The State of the Labour Process Debate after 25 Years: Some Reflections from Industrial Relations and Industrial

Sociology

Notes for remarks to plenary panel at the 25th International Labour Process Conference, Amsterdam, April 2007

Paul Edwards Industrial Relations Research Unit Warwick Business School University of Warwick UK & Advanced Institute of Management Research

Email: Paul.Edwards@wbs.ac.uk

There are two elements in my remarks. First, I want to look backwards but also forwards. Second, I will make some negative comments on the state of the debate but will be, overall, positive as to what has been achieved and what can be done in the future. In form, then, my remarks happily commit the frequently excoriated sin of dualism, and in terms of content I am also happy to engage in dualism and indeed see it as essential.1 The labour process debate is not the same as that approach that has become identified as Labour Process Theory. Gabriel, for example, lists six approaches to the study of a (perhaps the) central issue in the debate, namely, management control and worker resistance, and their relationships with worker identity.2 I make no pretence to assess all six, and should stress that my remarks come from the tradition of workplace industrial relations and industrial sociology. I see it as continuing to offer a key set of perspectives, and indeed as having established themes that other approaches seem to re-discover. It also offers a distinct view of the nature of workplace relations and helps to maintain a more exact focus on what the labour process is, and is not, than has been present recently.3 Looking back When we look back, perhaps the major negative comment stems from the number of contributions that have felt it necessary to re-assess or re-direct the terms of the debate, together with the content of those contributions. In addition to Gabriels re-think, there are the recent contributions by Mumby, May and others, as well as several earlier efforts at review, including the conclusion of Storey as early as 1985 that the debate had run into the sand.4 The number of contributions suggests a lack of clarity and the absence of any agreed paradigm.
1

The form-content distinction is also one that is often criticized, so there is a third dualist sin in the space of four sentences. 2 Y. Gabriel, Beyond Happy Families, Human Relations (52) 1999. His focus is in fact the controlresistance-identity problematic, to which he objects. At pp. 182-3 he argues, possibly tongue-in-cheek, that all the approaches suffer from fatal sins, such as essentialism, dualism and relativism. 3 I should also say that I am focusing on the labour process debate as debate, and not on the concrete experience of labour over the last 25 years. Numerous books from the ILPC have offered important analyses of the latter. Francis Greens Demanding Work (Princeton UP, 2006) provides a major overview, albeit from a tradition that connects less directly than it might with the stuff of the ILPC. 4 D. K. Mumby, Theorizing Resistance on Organization Studies: a Dialectical Approach, Management Communication Quarterly, 19 (2005); T. May, From Banana Time to Just in Time, Sociology, 33 (1999);

As to the content of the reviews, the dominant tendency has been to shift away from a distinctive view of the nature of the labour process. The term has often lost all semblance of definition and become no more than a synonym for work or occupation.5 Braverman and those who debated his work initially were clear that the labour process is that part of the mode of production in which workers productive capacity is deployed in order to produce use values and at the same time surplus value; there was an interest in the nature of the valorization process and in the dynamics of struggle and exploitation.6 This view has evidently been de-emphasized more recently, with the growth of interest in identity and subjectivity and with the rise of poststructuralist Foucault-inspired analyses. I would place myself among those offering an orthodox counter-critique and will say nothing here in overall terms about it.7 Two points need brief statement, however. First, I will focus on specific themes relating to the analysis of the labour process as just described. Whether or not identity and subjectivity are interesting topics is not the concern here, though I of course need to assert along with the counter-critics that any apparent neglect in orthodox approaches is not fatal. Second, this does not mean that orthodox views have all the answers. It is true that they neglected some major issues. Hyman identifies a failure to address gender and

J. Storey, The Means of Management Control, Sociology, 19 (1985); D. Spencer, Braverman and the Contribution of Labour Process Analysis, Work, Employment and Society, 14 (2000). Only the last of these argues for a return to the themes of Braverman and an emphasis on the distinctively capitalist properties of the labour process and on valorization. 5 P. Armstrong, Management, Labour Process and Agency, Work, Employment and Society, 3 (1989), p. 307. 6 H. Braverman, Labor and Monopoly Capital (New York: Monthly Review Press, 1974); Brighton Labour Process Group, The Capitalist Labour Process, Capital and Class, 1 (1977); S. Cohen, A Labour Process to Nowhere? New Left Review, 165 (1987); A. Friedman, Industry and Labour (London: Macmillan, 1977). 7 See S. Ackroyd and P. Thompson, Organizational Misbehaviour (London: Sage, 1999), pp. 150-65 and T. Nichols, Industrial Sociology and the Labour Process, in H. Beynon and P. Glavanis, eds, Patterns of Social Inequality (London: Longman, 1999). Nichols notes sardonically the arrival of management and OB scholars on the terrain of industrial sociology: such colonialists often do develop peculiar views of the countries that they invade (p. 115). For a substantial critique of Foucaultian approaches to power see S. Lukes, Power: A Radical View (Basingstoke: Palgrave, 2nd edn 2005), p. 88-107. My own observations are scattered: P. Edwards, Power and Ideology in the Workplace, Work, Employment and Society, 20 (2006), p. 573; P. Edwards and M. Collinson, Empowerment and Managerial Labor Strategies, Work and Occupations, 29 (2002).

