Transforming Development Knowledge GREEN POWER FOR AFRICA : OVERCOMING THE MAIN CONSTRAINTS

Recent literature on the effectiveness of donor programmes points to the importance of understanding the political context within which reforms are taking place. The characteristics of the power sector make reform intensely political in almost all countries and donor projects have sometimes failed because of an inability to navigate the local politics of reform. This article reviews what is known about how donors have taken politics into account in designing and implementing power sector reform programmes in sub-Saharan Africa. It illustrates the challenges which donors have faced with reference to a case study of donor attempts to support power sector reform in Tanzania. The article draws on documentary evidence from major donors as well as a set of qualitative interviews with experienced project supervisors to provide a set of lessons for donors about how to incorporate political context into the design and implementation of power sector projects.


Introduction
There has been a resurgence of interest in recent years, amongst both practitioners and the academic community, in the political economy of donor engagement in reform processes (Carothers and de Gramont 2013). A series of studies has suggested that projects that take a flexible and adaptive approach to reform have been more successful. For example, Andrews, Pritchett and Woolcock (2012) and Andrews (2014) have argued that programmes should focus on problem-solving through an iterative process; whilst Pritchett, Samji and Hammer (2013) have argued that projects should 'crawl the design space' for solutions. Similarly, Booth (2015) argues for the need to 'think and work politically', whilst  provides a theoretical framework for how reform can 'work with the grain' of the domestic political reality. There is also a growing literature of comparative case studies (e.g. Booth and Unsworth 2014) that argues that aid programmes which are flexible, long-term and locally owned are likely to be much more effective in achieving sustainable change than more traditional 'linear' programmes.
Donors have been paying attention to this literature and are increasingly recognising that an appreciation of the politics of the sectors and countries in which they are operating is essential for successful and sustainable reform. One of the consequences of this work has been considerably greater interest and uptake by donors in 'political economy analysis' (PEA) (Fritz, Levy and Ort 2014). Many donors now routinely conduct a PEA as part of the preparation of a project and some embed regular sectoral or issue-based PEAs into the operation of projects. Some have gone even further -for example, the Department for International Development (DFID) has sent several hundred of its staff on a training course about politics and institutions in order to try to change the mindset of technical staff about the importance of political considerations in projects. What is less clear is the extent to which such analysis is translating into different designs and different approaches to the implementation of projects. A recent review of DFID experience suggested that, whilst PEAs have been mainstreamed in many areas, this has resulted in relatively little change in the types of projects that are actually implemented (Piron et al. 2016).
This article examines how donors have tried to take political context into account in their support of reform efforts in one particular area and region -power sector reform in sub-Saharan Africa. In most countries, power systems are centralised, mainly based on a national grid, and although electricity is part of the development narrative of most countries, it tends to be controlled by a domestic elite, and is a significant source of rents. Thus, although there is politics associated with reform in all sectors, the characteristics of the power sector make accounting for political considerations particularly important.
There is already a literature on the political economy of power sector reform. Numerous papers have looked at the politics of reform (see Gratwick and Eberhard (2008) and Eberhard et al. (2016) for descriptions of reform in sub-Saharan Africa; Dubash and Rajan (2001) and Tongia (2006) describe the political economy of reform in India; Victor and Heller (2006) describe the politics of reform in five major developing economies; Scott and Seth (2013) provide a literature review of reform of electricity distribution in developing countries; Kojima, Bacon and Trimble (2014) review the literature on reform of power sector subsidies). However, this literature focuses on the political economy of the reform process within individual countries, where the main actors are government, business, and voters.
