The Political Economy of Financial Inclusion: Tailoring Donor Policy to Fit

Financial inclusion has recently become a globally acclaimed policy objective. This provokes the need to review policy in this sector, particularly in light of the tensions that arise between donor approaches founded on market modernism and governments with more activist leanings. This is done here in the context of efforts to move donor development policy beyond ‘best practice’ institutional blue‐prints to those which are ‘good enough’, which seek to understand underlying political economy dynamics in order to find space to engage with governments. In doing so, it is argued that there is scope for ‘working with the grain’ and harnessing the political economy of government policy in order to produce financial inclusion outcomes.


Introduction
Financial inclusion (FI) was adopted by the G20 as a policy goal in 2010 (GPFI, n.d.). It seeks to provide poor people with 'effective' access to credit, savings, payments and insurance services through formal financial institutions (World Bank, 2011) with the expected benefits of increased growth, efficiency and welfare. To achieve this goal at scale involves donors working with governments, as they have a critical role in setting the market conditions in which these formal institutions operate. However, despite three decades of donor-led pro-market reforms, 'activist' intervention policies and tendencies continue in many contexts, and governments frequently have quite different ideas about what their role in FI policy should be. The engagement of official donorsbilateral and multilateralwith recipient governments to develop the financial sector often, therefore, brings the motivations of donors and governments into tension. *Respectively, Senior Lecturer, Centre for Development Studies, University of Bath, Bath, BA2 5JB (s.z.johnson@bath.ac.uk); and Senior Consultant, Oxford Policy Management Group, Oxford. This article is one of the outputs of a two-year collaborative research project investigating financial sector policy development between the University of Bath and Oxford Policy Management. This project operated through a Knowledge Transfer Partnership (KTP) funded jointly by the UK Technology Strategy Board and the Economic and Social Research Council. We acknowledge this support, as well as comments on earlier versions from James Copestake, Ruth Goodwin-Groen, Joanna Ledgerwood and Robert Stone. Similarly, at a broader policy level, the dominant good governance agenda has been identified as failingespecially in Africaas a result of its inability to engage with local political and institutional realities (Grindle, 2007;Grindle, 2011). Moreover, it has been argued that it is necessary to understand the political economy of this failure and challenge donors to develop context specific approaches that 'work with the grain' of these realities (Crook and Booth, 2011;Booth, 2011a). This article therefore brings this challenge to bear on the specific arena of FI policy in order to examine the scope for engaging with governments in the pursuit of FI. Its purpose is to both understand the political economy constraints in the financial sector, and to go beyond this to consider the potential for a re-visioning of FI policies and approaches to engaging with governments. It proposes that 'working with the grain' means taking risks to harness the underlying political economy of government policy in order to effectively produce FI outcomes. In particular, this means finding ways to take advantage of activist policies for the purposes of extending the infrastructure and operations of the financial sector. For example, building on government transfer schemes to further FI aims, or working with state banks and subsidy regimes to harness the geographical and social footprint that they offer.
First, the origins of current FI policy and the tensions arising with recipient governments' own policy are outlined. Second, the political economy of the financial sector is reviewed for insights into the role of the financial sector in the economy as a whole as well as at the level of the institutions of the financial sector itself. Third, recent challenges to donor governance policy are reviewed alongside an outline of the spectrum of options for engagement with which donors are faced, discussing how these strategies have been operationalised in the FI sector to date. Our discussion is focused on Africa, as this currently combines a particularly strong policy focus among donors due to very low levels of formal sector access 1 (Demirg€ uc ß- Kunt and Klapper, 2012;Beck et al., 2011;Honohan and Beck, 2007) with incidence of poor governance. In particular, the financial sector is the site of ongoing and in some cases renewed activism: government owned banks and development finance corporations continue in many countries (SARB, 2014); postal banks are core to country FI strategies (Anson et al., 2013) and interest rate controls on microfinance exist or have been imposed in a number of countries (Mbengue, 2013). 2 Finally, we consider the challenge that an alternative and riskier 'working with the grain' policy strategy means for donor engagement.
2 Financial inclusion policy and the middle way of 'pro-market activism' In the last decade, there has been a move away from targeting poor people through specialist microfinance organisations to building the overall financial market and  (Mbengue, 2013). Zambia recently also introduced ceilings on non-bank financial institution loans. seeking their inclusion in it (World Bank, 2008). This is based on the view that competitive markets are the key means through which innovation and competition will result in falling transactions costs and the scale and sustainability required to overcome the significant levels of exclusion that remain (Porteous, 2004;Honohan and Beck, 2007). It has arisen out of a gradual convergence of the 'financial systems' approach to microfinance (Robinson, 2001) with broader development policy based in neoliberal market ideology (Johnson, 2009). The 'modernist' perspective emphasises the importance of institutional frameworks. It sees the government's primary (or for some commentators, only) role as providing a transparent and accountable system of contractual frameworks and property rights. It is particularly sceptical of relationships between finance and industry or government, which are seen as being likely to block entrepreneurship and innovation (Honohan and Beck, 2007). By contrast earlier activist policies emphasised the need to address market failures and use the financial sector to achieve particular development goals. This approach established specialist development finance institutions (DFIs) and government-owned banks and used lending and subsidies to address failures of the private sector to serve particular constituencies of borrowers, such as agriculture or the rural sector. Failures of many of these policies are now widely recognised, and modernist policy has driven the lifting of interest rate controls, removal of interest rate subsidies and extensive privatisation of state banks.