ethnicity as a key limitation of Marxist views of work;8 and other weaknesses could readily be listed. But this does not mean abandoning the core arguments. There has indeed been a tendency towards rhetorical distancing from some approach that is felt to be in error. Braverman is the usual target, but it is not uncommon to find his work bracketed with that of Burawoy and Richard Edwards as comprising labour process theory. This approach has two major limitations. First, it treats as uniform what is in fact varied; Burawoy for example specifically set his approach in contrast to Bravermans. Second, it finds a weakness in an argument and then suggests that the whole approach is faulty. Thus it is certainly true that Burawoy stressed the success of management in the factory that he studied in securing compliance and neutralizing resistance. But this was either a particular empirical limitation, in not seeking out evidence of deeper resistance, or an analytical mistake, in assuming that what was true of this one factory in Chicago is true generally. But it has no bearing on the wider approach that Burawoy was adopting. In particular, dualism is commonly seen as a central problem in Labour Process Theory, that is a mode of analysis in which control and resistance are treated as stark opposites. Yet Burawoy argued that, rather than there being a straight opposition between capitalist control and worker resistance, workers in fact produce consent as an integral part of the production process.9 Attempts at repositioning also seem to be rather weak. Consider Mumby. We need to
R. Hyman, Marxist Thought and the Analysis of Work, in M. Korczynski et al., eds, Social Theory at Work (Oxford: OUP, 2006). 9 M. Burawoy, Manufacturing Consent (Chicago: Chicago UP, 1979). The criticism is more relevant to Richard Edwards, especially Contested Terrain (London: Heinemann, 1979). This work developed ideal types of managerial control strategies and argued that each produced worker resistance and was successively replaced. But it could be criticized without abandoning LPT as a whole: P. Nolan and P. Edwards, Homogenize, Divide and Rule, Cambridge Journal of Economics, 8 (1984). A useful further case is the work of Ezzamel et al. (M. Ezzamel, H. Willmott and F. Worthington, Accounting and Management-labour relations, Accounting, Organizations and Society, 29 (2004)). This offers a substantive analysis that is consistent with labour process analysis and indeed uses the Burawoy phrase the politics of production: M. Burawoy, The Politics of Production (London: Verso, 1985). But its rhetoric repeatedly insists that LPT imputes interests to labour around resisting exploitation, assumes that people respond to work situations directly as a result of the economic categories they occupy, and that subjectivities are fixed by class location (pp. 271, 273, 274). No sensible analysis would do this. If by subjectivity is meant a concern to analyse why workers in particular situations think and act as they do in relation to the politics of production, then there is no problem, and indeed the substance of the analysis is consistent with that of studies cited below. If it means in addition the abandonment of a distinct focus on the politics of production and the adoption of a perspective based on the disciplinary society, then the concreteness of the analysis is diluted and ironically the alleged certainties of LPT are replaced by those of post-structuralism.
8

address the mechanisms by which the dialectic of control and resistance produces either at the level of the day-to-day or in the longer term critical reflection and transformation, and reproduction of the status quo. [And we should see how] social actors attempt to fix meanings in ways that resist and/or reproduce extant relations of power. Or May: Strategies should be viewed in terms of their abilities to mobilise dispositional powers that, in turn, create the conditions under and through which organisational interactions take place. Tactics of resistance to such strategies then draw upon episodic power within the spaces in which dispositional power is limited. The success of these local protests in being generally transformative will then depend upon their effects in exposing the antagonisms between strategies, as well as their unintended consequences.10 It is hard to disagree, but also hard to see how analytical advance has been made on the better extant studies. There are two points here. The first turns on the understanding of specific concepts such as strategy. May cites a number of works which have surely established, empirically as well as conceptually, the kind of argument just quoted. And there are many more that he does not mention, notably those of Batstone and colleagues who sub-title their major work the organization of conflict and accommodation precisely to highlight the dialectics of control and resistance and the ways in which strategies reflect but can also transform underlying structural conditions and Nichols and Beynon.11 There are also some well-defined treatments of strategy that not only say what May says but that also go beyond these core facts to establish how strategies vary according to economic and other conditions.12 I will not dwell further on the negatives of the past, though the appendix gives some further evidence for the claims just made.

Mumby, op. cit., pp. 23 and 24; May, op. cit., pp. 779-80. E. Batstone, I. Boraston and S. Frenkel, Shop Stewards in Action (Oxford: Blackwell, 1977) and The Social Organization of Strikes (Oxford: Blackwell, 1978); T. Nichols and H. Beynon, Living with Capitalism (London: Routledge,1977). 12 Notably R. Hyman, Strategy or Structure? Work, Employment and Society, 1 (1987). Hymans work seems to have received remarkably little attention within the labour process debate,, and certainly within its subjectivist and post-structuralist aspects. This is despite the fact that it addressed with great precision and
11

10

The second point relates to the conceptualization of the labour process. I return to this below, but in shifting towards the positives of the past I would argue that the many critiques notwithstanding there is in fact a reasonably clear set of core principles of labour process analysis. These have been adumbrated by Paul Thompson and they do not need repeating here.13 There are two further important positives. The first is empirical. We know a great deal about the evolution of the labour process in the past 25 years, as for example in the mass of evidence on the nature of work in Japanese firms both at home and overseas and on the effects of initiatives such as TQM. Think also of the alternatives, namely, the bland and misleading pictures from The Machine that Changed the World and the TQM gurus, that would have been all that we had to go on. Research has also helped to puncture myths of the new economy, the knowledge worker, and the like. This statement is more controversial than the previous one, in that some arguments on these topics have come from respected social scientists. But it does none the less seem to be the case that they have been in error, in part at least because of their search for the sweeping and the transformative, as opposed to seeing what was happening in concrete circumstances. Several labour process studies have indeed stressed explicit continuities with the past.14 And it is not just a question of piling up empirical detail that questions, as it is bound to question, large-scale predictions. In relation to TQM and employee participation schemes, for example, we know a great deal about the conditions under which they have certain effects. This research seems to me to escape the sin of essentialism in two respects. First, it is not argued that TQM is a distinct set of practices; rather it is analysed as a political project that combines a set of factors. Studies trying to compare TQM and

sophistication issues of the interplay of conflict and accommodation. Some of the key papers are reproduced in The Political Economy of Industrial Relations (Basingstoke: Macmillan, 1989). One obvious reason for this neglect is that Hymans work was explicitly Marxist in provenance and that it could thus be pigeon-holed rather than interrogated constructively. 13 P. Thompson, Crawling from the Wreckage, in D. Knights and H. Willmott, eds, Labour Process Theory (Basingstoke: Macmillan, 1990); P. Thompson and S. Ackroyd, All Quiet on the Workplace Front? Sociology, 29 (1995). 14 A. Scott, Willing Slaves? (Cambridge: CUP, 1994); M. Webb and G. Palmer, Evading Surveillance and Making Time, British Journal of Industrial Relations, 36 (1998). And more generally P. Thompson and C. Warhurst (eds), Workplaces of the Future (Basingstoke: Palgrave, 1998). This last is one of the 21 volumes generated from the series of Labour Process Conferences; several earlier volumes sustain the arguments in this paragraph about Japanese management and TQM.

