The “hollowing-out” of trade union democracy in COSATU? Members, shop stewards and the South African Communist Party 1

The South African labour movement’s strength has often been attributed to the vitality of local democratic structures and institutionalised mechanisms of worker control that govern unions affiliated to the Congress of South African Trade Unions (COSATU). Workers and shop-stewards continue to participate actively in local union activities and sustain democratic processes but, in an apparent contradiction, the arrival of political democracy has been associated with a decline in the intensity and vitality of local union democracy. As a result the model of union democracy is shifting from participatory to representative. This article draws on the results of the fourth leg of a longitudinal survey of COSATU members and shop-stewards and compares these results with surveys conducted in 1994, 1998 and 2003. The article explores the functioning of union democracy at the workplace and compares the attitudes of ordinary members and their elected representatives. At issue is whether this decline in the intensity of shop-floor democracy has resulted in a growing representation gap between workers and their elected representatives. The article concludes by examining the implications for union democracy of the recruitment of a large proportion of shop stewards into the South African Communist Party.


INTRODUCTION
The emergence of the independent labour movement in the 1970s led to the development of distinctive trade union structures that reflected the need to break with traditional bureaucratic organisational models.Many attributed the rapid growth of the South African labour movement in the 1980s to the vitality of local democratic structures and institutionalised mechanisms of worker control.While observers like Michels 2 and others had long bemoaned the decline of internal democracy as unions matured, periodic surveys of COSATU (the Congress of South African Trade Unions) members since 1993 have demonstrated the vitality of union democracy at the workplace. 3e constitutions of COSATU and its affiliates all provide formally that control lies in the hands of directly-elected worker representatives who dominate decision-making structures.The model of worker control was developed primarily to serve the interests of small local union structures, to ensure that officials operated with mandates from the membership and to protect the organisation from state and employer repression.As unions grew rapidly and merged in the 1980s, direct worker control became less practical and was replaced by a governance model of representative democracy that 1 The author acknowledges the support of Taking Democracy Seriously project leader Sakhela Buhlungu and co-ordinators Christine Bischoff and Malehoko Tshedu of the Sociology of Work Unit (SWOP) at the University of the Witwatersrand as well as colleagues from the universities of Natal, Port Elizabeth, Fort Hare, Western Cape and the Witwatersrand who collected the data.Funding for the survey was provided by SANPAD, South Africa-Netherlands Research Programme on Alternatives in Development.The author would also like to thank the two anonymous referees for their very valuable comments. 2Michels Political parties (1962 [1911]). 3Ginsburg et al Taking democracy seriously: worker expectations and parliamentary democracy in South Africa (1995); Wood & Psoulis "Mobilization, internal cohesion, and organized labor: the case of COSATU" (2001) 28 Work and Occupations at 293; Buhlungu (ed) Trade unions and democracy: COSATU workers' political attitudes in South Africa (2006).combined direct control of elected local representatives with regional and national committee structures where full-time officials were held accountable to elected representatives.
The constitutional entrenchment of worker control and these democratic practices in COSATU and its affiliates created direct linkages between the grassroots and national decision-making structures. 4In the late 1980s COSATU's increased political involvement was to a great extent in response to pressure from the membership.Despite this, direct worker control rarely extended beyond the workplace as ordinary members only participated indirectly through their elected representatives in the election of regional and national union leaders. 5Over time, as unions grew, full-time appointed officials and full-time elected leaders have become increasingly powerful and further removed from worker control.As they matured, COSATU and its affiliated unions seem to have taken on some of the oligarchic characteristics that their organisational design sought to avoid.
An overwhelming majority of members supported COSATU's participation in the national democratic struggle and the federation became an exemplar of social movement unionism, bridging the divide between economic, community and political struggles at local, regional and national levels. 6The rapid growth of COSATU and its affiliates following mergers in the late 1980s, coupled with a dramatic increase in issues on the labour movement's national bargaining and political agenda during the transition to democracy, put pressure on an organisational model that had primarily been developed to deal with local issues.
In an apparent contradiction, the arrival of political democracy in 1994 has been associated with a decline in the intensity and vitality of local union democracy.Similarly, the broader society has been characterised by the demobilisation of many social movements.While social movement theorists tend to consider movement and institution as separate, every movement carries with it the embryo of an institution.Movement success leads to a change of focus as the goal of any democratic movement is to capture power, institutionalise its policy agenda and consolidate its gains. 7though the intensity and vitality of union democracy may have waned, by international standards three previous iterations of this survey 8 found that workers and shop stewards sustained very high levels of participation in local union activities and supported a vibrant model of worker control at the workplace.The results reported here, which draw on the fourth leg of a longitudinal survey of COSATU members and shop stewards conducted in 2008/2009, suggest a continuation of this pattern.
From the late 1980s COSATU engaged directly in national politics, complementing its shop-floor strength and grassroots links to local community organisations with national alliances with other social and political organisations.These links were formalised through the Tripartite Alliance with the African National Congress (ANC) and South African Communist Party (SACP) after the unbanning of liberation organisations in 1990. 9During the first decade of democratic politics, earlier iterations of this survey indicated a steady decline in support for the Tripartite Alliance from an overwhelming 82% of respondents to 66%. 10 The results reported here suggest that this pattern continues, with clear evidence that COSATU has strengthened its links with the SACP.This paper explores the trends in the practice of shop-floor democracy over the past fifteen years and the relationship between worker experience of trade union democracy, their participation in union activities, their attitudes towards the involvement of COSATU in politics and their own conceptions of political democracy.In particular the paper compares the attitudes of ordinary members with those of their elected representatives who constitute a significant component of the respondents, providing an opportunity to explore the extent to which their interests coincide and diverge, and how these patterns have changed over the past decade.A particularly interesting new question on membership of the SACP provides an opportunity to explore the nature of the influence of the SACP and the influence of its members on union democracy.