However, debates over the role of activist approaches in overall economic development have been revisited (Lin, 2012). Development policy now increasingly offers an enhanced role for governmentcoined by de La Torre et al. (2007) in relation to the financial sector as 'pro-market activism' (see Figure 1 for a graphical stylisation of these three positions). This still places primacy on market mechanisms  Figure 1: Perspectives on government roles in the financial sector Source: Adapted from Napier (2008) in allocating resources (towards the left of Figure 1), but recognises that the state may be best placed to address important market development constraints. The leading donor consortium for pro-poor financial provision, the Consultative Group to Assist the Poor (CGAP), reflects this thinking on the role of government (CGAP, 2006;and Ehrbect et al., 2012). First, as a rule maker in setting up legal and regulatory frameworks to ensure prudential and consumer protection while creating space (and competition dynamics) for transformative models to broaden the financial system by, for example, allowing banks to use agents in rural areas. Second, as a promoter of FI, governments must provide essential infrastructure such as ID cards (to meet 'Know Your Customer' requirements), payment systems, land-titling (to allow easier use of collateral) and public goods such as feeder roads. Governments can also drive transaction volumes by putting their own social payments through the financial system (for example, using m-banking to deliver pensions). Third, it can advocate for FI through its own strategies and policies (for example, by setting national FI and microfinance policies) and co-ordination with the private sector through mechanisms that promote information sharing. There is less consensus on the use of moral suasion, such as pressuring banks to establish basic accounts for the poor.

MODERNISM
There is also broad consensus within the donor community about what the government should not do, specifically around resisting the direct provision of financial services through government programmes or state banks, which are seen as the least efficient approach to sustainable access (CGAP, 2009). However, middleway approaches between modernism and activism do argue a public choice rationale for some short-term subsidies, for example incentivising the downscaling of banks into rural areas given the potential positive learning externalities.
While this recognition of the potential role for government involvement is an important opening up of the middle ground, it presents new challenges. First, donors are more comfortable with activist approaches when they see them as rooted in strong governance structures, but few countries are seen as actually having 'adequate governance', especially in Africa (Honohan and Beck, 2007: 12). Second, governments themselves do not necessarily come to this middle way as a result of compromise, and continue to exert strong activist leanings, in part to fulfil objectives beyond a narrow FI lens. Indeed, there has been a resurgence of activism in general in development strategies in the post-economic crisis context (Birdsall and Fukuyama, 2011). This has included, for example, directed lending in domestic banking sectors, as well as providing financial services through government schemes in both developed and developing countries.
As donor focus on developing financial systems has gained ground, 3 this leads to a tension for donors working in these contexts. It raises the question of how donor 3. One particular donor approach to financial market development analytically divides the market between different levels; the rules of the game (for instance, policy and regulation), supporting market infrastructure (such as training institutions), and suppliers (such as banks) and consumers see (DFID/ SDC, 2009;Ferrand et al., 2004;Ledgerwood, 2013). This article focuses on the wider political economy dynamics that underpin the whole financial system and influence overall approaches to engagement rather than the micro political economy dynamics of engaging at each level.
policymakers and practitioners recognise and analyse the potential and scope for engagement with government. As a World Bank report into Africa's financial sector states, this 'requires looking beyond the dichotomy of modernist and activist approaches towards an approach that recognises that all financial sector policy is local' (Beck et al., 2011: 227). But as Zhang points out, 'the newly found state-friendly discourse has been framed in narrow economic terms . . . it neglects the political process through which state actors create and regulate the financial system. Despite its emphasis on governance, the paradigm seldom confronts the issue of government and the politics that underlie it' (Zhang, 2006: 170-1). The starting point is therefore to understand the political economy constraints to achieving market-led FI. The next section examines the two core dimensions of political economy in the financial sector: first, how it is influenced by the wider system of rent allocation and its role in the economy as a whole; and, second, how the rules within it are influenced by local political and social dynamics, and what this means for policy processes. 4 3 The political economy of financial inclusion policy

Structural factors: the role of the financial sector in development
A broad political economy perspective considers how the financial sector fits into the overall structure of politics and the economy. It considers how individuals and groups compete for power by investing in activities that uphold their survival (Whitfield and Therkildsen, 2011). A rational choice perspective along these lines is illustrated by Bates (2014), who describes how urban bias policies of African governments in the 1980s responded to coalitions of urban elites and farmers, with the financial sector being structured to channel rents to these groups, for example, through subsidised credit. This fits closely with a modernist perspective that argues that the financial sector is distorted by groups for personal gain, and, therefore, the influence of such groups (i.e. the state) should be limited. Another central tenet is that bargaining and contestation between groups seeking survival dictates policy outcomes including how rents are allocated. A rent distribution perspective recognises the use of redistribution to provide stability along with the deep-rooted social structures underpinning it. North et al. (2007: 1) argue, 'political elites divide up control of the economy, each getting some share of the rents . . . adequate stability of the rents and thus of the social order requires limiting access and competition-hence a social order with a fundamentally different logic'.