non-TQM workplaces found that there was often not this sharp contrast and that instead quality principles are expressed in different ways. Second, in relation to the effects, TQM is seen as having a contingent impact, shaped by specified contextual factors.15 Similar points could readily be made in relation to one of the major contemporary management practices, the High Performance Work System. Labour-process-inspired analysis has revealed the limitations of the HPWS model by showing something of how the black box of production works. And this is not just a matter of revealing the costs to workers of the relevant practices, important though that is. It has also been shown that the systems have inconsistent elements, that they may work only because of other conditions and not because of their inherent properties, and that their meaning is established and amended through social processes.16 Given the massive influence that the HPWS model has had, this critical role has been of major importance. In one key sense the implication is: keep up the good work. Many of the analytical lessons about HPWSs are in fact very similar to those that would apply to TQM or earlier initiatives. But managerialist research and practice continues to claim that things are different this time and the only way to counter such claims is to show that things may not in fact be that different. This optimism is, however, tempered by three points, all of which stem in various ways from reliance on the single case study focused on the immediate point of production. First is a failure to offer sufficient detail on a study to be able to place it in a wider context. Even some book-length accounts say little about key facts such as pay levels and the labour market experience of the workers studied. My own feeling is that this limitation is most marked among post-structuralist accounts that are interested in the creation of identity and that also deliberately eschew causal analysis. It is apparent for example in Barkers well-known study, which finds a distinct form of concertive control. But it

P. Edwards et al., The Determinants of Employee Responses to Total Quality Management, Organization Studies 19 (1998); C. Rees, Worker Responses to Quality Organisation, Work, Employment and Society 15 (2001). A feature of the latter study is its linking of material influences to subjective responses and its insistence that this dualism is a useful heuristic device that facilitates a clear picture of the nature of employee responses (p. 759). 16 H. Ramsay et al., Employees and High Performance Work Systems, British Journal of Industrial Relations, 38 (2000); A. Hesketh and S. Fleetwood, Beyond Measuring the Human Resources Management Organizational Performance Link, Organization, 13 (2006).

15

does not try to say how common this is or what conditions produce it. It also seems to me that there is a tendency for the sin of essentialism to re-emerge, in that new forms of teamwork are described as though this particular pattern must result and as though teamwork indeed has essential immanent characteristics.17 This argument receives some support when we look at Sewells analysis of patterns of teamwork, which in fact embraces only three cases and which admits that these may be extreme examples of extensive electronic surveillance. Yet if one were trying to analyse the meanings of teamwork, there is a larger set of cases available, from which it is possible to offer some causal analysis. There are clearly different kinds of teams, and where teamwork is combined with other conditions it is more likely to generate outcomes of value to workers as well as to management than is true when these conditions are absent.18 As Thompson and Ackroyd comment on another example of Sewells work, there is a tendency to strip out context and to produce an abstract control imperative.19 There is a need to address context much more clearly and to try to develop an explanatory and causal analysis, for why else do we study things? The second point is that sets of individual studies have not formed a coherent research programme, so that it hard to reach general conclusions. The ability to draw such conclusions around particular themes, like TQM, has emerged ex post and fortuitously, and there are many gaps in the record. Third, analysis at the point of production has not made sufficient connections to wider aspects of capitalism as a dynamic mode of production. I comment on this further below.

J. Barker, The Discipline of Teamwork (Thousand Oaks: Sage, 1999). G. Sewell, The Discipline of Teams, Administrative Science Quarterly, 43 (1998). Cf. J Blanger et al., Commitment at Work and Independence from Management, Work and Occupations, 30 (2002). For a formal model developing this point, see P. Edwards et al., The Bases of Compromise in the Workplace, British Journal of Industrial Relations, 44 (2006). 19 G. Sewell, Nice Work? Rethinking Managerial Control in an Era of the Knowledge Economy, Organization 12 (2005); P. Thompson and S. Ackroyd, A Little Knowledge is still a Dangerous Thing, Organization 12 (2005), p. 707.
18

17

Looking forward We need clearer analytical categories. Mumby dismisses one attempt that of Hodson to identify types of worker behaviour as tending the sin of reification.20 He seems to prefer to stress the deeply embedded connections between control and resistance. Yet he does not say what a research programme would then look like, and probably for good reason: celebrating complexity does not in fact generate any particular research questions. By contrast, as Gabriel has noticed, a focus on organizational control as a defining feature of post-modernity has tended to obscure the possibility that different organizations employ different strategies of control.21 This basic fact seems self-evident, and it was addressed theoretically in the early LPT works by Richard Edwards and Andy Friedman. Yet it does no harm to re-assert it in order to explore and explain empirical patterns. Four themes may be distinguished. Focus I noted above arguments that the labour process should not be a synonym for work and that accumulation and valorization should be (re-)placed at centre stage.22 This is a key point in any study that uses the term labour process other than rhetorically. The interest is in how labour power is deployed, how a surplus is generated, and what the consequences are. This does not mean that the focus is solely on direct producers, but it does mean that distinct tools are needed to understand groups such as managers. As Armstrong argued,

R. Hodson, Good Soldiers, Smooth Operators and Saboteurs, Work and Occupations, 18 (1991). Gabriel, op. cit., p. 186. 22 Nichols, op. cit.; Armstrong, op. cit., p. 309. Note that Armstrong argues that Braverman wrongly treated management as a labour process, but does not then dismiss Bravermans project, suggesting instead that this particular view of management was in fact in conflict with the project as a whole. It is also convenient to note here that, even in formal statements within Marxist theory, use and exchange values are both stressed. One important concomitant is that worker resistance is not reduced to a narrow economistic struggle over the price of labour power. Workers produce use value as well as surplus value, and they take pride in their work. Workplaces struggles are thus about the meaning and value of work as well as wages, though as many Marxists stress the functioning of capitalist markets makes it hard to sustain arguments about the former, and workers learn the rules of the game. See E. Hobsbawm, Labouring Men (London: Weidenfeld and Nicolson, 1964), ch. 17. Note also Armstrongs use of the agency framework. The principal-agent model is well-established in economics; it tends to treat the relationship in rationalistic terms. Implicit in Armstrongs use of the model are at least two things: a view of the principal-agent relationship as political and not just economic; and the location of politics within a view of the contradictions of the management process, as opposed to a view of politics as merely incidental.
21

