Balancing Rural Poverty Reduction and Citizen Participation: The Contradictions of Uganda’s Decentralization Program

Balancing Rural Poverty Reduction and

Citizen Participation: The Contradictions of

Uganda�s Decentralization Program

PAUL FRANCIS and ROBERT JAMES *University of East Anglia, Norwich, UK

Summary. —Uganda�s ambitious decentralization program is analyzed in terms of a ‘‘Dual-Mode’’system of local governance. Under a ‘‘technocratic mode,’’ conditional funding from the center isearmarked for particular programs but with little local participation. In contrast, the ‘‘patronagemode’’ is an elaborate system for local ‘‘bottom-up’’ planning, but with limited resources, which arelargely consumed in administrative costs and political emoluments. Along with the spoils of acommittee system controlling contracts and appointments, these resources provide the means forbuilding political alliances and loyalty. In the absence of a culture of transparency and civicengagement to assure downward accountability, it remains to be seen whether decentralization canpromote both efficient service delivery and local empowerment simultaneously.� 2003 Elsevier Science Ltd. All rights reserved

Key words — Africa, Uganda, decentralization, participation, poverty reduction, rural develop-

ment

1. INTRODUCTION

Since 1986, Uganda has embraced funda-mental economic and institutional reforms.One of the most ambitious has been its decen-tralization policy, held to be one of the mostfar-reaching local government reform pro-grams in the developing world. Several authorshave examined the history and formal struc-tures of the current decentralized system oflocal government in Uganda (e.g., Mamdani,1996; Nsibambi, 1998; Tidemand, 1994).However, little attention has been paid to theways in which these newly established systemsactually function at the local level, and thedegree to which the original objectives of pop-ular democracy and efficient service deliveryhave been achieved.The conventional conception of decentral-

ization comprises a national project, transmit-ted outward from the capital through theestablishment of a set of formal structures andprocedures. Actual local government systemsare often described in terms of these structuresas exemplifying devolution, deconcentration, ora hybrid of the two. In this paper, we comple-ment analysis of structure with attention toboth processes and resources. This leads us to

characterize Uganda�s decentralization policyin terms of a new model that is qualitativelydifferent from a mixed form of devolution anddeconcentration. We show how the combina-tion of formal organizations, resources, andprocesses of decision making and accountabil-ity yields specific distributions of managerialand political power at the local level, condi-tioned by corresponding public and privateincentives. We identify two contrasting formsof local governance in rural areas that wecharacterize as ‘‘technocratic’’ and ‘‘patron-age’’ modes. We further posit that these con-current modes of governance are founded upon

World Development Vol. 31, No. 2, pp. 325–337, 2003� 2003 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0305-750X/03/$ - see front matter

PII: S0305-750X(02)00190-0www.elsevier.com/locate/worlddev

*The research reported in this paper was undertaken aspart of the LADDER (Livelihoods and Diversification

Directions Explored by Research) research project un-

dertaken jointly by the School of Development Studies,

University of East Anglia, UK and the Economic Policy

Research Centre, Kampala, Uganda, and funded by

DFID.

Godfrey Kayobyo and Godfrey Turiho-Ahabwe

made particularly valuable contributions to fieldwork.

We also acknowledge Dr Godfrey Bahiigwa’s manage-

ment of the project in Uganda. Final revision accepted:

12 October 2002.

325

two potentially conflicting ideologies of devel-opment. The ‘‘technocratic’’ mode prioritizespoverty reduction, is driven by national targets,and is closely associated with poverty reductionstrategy plans (PRSPs). The ‘‘patronage’’ modedraws on the language of participatory plan-ning but, in the context of lack of resources andcapture by local elites, is reduced to a ritualizedperformance with little meaningful citizen in-volvement. This analysis leads us to questionwhether the objectives of poverty reduction andcommunity participation can be reconciled inthe absence of more effective accountability tothe local citizenry.The present section of the paper lays out the

general setting, providing some backgroundto the current system of local government inUganda and the research area. The operationof the decentralized system at district and lowerlevels is described in the next section, whichdraws largely on our research in rural Uganda.The final section presents our conclusions.

(a) The rationale behind decentralization––challenging the conventional view

Decentralization has several distinct aspects.A common categorization distinguishes politi-cal (or democratic) decentralization, adminis-trative decentralization (or deconcentration),and fiscal decentralization (Manor, 1999). Sincethe 1980s decentralization has been promotedas a solution to many of the problems of ad-ministration and governance constraining localand national development, as well as a meansof improving performance in poverty reduction(Blair, 2000; Crook & Sverrisson, 2001; Manor,1999). The benefits of decentralization areconsidered to include improved efficiency ofpublic service provision, more appropriate ser-vices, better governance, and the empowermentof local citizens.These benefits are held to arise in a number

of ways. Devolved decision-making mecha-nisms can facilitate the active participation ofcommunities, articulating local priorities andhelping to ensure that programs are appropri-ate to local needs. Political decentralization issupposed to carry the potential to create twodistinct kinds of accountability: downwardbetween electorate and local politicians, andhorizontally between democratically electedlocal politicians and local administrators.Decentralization is thus considered to be acornerstone of good governance both in

promoting local accountability and transpar-ency, and enfranchising local populations.Attractive though the potential benefits of

decentralization are, numerous studies haveshown that they are seldom realized (Crook &Manor, 1998; Moore & Putzel, 1999). Adam-olekun (1999, p. 58) goes as far as to concludethat, while decentralization has been includedin public sector reform in many sub-SaharanAfrican countries, ‘‘there are no real successstories as far as improved development per-formance at the local level is concerned.’’Nickson (1995) argues that despite the widelycited success of Latin American decentraliza-tion, there is a wide gulf between the rhetoricand reality of citizen participation.Why has the experience of decentralization

