introduction: decentralising service delivery? evidence

1 IntroductionDoes the devolution of responsibility for serviceprovision to elected local authorities improve thedelivery of services to the poor? This is the majorchallenge of democratic decentralisation and a keybenchmark by which its effectiveness should beassessed. Many governments across the developingworld are engaged in ambitious efforts to devolvepower and resources to local bodies which areincreasingly assuming responsibility for managing thedelivery of health, education and other essentialservices to poor people. Decentralised servicedelivery is now a key determinant of the scope forless-developed countries to achieve the MillenniumDevelopment Goals since many of these goals arepremised on outcomes that are increasingly withinthe realm of responsibility of elected localgovernments.

While many adherents of decentralisation value itspotential to increase accountability and participationat the local level, for poor people the critical litmustest lies in the scope for decentralisation to deliverimprovements in services and material well-being.

Improvements in democratic accountability andbetter service delivery outcomes are not mutuallyexclusive but can have complementary and mutuallyreinforcing effects. The problem is thatdecentralisation policy initiatives are often premisedon strengthening local democracy withoutconsidering fully the conditions under which servicedelivery improvements can be achieved. Thechallenge confronting proponents of decentralisationis that the evidence to support the case fordecentralised service delivery is fragmentary andinconsistent and the conditions for successfuldevolution of services are poorly understood. Thearticles in this IDS Bulletin seek to throw light on

these issues by marshalling evidence on servicedelivery outcomes from a range of developingcountries that are engaged in a process ofdemocratic decentralisation and highlighting theimplications for designing reforms that maximise theprospects for improvements in the quality and accessof services for the poor.

2 Decentralising service delivery: assumptionsand challengesMost efforts to promote democratic decentralisationare premised on the assumption that localgovernments will be more responsive to the needs ofthe citizens and take their preferences into account indetermining the type of services to be provided, thelevel of resources required, and the optimal means ofensuring effective delivery. Such efforts are alsopredicated on the expectation that power andresponsibilities will be devolved by benign centralgovernments to elected local bodies that areaccountable and responsive to their constituents. Afurther assumption is that financial resources will beavailable to support the provision of services at thelocal level through a combination of centralgovernment fiscal transfers and local taxation. Finally,most decentralisation initiatives assume that localadministrative capacity will be adequate to deliver theexpected increase in the production of local services.

However, as the articles in this IDS Bulletindemonstrate, these supportive conditions are oftenabsent in many contexts in which decentralisation istaking place. For this reason, efforts to improve thedelivery of health, education, drinking water andlocal infrastructure through elected localgovernments are often destined to fail. The evidencepresented here provides little succour to adherentsof decentralised service delivery by showing thereare very few cases where equity and efficiency

Introduction: DecentralisingService Delivery? Evidence andPolicy Implications

Mark Robinson

IDS Bulletin Volume 38 Number 1 January 2007 © Institute of Development Studies

1

1Intro38.1.qxd 11/01/2007 10:29

outcomes have improved as a result ofdecentralisation. Evidence of sustained improvementis very slight and is usually highly context specific,with improved outcomes resulting from acombination of locally supportive conditions. Thearticles in this IDS Bulletin draw attention to someimprovements in service delivery outcomes inColombia and Kerala and in the health sector in partsof Asia and Africa. But there is little convincingevidence from these articles and wider experience toshow that education outcomes are improvedthrough decentralisation.

In most cases reported from Africa, Asia and LatinAmerica the quality of public services has eitherdeclined or remained unchanged as a consequenceof democratic decentralisation. The evidence collatedand reviewed in this IDS Bulletin suggests thatdecentralised service delivery has not improved poorpeople’s access and improvements in quality have notresulted from a transfer of power and responsibilitiesto local authorities. Decentralisation also accentuateshorizontal inequalities between richer and poorerareas as a consequence of differential levels ofadministrative capacity and ability to raise localresources. From the evidence summarised byConyers, with some minor exceptions, theexperience of sub-Saharan Africa is especiallydisappointing with little improvement in the qualityof services provided through local governments,both to poor people and local citizens moregenerally, a fact confirmed by negative publicperception of decentralisation in many countries.

But this dismal evidence should not lead to theconclusion that decentralisation is inimical toachieving improvements in services that benefit thepoor. Nor does it necessarily lead to the conclusionthat centralised provision or deconcentration toarms-length central government agencies can ensureimproved delivery, or that private and non-governmental providers are better placed to deliverservices to poor people at the local level. Moreover,efforts to decentralise service provision are oftenmotivated in the first place by poorly performingcentralised state agencies that are unaccountable andunresponsive to local people and their electedrepresentatives in local governments. As Conyerscautions in her article, the results of decentralisationare mediated by a number of factors that lie beyondthe control of elected local representatives andofficials, such as the prevailing political context, the

balance of power at the local level, and then lack offinancial resources. Many of the problems withdecentralised service delivery lie in the design andimplementation of reform initiatives and insufficientattention to the feasibility of achieving majorimprovements without commensurate changes inbroader governance structures and underlyingsocioeconomic conditions.

A further difficulty in coming to definitiveconclusions on the potential for decentralisation todeliver improved outcomes is that the evidence isweak, partial and inconsistent. This finding emergesas a clear theme in the contributions by Conyers andRobinson. The case study evidence and survey data inthis IDS Bulletin adds considerably to the existingknowledge base and our understanding of theproblems encountered in decentralised serviceprovision. But the contributors acknowledge that it isdifficult to arrive at hard and fast conclusions in theabsence of comprehensive baseline or control dataon the state of service delivery at the inception ofdecentralised service delivery initiatives. The article byMohmand and Cheema on the decentralisationreforms in Pakistan is one of the very few exampleswhere such data has been collected. In most cases,the data draws on particular sectors for specific timeperiods and with limited geographical coveragewhich frustrates the task of generalisation. A relatedproblem is that the available data does not breakdown outcomes by socioeconomic status, gender,age or ethnicity which obscures the differentialimpact of decentralised service delivery on thequality and access of services. The article byShankland and Athias on the decentralisation ofhealth services to movements representingindigenous peoples in Brazil is a rare example of onesuch study. Finally, it is difficult to disentangle thedifferential effects of decentralisation from otherparallel processes such as economic stabilisation andprivatisation which have respectively resulted in fiscalausterity and reduced state provision of services. Ashighlighted by Robinson, for many Latin Americancountries in the 1980s, the combination ofdecentralisation with structural adjustment andprivatisation proved disastrous in terms of servicedelivery outcomes for poorer people and regions.

3 Decentralising service delivery: evidence andinsightsThe various contributions in this IDS Bulletin providefresh data and review available evidence on the

Robinson Introduction: Decentralising Service Delivery? Evidence and Policy Implications2

1Intro38.1.qxd 11/01/2007 10:29

impact of decentralised service delivery in localjurisdictions in Africa, Asia and Latin America. Thearticles by Conyers and Robinson respectively offerbroader comparative reviews from Africa anddeveloping countries as a whole while the other fivearticles draw on the experience of individualcountries. The coverage ranges from country-levelreviews in Mexico (Salazar) and Cambodia(Spyckerelle and Morrison), through to large-scalesurvey and comparative case study evidence fromdecentralisation in India (Johnson et al.) and Pakistan(Mohmand and Cheema), and the experience ofindigenous communities in Brazil (Shankland andAthias). In gathering evidence they draw on originalsurvey data, case study research and publishedsecondary materials. The articles cover a range ofservice sectors but with a particular focus on health,education, drinking water and sanitation, as well aslocal infrastructure, with considerable variation in theform and extent of decentralisation and availability ofresources. Most of the empirical material covers theperiod of democratic decentralisation from the late1980s and 1990s, with reference to earlier episodesof administrative decentralisation under centralisedand authoritarian governments.

In his contribution to this IDS Bulletin, Robinson surveysthe empirical evidence in the secondary literature onthe impact of decentralisation on service delivery,drawing on examples and case material from LatinAmerica, Africa and Asia. He finds that thecomparative evidence on equity and efficiencyoutcomes is very limited and uneven in coverage,rendering the task of generalisation difficult. Theavailable evidence suggests that the consequences forequity and efficiency are largely negative, with poorerpeople and regions being disadvantaged bydecentralisation reforms or receiving a much lowershare of the resulting benefits of improved servicedelivery. Contrary to theoretical expectations, there islittle evidence to suggest that the quality of servicesimproves with decentralisation, which means that localpreferences do not automatically translate intoimproved targeting and delivery. Whiledecentralisation can contribute to improvedparticipation and accountability, it is generally notdelivering improved service delivery outcomes. Thisdoes not invalidate the potential of decentralisation toimprove the delivery of quality services to the poor orto reduce inter-regional disparities. Robinson arguesthat a number of supporting conditions are requiredto enable decentralisation to deliver on this potential,

which includes political commitment, politicalmobilisation of the poor, institutionalised participationand accountability mechanisms, the availability ofadequate financial resources, and technical andmanagerial capacity in local governments.

In her article, Conyers focuses on African experienceof decentralisation and reviews the evidence onservice delivery outcomes available from a variety ofcountries and sectors. She also finds that, despitesome isolated examples of success, decentralisationhas not had a significant positive impact on thequality of public services in the region. However, sheargues that the main reason for poor outcomesstems from the fundamental characteristics ofcontemporary African states than decentralisation assuch. These include the centralisation of power,weak structures of accountability, and lack ofcountervailing pressure from civil society. For thesereasons, African governments have largely beenreluctant to devolve power and finance to localgovernments, which consequently lack the capacityand resources to deliver improved services. Conyersconcludes that the problems of decentralisationcannot be addressed in isolation from widerproblems of governance prevailing in many Africancountries, and therefore have to be addressed as partof a slow and gradual process of state-building.

Turning to the first of the Asian cases, Johnson et al.examine the impact that local governance structureshave on the ability of sub-national governments toimplement social policy in two states of India,Madhya Pradesh and Andhra Pradesh. They findsignificant differences in the functioning ofdecentralised local bodies in the two states whichhave a bearing on the implementation of creditdelivery programmes for the poor. In MadhyaPradesh, which had progressed further withlegislation governing the devolution of powers andresponsibilities, the government implemented itsself-employment and microcredit programmethrough elected local councils (panchayats). Incontrast, the government of Andhra Pradesh choseto implement its group savings scheme through self-help groups formed by poor rural women under thesupervision of officials in local governmentadministrations. Based on extensive field research inthe two states, the authors find that rates ofsatisfaction were much higher among members ofself-help groups in Andhra Pradesh than amongmicrocredit beneficiaries in Madhya Pradesh, and

IDS Bulletin Volume 38 Number 1 January 2007 3

1Intro38.1.qxd 11/01/2007 10:29

that perceived levels of corruption were much lowerin the former. The authors also report that levels ofparticipation in village assemblies (gram sabhas) weremuch higher in Andhra Pradesh, explained in part bythe incentives generated by the microcreditprogramme and the closer spatial proximity betweenvillagers and local government officials.

The personal involvement of the elected heads oflocal councils (sarpanches) in Madhya Pradesh indetermining eligibility and access to governmentschemes was found to give rise to patronage inbeneficiary selection, while in Andhra Pradesh theheads of elected councils lacked such discretion, withlocal government officials primarily responsible forselection, monitoring and implementation. Thesefindings demonstrate that devolving power andresponsibility to elected local councils in India maynot produce the improvements in service deliveryanticipated by proponents of democraticdecentralisation without commensurateimprovements in local accountability. Rather, theauthors highlight the value of close interactionsbetween local self-help groups representing thecollective interests of poor women and linedepartment officials operating at the local level inwhich the scope for graft was minimised by thedesign of the programme and the type of benefitsthat accrued to the participants in the scheme.

Mohmand and Cheema analyse the extent to whichdecentralisation reforms introduced in 2001 havebeen effective in improving the magnitude andprovision of health and education services in ruralPakistan. Survey data from four villages in Faisalabaddistrict of Punjab province gathered two years afterthe introduction of the reforms reveal that while theprovision of targeted sanitation and sewerageservices has increased substantially, the vast majorityof respondents in the sample villages report that thedelivery of universal health and education serviceshas either remained unchanged or has worsened.These results are confirmed by a national social auditthat used a baseline survey to track improvements inservice delivery. Utilisation of government healthservices has declined dramatically because ofshortages in the availability of medicines and doctorswhile those who were able to afford to do so optedfor private health provision on the basis of availabilityrather than considerations of quality. Mohmand andCheema find a similar pattern with state-providedprimary education, where the majority of

respondents report no improvement in teacherattendance or in school facilities since thedecentralisation reforms. This is especially marked forgirls’ primary schools where teacher attendance isworse than for equivalent boys’ schools, but wherealternative options for private schooling are far lessprevalent or accessible. An important consequenceof the reforms is that quality and equity in the localprovision of public services has worsened, and thatsmallholders and low caste citizens and their girlchildren are especially vulnerable as they cannotaccess alternative options. The authors locate theseproblems in accountability failures in the design ofthe new decentralised system and the consequentinability of citizens to hold local service providers toaccount for the shortcomings in the provision ofuniversal services.

In the third Asian case considered in this IDS Bulletin,Spykerelle and Morrison examine the approach takenin Cambodia to the delivery of small-scale localinfrastructure through newly empowered communesunder the provisions of the 2001 legislativeframework for devolved governance. A centralobjective of these reforms was to improve localinfrastructure and access to services, as well as topromote participation and good governance at thelocal level. Based on successful experience withdiscretionary funding before the new legislationcame into force, a special Commune Fund wasestablished to which the communes submit bids forlocal infrastructure development in the form offunding for the construction and repair of roads,bridges, irrigation works, water and power supply,markets, schools and health centres, which in turncould improve the provision of basic services.

Since 2003, one third of Cambodia’s population isestimated to have benefited from these investments.While it is too early to determine the effectivenessof these commune-level investments, Cambodia haslaid the foundations for decentralised service deliveryby establishing district-level planning andadministrative arrangements to oversee localimplementation and has created consultationmechanisms to ensure that the interests of the poorare represented in commune development planningprocesses. A key obstacle facing the successfuldecentralisation of service provision in Cambodia isthe limited financial resources available to thecommunes for local infrastructure investments (alittle over US$1 per capita), capacity constraints at the


1Intro38.1.qxd 11/01/2007 10:29

commune level, and the limited scope for mobilisinglocal financing. The remit of communes in localservice provision is also restricted in focus. Healthand education still remain the responsibility of centralline departments and the communes have no directrole in the implementation of these services underpresent legislative provisions. However, Cambodiahas created the legislative and administrativefoundations on the basis of which further devolutionof responsibility for service provision could occur,provided that commensurate resources and capacity-building investments are provided by the centralgovernment.