In African contexts in particular, such systems of resource allocation have been analysed as neo-patrimonial where authority structures are based on personal relations and tied to the distribution of resources. The provision of private goods, such as subsidised credit or debt write-offs, is a more effective tool for ensuring support than relying on the public provision of services or redistribution of domestic revenue (Khan, 2005). This suggests that building markets to produce competition and innovation faces significant constraints and the scope for coalition building is complex. The state is the creator of market structures through the extent and degree of regulation, and increased competition between actors and interest groups may be viewed by elites as a source of instability and threat to the social order. Bruhn et al. (2013), for example, have shown that countries with a high degree of bank concentration and limited contestability are less likely to adopt private credit registries as incumbent banks stand to lose their rents from sharing information.
Moreover, such patterns can often perpetuate themselves for a long time as vested interests and social norms mould the institutional framework to continue the status quo. For example, a 'rights' perspective has emerged in some countries regarding subsidisation of credit to the rural poor. A consequence of this, as in India, is that, even when formal rules create a low regulatory burden for banks to commit to outreach, they are discouraged by the belief that broader obligations will be placed on them (FICCI/M-Cril, 2011). While political institutions may help limit the worst kinds of capture (Haber et al., 2003), the links between democratic accountability and FI are not clear-cut. There is the classic collective action problem of numerous, but diffuse, winners from inclusion (namely poor people) having sufficient voice to bargain against concentrated interest groups, for example incumbent banks who are comfortable serving high income (often government) clients and do not see a 'market' of poor people to be served at sufficiently low cost and risk. Similarly, regulators tend to prioritise the safety rather than breadth of the system over the costs and risks associated with expansion. However, as expectations for FI become more widespread, those hitherto underserved in the emerging middle classes, as well as small and medium enterprises, may become a more powerful interest group for inclusion themselves (Rajan, 2006, cited in Demirg€ uc ß-Kunt andLevine, 2008). Indeed, Degryse et al. (2013) found that elites pursue economic opportunities by promoting capital raised on stock markets. In contrast, broader political participation empowers a middle class with different preferences, in which banks are favoured given that they limit the rights of a narrow set of elites. On the other hand is the concern that financial intermediaries become a key vehicle to buy support in youthful democratic regimes with short-term election cycles, such as in sub-Saharan Africa (Girma andShortland, 2005, cited in Milo, 2007).
The prevailing view therefore is that these neo-patrimonial features constrain the potential for reform in the financial sector in Africa, especially when exacerbated by ethnic fractionalisation and mineral dependence (Beck et al., 2011). However, recent analysis argues that countries in Africa that have managed to achieve development outcomes have demonstrated the ability to 'coordinate between "good" and "bad" rents by balancing the demands of politically strategic groups for handouts, . . . with the hand-ups required by genuine entrepreneurs' (Kelsall, 2011: 78;. Relations between state actors and the private sector provide the stability to stimulate investment, with rents transferred between competing elites to help stabilise the system for economic development to occur (Khan, 2005). The scope for this depends on: first, whether elites have long-term time horizons emanating from the strength of their coalitions vis-a-vis opponents (or a lack of groups to buy-off); second, the nature of political competition which affects these time horizons; third, the extent to which rent management is centralised (often associated with personalised leadership); fourth, incentives to broaden the tax base and, therefore, private sector (or limited access to easy rents such as natural resources); finally, a disciplined and competent bureaucracy (Kelsall, 2011;. Viewed through a narrower financial lens, how rents (and government intervention) are managed directs attention to the relationships underlying the structure and operations of the financial sector as powerful interest groups are able to shape the speed and nature of change (Rajan and Zingales, 2003). Variations in the state's relations with private capital are often clearly reflected in the structure of the banking sector and work through into the nature of banking sector reform and its outcomes, reproducing political economy structures, and often diminishing the impact of banking reforms (Boone, 2005). Changes in the organisation and strength of such groups therefore influence how policy and regulation is set. Even if 'best practice' rules are put in place, their implementation may be influenced by political forces, for example through inside pressure towards regulatory forbearance (Brownbridge et al., 2002).
