Bridging political economy analysis and critical institutionalism : an approach to help analyse institutional change for rural water services

This paper argues that approaches to understanding local institutions for natural resource management based on “critical institutionalism” (Cleaver 2012), which emphasises the importance of improvisation and adaptation across different scales, can be placed within broader political economy analysis frameworks for assessing challenges in public services delivery from national to local levels. The paper uses such an extended political economy analysis approach to understand the role of the international NGO WaterAid and its partners in Mali in relation to institutions for financing rural water services, drawing on collaborative research undertaken in 2010 and 2011. The case study shows that WaterAid’s approach can be understood through elements of both mainstream and critical institutionalist thinking. At local government level, WaterAid primarily promotes formal institutional arrangements, which exhibit the challenge of “reforms as signals” (Andrews 2013), where institutional reforms appear to happen but lack the intended function. However, the work of WaterAid’s partners at community level supports processes of “institutional bricolage” through which they try to gradually work with local actors to find ways of ‘best fit’ for financing rural water services which adapt existing local practices into new arrangements.


Introduction
The rural water sector in developing countries is shifting away from a reliance on community-based management towards institutional arrangements which involve wider sets of actors in an attempt to achieve sustainable rural water services (Lockwood and Smits 2011).In this context it is important to analyse how such institutional arrangements evolve.Key issues include the changing role of community participation and management, approaches for sharing long-term costs between different actors, and the role of external actors such as NGOs and donors.In this paper I link these practical questions to recent debates on theories of institutional change by using case study research on the financing and delivery of rural water supply in Mali which was carried out with the international NGO WaterAid in 2010 and 2011. 1 The paper responds to Cleaver's (2012) argument for placing detailed local-level analysis of institutional change within a broader framework which bridges different scales, considers the role of external actors and wider structural factors, and is "legible" to policymakers.To do this I link critical institutionalism and other academic literature on the influence of external actors in institutional change to forms of "political economy analysis" already used by international aid donors and think tanks.The overall aim of the paper is to illustrate a structured approach for analysing the complexities of institutional change in a way which is both academically rigorous and can provide useful guidance for practitioners and policymakers.
The term "political economy analysis" used here refers specifically to a variety of analytical approaches led or supported by international aid donors that have been developed since 2000 and draw in particular on the 'new institutional economics' of Douglass North and others (see North 1990). 2 These existing donorsupported political economy analysis approaches provide two useful starting points.Firstly, such forms of political economy analysis provide a systematic approach to analysing the relationships between key structural factors (such as historical processes and environmental issues), institutions (formal and informal rules, norms and arrangements) and actors in a given country or sector context.Secondly, these political economy analysis approaches also emphasise the roles of external organisations such as donors and NGOs themselves, and the need for these actors to adopt 'best fit' or 'good fit' approaches in their work.This means working with existing institutions in a way that is sensitive to the realities of the particular country and sector rather than trying to import 'best practice' institutions which may not fit the context (Landell-Mills et al. 2007;Booth 2012;Duncan and Williams 2012).
However, these forms of political economy analysis have been critiqued on both theoretical and practical grounds.Firstly, a series of authors argue that most political economy analysis is still based on over-simplified and sanitised understandings of politics and institutional change, in the broad belief that consensus can be built towards win-win situations that are beneficial for all involved (De Haan and Everest-Phillips 2007;Leftwich 2007;Hughes and Hutchison 2010).A second area of criticism points out the basic tensions between donors and NGOs attempting to engage in complex political analysis while at the same time responding to pressures to disburse money and achieve measurable results (Unsworth 2009;Robison 2010;Copestake and Williams 2012;Duncan and Williams 2012).Political economy analysis undertaken so far has also been hard to operationalize in terms of identifying specific practical recommendations or entry points for action (Chhotray and Hulme 2009;Unsworth 2009).
The combination of these challenges means that for international donors and NGOs "rejection of one-size-fits-all remedies, and the will to replace 'best practice' with 'good fit' approaches to institutional design, is at least a decade old … yet the new thinking still looks a lot like the old thinking" (Booth 2012, 92).Therefore I combine the work by donors and think tanks on political economy analysis with more in-depth academic approaches to analysing institutional change which can help understand how such processes of change really happen.
In particular, I use three areas of literature concerned with the role of external actors in institutional change in developing countries (summarised in Booth 2012;Cleaver 2012;Andrews 2013).These focus in turn on community-based natural resource management, decentralised local governance and the delivery of public services, and national-level public sector reform.All three of these areas of work are relevant to the case study topic of rural water services, and provide theoretical insights which help explain how financing and service delivery arrangements evolve at national, local government and community levels.In the next section of this paper I discuss and link these different areas of literature in order to develop the analytical approach.In the subsequent sections I then use the approach to analyse the issue of financing arrangements for rural water services in Mali.This analysis first considers structural issues and institutions at national level.The paper then examines the formation of institutions for financing water services at local government and community levels, and the particular roles played by WaterAid and its partners.