Coming to the first of two Latin American cases,Salazar examines the lessons emerging from Mexico’sexperience of decentralisation for the delivery ofessential public services since the late 1990s. Therehas been a major shift in responsibility fromcentralised government provision to the present-daysituation in which state and municipal governmentsare responsible for providing the majority of servicesin health, education and social infrastructure. But theevidence on changes in the delivery of public servicesdoes not point to a positive picture from theperspective of equity and efficiency. In particular,there are horizontal imbalances in levels of provisionand service delivery outcomes between the richerand more disadvantaged parts of the country andamong different sections of the population.Education achievement has not improved withdecentralisation while significant disparities remainamong social groups in terms of years of schooling.Considerable inefficiencies arise from uncleardemarcation of responsibilities between state andfederal governments. In the health and educationsectors there are significant disparities in theavailability of resources for service provision betweendifferent states. Increased funds have been madeavailable to finance social infrastructure in themunicipalities but the evidence suggests that there isno discernable relationship between the resourcesexpended and resultant improvements in servicedelivery. Salazar identifies two key problems thatcontribute to poor service delivery, namely the weakaccountability of public officials responsible forservice delivery and the high degree of fiscalcentralisation, both of which inhibit the scope forimproved outcomes.

The decentralisation of healthcare in Brazil is thefocus of the contribution by Shankland and Athias,

with a particular focus on the role of indigenouspeoples’ movements in the delivery of healthservices. Indigenous peoples suffer from much worsehealth problems than the average Brazilian citizen,with higher rates of infant mortality and high ratesof morbidity from infectious diseases. The Braziliangovernment’s efforts to decentralise healthcareprovision to municipalities and special indigenoushealth districts in the 1990s was designed to tackleproblems of centralisation and exclusion. But despiteimprovements in immunisation coverage and otherhealth indicators, efforts to subcontract the provisionof healthcare services in indigenous health districtsto non-governmental organisations (NGOs), churchgroups and indigenous organisations did not provesuccessful, with continued high rates of morbidityamong indigenous peoples from preventablediseases. Shankland and Athias examine theexperience of decentralised healthcare in a regionmainly inhabited by indigenous peoples through acase study of the Rio Negro region in the farnorthwest of the Brazilian Amazon, where anindigenous peoples’ movement assumedmanagement responsibility for health services.However, the movement continued to operatewithin the technical parameters of state healthprovision, with priorities and resource allocationslargely determined by non-indigenous healthprofessionals in line with central governmentpolicies. While the efficiency of healthcare provisionimproved under indigenous management, it faced aseries of difficulties caused by delayed financialtransfers and was unable to preserve an adequatelevel of health services which resulted in problemswith uneven drug supplies and erratic staffattendance in health centres. Hence, while theprinciple of decentralising healthcare delivery toindigenous management was initially viewed as awelcome development, it ultimately proved unableto achieve any significant impact on the nature ofthe decentralised service given the constraintsoperating in the wider healthcare system.

4 Lessons and implications The implications of the various articles in this IDSBulletin are of considerable significance for researchand policy. These emanate from a remarkablyconsistent set of findings based on the empiricalevidence gathered by the contributors from a rangeof countries and sectors. These are as follows: (1) theconsequences of democratic decentralisation interms of service delivery outcomes are largely


1Intro38.1.qxd 11/01/2007 10:29

negative, (2) poor and marginalised people have notgenerally experienced improved access and servicequality under democratic decentralisation; and(3) improved outcomes are contingent on asupportive set of conditions and mediating factors,some of which lie outside the control of electedlocal governments.

These findings raise important implications forresearch. The dearth of systematic, robust andcomparative evidence on decentralised servicedelivery outcomes is a striking gap in knowledge.Existing research remains partial, limited and contextspecific. Knowledge on what works well, where,how and why is still fragmentary. The implications ofdecentralisation for service delivery in differentsectors in a range of political and socioeconomiccontexts are far from clear. There are fewer areas ofdevelopment policy that are more in need ofresearch than strengthening the evidence base tomeasure the impact of policies designed to deliverservices to poor people through elected localgovernments.

The most immediate policy implication is an urgentneed to reappraise what democratic decentralisationcan realistically be expected to deliver in terms ofimproved service delivery. In some contexts, theexpectations may simply be excessive, and limitationsof resources and administrative capacity mean thatelected governments may not be well placed to takeon the increased responsibilities that have beendevolved to them. Countries emerging from conflictand protracted civil war may not be in a position torapidly devolve services to local governments.Equitable service delivery outcomes are unlikely tocome about when power is deeply contested andpowerful social groups control resources anddominate local politics, since these are precisely theconditions that give rise to élite capture indecentralised service delivery. Efforts to decentralise

service delivery in such environments are likely to fail,especially if they result from hasty and poorlydesigned interventions that are intended to generaterapid results in the form of improved equity andefficiency.

More stable political contexts accompanied by steadyeconomic growth and with relatively capable localgovernments offer a more conducive environmentfor the devolution of power and resources which inturn are indispensable requirements for successfuldecentralisation of service delivery. Devolution ofresponsibility for service provision withoutstrengthening local state capacity to produce or tocoordinate delivery by non-state actors at the locallevel produces inefficient and negative outcomes.Inadequate resourcing of services through restrictedfiscal transfers to local authorities from centralgovernments and commensurate local tax-raisingpowers limits the scope and capacity to generatesufficient funds for adequate levels of serviceprovision. Delegation of responsibility to non-stateand private sector providers can make up for some ofthe shortfall in provision but this is not a substitutefor enhanced local state capacity and adequate levelsof resources. Political commitment, effectivechannels of accountability and effective oversight ofservice provisioning by local non-state actors are allrecognised to be essential ingredients for efficientand equitable provision. Finally, even thoughconcepts of devolution and local governance have along pedigree, democratic decentralisation is still arelatively new phenomenon in most less-developedcountries and positive results will take a long time tomature. As with other types of governance reforms,short-term and time-bound interventions will notbring about the desired improvements in servicedelivery outcomes. Steady, incremental and well-resourced initiatives that build capacity and increaseaccountability are the surest route to realising thepromise of democratic decentralisation.


1Intro38.1.qxd 11/01/2007 10:29

1 IntroductionMany claims are made in favour of decentralisation,ranging from the democratising potential of increasedscope for participation and accountability through topoverty reduction and improved service delivery.Much of the literature and evidence centres on theintrinsic value of decentralisation as a desirable goal inits own right. But the arguments for thedevelopmental significance of decentralisation restprincipally on a series of assumptions and theoreticaljustifications. Proponents of decentralisation basetheir assumptions on widely differing criteria, rangingfrom expected improvements in allocative efficiency,welfare and equity, through to increasedparticipation, accountability and responsiveness on thepart of local authorities. Economists tend to frametheir analysis in terms of the costs and benefits ofdecentralisation, while other social scientists andpractitioners are generally concerned with processesand democratic aspects of the process (Blair 2000).

This article focuses on substantive developmentoutcomes, centred on how far decentralisationproduces improvements in service delivery for thepoor, drawing on evidence concerning equity andefficiency and the political and institutionalconditions which give rise to these outcomes. Theliterature on democratic decentralisation and servicedelivery generally falls into two distinct categories:opportunities for enhanced popular participation andincreased accountability of local authorities, or onforms of service delivery involving a plurality ofactors. A major problem with the empirical literatureis that there is no systematic or comparativeevidence on whether increased participation indecentralised local governance generates betteroutputs in terms of improvements in the provision of

health, education and drinking water and sanitationservices for poor and marginalised people. Theavailable evidence draws either on examples fromsingle countries and sectors, or is anecdotal,temporally specific and highly localised, thusrendering the task of generalisation problematic.Similarly, efforts to measure development outcomes,in terms of reduced poverty or improved socialindicators, and to attribute these to increaseddevolution and participation, are inconclusive andfraught with methodological problems (Crook andSverrisson 2003; Dyer and Rose 2005).

These data constraints pose a serious challenge toadvocates of participation and local governance,since the material benefits for the poor arising fromimproved service provision should be a keydeterminant of the effectiveness of democraticdecentralisation. The intrinsic value of increasedparticipation, accountability and responsivenessshould not be underestimated, especially whenpolitical rights have been previously curtailed undercentralised, authoritarian regimes. However, unlessthese process changes demonstrably translate intoenduring improvements in service provision andmaterial well-being, the claims made for the pro-poor potential of democratic decentralisation remainincomplete and cannot easily be sustained.

Some definitional issues are in order at this point.Decentralisation encapsulates three distinctelements: (1) fiscal decentralisation, entailing thetransfer of financial resources in the form of grantsand tax-raising powers to sub-national units ofgovernment; (2) administrative decentralisation(sometimes referred to as deconcentration), wherethe functions performed by central government are

7

Does Decentralisation ImproveEquity and Efficiency in PublicService Delivery Provision?

Mark Robinson


1Intro38.1.qxd 11/01/2007 10:29

transferred to geographically distinct administrativeunits; (3) political decentralisation where powers andresponsibilities are devolved to elected localgovernments; this form of decentralisation issynonymous with democratic decentralisation. Ourconcern in this article is mainly with democraticdecentralisation, but several experiments indecentralised service delivery have involved thetransfer of financial or administrative powers to sub-national units of government that are not subject todemocratic oversight through competitive elections.These variations in the form and content ofdecentralisation have an important bearing onprocesses of participation, accountability andresponsiveness and in turn on service deliveryoutcomes.

The focus of this article is mainly on equity and socialjustice concerns, and some elaboration is required ofthe meaning and application of these terms withrespect to service delivery. Equity outcomes have twomain dimensions: access to services across differentgroups of the population on the basis of income,gender and other categories, and inter-regional equityin terms of disparities in access within and across localgovernment jurisdictions (Litvack et al. 1998: 8). Theprovision of affordable, accessible and appropriateservices to all categories of a population in equalmeasure is a universal standard for determining suchoutcomes, but a social justice perspective on serviceprovision privileges the benefits that directly accrue toeconomically and socially marginalised groups (OneWorld Action 1999, 2001).

Services are often equated with public goods likehealth, education, drinking water and sanitation andthese tend to be the most common forms ofservices provided by local governments in developingcountries. Police, fire, transportation, housing andsocial welfare services also fall under localgovernment jurisdictions in many countries. Localgovernments are also given responsibility for a rangeof other public services, such as infrastructure in theform of roads and bridges, public buildings, andhousing, especially in larger jurisdictions and urbanauthorities. In many countries, specialised services forlow-income groups are the responsibility of localgovernments, such as social welfare, credit, andagricultural extension. Local authorities in rural areasoften perform a range of functions directed atagriculture and rural development, environmentalmanagement, disaster prevention and rehabilitation.

Our focus in this article is principally on health andeducation as services that impact most directly onthe well-being of the poor, as well as urban andrural infrastructure.

In most countries, public services are largely providedby the state, through government departments andspecialised agencies, while private sector provision isbecoming increasingly common in all areas of serviceprovision as a result of state failure, and throughprivatisation and contracting-out that encouragesmarket competition at the local level. Public servicedelivery is no longer the exclusive prerogative ofstate agencies in national and local governments, butinvolves combinations of state and private actors, andincreasingly civil society organisations that aredirectly engaged in the delivery of services.

2 Decentralisation and service deliveryoutcomesA leading rationale for decentralisation is that it cangenerate financial, efficiency and quality gains bydevolving resources and decision-making powers tolocal governments for the delivery of services. It isfinancially attractive to national governmentsbecause part of the burden of financing services canbe shifted to sub-national units and private providerswhich can produce these at lower cost. Theallocative efficiency argument is that productivity ofhealth, education and other services can bemaximised by enabling local governments to takedecisions on the allocation of scarce resources, sincethey have a better sense of local preferences. In theprocess, decentralised units of government canbecome more accountable in resource allocationdecisions. It is further argued that the quality ofservice provision can also be enhanced bydecentralisation since local governments will bemore sensitive to variations in local requirements andopen to feedback from users of services (Azfar et al.2004: 21–4).

At the same time, the literature draws attention tothe risks involved in decentralisation. First, there is noautomatic assurance that increased politicalautonomy for local governments will lead toimprovements in public services. Second, there is thewell-known risk of capture by local political élites,which can worsen equity in the delivery of services.Third, the technical capacities of local governmentstaff may be inadequate. Fourth, decentralisation canwiden regional disparities in the provision of public

Robinson Does Decentralisation Improve Equity and Efficiency in Public Service Delivery Provision?8

1Intro38.1.qxd 11/01/2007 10:29

services. Fifth, decentralisation posesmacroeconomic risks by increasing governmentvulnerability to financial deficits and over-expandingthe size of the public sector (Burki et al. 1999: 3–4).

Governments of very different ideological hues inLatin America, Africa and to a lesser extent, Asia,have experimented with decentralised servicedelivery over the past two decades. Initiatives havecentred on the transfer of powers and resources tolower tiers of government, through a combinationof measures involving a process of deconcentrationto sub-national agencies operating under central linedepartments, and devolution of power and resourcesto elected local authorities. National governmentshave devolved responsibilities for different types ofservices across countries and jurisdictionsaccompanied by different degrees of fiscaldecentralisation. It is therefore difficult to makegeneralisations across sectors and countries. Forthese reasons, the data on the impacts ofdecentralisation on service delivery outcomes arepartial and incomplete, and some caution is requiredin the interpretation of available evidence in theabsence of cross-national and cross-sectoral studies.In the remainder of this section, the availableevidence from different parts of the world isreviewed, to determine whether any general lessonsor patterns emerge on the relationship betweenpolitical devolution and equity and efficiencyoutcomes.

2.1 Latin AmericaThe decentralisation process has progressed furthestin Latin America, beginning with efforts in Chile andColombia in the early 1980s, to delegate increasedresponsibilities to municipalities (sub-nationaladministrations) for the delivery of health andeducation services. These reforms emanated from avariety of domestic circumstances that differedbetween countries in the region. In some cases,conditions of resource scarcity brought about bymacroeconomic crisis spurred countries to devolveresponsibility to lower tiers of government (Prawda1993). Governments in Colombia, Argentina andBrazil devolved powers to elected municipalities aspart of a wider process of political liberalisation,whereas the military regime in Chile favouredadministrative deconcentration to municipalitiesunder the control of non-elected administratorsappointed by the military (Nickson 1995). In Chile,where the reforms were far-reaching, the transfer of

responsibility for primary and secondary educationand primary healthcare to municipalities wasaccompanied by measures designed to expandprivate schools and healthcare facilities. According toStewart and Ranis (1994), ‘Municipal governmentsthus acted like “service delivery agents”, providinglocal public services on a cost-effective basis, withouthaving local governing power’.