TRADE UNION DEMOCRACY
The existence of democratic union structures and operations at the local or workplace level provides ordinary members with the opportunity to participate directly in exercising worker control over their elected shop stewards, and indirectly over the policies of the union and office-bearers at the local, provincial and national levels.The presence of a responsive and accountable shop steward structure should encourage further member participation in the union, both formally and informally. 11More frequent participation in union meetings not only enables members to exercise greater influence and control over union policies 12 but also empowers them and boosts their confidence to participate more broadly in politics and in their communities.This relationship is best regarded as reciprocal as participation in union meetings is also likely to be positively influenced by participation in political and community organisations. 13rticipation in local union organisation is also likely to influence the political outlook of members.Lipset, Trow and Coleman 14 argued that these serve as "arenas within which new ideas are generated" and "networks through which people may learn and form attitudes about politics."Through debate and discussion in union, community and political meetings at the local level workers are exposed to diverse ideas and develop shared beliefs and understandings of government policies as well as other social and political issues.
Participation in union activities and a responsive shop steward structure at the workplace builds loyalty to trade unionism through participation, socialisation and service delivery.The involvement of workers in community and political structures, coupled with their loyalty to the labour movement, provides the foundation for a set of interlocking organisational and political relationships that strengthen partnerships with like-minded allies to advance the interests of workers and the working class more broadly. 15

Social movement unionism and worker control
The label of social movement unionism was first applied to describe the model of union organisation practiced by South African unions that emerged in the 1970s and flourished in the 1980s.Combining the workplace organisational strength and collective bargaining focus of economic trade unionism with modes of collective action and societal consciousness typical of social movements, social movement unionism links factory-based production politics with community and state power issues. 16As an idealtype, social movement unionism embraces an enduring commitment to participatory democracy and political independence and is distinguished by a "bottom-up" organising model that places control of the union in the hands of the members and their representatives, in contrast to the "top-down" model of organising where union leaders and officials exercise control.
At the local level social movement unionism is not only embedded in the workplace but is linked to local political and community organisations and movements, where a common agenda to transform society in the interests of the working class is pursued.In this way unions are not just organisations but form part of a larger movement, and through these linkages a distinctive and transformative union identity may be forged and promoted. 17 Union democracy and worker control need to be understood in the context of the distinctive union social structure that has developed in South African workplaces. 18The union social structure comprises the formal institutional framework, including its constitution, offices, resources and the rules and procedures that govern shop steward elections and roles, as well as a range of informal relationships, practices and meanings 15 See Hirschsohn (2007). 16Lambert Political unionism in South Africa: the South African Congress of Trade Unions, 1955-1965  (1988). 17See Hirschsohn (1998). 18Von Holdt "Social movement unionism: the case of South Africa" (2002) 16 Work, Employment and Society at 283. around them.The union social structure is influenced by a range of societal and political factors beyond the workplace and "governs the distribution of power between members, shop stewards, officials and its various structures, and defines the practices, processes of decision making, strategies, goals and organisational culture of the union". 19nventional analysis of unions in advanced industrialised societies identifies bureaucratisation and the centralisation of power as inevitable features of mature trade unionism.In his seminal study of the social democratic party, Michels 20 concluded that even organisations with the most democratic ambitions, political parties, inevitably displayed bureaucratic tendencies.Michels's "iron law of oligarchy" argues that, as democratic organisations grow, they undergo subtle organisational changes: the decline of participation leads to member passivity and the professionalisation of the organisational machinery, which results in the ascendance to power of a bureaucratic oligarchy.The monopoly of political skills in the hands of the leadership is one of the chief factors perpetuating their power. 21o distinctive features characterised the organisational model of South African unions that emerged in the 1970s and thrived in the decade thereafter.The combination of high levels of direct participation in local union activities and worker control of union officials through elected representatives sought to counteract oligarchic tendencies.But, when COSATU affiliates merged to form large national unions, the democratic ideals of worker control and direct participation became impractical and representative democracy increasingly replaced direct participation. 22However, there are risks involved in this development.As Lange has argued, representative democracy is a pragmatic model that only comes close to these ideals if "tempered by a keen awareness of the desirability for the most rank and file participation possible consistent with effective promotion of interests". 23ter COSATU was established and its affiliates founded through a series of mergers in the late 1980s there seemed to be ample evidence that union growth does not necessarily lead to membership passivity or the ascendance of an oligarchy.Two decades later, COSATU's representative democratic organisational model still remains in place.However, the functioning of systems of democratic control depends on how these structures operate in practice, the extent of mobilisation and participation by rank-and-file membership (rather than their elected representatives), and whether or not a non-representative faction or oligarchy has ascended to fill the democratic space left by declining participation.
The origins of the organisational model in COSATU can be traced back to labour movement entrepreneurs, who revived unions in the 1970s and sought to avoid the 19 See Von Holdt (2002) at 287 20 See Michels (1962[1911]) 21 See Lipset et al (1962)  22 See Hirschsohn (1998) 23 Lange Union democracy and liberal corporatism: exit, voice and wage regulation in postwar Europe (1984) at 10 perceived failures of unions in the 1950s, when the demands of the political alliance to the ANC tended to be prioritised over the development democratic grassroots organisational structures.They also drew on the experience of the British shop steward movement and the participatory ethos of new social movements. 24In the absence of legal political organisations, and to protect themselves against state repression, the unions built democratic organisations based on worker representation and control. 25 building a democratic organisational culture from the bottom up, "worker control" sought to break with traditional representative organisational forms and emphasised direct democracy, accountability of delegates, open debate, education of activists, and worker participation in decision-making. 26Buhlungu notes that union democracy and democratic union traditions take on particular forms because of the context within which they emerge and develop and a wide range of cultural, traditional, political and intellectual influences. 27This democratic union culture comprised a set of practices and organisational values that permeated every aspect of union organisation and functioning, which entailed: (a) the emphasis on shop floor structures led by shop stewards; (b) the creation of representative decision-making structures with majority worker delegates; (c) mandated decision-making and regular report-backs to members; (d) firm-level bargaining so workers and stewards can control the bargaining agenda and agreements; (e) full-time officials subjected to control by worker-dominated structures; and (f) the involvement of workers in the employment of full-time officials. 28ce institutionalised, the organisational characteristics of unions, like other organisations,, tend to resist adaptation to new conditions. 29Consequently, these principles have endured in the governance structures of COSATU and its affiliates.Consequently, the original organisational models and the political culture that characterised the unions that emerged in the 1970s became imprinted in the organisational forms of COSATU affiliates today.