Concepts and method
Before discussing the areas of work which inform the approach, I clarify the use of the term 'institutions'.The definitions of institutions in the areas of literature under discussion differ according to whether: i) institutions are understood as rules, norms and arrangements ('rules of the game'), and organisations are understood as actors ('players of the game') (see Leftwich 2011); ii) institutions are understood as rules, norms and arrangements, which can exist or be represented in the form of organisations (see Cleaver 2012); or iii) institutions are understood as organisations, which are held together by and produce rules, norms and arrangements (see Lund 2006).Political economy analysis approaches broadly define institutions and organisations in the first way, as does Andrews (2013).Cleaver (2012) and the 'critical' institutionalist literature which she draws upon base their work on the second definition.In all cases, the authors concerned are actually interested in rules, norms, arrangements and their formation and interplay with different actors, regardless of which combination of these are labelled 'institutions'.I adopt an understanding based on (ii) for the purposes of this paper.In the next section I examine how each set of literature understands the processes of evolution and change of institutions.

Natural resource management, critical institutionalism and institutional bricolage
The first area of literature I draw on is that set out by Cleaver (2012), which is concerned predominantly with community-based natural resource management and the problems of approaches which focus too much on formal institutions and the search for general 'design principles'.Cleaver (2012, 16) sets out the differences between two schools of thought on institutions: "mainstream institutionalism" and "critical institutionalism".Cleaver argues that mainstream institutionalism seeks to identify common 'design principles' across different contexts and places too much emphasis on the local level without considering wider structures.Critical institutionalism refers to the body of literature which understands institutions as context-specific, dynamic and evolving, blurring across scales, and shaped by local history and politics.
Cleaver's comparison of mainstream and critical institutionalism highlights the key questions of decision-making, agency and the influences on actors' behaviour.Political economy analysis approaches have been criticised for containing limited analysis of how incentives and related political processes really work in relation to the nature of decision-making (Landell-Mills et al. 2007;DFID 2009).These are areas where political economy analysis approaches could adopt insights from critical institutionalism regarding the agency and behaviour of different actors.In particular, political economy analysis could benefit from engagement with what Copestake and Williams (2012) refer to as the "aidnography" literature3 (Mosse 2004;Eyben 2010).This work aims to understand the agency and actions of development workers themselves, as I discuss below in relation to the case study of WaterAid in Mali.
Cleaver proposes "institutional bricolage" as a concept to describe the way that institutions actually tend to emerge as a mixture of socially embedded (based on particular social and cultural practices) and bureaucratic (based on more formalised ideas and structures) (Cleaver 2002(Cleaver , 2012)).Cleaver identifies a series of key features of institutional bricolage, two of which are particularly relevant to the Mali case study.Firstly, bricolage involves improvising on existing practices with new ideas, and adapting innovations from elsewhere to fit a particular context.An example is taking formal state-endorsed ideas of water tariffs but then changing the rules on tariffs to suit local traditions.Secondly, the organisational forms of institutions (such as water management committees) often exist for multiple purposes rather than the single-purpose institutions imagined by mainstream institutionalism.
Finally, perhaps the most important element of comparison between mainstream institutionalism and critical institutionalism lies in what outcomes they see as possible.Critical institutionalism emphasises the possible unequal outcomes of institutional evolution rather than assuming that win-win solutions can be found as mainstream institutionalism and previous political economy analysis approaches tend to suggest.As I discuss in the following sections, the other areas of literature that I draw upon tend to be more optimistic about the ability of local actors and bricolaged processes of institutional change to deliver positive outcomes for the poor.