Latin American municipalities deliver services in fourways: directly through municipal secretariats anddepartments; indirectly through municipally ownedfoundations; through enterprises owned by themunicipalities or as joint ventures with the privatesector; and through contracts to private companiesor voluntary agencies (Nickson 1995). The absence ofcomparative data on the equity impact ofdecentralised service delivery in Latin America makesit difficult to derive well-founded conclusions, butthere is some cross-country data available forparticular sectors.

Prawda’s comparative review of educationaldecentralisation in four Latin American countries inthe 1980s – Argentina, Chile, Colombia, and Mexico –provides some insights into the equity and efficiencyimpacts of these reforms (Prawda 1993). From thiscomparative analysis Prawda concluded thatdecentralisation of education did not lead todiscernible quality improvements, but rather producednegative equity effects, with the result that the gapbetween better off and worse off schools actuallywidened. Educational expenditures fell in three of thefour countries (with the exception of Argentina) onaccount of sharp decreases in teachers’ salaries, underconditions of fiscal austerity, which may well haveimpacted adversely on teaching quality.

Four key lessons arise from Prawda’s review:(1) educational decentralisation does notautomatically accomplish productivity, equity, andquality improvements; (2) it requires a lengthygestation period before it starts producing benefits;(3) continuous changes of senior personnel in centraland local administrations are inimical to reform; and(4) an expansion in private provision has widened theperformance gap between schools and incomegroups (Prawda 1993: 262). He argues that fiscalincentives should be built into the decentralisationprocess to stimulate the performance of localgovernments by rewarding local revenue-raisingefforts and penalising severe budgetary deficits.


1Intro38.1.qxd 11/01/2007 10:29

Chile is the only Latin American country for whichdata is available on cognitive achievement in selectedsubjects, which serves as a proxy for determining theimpact of the reforms on quality and equity ineducation provision. According to Prawda, ‘asmeasured by cognitive achievement results, it is quiteclear that quality did not improve in the 1982–1988period. It is also clear that inequity widenedsignificantly at that time’ (Prawda 1993: 258). Thesefindings are corroborated by Parry, who found thatdecentralisation and privatisation have exacerbatedthe negative consequences of educationaldecentralisation, resulting in greater inequity inexpenditures and greater differences in theperformance of students from different incomegroups (Parry 1997: 116–7). Declining real per capitaexpenditures and competition for students betweenmunicipal and private schools also had some negativeconsequences for equity in the 1980s but remedialmeasures introduced by civilian governments after1992 counterbalanced these trends (Parry 1997:128–9). Municipal councils and mayors are nowelected and municipalities have been provided withadditional funds to cover service outlays, butdiscretionary power to raise additional resourcesremains limited (Stewart and Ranis 1994).

Regarding healthcare, evidence from six LatinAmerican countries indicates that the quality ofservice provision has worsened underdecentralisation. Transfer of financial resources andstaff to lower levels of government neither improvedservice delivery nor reduced the costs of care (Burki etal. 1999: 75–86). Chile provides some evidence on theequity effects of decentralisation and privatisation ofhealthcare provision under the military regime in the1980s (Gideon 2001). One review concluded that ‘Ingeneral, the transfer of primary care clinics tomunicipalities has not resulted in extending coverageor in improving the quality of services, largely becauseof a lack of professional supervision and poor healthplanning by the area health services’ (Montoya-Aguilarand Vaughan 1990). Despite vigorous efforts topromote private health provision and to delegateresponsibility for public healthcare provision to themunicipalities, two-thirds of all medical consultationsand 80 per cent of hospitalisations were still state-funded in the mid-1990s, supported by 7 per cent taxon earnings and pensions. Problems continued toaffect the quality of public healthcare provisionthrough municipalities, despite measures to improvetargeting and resourcing: ‘Although low-income

earners receive “free” healthcare, “access is difficult,waiting times are long, services are of poor quality,and facilities and provision of pharmaceuticalsmeager”’ (Gillion and Bonilla, cited in Tankersley andCuzán 1996: 113). However, since it is difficult todisaggregate the effects of decentralisation fromprivatisation and fiscal constraints the problems ofpublic health provision under the municipalitiescannot easily be attributed to local administrativearrangements alone.

Colombia is the one other Latin American country forwhich evidence on the impact of decentralisation onservice delivery is available. In response to growingsocial protests over the declining quality of publicservices, the Colombian government devolvedresponsibility for public services to electedmunicipalities, and sharply increased inter-governmental transfers and revenue-raising powersfrom the late 1980s (Forero and Salazar 1991: 122).Local governments assumed responsibility for theprovision of services in education, health, water,sanitation, roads and agricultural extension. Theevidence suggests that satisfaction levels withmunicipal governments increased after theintroduction of direct elections for mayors in 1988.Case studies of individual municipalities and opinionsurveys ‘found evidence of increased service coverage,citizen satisfaction, attention to rural areas and thepoor, cost consciousness and resource mobilizationefforts’ (Fiszbein 1997: 1030). There is some evidencefrom this research of a positive relationship betweenthe strength of community participation andgovernment performance: municipalities thatfollowed a more open and inclusive approach topolicymaking were positioned to achieve betteroutcomes. The majority of individuals surveyed in asample of 16 municipalities believed that municipalgovernments play a central role in the provision ofeducation, water and roads. An overwhelmingmajority reported greater trust in local than nationalgovernment and a larger number of individuals preferthe municipal government to be in charge of overallservice provision (Fiszbein 1997: 1035). Municipalitiesassumed responsibility for public education after 1991with the formation of councils composed of teachers,parents and students to run local schools. Councilswere given the right to elect principals, but hiringremained under the control of the Ministry ofEducation. Autonomous regions, communities andschools were given power to adapt curricula, raisingconcerns about fragmentation (Astiz et al. 2002: 75).


1Intro38.1.qxd 11/01/2007 10:29

Comparative evidence on health and educationaldecentralisation in Latin America in the 1980s and1990s points to several conclusions. First, it is difficultto derive hard and fast generalisations on the basis ofpartial and incomplete evidence. Second, theimplications for equity have been negative, withdivergences between poorer and wealthier groups inaccessing health and educational services. This findingis consistent with the broader literature oneducational decentralisation. In this respect, Dyer andRose (2005: 107) state, ‘The assumed benefits ofdecentralisation are … contested in relation to equity,for which it is often found to have negativeconsequences. Decentralization can widen qualitydifferences between schools, and performance gapsbetween students, in wealthy and poor areas’.Similar observations hold true for the health sector.Third, as the contrasting cases of Chile and Colombiademonstrate, political and institutional conditionshave a significant bearing on decentralisationoutcomes and levels of user satisfaction. This point istaken up at greater length in the concluding sectionof the article.

2.2 Sub-Saharan AfricaThe evidence from sub-Saharan Africa is very limitedand even more qualified as regards the equity impactof decentralised service delivery (Conyers, in this IDSBulletin). Despite the inclusion of decentralisation inpublic sector reform efforts in the 1980s and early1990s by countries such as Uganda, Botswana,Nigeria, Ghana, Côte d’Ivoire, Kenya and Tanzania,one leading commentator has stated that ‘there areno real success stories as far as improveddevelopment performance at the local level isconcerned’ (Adamolekun, cited in Francis and James2003). This stark finding is corroborated by Wunsch(2001), who attributes to failure of decentralisation inAfrica to problems such as the over-centralisation ofresources, limited transfers to sub-nationalgovernments, a weak local revenue base, lack oflocal planning capacity, limited changes in legislationand regulations, and the absence of meaningful localpolitical process. These dismal assessments arereflected in studies of local governance anddecentralised service provision from a number ofcountries in the region (Olowu and Wunsch 2004).

Uganda is one African country that has pursued apotentially far-reaching decentralisation experimentsince the late 1980s, with increased availability ofresources for national social service programmes,

especially for education, health and drinking waterinfrastructure channelled through local councils. Butthe evidence suggests that ‘Decentralization has notbeen able to arrest the deterioration in agriculturalservices, and that the improvements in social servicesare attributable to increases in central conditionalfunding rather than the very limited scope whichdecentralized institutions have provided for localdecision making’ (Francis and James 2003: 333).

In Côte d’Ivoire, new opportunities were created forpopular participation through the introduction ofmulti-party competition for local council (commune)elections, but the mayors continued to exertoverriding control and influence. As a result, thepreferences expressed by local people for roads,social facilities and water supplies did not correspondto spending priorities of the communes, whichfocused on municipal buildings and secondaryschools. In any case, most commune developmentprogrammes collapsed in the face of public spendingcuts during the financial crisis of the early 1990s. It istherefore unsurprising that only one-third of thoseinterviewed in four sample communes felt that thecommunes addressed their development needs(Crook and Sverrisson 2001: 26). A similar findingemerged from Ghana, where survey evidence fromtwo districts demonstrated that 70 per cent ofrespondents felt that the elected assembly did notrespond to their needs. Expressed preferences forroad repairs, health facilities, water supplies andelectricity were not reflected in district assemblyexpenditure priorities which focused on commercialtransport services, farming, manufacturingenterprises or markets, a situation exacerbated bythe dominance of recurrent expenditures in districtbudgets (Crook and Sverrisson 2001: 32). In Nigeria,a study of primary healthcare in the early 1990srevealed a complete lack of real participation indecision making despite devolution of responsibilityto elected local officials. Local residents saw primaryhealthcare as unreliable, ineffective and unresponsiveto their needs, while councillors were unclear of thehealth needs of their constituents, and had littleknowledge of health plans and activities (Crook andSverrisson 2001: 32).

This brief review indicates that the evidence on theservice delivery outcomes in Africa is even moreslender than for Latin America. Moreover, from thelimited evidence that exists, there is little to indicatethat the various decentralisation experiments under


1Intro38.1.qxd 11/01/2007 10:29

way across the continent are generating the expecteddevelopment dividends and that citizens have yet tosee any real improvement in service quality.

2.3 South AsiaEvidence from Asia is very limited, largely becausedecentralisation initiatives in countries of the regionare more recent in origin, and because in mostSouth Asian countries, health and education serviceshave only been devolved to a limited extent. Thereare few comparative studies of service deliveryoutcomes resulting from decentralisation in theregion, and limited insights from sector experience inparticular countries. Recent country studies ofhealthcare spending under decentralisation in China,India, Indonesia and the Philippines point to a declineor stagnation after decentralisation started in thesecountries. In China and India, local governmentswere unable to fulfil their new responsibilities forhealthcare provision in the absence of inadequateresource transfers from central government. But incontrast, health outcomes in Indonesia and thePhilippines improved significantly duringdecentralisation, reflected in a sharp decline in theunder-five mortality rate, largely because of reformsin healthcare funding (OECD 2006).

Drawing on survey data from 33,000 households invillages across India, Mahal et al. (2000) demonstratethat decentralisation of public service delivery inprimary healthcare and education services ispositively correlated with improved child mortalityand school enrolment. However, health andeducation services in India are generally under thejurisdiction of state governments and local councilshave limited influence over the use of resources ordeployment of personnel. Elected councils havelimited discretion over the use of resources fordevelopmental purposes, which are largelyearmarked for schemes and programmes determinedby state and central governments. It is only in theIndian states of West Bengal and Kerala thatdecentralisation of expenditures for basic services hastaken place on a significant scale, by placingsubstantial untied funds at the discretion of localvillage councils for developmental purposes. In theWest Bengal case, sample evidence points toimprovements in access to administrative and justicesystems and water provision in some areas, amid anoverall improvement in agricultural productivity andreduction in poverty levels in the 1980s, though it isdifficult to attribute these outcomes to

decentralisation and increased powers and resourcesfor elected local councils (Crook and Sverrisson2003: 243). Preliminary evidence from Kerala’sPopular Planning Campaign launched in 1996indicates that local council expenditures moreaccurately reflected local preferences, andinvestments in infrastructure were more orientedtowards the needs of the poor (Isaac with Franke2000; Chaudhuri and Heller 2002).

By comparison, successive decentralisation schemesin Bangladesh have all failed to deliver improvedservices. According to Crook and Sverrisson(2001: 46), ‘Material welfare, in terms of agriculturaloutput, did not increase, there was little evidence ofgreater equity at grassroots level, and a number ofstudies indicated that the beneficiaries were the richand the well-born. Instead, decentralization wasgenerally seen as a means to channel developmentresources into the hands of the better off’.Responsibility for implementation of disaster reliefprogrammes was devolved to local councils underthe military government in the mid-1980s, but withnegative impacts: flood rehabilitation programmessuffered from poor management, maldistribution,corruption and shortages of resources, while fewvery poor households received any benefits fromrehabilitation schemes that tended to focus on roads,bridges and buildings.

This brief and partial review of the experience ofdecentralised service delivery leads to the followingtentative conclusions. First, improved equity outcomeshave generally not been realised for poor and sociallymarginalised people. Second, the quality of publicservice provision has not improved as a result of thedevolution of power and resources to localgovernments. The gap in quality between wealthierand poorer areas has often increased underdecentralisation. Third, efficiency gains have beenrealised, usually as a result of the delegation of financialresponsibility for service provision from central to localgovernments, but resources have not been adequate toensure effective coverage and quality.

3 Improving equity through decentralisedservice deliveryIt is tempting to draw the conclusion that equity andsocial justice objectives are not well served bydecentralised service provision, and that centralisedprovision through deconcentrated state agencies is apreferable approach (Johnson 2001;


1Intro38.1.qxd 11/01/2007 10:29


Schneider 2003). At the same time, the availableevidence confirms that increased participation andbetter accountability can result from democraticdecentralisation, and that these substantive benefitsshould not be underestimated (Crook and Manor1998; Blair 2000). A poor record on service deliveryto date does not rule out the scope for improvedequity and efficiency outcomes. Rather, thechallenge is to identify the conditions under whichincreased participation in local governance isconducive to enhanced equity, quality and efficiencyof services. This will almost certainly require furthercomparative research but it is possible to outline aschema in which the potential for improvement restson a combination of political, institutional, financialand technical factors.

3.1 Political commitment and leadershipPolitical factors are of intrinsic importance todecentralised service delivery for several reasons. It iswidely accepted that political commitment on thepart of federal or state governments is a sine qua nonof effective democratic decentralisation, andespecially forms of decentralisation that arespecifically geared to the interests of the poor(Crook and Sverrisson 2001; Blair 2000). Successfulpro-poor decentralisation is associated withgoverning parties that are politically committed tothe democratic empowerment of local governments(Heller 2001; Escheverri-Gent 1993).

