decentralizing basic infrastructure services

Basic infrastructure services are those that house-holds and businesses tend to use every day. Theprimary components are roads and transporta-tion services, water supply and distribution, andsanitation—wastewater collection and removal, andsolid waste collection and removal. These servicesare often bundled together under the heading of“urban services.” However, parallel infrastructuresystems are found in rural areas, where roads, irriga-tion networks, and latrines are critical to life. Irriga-tion systems, in particular, are sometimes underesti-mated as infrastructure networks. In the Philippines,irrigation systems account for 80 percent of nationalwater consumption.

Efforts to decentralize infrastructure servicesraise distinctive issues regarding policy design andimplementation. Because of their capital intensive-ness, these systems require decision making at sev-eral different stages:

• Preparing capital investment plans and settingpriorities for individual capital projects.

• Operating a network system to provide servicesand maintaining facilities to sustain the physicalcapital.

• Financing the system by both mobilizing capitalto pay for the initial investment and generatingrevenues to cover operations and maintenance—that is, ensuring financial sustainability.

209

10

Decentralizing BasicInfrastructure

ServicesGeorge E. Peterson and Elisa Muzzini

Each of these decision-making points presentsan opportunity for decentralization, and differentcountries have responded by decentralizing differentstages of the decision-making process. The issue isfurther complicated because infrastructure projectsare subject to spillover effects and economies ofscale. When this is the case, there are often substan-tial benefits to be achieved by coordinating projectsacross subnational governments. Accounting foreconomies of scale and externalities is especiallyimportant when decentralizing decision makingand responsibility to low levels, and when infra-structure projects cover multiple jurisdictions, suchas in managing water resources across large water-sheds or trunk roads that connect regions. Promot-ing equity, harmonizing standards, and ensuringefficient revenue collection may also argue for lim-iting decentralization.

Infrastructure services are distinctive in anotherimportant respect. Because they are used so widelyand often, citizens are familiar with their benefitsand typically have strong opinions as to whichtypes of projects and service improvements shouldhave highest priority in their area. Choices amongpriorities for infrastructure investment oftenprovide local citizens with their first opportunity toparticipate in public decision making. Participatorychoice at the local level comes more naturally withsmall investment projects—which often quickly

east_209-236.qxd 5/17/05 11:43 AM Page 209

210 East Asia Decentralizes

yield local benefits—than with school curricula andhealth services, which often require professionalexpertise and longer waiting times to producebenefits. Thus, decentralization strategies ofteninclude participatory choice in (small-scale) infra-structure investments, not simply to better respondto local service needs but as a deliberate seedbed fordemocratic participation in governance, with theintention of strengthening civic commitment tothe entire decentralization program. The payoff isjudged only partly by whether infrastructure serv-ices improve; equally important is whether mecha-nisms for public participation in decision makingstrengthen citizen involvement in governance.

This chapter views decentralization in the infra-structure sector as a work in progress. The transferof service and investment responsibilities to thelocal level evokes an immediate response based onthe capacities and institutional practices of localgovernments relative to those of the central govern-ment. More important, however, this first round ofadjustment will reveal shortcomings in the newarrangements and stimulate responses by nationalagencies, local officials, local citizens, and interna-tional organizations supporting the decentraliza-tion process. The fact that East Asian countries haveexperience with both different strategies andimplementation periods for decentralizing infra-structure enhances the opportunity to learn fromcomparative results.

A final introductory observation is appropriate.Decentralization tends to be seen as a transfer ofresponsibilities and revenues from higher-level tolower-level governments. Many decisions aboutinfrastructure services, however, occur in a broadercontext that consists of unbundling the vast powersof previously centralized agencies responsible forinvestment and service provision. Some of thisunbundling entails transferring authority to lowerlevels of government. Other elements may includetransferring investment or responsibility for servicedelivery to the private sector or to public-privatepartnerships, including partnerships with non-governmental organizations (NGOs); and restruc-turing the public enterprises traditionally responsi-ble for larger-scale infrastructure services, such asurban water supply and wastewater removal. Decen-tralization typically entails restructuring publicenterprises by making them accountable to localgovernments rather than central line ministries.

This chapter has two purposes. The first is toreview the status of efforts to decentralize infra-structure services in East Asia, with a focus oncountries that are more advanced in the process,namely, China, Indonesia, and the Philippines.The second purpose is to highlight the efficiencygains in providing infrastructure services achievedin decentralized settings, and to underline themain challenges to realizing the full benefits ofdecentralization.

The next section examines the key features ofstrategies for decentralizing infrastructure adoptedby China, Indonesia, and the Philippines, and out-lines the impact of decentralization on the level ofinfrastructure investment. The third section ana-lyzes the available evidence from East Asia on theefficiency gains from more decentralized formsof providing infrastructure services. The fourth sec-tion reviews East Asian experience in enhancingcommunity-level participation in managing infra-structure projects, and investigates the scope forscaling up the benefits of community participationwithin a decentralized government structure. Thefifth section compares the approach to financinginfrastructure adopted by China, Indonesia, and thePhilippines, given their different strategies for decen-tralizing infrastructure. The sixth section discussesthe critical role of higher tiers of government indecentralized infrastructure, drawing on the experi-ences of China and of Indonesia and the Philippines,with the drawbacks of a “missing middle” in thearchitecture of decentralization being particularlymanifest in the latter two countries. The last sectionhighlights key policy issues that have emerged fromthe analysis of decentralization in the region.

Decentralization Strategies in the Infrastructure Sector

East Asian countries have followed two broadstrategies in decentralizing the infrastructure sec-tor. China exemplifies a principal-agent approach.The central government as principal has retainedand even strengthened its role in setting investmentpriorities across and within sectors, and has rein-forced this role by setting highly specific targetsand timetables for infrastructure coverage in differ-ent classes of cities. These targets and timetableshave extended beyond physical investments toinclude, for certain services, mandatory adoption

east_209-236.qxd 5/17/05 11:43 AM 10

by all urban governments of specific guidelines forservice charges, and, for all services, mandatoryseparation of asset ownership from service deliv-ery. In this respect, the infrastructure sectorremains highly centralized. At the same time, asagents for implementing centrally established poli-cies, local governments have full responsibilityfor actually executing investments and providingservices. They also have significant latitude in decid-ing how to mobilize funds to pay for capitalinvestment, which they now must finance entirelywithout central grant support. Local governmentsfurther have significant latitude in framing devel-opment plans, including where they will site majorcapital projects and how they will sequence invest-ments to meet nationally imposed standards mostcost effectively.

Indonesia and the Philippines represent analternative approach. As part of the Big Bang initia-tives launched in those countries, the central gov-ernments assigned virtually complete responsibilityfor urban and rural infrastructure services to localgovernments almost overnight. At the heart of thistransfer was local choice in investment priorities.Decentralization laws emphasize the importance ofcivic participation in making investment choices,and specify elaborate procedures designed toensure that citizens, as well as collective groups likeNGOs and civil society organizations, are repre-sented in the priority-setting process for capitalprojects. In fact, decentralization is clearly intendedto serve a dual purpose: to make investment choicesin the infrastructure sector more responsive tolocally perceived needs and thus more efficient; andto become a vehicle for introducing ordinary citi-zens to participation in governance.

Cambodia and Vietnam are in incipient stagesof decentralization, with broader policy drivenlargely by the design of infrastructure programs.Cambodia’s Seila Program, introduced in 1996, hascreated commune development committees in morethan 1,000 villages and 100 communes, with theprogram expected to reach three-fourths of all com-munes by the end of 2004 and the rest shortly there-after (Royal Government of Cambodia 2003b). Theprogram provides government and donor funds forsmall infrastructure projects selected by citizens atthe most grassroots level, with mechanisms fortransmitting their priorities for slightly largerprojects up to the commune level. Seila is expressly

seen as a way of engaging the citizenry in participa-tory governance. In a country as poor and rural asCambodia, Seila now accounts for the bulk of localinfrastructure investment. Vietnam has passed legis-lation on grassroots participation as part of its publicadministration reform. However, the national levelcontinues to set investment priorities for most infra-structure services and certainly for urban services,with provincial and local authorities viewed prima-rily as agents implementing national choices.

Investment Levels and National Standards

How has infrastructure investment fared duringdecentralization, and what is the role of nationalstandards and investment targets in sustaininginvestments? Both China and Vietnam, which haveemployed the principal-agent model, have experi-enced extremely high—almost unprecedented—infrastructure expansion within the areas given toppriority. During the latter half of the 1990s, Chinaassigned top infrastructure priority to road build-ing, implemented primarily by provincial govern-ments for national and provincial highway net-works, and by local governments for urbannetworks (see tables 10.1 and 10.2). Both levelsattracted some private investment under the direc-tion of provincial and local authorities. Almost85 percent of China’s road and highway investmentover the two decades ending in 2000 occurred dur-ing 1996–2000. Although national and provincialhighway networks—that is, expressways and class 1highways—expanded most rapidly, all classes grewat high rates, including major urban roads andclass 2 and other local roads.