FI objectives have added an extra layer of complexity to this interest group perspective with new trade-offs, for example, between growth and stability objectives for elites, the private sector and regulators (Porteous, 2006). For example, recent developments in technology and the potential for mobile banking to address inclusion have led to the respective power and interest of banks vis-a-vis telecommunications companies and their interactions with regulators becoming more prominent (Rhyne, 2014). Kenya has seen these core political economy characteristics play out as banks sought to protect their incumbent positions against the entry of Safaricom's M-Pesa mobile money transfer service. Now, as market leader, Safaricom is, in turn, seeking to stop Equity Bank moving into its market by blocking its use of an innovative mobile SIM card technology (Kimenyi, 2014).
In this regard, two main structural factors in the relationship between the state and the private sectorand especially bankshave been highlighted as affecting the efficacy of an activist approach (Dafe, 2011). First, is the degree of autonomy of the private sector and specifically of banks from government, and, second, is the degree of concentration of the private sector and specifically banks. Autonomy of the private sector from the government leads to the state having greater structural dependence on it, and it is argued that it increases the chances of coalitions across the divide to enhance financial sector development, while low concentration reduces the potential for their political capture. This reflects a fundamental tension that governments regulate, supervise and discipline financial intermediaries, but also rely on them for tax revenue and borrowing in addition to their being a channel for ensuring votes (through pressure to lend) and political support (through contributions to party financing) (Calamoris andHaber, 2011, cited in Lin, 2012). Where governments have significant political autonomy due to dependence on alternative sources of revenue (such as mineral income rather than tax revenue) this can prevent moves to develop coalitions in support of market development. Moreover, the extent of autonomy between groups making up coalitions may be difficult to establish (particularly by external actors) given that inter-relations between the state and private sector frequently operate through the influence of elites and their opaque social networks.

Institutions and financial inclusion
We now consider the specific dynamics of institutional reform in the financial sector. A focus on institutions has become central to recent understanding of financial system development : that is, how rulesboth formal rules such as policies, laws, regulations and informal norms, conventions, conduct and traditions -influence performance. A strong focus in donor practice has been the creation or adjustment of formal rules and a preoccupation with transplanting 'best practice' regulatory forms into developing countries aimed at lowering the risks and costs to market development. However, these often fail precisely because they take insufficient account of the underlying social, cultural and political institutionsboth formal and informal (Boettke et al., 2008;Chang, 2006;Rodrik, 2007;. Research by Lewis and Stein (2002) showed that regulatory change bringing about financial liberalisation in Nigeria in the mid-1980s, opened up new areas for rent seeking as banks had privileged access to foreign exchange auctions. Similarly a new FI or microfinance policy can be relatively easy to create, but often does not 'stick' and is left unimplemented. Goodwin-Groen (2012) shows how the Microfinance Act in Uganda was developed through a careful process of consultation and built on indigenous norms of small-scale saving by legitimising deposit taking and allowing their intermediation. She then analyses how the development of the regulations by the Central Bank to go with the Act directly undermined it by following more risk averse global 'best practice' that only allowed the use of savings as cash collateral for loans.
Additionally, institutional analysis highlights 'path-dependency' in institutional change. For example, the 'law and finance' perspective argues that the level of legal protection and quality of law enforcement is determined by a country's legal system origins, with common law systems providing better protection for investors through stronger contract and property rights than civil law legal systems (La Porta et al., 1998). Importantly, in terms of the dynamic landscape of FI it is argued that common law legal systems adapt more easily to changing economic and commercial needs so providing greater flexibility. This has been argued to be a constraint to the development of mobile banking in Latin America, in contrast to South Africa and Kenya (BFA, 2009). But on the other hand, civil law systems may lead to more rapid development once laws are adopted as, for example, in the case of correspondent banking in Mexico (ibid.). Factors beyond the financial sector, such as oil dependency, aid dependency and historical private sector influences, can also interact with it to determine, for example, how transformative Central Banks are in respect to FI (Dafe, 2013).
Formal rules and structures are underpinned by informal norms, conventions and practices along with mental models, ideologies and beliefs (Williamson, 2000;North, 1990). In the financial sector, trust and social networks are considered particularly important in reducing the costs of screening and enforcement where more formal rules such as debt creditor rights, property rights and collateral laws are absent (Armend ariz and Morduch, 2005). More broadly, relationship lending is strong throughout sub-Saharan Africa, with reputational networks built around culturally connected groups, leading Honohan and Beck (2007: 149) to ask, in relation to SME lending, whether 'by limiting themselves to financing the large and collateralized borrowers [and ignoring such networks], could banks be missing out on a lucrative market?'. This may be particularly important where banking systems dominate (over capital markets) and little credit information exists, allowing the harnessing of local 'soft' information through long-term relationships (Lin, 2012). However, social institutions such as those of gender, ethnicity and caste, also create systematic reasons for exclusion and the fragmentation and social regulation of markets through the discriminatory impact they have (Johnson, 2013).