The Indian state governments of West Bengal andKerala evince a strong commitment to decentralisation,reflected in supportive legislation and a significant flowof resources to lower levels of government. InColombia successive governments from the mid-1980shave systematically devolved powers and resources tomunicipalities, with positive consequences for servicedelivery. Brazilian experience demonstrates howpolitical commitment at the level of individualmunicipalities can explain a propensity for pro-poorreform initiatives, such as the participatory budgetingprocess in Porto Alegre and other municipalities(Baiocchi 2001; Heller 2001). In contrast, evidence fromAfrica and other Latin American countriesdemonstrates that weak political commitment todecentralisation opens up the possibility of élitecapture, limits the scope for participation, and results inineffective outcomes (Smith 1985).

Political leadership also plays an important role inshaping service delivery outcomes, since politicians in

local governments do not respond with equal vigourto the opportunities presented by high-level politicalcommitment to democratic decentralisation. This isespecially important in local administrations withpowerful, directly elected mayors who have theauthority to effect or block change mandated byhigher level political authorities. For instance, in theColombian case, mayors committed to deepeningthe process of municipal decentralisation throughpublic consultation, and enhanced resource flowregistered higher levels of public satisfaction withservice delivery outcomes. According to Fiszbein(1997: 1032), ‘competition for political office openedthe doors to responsible and innovative leadershipthat became the driving force behind capacitybuilding. It was the combination of the addedresponsibilities, more resources and political reformsthat created the environment conducive to theemergence of effective local governments’.

3.2 Political mobilisation of the poorThe political impetus for democratic decentralisationcreated by reform-minded political parties can createopportunities for collective action from below bymobilising constituencies that are traditionallyexcluded from national policymaking arenas. This canentail mobilisation of cadres and supporters bypolitical parties in local constituencies, andmobilisation of the poor by civil society organisations(NGOs, trade unions and social movements) to takeadvantage of political openings from above and toarticulate public protest and dissent.

Party-based mobilisation can assume two forms inthe context of democratic decentralisation:mobilisation of people though local units of politicalparties for electoral purposes and mobilisation ofsupporters to ensure effective implementation ofreform initiatives. Democratic decentralisation usuallyentails the devolution of power to elected localauthorities, which in turn widens the scope of politicalparticipation at the local level (Robinson 1998). Inmany Latin American countries, municipalities weretraditionally run by non-elected administratorsappointed by military or authoritarian regimes.Legislation introduced from the 1980s led to thecreation of elected mayors and local councils,providing opportunities for political mobilisationaround competing policy agendas. However, politicalparties are not always allowed to contest localelections (e.g. Uganda, India and Pakistan), whichmust be contested on an individual or no-party basis,

1Intro38.1.qxd 11/01/2007 10:29

thus limiting the scope for party-based mobilisation.However, in Brazil, Kerala and West Bengal, localgovernment elections serve as a basis for party-based mobilisation around competing politicalagendas, though the extent to which these hinge onservice delivery issues is not apparent.

Civil society organisations also mobilise constituenciesin local government jurisdictions to take advantage ofincreased powers and resources, to mobilise peopleto take part in consultative arenas, and to engage inpublic protest over the quality of public services. InKerala, a prominent social movement (Kerala SastraShitya Parishad – the People’s Science Movement)played a critical role in shaping and implementing thePeople’s Campaign for Decentralized Planning in thelate 1990s, though this has not been replicatedelsewhere in India (Isaac with Franke 2000;Chaudhuri and Heller 2002). Civil society mobilisationin response to increased powers to local governmentsis particularly marked in Latin America, which mayreflect traditions of political resistance toauthoritarian rule, but also resource availability at thelocal level. Social movements and trade unions playeda part in mobilising protest over the state ofmunicipal services in Colombia in the 1970s, whichserved as a catalyst for subsequent reform of localgovernment. In the city of Cochabamba in Bolivia,civil society organisations helped to articulate publicdemonstrations over water privatisation and servicecharges, resulting in remedial measures by themunicipal administration. Similarly, several SouthAfrican municipalities have witnessed civil society-ledprotests over service standards and fees.

3.3 Institutionalised participationLocal authorities in different countries haveexperimented with institutional arrangementsdesigned to facilitate public engagement, feedbackand oversight in recognition of the latent power oforganised civic protest. These include consultativebodies designed to provide citizen oversight overparticular services, taking the form of health councilsand school boards. Prominent Latin Americanexamples include the local administrative boards inColombia, local area boards in São Paolo and theneighbourhood councils of Montevideo, through tomore ambitious exercises designed to elicitparticipation in decisions concerning priority settingand resource allocations, exemplified by theparticipatory budgeting exercises in Porto Alegre andother Brazilian cities (Nickson 1995: 86–9).

Critics argue that such bodies serve to underminepopular resistance and oversight, while theirproponents claim that institutionalised participationfacilitates and widens public engagement at the locallevel in policy deliberation, planning andimplementation. However, in the absence of anycomparative evidence it is difficult to ascertain eitherthe prevalence of these bodies or their effectiveness ininfluencing resource allocations or service standards.

3.4 Adequacy of financial resourcesThe availability of financial resources is a criticaldeterminant of the equity, quality and efficiency ofpublic services, and the inadequacy of financialresources often contribute to poor service deliveryoutcomes. Devolution of responsibility for serviceprovision to local governments is usuallyaccompanied by some element of financialdecentralisation through resource transfers, usually asa share of central taxation, or enhanced powers toraise revenues through a variety of local taxes (Bahland Linn 1994; Bird and Vaillancourt 1999). Fiscaldecentralisation often renders local governmentsvulnerable to macroeconomic shocks and remedialmeasures to control public expenditures and nationalbudget deficits. Several Latin American countriesexperienced this phenomenon in the 1980s becauseof economic stabilisation measures, which sharplyreduced spending on the social sectors and the valueof transfers to local governments. The quality andreach of public services is bound to suffer in theabsence of complementary measures to raise localresources. The financial imperative has been a keyfactor underlying municipal privatisation initiativesand the introduction of cost-sharing measures in theform of user fees in local governments around theworld (Mawabu et al. 2001).

Another dimension of resource availability centres onthe financial powers of local governments. Salariesand recurrent expenditures tend to account for a largeshare of local government outlays on services,especially in the health and education sectors, withmore limited resources available for capitalexpenditures. Limited scope for discretionaryallocations across budget heads further restricts thebudgetary autonomy of local governments. Localgovernments may also receive financial transfers thatare earmarked for certain programmes orpre-assigned categories of expenditure. In India, forexample, local bodies receive grants-in-aid from stateand central government that are tied to specific anti-


1Intro38.1.qxd 11/01/2007 10:29

poverty and social welfare programmes, whilerecurrent expenditures account for a very highproportion of health and education budgets. Thefinancial autonomy of local governments is thus highlyconstrained. A major exception is Kerala, where theelected local councils have discretion over 40 per centof the state development budget, subject to broadguidelines on different categories of expenditure,which provides them with substantial scope torespond to locally determined development priorities.

3.5 Technical and managerial capacityThe provision of public services can be an enormouslycomplex exercise, especially in urban municipalitieswith large populations, and often requires a highlevel of technical and managerial capacity. However,decentralisation of responsibility for service provisionhas not always been accompanied by measures toensure effective capacity for planning, budgeting,implementation and monitoring in localgovernments, all of which have a critical bearing onservice quality. Efforts to strengthen the professionaland technical skills of local government employeesand to improve the internal organisation andmanagement style of local administration are oftencentral to building such capacity (Fiszbein 1997; Dyerand Rose 2005).

Managerial and technical capacity is not only a keydeterminant of the performance of local officials inrelation to service delivery, but also influences theirbehaviour towards users of services. Centralisedservice delivery through hierarchically organised linedepartments and deconcentrated agencies gives riseto behavioural norms that may not be conducive toparticipation and greater responsiveness. Creating anorganisational culture in local government that ismore citizen-friendly and receptive to activecommunity involvement, as well as performanceoriented, requires a combination of incentives andfocused capacity-building measures to complementthe strengthening of technical and managerial skills.

4 ConclusionThis article has sought to ascertain the impact ofdecentralised service delivery on equity and efficiency

outcomes. Subject to constraints of data availability,two main conclusions arise from a review of availableevidence in less-developed countries: (1) the qualityand equity of access have not improved with thedecentralisation of health and education services; and(2) equity and efficiency outcomes are closely relatedto the availability of financial resources and localgovernment capacity.

These insights tend to give rise to two types of policyprescription, neither of which is closely compatiblewith democratic decentralisation: (1) health andeducation services are better administered bydeconcentrated public agencies working under thedirect control of central line departments, and(2) expanding the role of private providers andintroducing user fees can improve quality andefficiency of resource use. However, experiencesuggests that while efficiency gains may be realised,neither of these approaches is conducive toparticipation in local governance, nor are theyguaranteed to produce outcomes that are morefavourable to equity and social justice objectives.

The challenge for proponents of democraticdecentralisation is to specify methods andapproaches by which equity objectives can berealised under decentralised forms of servicedelivery. Successful interventions are not premisedon participation and accountability alone, butrequire attention to political factors (such ascommitment, leadership and mobilisation),institutional arrangements, financial resources andtechnical and managerial capacity. Greater emphasisshould be given to measuring and monitoringservice delivery outcomes under decentralised formsof provision, to ensure that participation in localgovernance produces real gains for the poor interms of improved access and quality of services.Failure to do so will undermine the allure ofdemocratic decentralisation and encourage policyalternatives that run counter to the ethos ofparticipation in local governance and the potentialfor improving service provision for the benefit ofthe poor.


1Intro38.1.qxd 11/01/2007 10:29

ReferencesAstiz, M.F., Wiseman, A.W. and Baker, D.B. (2002)

‘Slouching Towards Decentralization: Consequencesof Curricular Control in National EducationSystems’, Comparative Education Review 46.1: 66–88

Azfar, O., Kähkönen, S., Lanyi, A., Meagher, P. andRutherford, D. (2004) ‘Decentralization,Governance and Public Service: The Impact ofInstitutional Arrangements’, in M.S. Kimenyi andP. Meagher (eds), Devolution and Development:Governance Prospects in Decentralizing States,Aldershot: Ashgate

Bahl, R. and Linn, J. (1994) ‘Fiscal Decentralizationand Intergovernmental Transfers in LessDeveloped Countries’, Publius – Journal of FiscalFederalism 24.1: 1–19

Baiocchi, G. (2001) ‘Participation, Activism andPolitics: The Porto Alegre Experiment andDeliberative Democratic Theory’, Politics andSociety 29.1: 43–72

Bird, R. and Vaillancourt, F. (eds) (1999) FiscalDecentralization in Developing Countries,Cambridge: Cambridge University Press

Blair, H. (2000) ‘Participation and Accountability atthe Periphery: Democratic Local Governance inSix Countries’, World Development 28.1: 21–39

Burki, S.J., Perry, G.E. and Dillinger, W.R. (1999)Beyond the Center: Decentralizing the State,Washington DC: The World Bank, World BankLatin American and Caribbean Studies

Chaudhuri, S. and Heller, P. (2002) ‘The Plasticity ofParticipation: Evidence from a ParticipatoryGovernance Experiment’, mimeo, Department ofEconomics, Columbia University

Crook, R.C. and Manor, J. (1998) Democracy andDecentralisation in South Asia and West Africa:Participation, Accountability and Performance,Cambridge: Cambridge University Press

Crook, R.C. and Sverrisson, A.S. (2003) ‘DoesDecentralisation Contribute to PovertyReduction? Surveying the Evidence’, inP.P. Houtzager and M. Moore (eds), ChangingPaths: International Development and the NewPolitics of Inclusion, Ann Arbor: University ofMichigan Press: 233–59

Crook, R.C. and Sverrisson, A.S. (2001) Decentralisationand Poverty-Alleviation in Developing Countries: AComparative Analysis or, is West Bengal Unique?, IDSWorking Paper 130, Brighton: IDS

Dyer, C. and Rose, P. (2005) ‘Decentralisation forEducational Development? An EditorialIntroduction’, Compare 35.2: 105–13

Escheverri-Gent, J. (1993) The State and the Poor:Public Policy and Political Development in India andthe United States, Berkeley and Los Angeles:University of California Press

Fiszbein, A. (1997) ‘The Emergence of Local Capacity:Lessons from Colombia’, World Development 25.7:1029–43

Forero, H. and Salazar, M. (1991) ‘Local Governmentand Decentralization in Colombia,’ Environmentand Urbanisation 3.2: 121–6

Francis, P. and James, R. (2003) ‘Balancing RuralPoverty Reduction and Citizen Participation: TheContradictions of Uganda’s DecentralizationProgram’, World Development 31.2: 325–37

Gideon, J. (2001) ‘The Decentralization of PrimaryHealth Care in Chile’, Public Administration andDevelopment 21.3: 223–31

Heller, P. (2001) ‘Moving the State: The Politics ofDemocratic Decentralisation in Kerala, SouthAfrica, and Porto Alegre’, Politics and Society 29.1:131–63

Isaac, T. with Franke, R.W. (2000) Local Democracyand Development: People’s Campaign forDecentralized Planning in Kerala, New Delhi:LeftWord

Johnson, C. (2001) ‘Local Democracy, DemocraticDecentralisation and Rural Development:Theories, Challenges and Options for Policy’,Development Policy Review 19.4: 521–32

Litvack, J., Ahmad, J. and Bird, R. (1998) RethinkingDecentralization in Developing Countries,Washington DC: The World Bank, PovertyReduction and Economic Management Network

Mahal, A., Srivastava, V. and Sanan, S. (2000)Decentralization and Public Sector Delivery of Healthand Education Services: The Indian Experience,Discussion Papers on Development Policy 20,Bonn: Centre for Development Research,University of Bonn

Montoya-Aguilar, C. and Vaughan, P. (1990)‘Decentralization and Local Management of theHealth System in Chile’, in Mills, A., Vaughan, J.P.,Smith, D.L. and Tabibzadeh, I. (1990) Health SystemDecentralization: Concepts, Issues and CountryExperience, Geneva: World Health Organization(WHO): 55–63

Mwabu, G., Ugaz, C. and White, G. (eds) (2001) SocialProvision in Low-income Countries: New Patternsand Emerging Trends, Oxford: Oxford UniversityPress

Nickson, R.A. (1995) Local Government in LatinAmerica, Boulder: Lynne Rienner


1Intro38.1.qxd 11/01/2007 10:29

Olowu, D. and Wunsch, J.S. (eds) (2004) LocalGovernance in Africa: The Challenges of DemocraticDecentralisation, Boulder: Lynne Rienner