By contrast, rather little has been written about the role of donors and the extent to which they have taken knowledge about the politics of reform into account in the design and implementation of their programmes in Africa. Our research questions are: to what extent have donors analysed the underlying political constraints that they face? Have donors significantly shifted the nature of the power sector reform programmes which they implement as a result of a better understanding of the political context? Are they taking on board the lessons from recent research on 'thinking and working politically' and, if so, how? Are there general lessons that can be learned about what sorts of approaches to power sector reform are more, or less, successful and how this varies by context? This article reviews the existing studies -both published and, where available, donor evaluations -looking at how donors have navigated the political economy of power sector reform. It complements this by collecting documentation on power sector reform projects implemented by the major donors in Tanzania over the last decade or so. This is supplemented with qualitative analysis based on a set of interviews with experienced donor officials responsible for the implementation of power sector reform projects. We conclude with a summary of the lessons, from the literature and our case study, about how donors might respond more effectively to the political challenges associated with power sector reform.

Methodology
Our methodology has three components.
First, we examine the literature about the political economy of aid in the power sector (including reviews conducted by donors such as Deloitte Touche Tohmatsu Emerging Markets (2004); Besant-Jones (2006); and Deloitte (2015)). We limit the scope of this review to studies that have examined the role of donors in power sector reform in developing countries and how they have designed and implemented such programmes in the prevailing political context in the countries of operation. Thus, we do not review all studies on the political economy of power sector reform, but focus on synthesising available literature on how this has influenced donor programming in sub-Saharan Africa in particular.
Second, we have attempted to obtain information and documentation about the power sector reform projects supported by donors over the last decade (or longer where appropriate) in Tanzania. Tanzania was chosen because of the length and intensity of donor engagement in power sector reform, providing a literature and experience to draw on. The documentation for these projects was then used to answer a set of questions including: how political considerations were taken into account in the design of the project; how challenges and blockages were handled during the course of implementation; what the overall performance of the project was; and what lessons were learned in terms of project design and implementation.
Third, we conducted a series of qualitative interviews with experienced project supervisors from the major donors who were supporting power sector reform for Tanzania. This was in order to obtain their views about the critical success factors for power sector reform projects and, specifically, how projects have attempted to take into account the politics of sector reform. 2 3 A brief history of donor engagement in power sector reform Although donors have been involved in supporting power sector investments since the 1950s, the nature of development partner involvement in power sector reform in developing countries is rooted in the transformations that affected the sector in Organisation for Economic Co-operation and Development (OECD) countries in the 1980s. During that time, a combination of political, financial, and technical factors triggered power sector reforms in the UK, Chile, and Norway (Gratwick and Eberhard 2008). These countries' experiences appeared to demonstrate the benefits of such reforms, which were then followed by several other industrialised and some developing nations from the early 1990s. Since then, power sector reform has been advocated by the development community, including the World Bank, the Inter-American Development Bank, the European Bank of Reconstruction and Development, as well as other international agencies, such as the World Energy Council (Bacon and Besant-Jones 2002). Thus, donors have played a major role as the architects of reforms in the power sector, and have often attempted to initiate reform through the provision of technical assistance and capacity-building programmes in developing countries (Wamukonya 2003; Dornan 2014).
At first, most development partners typically implemented a somewhat uniform approach to power sector reform. The approach taken in OECD countries crystallised into a 'standard' or 'textbook' approach to restructuring the sector, which aimed at fully unbundling and liberalising the power sector following a logical sequence of distinct steps: corporatisation, commercialisation, legislation, regulation, restructuring, privatisation, and competition (Gratwick and Eberhard 2008;Joskow 2006;Littlechild 2006;Hunt 2002).
The development community promoted power sector reform out of a belief that the standard 'model' would enable the transformation of poorly performing energy systems in developing countries, promoting growth and improving access for poor populations. It was felt that the recommended structure and regulation of the sector would benefit consumers by allowing their participation in the market and by ensuring consumer protection (United States Government Accountability Office 2005). In the words of one former senior donor official from a major multilateral donor, 'there was, in the early 1990s, a strong belief that one size did actually fit all.' However, the application of the 'standard model' of reform in developing countries yielded rather modest results, as it faced significant political barriers (Besant-Jones 2006;Choynowski 2004). The consensus in the literature is that the standard model failed for three main reasons.
