Of texts and practices: Empowerment and organisational cultures in world bank-funded rural development programmes

Abstract

The World Bank's recent concern for ‘empowerment’ grows out of longer standing discussions of participation, non-governmental organisations (NGOs) and civil society. While commitments to empowerment enter World Bank texts with relative ease, their practice within Bank-funded projects is far more contingent, and the meanings they assume become much more diverse. This paper considers the relationship between such texts and the development practices which emerge, using an analysis of the ‘organisational cultures’ of the Bank and the many organisations on which it depends in the implementation of its rural development programmes. The paper presents a framework for analysing these organisational cultures in terms of (a) the broader contexts in which organisations and their staff are embedded; (b) the everyday practices within organisations; (c) the power relations within and among organisations; and (d) the meanings that come to dominate organisational practice. A case study of a development programme in Bangladesh is used to illustrate the ways in which cultural interactions between a variety of organisations – the World Bank, government agencies, NGOs, organisations of the poor, social enterprises – mediate the ways in which textual commitments to empowerment are translated into a range of diverse practices.

I Empowerment?

Notions of empowerment have gained a foothold in key statements from the World Bank. The World Development Report of 2000–01 identified empowerment as one of the three pillars of poverty reduction (World Bank 2001). Subsequently the Bank went on to produce a ‘Sourcebook’ on Empowerment and Poverty Reduction (World Bank, 2002), a text intended as a resource that its staff can consult when conceptualising and monitoring projects. Its contribution to the UN Social Development Summit plus 5 (World Bank, 2000a) trod a similar line. The latest World Development Report (World Bank, 2006) on equity and development offers the Bank's most explicit discussion to date of agency and the inter-relationships that exist between power and the distribution of assets:

Bringing equity to the centre of development builds on and integrates the major emphases in development thinking of the past 10 to 20 years – on markets, on human development, and on empowerment. (World Bank, 2006: p.17; our emphasis)

These changes within the Bank beg the question as to why – and how – this concern for an increased engagement with questions of power has emerged? Part of the answer is that such statements on empowerment can be traced back to roots within longer standing concerns among some individual Bank staff and certain member states about enhancing the role of popular participation through the increased involvement of NGOs and other civic organisations in the Bank's programmes. The discussion of empowerment in some of the institution's flagship documents might then reflect real – albeit incremental – victories for this group of people within the Bank's internal processes and politics (Bebbington et al., 2004, 2006).

How far are such statements likely to engender significant practical commitments to the reworking of unequal relations of power? Reflecting on the way in which the 2000–01 World Development Report presented its ideas on empowerment, Moore (2001) notes that much Bank literature discusses empowerment mainly in terms of community organisation. This, he argues, is a type and level of empowerment that ‘poses no serious threat’ (2001: 323) and one that governments are consequently quite willing to applaud. Furthermore, he suggests, an emphasis on community empowerment gives yet more support for social funds as an instrument, and may also form part of efforts by the Bank to seek greater support for its continued existence from development activists and NGOs in both North and South. Cooke's (2004) doubts are more forceful still. The fact that Poverty Reduction Strategy Papers (PRSPs), supposedly produced through participatory processes, usually end up endorsing neo-liberal programmes (which he argues are yet further ranging than were earlier structural adjustment programmes) suggests to Cooke that their production remains primarily structured by the Bank. Seen this way, participation in the PRSP process is a form of social control more than it is a process of empowerment (Cooke, 2004; cf. Cooke and Kothari, 2001).

Even if textual commitments did signal a significant engagement with structures of power, how likely are they to be reflected in terms of meaningful changes within the practices of Bank-funded projects? This question of the relationship between the Bank's writings, and the actions that take place within the projects it funds, is important to investigate empirically. Answers to this question would have both political and analytical resonances. Politically, they would help activists assess the overall utility and relevance of advocacy strategies that focus on influencing the texts produced by the Bank. Analytically, they would have implications for the relative usefulness of those critiques or celebrations of the Bank that focus primarily on interpreting and deconstructing the texts it produces. The relationship between text and practice lies at the core of this paper and its case study of the inner workings and perceived effects of a Bank-funded project that espoused a commitment to fostering the empowerment of the poor. More specifically, the paper hypothesises that the ‘cultures’ of the various organisations involved in conceiving and implementing such projects have significant effects on any achievement of empowerment objectives. In taking this perspective we are not seeking to endorse the cultural determinism that has frequently marred discussions of culture and development (Rao and Walton, 2004; Harriss, 2005). Instead, we argue that more careful attention to organisational questions of power and difference is important for forms of critique that might contribute to any rethinking of empowerment strategies.

The following section discusses relationships between ideas about organisational culture, empowerment and development intervention, and goes on to suggest a conceptual framework in which the relationship between development texts and project practices can be analysed. This framework is grounded in a particular notion of organisational culture – one that emphasises questions of context, practice, power and meaning. The third section of the paper explores the methodological challenges that we encountered in our research as we attempted to address these questions of organisational culture. The fourth section presents an exploration of the relationships between organisational cultures and empowerment processes using a case study of a particular Bank-funded rural development project concerned with strengthening sericulture in Bangladesh. The discussion in the fifth and final section of the paper analyses the case study in relation to the concepts of power and organisational culture raised earlier.

II Practice, power and context in development interventions

The World Bank is much discussed and critiqued, but its inner workings remain relatively understudied. While there is a sizeable literature on the Bank, much of it focuses on questions of institutional history, the analysis of its activities and performance based on secondary material, or critiques of its performance and politics based on discursive analysis of texts produced with some form of World Bank imprimatur. Though analytically and politically important, these kinds of works shed little light on why Bank funded projects have the effects they have. Despite some discussion of explanatory factors, such accounts often leave unclear the ways in which projects operate in practice and lead to specific outcomes.

There are of course exceptions. David Mosse (2004a) recently published an account of the work of anthropologists as producers of knowledge within the World Bank that is based on a period of insider experience as a visiting fellow within the development economics research group. Robert Wade's studies of the production of the Asian Miracle (World Bank, 1993; Wade, 1996) and the World Development Report 2000 (Wade, 2001), as well as his analyses of the way in which the Bank incorporated environmental concerns into its work (Wade, 1997, 2003), clearly demonstrate the ways in which organisational dynamics and international political economy help produce certain policy statements and project outcomes within the institution. Methodologically Wade's work was also made possible by his own presence as a professional within the World Bank. The point is relevant in interpreting a further study that has also thrown important light on the dynamics of projects linked to World Bank-supported programmes: that of James Ferguson (1990). Ferguson's influential study of a rural development project in Lesotho was based on ethnographic work at the level of a project intervention. This enquiry illuminated both the perverse institutional incentives within development projects, as well as the processes that lead systematically and repeatedly towards a depoliticisation of development problems. But at the same time, Ferguson's analysis of the links between these outcomes and documents produced by the World Bank was based primarily on deconstruction of texts. Though remarkably suggestive, the account left hanging some of the ways in which these texts were produced, and how this may have related to project practices. In very different ways, Ferguson's and Wade's work suggests that if one is to more fully understand how ideas expressed within texts are produced, and the ways in which such ideas then relate to project implementation, then it is important to analyse practices, values and commitments both within the World Bank itself and within the many other government and non-governmental agencies that help implement Bank-funded programmes.