One World Action (2001) Developing Gender-SensitiveLocal Services, report of a seminar organised byOne World Action and the British Council

One World Action (1999) Influence and Access – LocalDemocracy and Basic Service Provision, report of aseminar organised by One World Action

OECD (Organisation for Economic Co-operation andDevelopment) (2006) Decentralisation in AsianHealth Systems: Friend or Foe?, Policy Insights 18,Paris: OECD Development Centre, May

Parry, T.R. (1997) ‘Decentralization and Privatization:Education Policy in Chile’, Journal of Public Policy17.1: 107–33

Prawda, J. (1993) ‘Educational Decentralization inLatin America: Lessons Learned’, InternationalJournal of Educational Development 13.3: 253–64

Robinson, M. (1998) ‘Democracy Participation andPublic Policy: The Politics of Institutional Design’,

in M. Robinson and G. White (eds), The DemocraticDevelopmental State: Politics and InstitutionalReform, Oxford: Oxford University Press

Schneider, A. (2003) Who Gets What from Whom?The Impact of Decentralisation on Tax Capacity andPro-Poor Policy, IDS Working Paper 179, Brighton:IDS

Smith, B. (1985) Decentralisation: The TerritorialDimension of the State, London: George, Allen andUnwin

Stewart, F. and Ranis, G. (1994) Decentralization inChile, Occasional Paper 14, New York: UnitedNations Development Programme, HumanDevelopment Report Office

Tankersley, W.B. and Cuzán, A.G. (1996) ‘Privatizationand Decentralization in the United States andChile’, Journal of Developing Societies XII.1: 104–18

Wunsch, J.S. (2001) ‘Decentralization, LocalGovernance and “Recentralization” in Africa’,Public Administration and Development 21.4: 277–88


1Intro38.1.qxd 11/01/2007 10:29

1 IntroductionAs Robinson’s article in this IDS Bulletin notes,decentralisation is frequently advocated as a meansof improving public service delivery, but there is verylittle evidence to determine whether or not this isactually the case. Nowhere is this paradox truer thanin sub-Saharan Africa. As Mutizwa-Mangiza(2000: 24) says, ‘many African countries have jumpedon the bandwagon of decentralisation andparticipation without even assessing their ownexperiences’. This article attempts to bring togetherthe evidence that does exist about the impact ofdecentralisation on service delivery in the region andgives general lessons that can be learned.

This article is divided into five main sections.Section 2 provides a brief historical overview ofdecentralisation in sub-Saharan Africa. It notes thatdecentralisation has played an important role in manystages of the region’s history, but that its form andobjectives have changed significantly over time.Section 3 considers the relationship betweendecentralisation and service delivery. It points outthat decentralisation’s impact on service delivery isindirect, in the sense that it affects a number ofintermediate factors (access to local information,locus of decision-making power, resource availabilityand administrative performance), which in turn affectservice delivery. Section 4 provides an overview ofthe nature and extent of the evidence on the impactof decentralisation on service delivery in sub-SaharanAfrica. It maintains that there is a dearth ofinformation about its ultimate impact on servicedelivery in sub-Saharan Africa, but a considerableamount of data about its impact on the intermediatefactors. Section 5 summarises the main findings thatemerge from both types of data, while the finalsection draws some conclusions about the factors

that affect the impact on service delivery and theimplications of this for policymaking. The mainconclusion is that, although decentralisation has notyet had a significant positive impact on the quality ofpublic services in the region, this is due primarily tothe wider policy environment rather than todecentralisation per se.

Before proceeding with the analysis, threequalifications about the scope of the article must bemade. First, any attempt to analyse the impact of aconcept as broad and vague as ‘decentralisation’ isinevitably fraught with problems. The term is used torefer to anything from the deconcentration ofadministrative responsibilities within a singlegovernment agency to the devolution of power overall basic local services to semi-autonomous localauthorities. It is also used to describe the transfer ofpower to a wide range of geographical levels, fromthe regional or state level to that of localgovernments or communities. Some restriction offocus is therefore necessary. This article thus, likeother contributions to this IDS Bulletin, focusesprimarily, although not exclusively, on devolutionrather than deconcentration and on the intermediate‘local government’ level. Second, sub-Saharan Africais a large and diverse region, so any attempt to drawgeneralisations is equally problematic. The articlefocuses on those countries for which secondary datais most easily available,1 and those of which thewriter has personal experience, and it is biasedtowards Anglophone countries. Third, the articledoes not pretend to provide a comprehensive reviewof all relevant literature. It aims merely to define thenature and extent of the evidence about the impactof decentralisation on service delivery andsummarise, with the help of examples, the mainfindings.

18

Decentralisation and ServiceDelivery: Lessons from Sub-Saharan Africa

Diana Conyers


1Intro38.1.qxd 11/01/2007 10:29

2 Decentralisation in sub-Saharan Africa: anhistorical overviewIt is neither necessary nor possible to provide adetailed account of the history of decentralisation insub-Saharan Africa.2 However, in order to understandthe relationship between decentralisation and servicedelivery in the region, some historical background isrequired.

At the risk of considerable oversimplification, sixmain historical periods may be noted:3

1 Pre-colonial During the pre-colonial period,African government was relatively decentralised,consisting either of small chiefdoms or of muchlarger but loosely organised kingdoms or states.However, within these administrative entities,there was an element of centralisation, in thatindividual leaders, such as chiefs or kings, held aconsiderable amount of personal power

2 Colonial Although there were significantdifferences between the various colonial powers,there were two common characteristics ofcolonial regimes: first, power was highlycentralised in the colonial authority; and second,this power was exercised through some form of‘indirect rule’,4 so elements of the decentralisedpre-colonial systems remained

3 Transition During the last few years of thecolonial period, Western-style local governmentswere established in many countries, particularlythose under British rule, where local governmentwas seen as a means of introducing people toWestern concepts of democracy and thuspreparing them for self-government

4 Post-independence 1: Centralisation In mostcountries the period immediately afterindependence was one of centralisation. This wasjustified (both by governments and by externalfunders and advisers) on the grounds that centralpolicymaking and planning were necessary tobring about the rapid economic and socialtransformation required

5 Post-independence 2: Deconcentration After afew years, however, many governments began toadopt some degree of decentralisation, primarilyas a means of improving the quality of localservice delivery. During this period,

decentralisation tended to take the form ofdeconcentration rather than devolution, in thatpowers were transferred not to semi-autonomous local governments but to institutionsover which the central government retainedcontrol. Particularly common, especially in theone-party states that characterised much of theregion at the time, was the decentralisation ofpower to regional and local developmentcommittees, comprising a combination ofcentrally appointed and locally elected officials

6 Post-independence 3: Devolution Over the lasttwo decades, decentralisation has maintained itspopularity but there have been significant changesin both its objectives and its form. Although stilladvocated as a means of improving servicedelivery, decentralisation has also been seen (bygovernments, external actors and the increasinglyinfluential civil society lobbies) as a means ofenhancing democracy and citizen participationand (by governments and external actors) as a wayof reducing the role, and in particular theexpenditure, of the central government. This hasbeen reflected in a change in emphasis fromdeconcentration to devolution (often known as‘democratic decentralisation’) during this period.

It is evident, therefore, that decentralisation is notnew to sub-Saharan Africa and that it has been animportant part of the development agenda for muchof the post-independence period. However, it isequally evident that there have been major variationsin the forms that decentralisation has taken and in itsobjectives. Of particular significance is the fact that,in recent years, decentralisation has been advocatedas a means of achieving three related, butsignificantly different, types of objective: improvedservice delivery, democracy and participation, and areduction in central government expenditure. Theimplications of this in terms of its impact on theformer are discussed below.

3 The relationship between decentralisation andservice deliveryThe theoretical relationship between decentralisationand service delivery is discussed in the article byRobinson in this IDS Bulletin (and in much of theliterature on decentralisation that relates to sub-Saharan Africa).5 This section merely highlights twokey points that affect the analysis in subsequentsections.


1Intro38.1.qxd 11/01/2007 10:29

First, decentralisation does not affect the quantity,quality or equity of public services directly, butthrough its effect on the following intermediate (or‘process’)6 variables:

Access to local information Decentralisation hasthe potential to increase access to informationabout local needs, conditions and priorities, whichare then incorporated into local development plans

Locus of decision-making powerDecentralisation should localise the power tomake and implement decisions, and thus totranslate plans into programmes of action

Resource availability Decentralisation mayincrease the amount of resources available forimplementing programmes, especially financialresources

Administrative performance Decentralisationmay enhance administrative performance and thusthe effectiveness of programme implementation.

Second, decentralisation does not necessarily havethe above effects. As Ludeki (2004: 19) says: ‘Acountry can adopt decentralised … structures but failto realise development. Conversely, a country canachieve breakthroughs in development at the locallevel under highly centralised, even authoritarian,administrative structures’. There are two mainreasons for this. One is that the actual effects ofdecentralisation depend on a number of factors,including the type of public service concerned, thedetailed ‘design’ of the decentralisation, the way inwhich it is implemented, the capacity of the variousindividuals and organisations involved, and the widereconomic, social and political environment. Theother is that decentralisation is not the only factorthat affects service delivery. Other governmentpolicies and the broader policy environment areoften equally, if not more, important.

4 Decentralisation and service delivery insub-Saharan AfricaThere is a vast literature on decentralisation in Africa.Most of it consists of detailed studies of individualcountries, often restricted in both temporal andtopical scope. However, there are a number ofcomprehensive studies, which attempt to makecomparisons between countries and draw broaderconclusions. These are of three main types:

Collections of country studies These consist of anumber of country studies (some includingcountries from other regions), together with oneor more synthesis chapters. The studies are oftennot strictly comparable and the quality of thesynthesis varies considerably

Proceedings of regional conferences There havebeen many conferences on decentralisation inAfrica, organised by regional and/or internationalorganisations and attended by delegates from anumber of countries in the region.7 Theproceedings vary considerably in terms of thedepth and objectivity of analysis

Systematic cross-country studies (e.g. Crook2003; Crook and Manor 1998; Mehrotra 2006;Ribot 2002, 2003). These are potentially the mostuseful but they are relatively few in number andvary in focus; moreover, some include case studiesfrom other regions.

Unfortunately, this vast and varied literature providesvery little specific information on the impact ofdecentralisation on the quantity, quality or equity ofpublic services in the region. Many studies do not tryto analyse the impact of decentralisation, while othersmerely make broad generalisations about, or briefreferences to, the impact on service delivery. Examplesof the latter are Crook and Manor (1998), Olowu andWunsch (2004), Oyugi (2000a) and Ribot (2003), allof which are concerned with the wider impact onwhat Oyugi calls ‘good governance and development’,and Crook (2003),8 which focuses on the impact onpoverty rather than service delivery per se.

The few studies that do provide detailed data onservice delivery tend to be confined to specificsectors or programmes within particular countries.Examples include Mehrotra’s (2006) analysis of healthservices in Benin, Guinea and Mali; Olowu andWunsch’s (2004) study of health services in Nigeria;Fass and Desloovere’s (2004) account of education inChad; studies of Kenya’s Local Authority TransferFund by Mitullah (2004b) and Smoke (2004); andreports on Uganda’s Local GovernmentDevelopment Programme by Kiyaga-Nsubugu (2004)and Onyach-Olaa (2003). Schroeder’s (2003) analysisof the division of functions between the variouslevels of government in South Africa is somewhatdifferent, in that it looks at all public services; but itfocuses on the factors that should be taken into

Conyers Decentralisation and Service Delivery: Lessons from Sub-Saharan Africa20

1Intro38.1.qxd 11/01/2007 10:29

account when allocating functions rather than onactual performance.

This dearth of information reflects the difficulty ofobtaining detailed, systematic data about the impacton service delivery – or about many other outcomesof decentralisation (Crook 2003; Ribot 2003;Therkildsen 1993a). There are two sets of problem.One is the logistical complexity and cost ofundertaking the necessary research, which ideallyrequires ‘before’ and ‘after’ studies in a number ofdifferent countries. The other is the complexity ofthe relationship between decentralisation and servicedelivery (discussed in the previous section), whichmakes it difficult to attribute any observed changesin the latter to decentralisation reforms.

Fortunately, however, there is a substantial amountof information about the impact of decentralisationon the intermediate variables – access to localinformation, locus of decision-making power,resource availability and administrative performance –which provide the hypothetical link betweendecentralisation and improved service delivery. It isnot possible here to discuss the literature on theseintermediate variables in any depth.9 However, thereare three comparative studies that warrant particularmention. They are Crook and Manor’s (1998) study,which, although covering only two countries inAfrica (Ghana and Côte d’Ivoire), provides unusuallydetailed information; Olowu and Wunsch’s (2004)collection of case studies, which are diverse in naturebut are analysed in depth using a common analyticalframework; and Ribot’s (2003) comprehensive anddetailed analysis of the theory and practice ofdemocratic decentralisation in Africa.

5 Decentralisation and service delivery insub-Saharan Africa: the findingsThis section summarises the main findings concerningthe impact of decentralisation on service delivery inthe region. It is divided into five subsections. The firstone summarises the limited evidence regarding theimpact on service delivery itself, while the othersconsider the evidence about each of the mainintermediate variables identified above, namely accessto local information, locus of decision-making power,resource availability, and administrative performance.

5.1 Impact on service deliveryThe main impression gained from the limited data onthe actual impact on service delivery is that

decentralisation has done little to improve thequantity, quality or equity of public services in theregion. For example, Walter Oyugi (2000b: 20), aveteran observer of decentralisation in Africa,concludes that decentralisation ‘has failed to act as aspur to democratic development management andefficient and effective delivery of services’, whileRibot (2003: 10) notes that ‘evidence thatdecentralization or deconcentration leads to betterservice delivery is thin’, and Crook (2003) andMitullah (2004a) find little evidence that it hasenhanced the position of the poor. Studies of publicperception of the quality of local government serviceprovision also tend to be negative. For example,Fjeldstad (2001: 294) found that at least two-thirdsof people interviewed in two local authorities inTanzania described council services as ‘bad’, whileCrook and Manor (1998: 186, 255) found that 70 percent of interviewees in Ghana and 54 per cent inCôte d’Ivoire considered their local authorities unableto ‘satisfy the needs of their area’.10

The few documented cases of a positive correlationbetween decentralisation and service delivery arelimited in scope and subject to qualification. Manycome from the health sector, where decentralisationhas been actively promoted at the international levelfor many years (Mills 1990). For example, Mehrotra(2006: 269, 278–9) reports that the decentralisationof primary healthcare services to locally electedhealth committees in Guinea, Mali and Benin and tolocal governments in Mozambique, has increasedaccess to affordable health services, which has in turnincreased immunisation rates and reduced infantmortality. Similar claims are made by Olowu andWunsch (2004) in relation to the decentralisation ofprimary healthcare services to local governments inNigeria between 1988 and 1995, and by Andrews andSchroeder (2003) in the case of Niger. However, allthese writers report variations in performance fromone area to another and many operational problems.Moreover, in all the cases decentralisation wasconfined to one sector and in only two (Mozambiqueand Nigeria) were powers decentralised to localgovernments. Another sector for which there is someinformation is road maintenance. Andrews andSchroeder (2003) quote evidence from the WorldBank (1994: 74–8) that decentralisation of roadmaintenance responsibilities can improve both thespeed and quality of service provision.11 However, theyemphasise that this is not necessarily the case andthat, as with health services, the impact depends on


1Intro38.1.qxd 11/01/2007 10:29

the extent and form of decentralisation and the wayin which it is implemented.