First, it often failed to take account of the vastly different circumstances prevailing in developing countries from those in the OECD countries where it was first implemented. For example, the textbook model for power sector reform did not make sense in many small African countries, where the necessary degree of competition in the generation segment cannot be obtained due to the small size of the power system and insufficient generation capacity (Barnett 2014). Second, the model struggled to map out a feasible pathway for reform. As Victor and Heller summarise, '[T]he standard textbook for reform focuses on the end point, namely an unbundled, privately owned and competitive power sector, not on the steps that governments need to take towards that end ' (2006: 21). In practice, governments tried a wide variety of approaches to reach the end point, not all of which were successful (Chikuni et al. 2011).
Third, and perhaps most important, many reform attempts failed to understand or, at least, to take account of, the underlying political constraints facing decision-makers. In almost all countries, reform of the power sector is an extremely sensitive area. Electricity is part of the development vision of all countries and therefore brings significant political benefits to leaders who can control the price of and access to a key developmental service. Economies of scale mean that large financial flows are involved in procuring power production, transmission, and distribution systems. The centralised nature of the technology concentrates control in the hands of relatively few powerful individuals. 3 As a result, the location of transmission and distribution lines can be driven by electoral considerations, power may be rationed to influence voters, and power generation may fluctuate with the election cycle (Tripp 2012; de Mesquita and Smith 2009). Utilities have historically been used to serve the broader patronage system and became large employers (Barnett 2014). As reforms often emphasise restructuring utilities, they have faced strong resistance from labour unions (Eberhard 2004;Dubash and Rajan 2001). Reforms also encouraged cost-reflective pricing, but the associated price increases have resulted in popular uprisings in several countries including Argentina, India, Indonesia, Ghana, and South Africa (Dubash and Rajan 2001). 4 Thus, rather than the 'standard model', the reforms of the last two decades have generated a wide variety of 'hybrid' structures (see Eberhard et al. (2016); Trimble et al. (2016) for a recent classification of different systems in Africa). In each case, these reflect the outcome of a complex and context-specific contestation both between domestic actors (utilities, independent power producers, regulators, finance ministries, energy ministries, and political leaders) and between domestic and international actors (independent power producers, donors, and other financiers).
Donors responded to the challenges of implementation in different ways. Some development partners had considerably more leverage than others due to the scale of the resources which they were providing. The literature points to examples, predominantly from the 1990s, where countries were effectively forced to implement the reforms suggested by donors because of the conditions included in loan packages (Wamukonya 2003; Lefevre and Todoc 2000).
There is also evidence that pressure to implement the recommended reforms was strengthened by coordination amongst donors. The World Bank, for example, refused to fund more capital investments until specific reforms had been implemented and most other donors would only support programmes that the Bank had approved (Barnett 2014). Some donors also responded by attempting to insulate reforms from politics. For example, Bouille, Dubrovsky and Maurer (2002) describe how in Argentina, a small group of politically powerful bureaucrats, supported by multilateral agencies, designed the reforms without engaging with other relevant public agencies or civil society. However, this opaque process created political opposition and resulted in a lack of ownership of the reforms by key stakeholders.
By the early 2000s, the donor community was conscious of the difficulties being encountered in the implementation of power sector reform in developing countries. The World Bank commissioned a major review of its power sector operations which has been extremely influential in shaping the approach taken by most development partners over the last decade (Besant-Jones 2006). It concluded that 'the most important lesson from reforming power markets in developing countries is that "cookbook" solutions for reforming their power markets are ruled out by the extensive range of economic and institutional endowments of these countries' (ibid.: 1).