20

management is not a labour process: the labour process is the means of securing the material base of human existence, capitalism is only one way of organizing the labour process, and the fundamental of management is that it is an agency relationship. This fact does not mean that management is simply a technical activity. On the contrary, Armstrong underlines contradictions in the agency relationship between controlling the agent through performance management and allowing trust and discretion. These contradictions might seem to be identical to those of control and autonomy that pertain to real labour processes, but for Armstrong they arise within an agency relationship and there are cases where management has not been de-skilled that the agency view can explain and alternatives cannot. A stronger argument would certainly be needed to sustain the point.23 For present purposes, the key is that a labour process perspective can say some distinctive things about management, and that we need to understand the labour process, not as work in general, but as a form of human activity that takes a particular character under capitalism. Here, managers and workers meet in a relationship of structured antagonism, and they define themselves in this relationship: there can be no manager without a worker, and the basics of social class lie in the dynamics and contradictions of the relationship.24 Now, the early labour process debate shared with Marxian debates at the time a concern to distinguish, usually in very abstract ways, productive and unproductive labour, a
23

Armstrong asks why, if the end result of the organization of managerial and non-managerial labour is the same, there is any point making the distinction. His answer is that it is not always the same: managerial work is not necessarily de-skilled even when this is technically feasible (p. 320). This seems a rather weak answer empirically, since not all non-managerial labour is de-skilled. It is also weak conceptually, in that trends in the organization of managerial work need to be treated as the result of contradictions in an agency relationship, whereas non-managerial work follows a different dynamic. Evidence that management retains its core supervisory nature can be found in the work of Colin Hales: Rooted in Supervision, Branching into Management, Journal of Management Studies, 42 (2005). But this does not dispose of the so what? issue. Part of an answer is in observations such as those of Tony Watson (In Search of Management, London: Thomson, 2001): middle managers recognize their distance from shopfloor employees even though they are also aware that they are removed from higher corporate decision-making. The wider theoretical point is that managers are agents of capital, and that they gain from this to the extent that they can move closer to the principals but lose if their role as agents can be performed in some other way. Paul Thompson (Introduction:Unmanageable Capitalism? in S. Ackroyd et al., eds, The Oxford Handbook of Work and Organization (Oxford: OUP, 2005) has re-stated the agency view of managers and used it to address some of the contradictions in the management process. 24 P. Edwards, Late Twentieth Century Workplace Relations in R. Crompton et al. (eds), Renewing Class Analysis (Oxford: Blackwell, 2000).

concern whose disappearance can only be welcomed. But there was the important analytical question of what a labour process debate is about. It is not about everything that goes on in employing organizations. Worker subjectivity is relevant only in so far as it affects the organization of conflict and co-operation in the work process. Managerial work is certainly not ignored in this perspective. But it becomes problematized in terms of its connection to the generation of use-values and surplus value. Is it the case for example that changing the role of the supervisor to that of coach and facilitator alters relationships with workers? Such a shift can be seen as one that stresses use-values (achieving a productive task) over surplus value (controlling workers) though the latter will of course have to be secured by other means. This does not mean that (1) the analyst directly wishes to categorize all behaviour as to do with use-value or surplus value or (2) some view is taken as to what a use-value really is. Let is suppose that the supervisor just imagined works is a weapons factory, and that becoming a coach means that workers are trained better and that weapons are produced more efficiently.25 The argument does not necessarily mean that weapons are inherently useful. It simply means that, under current social and political arrangements, there is demand for these products, that they have a use under these arrangements and that there are more and less efficient ways of providing the products. That deals with point (2). As for point (1), the idea of the two types of value is an analytical one, which allows us to ask about the ways in which work is configured and the changing composition of jobs. For example, what is the balance between the activities or control and co-ordination among a group of managers, and how is the agency relationship managed (e.g. in terms of performance targets or a reliance on other mechanisms such as trust)? A focus on the labour process and valorization may help to avoid an undue interest in the specific concept of resistance and whether or not it is (still) important. Such a question is, arguably, unanswerable: it is not as though in some mystical past there was a mass of resistant workers (and indeed a great of industrial sociology said why there was not); there is no way of comparing evidence in a reliable way; and whether or not any given
25

We need to treat efficiency here is its technical sense of producing more output for a given set of inputs. The production process is co-ordinated in a way that is technologically superior to its previous form. Workers are not working any harder.

10

act is a case of resistance requires careful scrutiny. On the last point, going absent, for example, may be a conscious act of hostility to management; it may be a pragmatic taking of benefits on offer; it may reflect excess conformity if physical and emotional exhaustion make the worker unfit to work; and so on. And absence will have different implications for fellow workers (shared expression of resistance, a cause of lateral conflict because those at work have extra work to do . . . . ) and for managers (a cost if work cannot be re-organized easily, or something relatively costless if other workers cover the relevant duties).26 The solution is to look at the organization of the labour process and the way in which a frontier of control is created and sustained. Thus workers may not resist but may obtain benefits, for example through legal rights. Thus Burawoy notes that workplace regimes are located in national contexts and that states take different views of the regulation of the labour process. To take a simple example, in the UK since 1971 there have been legal protections against unfair dismissal and indeed the number of issues on which employees can bring claims against employers has risen from one to 90. The operation of the rights is plainly far from automatic, but the terms of the labour process have changed. We thus need to look at workplace regimes and how they produce packages of costs and benefits for workers, and not seek out resistance for its own sake.27 Levels of analysis The second key issue is levels of analysis. The Foucaultian insistence that power and knowledge are parts of the same thing means, to the extent that it means anything, that at

P. Edwards and H. Scullion, The Social Organization of Industrial Conflict (Oxford: Blackwell, 1982), goes into these issues in detail, arguing that the salience of an action depends on its location in the workplace regime. Where managerial control is extensive, absence may pose few costs, while actions such as quitting can have benefits in removing those who might challenge the regime more directly. 27 My current research focuses on cases, small firms, where resistance is largely absent. But there are certainly interesting questions such as why apparently all-powerful managements are constrained to acknowledge workers concerns and why low-wage jobs continue to exist in advanced economies. The answer to the former question turns on such factors as: the fact that workers power is not zero; mutual obligations sometimes based in family or other ties; and the need for any employer to secure a minimal level of consent if the labour process is to continue to function. See M. Ram, Managing to Survive (Oxford: Blackwell, 1994). The latter question links to themes raised below about relating the labour process to wider structural issues; in this case, the jobs exist because of a supply of workers through legal and illegal migration, intense product market competition, and the fact that employers, lacking other options, are willing to occupy these highly competitive and insecure market niches. See e.g. P. Edwards and M. Ram, Surviving on the Margins of the Economy, Journal of Management Studies, 43 (2006).

