Urban Governance and Service Delivery in African Cities: The Role of Politics and Policies

This is an introduction to a special issue sponsored by UNU‐WIDER focusing on key governance challenges related to addressing gaps in urban service delivery in sub‐Saharan Africa. First, due to decentralisation policies in much of Africa, the provision of services is often transferred to sub‐national authorities. But complex layers of administration and high levels of poverty deprive local governments of adequate resources. Secondly, opposition parties now control a number of African cities, a situation known as ‘vertically‐divided authority’. Consequently, central governments do not always have an incentive to help municipal governments improve their performance. Case studies of Senegal, South Africa and Uganda highlight how and when such dynamics become problematic for the delivery of urban services, and offer important implications for the donor community.


Introduction
Urbanisation represents one of the main demographic transformations confronting sub-Saharan Africa today, with attendant implications for the region's long-term development trajectory. Although debate exists over whether Africa can be accurately classified as the fastest urbanising region of the world (see Potts, 2009;Satterthwhaite, 2010), it is undoubtedly true that many more Africans now live in urban areas than at any other point in history. The World Bank predicts that by 2030 Africa will finally be an urban continent, with more than 50% of its population living in cities (Kessides, 2006). Even by 2020, cities such as Dar es Salaam, Kinshasa, Lagos and Nairobi are expected to almost double in size from their 2005 levels (UN-Habitat, 2010). Increasingly, much of this urbanisation is fuelled more by natural population growth in cities or urban-to-urban, rather than rural-tourban, migration (Freund, 2007;Tacoli, 2001). 1 While African cities are undoubtedly vibrant locales that generate approximately 60% of the region's economic growth (Sisk, 2004), this increase in demographic pressure places These two governance concerns present important implications for the international donor community. Because of low resource bases, many municipalities rely on donor financing to fund their service-delivery needs, even though donors' support and modes of engaging with the urban sector have changed dramatically over the last few decades. In a number of African countries, however, only the central government can directly negotiate with donors. This places donors, which are often self-proclaimed apolitical entities, in a very difficult position in situations of vertically-divided authority. We therefore examine how donors have managed urban and decentralisation projects in such contested terrains.
To address the above themes in more detail, this special issue contains four studies. In all of them, the authors engaged in primary fieldwork and conducted interviews with a variety of knowledgeable stakeholders, including officials from central government ministries, municipal authorities, and members of the donor community. The first study, by Richard Stren, helps contextualise the demographic, economic and institutional elements of African urbanisation and, with a special focus on the World Bank, highlights how donor priorities and modes of engagement have changed over time.
In the subsequent three cases, the authors focus specifically on the challenges of vertically-divided authority in three of Africa's traditionally more decentralised countries: Senegal, Uganda and South Africa. All three demonstrate considerable variation in terms of the level of urbanisation, degree of overall economic development, party system, electoral institutions for municipal office, and decentralisation policies. Gina Lambright highlights the worrying re-centralisation in Kampala, Uganda where President Yoweri Museveni began removing a number of political and fiscal powers from the city's authorities who belonged to the opposition Democratic Party (DP). Danielle Resnick examines the case of Dakar, Senegal under the former president Abdoulaye Wade whose regime gradually lost control of the capital and other major cities to the Benno Siggil Senegal (BSS) opposition coalition. In both countries, the authors highlight that this has had negative implications for addressing issues such as urban flooding, trash collection, and infrastructure. By contrast, in Cape Town, South Africa, which is controlled by the Democratic Alliance (DA), Robert Cameron shows that service delivery has not been as vulnerable to inter-partisan struggles owing to the strong entrenchment of decentralisation laws in the 1994 Constitution and a variety of performance incentives at the national level. In fact, service delivery in Cape Town outperforms that in Johannesburg, which is controlled by the ruling African National Congress (ANC). Nevertheless, Cameron documents a variety of attempts by the ANC to undermine the political authority of Cape Town's councillors.
In order to elaborate on the findings from this collection, the following Section of this article discusses the trajectory of African decentralisation. Drawing on literature and examples from both Africa and other regions of the developing world, the particular difficulties of vertically-divided contexts and the potential for 'strategies of subversion' by central governments are then detailed. Subsequently, detailed examples of such strategies are summarised from some of the collection's case studies. The implications of such contexts for the donor community are then discussed before offering some final contributions of the collective research.