Local governance of public goods and practical hybridity
The second area of literature that I use is summarised in Booth (2012) and draws on the Africa Power and Politics Programme (APPP) research project led by the Overseas Development Institute (ODI).The key question of APPP was "which institutional patterns and governance arrangements seem to work relatively well and which work relatively badly in providing public goods, merit goods and other intermediate conditions for successful development?"(Booth 2012, vii).Local governance and the provision of public goods such as water and sanitation were key research areas.
This work is relevant to linking political economy analysis with better understandings of institutional change for two key reasons.Firstly, the APPP work addresses a similar set of problems to the critical institutionalist literature described by Cleaver, but widens the debate beyond natural resource management to the delivery of public goods and local governance above the micro-level of the community.Secondly, this literature seeks specifically to propose practical possibilities for action by external organisations because of a frustration that previous political economy analysis led to little changes in approach by aid agencies, as discussed above.Booth (2012) proposes the concept of "practical hybridity" to understand what is required for successful delivery of some public goods.Practical hybridity involves two elements.Firstly, the institutions that emerge address collective action problems in the particular context, rather than being externally-proposed solutions transferred from elsewhere (the argument of 'best fit' rather than 'best practice' already discussed).Secondly, such institutions will draw on "local cultural repertoires" (Booth 2012, 88), because this is less costly in terms of social disruption then creating new institutions from scratch.This argument is similar to Cleaver's (2012, 48) observation that institutions that socially fit "minimise cognitive energy" or (in economics terms) minimise transaction costs.
A further strength that the idea of practical hybridity shares with institutional bricolage is the focus on the actual outcomes that result in terms of access to public services or natural resources.Booth (2012, 81) argues that practical hybridity involves the formation of institutions that are not "merely palliative adaptations to the inadequacy of state provision."This is a crucial distinction that I explore further below in relation to the Mali case study.
However, a weakness of the concept of practical hybridity is that the APPP literature provides less detailed theorisation around issues of agency and the role of particular individuals than Cleaver's idea of institutional bricolage or the work of Andrews ( 2013) on institutional entrepreneurs.Finally, despite its policy-and practice-oriented motivations, the APPP work is still pessimistic about the role of external organisations in promoting practical hybridity, principally for the same reasons as the criticisms of political economy analysis which focus on donor constraints.

Public sector reform, bricolage and muddling through
The third area of literature discussed here focuses on externally-influenced public sector institutional reform, mostly at a macro level, such as reforms to public financial management systems.The key arguments are set out in Andrews (2013) and Andrews et al. (2012).Their argument is that the response to many intended public sector reforms in low-income countries, especially those reforms driven by external aid agencies, is that "governments and organisations pretend to reform by changing what policies or organisations look like rather than what they actually do" (Andrews et al. 2012, i).The authors call this phenomenon "reforms as signals" (Andrews 2013): governments prioritise form over function in order to satisfy donors and maintain aid flows.This is relevant to the discussion below of donor influence on national policymaking in Mali.
Instead, these authors argue that some form of "purposive muddling through" is required, suggesting that "incremental reforms focused on addressing problems frequently result in hybrid combinations of elements that work together to get the job done" (Andrews et al. 2012, 14).This requires supporting an environment for decision-making which encourages local experimentation and the engagement of wide sets of actors so that proposed reforms are feasible and implementable.
This literature is relevant in linking political economy analysis, mainstream institutionalism and critical institutionalism because of the way it takes ideas from new institutional economics as its starting point but then, like Cleaver and the critical institutionalist school of thought, uses concepts from other areas of social science.This literature also shows how ideas which are similar to those of the critical institutionalist school can be applied to national-level institutional changes.Andrews (2013), like Cleaver (2012), uses the concept of "bricolage" to refer to the process of recombining existing practices with new ideas to result in institutional change.This body of work also demonstrates a useful focus on development workers themselves, in a similar way to the "aidnography" literature introduced above.
Examining the strengths and weaknesses of these different areas of literature raises three key questions.The first is whether concepts such as bricolage and hybridity are analytically useful in understanding how and why institutional arrangements for public services have emerged in particular forms.The second is whether, by paying close attention to the outcomes of institutional arrangements (as Cleaver emphasises), we can distinguish between what Booth (2012, 81) refers to as "merely palliative" responses to a lack of state-delivered public services and "constructive" hybrid alternatives (Meagher 2012(Meagher , 1074)).
I address these first two issues by providing in Figure 1 a simplified representation of how these areas of literature on the formation of institutions relate to each other.The diagram shows how the concepts referred to lead to differences of 'form' (in terms of how institutional arrangements are mixed between the formal/bureaucratic and informal/socially embedded) and 'function' (in terms of access to public services for the poor).The diagram demonstrates the key difference between "practical hybridity" as described by Booth and "institutional bricolage" as set out by Cleaver in terms of the typical outcomes emerging from each process.
Thirdly, the literature raises the question of analysing both how institutional change at different levels actually happens (for example, through processes of bricolage), as well as understanding to what extent the approaches of WaterAid and other actors actively support bricolage and 'best fit' or attempt to promote more rigid ideas of 'best practice'.It is important to note that actors such as WaterAid may publicly claim to promote 'best practice' while being more sensitive to ideas of 'best fit' in their actual activities.The literature on the agency of development workers themselves is helpful in understanding these possible differences (Mosse 2004;Eyben 2010).
Before illustrating the approach with examples from research on financing rural water services in Mali, I summarise the key conceptual questions for the case study that arise from this approach.These questions overlap the different elements of a political economy analysis approach (structural factors, institutions and actors) and the different geographical scales of interest in Mali (national, municipal and community).
The first question is whether institutional reforms at the national level seem to be "reforms as signals" or to have emerged from "bricolage".Given the structural factors identified and discussed below, to what extent do policies and frameworks for rural water services delivery and financing represent "reforms as signals", where reforms are externally-driven and adopted on paper, but lack the intended functionality in practice?Is there any evidence of "bricolage" at national levels towards arrangements more likely to deliver effective services?
Secondly, how have local institutions at municipal and community levels for financing rural water services formed and what are the outcomes?To what extent do local institutions represent examples of attempts to implement 'best practice' principles, or examples of 'best fit' through processes of "practical hybridity" and "institutional bricolage"?From these observations, how can we understand the approaches of WaterAid and its partners at municipal and community levels in terms of "mainstream institutionalism" and "critical institutionalism"?What is the potential and limit of locally-driven solutions?
Thirdly, focusing on the relation between actors and institutions at local levels, what is the role of NGOs and their staff in institutional change?Can close engagement with local development workers help understand their decision-making and influence on institutional change, especially how they interpret national policy to fit their local context?