26

11

the immediate point of production there is a potential constant negotiation over claims to control resources. If workers argue about effort levels, they are asserting their power and their knowledge that they are aware that there is indeed a bargain and that they can change something. Such ideas do not of course need Foucault to be stated. Thus Edwards and Scullion went out of their way to criticize the toolbox theory of sanctions: the idea that workers have bargaining resources simply available for use like tools in a box. Deploying sanctions requires a language to express a sense of opposition from management and to take a view that what exists is not given for all time.28 At this level of the point of production one may wish to contrast different types of worker ideology as Hodson does. The point of this is not to reify behaviour, but to deploy some ways of capturing variation, rather than simply finding out the essence of worker behaviour anywhere. The categories will depend on the topic at hand. Thus Mars made very effective use of a different set of four categories, Goss in a very different context has also offered four models of types of employment relations in small firms, Burawoy offers another set of four types of workplace regime, and so on.29 Now, any of these categories can be interrogated for their logic and completeness, and in various places I have suggested that none of these is wholly adequate. But they certainly take us quite a long way. The reader of Mars will have a much clearer view of different types of fiddles and the conditions generating them than would have been possible without these analytical ideal types. Similar moves can be identified in more recent analyses of contemporary work, notably that of service employees. Understanding of this work began with accounts of the distinct nature of the service encounter, and has more recently identified different types of service work. Frenkel for example speaks of mass service and mass customized types of work. Related to this, there is now a huge amount of work on call centres, which is increasingly differentiating types of such work according to the market segment in which the firm is engaged and hence different levels of pay and different degrees of worker autonomy.30
Edwards and Scullion, op. cit.., p. 163. G. Mars, Cheats at Work (London: Counterpoint, 1982); D. Goss, In Search of Small Firm Industrial Relations, in R. Burrows et al. (eds), Deciphering the Enterprise Culture (London: Routledge, 1991); M. Burawoy, op. cit., ch. 3. 30 S. Frenkel, xx, in S. Ackroyd et al., eds, (Oxford: OUP, 2005);
29 28

12

Once we have categories of types of work situation, we can begin to ask about causal influences on them. Mars provided an extended and in many ways exemplary explanation of the four patterns of fiddle that he identified, which turned on the structure of the work tasks. We would also now want to give more attention to managerial strategies and the operation of the capitalist economy. Thus one of Marss types was the wolf fiddle, practised by gangs of workers with a strong sense of collective identity; dockers were his archetypal example. They were able to sustain their position because of a set of factors: their work was hard to monitor; they worked in gangs; and it was hard for management to find substitute labour. Containerization of dock work has challenged all these conditions and has led to major redefinitions of the job of a docker. This development in turn reflects competition between ports, the role of the state in allowing work that in many countries was legally defined as dockers work to be carried out by other workers, often well away from ports themselves, and so on.31 In short, changing structural conditions shape events within the politics and production, and at this level people make choices as to how to respond to these conditions, and out of the resulting actions, bargains and compromises a new pattern of workplace politics emerges. A related issue is the placing of a work regime in its organizational and economic context. As Thompson argues, we need to locate a workplace in corporate strategies and the development of the economy as a whole.32 A call for multiple levels of analysis is increasingly common, and rightly so given the pressures of globalization and the way in which market rationality impinges of the workplace. Some of the early studies, such as those of Friedman, were explicitly concerned with the historical evolution of forms of labour and the connections between workplace practices and competitive conditions. My only comment here is one of practical research design: if we are interested in the dynamics of workplace relations, we need a relatively micro focus, albeit one that is sensitive to material conditions and it is reasonable to ask, as argued below, that researchers give explicit attention to these conditions. But a complete placing of a workplace regime in the circuit of capital would be too much to ask.
31

See P. Turnbull and S. Weston, Employment Regulation, State Intervention and the Economic Performance of European Ports, Cambridge Journal of Economics, 16 (1992), and Co-operation or Control? British Journal of Industrial Relations, 31 (1993). 32 P. Thompson, Disconnected Capitalism, Work, Employment and Society, 17 (2003).

13

A reasonable approach is to pursue causal influences as far as is needed for the task at hand. Thus Delbridge argued that the pattern of workplace relations that he observed at Valleyco reflected the fact that this factory was a supplier to others and was thus under distinct kinds of customer pressure.33 Studies of small firms have also addressed how exogenous influences such as ethnicity and family shape the labour process.34 They show that co-ethnic ties influence the effort bargain by establishing relations of mutual dependence and moderating managerial control. Now, such research might be read as saying that ethnic and family identities are key, and that studies privileging the labour process are in error. But such a view makes sense only if one operates without analytical distinctions and wishes to treat social life as undifferentiated. If the research question is peoples sense of identity and meaning, then the intersections of class, race, and gender are central.35 But if we are concerned with the production of surplus at the point of production, then gender, ethnicity and so on can be treated as factors that shape particular labour processes but which are analytically separate from the labour process itself. Future research might take this theme further. From the Delbridge example, one might wish to explore supply chain effects more directly, and also to compare workplaces in different locations in a supply chain. From the ethnicity example, the obvious point is to try to compare otherwise similar workplaces with differing ethnic mixes. But the key point is not to dissolve the analysis and to keep a velar view as to what is to be explained. There is a body of work that addresses the changing structure of capitalism, some of it from scholars with an approach with strong affinities to labour process analysis.36 Establishing links between this level of analysis and workplace regimes is an important task. By this, I mean that analysis of corporate strategies and re-structuring cannot directly interrogate the effects of these activities at workplace level, for example whether