© The Author 2014. Development Policy Review © 2014 Overseas Development Institute. This is an open-access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Development Policy Review 32 (S1) 2 Africa's fitful decentralisation trajectory There are various forms of decentralisation, including deconcentration, delegation and devolution. While deconcentration involves the central government re-distributing some decision-making authority from the central to the field offices of national ministries, delegation entails the transfer of responsibility for certain public functions to sub-national governments. By contrast, devolution requires not only transferring responsibility but also authority over decision-making and accountability to autonomous and legally constituted sub-national governments (see Cheema and Rondinelli, 2007;Kathyola and Job, 2011). At the same time, there are political, administrative, and fiscal elements of decentralisation. The administrative element refers to the transfer of responsibilities for the delivery of certain services to sub-national governments, while fiscal decentralisation requires that local governments possess a high level of discretion over the formulation and allocation of budgets. Political decentralisation refers to the regular election of local leaders in secret and free and fair ballots. For the purposes of this special issue, decentralisation is equated with devolution, precisely because devolution cannot exist in the absence of political decentralisation (see USAID, 2009).
For two decades, decentralisation has been at least rhetorically embraced as a priority for Africa's development agenda. In fact, an analysis of all the current Poverty Reduction Strategy Papers (PRSPs) for Africa reveals that 25 countries list decentralisation as a key objective, typically categorised as part of their strategy for achieving 'good governance'. Improved governance has long constituted one of the two main justifications for decentralisation. Specifically, by bringing decisions closer to the people, decentralisation might empower local citizens and increase both upward and downward accountability between elected decision-makers and local citizens (see for example, Brinkerhoff, 2010;Rondinelli et al., 1983). Improved service delivery represents a second justification for decentralisation. In this perspective, decentralisation can facilitate the ability of local governments to tailor services to local preferences and needs as well as encouraging competition among local governments to retain residents by providing the most efficient services (Oates, 1972;Tiebout, 1956). In reality, the ability of decentralisation to always achieve both goals has been questioned based on both theoretical (for example, Prud'homme, 1995;Treisman, 2007) and empirical grounds (see Ahmad and Brosio, 2009;Ahmed et al., 2005).
In the African context, much of the impetus for decentralisation has been primarily political, and the election of sub-national authorities has been sequenced before other elements of the decentralisation process (Dillinger, 1994;Oxhorn, 2004). In South Africa, decentralisation was a way of maintaining stability at the time of negotiated transition away from apartheid in 1994. The apartheid-era National Party demanded that the ANC adopt a policy to promote strong provinces with guaranteed access to revenues to ensure that white Afrikaners would not be excluded from the political sphere (Dickovick, 2011). As Lambright discusses in this collection, decentralisation in Uganda was essentially a compromise position by President Museveni, who refused to allow multi-party competition at the national level until a referendum in 2005. Local participatory democracy corresponded well with the grassroots resistance councils that Museveni's National Resistance Movement (NRM) set up in the 1980s. In Senegal, the Parti Socialiste (PS) initiated decentralisation reforms in 1996 as a means of retaining party support at the local level, particularly in rural areas, even though it was losing credibility at the national level (Dickovick, 2005).
In all three countries, various forms of political decentralisation have existed at the municipal level. Previous to reforms in 2011, Kampala had a strong directly elected executive. However, in Cape Town and Dakar, the mayor is selected by city councillors who are elected by the broader populace. The electoral rules in each country also have important implications for the nature and degree of accountability that electoral officials have to their constituents. For example, in Cape Town, the councillors are nominated by party bosses rather than directly selected by voters. In Dakar, a constituency-based system is used but candidates are required to be affiliated with an existing political party and cannot run as independents. By contrast, given Uganda's 'no party' system, candidates for municipal office historically had to run as independents with no party affiliation.
Fiscally, while the cities in each country are all more economically viable than the rural areas, they nonetheless demonstrate important variation. As one of Africa's most developed economies, South Africa's cities have a larger tax revenue base on which to draw, particularly in Cape Town. In fact, Cameron shows that only about 15% of Cape Town's budget comes from government grants. Moreover, the funding that the city does receive is based on stipulations within the Constitution that provide local government with an equitable share of nationally-raised revenue. Only 20% of Dakar's revenue comes from the country's two main types of inter-governmental resources, the Fonds de dotation de la Décentralisation (FDD) and the Fonds d'équipement des collectivités locales (FECL). However, Resnick highlights that the distribution of these resources, particularly the FDD, is done on an ad-hoc basis, decided annually by a committee of officials that includes the President and the Prime Minister. Kampala depends more heavily on central government transfers, which comprise about 47% of its revenues. Like other city councils in Uganda, Lambright notes that Kampala should theoretically receive an unconditional grant from the central government based on a formula that takes into account its population and size.