This section therefore begins by considering this question in relation to the specific case of Bank commitments to empowerment. It then goes on to argue that recent work in organisational theory and social anthropology can help provide elements for an analysis of organisational practices within multi-agency World Bank programmes. These elements might be viewed as constituent parts of a concept of organisational culture, and we close the section with a framework for analysing organisational culture through such a lens.

Getting to grips with empowerment

Empowerment is an ill-defined concept used across a wide range of agencies within the development industry and also more widely in non-development settings. In his paper on the 2000–01 World Development Report, Moore is primarily concerned with political empowerment (Moore, 2001). But usages of the term ‘empowerment’ range from the radical philosophies of political and social work activists, through the world of private sector management-speak, to the more individualised sphere of personal self-help politics. Writing about the rise of such ‘buzzwords’ as empowerment, Cornwall and Brock (2004) point out that:

Empowerment has a … curious history, having gained the most expansive semantic range of all, with meanings pouring into development from an enormous diversity of sources, which include feminist scholarship, the Christian right, New Age self-help manuals, and business management … (Cornwall and Brock, 2004: 9)

While it may not be surprising that the term has such a range of origins and meanings, it is striking how these different meanings often seem to be unproblematically combined within development policy discourse. As Gledhill (1994) has argued, it is possible to detect a ‘functional ambiguity’ in such multiple meanings in which the very vagueness or malleability of the concept provides users of the term with a unique set of opportunities for negotiating and renegotiating its meanings flexibly among different groups or stakeholders. This ‘functional ambiguity’ may help explain why debates about the usefulness of ‘empowerment’ have so divided researchers and practitioners. Back in 1989, Caroline Moser identified an ‘empowerment approach’ as the most recent and potentially far-reaching policy approach within her conceptual framework that outlined different ideas about and approaches to gender planning. Yet ten years later we find the anthropologist Wendy James (1999) launching a scathing critique on the ways in which ‘empowerment’ within the ‘self-contained world of projects’ (1999: 13) has been largely stripped of its original meanings about the realities and distribution of power to refer instead to ‘responsibility delegated from above, or from the centre, to monitor others below or beyond one, for whose activities one has to be responsible’ (1999: 14).

It is significant that while the term ‘empowerment’ has emerged within the domains of policy and practice, it has little resonance within more theoretical debates about power. The classic analysis of the sociology of power by Lukes (1974/2005) contains no explicit references to or definitions of empowerment, even in its updated new edition. Following Locke, Lukes defines power in terms of capacity and potential and essentially as a dispositional concept – ‘being able to make or receive any change, or to resist it’ (69). Lukes argues that:

… social life can only be properly understood as an interplay of power and structure, a web of possibilities for agents, whose nature is both active and structured, to make choices and pursue strategies within given limits, which in consequence expand and contract over time. (Lukes, 2005: 68–9)

While such a notion of power as inherently relational is implied in some Bank discussions on empowerment, this is far from common. 1 Indeed, depending on which Bank document one is reading, the meaning of empowerment can run the full gamut of understandings, from those emphasising agency and social relations to those that understand power as an asset that can be built and bestowed on those without it. Until 2000, the two main broad usages that appeared within the Bank's documents were a primarily economic understanding of empowerment that emphasised market access, or a more managerial view that placed an emphasis on participation and decision-making.

The 2000–01 World Development Report, however, made the main contours of the concept seem clearer (World Bank, 2001). The report played a defining role in moving empowerment into the status of a common discourse within the Bank, and challenged the instrumentalist versions implied in earlier usages. 2 It depicted empowerment as a reworking of social relationships in a way that favours the less powerful, and argued that such empowerment can derive from two principal sources. The first is the reshaping of public and other institutions such that they become more accessible and responsive to the poor; and the second is the formation and strengthening of representative organisations of poor people. Significantly, the case study below examines a project claiming to pursue empowerment goals by working at these same two levels: fostering institutional reform, primarily within the public sector; and creating and strengthening certain types of (local) people's organisations, termed ‘organisations of the poor’ by the Bank.

The World Bank's 2002‘Sourcebook’ on empowerment was part of an attempt to carry this WDR agenda forward to a more operational level. The Sourcebook defines empowerment as ‘the expansion of assets and capabilities of poor people to participate in, negotiate with, influence, control, and hold accountable institutions that affect their lives’ (World Bank, 2002). Its empowerment framework speaks of operating at the same two levels identified in the WDR: ‘reform of state institutions’ and ‘investment in poor people and their organisations’ (World Bank, 2002: 23). It also talks of fostering ‘pro-poor market development’ as a means of fostering empowerment, in particular through the links between such markets and asset formation (World Bank, 2002: 50–66). However, the primary emphasis of the Sourcebook appears to be on national and local state reforms that foster empowerment through: (a) increasing citizens' access to information; (b) fostering inclusion and participation; (c) promoting accountability; and (d) fostering local organisational capacity (World Bank, 2002: 18–22). This focus on institutional reform may still, however, fail to satisfy Moore's (2001: 324–5) ‘robust initial set of tests’ of whether the poor are really empowered in such a process. For Moore, beyond any institutional reform, empowerment can only be possible when the poor ‘are organised politically’ and their ‘local organisations are integrated at higher levels’. At the very least, the types of reform referred to in such documents need not lead to any significant aggregated political organisation of the poor.

These two statements on empowerment, while considering questions of social relations, appear to lay greater emphasis on institutional arrangements and popular organisation in their understanding of power and empowerment processes. The 2006 World Development Report, however, takes these understandings somewhat further. Arguably, this change reflects, in part, the extent to which members of the 2006 writing team were also involved in the writing of the 2000–01 report and have reflected further on how to frame questions of power, both in a general sense as well as within the specific context of the Bank. 3 Whatever the case, this report – using the language of agency as much as that of power – pays far greater attention to the ways in which webs of a wide range of social relationships (institutional and not) structure the possibilities of human agency for different actors and social groups. In this report the notion of power as relational emerges more strongly than the notion of power as asset (cf. Rao and Walton, 2004). These changing and diverse uses over time alert us to the different ways in which a similar concept is used and understood in different parts of an institution.4 This in turn makes clear the possibility that notions of empowerment used in a given country programme may well be very distinct from those entertained by, for example, a WDR writing team. This diversity also leads to more distinct understandings of the types of policy and programmatic change necessary to foster empowerment, and rework power relations.5

If there are already different understandings of what a commitment to empowerment might mean within the World Bank, then there is scope for even greater diversity of meaning, interpretation and commitment among and within the various organisations that become involved in the implementation of a Bank-funded programme. This is the nub of the matter – for whatever the sorts of empowerment argued for in World Bank texts, the translation of such statements into bureaucratic development and political practices require particular national and institutional conditions. As Moore (2001: 328) notes, specific government programmes are more likely to create an environment that enables collective action of the poor when programme context and behaviour are characterised by tolerance, credibility, predictability and a recognition that the poor have rights to organisation, entitlements and programme benefits. The presence (or not) of such characteristics is to a considerable degree an effect of the dominant organisational cultures that influence programmes. The research reported here was predicated on the hypothesis that the ways in which programmes give meaning to the term empowerment, and the extent to which they might foster substantial changes in power relations, would depend considerably on the organisational cultures involved in programme conception and implementation.