A rather different example is Fass and Desloovere’s(2004) account of community control over primaryeducation in Chad. In this case, local residents inareas where the government had failed to provideprimary education decided to take the matter intotheir own hands and organised and funded theirown schools. However, one may question whetherthese are really an example of decentralisation, sincethe community involvement was a voluntary initiativethat stemmed from the state’s failure to provideservices. Moreover, when the state found out whatthe communities were doing, it restricted their roleby requiring that community schools meet minimumnational standards.12

There are also reports of positive outcomes from thedecentralisation of funds for capital development tolocal authorities. Two such examples are Uganda’sLocal Government Development Programme(Kiyaga-Nsubugu 2004; Onyach-Olaa 2003) andKenya’s Local Authority Transfer Fund (Mitullah2004b; Smoke 2003). In this case, however, theimpact is on infrastructure rather than servicedelivery per se; in fact, one could argue that thedecentralisation of funds for the construction ofinfrastructure without comparable measures toimprove operation and maintenance can create asmany problems as it solves.

Uganda also illustrates a rather different type of‘success’ story. Decentralisation of responsibility forprovision of most local public services to districtcouncils has been accompanied by significantimprovements in the quantity and (albeit to a lesserextent) quality of service provision, particularly in thecase of health and education facilities (Makara 2000;Uganda 2002a). However, it is difficult to prove acausal relationship between the two. In fact, it isquite likely that the increase is due primarily to thevast injection of donor funds that has been madeduring this period under Uganda’s Poverty AlleviationAction Plan, rather than to decentralisation.

The Ugandan case illustrates the problem (noted inthe previous section) of determining a causalrelationship between decentralisation and servicedelivery. It is important to note that this problemworks both ways. On the one hand, it means that, incases like Uganda, it is very difficult to attribute

improvements in service delivery to decentralisation.But it also means that, in countries wheredecentralisation has coincided with decreases in thequality of public service provision, there is little or noevidence to suggest that this is due todecentralisation. In other words, one should notblame decentralisation for the poor quality of serviceprovision in many African countries. As mostcommentators point out, the problems stem frommore fundamental characteristics of African states,which hamper any form of service delivery, whethercentralised or decentralised. This point will becomeclearer when we look at the evidence about theimpact of decentralisation on the intermediatevariables that in turn determine the impact onservice delivery – in other words, when we examinethe process of decentralisation.

5.2 Access to local informationThere is a considerable amount of information aboutthe extent to which and ways in whichdecentralisation enables information on local needs,conditions and priorities to be accessed andincorporated into local development plans. However,it is not always easy to extract this information fromthe broader debate about the relationship betweendecentralisation and popular participation, of whichit is a part. Our concern here is with participation asa means of enhancing the quality and relevance ofdevelopment plans – and thus potentially the‘allocative efficiency’ of resource use, rather than as ademocratic right or means of citizen empowerment.13

Several conclusions emerge from this information.First, the quality of information depends on whoparticipates, which in turn depends on thecomposition of the institutions to which power isdecentralised (Conyers 1999; Ribot 2003). Over theyears, powers have been decentralised to a widerange of institutions in Africa, including centralgovernment field staff, ‘arms-length’ managementbodies, political representatives of the centralgovernment, elected local authorities, ‘traditional’leaders and a variety of community-basedorganisations, and also to ‘composite’ bodies (such asthe regional and local development committeescharacteristic of ‘one-party state models’ ofdecentralisation) composed of any combination ofthe above.

Second, no one of these institutions is necessarily‘best’ in terms of representing and responding to


1Intro38.1.qxd 11/01/2007 10:29

local interests. Effectiveness depends not just on thetype of institution but on its structure andcomposition, the motivation and capacity of theindividuals involved, and local and national powerstructures. Of particular significance is the findingthat devolution is not necessarily more effective thandeconcentration. The move towards ‘democraticdecentralisation’ in many African countries over thelast two decades was initially welcomed both withinthe region and outside, since it had become evidentthat the earlier forms of decentralisation, which tookthe form of deconcentration rather than devolution,were little more than an instrument for nationalpolitical control (Wunsch and Olowu 1995; Oyugi2000b). In many cases, however, the results havebeen disappointing. The elected authorities haveturned out to be little more representative orresponsive than their predecessors and the centralgovernment has maintained much of its control overlocal development (Crook 2003; Olowu and Wunsch2004; Oyugi 2000b; Ribot 2003).

The problem stems from the manner in whichelected local government representatives achieveand maintain their political power, which in turnreflects the ‘patronage-based’ nature of bothnational and local politics.14 Most representativesobtain their support from a combination of twosources: the ruling political party, to which they willhave to belong in order to stand for election, andlocal élites, who more often than not are alsomembers of the ruling party. Consequently, theytend to be more concerned about maintaining theirallegiance to these groups, which in effect meanspromoting the interests of the ruling party, thanrepresenting the majority of their constituents. Inmany cases, the situation is exacerbated by lack ofpressure from civil society organisations, which areoften weak and sometimes part of the same politicalpatronage system (Kasfir 1998; Robinson andFriedman 2005). This problem is particularly welldocumented in Uganda (Francis and James 2003;Mwenda and Tangri 2005; Titeca 2005),15 but isprevalent in many other countries, including Kenya(Musyoki and Nyamu-Musembi 2005) and Ghana(Ayee 2004a,b). In Ghana, the problem isexacerbated by a high degree of central control overlocal governments, including the power to appointone-third of the members.

Third, there is an important difference between‘participation’ and ‘influence’. The evidence from

Africa supports the point made so clearly by Blair(2000), that it is relatively easy to increase thenumber and range of people who participate inlocal government and administration, but muchmore difficult to increase the extent to which theyinfluence decision making. Once again, the keyfactor is the local power structure, whichdetermines how decisions are actually made at thelocal level. The impact is particularly evident in thecase of deprived social groups, such as women,ethnic minorities, and the poor in general. Forexample, although several African countries (e.g.Mozambique, South Africa and Uganda) haveintroduced quota systems that guarantee women’srepresentation in local government, evidencesuggests that most women representatives have yetto have substantial influence over local decisionmaking because they lack the authority and self-confidence to participate on an equal basis withmen (Goetz 1998; Goetz and Hassim 2003; MDP-ESA 2003). Similar problems occur when power isdecentralised to bodies composed of very differenttypes of representatives, such as public servants andcommunity representatives or national and localpoliticians. In such cases, the more powerfulrepresentatives (in the above examples, publicservants and national politicians) tend to dominatedecision making (Cornwall 2004; Ribot 2003).

Finally, and on a more positive note, there isevidence to suggest that, despite the manyshortcomings identified above, the extent and qualityof participation and representation are graduallyincreasing. For example, most of the critics citedabove acknowledge that the establishment ofdemocratic local government institutions does atleast provide more space for participation than theformer deconcentrated structures, and that theprovision of quotas for disadvantaged groups is atleast a step in the right direction. Furthermore, manyefforts to promote more effective participation areunder way.16 They fall into two main categories, eachof which has a critical role to play in enhancing thequality of democratic decentralisation. Onecomprises attempts by local authorities (oftensupported by regional or international organisations)to establish more participatory forms of planning andbudgeting.17 The other category consists of efforts,usually by non-governmental organisations (NGOs)and civil society organisations, to increase thecapacity of local people to participate effectively.


1Intro38.1.qxd 11/01/2007 10:29

5.3 Locus of decision-making powerDebates about the quality of participation in locallevel planning are only meaningful if local authoritieshave the power to make and implement decisionsand thus translate plans into action. In this respect,the main conclusion that emerges from Africandecentralisation experience is that governmentshave, on the whole, been reluctant to decentralisesufficient power to local level governments toenable them to have significant impact on localservice delivery.

This problem is manifested in several different ways(Conyers 1999). In some cases, decentralisationconsists of no more than the decentralisation ofplanning powers. This is very common in countriesembarking upon decentralisation for the first time. Itwas a characteristic of many of the earlydecentralisation efforts, when ‘composite’ bodies(such as regional and district developmentcommittees) were the main decentralised institutionsand their principle role was to prepare ‘integrated’development plans (de Valk and Wekwete 1990). Insome countries, the limitations of this approach wererecognised and led to the decentralisation of moresubstantial powers; examples include Tanzania’s 1971decentralisation reforms (Conyers 1981) and Kenya’sDistrict Focus Programme (Barkan and Chege 1989).But in many other cases, there was no furtherdecentralisation and those involved in local levelplanning, frustrated by the inability to implement theirplans, often lost interest in the planning process; thiswas the fate of Zimbabwe attempts at decentralisedplanning in the 1980s (Makumbe 1999; Mutizwa-Mangiza and Helmsing 1991).

In other cases, local institutions are given the powerto make decisions – often over a wide range of publicservices – but are not given control over theresources needed to implement these decisions. Localgovernment legislation typically includes a long list oflocal government functions but control over theresources needed to exercise these functions remainscentralised. This problem, which is characteristic ofmany of the more recent devolution reforms, is welldocumented in Ghana (Ayee 2004a, 2004b; Crookand Manor 1998). Failure to decentralise financialcontrol is particularly critical – and particularlycommon (Conyers 1999; Olowu and Wunsch 2004;Ribot 2003; Smoke 2003).18 Moreover, the limitedresources that are made available tend to take theform of central government grants (most of which

are conditional) rather than revenue-raising powers.Uganda is one of the few countries where substantialfinancial powers have been decentralised, but evenhere, conditional grants constitute the main form oflocal government revenue.19 The other criticalresource, and one that has received much lessattention in the literature, is personnel. Very fewcountries have either decentralised control over thepublic servants currently engaged in service deliveryor given local governments the authority and (moreimportantly) resources to recruit their own staff. Themain exception once again is Uganda, where mostpublic servants at sub-national level are responsible todistrict councils rather than national line ministries.

Another manifestation of the problem is long delaysin implementing decentralisation reforms. There areoften long time-gaps between the political decisionto decentralise, the promulgation of the enablinglegislation, and the implementation of thislegislation. Moreover, the content of thedecentralisation reform is often ‘watered down’ ateach stage. Zambia is a good example of this.Decentralisation has been on the policy agenda sincethe 1960s, but each of the many decentralisationreforms has encountered implementation problems.Chikulo (2000) concludes that, although each reformmakes more progress than the one before, thecountry remains highly centralised and there hasbeen more deconcentration than devolution.

The final manifestation of the problem is a tendencyfor central governments to try to withdraw powersafter they have been decentralised. Uganda illustratesthis phenomenon. The decentralisation reformsimplemented in the late 1990s involved the transferof effective control over most local services to localgovernments, including control over financial andpersonnel resources. Subsequently, however, thecentral government has slowly but surely regained asignificant amount of control. This is most obvious inthe case of finance. The proportion of conditionalgrants has increased substantially (primarily due tothe provision of large grants for primary serviceprovision under a donor-funded national povertyreduction programme) and one of the main sourcesof local government revenue (the graduated tax) hasbeen abolished.20 However, there have also beenattempts to reclaim control over district councilpersonnel, most importantly by giving the centralgovernment the power to appoint chiefadministrative officers.


1Intro38.1.qxd 11/01/2007 10:29

5.4 Resource availabilityThe hypothetical case for decentralisation suggeststhat it may increase the availability of resources(especially financial resources) in four main ways:(1) opening up new sources of tax revenue thatcannot feasibly be exploited by a centralisedadministration; (2) improving the collection ofexisting taxes; (3) facilitating contributions from thegeneral public (including both user fees and voluntarycontributions of money, materials or labour); and(4) reducing the cost of service provision and thusgenerating surpluses that can be used for otherpurposes. Unfortunately, there is little substantiveinformation about the extent to whichdecentralisation has achieved these objectives in sub-Saharan Africa. The information that does exist ispiecemeal and consists largely of casual observationor ‘hearsay’, rather than objective statistical data. Thissection summarises the information that does existregarding each of the hypotheses posed above.21

There have been relatively few cases wheredecentralisation has opened up new sources of taxrevenue. There are probably three main reasons forthis: the limited number of additional taxes that canfeasibly be exploited, especially in poor and/orpredominantly rural areas; the reluctance of bothcentral and local governments to increase taxationfor political reasons; and the tendency for centralgovernments to want to retain as much control over,and access to, tax revenue as possible. Although localgovernments are often reluctant to increase taxationfor political reasons, the central government issometimes the main obstacle. For example, asalready indicated, the Uganda Government recentlyabolished one of the main sources of localgovernment revenue, despite strong objections fromlocal governments. Similarly, in Zimbabwe in the late1990s, local governments applied to the centralgovernment to introduce several new taxes (e.g.taxes on land), but were refused permission.However, there is evidence from both Zambia andZimbabwe of local councils introducing new taxeswhen allowed to do so (Mellors 2006).

There is circumstantial evidence of improvedcollection of existing taxes following decentralisationin a number of countries; for example, Uganda,Malawi and Zimbabwe.22 The improvement appearsto be due partly to increased motivation on the partof local government officials, but also to generalimprovements in administrative efficiency. However,

most of the evidence is less positive. Korsun andMeagher (2004) found the quality of revenuecollection in Guinea, Mali and Senegal to begenerally poor; Crook and Manor (1998) found similarproblems in Côte d’Ivoire and Ghana; and Therkildsen(1993b) reported declining revenue collection in anumber of countries. In many of these cases, thequality of revenue administration appeared again toreflect that of administration and governance ingeneral, while in Côte d’Ivoire and Ghana many ofthe problems seemed to stem from centralgovernment controls over local government revenueraising. Some of the most detailed information aboutlocal revenue administration comes from Fjeldstadand Semboja’s study of Tanzania’s local developmentlevy (Fjeldstad 2001). This suggests that councilemployees are more effective revenue collectorsthan local politicians or community-based taxcollection agents, since the latter are sensitive to thenegative impact of taxation on their political positionand susceptible to corruption and bias. There aresimilar findings about the shortcomings of localpoliticians and tax collection agents in Uganda(Francis and James 2003; Uganda 2002a) andSenegal (Juul 2006). However, evidence fromZimbabwe suggests that, when councillors are fullyinvolved in preparing the council budget, theirattitude often changes (Mellors 2006).