A caveat to the above emphasis on historic and social dynamics is that that they leave little room for understanding how individual agents or groups can shape, as well as be constrained by, their institutional context (North, 1990). Particular stakeholders may have a significant influence in pushing for, or subverting, specific reforms, with history showing that 'champions', have been successful in creating change. For example, progress in FI in Pakistan is argued to be partly a result of the then Governor of the Central Bank successfully advocating for a new FI unit, thus bypassing more conservative stakeholders in the government. These micropolitical dynamics, are well captured elsewhere in the political economy literature and to some extent apply to most sectors (Grindle and Thomas, 1991;Copestake and Williams, 2012; and are therefore not a focus here.

A framework for engagement
The need to understand underlying political economy dynamics of policy-making has been increasingly recognised by donors and is evident from the proliferation of approaches to analysing structures and incentives (DFID, 2009;World Bank, 2009). These approaches are also intended to offer more positive guidance through a focus on 'what works' rather than promoting normative ideals of good governance (Duncan and Williams, 2012). But, a concern has remained, especially with respect to Africa, that underlying political economy dynamics make close engagement with government particularly problematic.
However, Booth (2009: 3) challenges this arguing that 'we should at least consider the possibility that there are forms of the neo-patrimonial state that can combine patronage politics with quite a high degree of developmental effectiveness'. This has led to the idea of a spectrum of engagement with 'Big G' 5 governance reforms at one end which are constantly challenging for donors as they require working politically, identifying 'champions', and often operating in direct contrast to vested interests. Consequently, many donor approaches have tended to revert to the other end of the spectrum and operate within a more limited 'feasible reform space'. This involves a number of options, which ensure that engagement does not directly confront powerful interests. For example, working around government to deliver services through 5. This entails changing significant national-level institutions such as elected legislatures, the judiciary, centralised auditing authorities, ombudsmen, a free and vigorous media, and so on. independent or quasi-autonomous entities; or, slightly more ambitiously, engaging with 'small g' reforms which seek to build participation in the oversight of public services by interested stakeholders who can build some pressure from the bottom up (such as seeking to strengthen civil society). Or, finally, there may be scope at the micro level to support narrow policy reforms that operate in 'islands of excellence' (Levy, 2010). Booth (2011b) adds to the spectrum to incorporate more dynamic possibilities. He argues that stakeholder positions are not fixed and that there may be collective action problems such that there is scope for making 'intelligent interventions,' for example, to support coalition building between parts of government and the private sector. Such opportunities are rare given the difficulty of collective action to overcome the status quo, but they may evolve from short-or long-term shifts in power relationships between public and private actors, requiring practitioners to respond flexibly to exploit opportunities for change. Going beyond this, there is scope for what he calls 'working with the grain'. This recognises that there may be economic, political or social institutional factors that are not conducive to 'Big G' reforms, but that there are existing arrangements that can be built on incrementally to work towards desired objectives.
In bringing these perspectives to bear on FI policy, we replace the objective of 'Big G' reform with that of 'Big M' market reform to bring about broader and deeper financial systems. An important difference in the case of FI is that the 'means' of the market and the 'end' of inclusion have greater scope for separability than in the case of governance, in which 'good governance' is both the end and the means. Indeed, inclusion may be led by a government initiative, for example, through the delivery of welfare payments. Hence, the goal of inclusion may have more routes to its achievement than those that are solely market based. If the objective of achieving the end of 'Big M' markets is not sacrosanct, there are expanded opportunities for engagement by donors to move towards FI policy that fits.
Having identified the underlying political economy constraints for FI through market-based financial sector development, we now consider the options for engagement in the context of pro-market activist approaches and the spectrum of strategies for governance reforms presented in Figure 2. In doing so, we show that donor policyoriginating in microfinancehas evolved along this spectrum over the last 30 years, towards the current need to consider how to engage on the basis of more pro-market activist approaches. The following section will then identify the potential for more ambitious and challenging approaches to FI policy that entail 'working with the grain' and the considerations that underlie these.

Risk-mitigation in engaging with government
Modernist policy initially sought what we term 'Big M' market reforms through liberalisation and privatisation. These pursued a normative view of the governance arrangements needed for market development characterised by strong rule-based systems of property rights, robust competitive practices, well-functioning credit registries and so on. As highlighted in section 3, such reforms are often not feasible because they encounter alternative modes of resource allocation, vested interests and institutional or policy constraints. Despite this, 'Big M' objectives involving 'best practice' reforms still underpin much technical donor assistance and frequently fail to 'stick' as illustrated above. For example, a review of a number of government national microfinance polices concluded that key weaknesses emerged from them being donor-driven (and donor-funded) with a reliance on 'templates' that fail to reflect local realities (Duflos and Glisovic-M ezi eres, 2008).