Organisational culture and organisational theory

While for the purposes of systematic research the notion of organisational culture can appear vague, in our research and professional experience it is repeatedly invoked at a lay level in order to explain why organisations do what they do, and why projects end up taking the forms that they take. For example, in the Bangladesh sericulture case study discussed below, one of the participating NGO workers remarked that the staff of one of the other participating agencies, the silk foundation, were remote and did not know anything about ‘ordinary people's culture and conditions’. Our challenge, then, is to ground the concept of organisational culture in more specific terms that might allow its operationalisation in our effort to capture some of the issues to which such lay observers and participants seem to be referring when they use the term. There is a diverse literature within organisational theory that has aimed to do this. In much of this work, the starting point has been the study of the shared values, styles and practices that evolve within organisations. Such values can be understood as being the products of both wider societal cultures and (deliberately or not) of human agency within an organisation.

For Morgan (1997: 138), for instance, organisational culture is best defined as ‘a process of reality construction that allows people to see and understand particular events, actions, objects, utterances, or situations in distinctive ways’. Several management theorists also identify it as a force that helps shape aspects of an organisation's behaviour. For example, Peters and Waterman (1982) argued that businesses could operate more successfully if they built a strong unifying culture based on a ‘shared vision’. The enthusiasm for organisational culture that took root in the 1980s soon became extended beyond the business world into the not-for-profit or non-governmental sector as well. The idea of focusing on the ways in which people interacted within and between organisations and the values expressed in these interactions found favour with those analysts, such as Handy (1988), who were interested in improving the performance of voluntary or non-governmental organisations.

Such work was more prescriptive and normative than analytical in nature, and often conveyed organisational culture as something that could be easily created or manipulated. These more voluntarist conceptions also detached discussions of organisational culture from the history and political economy in which organisations are always embedded. Analytical approaches to the concept have focused more on questions of human agency and organisation. For example, Schein's (1985) model of organisational culture identifies it as an ongoing outcome of adaptive and integrative efforts of people to build the values and practices that can sustain an organisation. He identifies three interacting levels: artefacts, values and norms, and beliefs and assumptions, each of which help to shape an organisation's internal integration and external adaptation. Hawkins (1997) develops these ideas further in order to do justice to both the richness of the culture of an organisation and the complexity of organisational change processes. He identifies five levels of culture within an organisation: artefacts – such as policy documents, mission statements, dress codes, buildings; behaviour – such as what people do and say, what is rewarded, how conflict is resolved and how mistakes are treated; mindset – mainly the values and assumptions which inform and constrain behaviour; emotional ground – referring to the unconscious emotional states and needs which create a context within which events are perceived; motivational roots – the underlying sense of purpose which links the organisation and the individuals involved in it.

Once again, such work is arguably open to criticism for its instrumentalism and for downplaying questions of historical, social and political economic context. It can also be criticised for its conveyal of a predominantly integrative notion of organisational culture as a binding force within organisations with insufficient attention paid to negotiation, conflict and ambiguity. As Hatch (1997) points out, ‘sharing’ can also mean ‘dividing up’, implying issues of difference, tension and individuality. Following from this, a diversity approach to culture views organisations as microcosms of their wider environments, containing different cultural perspectives and power relations that lead to the reproduction of a range of tensions and conflicts within organisations. It also draws attention to the existence of sub-cultures within organisations – a view that reveals organisations to be far more complex and contradictory than the image conveyed in some management writing. Postmodern organisational theorists have gone furthest in this regard, arguing for a view of organisations as culturally fragmented, placing the concepts of ambiguity and flux at the centre of the analysis of organisational culture (Alvesson, 1994). Sub-cultures may serve to drive and support an organisation; but equally, they can conflict with a dominant organisational culture and challenge it (Dowling, 2001).

Organisation and culture in the anthropology of development

As management theorists engage with the notion of culture, anthropologists these days find themselves deeply ambiguous about the concept's explanatory value. For Douglas (1992) the main problem has been the way in which ‘culture’ has often tended to be used lazily as an all-purpose explanation. Kuper (1999) argues that a focus on culture can only bring partial explanation alongside equally important issues of power, politics and social institutions. Recent work within organisational anthropology has been deeply critical of the instrumentalist way in which organisational culture was taken up by management theorists, and Wright (1994) has argued that the reification of culture in such accounts can be challenged by focusing instead on the ‘process of organising’ – an approach which also acknowledges the need to engage with power, meaning and structure.

Despite this disciplinary ambivalence, debates within anthropology around the concept of culture both inside and outside organisations still stand in contrast to the lack of analytical engagement with the concept by many development researchers and practitioners. Anthropologists working on development issues have produced work relevant to our interest in power, history and agency within development organisations, although there has been little explicit engagement with the idea of ‘organisational culture’ as such. 6 Post-structuralist writings within the anthropology of development such as Ferguson (1990, and noted above) have analysed development as a discourse – a system of knowledge, technologies, practices and power relationships that serves to order and regulate people as the ‘objects’ of development intervention and which demarcates what can and cannot be said and done within such intervention processes. Ferguson's approach analyses project documents as texts, in order to show the way the language and concepts used limit certain ways of thinking while privileging others. The resulting ‘instrument effects’ impacts on the people who are ‘objects’ of intervention in ways that are generally advantageous to the interests of dominant institutions and the processes of capitalist development more widely. The approach helps reveal the system of power relationships that underlies not only the production of development texts but also the broader enterprise of development. Ferguson challenges our assumptions about the ways development organisations operate and highlights the embeddedness of organisations and practices within wider power relationships and practices. But the approach ultimately does not take us far in opening up the ‘black box’ of the workings of real development organisations.