The consequence of this reappraisal was the abandonment of the 'one-size-fits-all' approach. The World Bank issued new Operational Guidance to its staff (World Bank 2004), which emphasised context specificity and the importance of the political dimensions of reform. In particular, a stronger focus was put on identifying 'stakeholders with the incentive and influence to press for improved performance' and 'top-level political decision makers', who will be able to champion the reform process (Fritz et al. 2014: 134;. The last decade has therefore seen a considerable amount of experimentation by development partners and closer attention to understanding the political context. But donors still face considerable challenges in navigating the complex politics of power sector reform and it is not clear whether the deeper understanding of context has yet translated into donors undertaking different interventions in the countries that they support. Thompson and Bazilian (2014) assert that donor-funded technical assistance and capacity-building activities still avoid 'the fundamental political issue of who wields political power and how that power is wielded'. And Naqvi (2016), in examining the role of development actors in power sector reform in Pakistan, highlights how awareness of the complex political problems associated with reform is not sufficient to avoid them, pointing to a higher order of structural obstacles within development partners that remain unaddressed.
To understand these obstacles further and assess how donors have addressed them, the next section examines reforms in Tanzania and the role of development partners in the process.

The Tanzanian reform experience
The Tanzanian power sector is dominated by a state-owned and vertically integrated utility, the Tanzania Electricity Supply Company (TANESCO), which has been struggling financially for many years. Its precarious financial situation has led to chronic underinvestment in the sector, from both public and private investors, which in turn resulted in significant underperformance for decades, both in terms of quality of supply and coverage. The lack of reliable electricity supply has been identified as one of the three major constraints to growth in the country (Partnership for Growth 2011).
Reforms began in the 1990s, when the country's structural adjustment programme supported by the International Monetary Fund (IMF) and the World Bank, was expanded to the power sector through the inclusion of sector-specific conditions in general budget support. Restructuring and privatisation were deemed the solution to TANESCO's financial deficits and the generation segment was opened to independent power producers (Ghanadan and Eberhard 2007). In the mid-1990s, reflecting the expectations that the private sector would fill the investment gap in the generation segment, donors mostly withdrew their direct support to the development of power projects. Unfortunately, the relatively rapid reduction in donor funding for power supply and technical investments worsened the investment gap in the sector, as private capital failed to pick up.
Despite the changes associated with the reforms, neither TANESCO's financial situation nor the quality of supply improved. This led to the management of TANESCO being outsourced to a private company from 2002 until the end of 2006. The management contract placed an overwhelming emphasis on increasing the utility's short-term revenues, which may have deterred long-term investments. The results of this were also unconvincing -whilst commercial performance improved, there was little improvement in service quality or transmission and distribution losses. In 2005, the newly elected government terminated the management contract at the end of 2006 and de-specified TANESCO for privatisation.
Consequently, between 1997 and 2006, the utility suffered from a critical lack of investments, in its workforce, systems, and infrastructure, reflecting the view that long-term investments would not benefit the government or the managing entity, as TANESCO was expected to be privatised. The 'generation gap' left in TANESCO by the lack of recruitment in the last decade is still a major concern for the utility. The utility also requires a technological turnaround and related capacity building, as its systems have not kept speed with the revolution in industry-specific technologies.
Notwithstanding the difficulties in reforming TANESCO, the 2000s saw significant changes in the legal, regulatory, and institutional environment. The Energy and Water Utilities Regulatory Authority (EWURA) Act was passed in 2001 (although it only came into being in 2006). On paper, the institutional structure and governance of the sector is excellent, as the legislation guarantees the autonomy of the regulator and the participation of the public. However, in practice, the Ministry of Energy and Minerals can exercise political pressure and intervenes on regulatory issues, in particular, tariff setting. Other changes included the creation of a Rural Electrification Agency in 2005, to focus on extending access. And in 2008 an Electricity Act was passed that encourages private participation in the sector.
Whilst these legislative and institutional changes might signal a renewed commitment for creating a conducive environment for private investments (Kapika and Eberhard 2013), the deals struck with private investors have not always been transparent, which has heightened suspicion about private sector investments. For example, the lack of planning in the sector, combined with severe droughts, forced TANESCO to resort to emergency power producers in 2006 and 2011. The absence of competitive and transparent tendering resulted in one of the largest scandals in the country's history (Kapika and Eberhard 2013) and put TANESCO on the verge of bankruptcy, as the negotiated cost of electricity with emergency power producers significantly exceeded the tariff. This experience demonstrates that strong negotiation skills and an in-depth knowledge of competitive procurement in the utility are paramount to enable private sector participation in the long term.