Aside from Cape Town, the level of total revenue in each of these cities is insufficient, often meaning that city councils have unfunded mandates. In turn, this can have important implications for the delivery of services. Cape Town is South Africa's best managed city, even more so than Johannesburg. By contrast, Lambright notes that Kampala is considered a 'modern executive slum' whose performance in areas such as garbage collection lies well below the national average. In Dakar, resources are sufficient for addressing only 3 out of 9 mandated responsibilities, including health, education, youth and sports. Sectors that have not been decentralised, in particular water and sanitation, perform much better. A challenge in every setting is a high level of administrative complexity, resulting in overlaps in responsibilities between different layers of government.

Strategies of subversion in vertically-divided contexts
The disparate manifestations of the decentralisation process noted above influence how problematic vertically-divided authority becomes for service delivery and governance at the municipal level. Decentralisation typically alters relationships of accountability between citizens and elected politicians as well as between central and sub-national policy-makers © The Author 2014. Development Policy Review © 2014 Overseas Development Institute. This is an open-access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Development Policy Review 32 (S1) (Ahmed et al., 2005). Both central and local government actors will prefer a decentralisation process that maximises their autonomy, but they will prefer to be held accountable only when service-delivery outcomes are favourable.
The preference for greater autonomy and more malleable lines of accountability is exacerbated under conditions of 'vertically-divided authority' when different political parties are in control at the national and local levels. When the national and sub-national governments are in the hands of the same party, we would expect the national government to encourage better service delivery at the local level. 2 In such cases, the divestiture of autonomy is less egregious and there are strong, electoral spillover effects between constituents' perceptions of local and national leaders. This is especially true if elections are concurrent because such an institutional arrangement usually causes voters to view local and national politicians as equally responsible for performance (see Samuels, 2004). Under divided authority, the national government obtains little reward for helping sub-national officials from opposition parties. Based on observations from the Latin American context, where vertically-divided authority became more pronounced in the 1980s as mayors were increasingly elected, Myers (2002: 21) noted, 'When the capital city mayor and the president belonged to opposing political parties, the basic issues that divided their parties complicated relations between them.' Collectively, then, divided authority can encourage the central government to employ strategies of subversion, or tactics designed to limit the autonomy of sub-national officials while simultaneously increasing their culpability for poor service delivery (see Table 1). To obscure political motivations, these strategies are often justified via the use of a development discourse. In other words, the central government may stress that its interventions are driven by a desire to improve local development

Political
Resist relinquishing control to sub-national officials from opposition through increase in appointments Administrative Divest certain administrative responsibilities, unless they offer high levels of visibility to key voters Fiscal Limit ability of sub-national officials to finance their administrative responsibilities by reductions and restrictions in inter-governmental transfers and/or limitations on taxation or delays to improving donor funding As noted earlier, political decentralisation involves increasing the voice of local communities by shifting away from appointments of sub-national authorities and instead 2. Exceptions might occur when intra-party divides exist, because they could lead to an active undermining of independent-minded mayors whose individual success at service provision poses a threat to party leaders. For example, this was the case in Mozambique when Daviz Simango was considered by his RENAMO party to be 'too competent' in his job as mayor of Beira and therefore too threatening to the party's elite. Simango was expelled by RENAMO but ran as an independent candidate in the 2008 municipal elections and was re-elected mayor by a landslide vote.
allowing such authorities to be elected. Since elections are viewed as the key mechanism for enforcing accountability (Schmitter and Karl, 1991), political decentralisation offers sub-national officials the ability to be rewarded or punished for performance. Sub-national officials will prefer to be elected, rather than appointed, to their positions because they can then advocate in favour of those policies most relevant to their communities without fear of removal by the central government (see Falleti, 2005). By contrast, the ruling regime at the national level will be more concerned about manipulating power relations at the local level and therefore less favourable to political decentralisation if it involves relinquishing control, especially to sub-national authorities which are from the opposition (see Boone, 2003). For example, the ruling party in Namibia recently amended its Regional Councils Act to allow the President to appoint regional governors in those regions of the country where the ruling party lacks majority representation. According to the ruling party spokesperson, the change was needed 'because the opposition may block some government development projects…so that in the end, they will claim that government failed to deliver' (cited in Sharife, 2011: 35). Similarly, the ruling Botswana Democratic Party (BDP) often appointed additional councillors, who enjoyed the same privileges as their elected counterparts, to cities controlled by the opposition Botswana National Front (BNF). The Minister of Local Government was able to do this on the grounds that if a district had specific development needs, it could nominate additional councillors who possessed such skills (Kathyola and Job, 2011). Similarly, when the opposition gained control of Cameroon's major urban councils in the municipal elections of 1996, the ruling Cameroon People's Democratic Movement (CPDM) appointed CPDM delegates who had the power to manage council projects and finances (Nyamnjoh, 1999).