Defining union democracy
24 Kraak Breaking the chains: labour in South Africa in the 1970s and 1980s (1993). 25Maree "The emergence, struggles and achievements of black trade unions in South Africa from 1973 to  1984" (1985) 18 Labour, Capital and Society at 278. 26 Fine "Civil society theory and the politics of transition in South Africa" (1992) 55 Review of African Political Economy at 71. 27 Buhlungu "The rise and decline of the democratic organizational culture in the South African labor  movement, 1973 to 2000" (2006). 28See Buhlungu "The rise and decline of the democratic organizational culture" (2006). 29Warner "Organizational background and union parliamentarianism" (1970) 12 The Journal of Industrial Relations at 205.
What is meant by "union democracy" or by what standards should it be measured?Two of the four models of union democracy identified by Morris and Fosh,30 namely liberal pluralism and grassroots activism, predominate in the literature.The authors argue that each model considers five elements: constitutional arrangements, political organisation, representation, membership involvement and outcomes for members.Writing from a perspective that combines elements of liberal pluralism and grassroots activism, Stepan-Norris and Zeitlin 31 identify three basic features of union democracy: 1.A democratic constitution that guarantees of basic civil liberties and political rights; 2. The freedom of members to debate and criticise and union officials and to organise, oppose and replace officials through freely contested elections; and 3. Maximum participation by members in exercising power and in making decisions which affect them.
Conventional approaches to union democracy tend to emphasise the first two features, which are characteristic of the liberal pluralist approach which focuses on formal structures and measures such as the constitutional ability of the rank and file to affect decisions, influence policy and change leaders, or the responsiveness of leaders to members' demands.
However, as Flanders observed, unions are a mixture of organisation and movement, requiring organisation for their collective bargaining power and movement for their vitality. 32Undue stress on the institutional framework can deflect attention away from the importance of activist rank-and-file participation which is essential, not only to ensure that members' interests are advanced but also to defend the membership against oligarchy. 33 the South African context it was relatively easy to entrench a democratic union structure and culture because these principles complemented the democratic goals of the liberation movement. 34Unusually the movement's early full-time officials were political activists committed to building union democracy and developing a cadre of worker leaders.However, the rapid growth of COSATU affiliates following mergers in the late 1980s, coupled with the federation's rapidly expanding national bargaining and political agenda, put pressure on an organisational model that had been developed to deal with local issues.
Over time, continued union growth and maturity led to the erosion of direct democracy and increasing reliance on representative democracy through elected shop stewards.It also resulted in a growing gap between workers and shop stewards and 30 Morris & Fosh "Measuring trade union democracy" (2000) 38 British Journal of Industrial Relations at 95. 31 Stepan-Norris & Zeitlin "Insurgency, radicalism and democracy in America's industrial unions" (1996)  75 Social Forces at 1. 32 Flanders Management and unions: theory and reform in industrial relations (1970). 33Ramaswamy "The participatory dimension of trade union democracy: a comparative sociological view" (1977) 11 Sociology at 465. 34 See Buhlungu "The rise and decline of the democratic organizational culture " (2006)  between workers and the leadership at branch, regional and national levels.The steady erosion of worker control over policy issues that were remote from the shop-floor was already identified in the early 1990s by Marie, who also noted that democratic structures had often become top-down communication conduits rather than mandating channels. 35mocratic worker control is thus necessary, but not sufficient, to ensure that leaders remain accountable and the movement advances the interests of members.Unless supplemented by active rank-and-file participation, the institutionalisation of worker control is not a sufficient defence against oligarchic union leadership.It is the active involvement and participation of members that ensures that organised labour retains its movement character and advances its social purpose and the interests of the membership.
While mandating representatives and holding them accountable through periodic feedback and the power to recall are important checks and balances in any system of representative democracy, another important factor is the extent to which the views of ordinary members and their elected representatives are aligned.This may be particularly important when representatives act with broad mandates and are not expected to provide in-depth report backs.Consequently, an important dimension of the empirical analysis in this article, which has not previously been explored in the South African context, is to consider the similarity and differences between ordinary members and shop stewards in depth.
In discussing democracy in the union context it is necessary to draw on the conventional distinction between participatory democracy, where individual members are directly involved in decision-making, and representative democracy, where decisions are made by elected representatives.The line between the two is anything but clear, however, as participatory and representative democratic practices often complement one another in practice, for example when collective agreements are negotiated by elected representatives and ratified by ballot.
Union democracy can best be assessed by the opportunities that members have to participate in decisions that affect them, their ability to influence policy, exercise control over their representatives and hold them accountable, and by the responsiveness of representatives and leaders to their demands. 36As Strauss argues, union democracy is not only desirable because it enhances member control over the office-bearers, but also because "on balance democracy increases union effectiveness in representing members' interests and in mobilising these members to support its collective bargaining objectives." 37cause direct democracy is feasible only on the shop floor, where large numbers of members can participate personally, effective representative democracy depends on the accountability of representatives and the responsiveness of leaders.Ensuring 35 Marie "COSATU faces an organizational crisis" (1992) 16 South African Labour Bulletin at 26. 36 See Hirschsohn (2007). 37Strauss "Union democracy" in Strauss, Gallagher & Fiorito (eds) The state of the unions (1991).
responsive leadership requires that members are in a position to oppose their leaders' policies and to change their leaders if they become non-responsive.
The balance between representative and participative democracy depends on the mandate that members allow their elected representatives and the extent to which they are expected to consult with and report back to members.As unions have grown and representative democracy has necessarily replaced participatory democracy, so in assessing the extent of membership control one must distinguish between control over decisions and control over representatives. 38At the local level democracy may erode and an oligarchy emerge unless members actively enforce accountability of their representatives by requiring ongoing mandating and consultation, and replacing those who fail to perform and conform to expectations.
Even in highly democratic organisations the de facto power of members varies according to their personal interest in outcomes and the extent to which they are directly affected by decisions.Leaders are more likely to be held closely accountable on contractual negotiations than over policy and political issues. 39As the Organisational Report to the 10 th COSATU National Congress notes, internal democracy is weakened by "poor attendance at constitutional meetings, though affiliates still maintain effective mandating and report-back systems for collective bargaining". 40 their classical study, Lipset et al adopt a liberal-democratic perspective and identify the existence of competing parties or factions as essential to functioning democracy. 41The principle on which this critical factor is founded is the existence of autonomous centres of power that create the conditions necessary for opposition to officials and an independent organisational base for rank-and-file activists.Stepan-Norris argues that, even in organisational contexts that are not conducive to democracy, the introduction of radical ideology and factions based on outside organisations can provide structural support for internal democracy. 42If these factions are based on ideological differences that are aligned to worker interests they may stimulate and maintain interest and participation.The alliance between COSATU and the SACP highlights the potential importance of this dimension of influence over democratic processes within COSATU and its affiliates.Some proponents of business or economic unionism have argued that the purpose of unions is to protect and advance the living standards of members rather than provide members with experience in self-government. 43However, even from a functionalist perspective one can argue that democracy can increase union effectiveness in a number of ways and for a number of reasons.Elected leaders are more likely to know what their members want and thus better represent their interests.Democratic processes help to 38 Hochner, Koziara & Schmidt "Thinking about democracy and participation in unions" in Proceedings of the 32nd Annual Meeting of the Industrial Relations Research Association (1990), 39 See Hirschsohn (2007). 40COSATU Organisational report to the tenth COSATU national congress (2009) at 15. 41 See Lipset et al (1962). 42Stepan-Norris "The making of union democracy" (1997) 76 Social Forces at 475 recruit and develop voluntary leadership which can improve service delivery at the local level.The power to recall elected representatives makes it easier for the union to eliminate ineffective office-bearers or those who fail to represent member interests adequately.Finally, grassroots organisers may assist in organising and mobilising members.