In 2013, the Government of Tanzania, through 'Big Results Now', a programme aimed at fast-tracking projects of national importance in key sectors of the economy, including energy, promised a reform of the sector (Government of Tanzania 2013) including unbundling, commercialisation, and sale. This has been followed by a new Roadmap which was approved by the Cabinet in 2015. However, the latter was funded by the African Development Bank and was undertaken by a Western consultancy company. Its recommendations bear a strong resemblance to the 'standard model', promising to unbundle and liberalise the power sector by 2025. The election of President Magufuli in 2015 led to the dismantling of the Big Results Now programme. In January 2017, the managing director of TANESCO was removed and a tariff increase was reversed; more recently, the director-general of the regulator was suspended. Both moves point to the continued influence of political considerations in the management of the power sector.

The role of donors
Traditionally, the dominant approach to influencing power sector reform in Tanzania has been through general budget support. Combined commitments from donors were substantial and represented a potentially influential tool, since some donors have made the disbursement of their loans conditional on the adoption of reforms.
Since the Busan Partnership agreement on development cooperation in December 2011, 'a lot of donors were going for low-hanging fruits' to yield measurable 'results' quickly, 'instead of thinking of the long term,' according to one interviewee. 5 Also, despite some willingness amongst donors to develop a common Performance Assessment Framework to measure the effectiveness of aid, there has been a lack of alignment between the framework and the conditions of disbursement of further loans. This was due to both the complexity of the results framework and competing donor interests and approaches.
Coordination is also made more difficult by the large number of development partners working in Tanzania; and this is reinforced by the limited ability of the Tanzanian government to coordinate them. In an attempt to address this, donors have formed the Energy Development Partners' Group, intended as a sector-wide platform for dialogue with government. However, as each country's presence in Tanzania is founded on bilateral diplomatic relationships, the government has no incentive to deal with development partners as a united block of donors.
Commercial interests have not been absent from the considerations of bilateral donors when promoting reforms. In the words of one senior donor official 6 from a major bilateral donor, the 'dialogue with the main sector stakeholders started very good, and development partners managed to affect them in many many areas, until a number of development partners wanted to promote their interests more than that of the Tanzanians. Then the relationship became difficult.' This has undermined trust between the government and donors, since it is not always clear to the government that the approach being promoted is in the long-term interest of Tanzania.
Furthermore, some development partners have a particular ideological position about unbundling and liberalisation, despite the limited number of actors in the sector, which makes it difficult to achieve enough competition in the market, and the lack of high-level expertise necessary to manage the independent system operator required. In other cases, the relatively rapid rotation of staff in donor agencies can mean that there is an insufficient understanding of the country's politics combined with a lack of skill in supervising and monitoring reform recommendations produced by independent consultants to ensure that they are consistent with the local context. This might also be amplified by the fact that consultants are often contracted by donors, with limited inputs from the government, and, in turn, frequently recycle the standard model with relatively little modification for country context.
The experience of donor engagements in Tanzania exemplifies some of the key challenges faced by external actors in promoting power sector reform. Frequent changes in government policy and attitude towards reform, combined with a mistrust of the motives of foreign partners and an historical antipathy towards the private sector make it hard for development partners to support a reform process. For their part, a lack of serious analysis of the underlying political constraints, combined with rapid turnover of donor officials, has resulted in a reliance on a standard model of reform with seemingly little attempt to tailor this to Tanzania's context. The result has been de jure reforms which satisfy the donors and release funding, but little de facto reform in the way in which the sector operates. However, the efforts of specific individuals to build trusting relationships with key government counterparts and tailor interventions accordingly have yielded important practical reforms in some areas.

Lessons learned
Our analysis of the academic literature and documentation from donor projects, along with our interviews with donor officials suggest some general lessons for policymakers, both in development partners and developing country governments.