Although increases in user fees have often occurredat the same time as decentralisation in Africancountries, there is little evidence to suggest a causallink between them. This is probably because policyon user fees is generally made by central rather thanlocal governments, and tends to be determined bythe state of the national economy and/or donorpolicy. For example, in the 1980s and early 1990s,user fees were introduced in many countries, but aspart of structural adjustment policies rather thandecentralisation reforms. Similarly, in Zimbabwe thegovernment has decentralised control over someservices and increased user fees in recent years, butboth moves have been in response to its increasinglyacute financial situation and have not beenaccompanied by improvements in the quality ofservice provision (Conyers 2003).

There is often a positive link betweendecentralisation and voluntary contributions by thegeneral public, especially when decentralisationextends down to the community level (SNV 2006). Itappears that, given a choice, most people would


1Intro38.1.qxd 11/01/2007 10:29

prefer to contribute to specific projects from whichthey will directly benefit than to pay general taxes toa local authority. The case study of primary educationin Chad, cited earlier, is an extreme example of thelengths to which people are prepared to go toobtain goods or services that they value highly (Fassand Desloovere 2004). Similar evidence comes fromSenegal (Juul 2006). However, voluntarycontributions are not only associated withdecentralised government; well-organised centralgovernment campaigns and NGOs can both mobiliseself-help efforts. Furthermore, there is not always apositive relationship between the amount of self-help and the quality of service delivery. Thus, in thecase of Chad, it was a response to the government’sinability to provide basic services and it was merely acase of ‘some form of schooling is better than none’.Similarly, in Zimbabwe, the promotion ofcommunity-based maintenance of rural watersupplies was in large part a response to thegovernment’s inability to fund those services itself,and was again accompanied by a general decline inthe quality of service provision (Conyers 2003), whilein Ghana, where the promotion of self-help by localgovernments was again a response to financialconstraints at the national level, Crook and Manor(1998) found that the quality of self-help projectswas often poor.

Turning to the final hypothesis, there is little if anyinformation on the impact of decentralisation on thecost of service provision in Africa.23 However,experience in Uganda (Mwenda and Tangri 2005) andelsewhere suggests that, except in cases wheredecentralisation has coincided with a drastic reductionin public expenditure and thus in the quality of services,as was the case in Zimbabwe (Conyers 2003) andGhana and Côte d’Ivoire (Crook and Manor 1998), it isunlikely to result in significant cost savings. There arethree main reasons for this. First, the transfer offunctions to local governments is seldom accompaniedby a significant reduction in central government staff.This is due partly to general problems of ‘downsizing’,but also to the fact that decentralisation often createsnew functions for central government staff (such asmonitoring) or means that jobs previously undertakenby one field officer now require two – one toundertake the central government component of thejob and the other the local government component.24

Second, even in relatively well-run local governments,decentralisation tends to result in an increase inoverhead costs at the local level, because of the

additional administrative and political functions thathave to be performed. And third, decentralisation oftenresults in an increase in the number of local authorities,and thus in total overhead costs. In Uganda, forexample, the number of districts has more thandoubled since the decentralisation process began.25

5.5 Administrative performanceThe hypothetical link between decentralisation andadministrative performance is based on a number ofpremises. Of particular importance are the claimsthat decentralisation may increase flexibility andtherefore responsiveness, improve coordinationbetween the various agencies involved, andencourage integrity and responsibility among localofficials. Once again there is insufficient evidence toeither support or refute the validity of these claims insub-Saharan Africa. In this case, the problem is notmerely the quantity and quality of data but also thedifficulty of measuring the concept of ‘administrativeperformance’.

The main impression from the literature is thatadministrative performance under decentralisedsystems of governance is poor in most countries ofthe region. For example, Crook and Manor (1998)found the quality of administration in their twoAfrican case studies (Côte d’Ivoire and Ghana)generally inferior to that in the two Asian cases.Olowu and Wunsch (2004) are critical ofperformance in all but one of the seven countriesthey studied (the exception being Botswana), and (asalready indicated) Korsun and Meagher (2004)suggest that poor financial administration in thethree countries they studied was largely a reflectionof poor administration in general.

The potential benefits of decentralisation appear tohave been undermined by a number of factors(Crook 2003; Crook and Manor 1998; Olowu andWunsch 2004; Oyugi 2000b,c; Ribot 2003), inparticular:

Inadequate devolution of power, particularly overfinance and staffVague and/or inappropriate systems andproceduresInadequately qualified, underpaid and unmotivatedstaffPolitical ‘interference’, corruption and abuse ofpowerLack of ‘downward’ accountability.


1Intro38.1.qxd 11/01/2007 10:29

However, there is a need to put these findings intoperspective. There is, in particular, a need to askwhether administrative performance has actuallydeteriorated as a result of decentralisation andwhether the quality of administration at the locallevel is worse than performance at the national level.In most cases, the answer to both these questions isprobably ‘no’. The main conclusion that emerges isthat administrative performance at local level is, to alarge extent, a mirror of that in the country as awhole. As Oyugi (2000c: 16) concludes, ‘studies doneon decentralization tend to suggest that more oftenthan not, the problems that bedevil the national-level institutions and processes find similarexpressions at the local level in any scheme ofdecentralization’.

The Ugandan experience illustrates this point. Thereis no doubt that administrative performance at thelocal level has improved dramatically in Uganda overthe last two decades. In 1986, when the Musevenigovernment came to power and the presentdecentralisation policy was born, there was noeffective administration whatsoever at local level.However, it is difficult to determine how much (ifany) of this improvement can be attributed todecentralisation, as opposed to improvements in thequality of public administration as a whole and amarked increase in financial resources. Furthermore,the many administrative problems that remain at thelocal level, of which corruption and other abuses ofpower are perhaps the most obvious, are alsocharacteristic of the Ugandan administration as awhole.

Some evidence about the potential impact ofdecentralisation on administrative performance isavailable from the substantial number of pilot,donor-supported local government capacity-buildingprogrammes that have been implemented in anumber of African countries over the last twodecades (Conyers 2005; DIP 2002; Fjeldstad 2001;Romeo 2003; UNCDF 2003).26 In most of thesecases, there has been no national decentralisationreform, but unconditional grants for capitaldevelopment projects have been made available topilot local governments, together with capacity-building support in the form of training andorganisational development. The short-term impacton administrative performance has generally beenpositive, including measurable improvements in thequality of coordination, planning, project

implementation, and both financial and generalmanagement. Three main factors appear to havebeen critical to this success: the provision of general-purpose development funds (especially if linked toperformance criteria), which gave local authoritiesthe incentive and the power to improve resourceallocation procedures; the complementary capacity-building support, particularly if provided in a‘facilitatory’ rather than didactic manner and focusedon organisational change rather than just training;and the adoption of a flexible ‘process’ approach toprogramme implementation by donors. However,the longer-term results have been less positive.Difficulties have been experienced in ‘scaling up’ pilotprojects into nationwide programmes, in sustainingbenefits when capacity-building support is removed,and in replacing donor funds with permanentsources of local government revenue.

Finally, there is also positive evidence about thepotential for enhancing administrative performancethrough increasing downward accountability. Theimportance of downward accountability isincreasingly recognised and many attempts are beingmade to mobilise civil society organisations andcommunity groups to hold local governments toaccount. Most of these are being promoted byNGOs (both national and international) but some bynational and regional local governmentorganisations.27 Documentation of these efforts (see,for example, Conyers and Cumanzala 2004; Goetzand Gaventa 2001; Kajimbwa et al. 2005; Mushamba2000; Musyoki and Nyamu-Musembi 2005;Robinson and Friedman 2005: 15; SNV 2006;Uganda 2002a) demonstrates that pressure frombelow can have a positive impact on performanceand suggests a variety of possible techniques that canbe used.

6 ConclusionAt first sight, the main conclusion to emerge fromthis brief overview may appear to be a negative one.It seems that, as far as one can tell from the limitedevidence available, the many years of decentralisationexperience in sub-Saharan Africa have failed to havea positive impact on service delivery. However, it isnecessary to qualify this conclusion in four ways.

First, it is important to reiterate the problems ofdrawing any general conclusions about therelationship between decentralisation and servicedelivery, since so much depends on the type of


1Intro38.1.qxd 11/01/2007 10:29

service, the type of decentralisation, the way inwhich it is implemented, and the broader policyenvironment. As Olowu and Wunsch (2004: 123)aptly put it, ‘the devil is in the detail’.

Second, one could argue that, despite the manyyears of experience, decentralisation has not reallyhad a fair trial in Africa because there have been veryfew countries where significant powers, especiallyover finance, have actually been decentralised tolocal governments. The problem of centralgovernments’ reluctance to decentralise is notunique to Africa. It reflects the fundamental natureof decentralisation. As Smith points out,decentralisation is the ‘territorial dimension of thestate’ and the form that any decentralised system ofgovernment takes is ‘the result of political forces inconflict’ (Smith 1985: 201).

Third, it is not fair to blame decentralisation for thepoor quality of service delivery in much of the region

because most of the weaknesses of localgovernments – including their lack of power – are areflection of the problems of governance in general.Experience with decentralisation tells us a great dealabout the nature of governance in many Africancountries, especially the high level of centralisationand relative lack of accountability, and suggests thatthe problems of decentralisation (like so manyothers) cannot be addressed in isolation.

Finally, and on a more positive note, it is importantto note that the history of decentralisation has notbeen static. There have been a number of positivechanges over the years, including the move to moredemocratic forms of local governance, recognition ofthe need for fiscal decentralisation, and the manyrecent attempts to increase citizen participation anddownward accountability. This in turn suggests thatthere is a need to see decentralisation as part of along, slow process of state building – and thus to berealistic about what it can be expected to achieve.


Notes1 Those countries for which there is most

information include Côte d’Ivoire, Ghana, Kenya,Nigeria, South Africa, Tanzania, Uganda andZimbabwe (compare Crook 2003: 78).

2 For useful summaries, see Pasteur (1999); Olowuand Wunsch (2004, Ch. 2); Ribot (2003).

3 It is difficult to attach dates to these periodsbecause the timing varies from country tocountry, depending in particular on the date ofindependence. There has been a tendency forcountries that attained independence relativelylate to go through the same post-independencephases as their predecessors, albeit often morequickly. This suggests that, although externalfactors have undoubtedly played a part, theevolution of decentralised systems of governmentis part of a wider process of ‘state-building’.

4 That is, through pre-colonial administrativestructures or, where such structures were weakor non-existent, through structures set up toreplicate them.

5 See, for example: Andrews and Schroeder (2003);Conyers (1999); Kimenyi and Meagher (2004);Mehrotra (2006); Olowu and Wunsch (2004);Oyugi (2000a); Ribot (2003); Smoke (2003).

6 The term ‘process variables’ is used by Olowu andWunsch (2004).

7 Three regional organisations have played aparticularly important role in organising suchconferences, and in promoting decentralisationand local government in the region. They are theAfrican Union of Local Authorities, the MunicipalDevelopment Programme, and the African officeof the United Nations Centre for RegionalDevelopment.

8 Much of the data in this article is based on awider study, including countries in other regions;see Crook and Sverrisson (2001, 2003).

9 For a comprehensive bibliography, see Ribot(2003).

10 In the case of Ghana, a later survey by Ayee(2004b: 83) found that 52 per cent of peoplewere satisfied with the level of service provision,suggesting that performance may have improved.However, Ayee is somewhat surprised by this,since his general conclusion is that ‘decentralisedgovernment has fallen far short of reducingpoverty’.

11 The World Bank findings do not relate only toAfrica. They cover 42 developing countries andthere is no breakdown of findings by region.However, the report includes a case study fromEthiopia, where road maintenance wassuccessfully decentralised to a local community-based organisation.

1Intro38.1.qxd 11/01/2007 10:29

12 For other interesting examples of communityintervention when state service provision fails, seeGoetz and Gaventa (2001: 24) and Lund (2006).

13 The ‘allocative efficiency’ argument fordecentralisation, often used by economists, isbased on the premiss that ‘local governments willlikely be better able to match public goods tolocal preferences’, because they have betterinformation about local preferences and are morelikely to respond to local demands (Azfar et al.2004: 22). For further explanation of thisargument and discussion of its validity, see Azfaret al. (2004) and Mehrotra (2006).

14 The term ‘patronage’ refers to a situation wherepoliticians achieve and/or maintain their positionby giving favours (e.g. political positions, jobs,preferential treatment, local developmentprojects, cash handouts) to their supporters. InAfrica, it is often referred to as ‘neo-patrimonialism’, a term that suggests that itresembles and/or has roots in the pre-colonial‘patrimonial’ relationship between chiefs andcitizens. For various perspectives on its nature andimpact in the region, see Hyden (1983), Brattonand van de Walle (1997) and Chabal and Daloz(1999).

15 Titeca’s paper is particularly interesting. It providesa detailed case study of the way in which theNational Resistance Movement, which operatedfor many years as a de facto ruling party,dominates most local institutions in westernUganda, including local authorities and civilsociety organisations.

16 See, for example: ACPDT (2002); Conyers andCumanzala (2004); Goetz and Gaventa (2001);Kajimbwa et al. (2005); Mushamba (2000);Musyoki and Nyamu-Musembi (2005); SNV (2006).

17 The concept of ‘participatory budgeting’ has beenintroduced into Africa following the widespreadpublicity given to its adoption in the city of PortoAlegre in Brazil (Baiocchi 2003). It is beingpromoted by the Participatory BudgetingKnowledge and Action Support Facility for Africa,established by the Municipal DevelopmentPartnership for Eastern and Southern Africa(MDP-ESA), in collaboration with African localauthorities and the World Bank Institute (fordetails, see www.asaaf.org.zw/asaaf.htm).

18 Data on 22 African countries compiled by Bahland Smoke (2003: 13) revealed that in 15countries, the proportion of governmentexpenditure channelled through localgovernments was less than 6 per cent. It should,however, be noted that in some of the othercountries the proportion was much higher andthat in seven countries it was considerably higherthan one would expect from comparative analysisof other countries with similar conditions.

19 Information on Uganda in this section of thearticle is based primarily on personal observations.For further information on recent fiscaldecentralisation strategy, see Uganda (2002b).