According to Falleti (2005), administrative decentralisation represents the domain to which the central government is most favourable. Her reasoning is that the national government prefers to divest itself of as many expenditure responsibilities as possible. However, Andrews and Schroeder (2003) offer an important caveat to this claim. Based on cases of Tanzania and Zambia, they note that national politicians are least likely to relinquish control over providing those public services that are highly visible to political constituencies and which offer important opportunities for donor funding.
A more nuanced position most probably prevails in a context of vertically-divided authority. In such cases, the central government is only willing to give away those responsibilities de-facto to a local government when public services might fail, and takes control of them when there is the possibility for electoral gain. For example, the arrival of an opposition mayor, Ricardo Belmont, in Lima, Peru during the mid-1990s, became a source of concern for President Alberto Fujimori. He tried to deprive the mayor of any popularity by announcing that he would invest US$140 million in education and public works, despite Belmont's protests that such activities were really the responsibility of local government (Dietz and Tanaka, 2002). In Mexico, the arrival of democratic elections brought a mayor from the opposition Partido de la Revolución Democrática (PRD), Cuauhtémoc Cardénas, to power in Mexico City. The national ruling regime, the Partido Revolucionario Institucional (PRI), resisted helping Cardénas combat drug crime in the city because of fears that successfully tackling this problem would hurt the PRI in subsequent elections (Davis, 2002). Fiscal decentralisation interacts strongly with these challenges over administrative decentralisation. In general, sub-national officials will prefer inter-governmental transfers to support service delivery because they do not want to lose popularity with their citizens by raising taxes (Garman et al., 2001; see also Smoke, 2004). As Rodden (2002) notes, such transfers also allow local government to shift blame for poor service delivery to the central government, reducing accountability. Presidents, however, prefer that local government relies on taxes rather than transfers (Garman et al., 2001). Yet, divided authority may encourage sub-national authorities to prefer greater taxation powers in order to reduce their dependence on the central government. Indeed, focusing on the city of Caracas, Ellner and Myers (2002) observe that reliance on resources from the central government hindered each elected mayor's ability to build a municipal political machine. Alternatively, divided authority where local tax bases are low may encourage sub-national authorities to eschew tax mobilisation in preference for transfers that are not earmarked for specific purposes by the central government and that are delivered on a predictable and regular basis (see Falleti, 2005;Garman et al., 2001). Otherwise, such transfers are vulnerable to manipulation by the central government and augment the inability of politically-disadvantaged, sub-national governments to deliver services (see Ahmed et al., 2005;Gomez, 2003).
3 In Zambia, for example, efforts by the then-opposition Patriotic Front (PF) to fight cholera in Lusaka were in the past thwarted by the late delivery of resources from the central government (see Resnick, 2011).

Findings from the cases
The cases in this special issue highlight the manifestation of these strategies of subversion. All the cases reveal efforts to undermine the political autonomy of municipal authorities, though to different degrees. The most extreme example is in Uganda, where a parliament dominated by Museveni's NRM party passed the Kampala City Act in 2009. This Act removes the authority of the city council to govern Kampala and instead replaces it with a Kampala Capital City Authority (KCCA). The elected mayor has now essentially become a ceremonial position, while a new Executive Director, appointed directly by President Museveni, holds executive authority. Although justified on the basis of improving service delivery, the timing of the decision has clear links to partisan struggles. Overall, the move is a major form of backtracking in a country that was once known as being among Africa's most decentralised.
Though less extreme, a similar tactic was used under former president Wade in Senegal, particularly after the BSS coalition gained control of a majority of the country's urban councils in the 2009 local elections. Wade's government engaged heavily in the practice of découpage, which involves the sub-dividing of administrative units into smaller entities. This typically entailed removing elected officials from their posts and replacing them with an appointed 'special delegation,' which typically consisted of members from Wade's PDS party. Just as in Uganda, the justification for this has been to improve service delivery by bringing government even closer to the people. However, given that predominantly opposition strongholds had been targeted, the increase in découpage appeared to have strongly partisan overtones.