The ‘actor-oriented approach’ provides us with more openings because it focuses on the many ways in which meanings associated with development are produced, contested and reworked in practice. Everyday practices are, as Hilhorst (2003) argues from this perspective, essential for understanding organisational realities. Such a perspective allows us to understand that goals set out in policy documents, those pursued in operational practice, and the personal goals pursued by ‘clients’ and local bureaucrats in the course of implementing (or ‘performing’) projects, are interrelated in complex ways and should not be conceived in simple instrumentalist terms. It therefore becomes essential to analyse organising processes – what people do – on their own terms (Long, 1992: 34), since practices cannot be viewed simply as straightforward efforts by actors to ‘put into practice’ a text. Following from this, Arce and Long (2000) make the case for an anthropology of development which engages with the ‘counter-tendencies’ which arise within mainstream or dominant processes of development, as people attempt to reorganise values, cross social boundaries and otherwise challenge the ways in which externally imposed development processes unfold.

Conceptualising organisational culture

Taken together, culture has been invoked in these different literatures to refer to questions of history, power, everyday practice and the meanings invoked in and ascribed to those practices. This is a broad spread, but also provides a basis for grounding a concept of organisational culture in a way that respects the traditional concerns for history, structure and power that infuses much critical writing in development, while also drawing attention to questions of everyday practice and the possibility that human agency (be it in the form of management or some other) might foster organisational cultures that are more conducive to fostering empowerment.

In this research the notion of organisational culture was used to draw attention to each of the following (Lewis et al., 2003):

•	Context. The historical, political economic and sociological contexts within which organisations are embedded influence greatly what is and can be done through such organisations. Organisations cannot be viewed independently of this: management and design possibilities are thus deeply affected by this context.
•	Practice. This context brings considerable influence to bear on everyday practices within organisations, but does not determine them in every respect. Thus, the practices of people within organisations cannot simply be read off the history, political control and role of the organisation. Practice (human agency) is thus a critical domain of enquiry when aiming to understand organisations.
•	Meanings. The thrust of much of the organisational culture literature in management is to insist that the values and meanings that are dominant within organisations will influence the performance, effects and potential of organisations to be vehicles for particular types of social change. Assessing how meanings and values are conveyed, and influence practice, is therefore also an important area of analysis.
•	Power. Which values, meanings and practices come to dominate in an organisation – and also within a project involving multiple organisations – inevitably reflect relationships of power within and among organisations. While the focus on practice keeps open the possibility that such power can be reworked, the emphasis on context emphasises – more perhaps than much writing on organisational culture – that there are limits on how far this is possible.

Seen thus, the concept of organisational culture – while not providing a new ‘black box’ or explanatory variable – forms a useful entry point into understanding the workings of multi-organisation development projects, such as those funded by the World Bank. It can add value to the analysis because it helps connect the broader context in which a project is embedded with issues of power, practice and meaning within the processes that unfold during a project's lifetime and which operate within and between the organisational actors who take part.

III Methodological challenges

To explore the influences of organisational culture on the practice of empowerment, three World Bank funded rural development programmes were originally studied in Bangladesh, Burkina Faso and Peru ( Table 1). 7 This was not a selection of ‘best practice’ projects, but rather of projects that had had empowerment objectives but had not been subsequently promoted within the World Bank as success stories. The rationale for this selection strategy was twofold. First, examples of such best practice projects are discussed elsewhere (see Uquillas and van Nieuwkoop, 2003; Guggenheim, 2006). Second, we felt that a study of less visible, ‘everyday’ and ‘ordinary’ projects would give more reliable insight into the relationship between written commitments to empowerment and the organisational conditions under which empowerment practices are enacted. One of these case study projects is discussed in detail below.

Table 1.  Projects studied

Country	Project and dates of operation	Project aims
Bangladesh	Silk Development Pilot Project (1997–)	To reform public sector sericulture support institutions and to improve research and extension services to poor non-traditional female silk producers who have been organised by NGOs into grassroots producer groups.
Peru	Natural Resource Management and Poverty Alleviation Project (1998–), part of the World Bank's support to the National Program for the Management of Watersheds and Soils (PRONAMACHS)	To foster watershed-based management of natural resources and the strengthening of community based organizations linked to environmental management
Burkina Faso	Pilot component (1998–) within the Second National Agricultural Service Project (PNDSA II), along with sub-components of The National Land Management Program (PNGT Phase 1), where the implementing agencies in rural areas were NGOs rather than the state	To test new ways of channelling funds directly to farmers' organizations for the selection and implementation of villagers' own initiatives; support to community–level environmental management initiatives and to village-level natural resource management committees and their work
Source: Lewis et al. (2003).

Our approach to this analytical problem reflected a partial response to recent calls for detailed ethnographies of development organisations (Cooper and Packard, 1997; Watts, 2001; Mosse and Lewis, 2005). Field research was conducted at various points over a twelve-month period. Although we did not use the most intensive form of participant observation associated with more detailed anthropological studies, our approach did allow us to observe and interact with a variety of institutional processes. The research was qualitative in nature, based on: semi-structured interviews with key informants at organisational and village levels; organisational and village level focus groups; participant observation of project and organisational processes; multi-stakeholder workshops; and an analysis of key project documents. 8 In each case, interviews were conducted with World Bank staff (though see below), sector-specific public sector officials, staff of local agencies implementing the programmes (be these of NGOs, or local public agencies), and members of those poor people's organisations involved in the projects. Interviews and observation of organisational practices were thus conducted at a variety of locations: villages, district and provincial capitals, capital cities, and in Washington DC.

As we noted earlier, much discussion of the Bank lacks analysis of its internal processes and dynamics, for understandable reasons. Gaining such research access is not easy, and researchers can often be distrusted. We found that this was also the case in much of our experience on this research project. Notwithstanding the fact that the study was supported by funds managed by the World Bank, our greatest difficulties were in gaining access to World Bank officials and documents in two of the three case study countries.

In Bangladesh, for instance, as the research was taken further out of Dhaka:

… we noticed a declining level of interest from the Dhaka World Bank office. In part, this was a result of the heavy pressures of work faced by these staff. Our research had clearly been generated from the centre (Washington DC) and we were making extra demands on busy staff time. It was also perhaps partly a result of the lack of enthusiasm for our choice of project – it was clear that the Silk Project was seen by many as fairly marginal to the Bank's overall programme, and as a project that had not gone as well as expected. Nevertheless we argued that much could be learned from it in terms of ‘generic’ project relationship issues that could have wider application.

Although we had asked our two Bank contacts in Dhaka to comment on a short preliminary report (3–4 pages) of thoughts and ideas in June 2001 arising from the first part of our fieldwork we received no feedback, despite some efforts at ‘chasing’ by email. The crunch came with our mid-term workshop meeting in January 2002, when both staff members at first agreed to make themselves available for the agreed day to take part, along with staff from the two case study NGOs and from the Silk Foundation. 9 We were still without any comments on what we had done, but we sensed disapproval.

Furthermore, as the day grew nearer there were hints from both staff that they might not be able to spare a day for the workshop … There were also signs that the Task Manager in particular disapproved of the research, while the Social Development adviser seemed to have changed her mind about its possible relevance or usefulness, perhaps partly as a result of being ‘caught in the middle’.