proved so disappointing? Three explanationsdominate in the literature: inadequate capacity,insufficient fiscal decentralization, and a lackof accountability to citizens (Johnson, 2002).Certainly, the difficulty of recruiting and re-taining skilled staff at district level and below isa widely recognized constraint. But, our re-search has led us to focus more closely on theinterplay of the other two factors: the degree offinancial autonomy and the character of localgovernance.The extent to which elected local represen-

tatives actually control district finances freefrom central interference is fundamental to therealization of local autonomy. Yet the degree ofcontrol which local politicians have over eitherlocally raised revenues or central transfers isnever entirely unconstrained, and varies widely.In rural Africa, the local revenue base is oftenso weak that central transfers dominate districtbudgets. While a high proportion of centraltransfers is not as such incompatible with localautonomy, the conditions frequently attachedto these transfers can undermine genuine localdecision making. Manor (2000), for example,notes that in South Africa the pursuit of re-distribution and efficiency goals by centralgovernment seriously reduces local govern-ment�s independence.The third constraint identified above, that of

accountability, is perhaps the most problematic(Blair, 2000). Political decentralization makesdemocratic mechanisms the means to ensureand promote downward accountability. Yet,the ballot box is only a part of the wider in-stitutional context: democracy also presupposesaccess to information, transparent proceduresof government and an effective media (Gaventa& Valderrama, 1999). Certainly politicians or

WORLD DEVELOPMENT326

administrators can only be held to account by acitizenry that is active and informed. But, asMamdani (1996) has argued, the African pat-rimonial state has perpetuated a rule over sub-jects rather than a rule by citizens. In such anenvironment, decentralization may result, inthe words of Cross and Kutengule in ‘‘repres-sion being brought closer to the people’’ (2001,p. 6). Introducing decentralization into a po-litical environment characterized by clientagerisks strengthening ties of patronage and fur-ther entrenching local elites.

(b) Decentralization in Uganda: origins andstructure

Under the colonial system of indirect rule,District Commissioners (DCs) governedthrough ‘‘native authorities’’ headed by favoredtraditional chiefs, a system characterized byMamdani (1995) as ‘‘decentralized despo-tism.’’ 1 During the early independence period,native authorities were abolished, but DCs re-mained both powerful and centrally appointed.The powers of local government significantlydiminished as a result of the 1967 republicanconstitution and yet further under Amin�s mil-itary regime. On its accession to power in 1986,the National Resistance Movement (NRM),saw decentralization as a ‘‘necessary conditionfor democratization’’ and hence central to thefulfillment of their goal of establishing a‘‘popular democracy’’ in Uganda (Kisakye,1997).The political context of the NRM era

has been a unique system of ‘‘no-party’’ de-mocracy. Given Uganda�s turbulent politicalhistory, the NRM leadership has held thatmulti-partyism would revive ethnic and reli-gious cleavages. The movement system, underwhich no member can be expelled (in contrastto a one-party system), is meant to ensure thatmerit rather than political affiliation is the basisof representation. Uganda�s international pa-trons have looked askance at this system, butbrought little pressure on the government torealize pluralism (Hauser, 1999). In these cir-cumstances, decentralization has provided ademocratic gloss in the eyes of both interna-tional donors and local actors. Over time,however the movement style of politics hasbegun to take on some of the characteristics oftraditional one-party rule (Mamdani, 1995;Odongo, 2000). This includes the use of stateresources for the purposes of political mobili-zation in order to sustain support for the ‘‘no-

party’’ system at grass root levels. If, as Blair(2000) argues, the existence of competitive po-litical parties is a necessary requirement forencouraging a culture of local accountability(cf. Crook & Manor, 1998), then the perpetu-ation of the no-party state in Uganda may se-verely compromise its development.The legislative framework of decentralization

is provided by the Local Government Statuteof 1993, the 1995 Constitution and the LocalGovernment Act 1997 (Uganda, 1993, 1995,1997a). These acts converted the existing sys-tem of Resistance Councils (RCs), with theirorigins in the civil war period, to a pyramidalstructure of Local Councils (LC) at village(LC1), parish (LC2), subcounty (LC3), county(LC4) and district (LC5) levels. 2 Table 1summarizes key aspects of the decentralizedlocal government system in rural Ugandawhich has devolved functions, competency andresources to elected local government councils.Administrative and technical personnel arefound at the district and subdistrict levelsheaded respectively by a Chief AdministrativeOfficer (CAO) and a Subcounty Chief (SCC).

(c) The ‘‘macro’’ policy context

Decentralization comprises an importantcomponent of a much wider program of reformundertaken by the Government of Ugandasince 1987, encompassing the economic andjudicial as well as the administrative and po-litical spheres. In the late 1980s the Govern-ment undertook a far-reaching economicrecovery program embracing market liberal-ization through the removal of price controls,the privatization of state industries and thedisbandment of agricultural parastatal boards.Average GDP growth of 6.5% per year since1990 is cited as proof of the success of theseprograms. There is evidence that the incidenceof poverty has fallen from 56% in 1992–93 to35% in 1992/3. Appleton (2000), attributes thisdrop entirely to the growth of GDP, thoughadds a cautionary note that urban-rural in-equality over 1992–2000 has increased.There are three key elements to Uganda�s

poverty reduction policy: the Poverty Eradica-tion Action Plan (PEAP), the Poverty ActionFund (PAF), and the Plan for the Modernisa-tion of Agriculture (PMA). Decentralizationprovides the institutional framework for theirimplementation. The PEAP represents theUganda version of the Poverty ReductionStrategy (PRSP) process common to many

RURAL POVERTY REDUCTION AND CITIZEN PARTICIPATION 327

Table 1. Decentralization in Uganda: key features of the institutional structure

Local councillevel/area

Political head Procedure for selection of representatives Status of LC leveland Administrative

Head

Technical staff Percentage of locallygenerated revenue

retained

LC5/District District CouncilChairman

(a) Chairman elected by universal adult suffrage(UAS)