Analysing the underlying politics of change in a country is valuablebut only if used
One of the surprising findings that emerged from our interviews was the absence of any significant formal analysis of the political context, in which reforms were being undertaken. Reforms were viewed as technical with the result that the only analysis undertaken prior to the implementation of projects were assessments of the different technical aspects of the project, rather than assessment of the wider political context and the incentives facing the various stakeholders. 7 Even where assessments are done, there seems to be little connection between the theoretical assessments provided by political economy analysis and operational programmes implemented on the ground (Levy and Palale (2014) provide an exception for Zambia). There is also a tension between those that believe that developing a deep understanding of the local political context is essential to success and others who believe that remaining politically 'neutral' requires focusing only on technical aspects.
Notwithstanding this, almost all donor officials interviewed displayed a detailed understanding of the political context in which they operated, with many arguing that success depended far more on intensive engagement with counterparts in government than any formal analysis of the context. As one respondent put it, ' [H]aving key staff and development partners work intensively and over a long period of time [with the government] is important.' 8 However, the instruments at their disposal, the pressure to disburse, and the processes with which they had to comply seemed to, at best, discourage taking a slower and more reflective approach to engagement. For all donors, it appears to be difficult to not pursue reform, even if there is evidence that the particular pathway being proposed is unlikely to be successful.
Indeed, most projects provide an optimistic view about what will be achieved and only mention political interference as a potential risk to implementation. This is despite the fact most donor officials accepted that political interference in the power sector was a near certainty, and that many previous projects had cost two or three times as much as originally anticipated and taken two or three times longer than originally planned. It is not entirely clear why development partners are driven to pursue reforms; this may relate to the 'mental model' adopted by the donor (see Section 5.4), or to the belief that funds are only likely to be effective if reforms take place. However, it would appear that the internal approval processes for such projects may sometimes discourage honesty about the practical political challenges that they entail.
There would also appear to be a reluctance to apply such analysis ex post as well. With the exception of the World Bank, no other development partner had readily available project completion reports for power sector projects (including those supporting policy reforms) and virtually no donors appear to have assessed in a systematic fashion the political factors that determined success or failure. There would appear to be a strong case for more systematic ex post evaluation of power sector projects.

Flexibility is important -but there are reasons why it is difficult
Much of the recent mantra about project design is that donor programmes (in all sectors) should be flexible and adaptive (Andrews et al. 2012). Indeed, some of the successes reflect the ability for development partners to act in a flexible and opportunistic way. Flexibility also allows donors to shift focus to take account of changes in government and key personnel in counterparts, as well as enabling implementing partners to experiment with alternative approaches to solving problems. 9 However, respondents also pointed to significant structural and procedural reasons as to why flexibility is difficult. For example, USAID receives funding from Congress that is allocated by sector and by country, inhibiting the flexibility to shift resources from less effective to more effective areas. Internal structures can also mattercomplex reforms often require multidisciplinary approaches, but donor staff are often arranged by sector and discipline, making building a multidisciplinary team difficult. Different types of donors can also work with different counterparts. Bilateral donors generally have the capability to work with a range of counterparts in developing countries, including sometimes, the private sector and civil society and to work through contractors. Multilateral organisations, by contrast, are typically constrained to work through government which can restrict the nature of programmes that they can operate. This is a particular problem in the power sector, where projects often must work through the Ministry of Energy, even where political influence by the ministry is a key constraint in progressing reform.
Flexibility is particularly difficult when it comes to stopping ongoing initiatives. Once approved, the obligation on donor officials and implementers is to make a success of a project. This provides strong incentives to continue activities, even when external, often politically induced, changes have made the chances of success slim. Conversely, officials have greater flexibility to add components to programmes, with the result that programmes evolve by growth rather than by re-allocation.