Results and discussion
The first question posed in the extended political economy approach requires an analysis of key issues at a national level that affect the financing and service delivery arrangements in the Mali rural water sector.This means looking beyond the water sector in order to understand broader structural factors of aid dependency, the influence of international donors on policymaking, and ongoing processes of decentralisation.

Aid dependency and the influence of donors
Since the early 2000s, aid received by Mali has consistently represented 10-15% of GNI and up to 50% of the government budget (OECD-DAC 2012).This aid dependency has resulted in policymaking processes dominated by donors, lack of national ownership of plans for development and poverty reduction, and national planning capacities which are weak in comparison to what is needed to satisfy donor demands (Dante et al. 2003;Bergamaschi 2009; Van de Walle 2012).Van de Walle (2012) notes that donor reports comment that national capacities for policymaking and public administration are improving, but that similar observations have been made as far back as the 1960s.This suggests that Mali is a case -in the words of Andrews (2013) -where history repeats itself through a series of attempted reforms in which donors blame contextual factors (such as lack of capacity) which they overlooked or chose to ignore.Failing to properly consider the context for possible reforms is one element of what Andrews calls "reforms as signals".
Some observers extend this argument by suggesting that government agreement with most donor demands has been due to a deliberate strategy by Malian leaders for ensuring the continuation of aid flows rather than solely a lack of administrative capacity (Bergamaschi 2008;Magassa and Meyer 2008).Therefore rather than demonstrating a lack of government leadership, this approach in fact serves a political function.This process seems to be an example of how "reforms as signals" result when donor funding is conditional on the implementation of certain reforms on paper by narrow sets of government agents, rather than engaging with the wide set of actors that would be needed to implement institutional change in practice.

Decentralisation reforms
However, one policy that observers suggest has been nationally-led rather than donor-driven is the process of decentralisation reforms since the 1990s (Baudais and Chauzal 2006;Bergamaschi 2008;Magassa and Meyer 2008).Three key motivations for the Malian government pursuing decentralisation are identified in the literature.Firstly, to respond to the demands of the National Conference of 1991 that decentralisation was a necessary part of democratisation, fitting both traditional ideas of bringing power back to the people and more modern arguments for the economic and social benefits of local governance (Rawson 2000;Pringle 2006).Secondly, the need for the central government to maintain its own political power and legitimacy was an important driver (Seely 2001;Smith 2001;Ouedraogo 2003).A related third reason was that decentralisation was also used as one way of responding to the Tuareg rebellion in northern Mali in the 1990s (Seely 2001;OECD-DAC 2012).
These reasons suggest an extension to Andrews' idea of governments undertaking "reforms as signals" to satisfy external donors.In the case of decentralisation, it appears that 'signalling' by central government was as significant in its relations to the population as in its dealings with the donors.The result is apparently similar: a process which privileges form over function, where municipalities exist but have little capacity to fulfil their roles.For example, one of the key challenges identified for decentralisation in Mali is the slow transfer of financial and technical resources from central to local government (Djiré 2004;Le Bay and Loquai 2008), highlighting the lack of financial authority that restricts decentralised governments (Johnson 2001;Smoke 2003;Conyers 2007).I discuss below the limited funds available to municipalities and the possibilities for seeking further financing in reference to the case study of the rural water sector.