Vietnam has assigned priority to piped water dis-tribution in urban areas, with implementation inthe hands of provincial water authorities and, in the

Decentralizing Basic Infrastructure Services 211

TABLE 10.1 Road and Highway Investmentin China

Billions of Share of Period yuan total (%)

1981–1989 19 21990–1995 153 141996–2000 881 84

Source: Mitchell Stanfield & Associates, asreported in Bellier and Zhou 2003.

east_209-236.qxd 5/17/05 11:43 AM Page 211

largest cities, water enterprises attached to local gov-ernments. Despite high rates of urban populationgrowth, coverage rates have expanded remarkablyin a short time (see table 10.3). In 2000, 41.5 percentof the average urban water system was five years oldor younger.

China’s government has recently given especiallyhigh priority to urban wastewater collection andtreatment, reflecting heightened concern over thecondition of urban water bodies. Higher standardsimposed on local governments show how thecenter transmits signals to local authorities in aprincipal-agent relationship.

In May 2000, China’s Development PlanningCommission—under the Environment ProtectionAgency in the Ministry of Construction—circulated“City Dirty Water Treatment,” which specified thatthe wastewater treatment rate in all towns and citieswould be at least 50 percent by 2010. The treatmentrate for cities would be at least 60 percent, andthat for provincial capitals and other major cities atleast 70 percent. The circular also defined treatmentquality standards for different types of cities(Government of China 2000). Given that the vastmajority of China’s cities then had no wastewater

treatment of any kind, the targets imply a massiveinfrastructure investment program.1 Recognizingthe need to mobilize capital for such an effort, theDevelopment Planning Commission in October2002 stated that “cities with existing wastewater andgarbage treatment facilities shall start to immedi-ately charge a treatment tariff,” and that all othercities should do so by the end of 2003. The tariff“shall cover operations cost and a reasonable invest-ment return” for wastewater treatment plants, togenerate revenue for raising commercial investmentfunds. Cities in better economic conditions wereurged to set tariffs high enough to cover the cost ofconstructing wastewater collection networks.

In a system with strong upward accountabilitysuch as China’s (and Vietnam’s), local authoritiestake national investment targets seriously.2 Eachmunicipality incorporates specific targets for infra-structure coverage into its five-year developmentplan, approved by the provincial government andultimately by the Development Planning Commis-sion. The political careers of local officials in theCommunist Party hinge on meeting or surpassingthe goals. As a result of either conscientious plan-ning or competitive zeal, local officials often set tar-gets that exceed national standards. This is true inthe wastewater area. Analysis conducted for CityDevelopment Strategies has found that the majorityof covered cities—although located in the poorerwestern provinces—are on their way to meetinginvestment targets that surpass state-mandated lev-els (Chreod Ltd. 2003). Local development plansemphasize these ambitions throughout urbaninfrastructure: municipalities have set targets evenfor square meters of green space per capita, trigger-ing local investment in parks and other green areas.


TABLE 10.2 Road and Highway Investment in China, by Type

1990 2000

Road type Billions of yuan Share Billions of yuan Share

Express-ways 0.01 0.1% 56 28%Class 1 0.02 0.3% 30 15%Class 2 0.20 4.0% 60 30%Class 3 0.84 16.5% 12 6%Class 4 2.60 51% 27 13.5%Unclassified roads 1.43 28.1% 15 7.5%

Source: Mitchell Stanfield & Associates, as reported in Bellier and Zhou 2003.

TABLE 10.3 Piped Water Coverage withinthe Urban Population, Vietnam

Region

Year North Central South

1997 42.0% 30.4% 40.4%2000 52.1% 38.9% 42.7%

Source: World Bank 2002b.

east_209-236.qxd 5/17/05 11:43 AM Page 212

Although driving higher infrastructure investmentlevels, standards that emphasize the capacity ofcapital facilities—whether or not they actually areoperating or doing so economically—have often ledto significant inefficiencies in operations and main-tenance. The next section discusses these impacts.

China’s intergovernmental system is graduallymoving toward more sophisticated and meaningfulmeasures of infrastructure performance, such asoutcome measures. In the wastewater area, forexample, the national government now requireslocalities to test the quality of receiving bodies ofwater, which are subject to quality standards meas-ured along seven dimensions. Some cities voluntar-ily sample discharge quality from wastewater treat-ment plants and have included locally definedtargets in their five-year development plans. Thus,the principal-agent relationship has proved to bemore than a one-way street. The implementingagent not only incorporates mandated standardsinto its planning but may also set higher standardsthat become the basis for upward accountability.

Lessons Learned. The strong investment perfor-mance of subnational governments in China reflectsmany factors. The same national investment priori-ties communicated to local governments have beentransmitted to the state-controlled banking system,clearing the way for lending that has financed muchof the expansion in infrastructure coverage. Com-mercial banks have lent these funds for three- to five-year periods, creating the need for municipalities toroll over short-term debt. Local governments aretherefore beginning to face high debt service bur-dens, which may exacerbate the uncertain credit-worthiness of loan portfolios in the banking system.China also has a tradition of strong policy directionfrom the center, coupled with a high degree of defacto freedom in local budget management notfound elsewhere.

One lesson that can be generalized, however, isthe power of performance measurement andaccountability in China. Measurable performancetargets tied to upward accountability have drivenChina’s infrastructure investment. The quantifiedstandards have sometimes proved unduly rigid,upward accountability has substituted for account-ability to clients, and national standards have cur-tailed local investment choice. But the effectiveness

of infrastructure performance standards in steeringbudget choices at the local level is clear.

“Autonomous” Decentralization

Countries such as the Philippines and Indonesiahave opted for political decentralization, with localauthorities formally recognized as autonomousbodies. Inherent in their powers is setting prioritiesfor local budgets, including capital budgets. Con-cern has arisen in both countries as to whether thistype of decentralization can sustain capital invest-ment and maintenance. In particular, the transfer oflarge numbers of central government employees—subject to wage protection—to local rolls, and thelegal and political difficulties of raising local rev-enues, subject subnational governments to budgetpressures. In the face of such pressures, local gov-ernments are thought more likely to maintainemployment levels rather than adjust their budgetsto sustain investment. Within capital programs,spending on maintenance and repair is believed tobe particularly vulnerable. Displacement of localinvestment has potentially serious consequences.The World Bank has estimated that, in Indonesiafor example, some 60 percent of total developmentexpenditures are now a local responsibility (WorldBank and Asian Development Bank 2003; WorldBank 2003b, 2003c).

The Philippines has the longest experience withlocal budget allocations following decentralization.The share of capital spending in city and municipalbudgets surged in 1993, shortly after new revenueand expenditure assignments took effect (seetable 10.4). That occurred because initial revenueallocations exceeded the cost of transferred func-tions, leading to sizable local surpluses, which werethen drawn down by capital investment. Since1993, however, the share of local budgets devoted tocapital spending has fallen steadily, with declinestotaling more than 50 percent. These declines accel-erated in 1998 with the Asian financial crisis andthe consequent loss of public revenues before reviv-ing briefly the following year. Such spending doesnot tell the entire story, as central line ministriescontinue to pay for some capital projects at thelocal level, as do some congressional allocationsthat are treated off-budget. However, the overalltrend in local investment is clear.


east_209-236.qxd 5/17/05 11:43 AM Page 213

Indonesia’s decentralization process is too youngto draw comparable conclusions about its impact oninvestment spending. Development spending onroads and mainland transportation fell sharply dur-ing the Asian financial crisis—from 15 percent oftotal development spending in 1994–97 to a littlemore than 5 percent in 1999–2001—before recover-ing somewhat. Another potential factor affectingthis slowdown is significant deconcentrated spend-ing by line ministries, and the difficulty of adaptingthe DAK (dana alokasi khusus, a conditional equal-ization grant) and providing funds within a decen-tralized environment. The World Bank has expressedconcern that road maintenance at the Kabupatenlevel has suffered from underfunding, and thatdecentralization may exacerbate neglect of roadmaintenance, with consequences that are not imme-diately visible (World Bank 2003d). Whether localmaintenance and repair budgets have actually suf-fered disproportionately from local budget adjust-ments under decentralization is unknown. However,protecting maintenance funds may require a greaterrole for professional planners and engineers, asopposed to local political officials and civic groups, inallocating expenditures.

The apparent decline in local infrastructurespending in the Philippines, and concern over localinvestment levels in Indonesia, have generateddebate about the role of national standards and per-formance measures. This debate is occurring mostvigorously in Indonesia, where decentralization leg-islation calls for national agencies to develop guide-lines rather than mandatory standards promulgatedvia the provinces, whose capacity for oversight hasgreatly weakened.