Flexible programmes are also perceived to be more difficult to evaluate. Building a good relationship may be key to achieving reform, but is hard to assess in a results matrix. Flexible programmes typically consist of multiple initiatives which are not known in advance, making it much harder to design a monitoring and evaluation framework for such programmes. 10 Donor programmes are evolving to address some of the challenges created by flexible programming, but difficulties remain. For example, some of the recent programmes designed by DFID in other sectors have deliberately built-in flexibility to allow a response to changing and emerging needs of the host country government. This may be easier in larger programmes, especially those backed by a long-standing relationship between donors and host country governments. At the same time, the monitoring and evaluation can be facilitated by retaining a clear sense of the overall outcomes that the donor programme is seeking to effect. But the stakeholders interviewed for this case study suggested that, despite some progress towards more flexible programme design, the structural and procedural constraints outlined previously remain a significant barrier to more flexible and politically informed ways of working.

Dialogue, trust, and personal relationships are of critical importance for reform
Perhaps the most consistent message from our interviews related to the critical importance of dialogue, trust, and personal relationships with key decision-makers. Again, this is true for all sectors, not just the power sector 11 -but the long-term nature and sunk costs of power sector reforms make the issue of particular salience. Respondents often attributed programme failures to a lack of trust between the government and donors. Almost every instance of successful reform was related to the construction of an effective working relationship through intensive and repeated interaction, often over a long period of time. A number of respondents pointed to the need to understand the nature of the constraints facing policymakers and the need to 'listen more and lecture less'. One respondent argued that loan conditionality was only effective when the conditions were suggested by the government as a mechanism to help them to handle domestic resistance. Overall, there was a consensus that 'there is no substitute for an experienced and credible [donor] staff member who has the trust and confidence of the key decision-maker.' 12 Notwithstanding the almost universal support for the importance of building trust, there was a sense from some respondents that donor structures and training are making this more difficult than before. Some officials highlighted the burden of process compliance taking time that might otherwise be spent engaging key stakeholders. Some related the bureaucratic burden to risk aversion associated with (donor country) domestic political concerns; others pointed to the results agenda as a source of additional administrative work to 'prove' that aid is working.
A number of respondents suggested that the frequent rotation of donor staff undermines the ability to build long-term relationships with key government counterparts and noted the difficulty of maintaining institutional memory as well as a good understanding of the country's recent history. For example, one official suggested that the reticence of Tanzanian officials to pursue certain types of reform was rooted in the experience of reform attempts in the early 1990s; failing to understand this might result in pursuing reform models which, although technically sound, were politically unfeasible.

Mental models matter
As noted in the literature review, the application of the standard model for power sector reform has not generally been successful in developing countries. Notwithstanding the growing body of evidence supporting this, the core elements of the standard model -unbundling, commercialisation, privatisation, competition -still appear to hold sway in the thinking, both of development partners and some government officials. However, some senior donor officials argue that this view is mistaken, firstly because the standard model was rarely actually implemented in practice and so cannot be deemed to have failed; and secondly, because adherence to the model was abandoned in the early 2000s, notably by the World Bank, as evidence of its unsuitability came to light.
Although both of these are true, our research leads us to believe that the 'standard model' still holds considerable sway as a 'mental model'. The standard model is clear, coherent and logical. It was generally regarded as a success in England and Wales, Chile and Norway. Despite accepting the need for context specificity, development partners have struggled to adjust because there is no other compelling 'mental model' to which to refer. Rather, a set of path-dependent 'hybrid' models has begun to develop (Gratwick and Eberhard 2008) based on local experimentation within different political contexts. However, most international consultants, on whom donor officials depend, are still most familiar with the standard model. As a result, it has been difficult for donors to shift their advice, whilst developing countries have found it hard to gain donor acceptance of the somewhat expedient and path-dependent approaches that have been implemented in practice.

There may be an opportunity to build domestic demand for reform
One of the surprising findings from the project documentation and interviews was that development partners have put relatively little effort into building a wider domestic constituency for reform in the countries in which they operate. In general, donors appear to have interpreted the 'country ownership' principle of the Busan Partnership for Effective Development to mean government ownership. None of the projects reviewed in Tanzania included activities to build support for reform outside of government.