Matters came to head in January 2002 when, prior to the workshop we tried to make contact at the Bank office in Dhaka, and eventually were summoned to the World Bank office for a meeting at 8.30am on the morning our workshop was due to begin at 10 am – on the other side of town! Only now, at the last minute, did it seem that there was there a serious interest being taken in our work. After the earlier apparent interest and consent, we were now being asked ‘Why are you looking at organisational culture? What has that got to do with anything?’ Indeed, there was a strong implication that our research was actually a big waste of everybody's time. While there was still no feedback on our earlier document, concerns were voiced at certain phrases it contained, such as ‘organisations of the indigenous poor’. There were worries that we might be interested in stirring up ‘minorities’, then there were concerns that there were no ‘organisations of the poor’ in Bangladesh and ‘what did we mean’ by this.

At the 8.30 am meeting that took place on the day of the workshop, we were asked about what the research was about and we were again informed that the Bank ‘did not work with NGOs’. 10 We had to remind them quite forcefully that (a) the culture and poverty research was a legitimate World Bank research project that had been discussed with them in detail at inception and (b) our use of terminology such as ‘organisations of the poor’ was in fact drawn from the language of the Bank itself. In fact, the meeting itself seemed to be more about demonstrating to us the Bank's power (getting us down to the office on demand at a very inconvenient time to give some kind of account of ourselves) than about showing any real interest in the research itself. … In the end, the Task Manager did not attend the workshop at all (although she did not have the courtesy to tell us this in advance) and the Social Development person only attended the workshop for a couple of hours in the middle. (Lewis and Siddiqi, 2002: 26–7)

Similar issues arose in the other cases that we studied. In Burkina Faso, repeated efforts to engage World Bank staff and task managers in the research were unsuccessful. In Peru, where we had the best access to the World Bank, difficulties arose more with the government agency involved in implementing the Bank-funded programme. Interviewees' concerns about job security and an apparent strategy of upholding an institutional ‘code of silence’ led to missed meetings, undelivered documents, and off-the-record statements.

However, though limiting, these ‘methodological’ problems also provided substantive insights into the dynamics at play in the process of turning textual commitments to empowerment into practice. They seemed to reflect a lack of interest in the empowerment and organisational issues being pursued in the research, along with a general resistance on the part of a number of Bank staff to being an object of research and possible critique. And in the public agencies, they also reflected a desire not to address issues that might in some sense endanger continued Bank funding.

IV Bangladesh case study: The silk development project

In this paper we report on findings from one of the case studies conducted in this project – that in Bangladesh. 11 Bangladesh has a strong and longstanding non-governmental tradition of income generation and empowerment work as well as a long history of more conventional public sector project intervention. The Bangladesh Silk Development Project (which we refer to as the ‘silk project’) was an attempt to link these two elements of organisational heritage, by attempting to foster grassroots empowerment through involving NGOs and grassroots organisations in a traditionally public sector programme (Lewis and Siddiqi, 2002).

Sericulture is a well-established sector in Bangladesh, but it has long been characterised by top-down bureaucratic interference and has remained relatively uncompetitive in international markets compared with other production centres in China and Thailand (Van Schendel, 1995). 12 The silk project was approved with a US$11.35 million IDA credit by the World Bank in November 1997 as a five-year initiative which was designed to help revitalise silk production. The total cost of the project was US$13 m, with the difference made up from contributions from the government and participating NGOs. The project had two main objectives. The first was to assist in increasing the incomes of small-scale silk producers, through the introduction of improved technology, and the creation of institutional and policy improvements which would encourage sustainable development of the silk sector. This aim had both an income generation component at the household level and an empowerment element at both an individual and grassroots group level.

The second objective was to address the wider institutional, economic and technical constraints on silk development in Bangladesh. In order to shift the silk sector away from a traditional dependence on public sector production and marketing agencies, the project created a new autonomous organisation, called the Bangladesh Silk Foundation (which we refer to as the silk foundation), to provide technical assistance to private sector and non-governmental work in the sericulture sector. The silk foundation worked with nine NGOs that supported sericulture as an employment generation activity for low-income rural people. The ‘target’ group of the project was primarily poor rural women, who had already been organised by the NGOs into multi-purpose, semi-autonomous grassroots membership groups known as ‘shomitis’.

The project was relatively successful in two main areas. First, it increased silkworm rearing productivity and provided useful additional income for female producers. 13 Second, both women participants and the project staff with which they work, perceived some positive changes in women's status as they gained greater knowledge about sericulture production, increased local mobility and improved access to financial resources (Lewis and Siddiqi, 2002). Shomiti members spoke of changes in the balance of power within the household: ‘Since we came to the group, we do not have to ask our husbands for money in our hands, and we have something to spend’.

The project was rather less successful in two other main areas. The first was in the wider institutional restructuring objective, where resistance by both public sector and, to some extent non-governmental, interest groups remains high. For example, the silk foundation was given insufficient access to government-controlled hatcheries that are the only local source of high quality silkworm eggs. NGOs, too, seemed to have paid less attention than the overall project framework implied to the challenge of linking organisations of the poor with the market for silk products. The second lack of success was the project's inability to sustain coherent meanings around sericulture as a source of sustainable livelihoods which could help to strengthen the autonomy of organisations of the poor at the grassroots level, in relation both to the NGO intermediary organisations which originally founded the these groups, and through making use of emerging local and international markets for sericulture inputs and products. For example, some NGOs seemed to place as much emphasis on sericulture as a potential source of resources to support their own organisational sustainability as they did on the idea of building sustainable, autonomous grassroots groups.

Context

The project can be usefully understood in the context of a long history of sericulture policies in both the Indian state of West Bengal and Bangladesh – policies which Van Schendel (1995) has characterised in terms of ‘authoritarian developmentalism’. 14 Various international donors and agencies have had a long history of involvement with the promotion of sericulture – from the Salvation Army in the early part of the twentieth century to the Swiss government from the late 1970s. The evolution of a group of ‘silk bureaucrats’ within the Bangladesh Silk Board, combined with the ideas of the international agency experts, contributed to the definition of the ‘problem’ of sericulture and poverty chiefly in technical terms, in need of ever more scientific research and administrative intervention and driven primarily by the availability of external resources and little attention given to ‘the view from below’ (p. 181).15

In addition to the wider historical legacy of sericulture development work, there have been more recent pressures linked to the rapid growth of Bangladesh's now well known and high profile non-governmental sector (Lewis, 1997; Wood, 1997; Hashemi, 1999). NGOs view sericulture as a potentially attractive component of their poverty reduction programmes. As a high value commodity that lends itself to small-scale, household-based production, silk can form a useful addition to the rural non-farm livelihood portfolio of poor households. 16 Furthermore, in a society where women's subordinate position in the labour market is widespread, sericulture is an activity which is relatively gender-neutral and it is possible for NGOs to create a range of employment opportunities for women at different stages of the silk production cycle, from silkworm rearing to silk thread spinning.³ At the same time, as NGOs become less dependent (either by choice or by necessity) on foreign aid, they are searching for additional sources of income and see sericulture as a potential source of institutional income. The sale of locally produced handicrafts and other products has become one possible route to improved NGO financial viability (Stiles 2002).