Measures of union democracy
Strauss identifies a number of operational measures of union democracy. 44At the most basic level South Africa unions all pass the legal test in that their constitutions provide for the regular election of officers.Another test relates to whether or not members can run for office and speak in opposition to the leadership without fear of reprisal.Third, various measures reflect whether or not the right to participate is used and whether incumbent representatives or officer-bearers are defeated.
No empirical research has systematically documented the extent of contestation or whether office-bearers in South African trade unions have been defeated in elections.Constrained by the focus of the survey on the views of ordinary members, this article considers measures of individual participation as well as measures of responsiveness and influence.Do members feel they have some "say in how things are decided"? 45To what extent can they influence union decisions? 46Do shop stewards and office-bearers share the values and priorities of their members?
We begin by considering the trends in member attitudes towards steward accountability and an analysis of participation levels in union activities.In order to better understand the extent to which stewards are true representatives, the paper compares the responses of ordinary members and shop stewards on a wide a range of issues.Finally, the potential influence of SACP membership is explored to better understand the impact of the alliance on union democratic practice.

METHOD
The data reported here represents the fourth iteration of a project titled Taking Democracy Seriously, the only regular nationwide survey of members of affiliates of South Africa's pre-eminent trade union federation, COSATU.The study was initiated by a group of labour scholars at the universities of Cape Town, Natal, Port Elizabeth, Rhodes and Witwatersrand in 1994 47 and was repeated before each general election since then that is, in 1998 48 , 2004 49 and between November 2008 and February 2009.The most recent survey was coordinated by the Sociology of Work Unit (SWOP) and 44 See Strauss (1991). 45Tannenbaum & Kahn Participation in union locals (1958), cited in Strauss (1991). 46See Anderson (1978). 47See Ginsburg et al (1995). 48See Wood & Psoulis (2001)  49 See Buhlungu Trade unions and democracy (2006).
Because of redundancies and job mobility, a panel study could not be conducted with the same respondents.Instead, a multi-layered sampling approach was adopted.In previous iterations of the survey large unionised workplaces were identified in each province and then randomly selected to ensure that the major COSATU unions operating in each province were represented.
Face-to-face interviews were conducted in late 2008 and early 2009 at the same sites as the 2004 study where employers granted permission.Where this was not possible, workplaces were selected in the same sector to ensure that sectoral coverage woud be consistent with previous surveys.After employers granted access, a sample of ten union members was drawn randomly from each workplace.Although not ideal, the sampling strategy ensured that general trends could be identified.

Accountability of shop stewards
For ordinary members the union is given shape and meaning through their shop stewards who service grievances, provide information, influence how members vote and whether they attend meetings, and thus provide opportunities to interface with the union.If stewards are responsive to members' desires the union itself will be more effective and responsive.Furthermore, when members elect their shop stewards they are effectively evaluating their responsiveness and that of the union to their needs. 51he questions in the survey primarily relate to the culture of participation between union members and shop stewards.The questions address the mandate that stewards operate with, the extent to which they are accountable to members and are expected to report back on decisions, and whether member should be able to replace stewards who do not represent their interests.
Consistent with the surveys in 1998 and 2004, 92% of respondents reported that shop stewards were elected by workers rather than being appointed by the union or management.Tables 1, 2 and 3 reflect the expectations of workers on how the process of worker control should operate at the workplace.A comparison of the views of shop stewards and members on all three questions shows no statistically significant 50 An in-depth discussion of the survey methodology is provided in Buhlungu Trade unions and democracy  (2006). 51See Strauss (1991).differences, reinforcing the view that shop stewards and members have similar views on these questions.52The results in Table 2 indicate that minimal change has taken place over the past three surveys in the expectations that shop stewards need to be accountable and report back.However, a clear shift from specific to open mandating is evident in Table 3. Almost all members expect stewards to consult periodically, particularly regarding important issues.It is noteworthy that while the majority (54%) still expect stewards to consult every time they act on workers' behalf, this has declined significantly since the first survey (76%), suggesting that the style of democracy on the shop-floor has been shifting from mandating to delegating.The answers to these three questions indicate the extent to which the fundamental principle of steward accountability to members has become an ingrained expectation of shop-floor union culture, despite the shift from direct mandating to delegation.This stable pattern of accountability is reinforced by the finding that an overwhelming majority of respondents (93%) believe that shop stewards may be recalled by the membership if they do not do what the members want.This is consistent with previous surveys, and these views are shared by stewards and members alike.Members continue to replace non-performing stewards who have failed to fully represent their interests, as is shown in Table 4.There is an oscillating pattern over the four surveys indicating that members continue to exercise their power to remove shop stewards.What is unclear from the question is whether the removal of non-performing stewards has resulted from the withdrawal of the stewards' mandate during his/her term of office or has taken place during periodic elections.