National-level water policy and institutional framework
Although an official institutional framework for rural water supply in Mali exists, there is a lack of clarity over the responsibilities of different actors and a lack of capacity to fulfil their roles (Jones 2013b).In relation to financing arrangements for the sector, national policy is ambiguous about when exactly municipalities or central government can or should contribute to financing the costs of renewal or replacement of rural water infrastructure, a key factor in effective service delivery.In the face of this ambiguity, local actors have adapted within their own contexts as suggested by the idea of "bricolage", as I discuss below.Although it is difficult to say to what extent donors have influenced specific elements of policy relevant to the water sector, Mali's dependency on aid and the motivations of the political class to maintain aid flows have given international donors a strong influence over national policymaking in general.Therefore it is plausible that similar processes have occurred regarding policies and frameworks for the water sector.Policies such as community-based management of rural water supply and the recovery of costs from users have been adopted without sufficient consideration of the limits to these approaches and how support could be provided from higher levels of government (Jones 2013a).
In the remainder of this section I undertake more in-depth analysis of the roles of different actors at municipal government and community level.Given the observation that a lack of capacity and coherence hinders the implementation of national policies, it is necessary to analyse how organisations such as WaterAid and its partners respond to these challenges at local level.This requires assessing the extent to which these responses can be understood as "practical hybridity" and "institutional bricolage".In turn, this helps understand the potential and limits of local actors to develop effective systems for financing and running rural water services.I address two key questions.Firstly, how have local institutions for financing rural water services formed?Secondly, what is the role of NGOs and their staff in such processes of institutional change?Although these questions overlap the municipal government level and the community level, I structure the analysis by addressing each level in turn.