Despite the achievement of some 'Big M' reforms (Demirg€ uc ß-Kunt et al., 2008), many difficulties and failures emerged leading policy to bounce to the opposite end of the spectrum: either complete disengagement from the financial sector or working around government. In the wake of structural adjustment policy failures, donors supported building independent microfinance institutions initially based on NGO development programmes. Bangladesh was a key example (Devine, 1999;Wood, 1997) setting off a major wave of innovation and development in microfinance practice. Similarly, but more recently, a number of major donors disengaged from supporting the microfinance sector in Uganda, partly in the context of strong government activist policy towards the use of the savings and credit co-operative sector for FI (Goodwin-Groen, 2012).
Donor strategies then turned to developing financially sustainable microfinance sectors involving cost covering interest rates and eventually direct support to private sector financial institutions (including banks). While initially enabling donors to work around the state, this approach has itself evolved to require ever-greater engagement with governmentfor example, in the need to develop microfinance legislation to adequately regulate and supervise an emerging sector. As section 3 illustrates, this situates practitioners in a messier position where working with coalitions of governments and private actors may be risky. This has led to some more granular options in mitigating these risks through identifying a feasible, albeit more limited reform space, attempting to build credibility along the way by not fully confronting activist models.
These tend to focus on technocratic changes often by Central Banks and be less central to short-term political economy dynamics. 6 For example, German technical co-operation works with regulators to assist them to conform to standards such as those of Basel and the Financial Action Task Force. 7 This is not to say that Central Banks, are not subject to political economy dynamics, with varied degrees of independence from other ministries and regulation being subject to coalitions of interests. But they may also exhibit characteristics similar to Levy's 'islands of excellence' (2010). As Dafe (2013) notes, the Central Bank of Nigeria was one of the main reformers in the last few years, with the Governor of the time winning various awards from banking consolidation and branchless banking strategies, despite the country as a whole performing poorly on governance indicators.
At the level of supply, donors are often prepared to support subsidies to the financial sector that offer market inducing modalities outside of direct government participation such as challenge funds (O'Riordan et al., 2013) or special purpose 6. In the longer-run these may, of course, strengthen Central Banks in relation to other government actors to protecting the stability of the financial system in the face of proposed activist reforms. 7. For a description of their strategy in working with Central Banks, http://www2.gtz.de/dokumente/bib-2011/giz2011-0057en-financial-systems.pdf vehicles with separate organisational arrangements. 8 Similarly, credit guarantee schemes are perhaps the archetypal pro-market activist intervention in that they marry fiscal resources to overcome market externalities (for example, asymmetric information in regard to SMEs previous performance) with market discipline. By focusing on technocratic reforms to the wider enabling environment and on new approaches to 'smart' subsidies at the micro level, current donor strategies therefore tend towards the less risky end of the spectrum of Figure 2 that avoids direct engagement with underlying political economy dynamics. In these cases scale is compromised and opportunities to be more responsive may be missed. This, it could be argued, characterises much of where donor strategies currently operate.

'Working with the grain' and embracing risk
For donors to embrace risk and go beyond these approaches to 'work with the grain' is challenging. Information on political economy dynamics is nearly always incomplete and incentives for more recognisable and risk-averse strategies persist. It therefore appears necessary for donors to 'seek out incremental reform options that are feasible given political economy realities'; and to see which existing institutions may be strengthened (Beck et al., 2011: 232). This section discusses what more proactive engagement would look like.
Facilitating collective action may occur through stakeholder analysis to identify coalitions who are seeking, or who will back, reform. The role of the donor is to inject information and evidence to key actors, through highlighting key demonstration models or regulatory approaches in different countries, for example. This may involve establishing co-ordination mechanisms between like-minded groups such as the Ministry of Finance, the private sector and other members of the development community such as the FI working groups or 'round tables' used in Latin America, or through long-term conveners such as the financial sector deepening (FSD) trusts in Africa. This co-ordination function is crucial given the differing capacities, mandates and objectives of those involved in FI (BFA, 2009). For example, there are often trade-offs between a Ministry of Finance seeking medium-term growth with that of other ministries, who may primarily be concerned with shorter-term livelihood or welfare issues, and Central Banks interested in the stability of the system. It may also require identifying and engaging with opposition groups who need convincing of the benefits of a more market-oriented approach, including perhaps how it may fulfil, or not interfere with, certain rent allocation functions, at least in the short term.