Promotion of Stewards into Management
The introduction of questions on the promotion of shop stewards into management in the 2004 survey for the first time recognised the well-established role of the position as a mechanism for climbing the organisational hierarchy.The practice has become more prevalent as the majority of respondents (52% versus 37% in 2004) now report instances of the promotion of stewards in their workplace (see Table 5).This also reflects the increasing preparedness or willingness of shop stewards to make themselves available for promotion into the managerial ranks.As the practice has become more common, the knowledge gap between stewards and members about the existence of promotions into management also has diminished.Table 6 indicates a moderate positive shift in the attitude of both shop stewards and members towards the acceptability of stewards being promoted into the managerial ranks.It is noticeable that the vast majority of shop stewards (74%) now agree that this behaviour is acceptable.The growing practice of promoting stewards into management not only raises the possibility of the erosion of internal solidarity but also brings into question the accountability of stewards to the membership.Clearly, the historical animosity between unions and management ranks appears to be eroding as the practice of promotion becomes more acceptable and more common.
However, if shop stewards are acting to advance their own careers rather than representing the interests of their members, the process of democratic accountability to the membership will necessarily be eroded over time.

Member participation and participatory democracy
In her seminal work on participatory democracy, Pateman argued convincingly that this form of democracy becomes self-sustaining because it socialises members into continuous participation by developing and fostering the skills they require to engage further. 53While attitudes and behavioural intentions of members are useful to understand the normative expectations of union members, it is member participation in a variety of union activities that determines whether the democratic culture is primarily participatory or representative.If the purpose of unions includes empowering members to help determine their employment conditions, one test of union effectiveness (not just democracy) is the extent to which members participate individually. 54ions depend on the support of the majority of members to support the activist core.This is particularly the case for COSATU affiliates that require high levels of participation to sustain the democratic culture discussed earlier.The range of measures that have been used to assess member participation typically focus on voting in elections for union representatives, office-bearers and officials, the frequency of attendance at union meetings, holding office as a shop steward or other official, and engaging in collective action.Where unions have formal and informal linkages with political parties and the community, union members, particularly shop stewards, wear a representative hat when they engage in these organisations.Participation of union members in other organisations, whether or not they wear a union hat, is integral to the notion of social movement unionism, which is founded on the articulation of labour, political and community struggles. 53Pateman Participation and democratic theory (1970). 54See Strauss (1991).
The frequency of shop steward elections is measured using two questions.While the results in Table 7 suggest that the frequency of steward elections has declined steadily over the past three surveys, the results are not clear-cut when respondents were asked when they last elected their shop steward (Table 8).73% indicated that they had voted within the past two years.No clear trend is evident as more respondents participated in elections in the past two years, compared with the 2004 survey, but the frequency of voting is lower than the 1998 survey.Overall, the results suggest less frequent elections, and presumably a more stable cohort of elected stewards.The results in Table 9 indicate very clearly that weekly attendance at union meetings by both shop steward and members has declined markedly between 1998 and 2008.While the typical shop steward attended union meetings on a weekly basis in 1998, by 2008 this had reduced to attendance on a monthly basis.Similarly, while one-third of ordinary members were attending meetings every week in 1998, this ratio has fallen to about one in six ordinary members a decade later.The steady decline in intensity of involvement by ordinary members in the past decade points to a significant decline in union mobilisation and has important consequences for the sustainability of the democratic culture that sustained the model of worker control.
It is also indicative of a steady shift from participatory democracy to a more representative style.Despite these significant declines, about two-thirds of ordinary members attend union meetings at least on a monthly basis, a high ratio by international standards.Regular meeting attendance is consistent across the sample with the majority of members in 12 of the 14 unions surveyed reporting that they still attend at least one union meeting per month.
The participation of members in industrial action (see Table 10) represents a further measure of engagement in union activities.Although respondents were not asked whether they had personally participated in industrial action since 2004, two-thirds reported that there had been some strike activity at their workplace in the past four years.While there is no clear trend over the past decade, the results also do not support public perceptions that the level of strike activity in the public sector is much higher than activity in the private sector.The results also suggest that although large scale strikes in the private sector may not be common, industrial action persists in this context.There are two ways to interpret this data.On the one hand, the overall pattern of participation in structured activities -shop steward elections, union meetings and industrial action -remains relatively high by historical standards and very high by international standards.This suggests that members of COSATU affiliates remain active participants in their unions.On the other hand, the significant decline in the intensity of activist-type participation, represented by attendance at weekly meetings, is suggestive of the erosion of the culture of participatory democracy.Just as active engagement becomes self-sustaining and creates an empowered and socialised membership, the steady decline in participation by ordinary members is disempowering and weakens the processes through which elected representatives can be held accountable, even if attitudes towards accountability remain largely unchanged.