Local Government Full team 35%

(b) One councilor elected from each subcountyby UAS

Chief AdministrativeOfficer (CAO)

(c) Women make up 1/3 of council(d) Special councilors for youth and disabled

LC4/County LC4 Chairman (a) Council made up of all LC3 executives,who then elect LC4 executive

Administrative UnitAssistant CAO

– 5% of 65%

LC3/Subcounty Subcounty CouncilChairman

(a) Chairman elected by UAS Local Government Subaccountant,extension and othertechnical staff

65% of 65%

(b) Councilors elected by UAS from eachparish

Subcounty Chief

(c) Women make up 1/3 of council

LC2/Parish Parish CouncilChairman

Selected by LC1 Executive members Administrative UnitParish Chief

– 5% of 65%

LC1/Village LC1 Chairman Direct election by universal suffrage Administrative Unit – 25% of 65%

Source: Adapted from Uganda (1993, 1995, 1997a).

WORLDDEVELOPMENT

328

developing countries and a prerequisite forqualification for Highly Indebted Poor Coun-tries (HIPC) debt relief. Initiated in 1997,PEAP has four central pillars: creating anenabling environment for economic growth,ensuring good governance and security, pro-moting the ability of the poor to raise theirincomes, and increasing their quality of life(Uganda, 1997b). The PAF finances key pov-erty eradication programs using funds from theHIPC debt relief initiative and resources mo-bilized directly from donors. Finally, the PMAis a multi-sectoral program aimed at reducingrural poverty through the commercialization ofagriculture (Uganda, 2000). Key features in-clude a grant mechanism for subcounty-levelinvestments, and a shift toward demand-drivenextension by vesting budgets in farmer user-groups.

2. DECENTRALIZATION: BRINGINGCONTROL CLOSER TO THE PEOPLE

The fieldwork on which this section is basedwas undertaken in three districts, Mbale, Ka-muli and Mubende, each representing a majorrural livelihood system. Three villages were se-lected in each district, and qualitative andquantitative data collected by a multidisciplin-ary team of researchers. This research wascomplemented by interviews at subcounty anddistrict levels with administrators, technicalstaff and politicians.

(a) Local government institutions: the view fromthe community

LC1 Chairmen are simultaneously leaders ofa local community and the lowest level of aformal administrative system. In fact in localperceptions it is their internal role in regulatingvillage life that predominates: adjudicating mi-nor offences and disputes and witnessing landtransactions. As the point of contact with thehigher levels of the administration, chairmenmay be asked to assist in graduated tax col-lection, receive official visitors or write lettersto certify residence or ownership of livestock.Chairmen charge informal fees for most ofthese services. The adjudication of a minordispute may cost the guilty party Sh. 5,000 whilewriting the certificate of ownership necessary tosell livestock costs about Sh. 1,000. 3 In con-junction with their councils, chairmen makedecisions about how to spend any locally gen-

erated revenues remitted back from the subco-unty level, and mobilize the community forcollective contributions to development projects.Although their members are subject to five-

yearly elections, LC1 councils tend to haverelatively stable memberships. The councilorsare drawn almost exclusively from householdsin the highest income tercile. Poorer individualscannot afford the ‘‘goodwill gestures,’’ such asbeer, soap, or salt, handed out as an induce-ment to potential voters in elections to all levelsof local government. Better-off candidates arealso considered more likely to have the neces-sary connections and skills (e.g., literacy). Theformal roles of LC2s are to co-ordinate andmonitor LC1 activities and provide a link be-tween village and subcounty in the decision-making process.Villagers generally feel distant from the sub-

county level (LC3), whose officials are identi-fied mainly with graduated tax collectiondrives. Taxation being the main way in whichlocal administration manifests itself in the livesof community members, in all communities themost adverse aspect of the institutional changesaccompanying decentralization was consideredto be the increasing burden of taxation andother formal and informal fees. Villagers seelittle evidence of the locally collected revenuesbeing used for the general good and consideredthem to be frequently misused. Even their LC1counterparts did not consider that LC3 coun-cilors took their local priorities seriously. Inaddition, the performance of production staffposted at subcounty level in most sectors isconsidered so poor as to be largely irrelevant totheir needs.

(b) Local government finance

District resources come from locally gener-ated revenues and central funding of threekinds: unconditional (UCG), conditional (CG)and equalization grants (EG). Table 2 showsthe trends in transfers to local authorities since1997. Total transfers more than doubled during1997–2000. Conditional funding increasinglydominates, now accounting for over 80% of allcentral transfers. CGs are programmed at thenational level, and earmarked to support spe-cific national sectoral programs in the districts.Eighty percent of CGs are channeled throughthe PAF which in 2000–01 amounted to 30%of the total Government Budget. Three quar-ters of these funds take the form of condi-tional grants to districts principally in the


health, education water, roads, and agriculturesectors (Uganda, 2001). Of late the complexityand inflexibility of the CG system has beencriticized, as district councils demand moredecision-making power over centrally-derivedresources. The use of UCGs is not defined bylegislation, and they carry minimal reportingrequirements to central government: they are infact largely spent on general management andadministration. While UCG has increased inreal terms since 1997, as a share of all transfersit has declined, from 24% to 15%. EGs supportdistricts lagging behind the national averagestandard for a particular service. But, they ac-counted for less than 1% of all transfers in2000–01. In addition to these three channels,the Local Government Development Program(LGDP), initiated in 2000–2001, transfers fundsto both subcounty and district levels on thebases of agreed performance criteria and a re-quired counterpart cash contribution of 10%.The 1997 Local Government Act gave ex-

tensive powers to the district to raise revenuelocally and set out rules governing its distri-bution. These revenues are controlled by dis-trict actors and are the main source of localautonomy and patronage. Their level is con-strained, however, by a weak revenue base andthe inevitable political costs of imposing localtaxes. Reflecting these constraints, levels of lo-cal revenue are, in real terms, static or in manydistricts falling (Uganda, 2001). Analysis of thebudgets of the three research districts showedthat the proportion of revenue raised from localsources was small: 5% in Mbale, 4% in Kamuliand 10% in Mubende. Centrally allocatedfunds, accounting, as a national average, for90% of income, therefore dominate district fi-nances. It is this combination of centrallyoriginating conditional grants and limited localresources which gives decentralization in ruralUganda its specific and contradictory charac-ter.