A step further involves exploiting change opportunities with more dynamic approaches to market development. FSDs established in eight African countries have been established to work directly with private sector institutions as well as with government (who sit on their Boards) to address constraints to FI. In a way, therefore, they operate simultaneously at different points on this spectrum in 8. For example, there are a number of financial sector deepening trusts, predominately funded by the UK Government throughout Africa, which operate independently, albeit with government involvement on governing boards. Figure 2 by seeking to build retail capacity and competition in the market while developing support services (such as training institutions or market information) and addressing the institutional regulatory and supervisory environment. By building a presence as a supportive and engaged player in the sector over the longer term, this has enabled trusts such as FSD Kenya to assist at strategic moments by providing inputs into decision makingfor example, in providing technical advice to the Central Bank to allow the development of Safaricom's M-Pesa product and approach it with a potentially more risky light regulatory touch (Stone et al., 2010), so prioritising the potential for innovation and outreach over concerns about stability. An example of exploiting change opportunities to work with the explicit objectives of activism while developing a market focus is the role of the donor consortium FinMark Trust in South Africa. As part of its post-apartheid black empowerment agenda, the government of South Africa was intent on regulating the banking sector to provide a simple and cheap basic bank account. After extensive discussions, it was agreed that the banks would provide such an account (called 'Mzansi') voluntarily. The donor-funded FinMark Trust responded to this change in the political economy dynamics between the state and banks by helping to supply market information to the retail sector in order to monitor the take-up of the supply response (Porteous, 2004). Through the engagement with Mzansi and the market information that gradually became available, banks have developed better products and services and learned about the needs of this market segment and the challenges to be overcome in effective provision (Mittner, 2010).
A further step on the spectrum towards 'working with the grain' more directly considers what local economic, political or social structures and incentives could be worked with to achieve FI. This recognises that FI as an end can be achieved through multiple routes and not solely through competitive market provision based on strong rule-based frameworks. Dafe (2011) to some extent moves in this direction with the suggestion that an optimum mix of banking autonomy and concentration may create governance dynamics that are more conducive to activist approaches. Within current pro-market activism this tends to include still fairly risk-averse intervention, for example timebound subsidy or risk sharing through partial credit guarantees These still fit, therefore, with the desire for 'Big M' solutions. More interesting is Dafe's suggestion of leveraging social networks and local networks through building on relationship lending in banking, although again these are viewed as a short-term 'second best' option (until the first best presents itself) rather than a long-term strategy for donor engagement. As Booth argues, 'pre-existing institutions need to be treated as a potential resource for reforms that improve development outcomes, not swept [them] aside regardless of their ability to contribute' (Booth, 2011b: 2).
Moving across the central space of Figure 2 towards FI policy that fits involves recognising that the state sector may have particular benefits to offer the FI agenda. This may include working directly with state banks or directed credit schemes to improve their performance, while bearing in mind that what constitutes 'success' is likely to go beyond the crowding in of private actors and include objectives of targeting specific groups consistent with a government's wider development strategy. This is where policy starts to take significant risks.
Referring only to East Africa, in recent years governments have become more activist, using state resources to expand semi-formal financial services. In Uganda, the Bona Bagagawale scheme was set up in 2007, including the re-establishment of co-operatives to provide government funds and loans (Goodwin-Groen, 2012), and it was in the wake of this that donors withdrew their support from the sector (ibid.). In Ethiopia, the government has used the extensive co-operative sector to pursue its wider rural development policies. Even in Kenya, where the development of the financial sector is seen as exemplary in the region, the government established a Youth Enterprise Fund in 2006 with a majority of the funds allocated to financial institutions with a capped interest rate of 8% per annum, while a minority was directly allocated by district committees at a zero interest rate and a one-off management fee of 5%. These are cases where the extension of subsidies involves fairly direct attempts to establish the legitimacy of incumbent politicians, while frequently appealing to discourses about marginalised groups of rural, young or women voters.
However, the donors' response in Uganda contrasts with that in Rwanda, where the government set up the Ubudehe Credit Scheme (UCS) in 2009. This was a directed credit programme, as part of a broader social protection scheme at a capped interest rate of 2% per annum, alongside another programme mandating the establishment of regional savings and credit co-operatives, known as SACCOs. Donors were similarly concerned about the effect on the financial market, but recognising the strength of the government's intention, they worked to improve the scheme's performance and connect it to the financial system, within a wider process of reform. For example, Ansoms (2008) has situated this top-down modernisation process within the post conflict desire for state building. So, despite an evaluation of UCS showing only 50% repayment rates, and the potential to undermine commercial financial services (through the provision of below market interest loans), the programme was generally considered a success in social protection terms, with strong government ownership despite external funding (Arora et al., 2012).