Alignment between values and priorities of shop stewards and members
The third question that Strauss raises is whether or not shop stewards and officebearers reflect the values and priorities of their members.The reasons for alignment or misalignment may be varied, but the efficacy of democratic processes and practices is reflected in an alignment of views between members and their representatives.To some extent one would expect that the attitudes of elected stewards will be affected by exposure to different viewpoints and influences through the performance of their representative roles.
In instances where democratic procedures are absent or weak, leaders and shop stewards may systematically misinterpret the preferences of workers because they base their choices on the preferences of an activist group or those who attend meetings most frequently.Similarly, the preferences of the majority of workers may not be represented by their elected representatives if they are expressing views on issues that have not been subjected to a process of validation through a worker vote. 55ile the majority of survey respondents are ordinary members, the survey includes a disproportionately large representation of shop stewards (26%).This provides a large enough group to compare their attitudes and behaviours with those of ordinary members, and thus establish to what extent shop stewards represent the interests of members.The areas of commonality are clearly as important as those where they differ.
The survey results reveal that members and shop stewards share expectations about the relationship between members and their elected representatives.Both members and their representatives agree on the need for shop stewards to obtain mandates from the membership, to consult on decisions and on the frequency of feedbacks required.Moreover, an overwhelming majority of stewards and members (over 90%) recognised the right of members to remove stewards who have lost the confidence of the members.Together, these questions clearly suggest that procedural democratic principles are widely shared across COSATU affiliates, with the exception of SASBO (South African Society of Bank Officials) where there is limited evidence of a functioning structure of elected and accountable shop stewards.
It is also noteworthy that ordinary members and shop stewards appear to be equally aware of the various committees or structures for workers to participate or communicate with management.However, there is a marked difference in beliefs about the impact of participation.72% of shop stewards believe that that these bodies have some or extensive influence over management, while only 54% of members share these views.Almost half of shop stewards, who are more likely to be directly involved in these bodies, believe that participation results in extensive influence over management.Their experience of direct participation with management and the belief that they help workers gain some influence over management decisions, may determine the significant difference between ordinary members and shop stewards concerning promotions into management discussed earlier.
These findings support the view that participation in employer consultation structures increases the distance between members and shop stewards.Not only do stewards appear to perceive that relationships with management are more effective but the attractiveness of a management position is likely to increase as perceived influence over management decisions increases.
Levi et al 56 further emphasise that, while the benchmark of participatory democracy is a high degree of participation in voting, meetings and policy debates, this is only truly meaningful if members are well informed.Knowledge of past policies and decisions and access to relevant information is essential for contributing to debates.This is consistent with the programme of Organisational Renewal that COSATU and its affiliates committed themselves to following, as recommended by the September Commission of 1997. 57The programme's aims include ensuring (a) that members are provided with excellent service in terms of political and labour-relations education and (b) that every affiliate maintains internal democracy through regular and well-attended constitutional meetings, support for shop stewards, and well-defined and effective report-back systems.As indicated in Table 11, there has been a precipitous decline overall in policy knowledge levels, with the exception of the Reconstruction and Development Programme (RDP), among both shop stewards and ordinary members.NEDLAC, the National Economic Development and Labour Council, provides a critical forum for organised labour to participate in the formulation of legislation and policy directly related to the interests of individual workers.Over the past decade the prominence of NEDLAC has declined as the volume of legislation and policy has receded.It is to be expected that, despite NEDLAC's potential importance, there would be a decline in the proportion of members who are familiar with the institution or have attended meetings where there has been a report-back.COSATU's deputy general secretary, Bheki Ntshalinshali, acknowledges that implementing democracy is complex and that continuous report-backs and mandating on issues dealt with at NEDLAC has not been part of union culture. 58e declining level of knowledge among shop stewards, however, suggests that union educational efforts are having less impact and that grassroots leaders are increasingly marginalised from union decisions concerning broader policy issues.Not surprisingly, the data also reveals significant differences between shop stewards and ordinary members in all the questions relating to knowledge about NEDLAC, GEAR and the RDP, as well as attendance at meetings where government policy and policy-making bodies were discussed.Although individual members are far removed from policy negotiations at NEDLAC, the limited participation of their representatives in meetings where policy issues are discussed raises substantive questions about the extent to which effective mandating processes are in place.
The only question where the views of members and shop stewards are similar related to whether the government was achieving the goals of the RDP.Despite these differences, even among shop stewards a relatively low proportion had attended meetings where there had been any discussion of these issues.This highlights the likely gap on matters of policy and politics between union official, elected representatives and the membership.
In comparison with ordinary members, shop stewards are more likely to (a) believe that unions should have active links with community organisations, civil society groupings, or social movements, (b) be involved in local government or communitybased development initiatives, and (c) be a member of the SACP.While only 18% of the ordinary members surveyed are signed-up or paid-up members of the SACP, more than 40% of shop stewards are SACP members.This distinction is explored in depth later in the paper.Surprisingly, shop stewards are less likely than ordinary members to have attended a meeting of the political party they support.As the data in Table 12 suggests, support for the ANC remained remarkably stable until the pronounced decline in committed ANC support in 2008.Some of this can be attributed to the fact that 23% of the 2008 sample refused to answer the question, compared to only 16% in 2004.Further analysis of the provincial breakdown in support for the ANC reveals that the Western and Eastern Cape were solely responsible for the dramatic decline in expressed support.In the Western Cape only 20% declared their intention to vote for the ANC, while in the Eastern Cape only 31% did.The emergence of an ANC breakaway, the Congress of the People, and the strong shift from the ANC to the Democratic Alliance in the Western Cape, clearly played a role in the willingness of respondents to address the question in the run-up to the national election.
To some degree the shift can also be ascribed an environment of political instability at the time of the survey, as approximately 40% of respondents in these provinces refused to answer this question.In the remainder of the provinces ANC support levels remained steady compared to the previous survey.While the gap between ordinary members and shop stewards remains about 12%, it is also significant that declining levels of support for the ANC among shop stewards was already evident in 2004.
There is a potentially significant shift occurring in patterns of support for the Tripartite Alliance and support for the ANC among both members and stewards.Whereas previously support for the ANC was significantly higher than current and future support for the Tripartite Alliance, expressed support for the alliance is now stronger than support for the ANC itself.Given the professed importance of the alliance to COSATU and the high membership of the SACP, particularly among shop stewards (see Table 13), this may indicate that support for the alliance is linked to primary political allegiance to the SACP rather than the ANC.