20Moves to reduce the degree of conditionality ofthe grants within sectors are currently under way,but they will remain conditional.

21 Where no references are given in this subsection,information is based on my own personalobservations.

22 This assertion is based largely on unreported data,including personal visits to local authorities inthese three countries and personal communicationfrom colleagues working at this level.

23 This is part of a wider problem of lack ofinformation on the cost of service provision ingeneral, which, as Bahl and Smoke (2003) note,hampers any attempt at fiscal decentralisation.

24 An example of this is education, where generaladministration becomes a local governmentfunction, while inspection remains a centralgovernment function.

25 According to Mwenda and Tangri (2005: 457), thenumber of districts increased from 33 in 1986,when the NRM Government came to power, to56 in 2003; and at the time of writing (2006) ithad increased to 76 (Larok, pers. comm. 2006).

26 The countries concerned include Zambia andZimbabwe (supported by the UK’s Departmentfor International Development), Malawi andUganda (supported by the United Nations CapitalDevelopment Fund), and Tanzania (supported bythe Netherlands Government).

27 An example of the latter is the MunicipalDevelopment Partnership for Eastern andSouthern Africa (MDP-ESA), which has workedwith local authorities in a number of countries topromote such activities.


1Intro38.1.qxd 11/01/2007 10:29

ReferencesACPDT (Africa Community Publishing and

Development Trust) (2002) Local Governance andParticipation, Harare

Andrews, M. and Schroeder, L. (2003) ‘SectoralDecentralisation and IntergovernmentalArrangements in Africa’, Public Administration andDevelopment 23.1: 29–40

Ayee, J. (2004a) ‘Ghana: a Top-Down Initiative’, inD. Olowu and J.S. Wunsch (eds), Local Governancein Africa: The Challenges of DemocraticDecentralization, London and Boulder: LynneRienner

Ayee, J. (2004b) ‘Decentralised Governance andPoverty Reduction at the Local Level in Ghana’,Regional Development Dialogue 25.1: 71–86

Azfar, O., Kahkonen, S., Lanyi, A., Meagher, P. andRutherford, D. (2004) ‘Decentralization,Governance and Public Services: The Impact ofInstitutional Arrangements’, in M.S. Kimenyi andP. Meagher (eds), Devolution and Development,Aldershot: Ashgate

Bahl, R. and Smoke, P. (2003) ‘Overview of FiscalDecentralization in South Africa’, in R. Bahl andP. Smoke (eds), Restructuring Local GovernmentFinance in Developing Countries: Lessons from SouthAfrica, Cheltenham: Edward Elgar

Baiocchi, G. (2003) ‘Participation, Activism andPolitics: The Porto Alegre Experiment’, in A. Fungand E. Wright (eds), Deepening Democracy,London and New York: Verso, Ch. 2

Barkan, J. and Chege, M. (1989) ‘Decentralising theState: District Focus and the Politics of Re-allocation in Kenya’, Journal of Modern AfricanStudies 27.3: 431–53

Blair, H. (2000) ‘Participation and Accountability atthe Periphery: Democratic Local Governance inSix Countries', World Development 28.1: 21–39

Bratton, M. and van de Walle, N. (1997) DemocraticExperiments in Africa: Regime Transitions inComparative Perspective, New York: CambridgeUniversity Press

Chabal, P. and Daloz, J.-P. (1999) Africa Works:Disorder as Political Instrument, London:International African Institute, in conjunction withJames Curry and Indiana University Press

Chikulo, B.C. (2000) ‘Decentralization for GoodGovernance and Development: The ZambianExperience’, Regional Development Dialogue 21.1:26–48

Conyers, D. (2005) ‘The Role of Aid in the MDGLocalisation Process’, paper presented at a

conference on ‘MDG Localisation in Africa:Options and Experiences’, Kampala, August

Conyers, D. (2003) ‘Decentralisation in Zimbabwe:A Local Perspective’, Public Administration andDevelopment 23.1: 115–25

Conyers, D. (1999) Decentralisation: A ConceptualAnalysis, Ministers’ Conference on LocalGovernment in Eastern and Southern Africa,Victoria Falls, Zimbabwe (reproduced in MDP-ESANewsletter 2000)

Conyers, D. (1981) ‘Decentralisation for RegionalDevelopment: A Comparative Study of Tanzania,Zambia and Papua New Guinea’, PublicAdministration and Development 1: 107–20

Conyers, D. and Cumanzala, F. (2004) ‘CommunityEmpowerment and Democracy in Zimbabwe:A Case Study of Binga District’, Social Policy andAdministration 38: 383–98

Cornwall, A. (2004) ‘New Democratic Spaces? ThePolitics and Dynamics of InstitutionalParticipation’, IDS Bulletin 35.2: 1–10

Crook, R.C. (2003) ‘Decentralisation and PovertyReduction in Africa: The Politics of Central–LocalRelations’, Public Administration and Development23.1: 77–88

Crook, R.C. and Manor, J. (1998) Democracy andDecentralisation in South Asia and West Africa,Cambridge: Cambridge University Press

Crook, R.C. and Sverrisson, A. (2003) ‘DoesDecentralization Contribute to Poverty Reduction?’,in P. Houtzager and M. Moore (eds), Changing Paths:Institutional Development and the New Politics ofInclusion, Ann Arbor, MI: University of Michigan Press

Crook, R.C. and Sverrisson, A. (2001) Decentralisationand Poverty Alleviation in Developing Countries: AComparative Analysis, or is West Bengal Unique?,IDS Working Paper 130, Brighton: IDS

DIP (Development in Practice) (2002) CapacityBuilding for Decentralised Local Governance inZimbabwe: Lessons from PDSP and RDCCBP,Nottingham: DIP

Fass, S.M. and Desloovere, G.M. (2004) ‘Chad:Governance by the Grassroots’, in D. Olowu andJ.S. Wunsch (eds), Local Governance in Africa: theChallenges of Democratic Decentralization, Londonand Boulder: Lynne Rienner

Fjeldstad, O.-J. (2001) ‘Taxation, Coercion andDonors: Local Government Tax Enforcement inTanzania’, Journal of Modern African Studies 39.2:289–306

Francis, P. and James, R. (2003) ‘Balancing RuralPoverty Reduction and Citizen Participation:


1Intro38.1.qxd 11/01/2007 10:29

Contradictions in Uganda’s DecentralizationProgram’, World Development 31.2: 325–38

Goetz, A.-M. (1998) ‘Fiddling with Democracy:Translating Women’s Participation in Politics inUganda and South Africa into Gender Equity inDevelopment Practice’, in M. Robinson andG. White (eds), The Democratic DevelopmentalState, Oxford: Oxford University Press

Goetz, A.-M. and Gaventa, J. (2001) Bringing CitizenVoice and Client Focus into Service Delivery, IDSWorking Paper 138, Brighton: IDS

Goetz, A.-M. and Hassim, S. (eds) (2003) No Shortcutsto Power: African Women in Politics and Policy-Making, London: Zed Books

Hyden, G. (1983) No Shortcuts to Progress, London:Heinemann

Juul, K. (2006) ‘Decentralisation, Local Taxation andCitizenship in Senegal’, Development and Change37.4: 821–46

Kajimbwa, M., Lwaitama, A. and Kaijage, P. (eds)(2005) Capacity Implications for Good LocalGovernance: Experiences and Prospects, Dar esSalaam University Press

Kasfir, N. (ed.) (1998) Civil Society and Democracy inAfrica, London: Frank Cass

Kimenyi, M.S. and Meagher, P. (eds) (2004) Devolutionand Development, Aldershot: Ashgate

Kiyaga-Nsubugu, J. (2004) ‘Local Governance forPoverty Reduction in Uganda: An Overview’,Regional Development Dialogue 25.1: 175–94

Korsun, G. and Meagher, P. (2004) ‘Failure byDesign? Fiscal Decentralization in West Africa’, inM.S. Kimenyi and P. Meagher (eds), Devolution andDevelopment, Aldershot: Ashgate

Ludeki, C. (2004) ‘The State and Rural Development:Transcending the Centralization–DecentralizationDebate’, Regional Development Dialogue 25.1: 18–32

Lund, C. (ed.) (2006) ‘Twilight Institutions and LocalPolitics in Africa’, Development and Change 37:4,special issue

Makara, S. (2000) ‘Decentralization for GoodGovernance and Development: Uganda’sExperience’, Regional Development Dialogue 21.1:73–92

Makumbe, J. (1999) ‘Zimbabwe: Decentralization,Development and Democracy’, in P.S. Reddy (ed.),Local Government, Democratisation andDecentralisation: A Review of the Southern AfricanRegion, Cape Town: Juta

Mehrotra, S. (2006) ‘Governance and Basic SocialServices: Ensuring Accountability in ServiceDelivery through Deep Democratic

Decentralization’, Journal of InternationalDevelopment 18: 263–83

Mellors, R. (2006) Personal communication, basedon experience of local authority capacity-buildingprogrammes in Zambia and Zimbabwe

Mills, A. (1990) Health System Decentralisation:Concepts, Issues and Country Experience, Geneva:World Health Organization (WHO)

Mitullah, W.V. (2004a) ‘Local Governance andPoverty Alleviation in Africa’, RegionalDevelopment Dialogue 25.1

Mitullah, W.V. (2004b) ‘Participatory Governance forPoverty Alleviation in Local Authorities in Kenya:Lessons and Challenges’, Regional DevelopmentDialogue 25.1: 88–105

MDP-ESA (Municipal Development Partnership forEastern and Southern Africa) (2003) Africa LocalGovernance Programme: Regional Scan ofDecentralisation Issues and Cross-Cutting Themes inthe Eastern and Southern African Region, Ch. 5,Harare: MDP-ESA

Mushamba, S. (2000) Programme on CommunityEmpowerment and Social Inclusion ThroughStrengthening Civic Participation in MunicipalGovernance, Harare: Municipal DevelopmentPartnership for Eastern and Southern Africa

Musyoki, S. and Nyamu-Musembi, C. (2005)‘Defining Rights from the Roots’, IDS Bulletin 36.1:100–9

Mutizwa-Mangiza, N.D. (2000) ‘Decentralization forGood Governance and Development: Comment’,Regional Development Dialogue 21.1: 23–5

Mutizwa-Mangiza, N.D. and Helmsing, A.H.J. (eds)(1991) Rural Development and Planning in Zimbabwe,Aldershot: Avebury

Mwenda, A. and Tangri, R. (2005) ‘Patronage Politics,Donor Reforms and Regime Consolidation inUganda’, African Affairs 104/416: 449–67

Olowu, D. and Wunsch, J.S. (eds) (2004) LocalGovernance in Africa: The Challenges of DemocraticDecentralization, London and Boulder: LynneRienner

Onyach-Olaa, M. (2003) ‘The Challenges ofImplementing Decentralisation: RecentExperiences in Uganda’, Public Administration andDevelopment 23.1: 105–14

Oyugi, W.O. (ed.) (2000a) ‘Decentralization andCitizen Participation in Africa’, RegionalDevelopment Dialogue 21.1

Oyugi, W.O. (2000b) ‘Decentralization for GoodGovernance and Development: The UnendingDebate’, Regional Development Dialogue 21.1: iii–xix


1Intro38.1.qxd 11/01/2007 10:29

Oyugi, W.O. (2000c) ‘Decentralization for GoodGovernance and Development: Concepts andIssues’, Regional Development Dialogue 21.1: 3–22

Pasteur, D. (1999) ‘Democratic Decentralisation: AReview of the African Experience’, in P.S. Reddy(ed.), Local Government, Democratisation andDecentralisation: A Review of the Southern AfricanRegion, Cape Town: Juta

Ribot, J. (2003) African Decentralization: Local Actors,Powers and Accountability, Geneva: United NationsResearch Institute for Social Development

Ribot, J. (2002) Democratic Decentralization of NaturalResources: Institutionalising Popular Participation,Washington DC: World Resources Institute

Robinson, M. and Friedman, S. (2005) Civil Society,Democratisation and Foreign Aid in Africa,Discussion Paper 383, Brighton: IDS

Romeo, L.G. (2003) ‘The Role of External Assistancein Supporting Decentralisation Reform’, PublicAdministration and Development 23.1: 89–96

Schroeder, L. (2003) ‘Municipal Powers andFunctions: The Assignment Question’, in R. Bahland P. Smoke (eds), Restructuring Local GovernmentFinance in Developing Countries: Lessons from SouthAfrica, Cheltenham: Edward Elgar

Smith, B.C. (1985) Decentralisation: The TerritorialDimension of the State, London: George Allen andUnwin

Smoke, P. (2004) ‘Kenya: Erosion and Reform fromthe Centre’, in D. Olowu and J.S. Wunsch (eds)Local Governance in Africa: The Challenges ofDemocratic Decentralization, London and Boulder:Lynne Rienner

Smoke, P. (ed.) (2003) ‘Decentralisation and LocalGovernance in Africa’, Public Administration andDevelopment 23.1

SNV (Netherlands Development Organisation)(2006) Towards MDG Localisation in Africa: Optionsand Experiences, The Hague: SNV

Therkildsen, O. (1993a) ‘Successful African LocalGovernment: Some Methodological andConceptual Issues’, Public Administration andDevelopment 13: 501–5

Therkildsen, O. (1993b) ‘Legitimacy, LocalGovernments and Natural ResourcesManagement in Sub-Saharan Africa’, inH.S. Marcussen (ed.), Institutional Issues in NaturalResources Management, Occasional Paper 9,Copenhagen: Roskilde University, InternationalDevelopment Studies

Titeca, K. (2005) ‘Civil Society Organisations, LocalGovernment and Political Struggles: FormalInstitutions and Informal Practices’, paperpresented at a conference on ‘MDG Localisationin Africa: Options and Experiences’, Kampala

Uganda (2002a) Uganda Participatory PovertyAssessment Process: National Report, Kampala:Ministry of Finance, Planning and EconomicDevelopment

Uganda (2002b) Fiscal Decentralisation in Uganda,Kampala: Fiscal Decentralisation Working Group

UNCDF (United Nations Capital Development Fund)(2003) Local Government Initiative: Pro-PoorInfrastructure and Service Delivery in Rural Sub-Saharan Africa, New York: UNCDF

de Valk, P. and Wekwete, H.K. (eds) (1990) Decentralisingfor Participatory Planning, Aldershot: Avebury

World Bank (1994) World Development Report 1994,Washington DC: World Bank

Wunsch, J.S. and Olowu, D. (1995) The Failure of theCentralized State: Institutions and Self-Governance inAfrica, Boulder: Westview Press


1Intro38.1.qxd 11/01/2007 10:29

introduction: decentralising service delivery? evidence

Documents