Political interference was still present but least extreme in the South African case. For instance, sensing that it would lose Cape Town to the opposition DA, the ANC attempted to renew the contract of the party-appointed city manager of Cape Town just days before local elections in 2006. This violated legislation stating that the city manager needed to be installed by the full city council. Moreover, the ANC has considered changes to the electoral system that would require a shift to holding local elections at the same time as national ones. This would create an advantage for the ruling party in Cape Town because the ANC typically wins a majority of votes in national elections within the city even though it has lost municipal elections to the DA in recent years.
Due to South Africa's Constitution, fiscal decentralisation appears to be the least susceptible to the challenges of vertically-divided authority because of clearly specified rules regarding mandated transfers. Indeed, as Dickovick (2011) highlights, transfers mandated by constitutions are the most preferred by sub-national governments because they are the least susceptible to political machinations by the central government. Although there are some concerns that Cape Town has received a lower share of unconditional grants as a share of its population than Johannesburg, this could be because the city's population has grown substantially since the last national census on which the allocations are based. The city's high tax base nevertheless mitigates the potential impact that any type of reduced inter-governmental transfers has on service delivery.
Uganda provides inter-governmental transfers based on a formula derived by population and size. Yet Kampala does not receive funds commensurate with its responsibilities and size, and its high dependence on central-government funding creates space for political manipulation. Since many of these transfers are conditional, they also limit the policy options of local government. More recently, Museveni has even eliminated important sources of local revenue for Kampala, including the graduated tax exempting owner-occupied residences from property taxes, and the tax on boda bodas or motorcycle taxis. These reductions have negative implications for key services, such as garbage collection.
The system of inter-governmental transfers in Senegal is probably the most vulnerable to political machinations since funding decisions are decided on an annual basis according to a committee presided over by the President. Government transfers to Dakar have declined since the BSS coalition came into office, despite the city's large population and growing service needs. Often, the transfers are released too late in the year to be properly included in the budgets of the local municipalities.
In the administrative domain, the cases suggest that central governments do try either to rescind responsibilities for highly visible services that could bolster the popularity of the opposition or to offload important tasks to local government that could undermine the latter's reputation. In Senegal, Wade's government launched a series of high-profile, but relatively unsuccessful, projects to support victims of urban flooding in Dakar during most of the President's tenure in office. However, the responsibility for this matter was then offloaded to local government after the BSS came to office in 2009. This occurred despite the fact that local government tends to lack the requisite capacity to handle such a perennial development challenge which is a major source of discontent for poor Dakarois. At the same time, Wade's government removed the city's responsibility for garbage collection soon after local-government officials devised a new strategy for a more effective mode of solid-waste disposal. In Kampala, the central government constructed a sewage plant in one of the city's wetlands, which directly undermined the city council's responsibility for preventing illegal construction in the area. This then opened the door for thousands of residents to also construct illegal homes in the area, thereby undermining the city's authority. As Devas (2005: 353) observes in the Indian context, rescinding localgovernment responsibilities is problematic precisely because it erodes the degree of accountability between local citizens and elected municipal officials; 'removing key functions from local level decision-making risks weakening whatever influence the urban poor have on the services that affect them.' The South African case is much more nuanced. The ANC did remove Cape Town from participating in the N2 Gateway housing scheme in March 2006 just after the DA was elected. The scheme was a highly visible project aimed at providing disadvantaged populations with proper housing. According to X, one motivation for this move was that the ANC feared that the opposition would uncover evidence of patronage-based allocations of housing to the ruling party's supporters. Yet, compared with the other cases, other examples of interference at the administrative level have been less apparent in South Africa.
There are a number of reasons why such patterns emerge. First, the degree of administrative ambiguity that already exists across different levels of government creates the opportunity to shift responsibilities for certain services. The vaguer the laws related to administrative decentralisation are, the more likely that there is space for political manipulation. Secondly, in certain countries, there are important incentives that mitigate excessive political interference in local service delivery. For instance, South Africa's national ministers have performance contracts with the President in the area of service delivery. Since Cape Town has a good reputation for capacity and delivering services, even ANC ministers often support the efforts of the city because this could, in turn, improve the ministers' own image.

Implications for the donor community
The findings from the cases are particularly relevant to the international donor community. Many donors support decentralisation as a means of improving service delivery in developing countries. In Africa, this support is reinforced by the fact that a number of countries include decentralisation as a key objective in their country's poverty-reduction strategy papers, which are the documents that typically guide donors' own interventions in a specific country. 4 The World Bank, for example, allocated approximately US$10.6 billion to decentralisation projects worldwide over the period spanning 1990spanning -2006spanning (IEG, 2008.