SACP membership and union democracy
The formation of the Tripartite Alliance formally established the association between COSATU and the SACP in 1990.Initially, however, much of the political work of the SACP was focused on building support for the ANC and many SACP leaders also held leadership positions in the ANC.Similarly, COSATU was instrumental in mobilising voter support for the ANC in the general elections.Despite this, the political influence of the SACP and COSATU in the Tripartite Alliance has often been marginal.This was most evident with the adoption of GEAR, a neo-liberal macro-economic policy, for the decade after 1996.
The strengthening of the relationship between the SACP and COSATU in recent years (see Table 14 below) has been a natural and strategic result of a common agenda to advance a "Left strategy" devised to "capture the ANC" from within, "with the objective of shifting government strategy in a pro-poor and pro-working class direction". 60The introduction of a question on SACP membership in the current survey enables us to explore in some depth the inter-relationship between SACP membership and trade union activism, and the role that left-wing shop stewards and activists may play in leading and sustaining shop-floor democracy and mobilisation.The 12th Congress of the SACP revealed that the industrial employed working class represents under 40% of SACP membership. 61COSATU's leadership highlighted that in order to strengthen their alliance with the SACP more work was required to convert members into staunch socialists who are active in the SACP. 62If membership of the SACP is accompanied by education and mobilisation, one might expect enhanced levels of political understanding among union members and activists.
The most striking feature of the data in Table 13 is the high level of signed-up and paid-up membership of the SACP reported by shop stewards (45%) outside the Western Cape, compared to 21% for ordinary members. 63Of those who reported to be SACP members, 72% claimed to be paid up while 28% claimed to be signed up.If the data for ordinary members is to be believed and the sample is representative of COSATU as a whole, then the SACP has about 215,000 paid-up members and another 85,000 signedup members within COSATU.While the close relationship between COSATU and the SACP in recent years may have translated into significant cross-membership, this amounts to four times the claimed membership of the SACP at 73,000. 6461 COSATU Political report to the tenth COSATU National Congress (2009) at 26. 62 See COSATU Political report (2009). 63In aggregate over 70% of all respondents who indicated that they were SACP members claimed to be paid-up. 64Bell "Inside labour: Debate needed on COSATU's SACP agenda" (2009) Business Report.
While the dominance of Marxist rhetoric is clearly evident in COSATU's political discourse 65 and SACP members are prominent in COSATU leadership positions, the data in Table 14 reveal the extent to which the SACP has recruited grassroots leadership throughout COSATU.This reflects the strengthening SACP-COSATU linkages within the Tripartite Alliance more generally.In the run-up to the ANC's National Conference in Polokwane in December 2007, SACP and COSATU members made a concerted effort to infiltrate ANC branches across the country in order to ensure that their interests were fully represented in the ANC leadership race. 66ile the intensified alliance between the SACP and COSATU at top and grassroots leadership may strengthen their position in the Tripartite Alliance, it also raises questions about a potential representation gap on political issues, particularly in unions (such as NUMSA and SATAWU) where the majority of shop stewards are SACP members but only 20% of members claim to be.The data in Table 15 also suggest that SACP members are generally much more active participants in union affairs than workers who are not SACP members.The majority (60%) of shop stewards who attend union meetings on a weekly basis are SACP members, highlighting either the effectiveness of the SACP in recruiting among this activist group, or the importance of the SACP to sustaining relatively high levels of union participation among an activist group, in a context where the overall intensity of participation has declined markedly (see Table 9).In the absence of data on SACP membership in the earlier surveys it is not possible to establish patterns more conclusively.
The question also arises as to why a large proportion of shop stewards and ordinary members indicate that they are SACP members when they probably are not.The idea that these SACP members are all genuine left-wing activists, who adopt an oppositional 65 See COSATU Political report (2009). 66See Southall & Webster (2010).
attitude towards management and exclusively advance the interests of the working class, should probably be dispelled as an overwhelming majority (73%) of shop stewards now believe that it is acceptable for shop stewards to be promoted into management (see Table 6).
Buhlungu's study of full-time officials has highlighted the emergence of entrepreneurial and career unionists, in addition to the more traditional ideological unionists. 67There appears to be a similar pattern at the workplace where the position of shop steward may not only be utilised as a stepping stone into a full-time position as union official but increasingly to access opportunities in managerial ranks or in government.This suggests that in the majority of cases the decision to affiliate with the SACP is more likely to be pragmatic or opportunistic than ideological.As suggested by Southall and Webster, there are strong indications of political factions within COSATU behind the façade of unity.They argue that "factionalism could easily translate into a scramble for jobs" in government. 68Similarly, shop stewards may be using their SACP membership opportunistically to align with a union faction where the leadership is dominated by SACP members.Identification with an SACP-led faction may partly explain why so many respondents claim to be paid-up or signed-up SACP members, when they probably are not.

Implications for union democracy
Does the presence of a high proportion of SACP members among the shop stewards surveyed threaten or strengthen union democracy?On the one hand, from a liberal pluralist perspective one could argue that the presence of organised factions or platforms contesting for power within the union might represent the best defence against oligarchy.This argument presupposes that the existence of factions is well known to representatives with voting rights.Of course, many liberal pluralists may not be comfortable with the notion that a leading faction was communist.On the other hand, the election of representatives who may not directly reflect the interests or views of the members they represent may create problems of legitimacy and the potential for the union to pursue policies that do not reflect those of the majority of members.
Alternatively, from an activist perspective, one may argue that an outside organisation like the SACP may sustain, stimulate or revive activism with the labour movement, enhancing the interests of all members.However, in COSATU affiliates there is no pure electoral democracy where members can vote directly for policies or national leadership candidates they prefer.Consequently, in participatory democratic systems where leaders are selected by elected representatives, the most active members dominate and may misrepresent the majority interests.
Writing about the British experience, Darlington notes that most shop stewards are elected because of their commitment to fight to advance workers' interests at the workplace rather than because of their broader political beliefs. 69The more adversarial approach adopted by left-wing stewards may be valued by members irrespective of their political affiliation.Where stewards have been removed, the main reasons given in the survey is because they were not doing their job properly (60%) or were regarded as being too close to management (24%), rather than on ideological and political grounds (4%).
Similarly, Bacarro 70 argues that if unions pursue the "logic of mobilization" rather than the "logic of representation" and take action based on the members or representatives who are most willing to engage in collective action, the interests of the less-active majority may be ignored.In addition, activists and elected leaders use persuasion to influence members' views as workers often rely on their leaders to evaluate alternative policy options, particularly when specialist knowledge is required.In the dialogue associated with democratic decision-making, leaders may also strongly influence members by (re-)shaping their perceptions of their own interests. 71ere is a significantly greater proportion of SACP members among shop stewards than among ordinary members in most COSATU unions.Consequently, some shop stewards may be advancing the interests of the party rather than the preferences of the majority of union members.This survey has not explored their reasons for joining the SACP or their impact of membership on the role as shop stewards.It thus remains a subject for further research to identify whether or not SACP membership serves to advance the interests of the members they represent.

