the skills and information to participate in the firm, and by promoting a climate of cooperation and mutual gains-sharing.[5]

The key finding to emerge from columns 1 and 2 is that participatory practices are most effective when present alongside extensive individual and organizational supports, suggesting that the systematic adoption of participation exerts the greatest impact on productivity. But to what extent are these relationships influenced by the threat of job loss? It is to this question that we now turn.

Threat Effects

In the remaining columns of table 2, we include variables which measure the ‘threat of job loss’, captured through travel-to-work area unemployment. In column 4, we include measures of relatively high and relatively low unemployment alongside the four participatory and six support variables. The positive and negative sign for relatively high and relatively low unemployment, respectively, could be viewed as weak evidence of a ‘threat effect’, in that establishments in areas of relatively high unemployment report higher levels of productivity, but neither variable is statistically significant. The inclusion of these terms does not alter the significance of our participation or support variables.

Column 5 interacts relatively high and relatively low unemployment with measures of the depth of participatory practices and supports to investigate whether the earlier findings apply consistently according to local labour market conditions. The results reveal that there is no significant difference in productivity between establishments with high organizational supports in areas of relatively low or relatively high unemployment (here the reference category is workplaces with high levels of organizational supports in areas with average unemployment). This would seem to suggest that the relationship between organizational supports and productivity is not sensitive to the effects of unemployment. Against this, the interaction term between relatively low unemployment and high levels of individual support is negative and significant. Establishments situated in areas of relatively low unemployment with high levels of individual supports thus appear to be associated with significantly lower levels of productivity compared to those in areas with average levels of unemployment. Further analysis, in which we interacted each of our individual support variables with relatively low unemployment, revealed that job discretion is a particularly important factor in this relationship, with a significant, negative interaction with relatively low unemployment. This result is not unexpected. Hence, where workers have discretion over their work and where jobs are relatively plentiful, a stronger challenge may be mounted by workers against management efforts to intensify work, leading to lower reported establishment productivity levels. This provides some limited evidence that the effect of participation may operate differently according to external labour market conditions. However, as already discussed, combinations of participatory practices together with individual and organizational supports show a stronger relationship to productivity than concentrations of particular sets of practices. The question remains, how sensitive is the relationship between productivity and groupings of participatory practices and supports to unemployment?

This question is addressed in column 6. Even after including the unemployment interactions, it remains the case that our measure of high participatory workplaces is associated with higher than average productivity whereas interactions between high participatory workplaces and relatively high and relatively low unemployment are insignificant. There is thus no significant effect on productivity from relatively high or relatively low unemployment in high participatory workplaces, indicating that the productivity effect from high participation is independent of unemployment. The results still support the hypothesis that clusters of participatory practices and supports are associated with better performance outcomes, since productivity levels are significantly higher in high participatory workplaces in areas of average unemployment when compared with low participatory workplaces. Unemployment does have an impact on productivity in mixed participatory workplaces, however. Both relatively high and relatively low unemployment interacted with these workplaces generate a positive coefficient, but this is only significant in the case of relatively high unemployment. This suggests that any positive relationship between mixed participation and productivity is conditional upon the presence of relatively high unemployment. This finding implies that, where mixed participation exists, higher productivity is driven partly by the threat of job loss.

Participatory programmes of all kinds—low, mixed, and high—may be associated with increased pressure to raise effort levels and a high-pressure work environment, as managers seek to ensure a quick return on their investments (Godard, 2004: 368). Our results show that relatively high levels of unemployment may give managers the ‘stick’ they need to reap the advantages of a mixed system of participation in particular. It is important to emphasize that the majority of firms in our sample (61 per cent) could be characterized as adopting a ‘mixed’ approach to participation, in other words, implementing participation in a partial or relatively piecemeal fashion. This may be due to the high costs associated with introducing and maintaining participatory initiatives (Godard, 2004: 367). Where workers are fearful of losing their job, in the presence of high levels of local unemployment, this partial form of participation may be enough for management to secure higher levels of productivity. In this case, at least, enhanced productivity will involve some involuntary effort and will not be solely dependent upon increased commitment from workers. Instead, ‘speed-up’ and intensification, created by high-performance work norms and pressures that accompany participatory initiatives may be the route through which increased productivity is achieved (Barker, 1993; Edwards and Wright, 2001). Our finding that interactions between high participatory workplaces and relatively high and relatively low unemployment are insignificant suggests that where participation and supports are extensive, participatory initiatives are ‘insulated’ from any unemployment effect; however, a relatively small proportion of workplaces—around one in twenty—implement such a comprehensive and coherent system of participation in our sample.

Taken together, our results offer support to the idea that the impact of participation on productivity depends upon the threat of job loss, specifically within workplaces adopting a mixed approach to participation. The relatively limited adoption of participation by employers may be accompanied by efforts to intensify work underpinned by the threat of job loss. But we also find that the high productivity exhibited in high participatory workplaces is independent of the level of unemployment, suggesting that in these workplaces, participation can be adopted effectively in tight as well as slack labour market conditions.

Conclusion

This paper has shed new light on the interrelationship between participation, the external labour market and establishment-level productivity. It has shown that participation can work to the advantage of productivity. Whilst only a few individual participatory practices have been found to exert an independent effect on productivity, significant positive productivity effects have been observed where these practices are implemented alongside individual and organizational supports. Importantly, this relationship appears to hold, irrespective of the level of unemployment.

Yet, we have found that the positive effect of participation on productivity in establishments that adopt participation in a relatively piecemeal fashion—the ‘mixed’ participatory workplaces in our analysis—is dependent upon the presence of relatively high levels of unemployment. To the extent that ‘fear’ plays a necessary part in the success of participation in these workplaces, we are not able to rule out the possibility that participation more generally involves some element of coercion and that some of the gains from participation stem from higher work intensity, rather than simply increased levels of commitment.

Further research is needed to examine in more detail the nature of the relationship between participation and performance under different external labour market conditions. Little consideration has been given to the role of changing external labour market conditions on any participation-performance link. It may be the case that participatory initiatives are reversed or cut back as external labour market conditions worsen, for example. Alternatively, the performance effects from participatory initiatives may change over time, as variations in the level of unemployment permit management to use participation as a means of intensifying work to a greater or lesser extent. The initial attempt made in this paper to examine the impact of the threat of job loss on the relation between participation and productivity confirms that such issues warrant close attention. At the very least, our results suggest that a failure to consider the role of external labour market conditions may result in an overstatement of the performance impact of participation, and a neglect of the role that coercion might play in the apparent success of the ‘participatory workplace’.