This example of building on community-development policy that integrates financial services into wider intervention strategies to deliver outreach, even though these may have political objectives and be poorly implemented, is also evident in the Indonesian National Program for Community Empowerment (PNPM Mandiri). This programme has reached 1.5 million borrowers through community revolving loan funds, again with poor returns when seen through a narrow financial lens (Bebbington et al., 2006;Micro-Credit Ratings International Limited, 2012). Indeed, it is perhaps ironic that much NGO adoption of group-based microfinance was a result of the need to enable poor people to organise collectively for a range of purposes integrated into other development programmesa perspective that the narrower financial systems approach now rarely engages with.
An even more explicit approach would be to build on broader activist approaches to FI, not only because they are the politically feasible option, but also because there might be additional advantages when implemented appropriately. For example, the presence in rural areas of government institutions provides important infrastructure, and government may be more expected and accepted as a local player. A recent systematic study into formal access, albeit from a small selection of studies focusing on India, concluded that, 'state-led expansion of the banking sector in rural areas can reduce rural poverty, increase rural wages and increase agricultural investment' (Pande et al., 2012: 1). The Bill and Melinda Gates Foundation aimed to leverage the state's footprint by funding a large programme through working with state postal banks to modernise their products and services and make savings accounts available to low-income and rural populations. A further example is the use of external management contracts used with Tanzanian and Mongolian state-owned banks to turn their operations around as a strategy to 'work with rather than around' the government and maintain the strengths of the networks that already existed (Dressen et al., 2002) and which were later sold to private overseas investors.
It is possible to envisage more radical options. Working with cash-transfer schemes to deliver FI offers one route that is now accepted by donors as a means to develop financial service infrastructure in remote and rural areas (OPM, 2008). While this may initially offer payment facilities rather than direct access to savings accounts and credit, the development of digital technologies and the rapidly expanding use of mobile money, especially in Africa, is rapidly breaking down the barriers between services, and is now being seen as offering the potential for charting routes to inclusion into the full range of services. In Zambia, input and post-harvest finance, to which over 60% of the Ministry of Agriculture's funding is allocated (Mason et al., 2011), is consistently derided by donors for its deep inefficiencies (such as a lack of targeting, price distortion in input and output markets and overall fiscal cost). This again ignores the wider importance of this allocation of resources in connecting the state to rural elites and voters and across ethnic lines. Such schemes can be envisaged with FI objectives in mind. They might include, for example, delivering and managing these input subsidies through evouchers delivered to mobile phones (SciDevNet, 2010), which further engage poor people with the newly developing infrastructures of the financial system and the potential to uncover new routes to FI.
Building on this, donors need to remain open-minded as to the potential for providing longer-term subsidies that are not strictly 'smart'or designed purely to right the marketbut intended to create new opportunities for the entry of new private sector providers. Indeed, from an historical perspective, the current vision of FI has in part been forged as a result of the demonstration effects that the microfinance sector, which relied on long-term subsidies for many years, has provided to the financial sector more broadly (Roodman, 2012). In particular, there is now scope for a new wave of innovation using the infrastructure of mobile technologies to catalyse approaches which harness the activist leanings of governments in new ways.
It is often also the case that interventions operate in an ongoing way and with smaller spaces for opportunities where decisions as to feasibility are made within the framework of wider programmes. In other words, choices of when to 'work with the grain' do not only occur at this macro political economy level. Donor engagement often ebbs and flows with donor assistance trends as well as the difficulties of operating in a particular sector, therefore decisions about programme design and options operate within cycles of engagement on a more micro basis. Consequently, the wider political economy focus of this article will need to be augmented with assessing space for engagement through informed and reflexive understanding of political economy that cuts through this to the micro-politics affecting particular programmes (Copestake and Williams, 2012).

Conclusion
This article has explored the tension that exists between FI policy founded in market modernism, and the need for donors to engage with governments with more activist leanings and weaker governance capabilities in order to achieve the new global policy objective of financial inclusion. It has reviewed the insights that a political economy perspective can offer in revealing the underlying dynamics constraining policy development in this sector.
Emerging thinking about donor policy engagement in the governance field has been applied to suggest that this also offers insight into how donors can 'work with the grain' of activist government approaches in the FI field to achieve developmental outcomes rather than to assume they cannot. We further argue that donors may need to be prepared to take greater risks once they engage with developing this deeper understanding. Moreover, the appetite for such risks can sometimes grow from the successful implementation of smaller-scale initiatives within 'small M' reforms. Additionally, thought can be given to adopting approaches that build on local social institutions, which may abandon what are externally perceived as 'best practice' approaches, but solve particular problems of financial services provision.
This implies that donor practitioners should 'pause' before immediately attempting to strengthen more formal institutional processes (Rodrik, 2008) or adopt best practice institutional change, as this is unlikely to 'stick' if not adequately embedded into broader political and social relationships. A solid understanding of local institutions and incentives is a prerequisite for sensitive and successful engagement, and developing the space for inclusion opportunities requires reflexivity on the part of the practitionerbe they development agency, consultant or government official.