CONCLUSION
At twenty-six years old, COSATU and its affiliates are confronting the challenges of adulthood and organisational maturity that Michels recognised as the "iron law of oligarchy".Buoyed by the exhilaration of youthful development during the democratic struggle era, many, including the author, 72 drew overly optimistic conclusions about the ability of the adolescent COSATU to avoid the challenges of maturity and oligarchy.The federation of unions that has emerged into adulthood is more diverse, comprising an increasing proportion of public sector and better-educated workers, together with COSATU's traditional blue-collar, semi-skilled industrial working class base.
These unions have successfully embedded many of the founding principles of union democracy and worker control.However, although representative structures and democratic organisational principles remain in place, the vibrancy of intensive participation has transformed the type of union democracy that is practiced.Coupled 69 Darlington "Shop stewards leadership, left-wing activism and collective workplace union organization" (2002) 76 Capital & Class 95. 70 Baccaro "The construction of 'democratic' corporatism in Italy" (2002) 30 Politics and Society at 327. 71 See Bacarro (2002). 72See Hirschsohn (1998).
with the erosion of participation in civil society organisations in the democratic era, the displacement of established traditions of participatory union democracy with representative democracy in turn undermines the democratic control that members hold over their union leaders and federation leaders.
Following the September Commission Report of 1997, COSATU's programme of Organisational Renewal highlighted the importance of ensuring that affiliates maintain internal democracy through regular and well-attended constitutional meetings, support for shop stewards and well-defined and effective report-back systems. 73This strategy aimed to reverse the steady shift from participatory union democracy to a representative democracy as reported here.The inability of affiliates to effectively implement programmes that addressed these goals has been reiterated in COSATU Organisational Renewal reports that note the erosion of progressive union movement traditions and practices such as worker control. 74While internal democracy appears to have survived at the workplace level, the findings reported here are generally consistent with a recent NALEDI survey of COSATU affiliates highlighting that the "maintenance of internal democracy is lacking as there is poor attendance at constitutional meetings, [al]though affiliates still maintain effective mandating and report-back systems for collective bargaining." 75 affiliates mature, many members have become occasional participants in organised labour's body politic rather than the pistons and cogs in the engine-room of the union machinery.As members have disengaged from daily union activities, "ownership" and control of the union has steadily shifted from members to elected representatives, factions and officials.Similarly, citizens express dissatisfaction about the remoteness of their political representatives, resulting in disempowerment and disaffection.
In contrast, the power of participatory democracy lies in the virtuous cycle of empowerment that it fosters, particularly in developing the characteristics required of engaged citizens in a democratic society.Participatory democracy in unions in the 1980s provided a classroom to strengthen commitment to democratic practice in the post-apartheid polity.As it has eroded over the past two decades, it remains unclear whether oligarchies have usurped power or are filling voids that were abandoned by the hyper-active membership in the 1980s.Irrespective, the rapidly declining intensity of participation among shop stewards, as measured by the frequency that they attend meetings, is of greatest concern for the prospects of union democracy.
While attendance at union meetings by shop stewards and ordinary union members has declined markedly, members still hold shop stewards accountable, expect regular consultation and report-backs, and remove non-performing shop stewards.As frequent meeting attendance is replaced by periodic attendance, expectations of accountability necessarily transform and result in the hollowing-out of participatory democracy.As 73 Munakamwe The State of COSATU Affiliates (2009) 74 Naledi COSATU: Phase one report (2006) 75 See Munakamwe (2009) at 6 stewards become representatives rather than delegates, members can only hold them accountable at election time or when they periodically need to remove non-performers.
Whether members can expect that their representatives primarily serve their interests, rather than their personal ambitions, is moot.As shop stewards become increasing attracted into the managerial ranks, the emergence of careerists and entrepreneurs among union officials, as identified by Buhlungu,76 seems to be replicated among elected representatives.Given that shop stewards overwhelmingly support the promotion of shop stewards into the managerial ranks, SACP membership be inspired by opportunism rather than ideology.Consequently, unions are not experiencing the kind of democratic revitalisation that one would expect to result from factional competition for grassroots leadership positions.
For a movement as large as COSATU, judgment about the emergence of an oligarchy requires differentiated criteria appropriate to assessing participatory democracy in local decision-making and representative democracy over policy issues at the union or federation level.Any assessment of union democracy must take cognisance of the crucial distinction between participatory and representative democracy; i.e. between control over decisions and control over representatives. 77alysis needs to focus, firstly, on the extent to which elected representatives on COSATU's and its affiliates' Central Executive Committees are able to exercise effective control over full-time officials and, secondly, whether members broadly concur with the main policy decisions that the federation and its affiliates are pursuing.With support for the Tripartite Alliance remaining consistently high (close to two-thirds) over the past three surveys, it is hard to conclude at present that the policies of COSATU no longer have member support.
There is no question about whether the governance structures of COSATU and its affiliates meet most criteria of constitutional democracy.However, other traditional measures of union democracy should be deployed to establish the extent to which democracy is practiced.This requires further research to address questions including the following: are members free to criticise and debate union officials and to organise, oppose, and replace officials through freely contested elections? 78Does the SACP operate openly as an organised faction or party within COSATU and affiliates?Are there opportunities for competing factions to compete for power, as suggested by Lipset et al? 79 How close are elections for senior posts and how often are incumbent office bearers and officials voted out of office? 80 There is no doubt that the traditional model of worker control that characterised the early days of COSATU and its affiliates is being hollowed out.The structure of constitutional democracy remains in place but the effectiveness of both participatory 76 See Buhlungu (2002). 77See Hochner et al (1980). 78See Stephan-Norris & Zeitlin (1996). 79See Lipset et al (1956). 80See Strauss (1991).
and representative union democracy depends on the mobilisation, participation and commitment of ordinary members to sustain it and to hold its leaders accountable.From a service-delivery perspective, declining participation may well reflect member satisfaction that union leaders are effectively advancing their interests.If members view unions instrumentally rather than as opportunities to practice democracy, then the idealised model of worker control built on the hyper-mobilisation of the 1980s may no longer be perceived as necessary in a labour movement that has reached maturity and helped to secure a democratic dispensation.

Table 1
indicates that the largest group of respondents still expects shop stewards to act as delegates who operate according to a direct mandate.What is significant, however, is that the proportion of respondents prepared to allow stewards to operate with an open mandate has declined markedly in the past two surveys.As shop stewards increasingly pursue their own interests rather than those of the membership, this may suggest lower levels of trust than in the first two surveys.

Table 9 : Frequency of attendance at union meetings How frequently do you attend union meetings?
* Note: Shop stewards were not identified in the first survey

Table 11 : Members and shop stewards: Knowledge of government policy
56SeeLevi et al (2009). 57A Commission of Inquiry was appointed by COSATU in 1996, headed by Connie September, to assess the trade union movement's position in relation to changes in the labour market: see A Bezuidenhout Towards global social movement unionism?Trade union responses to globalization in South Africa Discussion Paper DP/115/2000 (International Institute for Labour Studies, Geneva, 2000).

Table 12 : Members and shop stewards: Voting ANC and the Tripartite Alliance Ordinary Members Shop stewards Total 1998 2004 2008 1998 2004 2008 1994* 1998 2004 2008
* Note: Shop stewards were not identified in the 1994 survey.