By contrast, in his contribution to this collection, Stren notes that many key agencies have recently reduced the degree of finances and levels of staff devoted to the urban sector because of competing demands for attention to other important areas, such as combating HIV/AIDS, climate change, and food insecurity. He highlights that this is problematic, given that donor financing is critical for a number of municipal governments in Africa due to their low resource base, which in turn limits their expenditure capabilities in key servicedelivery areas. The World Bank remains the most active donor in Africa's urban sector, although its approach has shifted significantly over the years. While initially supporting 'sites and services' housing projects during the 1970s, the Bank then moved to an emphasis on improving local management practices and viewing the city as a catalyst for wider development. More recently, the institution has shifted to highlighting the importance of local participation in urban service-delivery projects, thus complementing the Bank's broader focus on 'good governance'.
Throughout these trends, however, the Bank and many of its donor partners have espoused an apolitical view on urban service delivery and decentralisation. Yet, recognising and addressing the ongoing contestation of power between local and national authorities in contexts of vertically-divided authority is crucial for ensuring the effectiveness of development assistance. As Eaton et al. (2010: 19) note, 'When development partners engage with local officials like mayors and municipal councillors, they are likely to gain greater analytical leverage if they simultaneously approach these officials as actors who are engaged in partisan struggles---both with national leaders of their own parties and with competitors in rival parties. ' Even under normal circumstances, the fact that donor resources must often be approved by the central government and transferred to the national treasury can constrain decentralisation policies, and even more so when assistance is provided via budget support or sector-wide approaches (see Andrews and Schroeder, 2003;Romeo, 2003). In a context of vertically-divided authority, donor involvement to support either decentralisation or urban service-delivery initiatives can become even more politicised. For instance, in Mexico City during the late 1990s, Cardénas requested assistance from the World Bank to help fund an urban housing project. According to Davis (2002: 251), the PRI national government was 'slow, if not entirely negligent' in approving foreign aid for the project. Likewise in Lima, mayors required the support of the Peruvian national government to obtain loans or credit. However, former president Alberto Fujimori prevented an opposition mayor from gaining World Bank credit for investments in the city out of fear that this would allow the mayor to create an independent power base (Dietz and Tanaka, 2002).
In the cases presented here, we observe a variety of approaches used by donors to navigate this unusual context. In South Africa, certain donors do not discriminate in their interventions, while others prefer to allocate scarce resources to the best-performing cities, including the DA-led Cape Town. In Senegal, key donors in the urban sector, including not only the Bank but also the French Development Agency, have supported the creation of a Municipal Development Agency and rely on medium-to long-term municipal contracts that involve ex-ante commitments by both central and municipal governments. To a large extent, Resnick notes that these contracts prevent backtracking by the central government even when an opposition party comes to power at the local level. In Uganda, however, donors appear to be held relatively hostage to the decisions of Museveni's government. As Lambright highlights, they have halted funding to the main decentralisation programme, the Local Government Sector Investment Plan (LGSIP), in the wake of the government's recentralisation. Other interventions, including the use of sectoral budget support for service delivery, reinforce the country's gradual reversal of the decentralisation process.

Conclusion
Jointly, the research in this collection poses important questions about how partisan politics intersects with local governance structures and urban service delivery. In doing so, the research reconciles three literatures that often do not communicate with each other. Specifically, the decentralisation literature rarely focuses explicitly on cities, even though both the service demands and the political importance of such agglomerations differ markedly from those in rural areas. At the same time, political-science scholarship on partisanship focuses predominantly on dynamics at the national level and on electoral periods. By contrast, work on urban studies sometimes fails to consider how the macro political environment influences outcomes at the city level. In addition to these contributions, the studies provide a set of rich data on municipal finances, urban service delivery, and local elections that typically are difficult to obtain for African cities.
As Africa's rapid urbanisation continues apace and as its nascent democracies become more consolidated, the intersection between governance and service delivery will undoubtedly become more pronounced. The challenge will be to determine whether isolated attempts at recentralisation are truly motivated by concerns over poor service delivery and low capacity at the municipal level or by political considerations by central governments. Through both cross-country and cross-city comparisons, this collection begins to address these questions by highlighting what policy interventions and decentralisation processes best ensure that Africa's governments can begin to meet the basic service needs of the region's growing numbers of urban poor.