Upward accountability for complying withmandatory infrastructure standards seems funda-mentally at odds with strategies that place primaryimportance on local choice in investment projectsand priorities. The challenge is to capture the powerof performance measurement and accountabilitywithin a framework for political decentralization.

The first step is agreement on performance meas-ures in the infrastructure sector. National guidelinescan establish a handful of basic measures reportedby all local authorities and monitored by centralinstitutions. Such measures would include funda-mentals such as:

• Hours per day or week of water provision.• Coverage of road networks (kilometers per

1,000 persons, kilometers of road per squarekilometer of territory); quality of roads (percentin good condition, percent of all-weather roads).

• Wastewater removal rates.• Expenditure on road maintenance as share

of gross domestic product (GDP) or kilometersof roads.

• Affordability of transport services (freight rateper ton per kilometer, average bus fare per pas-senger per kilometer).

The absence of such measures handicaps bothnational decisions about infrastructure prioritiesand local attempts to measure and improve invest-ment efficiency.

Decentralization frameworks now provide ampleopportunities for incorporating these measures intoan accountability system. At the national level, cate-gorical grants such as Indonesia’s DAK alreadyrecognize performance measures, although so farthey do not actually take them into account. Wherenational authorities have clear investment prioritiesnot adequately captured by local decision making,they can introduce standards—either absolute levelsor improvement from a baseline—as a condition ofcapital grants. The ratio of matching grants canreflect both national priorities and local perfor-mance. Simple benchmarking of infrastructure per-formance and transparent reporting of performancemeasures can enhance efficiency. Individual localauthorities, for example, can cite the range of actualcosts per kilometer of standardized road construc-tion elsewhere in conducting competitive bidding


TABLE 10.4 Capital Spending as a Share of Local Government Expenditure, Philippines

1992 1993 1994 1995 1996 1997 1998 1999 2000

Cities 10.7% 17.3% 17.6% 16.4% 11.1% 8.8% 8.7% 8.9% 8.2%Municipalities 8.7% 9.8% 8.9% 7.5% 6.5% 6.8% 4.2% 5.1% 4.7%

Source: Orial 2002.

east_209-236.qxd 5/17/05 11:43 AM Page 214

for local construction.3 Civic groups can comparebasic output levels with those of other local authori-ties in setting up their own reporting systems.

For other types of infrastructure projectsand service delivery, formal accountability to thecitizenry through contractual agreements and per-formance monitoring appears to be the most effec-tive way to use standards under autonomous decen-tralization. Experience with 24 water utilities inIndonesia, known as PDAMs, shows that NGOscan monitor business plans that specify improve-ments in water coverage, quality, and reliability—established after public debate as part of a socialcontract. Fulfilling these goals then becomes thebasis for raising service tariffs and supportingfinancial sustainability (Urban Institute 2003). Sev-eral local governments in the Philippines have sim-ilarly published performance goals and invitedmonitoring by the nongovernmental sector afterpublic participation in setting priorities for infra-structure services (World Bank and Asian Develop-ment Bank 2003).

Lessons Learned. In the politically decentralizedsystems of East Asia, a lack of standardized perfor-mance monitoring severely hampers understandingof local infrastructure. No country now has a rou-tine monitoring and reporting system for the sector.Such systems can be built gradually and improvedover time. However, national guidelines shouldquickly establish a rudimentary system of reportingthat can be built into budgeting at all levels. Upwardaccountability is more difficult to establish in politi-cally decentralized systems than in systems thatremain centralized. Decentralized systems need toincorporate performance goals into local budgets,local corporate plans for water utilities, and localdevelopment planning—in a form that can be mon-itored by civil society organizations. Quantitativeaccountability to informed civic monitors can sub-stitute for traditional upward accountability, butonly if public agencies provide specific and verifi-able information.

Efficiency Gains from Decentralizing Infrastructure

A fundamental argument in favor of decentralizinginfrastructure is that moving decision making oninvestment and implementation closer to clients

will yield efficiency gains. These gains can comefrom two sources. First, production efficiencyimplies that local entities can build and operate agiven infrastructure package less expensively. Costsavings may derive from cheaper local buildingmaterials, less expensive local labor, more efficientproject design, fewer layers of bureaucratic over-sight, and less corruption, among other sources.Sustainability is an important aspect of productionefficiency. Second, allocative efficiency implies thatlocal investment priorities will reflect the prefer-ences of citizens more than those of the centralgovernment, and that households will thereforevalue each unit of infrastructure spending morehighly. As this chapter later shows, governmentsneed to balance such efficiencies with possibleeconomies of scale (which for some infrastructuresectors can be significant) and externalities (bothnegative and positive) across local jurisdictions.

Analysts have made several attempts to test, inEast Asia, the hypothesis that decentralizationenhances the efficiency of infrastructure services.Rigorously evaluating the gains from productionand allocative efficiency is a demanding exercise.Evidence so far suggests efficiency gains, but it is farfrom conclusive, as it is drawn largely from casestudies (this differs from the health and educationsectors, where systematic monitoring and householdsurveys are far more common). As a result, we knowa good deal more about how to extract efficiencygains from decentralized infrastructure throughwell-designed implementation strategies than we doabout broad generalizations as to whether decentral-ized approaches, on average, are more efficient.

Production Efficiency

The World Bank has reported that village infra-structure constructed under Indonesia’s KecamatanDevelopment Program, which involves community-level planning and implementation, “cost signifi-cantly less—on average about one-third but inmany cases more than half less—than equivalentworks built through Ministry of Public Works con-tracts.” Maintenance costs were also reportedlylower because communities provided the labor.Unfortunately, the analysis supporting this conclu-sion has not been published, leaving open the ques-tion of how the study compared costs. In particular,such comparisons often do not take into account


east_209-236.qxd 5/17/05 11:43 AM Page 215

the social infrastructure costs of supporting localproject development (World Bank 2003e).4

In collaborating with 24 PDAMs in Indonesia,the Urban Institute reported that the PDAMsbelieved they could substantially reduce investmentcosts when they handled construction, land acqui-sition, project design, and scheduling, as opposedto complying with central specifications. Savingson projects that produced a given increment in thecapacity of the daily water supply reportedly run ashigh as 50 percent. Some PDAMs have expressedreluctance to accept low-cost loans from centralsources, should these become available, if they mustadopt centrally imposed project standards, believ-ing that the costs of complying with those ruleswould more than offset the savings from low-costcredit (Urban Institute 2003).

Loehr and Manasan (1999) conducted one of themore ambitious attempts to compare productioncosts for standardized projects in the Philippines.Drawing on World Bank and other data, the authorsfound costs in the range of P= 180,000–235,000 whenlocal governments built their own schools, com-pared with P= 305,000 per classroom for the centralDepartment of Public Works and Highways. Alonzo(1998) has reported comparable differentials in thecosts of locally versus centrally built classrooms inthe Philippines, as well as local savings per kilome-ter of road construction.

Corruption is a major source of cost escalationin infrastructure projects throughout East Asia.Azfar et al. (2000) have estimated that it adds20–40 percent to the cost of infrastructure projectsin the Philippines. Respondents to a 1999 surveyconducted by Social Weather Stations ranked the

central Department of Public Works and Highwaysas the most corrupt organization in the Philippines(Azfar et al. 2000).

Because a reduction in corruption costs is anexpress rationale for decentralizing infrastructureservices, the findings of household and other sur-veys asking respondents to compare the severity ofcorruption at local and central levels therefore holdspecial interest. In a survey of 468 respondents in13 randomly selected Kabupaten/Kotamadya in WestJava, Indonesia, Azfar (2002) found that 29 percentpaid fewer bribes following decentralization, whileonly 5 percent reported that they paid more.Respondents attested to similar changes in the costof bribes, or “unofficial payments” (see table 10.5).At least in the eyes of citizens, corruption becomesmore widespread and costly the further removedgovernment agencies are from the local level.

Although this evidence supports the hypothesisthat decentralization reduces corruption, muchdepends on how reform is implemented. In thePhilippines, infrastructure projects selected by LocalDevelopment Councils but built by the Departmentof Public Works and Highways or other centralagencies tended to trigger a cascading effect ofunofficial payments at each layer of government(Hofman and Kaiser 2002). Analysts have alsoreported examples of corruption and rent seekingamong local councils, including cases where legisla-tors have voted themselves large salary increases andautomobiles, and where a local council has voted tosimultaneously approve an investment project andname the party to be awarded the contract. Thus,blanket generalizations about comparative corrup-tion are suspect.