Locally raised revenues include graduatedpersonal tax (GT), produce taxes, market dues,licenses and taxes specific to individual districts.Graduated tax is the most significant of these.It is payable annually by all adult males andsalaried women according to a scale based onimputed incomes, which take into accountownership of productive assets. Annual licensefees are payable by a wide range of trades andbusinesses. In addition to market dues, parishtaxes are payable on all transactions and busi-nesses at the village level. Permits are requiredfor the movement of livestock. As well as tothese taxes and levies, individual districts sup-plement graduated tax with levies for specificservices such as feeder roads and education.The Graduated Tax is very unpopular.

Community leaders complained that they werenot consulted in assessment, and the way inwhich the tax was collected also led to localresentment. During the annual round of taxcollection visits to villages by subcounty offi-cials and armed police, force was often used,and defaulters even said to be dragged fromtheir homes at night and imprisoned. For theirpart, administrators considered that decentral-ization, in making them subordinate to localcouncilors, had made it more difficult to im-pose discipline in revenue collection. Localpoliticians, they claimed, now protect theirelectorate from the sanctions of the law. Espe-cially during election years, politicians at alllevels encourage default. 4 In Kamuli, the localrevenue generated in the election year 2000–01fell to less than a third of its levels in 1999–2000, as all candidates had promised the re-duction or abolition of GT during their cam-paigns. These local forms of resistancemean that a high proportion of the tax is con-sumed in its own collection (40% in Mbale, forexample).With the exception of GT and licenses, the

collection of most other taxes has been priv-

Table 2. Transfers from central government to local government (UG S bn.)

Grants 1997–98 1998–99 1999–2000 2000–01

UG Shs bn % UG Shs bn % UG Shs bn % UG Shs bn %

Unconditional Grants 54 24 64 23 67 17 79 15Conditional Grants––Recurrent 168 75 202 71 275 71 321 63Conditional Grants––Development 2 1 19 7 45 12 107 21Equalisation Grants 0 0 0 0 2 0 4 1

Total 224 100 285 100 389 100 512 100

Source: Adapted from Uganda (2001).


atized since decentralization. This policy hasbeen justified as a means of increasing the yieldand efficiency of tax collection. From the pointof view of the collectors, it is a policy that canbe extremely lucrative. Estimates of parish-leveltax collection analyzed in one community, forexample, suggest that a licensed revenue col-lector might take up to Sh. 300,000 per month,only Sh. 30,000 of which is passed on to thepublic purse. Frequently even the agreed fee isnot actually paid by the collector. Tax farmingcontracts are allocated by the District TenderBoard (DTB), which chooses among biddingprospective collectors, a procedure widely feltto be subject to political interference. In sum,while the privatization of tax collection hasintensified the burden of taxation on the ruralpoor, it is doubtful whether it has broadenedthe local revenue base. To make matters worse,producers and traders face demands for infor-mal payments by administrative or politicaloffice holders in addition to these official taxes.In the eyes of community members, the taxes,licenses and levies imposed on agriculture andtrade combined with unofficial rents, constitutea dense thicket of fiscal oppression and a dragon their livelihoods.Goods going to market may be subject to

several taxes, especially if they cross adminis-trative boundaries. In Mbale district, for ex-ample, a fee of Sh. 200 is payable to the holderof the parish tax tender on every bunch of ba-nanas transported out of the village. A furtherfee of Sh. 200 per bunch is payable in eachmarket in which the bananas are sold. In ad-dition, if the bananas pass through other dis-tricts, yet another Sh. 200 per bunch will becollected in each, even if the bananas are nottraded. In this way, taxes can easily amount to30% of the farm gate price of Sh. 2,000. Similarlevies apply to sales of livestock and milk.Farmers complain that, because of their lack ofliquidity and ignorance of the system, thesedisincentives result in middlemen dominatingagricultural marketing beyond the farm gate.Similarly, fishermen have to make a range ofpayments, both official and unofficial, to publicbodies and institutions. In Iyingo on LakeKyoga, for example, these include income taxto central government, boat license paymentsmade directly to districts, and daily parishtender payments exacted by private collectors.In addition, there are ‘‘informal’’ payments tothe gabunga (leader of the fishers organiza-tion), the fisheries officer, and a fisheries ‘‘taskforce.’’

(c) Participation by communities in localplanning

To what degree have the new institutionscreated by decentralization promoted localparticipation in decision making? The formalsystem of planning is supposed to proceed in anintegrated bottom-up manner. Each villageproduces a Community Action Plan (CAP)based on local needs and priorities. Parishcouncils incorporate these plans into parishplans, which are in turn synthesized into sub-county plans. The District Technical PlanningCommittee is then supposed to produce an in-tegrated plan based on subcounty plans and thedeliberations of a budget conference of keystakeholders, and this plan is ratified by theDistrict Council. In practice, however, the sys-tem scarcely articulates at all between levels.First, priorities from LC1 and LC2, if theyreach higher levels, are rarely actually incor-porated into subcounty plans. Second, giventhe limited capacity and experience of politi-cians and civil servants, at the subcounty levelplans of any quality or realism rarely materi-alize. When they do appear, plans are generallybased on over-optimistic estimates of revenue,which, as the following analysis shows, isscarcely available for ‘‘development’’ at all.The 1997 Local Government Act defines a