Practice, power and meaning

Analysis of the everyday practices of key organisational actors within the silk project reveals a number of important tensions that help us understand ways in which actors within the project gave meaning to and engaged with a notion of empowerment (Lewis and Siddiqi, 2002). These tensions were revealed in various domains. The sale of sericulture inputs and outputs between producer groups and NGOs, for instance, was characterised by a lack of clarity and transparency. When group members were ready to sell their produce, they found themselves locked into what amounted to a ‘tied’ relationship with the NGO which provided the inputs and credit, and were therefore unable to access wider markets:

Because we take the silkworm eggs from the NGO, the condition is that we must give them what we produce. Even if another organisation offers a higher price, we cannot sell our cocoons to them.

There were also frequent criticisms made of the style in which transactions were conducted, with complaints about poor transparency and arbitrary pricing:

My husband was offered less than Tk100 by one staff member. Then another staff member said ‘Let's not give him so little, let's make it Tk100’. The NGO people don't weigh the produce in front of us, they just give us a lump sum.

Such comments highlight the ways in which everyday and organised practices undermined the project's stated wider concerns to achieve empowerment effects in the relationship between ‘organisations of the poor’ and NGOs.

At the same time, certain internal NGO practices also conflicted with wider empowerment objectives. Decision-making and management within both NGOs that were studied were found to be strongly hierarchical. This became apparent during discussions with junior staff about the roles of individuals within the NGOs' administrative systems. One of the NGOs was found to possess some of the characteristics of a ‘role culture’ (cf. Handy) which allowed systems to operate relatively independently of individuals: ‘It is not that we are led by a leader – rather, our systems drive people so even if the managers change, there will be others to implement the systems’.

But within the second NGO, workers were reluctant to step outside clearly circumscribed roles associated with particular individuals: ‘For policies we have different people in the administration – they know about policy, so you should talk to them. We do sericulture and if you want to know about that you will get all the information from us’.

These observations are important for two reasons. First, for the project to sustain its overall empowerment objectives the ability of an NGO to deal clearly and transparently with ‘its’ groups is likely to be a precondition for the emergence of strong, autonomous structures at the grassroots. Yet unequal power relationships within NGOs can work against building more equal and empowering partnerships between NGOs and grassroots groups. Second, they make clear the structural and cultural differences among NGOs, contradicting the commonly held view that NGOs share certain common characteristics and working methods. This diversity of NGOs is a challenge to the common tendency to assign them particular functional roles within inter-agency projects. Our fieldwork also uncovered considerable diversity within organisations of the poor, suggesting that a more nuanced view of organisations – often presented in development literature simply as ‘black boxes’ – is needed to reflect different sets of values and cultures.

Within the groups associated with the second NGO, there were signs that an authoritarian leadership culture was sometimes present which ran counter to the norms of democratic decision making (implied by the NGOs' presentation of group dynamics) and this was often resented by other members:

Look, I am the shomiti leader and I have to be responsible for any non-repayment … if necessary, I apply force to get the money paid back. For example, if I say that the NGO will be forced to take the roof off a person's house, or take away some of their household utensils, people get scared and then they make the repayments any way that they can.

At the same time, there was a strong view of the NGO as protector or patron. This was manifested in a reluctance to see a more autonomous future for the groups, either as producer cooperatives or otherwise freestanding entities: ‘If the people from the organisation are not there, having only money will not be enough. The root of a tree is very important. If Organisation 1 is not there, it will be hopeless’.

This was also reported from the first NGO's group members, one of whom remarked that ‘[the NGO] has looked after us like children, so we cannot leave them’.

In keeping with market-centred values held by dominant groups at the World Bank, and to a large extent reproduced within the new silk foundation, the project aimed to encourage grassroots groups to evolve over time into autonomous sericulture producer groups or cooperatives. However, such a view of NGO groups as future specialised producer cooperatives conflicted strongly with the original NGO goal of multi-purpose, developmental group-building: ‘Our groups are not sericulture groups or apiculture groups or fishing groups or anything else, they are formed for the development of the members as people’.

One reason why the NGOs were reluctant to encourage such autonomous, market-based activity within their pre-existing grassroots groups was this developmental focus, and they did not wish to see them become merely vehicles for business. Yet even here there is no easy distinction between an ‘NGO’ view and a ‘Bank-foundation’ view – for some grassroots groups are also quite interested in becoming cooperative businesses, and some NGO field workers had ideas which ran counter to the view of senior NGO staff. Indeed, some staff in the first NGO were actively giving their support to a local grassroots group who wished to explore the business option:

[The producer association] … is seven months old. Now they have 25 members and the savings rate is Tk20 minimum per month. 17 They meet monthly to discuss problems and possible solutions. They set up the association because they think they can do some collective buying of mulberry inputs like fertiliser and other equipment through loans from the association.

The design of the silk project is predicated on the notion that NGOs in Bangladesh are organisationally and culturally predisposed to fostering the social and economic empowerment of poor people, but that this empowerment will only be long-lasting if there are also changes in higher-level institutions. Yet these views are not shared equally or easily by all involved in the project. In the World Bank's Dhaka office, the official most closely linked to the project, while generally sympathetic to NGOs, also felt they were overly ‘controlling’ of producers, and that they were blocking the project's attempt to form producer associations because they wanted to benefit from silk production themselves. NGOs had ‘a lack of seriousness in promoting people's organisations through the sericulture project’. The General Manager of the silk foundation was likewise conflicted. While positive about the role that NGOs can play in the promotion of sericulture and clear that the foundation could only act as an advisory or coordinating body, he also argued that NGOs have their own agendas and tend to foster dependency among their groups of beneficiaries: ‘NGOs don't really want their groups to be empowered. If they did, they would have left them by now …’.

For their part, the participating NGOs differed in their perspectives on the silk foundation. Most generally concurred that there was now a more responsive agency in place to support and coordinate developments in the sericulture sub-sector, at least compared with the old bureaucratic and top-down silk board, which had largely ignored the NGOs. They also noted that the foundation deals more fairly with the NGOs and the producers. There were allegations that the earlier silk board did not do so, and had delayed payments for cocoons and undersold scarce silkworm eggs. Some also viewed the foundation as a powerful ally because of its privileged relationship with the World Bank: ‘The foundation has the backing of the government and the World Bank. So, it has the potential to play a strong policy and enforcement role in stopping illegal importation of foreign silk to Bangladesh’.