Mainstream and critical institutionalism at municipal level
Firstly I consider the processes of institutional change that are illustrated by examining the role of local governments in rural water services.Under decentralisation reforms, rural municipalities in Mali have responsibility for organising the construction of drinking water infrastructure and then ensuring that community-based or private operators are in place and monitored (Diarra et al. 2004).As of 2011, WaterAid works in 15 rural municipalities to support them in their role of planning, securing financing and organising the implementation of new infrastructure, as well as emphasising the role of municipalities in providing ongoing post-construction support to community management, such as monitoring, technical support and conflict resolution (WaterAid Mali and GERAD 2008).
However, despite their theoretical role, municipalities still have to look to central government for most capital investment financing because of their limited ability to raise local taxes (Coulibaly et al. 2010).As part of the conditions for applying for transfers from central government, municipalities have been required to elaborate local development plans, known as PDSECs (Social, Economic and Cultural Development Plans).These plans set out the public investment needs in the area.In theory, these plans were supposed to be developed through a participatory process with the local population.However, most municipalities relied extensively on NGOs and consultancy firms to help them develop these plans, and they focused more on meeting the technical demands of the documentation required than facilitating local input (Coulibaly and Hilhorst 2004).This process represents a local-level example of "reforms as signals"; the actual people whose role it is to promote change -in this case, local councillors, civil servants and citizens themselves -were left out of the conversation about what changes were desired.
Given this challenge, WaterAid has tried to help municipalities raise additional investment funds from other sources, principally international aid donors.In each of the municipalities where it works, WaterAid has helped to develop local Sector Development Plans for water and sanitation.These followed a similar principle to the process of elaborating the general local development plans (PDSECs).However, in a 2011 workshop on using the Sector Development Plans to demonstrate investment needs to potential donors, the majority of local government representatives participating admitted that they were not familiar with the plans for their own municipalities.
There were two key reasons for this lack of knowledge of the sector plans.Firstly, the process was dominated by the consultants employed by WaterAid, in a similar way to the criticisms made of the original PDSECs.Secondly, some officials had only been elected since the plans were first developed and there was no handover from one set of officials to the next.One councillor stated that he "saw lots of copies of the Sector Development Plan at the Mayor's office, but they were never distributed after the handover [to the new council]" (interview with municipal councillor 22 November 2011).This suggests that in some respects WaterAid's approach can also be considered as promoting "reforms as signals": priority was given to the 'form' of the processes rather than the actual 'function' of developing plans which would be locally-owned and used.WaterAid was aware of these challenges and had planned to support revisions of the Sector Development Plans in 2012 with greater involvement of each municipality and less reliance on external consultants, but this was put on hold after the coup d'état in March 2012.
However, feedback from local government representatives was that WaterAid's support to municipalities in developing the Sector Development Plans had some beneficial effect in terms of local coordination.Representatives in two of the municipalities gave the example of another international NGO working on water and sanitation which had previously developed relationships with particular villages without consulting representatives of the municipality or considering the municipalitywide priorities described in the Sector Development Plans.The interviewees explained that this NGO had now begun to start consulting with representatives of the municipalities concerned, and had indicated that in future it would work in line with the priorities expressed in the Sector Development Plans (interviews with municipal councillors and civil servants 4 November and 23 November 2011).
This example demonstrates Wateraid's strength in trying to address the challenge of a lack of coordination at municipal levels as well as a lack of resources.In effect, WaterAid tries to support institutional change from what Olivier de Sardan (2011) calls a "project-based mode" of local governance (where donors and NGOs represent the main sources of investment, but with low levels of coordination) towards a mode of governance with stronger "municipal" features (where local government can begin to provide the desired coordination in investment and service delivery, even if its own resources are limited).These issues demonstrate how WaterAid's approach can be understood partly through the ideas of mainstream institutionalism -for example, promoting formal processes, decision-making and incentives -and partly critical institutionalism, in its sensitivity to the ways in which institutions change through gradual adaptation and building on what already exists.
This evidence of characteristics of both mainstream institutionalism and critical institutionalism is also apparent when examining the second key role of municipalities in Mali in relation to rural water services, which is ensuring ongoing support to community management organisations.As a way of helping municipalities to ensure support to communities as well as the implementation of new infrastructure, since 2008 WaterAid has begun introducing a system of direct budget support to its partner municipalities.This involves creating a water and sanitation Technical Unit within each local government, made up of one to two members of paid staff who are employed as civil servants of the municipality and report to the elected mayor.However, their salaries and overheads are financed by WaterAid.
The municipal Technical Units aim to provide all the elements of support to communities for managing rural water supply which are proposed as necessary in the international water sector by Smits et al. (2011) andWaterAid (2011), and the cost of the Technical Units is roughly in line with the expenditures that international benchmarks suggest are required for supporting rural water services (Jones 2013a).However, this attempt to set up a new institutional arrangement starts from a position where almost nothing existed before (in terms of local government capacity for supporting water services) and requires financing beyond what the Mali water sector can currently provide (Jones 2013a, b).These observations can be understood as a mainstream institutionalist approach to creating new, formal institutions, rather than more gradually building on what exists locally.
Yet there are still aspects of critical institutionalism within this approach: WaterAid is trying to support the role of municipalities as actors who can promote problem-solving within their local contexts.In particular, the approach of one of the Technical Units again highlights how the work of WaterAid and its partners can be understood through a combination of ideas from mainstream and critical institutionalism.This particular Technical Unit has adopted a specific approach to sharing the long-term costs of operation and maintenance between water users and the municipality by developing its own local interpretation of the ambiguous national policy (Jones 2013b).The local approach is based on what the Technical Unit sees as realistic from its knowledge of communities in the municipality.In this way they actively promote ways of working which can be seen as "institutional bricolage" (a process which I discuss further at community level in the next section).
Yet despite this hybrid approach of trying to solve local problems in a way which responds to policy incoherence and builds on existing cultural practices, in some ways it is different to the examples of "practical hybridity" used by Booth (2012).Booth argues that such examples have tended to emerge independently of external interveners, and are not reliant on external funding.Yet the case of WaterAid's partner is possible because of the funding WaterAid provides (for the staff of the Technical Unit itself, as well as some discretionary budget for the Unit to use as part-funding for repairs to water infrastructure).The functionality rate of handpumps in this municipality at the time of research (90% functional) was significantly greater than the Mali average for rural areas of 71% (DNH 2010) (see Jones 2013b).This observation acts as a counterpoint to the pessimism about the ability of external actors to support local problem-solving expressed by Booth.