statutory formula for the distribution of locallygenerated revenue between levels: 35% is re-tained by the district level, while the remainderis divided between the other four levels, withthe subcounty receiving two thirds of it (fordetails, see Table 1). Communities� share oflocal revenue is modest: for a typical village of100 households, the LC1�s share of graduatedtax, assuming a payment rate of 50%, wouldamount to about Sh. 89,000 (or only US$52)per year. This is further depleted by the costs oftravel to collect it from the subcounty head-quarters. More seriously, in most villages, LC1committee members complained that they werenot receiving their full share of locally collectedtaxes (especially taxes other than GT), in spiteof efforts to follow up through the local councilhierarchy, and that this situation was worsen-ing from year to year. The chair of Kiribairyavillage, for example, had gone to the SubcountyChairman and Cashier with the receipts andserial numbers from all of the GT and the boatlicenses issued in his community, but had stillnot been able to obtain the village�s entitlement.More often, LC1 and LC2 councilors were notaware of the actual amount to which they were


entitled. Administrators, for their part, felt thatlocal councilors did not account properly forthe funds that they did receive.The use of resources that do accrue at village

level is discussed at village meetings. Formally,all adult members of the community are mem-bers of the LC1, but attendance at these meet-ings is not high, in part because of the limitedresources at stake and in part due to a per-ception that such decisions were the prerogativeof elected representatives. Where funds wereavailable, they were most often spent on pri-mary schools or as counterpart contributions towater projects. Typically, however, far frombeing a catalyst for local development initia-tives, funds were insufficient even to undertakepetty works such as the maintenance of a vil-lage well.At the subcounty (LC3) level, most locally

generated revenue is in fact consumed by thesalaries and allowances of the council itself (inspite of the fact that, according to the LocalGovernment Act, these charges are supposednot to exceed 15% of revenue). 5 For example,Butiru SubCounty, Mbale District, in 1999–2000 spent Sh. 13.6 million, raised through lo-cal fees and taxes. Almost Sh. 5 million (36%)was spent on sitting fees and allowances forcouncils, committees and boards and the LC3chairman�s salary, with a further Sh. 4 million(28%) being spent on administrative support(much of it on GT collection).At district (LC5) level, this pattern is even

more marked. Councilors� salaries, allowancesand emoluments frequently consumed most orall of locally generated revenue, or even ex-ceeded it. The use of scarce funds for admin-istrative overheads and political emolumentsmeans that very little is left to finance produc-tive activities. In Mbale, some 362 millionshillings of locally generated revenue wasavailable to the district level in 1999–2000 while371 million was spent on ‘‘Commissions,Committees and Boards.’’ In Kamuli, politicalemoluments encroached even more seriously onrevenues. Sh. 112 million of local revenue wasavailable but Councils Commissions andBoards consumed a remarkable Sh. 369 mil-lion. Such shortfalls can only be met by plun-dering the unconditional grant.If participation by communities in decision

making about locally generated resources islimited by their scarcity, their influence overcentrally allocated funds is seriously con-strained by the conditions to which such a highproportion are subject. Recurrent funds are

already earmarked for specific salaries, whilstthe capital grants (which are smaller) are pre-allocated by sector. Thus the only leeway forlocal decision making with regard to condi-tional grants is in the siting of capital projects.The ‘‘unconditional’’ (block) grant is mostly

consumed by administrative and operationalcosts. Councilors may at times also accessUCG transfers for their own allowances. Intheory, any funds that remain are allocatedbetween departments according to the ap-proved district budget. In practice, however,funds are rarely available, and where they are,are generally allocated on an ad hoc basis andwithout consultation. There is thus very limitedscope for local decision making in the use of theunconditional grant, and senior administratorsand councilors close off even the limited possi-bilities that exist.It is clear from the above that whatever in-

stitutions, procedures and rhetoric exist for thepromotion and realization of a wider ‘‘policyspace’’ at local level, the resources that theparticipatory process can actually control areminimal. Partly to address this problem andpromote the scope for local decision making, in2001 districts became eligible for LGDP grantschanneled directly to district, subcounty andparish levels.

(d) Decentralization and service provision

What has been the impact of decentralizationon the quality of public services? Through thePAF, there has been a large expansion inmoney coming to the district through nationalprograms for particular social services. Thecapacity of the primary school system wasdoubled during 1996–97 following the intro-duction of Universal Primary Education policy,while the provision of health and water infra-structure has also been extended (Ablo & Re-inikka, 1998).The situation in the productive sectors, es-

pecially agriculture, is less positive. There are anumber of interlocking reasons for this, in-cluding: the low level of resources allocated toproductive sectors, confusion over responsibil-ities for management and operational funding,the limited appropriateness of the advice of-fered to farmers, and the transitional status ofagricultural policy pending the full implemen-tation of the PMA.The proportion of district budgets allocated

to production and marketing (which includesagricultural services) is small: 3.2% in Mu-


bende; 2.5% in Kamuli; and less than 1% inMbale. This level of support, most of whichcomes from conditional grants for salaries ofagriculture staff, seems inconsistent with theobjectives stated in the PEAP which stresseconomic growth and the incomes of the poor.At the subcounty level, the proportion of fundsspent on production is even smaller (e.g., 0.3%in Buyende SubCounty, Kamuli).Decentralization has resulted in a complex

matrix management situation in which fieldpersonnel are responsible to a local managerwho has no technical knowledge of their role,yet dependent for programming and opera-tional funds on a technical manager in thedistrict headquarters. As a result agriculturalstaff are poorly supervised. The lack of opera-tional funds and supervision means that agri-cultural extension officers have a minimalimpact at village level. They rarely, if ever,visited sample villages. In Buwopuwa village,Mbale it was reported that ‘‘the last time anagricultural officer was seen in this village wasin the 1980s.’’ Farmers who had actively soughtout extension workers for problems such ascoffee wilt, cassava mosaic virus and stregacomplained that appropriate advice was notoffered. The decentralized system of manage-ment has thus not been successful in articulat-ing and responding to economic needs. Thefinal factor underlying poor services is theprovisional and transitional status of agricul-tural policy at the time of our research, pendingthe implementation of the reforms envisaged inthe PMA. This has had a very damaging effecton staff motivation.In summary, the evidence suggests that, de-

centralization has not been able to arrest thedeterioration in agricultural services, and thateven the improvements in social services areattributable to increases in central conditionalfunding rather than the very limited scopewhich decentralized institutions have providedfor local decision making.