Yet at the same time there was a widely held perception among NGO staff that the silk foundation was a remote body that was out of touch with both the NGOs and the grassroots. One staff member commented: ‘We appreciate the foundation because we need it to play a coordination role in sericulture. But we'd like it to sit regularly with producers and discuss problems with them’.

The silk foundation is based in Dhaka in a tower block in one of the city's wealthier neighbourhoods, where formal office attire and air-conditioned work-spaces contrast starkly with the conditions under which silk producers and NGO field staff normally work. Another NGO staff member reported: ‘The foundation people don't really know anything about ordinary peoples' culture and conditions. They are educated people, but they don't know anything about the poor’.

And the lack of knowledge is mutual. Few of the beneficiaries we spoke to had heard of the silk foundation or had positive words to say about it. The following exchange took place with a group of rearers at the second NGO:

	Question: Do you know anyone from the BSF or the World Bank?
	One group member: No.
	Most senior rearer: Yes, they come to see our mulberry and our silkworms.
	Q. When they come, what do they see, what do they do?
	Answer: They go to see the trees. They sign papers, they ask us how it is going, but that's it. None of them ever came to ask about our problems like the way you are doing today.

If, as the Bank's Sourcebook on empowerment implies, information is (one element of) power, then power is unequally distributed within the Bangladesh silk project. Shomiti members continue to know little or nothing about the Bank or the foundation, and even those few who sit on the foundation's board play no real role in decision-making. Meanwhile, dependency still follows resource flows – leaving shomiti members the most dependent in this network of relationships, and the Bank the least. This is perhaps unsurprising, but it is significant. Even with an explicit empowerment aim, and significant NGO involvement, progress on empowerment has been poor.

V Empowerment, cultures and practices

A key conclusion from the Bangladesh case study is that there can be no straightforward ‘translation into practice’ of the Bank's textual commitments to empowerment as set out in its project and policy documents. 18 How far empowerment objectives are‘put into practice’ varies over the life of a project, and across the geographical space it covers. Perhaps this is not surprising in the light of increasing evidence from anthropological accounts of the policy process that increasingly question the relationships between policy and implementation. Indeed, the study leads us to question further the form of linear thinking implicit in thinking which might suggest simple causal linkages between policy ideas and practices or between the life of ideas and texts and the systems of meanings and practices which emerge (cf. Mosse, 2004b).

The reality is far more complex, since there are many factors that help determine a mosaic of varied meanings and practices of empowerment found across these projects among different actors. The analytical lens of organisational culture can help us to understand better the ‘process of organising’ (cf. Wright, 1994) that takes place within World Bank and other projects and, by extension, the ways in which meanings around the concept of empowerment – which were never particularly unified in the first place – find expression, refraction and ultimately fragmentation at the project level. Organisational culture is understood in relation to the interactions among the deeper socio-political context of organisations, power relationships between and within them, everyday practices and the dominant meanings and values in each organisation. In the Bangladesh case, the socio-political context of the programme intervention conditioned the ways in which empowerment ideas have taken shape, since NGOs activities around social mobilisation are a sensitive issue with the government. At the same time, the concept of empowerment itself was not sufficiently coherent as to become a viable organising principle for such mobilisation (cf. Moore, 2001). Finally, more deeply rooted professional cultures within partner organisations also reduced everyday commitment on the ground to stated empowerment objectives (cf. Chambers, 1993). The technical culture of the silk foundation made it difficult to engage with the empowerment dimensions implied within project documents, and despite the rhetoric of empowerment in many of the participating NGOs, there was little interest in extending this idea to the ‘beneficiaries’ of the silk-related activities. Indeed, NGO interpretations of empowerment, centred on the idea of building capacities within relatively dependent shomiti groups, differ significantly from local Bank-sanctioned forms of market-based empowerment.

And in the Bank itself, what Mosse (2004a: 78) has described as ‘the relentless demands and short-term pressures of meeting lending targets, managing a pipeline of projects, maintaining clients and avoiding a delivery ‘crunch’’ was found to be central to the organisation of Bank practices. This means that the goals of project approval, fund disbursement and the management of relationships with host governments took precedent over empowerment goals in the day to day practices of staff. The practices, values and power relationships in which organisations are embedded therefore influence what those organisations are able to do and the values that dominate their organisational practice. The relationships of power among organisations also influence the values that come to dominate overall programme intervention. In this web of relationships, perhaps the most influential power is that of the World Bank to sanction programmes through the potential reduction or withdrawal of financial support. Yet it is also the case that both in Bangladesh and the other cases studied, the power of the Bank to insist on practices aimed at reworking existing power relationships was limited. This reflects both the ability of government and NGOs to resist any such insistence, but also the fact that Bank commitment to such goals is frequently more individual than institutional, making it far more likely that government resistance strategies will be successful. Indeed, in some cases Bank commitment is limited to a few individuals who, even if they are task managers, are not always able to follow through on their commitments because other staff and managers are far less committed to empowerment or because, in the case of Washington DC based staff, they are too distant from everyday practices to be able to exercise much leverage. And finally, Bank staff members are also subject to other incentives and pressures – above all to disburse loans – which ultimately trump empowerment commitments.

Power of course is also exercised in the daily conduct of relationships within organisations in a way that influences the values that are present in project practice. Here there is clearly much more variability, and the sui generis characteristics of particular local organisations and above all, particular local directors becomes much more influential. In the silk project, local practices varied according to the style of the particular subcontracted NGOs. Those that were more horizontal more committed to local organisational development and more participatory influenced the local practice of the project in those ways. Again, then, who is hired, and which organisations are subcontracted, matters a great deal because the commitments and organisational cultures involved influence overall project practice and outcomes. Similar variation can be seen among the community-based organisations involved in the silk project. Some are dominated by more authoritarian cultures, others by more horizontal ones; some are more patriarchal, others far less so. Dominant values within these organisations affected project outcomes.

All this said, power relationships around the flows of financial resources in project interventions go a long way in structuring the possibility that pro-empowerment values and practices will become prominent in the organisations involved in programme implementation. Here, the Bank's control over project monies is clearly determinative, though differential patterns of financial control elsewhere in the project chain were also influential. The silk project, for instance, required implementing NGOs to pursue entrepreneurial or business-oriented approaches. To an extent, these NGOs had already been moving in this direction as part of their own income generating strategies, but the further insistence on this deepened tensions within the NGOs between value systems based on solidarity and philanthropy, and ones based on market orientation. Certainly the influence of the project's business values upon NGOs was greater than any influence that NGO commitments to solidarity had upon the values of the project.