TABLE 10.5 Citizen Perceptions of Corruption in Different Layers of Government in Indonesia

Rare/ Quite/veryCorruption Never infrequent Common common

Local government 17.7% 25.4% 23.3% 11.3%Kabupaten/Kotamadya 1.7% 35.7% 28.2% 18.0%Provincial government 0% 19.5% 29.9% 33.3%Central government 0.2% 4.5% 37.0% 48.1%

Source: Azfar 2002, p. 7.Note: Percentages do not sum to 100 because the figures exclude “don’t know” responses.

east_209-236.qxd 5/17/05 11:43 AM Page 216

Probably the strongest evidence for gains in pro-duction efficiency from decentralizing infrastruc-ture comes from studies designed to determinewhether latrines and small-scale water distributionprojects were still functioning and actually used byvillagers several years after installation. Studies inCambodia, Indonesia, Lao People’s DemocraticRepublic, and Vietnam all concluded that projectsthat relied on community consultation on design,and community organizations for maintenance andoversight, were significantly more likely to be sus-tained than projects built from a central designwithout such consultation, or that relied on outsideexpertise for maintenance and management (Gross2003; Chanthaphone and Lahiri 2003; WorldBank 2002a). Production efficiency overlaps withefficiency in allocative choice and community par-ticipation (discussed later in the chapter). RuralIndonesians, for example, were found to have astrong preference for pour-flush latrines, whichwere far more likely to remain in operation thanalternative designs. Women assigned a high priorityto small-scale water and sanitation projects, andcommunity-scale water management organizationswith strong participation by women proved moresustainable than organizations operated solely ordominated by men.

Such evidence tends to confirm that decentraliz-ing projects to the village or commune level allowscommunity involvement in support of sustainabilityto emerge. However, simply decentralizing invest-ment and management decisions does little to pro-mote sustainability unless primary users participatein maintenance and management decisions.

On a larger scale, the efficient use of capital inmunicipal utilities relates directly to performancemeasures targeted and rewarded under nationalaccountability systems. One of the apparent para-doxes of China’s investment in municipal waste-water treatment plants is that local governmentsare investing heavily in new plants while existingfacilities operate at 50 percent or less of designedcapacity because of lack of funds for operationand maintenance. This phenomenon has beenreported in Hunan province, among other loca-tions (Chreod Ltd. 2002). It reflects the fact thatinfrastructure targets initially included in localdevelopment plans and monitored by higher-levelauthorities were based on the treatment capacity of

completed wastewater treatment plants. Under thisyardstick, localities met coverage targets whetheror not treatment plants were actually operating, asno one measured the volume of treated dischargeor the quality of receiving water bodies. Yet inChangsha, the capital of Hunan province, waste-water treatment plants operated at 50 percentcapacity for two years because the city simply shutoff intake valves and diverted incoming flowsdirectly into the river. Meanwhile, the municipalitywas planning large-scale construction of newplants. This experience reveals the power of per-formance measures tied to accountability proce-dures: poorly selected performance measures canfail to capture the intended effects. Similarly, cen-trally imposed cost targets may jeopardize produc-tive efficiency gains if they fail to account for localvariations in exogenous costs that are not underthe control of municipalities.

When wastewater treatment facilities feed intothe same body of water, coordinating projects acrosslocal jurisdictions can yield better water quality atlower cost. The top-down system in China, com-bined with strong decentralized authorities, how-ever, does not provide enough incentives for localgovernments to consider economies of scale andcoordination. In the Guangdong River delta, forinstance, each of 51 administrative districts has awastewater treatment facility, although cooperativeagreements based on economies of scale could haveyielded significant savings. In Vietnam, in contrast,flexibility in local management and timing of waterdelivery has resulted in significant drops in wateruse per hectare, revealing the gains from productionefficiency that can result from decentralizing ruralirrigation systems (see box 10.1).

Allocative Efficiency

A full argument for gains in allocative efficiencyfrom decentralization would marshal several linesof evidence. First, central and provincial officialsfail to correctly identify the spending priorities oflocal citizens. Second, choices on spending priori-ties made by local officials, as well as choices madeby citizens themselves through participatory proce-dures, better reflect true local preferences. Forexample, reliance on central quality standards, as inthe Chinese model, may significantly reduce the


east_209-236.qxd 5/17/05 11:43 AM Page 217


BOX 10.1 Vietnam’s Red River Delta: Efficiency Gains from Decentralizing Irrigation

The Red River Delta (RRD) in Vietnam has one ofthe highest-density rural populations in theworld. The delta depends on irrigation for cropproduction. Experience shows the efficiencygains that can be obtained from decentralizingmanagement of an infrastructure network, aswell as links between gains from production andallocative efficiency.

Since Vietnam decollectivized its agriculturalsector in the 1980s, two types of institutionshave provided irrigation in the RRD to farmers.One is state-owned Irrigation and DrainageManagement Subsidiary Companies (IDMSCs),which centrally manage pumping stations andirrigation networks. The other is cooperatives,which operate on a smaller scale through jointmanagement and localized pumping stations.Cooperatives entered the irrigation business inresponse to inflexibilities in water provision fromthe centrally managed state companies, whichscheduled uniform water flows in advanceregardless of local conditions or planting pat-terns. Cooperatives obtain raw water suppliesthrough contracts with regional irrigation man-agement companies. The cooperatives are, ineffect, decentralized alternative suppliers of irri-gation services to farmers. In the RRD, IDMSCsand cooperatives each serviced about half ofthe land, facilitating comparisons between theinstitutions.

Cooperatives have improved allocative effi-ciency by making water delivery more flexible inresponse to farmers’ demands. Empirical obser-vations over an irrigation season found that, onaverage, cooperative pumping stations providedwater flows within 2 days of a request, comparedwith 11.5 days for centrally managed systems,which work off a predetermined rotational sched-ule. The shorter delivery time and greater flexi-bility of cooperative supply give farmers morechoice in cropping patterns, rice varieties, andgrowing conditions. Production efficiency can bemeasured in terms of water use per hectare—table 10.1B below shows observed differences inwater use for comparable paddy fields.

Part of the observed differential in water usereflects the shorter canal networks of local coop-eratives, which reduce water losses and presentfewer opportunities for diversion to illegal users.Part of the savings comes from managementefficiency motivated by the desire to save oncontracted costs for water supply. An importantpart of the savings, however, comes fromreduced corruption. Staff of the state-ownedcompany reportedly made illicit agreements tosell water on the side, adding to the amountspumped per eligible hectare.

Sources: Fontenelle and Molle 2002; Fontenelle2000.

TABLE 10.1B Average Volumes of Water Pumped per Hectare, Spring Season 1996,Vietnam (cubic meters)

Water used to Water used during Supplier prepare land growing season Total

Local station 1,600 2,400 4,000Centrally managed 3,900 5,900 9,800

Sources: Fontenelle and Molle 2002; Fontenelle 2000.

gains in allocative efficiency from decentraliza-tion if it prevents municipalities from adjustingthe quality of service to the preferences of theirconstituencies.

Analysts have collected evidence on parts of thisargument. For example, Azfar et al. (2000) foundthat Filipino households strongly favor spendingincremental funds on roads, but that household

preferences vary substantially across locations, pre-sumably reflecting differences in both local valuesand conditions. Municipal officials had a statisticallysignificant ability to identify local preferences, whileprovincial officials had no ability whatsoever toidentify local preferences—with a negative correla-tion between their predictions and actual householdpreferences. In particular, provincial officials vastly

east_209-236.qxd 5/17/05 11:43 AM Page 218

underestimated local demand for spending on roadsand other local infrastructure.

Household surveys in Indonesia and Cambodiahave also found strong preferences for road con-struction as the top investment priority. Interna-tionally funded programs that involve communitychoice report a significantly higher level of invest-ment in roads than among projects whose outputsare negotiated at the central level. Local develop-ment projects that incorporate citizen participa-tion in project selection appear to more accuratelyreflect both the general preference for roads andvariations across communities. For example, duringthe first two years of the Kecamatan DevelopmentProgram in Indonesia, road projects were by far themost popular local investment choice (62 percent),compared with bridges (10 percent), irrigation(8 percent), and clean water (7 percent). Follow-upsurveys found that even given these percentages,households felt that too much had been expendedon clean water. Evidence from these and other stud-ies, in short, is that citizens have clear-cut prioritiesfor spending, and that distant representatives andbureaucrats do not grasp these priorities or assigngreat importance to local priorities. Moreover,local demand for simple infrastructure projects—primarily roads—is high compared with alternatives.Of course, how much deference central governmentsshould pay to household preferences in allocatingspending across sectors is an open question, espe-cially when sectors like education and health generatepositive externalities not fully captured by local resi-dents. Nonetheless, for spending assigned to localinvestment choice, evidence supports the conclusionthat decentralization is closing the gap between localpreferences and project selection.

Lessons Learned

More important than the first-round impacts ofdecentralization on the efficiency of infrastructureservices are the lessons that can be learned to betterextract potential gains.