(e) Conflict, collusion and the means ofpatronage

Decentralization has created a new set ofrules, procedures and incentives for the alloca-tion of development resources which createcharacteristic lines of competition, conflict andalliance. One axis of conflict is between thevarious levels of local administration. There isresentment at the village and parish at thefailure of higher levels to include them in de-

cision-making processes in more than a tokenway, or to release the funding which they aredue. Likewise, the subcounty level, althoughnow the lowest level of government, is chroni-cally underresourced and tends to blame thedistrict for this (e.g., for its failure to remit theGT of public employees). Higher levels justifytheir reluctance to devolve resources by the lackof capacity and accountability mechanisms atlower levels. Indeed, while subcounty booksare audited, sanctions are rarely brought incases of irregular spending.The second major dimension of conflict is

that between politicians and civil servants. Civilservants are better educated than the politicalleadership of the district, yet their salaries andallowances are considerably lower. For exam-ple, the salary of a graduate assistant CAO is inthe region of Sh. 210,000 per month, while anon-graduate chairperson earns Sh. 1,300,000.Decentralization has also increased the powerof local politicians vis �aa vis district staff. Twocrucial arenas of decision making which carryenormous scope for patronage are the DistrictTender Board, responsible for awarding con-tracts and tax collection rights, and the DistrictService Commission (DSC), which appoints alldistrict staff. The District Council (consisting ofelected politicians) on the advice of its DistrictExecutive, appoints the members of both ofthese bodies, who are unelected citizens.Both community members and administra-

tors alleged that the patronage of the DistrictExecutive Committee over membership of theDTB, with its privileges including generoussitting fees, enabled them to influence the de-cisions of the board unduly. Tenders (e.g., fortax and construction contracts) are supposed tobe allocated on the basis of a points system thattakes account of a range of criteria includingprice, experience, and record of tax payments.In practice though, a letter of recommenda-tion from a politician is believed to be a cru-cial prerequisite, and it is widely believed thatsuccessful tenderers are friends, relatives orprot�eeg�ees of the political class, or proxy com-panies operating on their behalf.The DSC appoints, disciplines and has the

power to remove, all district staff. Civil servantsfeel that this makes them vulnerable to unduepressure or even victimization should they goagainst the wishes of local politicians. TheDSC of one district, under pressure from theDistrict Executive, had dismissed a DeputyCAO after he questioned the use of the un-conditional grant for politicians� allowances.


Other administrators had similar concerns,and, though reluctant to voice them publicly,did express the view that the problem couldonly be countered by making a still higherproportion of central transfers conditional.Conflict is one response to this situation,

collusion is another. In what is sometimestermed the ‘‘sons of the soil’’ phenomenon,local applicants are favored for administrativeappointments over candidates from other partsof the country. Politicians exert pressure on theDSC to employ indigenes, who are likely to bemore malleable, or easily enmeshed in localstructures of patronage. This practice not onlyfurther weakens technical capacity, but alsoundermines the national character of publicadministration.

3. CONCLUSIONS

On the surface, the mechanisms of decen-tralization are established and functioning inUganda, with a five-tier structure of localcouncils, deconcentrated staff, a bottom-upplanning process, and powers to raise andspend local revenue. We have seen howeverthat these structures and processes do notconstitute a genuinely participatory system oflocal governance.The first problem is one of resources: the

local revenue base is weak, and central trans-fers, as we have seen, are predominantly con-ditional. In rural districts, the tax burden fallslargely on poor farmers, discouraging the ex-pansion and commercialization that are theostensible objectives of two of the government�scentral policies, the PEAP and the PMA. Pri-vate tax collection systems inflate the burden oftaxation yet deliver a low yield to the districtcoffers. The failure to remit tax yields back tolocal level, and the predisposition of politiciansto sanction the nonpayment of GT duringpolitical campaigns, conspire to reduce localrevenue still further. The legitimacy of localtaxation is eroded by the nonproductive, andoften nonstatutory, use of revenue: communi-ties see few concrete benefits coming from thetaxes that are extracted from them.The transaction costs incurred by the multi-

layered pyramid of planning are high in termsof both time and their consumption of revenuefor allowances. Yet the decisions which are theproduct of this system often fail either to reflectthe priorities of lower levels or to enhance theflow of information downward to communities

in ways which would enable them to hold theirrepresentatives accountable. More seriously,given the conditional nature of most transfersfrom the center, the system is devoid of realresources to control. The participatory plan-ning process is thus more a matter of form thansubstance––a ritualized performance simulatinglocal decision making. While frequently iden-tified as a constraint, the lack of local capacityfor planning on the part of local politicians andadministrators is less a problem than the lackof resources that could make local decisionmaking meaningful.In order to accommodate the contradiction

between centralized control and local partici-pation, a system has developed with two par-allel but distinct modes of operation, which wecall the ‘‘technocratic’’ and the ‘‘patronage.’’This system should not be seen as simply ahybrid between deconcentration and devolu-tion: in the first place personnel are employeddirectly by the districts rather than beingmerely deconcentrated. Second, responsibilityfor the provision of most services no longerrests with line ministries but with the districtsthemselves (Uganda, 1997a). Yet, despite thesestructural reforms toward a formal system ofdevolution, central control has been main-tained, not through deconcentration, but byplacing conditions on the use of centrally de-rived resources. We term the ‘‘technocratic’’and the ‘‘patronage’’ components of the dualsystem ‘‘modes,’’ because they consist of notonly an organizational framework, but alsothe political and financial resources, norms ofcompetition and conflict, legitimating dis-courses, and accountability mechanisms thatsurround them. Their essential characteristicsare laid out in Table 3.The ‘‘technocratic mode’’ is resourced by