The role of NGOs and their staff in institutional change
A series of workshops was held with all of WaterAid's partner Technical Units and NGOs in 2011 to discuss together this issue of cost-sharing and whether the other partners had also adopted similar approaches to that of the particular Technical Unit discussed above.WaterAid's other partners had not developed such clear specific procedures for determining the contribution of users in case of breakdowns.Instead, they used more informal discussions on a case-by-case basis.This lack of clarity over the exact division of responsibilities between users and other actors reflects both the ambiguity in national policy, and the responses of WaterAid's partners to different levels of willingness and ability to pay amongst different communities (Jones 2013a).
I argue that the differences between policy and practice represent an example of Mosse's (2004, 639) argument that 'good policy' "legitimizes and mobilizes political support" while often being "unimplementable" in practice.As discussed above, it is reasonable to suppose, given wider donor-government relations and the dependence of the water sector on foreign aid, that mobilising domestic and international political support and therefore financial assistance was a plausible reason for water legislation and policy adopting the approach of cost recovery from users to the extent that they did.
However, while national cost-sharing policies may be 'good' in terms of gaining the necessary political support, the experiences and reactions of WaterAid's partners suggest that these policies are "unimplementable" (Mosse 2004) in practice.Mosse argues that development actors faced with this challenge try "to maintain coherent representations of their actions as instances of authorised policy" (Mosse 2004, 639).Eyben (2010) makes a similar argument, pointing out that if development workers are restricted by inappropriate policies or management approaches, they may try to work around these rules for the benefit of those their organisation is trying to support, while reporting their actions to managers and donors in ways that are framed according to the original policies.In this way they combine performance-related ways of working with the self-interest requirements of appearing to conform to instructions.The evidence suggests that WaterAid's partners are performance-oriented in the way they work with local communities to help develop forms of "institutional bricolage" which combine some aspects of national policy with local interpretations in practice.These processes of bricolage are explained in the next section in relation to understanding how communities really raise money for their contributions to water services.
However, the issue then arises of the extent to which WaterAid and its partners actually question national policy, given that they significantly adapt the policy in their own work.The links between community and local government levels are important in relation to this question.As described above, WaterAid's model of Technical Units at local government level can be understood more through a mainstream institutionalist view of how decentralised public services can work than a critical institutionalist perspective.Yet the actions of WaterAid's partners can be seen as a critical institutionalist approach to working with communities (as I discuss in more detail in the next section).It is possible to argue that if WaterAid questions the cost recovery model in national policy, it would raise awkward questions about the ability of their own Technical Unit approach to respond to this challenge at local government levels, since for the moment this model relies on WaterAid funds for key expenditure.As discussed in Jones (2013a), it would be helpful for WaterAid to discuss over what timeframe they think municipalities will continue to rely on donor funds, or if there are lower cost approaches which could be used until a future point where local governments are able to access sufficient revenue from taxes and central government transfers.

Institutional bricolage at community level
A further part of the research on financing arrangements involved the identification by WaterAid's partners of eight case study villages which were considered particularly successful at raising money for their contributions to the long-term costs of maintaining water infrastructure.Analysis of the case studies sought to understand whether their mechanisms for raising money were more successful because they employed the recommended national policies (paying per container collected or paying a regular tariff) or whether they were successful for other reasons.The analysis showed that villages actually use a combination of methods to raise money; some of these are suggested by policy but others are not.These communities represent examples of fundraising that are more successful than average because of the particular features of "institutional bricolage" that have emerged.Two of the key features of institutional bricolage are particularly relevant to understanding how communities pay for water in the case study villages: the idea that institutional bricolage commonly involves combining existing practices with ideas adapted from other contexts, and the result that bricolaged institutions are often multi-purpose and dynamic (Cleaver 2012).
To illustrate the first point, some villages adopted payment mechanisms which were similar to one of the official policy options: a regular tariff paid per household.However, in three of the case study communities the idea of a regular tariff had been borrowed or adapted from the existing practices of women's savings groups rather than being used directly as policy might suggest.In these villages, the regular tariff contributions were made by women depending on the frequency of the parallel system used by the women's savings groups.For example, in one village payments were made during the 6 months of the year when the women's group could raise income from the sale of fish.In these villages, the contribution from men was via an annual tariff or through a contribution to collective farming work if additional money was required beyond the amount raised by women.This use of elements of traditional practices in the community has been combined with more formal procedures that water management committees are supposed to adopt.This includes the committee formally recording this income and openly sharing its accounts, in order to demonstrate transparency to the users and show the council and the Technical Unit that it has raised its own contribution when it requires additional support (Sidibé and Jones 2011).
These observations also relate to Cleaver's argument that institutional bricolage commonly leads to institutions which serve multiple purposes rather than the typical single-purpose institutions which are often promoted by NGOs, such as water management committees.In some cases, the women's groups have not just lent ideas to the payment systems adopted for water management, but the women's groups themselves combine the facility for individual members to save and take loans with the possibility to jointly contribute to other collective community expenditures.For example, the women's group in one community had contributed money to the construction of a micro-dam for the village and the women's group in another had contributed to the cost of building a school classroom.The surplus deposited in a bank account could also be accessed for other unexpected expenditures even apart from repairs to water points.
This evidence also provides insights into the role of NGOs and external organisations such as the municipal Technical Units in making links between women's groups and other parts of communities involved in financing water services.This required repeat visits over a long period with continued discussion about what combination of practices might work to raise the money needed.By acknowledging and supporting practices of bricolage, the work of WaterAid's partners can be understood as a critical institutionalist approach.Instead of assuming that water management committees can be formed through applying consistent design principles, WaterAid's partners work with the community institutions that do exist and adapt their practices by drawing on both tradition and more modern ideas.