R. Delbridge, Life on the Line in Contemporary Manufacturing (Oxford: OUP, 1998). See note 27 above, and M. Ram et al., Making the Link: Households and Small Business Activity in a Multi-ethnic Context, Community, Work and Family, 4 (2001). 35 E.g. M. Lamont, The Dignity of Working Men (New York: Russell Sage, 2000). 36 E.g. Lazonick, whose early work examined labour processes historically and who more recently had addressed the financial structure of capitalism, and Karel Williams and colleagues: W. Lazonick, Industrial Relations and Technical Change, Cambridge Journal of Economics, 3 (1979); Business Organization and the Myth of the Market Economy (Cambridge: CUP, 1991); K. Williams et al., Why Are the British Bad at Manufacturing? (London: RKP, 1983), Cars (Providence, RI: Berghahn, 1994), J. Froud et al., Caterpillar: Two Stories and an Argument, Accounting, Organizations and Society, 23 (1998).
34

33

14

new accounting regimes actually secure the tighter control of production costs that they seek; workplace analysis remains an important element here.37 Effects of worker behaviour It is important to continue to ask explicitly about the implications of worker activity.38 This would be my one qualification to Ackroyd and Thompsons insistence on the continued presence of worker misbehaviour. As well as showing that it exists, and that managerial claims to secure control and commitment are empirically hollow and conceptually impossible, we need to continue to ask what it does. Does it represent Burawoyian consent to ones own exploitation and even if it does, what might the practical implications be?39 Burawoys workers (in mid-1970s US manufacturing) were doing reasonably well in having secure and relatively well-paying jobs. At the time, their interests may well have lain in the sense of the balance of costs and benefits of alternative lines of action in continuing as they were. We now know with the benefit of hindsight that many threats were on the horizon, and it is a perfectly proper question to ask whether these might have been anticipated and what if anything might have been done about it. It is also analytically important to ask how patterns of behaviour reproduce existing relationships. Thus high quit rates in low-wage and insecure jobs tend to undermine any worker collectivity and, along with other factors, to reproduce this form of labour process. In other circumstances, worker action has clear effects. Among the most common noted in labour process studies is some kind of tacit disobedience that means that a managerial initiative fails to achieve its ends. Another is the exploitation of space to bend rules. Thus call centre studies reveal workers who spend longer than they are supposed to on calls. This gives them some personal satisfaction and is not necessarily consciously resistance against management. It may also have the result in an echo of Roys finding that

37 38

As in the work of Ezzamel, op. cit. I recall Theo Nichols arguing I think at the first Labour Process Conference in 1983 that there was undue interest in what went on at the point of production, and not enough attention to what this meant for outcomes such as productivity. His point remains important. It is developed in Nichols, op. cit. and also in his The British Worker Question (London: RKP, 1986). 39 Burawoy, Manufacturing Consent.

15

workers break formal rules to achieve substantive objectives of improving customer service.40 It is within such ambiguities that resistance takes its meaning. It is also important to note longer-term effects. Ackroyd and Thompson focus on specific and small-scale acts if misbehaviour, as do Noon and Blyton in their discussion of five survival strategies.41 These acts may in themselves come to shape how a workplace regime is defined. This was the burden of many of the classic studies, which demonstrated that there was an established space for workers that managers attempted to change at their peril. Such clearly established custom and practice may be less common than it was, but it illustrates the wider point that socially constructed expectations shape the extent to which managerial objectives can be met. Overt resistance may be limited, but, as the limited effectiveness of TQM and HPWS models shows, management intentions are often not realized. The reasons turn on the contradictions inherent in the organization of work: establishing control while eliciting consent, meeting customer needs while also hitting financial targets, and so on. Beyond specific acts that come to define workplace regimes there are more deliberate efforts to alter the terms of the labour process. These embrace bargaining over pay and conditions, strikes, and at the extreme quasi-revolutionary protests. Such acts, or the threat of them, can come to define a certain terrain of workplace relations. Terrains cannot necessarily be ordered in terms of whether they are better or worse from the workers (or the managers) point of view. Gallie, in his classic comparison of French and British oil refineries, showed that in some respects for example shift work and manning levels the French workers were the more militant, but also that French management retained much greater freedom of action than did its British counterparts.42 In some circumstances, an ordering may be possible if the frontier of control is unambiguously more in favour of workers in one place than in another. But it may also be that workers who have won workplace battles thereby expose themselves to

D. Roy, Quota Restriction and Goldbricking in a Machine Shop, American Journal of Sociology, 57 (1952); Work Satisfaction and Social Reward in quota Achievement, American Sociological Review, 18 (1953); Efficiency and the Fix, American Journal of Sociology, 60 (1954). 41 M. Noon and P. Blyton, The Realities of Work (Basingstoke: Palgrave, 2nd edn 2002). 42 D. Gallie, In Search of the New Working Class (Cambridge: CUP, 1978), esp. pp. 300-317.

40

16

counterattack. And so on. The point is that we need to address such issues directly if the significance of misbehaviour, resistance, and so forth is to be grasped. One of the benefits of research around the HPWS is interest in a range of outcomes of workplace relations, embracing wages and job satisfaction but also the relationship between the workplace and family and other spheres. A labour process perspective might build on this approach by looking systematically at outcomes and the quality of jobs. Comparative analysis The key future development, which draws together the preceding three themes, is comparative analysis. The most effective studies have always been comparative in at least one of several senses: direct comparison within the same study (Lupton, Gouldner, Mars, Gallie); comparison of one study with specifically comparable data from other studies (Burawoy); a use of a new study in the light of previous expectations in the literature (e.g. Dittons study of the control of time by workers paid flat rate, taking as its foil evidence of the practices of workers paid by the piece).43 Through the imagination and effort of Hodson and his collaborators it has also been possible to draw together workplace ethnographies into a comparative data set, now numbering 204 cases.44 But such a collection depends on what was in the available ethnographies, and it necessarily looks backwards to what has been done.45 As argued above, it has also been possible, in more discursive fashion, to extract from studies of teams or TQM some factors that seem to explain patterns of variation. But more planned and systematic comparisons would have obvious benefits of generating directly comparable data. Comparative analysis in this field is bound to be hard because the phenomena of interest do not vary independently of each other or in ways that are easily observable in advance.
In addition to studies cited above: T. Lupton, On the Shop Floor (Oxford: Pergamon, 1963); A. Goulder, Patterns of Industrial Bureaucracy (New York: Free Press, 1955); J. Ditton, Baking Time, Sociological Review, 27 (1979). 44 R. Hodson, Dignity at Work (Cambridge: CUP, 2001). Recent studies include V. Roscigno and R. Hodson, The Organizational and Social Foundations of Worker Resistance, American Sociological Review, 69 (2004). This last study makes imaginative use of Qualitative Comparative Analysis, a method that might well be used to develop systematic comparative workplace studies. 45 It should also be noted that the richness of the data is often limited. Thus Hodson is constrained to record wages, job security, and other key measures into two or three categories. And there must be some question as to the comparability of categories over time and space. For other conceptual and empirical issues, see P. Edwards and J. Blanger, Generalizing from Workplace Ethnographies: From Induction to Theory, unpublished paper, 2006.
43