VI Conclusion: Prospects for empowerment

While there are many factors that mediate the extent to which different projects put empowerment objectives into practice, it is also the case that each project that we studied in different ways had certain empowerment effects. Some of these effects appear at a personal and household level – for instance, in terms of changing gender relationships. Such effects derive as much from the material aspects of the projects as from their organisation building work, giving some credence to the notion that empowerment is also more generally about the expansion of poor people's assets (World Bank, 2002). This was clear in Bangladesh, where the explicit attempt to increase women's income with silk production enhanced their bargaining power within households and local organisations – though the extent to which this was so varied according to the prior gender dynamics within these organisations.

In some cases, local organisations appeared to have become stronger and more assertive as a result of engaging in participatory and – in some cases – self-management processes within the projects. Where this was the case, it depended a great deal on the particular approaches of local project staff that, for reasons of local agency culture and personal commitment, worked more closely with local organisations. However, this seems to have been more the exception than the rule, a subculture or countertendency within the dominant culture of the organisation. The more general image is one of local organisations being dependent on the chains of relationships involved in the projects and especially the resources running through these relationships. When these resource flows were interrupted, local organisations generally encountered internal problems. Nor did it seem to matter whether these organisations were working directly with government or non-governmental agencies – their level of dependence was equally severe. This was apparent in the Bangladesh case where local project participants were quite clear – and resentful – that they depended so much on the NGOs working within the silk project, for resources, training and market access.

The Bangladesh programme case study offers very little solace for those who would call for more attention to making public sector institutions and other intermediaries more pro-poor. Project specific efforts to reform public sector institutions have not fared well in the face of far more deeply sedimented elements of organisational culture – whether these are related to the dominance of technical over social concerns, or to the deeply held view that public agencies can be used for regime specific political ends. Efforts to by-pass these problems by working through NGOs do not necessarily change this picture. While our analysis does not concur with Moore's claim that ‘all else being equal, the use of NGOs as implementers of public programmes is likely to demobilise the poor’ (Moore, 2001: 327), they do at least support his observation that ‘dependence on NGOs as empowerment agents is more problematic than we conventionally assume’ (Moore, 2001: 325).

Fostering empowerment was never going to be an easy road – the history of liberation theology and critical pedagogy ought to have told us as much. To take empowerment seriously is to go against entrenched values, interests and powers. Even more problematically, the culture of each of the organisations through which empowerment initiatives are pursued is itself embedded in these same values, interests and social relationships. Understanding better the ways that counter-tendencies and subcultures can emerge and grow within these organisational and institutional contexts is thus a critical task: for theory and for practice.

Acknowledgements

This paper reports on research funded under the Culture and Development research programme managed by the Poverty and Social Development groups of the World Bank, with funds from the Trust Fund on Culture and Poverty, financed by the Netherlands Ministry of Foreign Affairs. We are very grateful for the assistance received from staff of the different agencies and projects discussed in the paper. Thanks also to the many people (some who must remain unnamed) who helped us in the course of the research. The views expressed are the authors' own.

Additional information

Notes on contributors

Anthony Bebbington

Bebbington and Lewis are joint first authors of this article.

Notes

1. More recently Eyben (2005), commenting on WDR 2006, still finds an over-reliance on externally-defined categories and pre-constituted individuals in relation to poverty concepts and analysis, rather than what she terms a ‘relational’ concept of society that is informed by a concept of power.

2. For instance, a search of the 1998–99 WDR reveals only one mention of empowerment in the entire text. This was a highly instrumental usage which spoke of the power of technology to teach ‘governments and institutions, about the poor, for designing programs that benefit them, and for enhancing their participation and empowerment’ (World Bank, 1999: 60).

3. As one of us has observed within the Bank (Bebbington), different understandings of concepts such as empowerment are probably better explained by staff's personal views of the world than by their particular disciplines or indeed their formal job titles.

4. There are, of course, potentially significant new changes at the Bank taking place as we write. The arrival of Paul Wolfowitz may signal another possible shift in the way that the concept of empowerment may be used – for instance, as a justification for funding processes of regime change.

5. In a somewhat similar sense, Bebbington and Barrientos (2005) have explored the extent to which different parts of a World Bank country programme understand poverty differently.

6. Although it does not address the issue of organisational culture directly, an exception is a recent paper by Rossi (2004) on agency and power that discusses the strengths and weaknesses of a Foucauldian analysis of institutions and practices within a development project in West Africa.

7. The countries were chosen according to researcher expertise. Given the range of meanings of the term, the research focused on projects that had relatively modest empowerment objectives. At a minimum we were interested in interventions that sought to: increase local organisational capacity, with or without some degree of higher-level federation of local organisations; and foster empowerment through asset formation. These were therefore projects that combined material objectives (income enhancement, sustainable natural resource management) with varying types of socio-political objectives (strengthening landless peoples' groups, changing gender relations, fostering local governance of natural resources, or increasing the access that base groups have to public resources). The projects also involved a range of organisations (allowing us to look at the effect of organisational cultures on relationships among actors).

8. Fieldwork was conducted by a combination of locally-based local and expatriate researchers and the project coordinators.

9. In fact, the workshop date itself had been previously fixed after consultation as to the convenience and availability of our two main World Bank contacts.

10. The research proposal had been sent to both Bank staff numerous times, and we had discussed it personally with each of them on several occasions when we met at different stages of the research.

11. See Olson and Bebbington (2006) for the Peru case study.

12. The sericulture sector in Bangladesh has two separate sub-sectors: a ‘traditional’ sub-sector in which households have been active as silk producers for many generations and a non-traditional sub-sector – the main focus of the project – in which development organisations such as NGOs have promoted sericulture through extension and training to low income households as a new supplementary income source.

13. Silk production involves a complex chain of seven interlinked stages: mulberry sapling preparation at nurseries, mulberry plantation (in land ‘blocks’ or by the roadside), silkworm rearing, cocoon production, silk yarn reeling, weaving the silk thread into fabric, and finally, the production of garments. Because they are difficult to produce, silkworm eggs (dim) are not available on the market but only from the government silk board hatcheries and now, also from the silk foundation. The NGOs buy a quantity of eggs and then pass them on to their producers as part of an overall credit and training package. The eggs cannot be bought by producers individually, due to the economies of scale involved in production. Once the silk worms hatch, they are fed on mulberry leaves.

14. Recent cooperative sericulture initiatives driven ‘from below’ in West Bengal and showing benefits in terms of increasing the power and visibility of adhivasi women (Webster, 2002) may constitute an alternative – and potentially more sustainable – trajectory within the ongoing struggle to make sericulture a tool for poverty reduction.

15. There is a striking parallel here with the case of aquaculture in Bangladesh, which has been constructed as a policy problem primarily in technical terms at the expense of broader social and economic concerns (Lewis, 1998).

16. Some NGO staff did also suggest an alternative explanation – namely that many NGOs had given up on sericulture as a sustainable income generation activity, but had kept it going because of the availability of subsidies through projects such as this one.

17. One US $ is approximately 65 Bangladesh taka.

18. The research undertaken in Peru and Burkina Faso produced broadly similar sets of findings.