conditional grants from the center, which cur-rently finance the delivery of most services atdistrict level. The key institutions at the centerare not only the sectoral ministries, but alsothe increasingly dominant Ministry of Finance,Planning and Economic Development throughwhich all donor funding, the most significantsource of capital spending, passes (Harrison,2001). The conditionality of funds closes off grass-roots control, giving the local population andits political class limited voice. Needs are readfrom the top, and programs imposed downward.The ‘‘patronage’’ mode, in contrast, is very

much enmeshed in the local political process. Itis fuelled directly by locally generated revenuealong with unconditional funding channeled


into a structure of petty patronage. Hence,while carrying the potential to empower localpeople, in reality this mode rarely involves reallocal decision making, simply because the lim-ited available resources are largely consumed inthe performance of participatory planning it-self. The spoils that arise from the control ofcontracts and appointments provide less directopportunities for patronage and even rent-seeking.Each mode has its own discourse. That of the

‘‘technocratic’’ mode revolves around sectoraltargets and poverty priorities; that of the ‘‘pa-tronage’’ mode evokes popular democracy andbottom-up planning. Yet, as we have docu-mented in some detail, these discourses corre-spond imperfectly to actual processes: that ofthe ‘‘technocratic’’ mode masks a perpetuationof central control while that of the ‘‘patronage’’mode coexists with a latent function of ex-tending links of clientage and ensuring politicalloyalty.

For each mode, there are three potentialmechanisms of accountability: upward (tocentral government), horizontal (to electedrepresentatives) and downward (to the citi-zenry). These are represented diagrammaticallyin Figure 1.In the ‘‘technocratic’’ mode, the dominant

mechanism of accountability is upward, framedby centrally determined targets and audit con-trols. There is limited accountability of ad-ministrators to local politicians, or of eithergroup to the local population. Local partici-pation is limited to counterfeit mechanisms ofenfranchisement such as the ‘‘ParticipatoryPoverty Assessments’’ so alluring to Uganda�sdonors, which provide the desired facade ofconsultation.In the ‘‘patronage’’ mode, by contrast, ac-

countability to the center is limited, with theinspectorate of the Ministry of Local Govern-ment having little power to sanction inappro-priate behavior. Politicians do have a degree of

U

D

HManipulative

Limited

Electoral

U

D

HLimited

Audit

Token

Technocratic mode

Patronage mode

Figure 1. Upward, horizontal, and downward accountability in the Dual-Mode system.

Table 3. The political economy of decentralization: the Dual-Mode system

Characteristic Mode

Technocratic Patronage

Financial resources Conditional Grants (c.80% district budget) Unconditional Grants plus locallygenerated revenue

Political resources Control of District Tender Board,District Service Commission

Uses Poverty reduction priority (mainlysocial sector)

Largely politicians� emoluments

Local participation in planning Participatory poverty assessments, etc. Five-tier local council systemKey stakeholders Donors, central government Party, politiciansOrienting discourse Delivery of services according to PEAP

targetsParticipation, (enfranchisement)

Upward accountability Audit LimitedHorizontal accountability Limited ManipulativeDownward accountability Token Electoral accountability of councilors


control over administrators, but this tends to bemanipulated in order to further their indi-vidual, rather than the public, interest. Whilein theory, downward accountability existsthrough the ballot box, this is ineffective in asystem where there is very limited publicknowledge about either resources or decisions,and votes are regarded as a form of reciprocityin return for ‘‘goodwill’’ gestures. Hence, be-hind the manifest function of promoting localdemocracy is the latent function of perpetuat-ing a network of patronage for political mobi-lization.The Dual-Mode system allows central con-

trol to exist alongside a simulation of populardemocracy. This is particularly opportune inpresent day Uganda, where decentralizationplays a crucial role in justifying the unique ‘‘no-party’’ system both to a highly politicizedpopulation and the international donors whohave elsewhere made so much of multipartydemocracy as the hallmark of good gover-nance. For the NRM, decentralization in thecontext of the no-party state serves the fur-ther purpose of entrenching the party ma-chinery into the organs of the state, facilitatingthe use of public resources for political pa-tronage.Many policy makers have seen decentralized

local government as a key vehicle for achievingthe aims of the PEAP. Our analysis has how-ever brought out the contradiction between thecurrent means of achieving PEAP targets andthe objectives of participation and local deci-sion making that supposedly underlie the drivefor decentralization. Rather than arising, asmany policy makers perceive it, from the con-straint of limited fiscal decentralization, thisis a true contradiction between two potentiallyconflicting development ideologies––bottom-updecision making and nationally (and interna-tionally) imposed poverty reduction blueprints.As Johnson (2002) argues, central governmenthas a vital role ‘‘in ensuring the developmentand implementation of substantive pro-poorpolicies’’ (p. 529). The challenge is to definehow this can coexist with local participationand autonomy.

Our Dual-Mode model implies that mean-ingful decentralization could be advanced inimportant ways. The main needs are to makeresources available for local decision making,to build capacity for planning, and to ensurefirmer accountability for resources. There al-ready exists scope for increasing transparencyby making simple changes to procedures and,as important, ensuring that existing regulationsare enforced. Such measures include: ensuringthat information is disseminated publicly aboutlocal revenue and budgets, bringing sanctionsagainst defaulting private tax collectors, en-forcing the capping of political emoluments,enforcing receipting, simplifying the calculationof village and parish entitlements, and stream-lining their remittance. Such unspectacularmeasures could do much to shift the terms ofthe political economy of information at thelocal level.The recently initiated Local Government

Development Program is designed to address anumber of the problems identified here. As partof a reformulation of local government fi-nancing, it will make resources available forinitiatives at local levels, together with the ca-pacity to plan and manage them, sanctioned byperformance incentives and penalties. Althoughnot without its problems (difficulty in raisingthe community contribution in areas of weakrevenue base, proposals overlapping with thosecovered by existing conditional grants, inade-quate support and monitoring of subcountiesby districts and inflated tendering), this pro-gram does hold out the prospect of increasinglocal autonomy.Nevertheless, genuine downward account-

ability will require more than a new set ofprocedures and institutions. As we have seen,those with vested interests are capable ofturning the institutions and opportunitiescreated through decentralization to their ownadvantage. True local democracy and ac-countability can only be founded on a shift invalues and awareness, and the emergence ofactive citizenship. It is doubtful whether such adeepening of democracy can be imposed fromthe top downward.