17

As noted above, studies of TQM have in fact found it hard to find firms that clearly had or did not have the relevant practice; quality principles need not go under a particular label. There is no simple solution to this issue, but sensitivity to previous studies and to likely sources of variation can give some pointers. Thus studies of call centres have shown variation according to market segment, and future studies might be more systematic in comparing within a segment. Similarly, this sector is one that has seen rapid globalization, and as research is already showing it is productive to compare new countries with old ones in terms of how the same work is organized and experienced.46 Studies in this vein have made useful advances in deploying quantitative surveys within case studies of particular settings. This has been of great value in demonstrating the representativeness of the results and also in showing that claimed differences between cases are real. None the less, there are limits to the value of self-report data on such things as work intensification, and independent direct observation can also be valuable. It is for example common to read that performance management systems put new pressures on workers and run counter to claims about empowerment. This is a perfectly sensible argument. But there are benefits in substantiating it in more detail. In any particular case, the exact ways in which the effect works need demonstrating, and the extent to which the effect is real in the sense of being substantial and affecting a significant number of workers needs evidence. Illustrative quotations do not show just how important the effect really is. And across cases it is important to know whether all PMSs have this effect, or only some, and if the latter why. The classic studies were able to address these issues through the presence of the researcher in the workplace for a substantial amount of time, so that it was possible to find out what really happened as opposed to what respondents might report. Some were based on structured observation, as in the Batstone et al. studies and, with a different focus, the work of Blackburn and Mann.47 Observation is time-consuming and expensive, and there are also issues of access for such studies as firms become leaner and outcome46

P. Taylor and P. Bain, India Calling to the Far Away Towns: the Call Centre Labour Process and Globalization, Work, Employment and Society, 19 (2005). 47 R. Blackburn and M. Mann, The Working Class in the Labour Market (London: Macmillan, 1979).

18

oriented. But the pay-offs are considerable and the demands on those being researched are less, in terms of time if not exposure, than is the case with survey and interview methods. Conclusions: role of labour process analysis A distinctively labour process analysis of work needs to retain its core strengths. These include an empirical interest in the experience of work at the point of production and a theoretical concern with the contradictory relationships between capital and labour. And, in terms of method, detailed case studies and ethnographies have been, and should be, the preferred approach. Large-scale comparative surveys do not fit well with these preferences though it certainly makes sense for them to be informed by labour process issues in terms of the questions that they ask. Within the labour process tradition, I would argue, first, for more explicit comparative analysis, which can be within a country as well as being internationally comparative. Second, it would be a great help if analysts took it upon themselves to collect and report key data on such things as pay levels, quit rates, local unemployment rates, and the labour market experience of employees. Third, systematic observation can reveal important information about the nature of the work task and other key issues such as the ways in which managerial controls are deployed (e.g. extent of direct supervision, frequency of reporting of data). It would also be desirable, if difficult, to observe processes such as performance appraisals. There are now useful studies that give workers self-reports of how far performance targets are agreed or imposed, and what the effects of the targets on subsequent behaviour are. But it is also desirable to have direct evidence on what goes on, how far workers talk among themselves about appraisals and develop some collective norms about them, and if possible how far appraisal really shapes the effort bargain.48 Fourth, the time dimension is central. A question about the large number of one-off case studies that now exist naturally concerns how stable the relationships reported are.

Consider here for example S. R. Barley and G. Kunda, Gurus, Hired Guns and Warm Bodies (Princeton: Princeton UP, 2004). This is a work by two leading ethnographers, studying itinerant experts such as computer programmers who move between companies on a freelance or contract basis. It embraced periods of participant observation in staffing agencies, but most of the information on work itself seems to have been gathered through interviews. The labour process itself was naturally hard to observe, but observational detail would, in principle, be highly desirable.

48

19

Understanding a workplace over time can surely reveal a great deal about processes of change (and continuity).49 As for multi-level analysis, a basic need is that the context of a workplace be described in terms of such features as the type of product, the intensity of competition, and the place of the workplace in the (exchange) value chain. We can then obtain some view of the causal forces at work. Just how much further one should go in developing a multi-level analysis must then surely depend on the task at hand. The danger of being too ambitious is that analysis becomes superficial. Influences from outside the workplace itself need to be traced to the extent that they are likely to shape events. As to what might be studied, there is a huge range of choice but also challenges. Workplace sociology developed from studies of large groups of workers assembled in distinct work sites. As the character of work changes, the idea of a fixed work place also loses resonance. A solution may lie less in multi-level approaches and more in the use of multiple methods embracing interviews and systematic observation. To mention just two candidates: understanding the re-structuring of work at the bottom of the labour market, in the light for example of migration to developed economies, is an important issue; and the nature of managerial work continues to deserve attention. On the latter, we know something of middle managers but it would also be very instructive to learn more about higher levels of management, for example how identities and loyalties are created and the way in which the agency relationship with capital is understood and played out. Sklair for example identifies a transnational capitalist class of elites in multinational companies and state agencies, though his work seems largely a hypothesis about the presence of such a class rather than detailed substantiation of its existence as an economic category, still less as a class with shared awareness and interests.50 It would be extremely valuable, if challenging, to analyse the work of putative members of this class, their agency relationships, their roles in planning corporate re-organizations, and so on. There is some well-established research on the structures of corporate power and managerial elites that

49 50

E.g. B. Ahlstrand, The Quest for Productivity (Cambridge: CUP, 1990). L. Sklair, The Transnational Capitalist Class (Oxford: Blackwell, 2001).