Extending political economy analysis
The extended political economy analysis approach used in this paper builds on existing work to show how political economy analysis can incorporate analytical concepts from more detailed theoretical literature, in line with the proposal outlined by Harris (2013).By investigating these relationships between actors and institutions across different scales, I have shown how WaterAid's partners develop local interpretations of national policy on financing rural water services, in a similar way to how Mosse (2004) and Eyben (2010) examine the responses of development workers to "unimplementable" policies or processes.The use of the wider political economy approach also demonstrates the effect of structural factors on how these arrangements emerged, such as the influence of Mali's aid dependency and decentralisation reforms.Putting this approach into practice through the partnership with WaterAid shows one way of considering within a political economy analysis approach what Copestake and Williams (2012) call the "micro-politics" of aid agencies, by trying to understand the day-to-day decisions of development workers such as the staff of WaterAid's partners (Jones 2014).This should be an essential part of political economy analysis.

Understanding institutional change
The use of this approach has helped both to explain the emergence of the institutions observed and to raise further questions for theories of institutional change.The analysis of the approaches to influencing institutional change adopted by WaterAid and its partner organisations at municipal and community levels shows that WaterAid's approach can be understood through elements of both mainstream and critical institutionalism (following Cleaver 2012).
At local government level, the formal institutional arrangements that WaterAid promotes (such as the creation and use of Sector Development Plans as tools to seek further financing from donors) exhibit the challenge of "reforms as signals" (Andrews 2013), where institutional reforms appear to happen but do not fulfil all of their intended function.However, some parts of WaterAid's approach at municipal levels can be understood through critical institutionalism, where WaterAid's partners try to support institutional changes through gradual processes of "bricolage" (following Cleaver 2012 and Andrews 2013).
Furthermore, one of WaterAid's partner Technical Units demonstrates at municipal level an approach close to Booth's (2012) idea of "practical hybridity" in the way it has developed a local system of cost-sharing between communities and local government in response to an unworkable national policy.This suggests an additional possibility to Booth's examples of practical hybridity, which generally rely on mobilising local resources in the absence of government or donor funds.The approach adopted by the Technical Unit is possible specifically because of the additional discretionary funds that WaterAid has made available, although as discussed below it can help mobilise some further resources at community level.
At community level, case studies of different villages show how institutions for financing the long-term costs of water services emerge through institutional bricolage, as a mix of both traditional arrangements and imported ideas.More importantly, this research helps show how external organisations can support processes of bricolage.The work of WaterAid's partners in different communities shows how they try to gradually work with local actors to find ways of 'best fit' for financing rural water services.However, there are also limits to this approach.Firstly, even communities identified as more successful in terms of fundraising rarely achieve the sums of money suggested as necessary by national policy.Secondly, working with communities in this way is a time-intensive process requiring frequent visits and follow-up.Both these factors mean that greater finances are needed at local government level.This is a key area where WaterAid can use its local experiences to promote debate at national levels about what financing policies are realistic over what timescales.
Overall, I draw together the observations about the role of WaterAid and its partners in institutional change (regarding the form of the institutional arrangements that they promote) and the potential and limits of these processes in terms of service delivery (the function that results) in Figure 2.This diagram builds on Figure 1, presented above, which I used to conceptualise the institutional arrangements and outcomes for public services described in the three areas of literature used in the analytical approach.Figure 2 therefore demonstrates how different elements of WaterAid's work reflect different approaches and results.On the left of the diagram are those which can be understood as "reforms as signals" (Andrews 2013), where institutional arrangements have a bureaucratic form but lack the intended function.On the right are parts of WaterAid's work which approach the ideas of "practical hybridity" (Booth 2012) through supporting institutional arrangements which are more socially embedded, but in this case where the potential for locally-driven solutions to effectively deliver services is limited by wider constraints such as funding.The diagram provides a way of thinking more broadly about what forms of institutional change external organisations such as WaterAid are able to support.Can they help promote changes further along the dotted arrow, whether at the level of national policy reform, local government, or communities? 4This is the key question for future analysis and for other organisations seeking to influence institutional change.In the Mali context, the approach developed here could be used to analyse in greater depth the changes since the coup d'état and political crisis in 2012, such as the weakening of the government and the arrival of new NGOs and donors, and what these factors might mean for the future of the rural water sector.This approach for analysing institutional change could also be applied to other sectors and/or organisations as a way of mapping other approaches and outcomes.

Figure 1 :
Figure 1: Conceptualising institutional arrangements and outcomes for public services.
Figure 2 sets out key examples from the work of WaterAid and its partners observed in this research in comparison to the concepts identified in the literature.

Figure 2 :
Figure 2: Conceptualising institutional arrangements and outcomes for public services: examples from this research.