NOTES

1. The political history of Uganda during the colonial

and early post colonial period is well documented:

Karugire (1980), Mamdani (1976) and Mutibwa (1992).

2. Although a formal administrative unit, the LC4

is largely a legacy of previous systems with few

functions.


3. Exchange rate at time of fieldwork: US$ 1 ¼ Sh.1,700.

4. In the 2001 presidential election campaign, the

incumbent directed the Ministry of Local Government

to reduce the minimum level of GT from 11,000 to 3,000

Sh.

5. Increased to 20% since 2001 (Local Government

Revenue Amendment Regulation).

REFERENCES

Adamolekun, L. (1999). Public administration in Africa.Boulder, CO: Westview Press.

Ablo, E., & Reinikka, R. (1998). Do budgets reallymatter? Evidence from public spending on educationand health in Uganda. Washington, DC: World Bank.

Appleton, S. (2000). Poverty reduction during growth:the case of Uganda, 1992–2000. Nottingham, UK:University of Nottingham, Department of Econom-ics, Mimeo.

Blair, H. (2000). Participation and accountability at theperiphery: democratic local governance in six coun-tries. World Development, 28(1), 21–39.

Crook, R., & Manor, J. (1998). Democracy and decen-tralization in South-East Asia and West Africa:participation, accountability and performance. Cam-bridge, UK: Cambridge University Press.

Crook, R., & Sverrisson, A. (2001). Decentralisation andpoverty-alleviation in developing countries: a compar-ative analysis or, Is West Bengal unique? IDSWorking Paper 130. Brighton, UK: Institute ofDevelopment Studies.

Cross, C., & Kutengule, M. (2001). Decentralization andrural livelihoods in Malawi. LADDER WorkingPaper No. 4. Norwich, UK: University of EastAnglia, Overseas Development Group.

Gaventa, J., & Valderrama, C. (1999). Participation,citizenship and local governance. Background noteprepared for workshop on Strengthening participa-tion in local governance. Brighton, UK: Institute ofDevelopment Studies.

Harrison, G. (2001). Post-conditionality politics andadministrative reform: reflections on the cases ofUganda and Tanzania. Development and Change, 32,657–679.

Hauser, E. (1999). Ugandan relations with Westerndonors in the 1990s: what impact on democratisa-tion? Journal of Modern African Studies, 37, 621–641.

Johnson, C. (2002). Local democracy, democratic de-centralisation and rural development: theories, chal-lenges and options for policy. Development PolicyReview, 19, 521–532.

Karugire, S. R. (1980). A political history of Uganda.London: Heinemann.

Kisakye, J. (1997). Political background to decentraliza-tion. In S. Villadsen & F. Lubanga (Eds.), Demo-cratic decentralization in Uganda. A new approach tolocal Governance. Kampala: Fountain Publishers.

Mamdani, M. (1976). Politics and class formation inUganda. Kampala: Fountain Publishers.

Mamdani, M. (1995). The politics of democratic reform?critical reflections on the NRM. Kampala: MonitorPublications.

Mamdani, M. (1996). Citizen and subject: contemporaryAfrica and the legacy of late colonialism. Kampala:Fountain Publishers.

Manor, J. (1999). The political economy of democraticdecentralization. Washington: World Bank.

Manor, J. (2000). Local government in South Africa:potential disaster despite genuine promise. Brighton,UK: IDS background paper.

Moore, M., & Putzel, J. (1999). Politics and poverty: abackground paper for the World Development Report2000/1. Brighton, UK: Institute of DevelopmentStudies.

Mutibwa, P. (1992). Uganda since independence. Kam-pala: Fountain Publishers.

Nickson, A. (1995). Local government in Latin America.Boulder, CO: Lynne Rienner.

Nsibambi, A. (1998). Decentralization and civil society inUganda: the quest for good governance. Kampala:Fountain Publishers.

Odongo, J. M. O. (2000). A political history of Uganda:The origin of Mueveni�s referendum 2000. Kampala:Monitor Publishers.

Tidemand, T. (1994). Resistance councils in Uganda––Astudy of popular democracy and rural politics. Ros-kilde, Denmark: Roskilde University, Institute ofDevelopment Studies.

Uganda, (1993). Local government statute. Kampala:Government Printer.

Uganda, (1995). Constitution of the republic of Uganda.Kampala: Government Printer.

Uganda, (1997a). Local government act. Kampala:Government Printer.

Uganda, (1997b). Ministry of Finance, Planning andEconomic Development, Poverty eradication actionplan, a national challenge for Uganda, Vol. 1. Kam-pala: Government Printer.

Uganda, (2000). Ministry of Agriculture, Animal Indus-tries and Fisheries and Ministry of Finance, Planningand Economic Development, plan for modernisation ofagriculture and eradication of poverty in Uganda.government strategy and operational framework.Kampala: Government Printer.

Uganda, (2001). Government of Uganda and donorsubgroup on decentralization, fiscal decentralizationin Uganda: the way forward. Kampala: GovernmentPrinter.


Documents

Balancing Rural Poverty Reduction and Citizen Participation: The Contradictions of Uganda’s Decentralization Program