20

could offer some foundations here.51 A less challenging study might take some occupational groups and examine their positions within management and how they define the nature of the managerial task pursuing the agenda on these issues established by Armstrong.52 Such a study would take the Foucaultian interest in knowledge and render it in a reasonably precise way by asking about knowledge claims and how they are sustained, together with the relationship between these claims and the material demands of a capitalist enterprise: some claims to knowledge are more successful than others and it is important to understand why. Plainly, comparative analysis would be of central value here. In short the labour process debate, qua debate, has entailed a fair number of false starts, re-inventions of wheels, and attempts to establish stark differences between perspectives where there is also common accord. It has also generated important empirical evidence and sustained critical engagement with major changes in the experience of work over 25 years. Analysis now needs to build on these strengths by examining workplace regimes and their costs and benefits, and placing such examination in the context of capitalism as a mode of production.

Appendix: Remarks on the Post-structuralist Turn Ackroyd and Thompson have made a particular point of re-asserting an approach based on an appropriately explicated labour process analysis, and I agree with much of their argument. I want here to underline a few key points. I take May as an example, because this explicitly goes back before Braverman to try to argue that an LPT analysis is insufficient. May has an impoverished view of what was actually found out. His point to frame the analysis is Roys famous piece on banana time. To begin with this study itself, it was used to illustrate the ways in which workers alleviate boredom by playing tricks on each other. So far, so good. But, first, what is this piece saying about resistance? Its key point
J. Scott, Corporate Business and Capitalist Classes (Oxford: OUP, 1997); R. Whitley et al., Masters of Business (London: Tavistock, 1981); J. Fidler, The British Business Elite (London: RKP, 1981). 52 P. Armstrong, Professional Knowledge and Social Mobility,Work, Employment and Society, 7 (1993), and other papers cited therein.
51

21

is that workers games allow them to tolerate boredom. In what sense this represents resistance is unclear. Second, Roy was not seeking out eternal truths of workers everywhere but asking how largely powerless workers make sense of their worlds and create some meaning. In his other studies, not cited by May, he showed that other workers bargain extensively over the terms of their effort and presumably do not need banana times to establish meaning.53 Turning to Mays wider arguments, he acknowledges the work of Hyman, Friedman and others but seems to have a limited grasp of the richness of these contributions. He sums up my own as regarding informal practices as symptomatic of a general, but also locally variable, desire to assert control (p. 770). This is true at some level, but I also went out of my way to reject the idea that there is simply a reservoir of latent discontent ready to burst out, and went into great detail as to the conditions that generate discontent and lead to its emergence in one form or another. Strikes and absenteeism for example are not expressions of the same thing. I was also interested in a lot more than informal practices, covering a range of formal ones in original empirical analysis and addressing in more general works long-term issues of class conflict.54 And, together with Hyman and many others, I stressed that any existing balance of relations between managers and workers was the crystallization of previous overt and tacit struggles and represented a historically generated compromise. Such views established, and illustrated empirically, what May seems to see (in the passage quoted in my main text) as new. May then moves on to his major theme. The question now arises as to whether the above perspectives are sufficient in accounting for indeterminacy and observed variations in local conditions. [A Foucaultian approach questions] the adequacy of the dichotomy between the subject and power at work, as utilised in the above accounts (p. 771). It is not shown that the accounts in question operate with this dichotomy. Studies such as Luptons not discussed by May demonstrate a clear interplay between workers and the power relations in which they are embedded. Some workers developed an extensive repertoire of means to bargain with management and in the process defined who they
53 54

D. Roy, Banana Time, Human Organization, 18 (1958); cf works cited at note 40, above. P. Edwards, Conflict at Work (Oxford: Blackwell, 1986).

22

were, in terms of their sense of shared collectivity and distance from management. Others did not, and were more isolated and individualized. Lupton, moreover, offered an explicit explanation of the differences that he observed. Finally, he stressed that workplace practices cannot be divided into resistance or consent though he did not use these terms but, rather, constitute complex wholes whose meanings need to be grasped in context. In my view, not only did work such as Luptons directly address indeterminacy and local conditions, but attempts to use Foucault to correct an alleged deficiency are far less successful. They are very little interested in explaining empirical variation and much more concerned with the fundamentals of power and subjectivity. They thus seem to make exactly the kind of error attributed to labour process writing. They are also so concerned with subjectivity that they often say very little about the concrete organization of work such as how tasks are allocated and how performance is monitored and negotiated. They also shift between the micro and the macro too easily. Thus May speaks of a shift from the space of banana time to the intensity of controls under just-in-time regimes as though there has been such a simple unilinear and universal trend, and as though the approach of Foucault allows us in some way to know that this has indeed been the case. A final point here concerns meaning and pride in work. Consider a recent study of resistance and its links with aesthetics and performativity.55 This argues that computer programmers take pride in the quality of work and that they resent, and resist, managerial concerns to deliver a product within a budget. This seems to be sensible if scarcely surprising. But, first, pride in work is not a new phenomenon. The theme runs through the classic studies, which emphasize the satisfactions of doing a good days work. This is true even in very repetitive jobs: Kusterer for example revealed the pride that producers of paper cups took in turning out a good product, and also the fact that this commitment to quality was often greater than that of managers concerned with output targets.56 And a
P. Case and E. Pieiro, Aesthetics, Performativity and Resistance in the Narratives of a Computer Programming Community, Human Relations, 59 (2006). 56 K. Kusterer, Know-how on the Job (Boulder: Westview, 1978). Contemporary workplace ethnographies abound with contextually rich accounts of the importance of feelings of pride in ones accomplishments: Hodson, op. cit., p. 45. There is a difference between the corporate and the working class. Corporate is
55

23

central theme in Roys work was that workers would find ways of achieving a task that were more efficient than those entailed in managements own rules. Second, the study of programmers lacks context in explaining just which workers resisted managers and under what conditions. The sense of resistance is presented as a characteristic shared by all the group and as a universal condition. Did in fact all or only some workers resist and why? Traditional workplace studies, by contrast, managed to deal with variation, in showing, for example, why some workers restrict effort, why some do not, and the conditions producing these results as well as developing a critical analysis of the very concept of output restriction.

only concerned with the dollar. The working class is concerned with doing the job right, the feel, and getting the job done (US worker quoted by Lamont, op, cit., p. 1). See also note 22 above for the theoretical base of this theme.

24