Because the costs of corruption are high in pro-viding infrastructure, countries should take practi-cal steps to reduce corruption and other inefficien-cies under decentralized management. In most EastAsian countries, central authorities have tradition-ally handled infrastructure procurement, even forlocally selected projects supported by significantlocal funding.Meanwhile,however,decentralization

has given local authorities responsibility for definingtheir own procurement rules. This situation has notonly created confusion and inconsistency across thelocal government landscape, but has also led to pro-curement abuses, such as lack of competitive bid-ding and technical evaluation of proposals, steeringof contracts to particular firms by elected officials,and extensive price renegotiation after contracts areawarded. Adoption of a uniform local procurementcode should be a top priority for decentralization, tobuild in competition and transparency. The Gov-ernment Procurement Reform Act of 2003 in thePhilippines offers a model of such support. An over-sight and regulatory board empowered to investi-gate and punish procurement abuses needs to sup-plement formal procedures.

Best-practice local examples suggest othersteps for improving procurement. Naga City in thePhilippines, for example, now publishes in the news-paper and on the Internet winning per unit pricesfor all procurement contracts. This simple actionboth supports transparency and encourages pricecompetition among suppliers (World Bank andAsian Development Bank 2003). In administeringthe Kecamatan Development Program in Indonesia,the World Bank has found that the simple expedientof requiring that an independent (local) third partyas well as contracting principals sign off on all con-tracts and procurements, however small, saves fundsand fosters a culture of transparency.

Excess capacity in infrastructure design is a majorsource of cost inefficiency. Paradoxically, at a time oflarge infrastructure backlogs, a number of projectssuffer from substantial overcapacity, raising costsunnecessarily. In some cases, excess capacity hasresulted from centralized application of standardproject designs or per capita use estimates thatdo not take local conditions into account. SomeIndonesian PDAMs are operating at only one-thirdof designed capacity because abundant groundwa-ter sources are available and households prefer tocontinue pumping their own water rather than payfor connecting to the piped water system.

Many Chinese cities suffer from overestimatedwater demand because authorities failed to antici-pate the drops in demand that would occur with full-cost water tariffs and the rapid decline in state-owned firms—often the most inefficient water users.This situation has left some local governments with“take or pay” contracts with private companies,


east_209-236.qxd 5/17/05 11:43 AM Page 219

which require governments to either purchase morewater than consumers will use or displace lower-costmunicipal water with higher-cost supplies. As inall countries, grant and low-cost loan financing alsopromotes excessive scaling of infrastructure facili-ties. When investment appears to be free to localauthorities, they tend to base estimates of futuredemand on the most optimistic assumptions. Com-parable projects financed from own-source fundsreflect more realistic growth projections and thetime and cost entailed in tying up funds. Again, thepractical lessons are straightforward. Whereverpossible, major infrastructure projects should occurat the local level after reforms in user prices, so offi-cials can estimate the impact on demand more accu-rately. Uniform per capita projections of use shouldbe scuttled in favor of demand studies based on localconditions. All infrastructure projects of significantscale should require substantial own-source financ-ing, to create incentives for realistic cost projectionsand savings.

Predicting a country’s optimal infrastructurecapacity compared with its long-term investmentneeds is admittedly complex, given the “lumpiness”of investments and the uncertainties associated withlong-term planning. Hence, whereas today’s use ofinfrastructure facilities points to overcapacity, theextent to which that overcapacity is likely to persist isdifficult to gauge.

Gains from Citizen Participation

For countries pursuing autonomous decentraliza-tion, the hallmark of reform has been the promise ofgreater citizen participation in local budget deci-sions, investment choices, and development plan-ning. Such participation is supposed to yield greatercitizen satisfaction with basic services and more cov-erage for previously excluded groups—particularlythe poor, ethnic minorities, and women—while lay-ing the groundwork for broader democratic partici-pation in national government.

Cambodia’s 2003 Poverty Reduction Strategyillustrates the weight of expectations placed ondecentralization as a tool for achieving participation:

“Decentralization has three objectives inCambodia:

• Promote pluralist participatory democracy atlocal level . . . by creation of popularly mandated

and autonomous local governments that areresponsible to the citizens . . . and make deci-sions over delivery of public goods and services.

• Promote the culture and practice of participatorydevelopment (planning, management, resourcemobilization) at local level.

• Contribute to reduction of poverty in thecountry through improvement of service condi-tions . . . and [service] improvement to poorand deprived ones” (Royal Government ofCambodia 2003a, p. 108).

Countries have introduced autonomous decen-tralization primarily through grants to local com-munities that allow citizens to directly choose smallinvestment projects and provide for communitymanagement of the projects once installed. The SeilaProgram has brought citizen choice over small infra-structure projects to Cambodia; the KecamatanDevelopment Program has done the same for sub-districts in Indonesia; as have barangay investmentprograms in the Philippines for urban subdistricts.Several other programs follow the model of provid-ing communities with investment resources thatthey can allocate among eligible projects. Suchprograms have typically won strong support fromthe World Bank, the Asian Development Bank, theUnited Nations Development Programme, andother international organizations seeking to imprintlocal citizen choice as the bedrock support fordecentralization.

Experience with citizen participation in East Asiaraises two critical questions. First, does participa-tion in fact improve sustainability and coverage ofbasic infrastructure services at the village, urbansubdistrict, and neighborhood scale? Second, canexperience in direct participation be successfullyscaled up to larger political units and infrastructurenetworks? As the following evidence reveals, theanswer to the first question is yes, while the jury isstill out on the second question. Despite some prom-ising local experience, scaling up has proved moredifficult than optimistic advocates of participationforesaw.

Community-Scale Participation

A large body of evidence supports the importanceof community-scale participation in infrastructurechoices and management. This evidence also


east_209-236.qxd 5/17/05 11:43 AM Page 220

provides guidelines on specific responsibilities thatcommunities can handle to extract the maximumvalue from participation. For example, in a studyof 88 community-managed water supply projectsworldwide, Gross et al. found that:

• Community-based water supply projects whichwere more demand-responsive—that is, thosethat involved greater community choice in thetype of water supply, households to be covered,and the method for paying for operations andmaintenance—were more likely to be sustainedand better maintained.

• The more broadly participatory and gender rep-resentative the decision-making procedures, thehigher the rate of cost recovery.

• Communities that set up water managementorganizations (WMOs) had better project sus-tainability records than communities that par-ticipated only in initial project selection; themore equitably WMOs represented women andthe poor, the greater their sustainability.

• Household contributions to construction weresignificantly associated with better-sustainedwater supply services only if the communityactively participated in project selection and sub-sequent management (Gross et al. 2001).

Hopkins’ study of 33 sites in Flores in the Philip-pines found higher rates of sustainability whereplanning included both women and men, as wellas poor households. The World Bank has reportedhigher rates of sustainability of locally selected in-frastructure projects when project choice reflecteddirect community participation (Hopkins 2003;World Bank 2003d).

Some of the benefits to be reaped from localparticipation and choice require careful listeningby outside experts, whether national officials orinternational advisors. In 2001 Lao PDR adopted anew policy of allowing communities to choosetheir hygiene solutions and influence their design.Follow-up work found that villagers primarily val-ued the “comfort and convenience” of latrinesrather than the health benefits, which were fore-most in the minds of external experts (Meadley2003). This finding influenced both the design ofthe latrines and the strategy for disseminating theprogram, which relied on “champion families”respected by other families. After the champion

families receive latrine facilities, photos showingtheir convenience and the families’ pride in owningthem are used to engage the widely dispersed ruralcommunity in discussing the benefits of latrinesand gain support for their adoption.

The literature on decentralized fiscal choice hasemphasized the gains from allowing clients tochoose among a wide array of project options, con-strained by either a fixed budget or the requirementthat users pay for the service they choose. ManyEast Asian experiments in decentralized projectselection have followed this model, offering localresidents a broad initial choice of investment prior-ities. The Seila Program in Cambodia and the Keca-matan Development Program in Indonesia are twoexamples that offer relatively open-ended choice.

However, some national programs have definedcommunity choice and participation differently,particularly those that retain strong roles for gov-ernmental planners and service providers. In thepoor rural province of Guizhou, China, for exam-ple, prospective township clients can choose toreceive private water connections at a tariff levelthat covers operating costs plus some 75 percent ofcapital costs, including all debt service. Less expen-sive alternatives, such as public stand posts, are notoffered, nor does the community have a voice intariff policy or technology. Community consulta-tion consists of carefully explaining to residents thekind of water service they will receive, the tariffcosts, and the procedures for collecting the tariffs,which entail house-by-house collection (Zhixiong2003). The community may then embrace the pro-gram and its rules or, in principle, express reluc-tance to pay for it. Village committees identifydelivery problems and exert collective pressure oncitizens to pay water bills. However, the plant man-ager remains in control of all aspects of manage-ment. A portion of his salary is deducted if tariffcollection rates fall below 90 percent.