There are several reasons for this. First, in some countries donor projects are tightly controlled by the government, which may be unsympathetic towards activities that attempt to support advocacy activities. Moreover, many donors, particularly the multilateral development banks, typically work directly with government counterparts, making work with other stakeholders more challenging. Indeed, some respondents argued that, in certain countries, it was neither possible nor necessary to engage with wider civil society since all the key decisions were taken by a narrow group centred on the government.
Second, several donor officials said that engaging groups outside of government in favour of support was not possible because there simply was not a constituency of support for reform. For example, large industrial customers, who constitute the majority of electricity demand, can have significant influence over policy -but their interest is primarily in maintaining low prices. Small businesses without electricity might be expected to support improvements in access, but generally do not wish to pay a significantly higher tariff to obtain it. And the majority of households in Tanzania are poor and are not believed to support cost-reflective tariffs that would significantly increase the price of access and supply.
This lack of interest in building coalitions in support of reform is nonetheless surprising, given the growing literature on the importance of building broader coalitions to achieve sustainable reform (see Faustino and Booth 2014;Buckley, McCulloch and Travis 2017). Whilst no one is likely to be in favour of price increases in the short term, there are constituencies that would benefit from reform in the longer term (e.g. businesses that obtain more reliable supply; households and communities that can be connected because the utility is able to invest in expansion; civil society groups campaigning against corruption in the sector, etc.). There are strong parallels with the literature on fuel subsidy reform, where effective communication about the issue to build a wider constituency of support has been an essential part of successful reforms (Beaton et al. 2013;Kojima et al. 2014).

Conclusion
Our central finding is that, whilst development partners are often fully aware of the political nature of reform of the power sector, they appear to devote much less attention to the analysis of these political constraints and that what analysis there is has little or no influence on the nature of programmes that are put in place. Moreover, many development partners appear to be unaware of the emerging aid modalities for engaging with such dynamic political contexts or, for structural or political reasons, are unable or unwilling to adopt them.
We suggest that, at a minimum, future programme designs should start with a detailed analysis of the underlying motivations of the key actors and institutions in order to identify reform pathways that are politically feasible, rather than just those that are technically desirable. In general, donor interventions in this field should be based around the concept of 'politically feasible and technically sound' rather than the present approach of proposing technically sound solutions with the lack of political feasibility treated as a risk to be mitigated. Our work also suggests that development partners face a difficult choice between two rather different principles. One the one hand, where the political analysis indicates that certain reforms are likely to be impossible, they must find ways to 'work with the grain' , i.e. to identify useful activities that are consistent with the current political equilibrium to maximise the chances of success. On the other hand, sometimes long-term sustainability requires a change in the political equilibrium. This suggests that development partners should consider ways in which they can support legitimate, credible domestic actors to challenge the status quo. Such interventions are common in the wider governance field, but rare in power sector reform.
Finally, our work poses an empirical challenge. Our theory is that new, more politically savvy approaches may help development partner interventions on power sector reform to be more successful. Insofar as such approaches are adopted going forward, there is a need to test this empirically to see whether such programmes are actually more effective and how their success or failure depends on the nature of the political context and the way in which they are implemented.
Notes 5 Interview, September 2016. 6 Interview, September 2016. 7 It is possible that donors may undertake these assessments but not publish them because of their sensitivity. Whilst this may be true, our interviews suggest that formal assessments of this kind are rare and that these issues, if tackled at all, are typically addressed informally through discussions around assumptions and risks. 8 Interview, September 2016. 9 This said, flexibility is not always seen as positive -one experienced donor official commented that 'flexibility in reform tactics is often necessary, whereas flexibility in reform strategy is seldom advisable' (interview, September 2016). 10 Although not impossible -see Ladner (2015) for recent work in this area. 11 Indeed, one of the challenges in building these relationships can be that work in the power sector may be a small component of the overall work of some donors and so resources are devoted to building relationships elsewhere. 12 Interview, September 2016.