Scaling Up Participation

The difficulties of scaling up successes in communityparticipation are widely recognized (World Bank2003b). Ironically, scaling up community and ruralvillage projects has proved easier in some respects ina centralized rather than a decentralized environ-ment, where many levels of government exertauthority. Scaling up means multiplying the number


east_209-236.qxd 5/17/05 11:43 AM Page 221

of villages and urban subdistrict communitiesreached using the same implementation principles.Such scaling up is constrained primarily by theresources of a program and the central government’swillingness to support it through its own resourcesor agreement with international donors. If the finan-cial and human resources are replicable, projects canbe reproduced throughout the country. Seila inCambodia is an example of a community participa-tion program that is rapidly scaling up to reach theentire country with direct central support, withoutmuch interaction with other levels of government.The Kecamatan Development Program in Indonesiais another program that has spread quickly by repli-cating the same neighborhood approach, with cen-tral government and donor support. Only now, in itsthird generation, is the program attempting themore difficult task of integrating priority setting atthe community level with the formal planning pro-cedures of the decentralized system.

Scaling up community-identified prioritieswithinadecentralizedgovernmentstructurerequiresfinding ways to transmit community preferences oninfrastructure investments to successively higherlevels of government. Community “demand” mayinclude both community-scale projects that needhigher-level financial support and communityviews on the priority of village, district, and munici-pal investment projects that affect the community.The transmission of community preferences aboutcapital projects to higher levels of government hasproved problematic. Part of the difficulty stemsfrom distrust of representative government and sus-picion of the willingness of municipal officials torespect community investment priorities. This dis-trust has been compounded by the difficult interfacebetween top-down national investment planningand bottom-up community and local planning.Both the Philippines and Indonesia have attemptedto address this challenge by directly involving com-munity groups in progressively higher levels of deci-sion making.

Indonesia illustrates the complexity of transmit-ting community preferences.5 The urban planningprocess begins with village development meetingsattended by the village representative council,NGOs,and a representative of the subdistrict (Kecamatan).A major objective is to submit project proposals tothe Kecamatan subdistrict level. There officialsreview and weed out community proposals and add

new proposals, and then submit a priority list to thenext level of government (Kabupaten/Kota), whichadds proposals from technical officials. The projectpreferences of the Kabupaten parliament alsobecome part of the mix and may override other rec-ommendations. Finally, an umbrella system is sup-posed to coordinate local investment priorities withprovincial and national priorities. As part of thisstructure, the Indonesian government has promul-gated general guidelines for participatory planning atthe Kabubaten and Kota level, including open meet-ings that bring together representatives of communi-ties, NGOs, and technical bureaus as well as munici-pal elected officials. The entire process—facilitatedby a government-provided scoring sheet—shouldyield a consensus list of local priorities.

Case studies reveal that this process plays outdifferently in different locations. Municipal parlia-ments, technical agencies, and NGOs typically havedifferent priorities (Pratikno 2002; IndonesianPartnership on Local Governance Initiatives 2002).Negotiations leading to final project prioritizationat the municipal level depend on the relative cloutof these parties and the role the mayor chooses toplay. Most case studies have concluded that thepreferences of local elites, the municipal parlia-ment, and technical agencies tend to drown out thepreferences of community groups. The mandatedparticipation of NGOs has not resolved this issue,because—instead of representing a consensus oflower-level priorities—NGOs have more oftenproved to be splintered advocacy groups for partic-ular priorities. A similar system of planning andpriority setting operates in the Philippines, includ-ing the mandated participation of communityNGOs in local development councils. Similar diffi-culties in sustaining grassroots participation inmunicipal priority setting have been reported.

A realistic reassessment of what community par-ticipation in infrastructure decision making meansat the municipal scale is in order. Intermediation isplainly required in scaling up from the communitylevel to the municipal level. Elected representativesin the municipal parliament provide one form ofintermediation; NGOs provide another. In themost successful examples of public inclusiveness insetting infrastructure priorities, NGOs have playedthe role of intermediary between community andgovernment. This has required continuous involve-ment by NGOs, starting with community-level


east_209-236.qxd 5/17/05 11:43 AM Page 222

meetings and extending to collaboration with thetechnical agencies of municipal government.Against this preparatory background, structuredmeetings on investment priorities and budget allo-cations, such as citizen forums, can succeed. Theconditions for success, however, are demanding:NGOs must be willing to see themselves as partnerswith local government rather than antagonists,and local officials must be open to input from insti-tutions outside the political and governmentaltechnical sphere. Clear examples of successfulimplementation of this vision do exist. Theseinclude the involvement of the Indonesian Partner-ship on Local Governance Initiatives with bothmunicipal government and communities in urbanforums in Indonesia, and the culture of partnershipin all decision making in Naga, the Philippines.Efforts to include community expression under theKecamatan Development Program and ordinarydecentralized priority-setting procedures inIndonesia are other examples. Whether scaling updirect community participation in setting invest-ment priorities is widely workable remains to beseen, however. Many communities have reported apublic stalemate that gives rise to the older patternof nontransparent decision making by elites.

Lessons Learned

Community participation is essential to the success ofinfrastructure projects at the village and subdistrictlevel. Participation is required at the point ofproject selection as well as in continued projectmanagement. Meaningful participation requiresthe involvement of a cross-section of project users,especially women and minority groups, who areoften excluded from project management.

Scaling up participation in infrastructure choicehas proved difficult. Other mechanisms, such asgenuinely representative municipal governments,and NGOs willing to serve as intermediaries toboth municipal government and local communi-ties, are needed to make the process work.

The most promising approach involves NGO par-ticipation from the start of the priority-setting processthrough the municipal meetings that establish localinvestment priorities. Only NGO involvement at thelast stage—such as through the Local DevelopmentCouncils prescribed by Filipino law—has provedunsuccessful.

The idea that community participation in settinginfrastructure priorities will breed a national cultureof democratic decision making has seen a modestamount of empirical support thus far. Just as scalingup infrastructure priorities can be difficult, so canscaling up expectations about democratic partici-pation in governance. The experience of allocatingbudgets to communities undoubtedly empowershouseholds and raises their expectations aboutresponsive government. Whether such participa-tion helps consolidate national democracy remainsto be seen.

Paying for Infrastructure Services

Infrastructure services must be financed at twolevels. Capital resources must be mobilized to payfor the initial investment in facilities. Then recur-ring revenues must cover the annual cost of opera-tions and maintenance, plus contributions to serv-icing the debt incurred to finance the initialinvestment. In an economically efficient world, thefull cost of infrastructure facilities, including depre-ciation, would be recognized and recovered throughuser fees and—given positive externalities—explicitsubsidies from government, in the form of capitalgrants or targeted subsidies.

Much of the institutional unbundling of largeinfrastructure utilities in East Asia has been moti-vated by a desire to generate more reliable financingstreams, and to make infrastructure services moreattractive candidates for commercial investment orlending. The latter goal has sometimes conflictedwith the goal of making monopoly utilities moredirectly accountable to local governments.

Asia faced a formidable infrastructure pricingchallenge at the beginning of this decade, as it hadthe lowest water and sanitation tariffs—both inabsolute terms and as a percentage of the costs ofproduction—of any other region (see annex 10.1).The median urban tariff for water supply report-edly covered less than 85 percent of operating andmaintenance costs, with no contribution to the costof capital. The median tariff for sanitation coveredan even lower portion of operating and mainte-nance costs. If anything, the tariff ratios reportedby governments are likely to underestimate the truecost gap.

In examining financing approaches, we canextend the distinction introduced at the beginning


east_209-236.qxd 5/17/05 11:43 AM Page 223

of this chapter between China and (in incipientterms) Vietnam, on the one hand, and the Philip-pines, Indonesia, and (in incipient and less clear-cut terms) Cambodia, on the other. Aided by strongeconomic growth, China has devised a broadmodel for self-financed infrastructure investmentand market-based capital financing. Both thePhilippines and Indonesia have been handicappedby a stronger impact from the Asian financial crisisand less robust economic growth. However, theyhave compounded the difficulty by retaining cen-tral control of lending to local authorities for infra-structure investment—an anomaly in their other-wise sweeping embrace of decentralization.

China’s Infrastructure Financing Strategy

China has pursued a clearly defined sequential strat-egy for financing local infrastructure. Although themodel has important weaknesses, it contains les-sons for the rest of the region. To finance the firstwave of investment in local infrastructure net-works, the government relied primarily on localtaxes and fees, supplemented by borrowing frominternational donor agencies and capital grants andbudget assignments from central government. Start-ing in 1998, the government began to borrow heav-ily from the domestic market to finance infrastruc-ture. Between 1998 and 2002 it issued ¥ 660 billion(US$79.5 billion) in infrastructure bonds—some30 percent of which was then transferred to localgovernments, half as subloans and half as grants.Along with this use of the domestic bond market,the government communicated the priority of localinfrastructure lending to China’s banks, all ofwhich are publicly owned. Short- to intermediate-term loans from banks have been a principal sourceof capital financing for local governments investingin infrastructure.

One undesirable effect of the surge in local bor-rowing to finance infrastructure investment hasbeen a high and rising level of municipal indebted-ness. Under existing arrangements, municipalitieshad to repay outstanding debt from their generalbudgets, placing a high degree of strain on theirfinances. This was especially true in light of theinadequate structure for service fees, which did notrecover operating and maintenance costs, muchless the costs of debt service. Further squeezingmunicipal budgets was the short maturity of

infrastructure bank loans—typically three years,sometimes five years.

In response, the central government took threeimportant steps designed to ready local govern-ments to further finance local infrastructure withina fiscally responsible framework. First, it announcedthat most local governments would be responsiblefor obtaining capital financing for infrastructureinvestment from the market. State onlending andgrants from the proceeds of state infrastructurebonds would be limited to the economically laggardwestern provinces—other local governments hadto be self-financing. Second, to support this self-sufficiency, the government decreed that municipal-ities should adopt full-cost tariffs for water supply,solid waste, and wastewater treatment, on a highlyaccelerated timetable. Full-cost pricing was definedto include all operating and maintenance costs plusdebt service and a competitive return on newlyinvested capital. Better-off municipalities were sup-posed to set tariffs to allow for full recovery of sys-temwide capital costs. This regime was designed toprovide adequate revenue streams to cover debtservice and attract private-sector capital into thelocal infrastructure sector, via either direct invest-ment or lending.

In a third—and in some respects mostinteresting—initiative, the government decreed thatlocal government should be restructured to separatethe ownership of infrastructure and other assets fromoperating responsibilities. This last initiative had twogoals. First, it was intended to yield more efficientmanagement of municipal assets, following so-calledNew Government trends established in Australia,New Zealand,and other countries, in which the asset-owning institution levies a capital charge on users toallocate costs more efficiently. In China’s case, how-ever, a more important motivation was to placeunder a single institutional umbrella assets used ascollateral for municipal loans.As part of governmen-tal restructuring, only the asset-holding institutions,known as Urban Development Investment Corpora-tions (UDICs),may now borrow on behalf of munic-ipal government. The restructuring was supposed toensure that municipal borrowing did not exceed thecollateral capacity of the municipality’s asset base.Moreover, as only UDICs are legally authorized toborrow, the restructuring was supposed to insulatethe general municipal budget from debt serviceclaims while providing a powerful incentive to


east_209-236.qxd 5/17/05 11:43 AM Page 224

UDICs to implement fees that would cover the cost ofcapital.

In some respects, China’s infrastructure financereforms are less sweeping than they may at firstappear. Although municipalities are prohibitedfrom direct borrowing, they continue to providecomfort letters to local UDICs. These in effect com-mit a municipal government to use general budgetrevenues or income from municipally owned prop-erty to help meet the UDIC’s debt service obliga-tions, should such support become necessary.Although UDICs tap the general corporate debtmarket, their debt instruments are a form ofmunicipal borrowing. The change in the institu-tional name of the borrower does not relieve localpublic institutions of the debt service burden cre-ated by short-term borrowing to finance long-terminfrastructure projects. From the banking sector’sperspective, the large amount of local assets held inthe form of loans to local governments represents acredit risk of unknown magnitude. No defaults onmunicipal borrowing from banks have beenreported. However, banks routinely roll over short-term loans as they come due. Questions remain asto whether banks will continue this policy as for-eign competition enters the banking sector underWTO rules, how much of their outstanding debtmunicipalities could actually pay under existingschedules, and how politically feasible and eco-nomically rewarding it would be for banks to fore-close on assets offered as collateral.

One finding of potential significance to otherEast Asian countries has emerged from UDICrestructuring, however, especially for periods ofstrong economic performance. Municipal govern-ments in China possess undeveloped or redevel-opable land with great market value, which couldfinance a substantial proportion of the local infra-structure investment burden. Changsha, the capitalof Hunan province in China’s interior, illustratesthis situation. The municipality holds title to some1.33 million hectares of land, valued at the munici-pality’s minimum long-term leasing price in 2001 atsome ¥ 105 billion. Of this total, about ¥ 85 billion(more than US$10 billion) corresponds to land notoccupied by the municipality itself that could beleased. Changsha officials estimate that some 60 per-cent of the gross price of leased land represents costsfor land that must be set aside as matching openspace under planning regulations, resettlement

costs, and revenues that must be shared with higherlevels of government. Still, with 40 percent of thegross value representing net profit, the net value ofChangsha’s land inventory is some ¥ 34 billion,which could finance a large part of the municipality’s10-year capital investment plan.

The potential for converting Chinese land valuesinto infrastructure assets is even greater if one con-siders that a large share of local infrastructureinvestment is capitalized into the value of munici-pally owned land, and that municipalities’ land-holdings are far from static. As population growthpushes the urban boundary outward, more andmore land reverts from collective rural ownershipto municipal ownership, providing a continuingbasis for capturing land value. Rural land at theedge of urban centers is indeed seriously under-priced, as the replacement cost of rural land isbased on its agricultural use without the premiumreflecting proximity to urban centers. This givesmunicipalities strong incentives to convert landfrom rural to municipal ownership. And munici-palities are in fact converting land values into infra-structure investment throughout China. In somemunicipalities studied under the City DevelopmentStrategies (Cities Alliance), land has financed asmuch as 70 percent of local infrastructure invest-ment, either directly through proceeds from leasesor indirectly by serving as collateral for infrastruc-ture loans.6 However, concerns are growing thatreliance on revenues from periurban land may leadto unsustainable urban planning and degrade thequality of urban life, calling for the development ofand integrated approach to land use and urbanplanning at the municipality level.

The combination of full-cost tariffs for watersupply, sanitation, and solid waste, coupled withincrements in land value created by road construc-tion and expansion of urban boundaries, providesa potentially sound basis for financing municipalinfrastructure.

Infrastructure Financing in the Philippines and Indonesia

The structure of lending to local governments forfinancing infrastructure has become an importantbottleneck to decentralization in both the Philip-pines and Indonesia. Despite plans to graduatecreditworthy municipalities and local utilities to the


east_209-236.qxd 5/17/05 11:43 AM Page 225

competitive credit market, the central governmentremains in control of credit channels in both coun-tries, acting as a monopoly intermediary betweenloans provided by international financial institu-tions and local governments. This position hasfrustrated development of sustainable sources ofdomestic financing while allowing central govern-ment institutions to restore—through loan condi-tions and discretionary loan approvals—some con-trol over the local infrastructure sector formallyrelinquished in the decentralization process.

The local credit market in the Philippines illus-trates the unequal playing field established by thegovernment. The Development Bank of the Philip-pines and the Land Bank of the Philippines obtainfinancing from international organizations andfrom the National Bank at below-market rates.Their onlending to local governments is securedby the authority to intercept intergovernmentalrevenue-sharing allotments—authority that is notavailable to private lenders. Reliable local loanrepayment to government financial institutionswas intended to introduce commercial banks tomunicipal lending as a creditworthy activity. How-ever, commercial banks cannot match the cost offunds of the government financial institutions, asthey are prohibited from serving as depositoryinstitutions for municipalities, which wouldstrengthen their ties to municipal budgets and pro-vide a lower-cost source of financing. As a result,commercial bank lending to municipalities forinfrastructure has yet to get off the ground, despitethe formal policy of promoting creditworthymunicipalities to the private credit market.

Indonesia illustrates the legacy power of badloans in thwarting development of a local creditmarket. As of March 31, 2000, 63 percent of theborrowing accounts of water utilities through sub-sidiary loan agreements and the Regional Develop-ment Account were reportedly in arrears (WorldBank 2003b). Although the country has launched aprogram of debt restructuring, it has made littleheadway in straightening out legacy borrowing.Resolution of the inherited debt runs straight to thefundamental issues raised by political decentraliza-tion. If central authorities made past investmentdecisions and mandated loan agreements, should adecentralized water utility be required to honorthat debt, and, if so, how will it recover the debt ifmunicipal governments are unwilling to impose

the required tariff increases? On the other hand,writing off these loans implies a substantial fiscalloss to the central government as well as an unde-sirable precedent for future onlending. These con-ditions seem to lay the groundwork for loanrestructuring, but reaching comprehensive agree-ment has proved difficult.

Lessons Learned

A self-sustaining local credit market is essential forsuccessfully decentralizing the infrastructure sec-tor. National government, as the original onlenderto municipal governments and utilities, needs tohave a strategy for developing a domestic localcredit market from the outset. This strategyrequires three components:

• Establishing a track record of timely debt repay-ment by local authorities. The injection of politi-cal considerations into debt repayment to gov-ernment financial institutions creates a creditrisk in local lending that can set back marketdevelopment for decades. The institutionalweight of bad loans deters entry by privatefinancial institutions into the subnational creditmarket.

• A policy that promotes replacement of governmentlending by private lending as quickly as feasible.This, in turn, requires a level playing fieldregarding revenue intercepts, depository func-tions, and other regulations. In other regions,internationally sponsored municipal develop-ment funds have proved successful by onlendingto municipalities through commercial banks,which fully accept the credit risk. This strategyintroduces commercial banks to municipallending while giving them access to longer-term,lower-cost funds than are available on thedomestic market. This approach requires a will-ingness on the part of government financialinstitutions to introduce decentralization to thefinancial sector by eliminating their monopolis-tic role as sole municipal lenders.

• A policy of substantial capital cost recoverythrough service tariffs. For services such as watersupply and wastewater removal, which cannotdirectly generate gains in land value, the onlyreliable recurring source of revenue is servicefees. One of the most useful standards that


east_209-236.qxd 5/17/05 11:43 AM Page 226

national government can set in the infrastructuresector is model tariff agreements between munic-ipal authorities and water utilities. Under these,the municipality agrees to sanction cost-recoverytariffs if the water utility meets performancetargets for service delivery and coverage.

How Far to Decentralize?

Decentralization requires unbundling the functionsassociated with providing infrastructure servicesand allocating them among different tiers of govern-ment. The optimal level of decentralization will varywith government’s policy goals and the types ofinfrastructure. This section discusses the mandate ofhigher tiers of government in the context of decen-tralized services.

The Role of Higher Tiers of Government

Even with aspects of infrastructure that are essen-tially local, an argument can be made for allocatingspecific functions to tiers of government higherthan the municipal level given one of the followingconditions:

Spillover effects. Interjurisdictional spillovers—orexternalities—arise when the activities of one juris-diction affect the welfare of people in surroundingjurisdictions. If municipal services produce spilloverbenefits or costs, service provision will be inefficient

without intervention by higher-level government, aslocal governments would ignore these impacts. Forexample, local decisions on regulating effluent dis-charges into rivers have implications for users inother jurisdictions that are part of the same catch-ment area. Similarly, spillover effects occur whenmunicipalities are responsible for managing feederroads whose benefits accrue to more than onejurisdiction.

To correct for such spillover effects, the tier ofgovernment whose jurisdiction encompasses all theusers benefiting from such services should decide oninvestment priorities and allocate resources.For example, managing environmental resourcesaccording to water basins is becoming more com-mon to correct for externalities in shared waterresources. The functions entrusted to water basinauthorities include managing and conserving water-sheds, controlling floods, reducing pollution, andlicensing water extraction. On the other hand, over-use and degradation of natural resources may occurwhen an intermediate tier of government does nottake responsibility for integrated water resourcemanagement (see box 10.2). Spillover effects simi-larly call for devolving management of secondaryroad networks to intermediate tiers of government.

Economies of scale. A municipality responsiblefor providing basic infrastructure services may besmaller than the minimum scale required to ensure


BOX 10.2 Vietnam: Watershed Management

In Dak Lak, Vietnam, groundwater is in highdemand to feed the expanding cultivation ofcoffee plantations. Groundwater resources in theEa Tul and Quang Phu catchments have so farbeen freely accessible, leading to overuse anddegradation of the natural resource base. Grow-ing competition in water use has led to conflictsbetween upstream and downstream users.

Local agencies have failed to mediate theseconflicts. For example, attempts by irrigationofficers to introduce irrigation calendars failedbecause communes could not consolidate theircropping calendars to fit the desired schedule.Under the new Water Law, Water Users Associa-tions have emerged to make decisions and coor-dinate water resources. At the provincial level,the Province People’s Committee established a

Participatory Irrigation Management SteeringCommittee to provide guidance to the WaterUsers Associations, along with supporting com-mittees at the district level. The water associa-tions, which cooperate with local line agencies,have encouraged farmers to view watershedproblems more holistically and mobilized newforms of collective action to address overuse anderosion.

The focus on participatory irrigation manage-ment is seen as a starting point for a more envi-ronmentally integrated approach to managingwater resources. However, even in the new insti-tutional landscape, no effective regional institu-tions regulate access to and use of groundwater.

Source: Dupar and Badenoch 2002.

east_209-236.qxd 5/17/05 11:43 AM Page 227

technical efficiency, especially for services that arelocal in nature but require large capital invest-ments, such as water supply, electricity distribu-tion, and public transport.7 When excessive frag-mentation of service provision is a concern,clustering municipalities to provide regionalservices can boost efficiency.8 However, regionalutilities require an institutional interface at a higherlevel of government in charge of setting investmentpriorities and regulating services. Spontaneouscoordination across municipalities is indeed diffi-cult to achieve and may be unsustainable when nohigher tiers play a coordinating role. As an example,in Caracas, 23 municipalities agreed to cooperate toaward a single water concession. However, theresulting agreement lacked credibility to investors,and the group received no responsive bids fromprivate operators (Triche et al. 1993).

Scarcity of human resources. A scarcity of specificskills may also make multiplying the number ofservice providers undesirable. In such a context,fewer larger entities may be in a better position toattract the minimum required skills than morenumerous, smaller service providers. A similar argu-ment can be made for limiting the number of regula-tory entities to enhance their capacity when humanresources are scarce (see, for example, Smith 2000).

Equity considerations. Fiscal decentralizationmay conflict with equity goals if the poorest regionshave limited leeway to mobilize financing and raiseown-source revenues to meet their infrastructureneeds, such as through local taxation, user fees, andaccess to capital markets. This may argue for limit-ing fiscal decentralization to allow higher tiers ofgovernment to redistribute resources to areas

lagging behind in economic development. In thiscontext, intergovernmental transfers are instru-mental in ensuring that all localities can afford toinvest in infrastructure. As an example, the govern-ment (or regulatory authority) may impose a levyon all firms operating in a market, and redistributethe revenues to companies connecting new users inpoor regions that cannot afford steep user charges.Higher tiers of government may also need to retainsome discretion in setting investment priorities toensure that local projects contribute to nationaland regional strategies for reducing poverty.

Distortion of interjurisdictional trade. Local regu-lation of basic infrastructure services may affectinterjurisdictional trade, adding transaction costsfor operators. For instance, local regulations gov-erning transportation safety may conflict and thuslimit or distort opportunities for trade. When localregulations impede trade across jurisdictions, thereis an economic argument for setting homogeneousquality standards throughout the area. As an ex-ample, in catchment areas cutting across severalmunicipalities, water basin authorities may play arole in harmonizing environmental standards andregulating inland waterways. Higher tiers of gov-ernment could similarly be entrusted with respon-sibility for setting quality standards for secondaryroad networks (see box 10.3).

Destructive competition. Decentralization mayincrease efficiency by promoting competitionamong local governments. However, devolution ofdecision-making powers to the lowest tiers of gov-ernment may turn the potential for competition intoa “race to the bottom,” where competition amonglocal governments to attract foreign investment in


BOX 10.3 Indonesia: When Transport Regulations Distort Trade

One area where transport regulations may dis-criminate against outsiders is the introduction oflicenses for use of roads within a certain region.The Izin Trayek rule in South Sulawesi, Indonesia,for example, requires that all transport truckscarry one of three specific licenses: for inter-provincial transport, for intraprovincial trans-port, and for entering the regencies (Keca-matan). The first two licenses are issued at theprovincial level in accordance with gubernatorial

decree (Keputusan) No. 10 1996, while localgovernments issue the third type of license.Trucks not carrying licenses are typically finedRp 35,000. This regulation clearly discriminatesagainst trucks from other areas, particularlybecause licenses are not available outside SouthSulawesi.

Source: Goodpaster and Ray 2000.

east_209-236.qxd 5/17/05 11:43 AM Page 228

infrastructure can prompt municipalities to biddown taxes or other regulatory obligations (or bidup subsidies or regulated rates of return).9 Excessivecompetition may induce inefficient allocation ofresources and overinvestment, with municipalitiesbuilding or upgrading ports or other infrastructurefacilities in their own areas to enhance their prestige,rather than relying on facilities in adjacent regions.

Efficiency of revenue collection. The scope fordecentralizing financial powers may be limitedwhen the central level can collect budgetary rev-enues more efficiently, and when there is littleopportunity for collecting cost-covering charges atthe point of service. This is often the case in the roadsector, for example, where financing comes largelyfrom fuel taxes and vehicle operating fees, which aremore efficiently collected by higher tiers of govern-ment. Higher levels need to redistribute those rev-enues to lower tiers where services are provided. Anexample is dedicated road maintenance fundsfinanced by user charges collected at the nationallevel (mainly through fuel taxes). Sophisticated cost-sharing formulas can allocate these funds among dif-ferent road networks (and corresponding levels ofgovernment), and robust accountability mecha-nisms can oversee use of the funds.

Interjurisdictional Coordination in China

In China, sustained economic growth is spurring afew major cities to develop into metropolitan areasthat cut across more than one jurisdiction. Theseareas include the Pearl River Delta region (centeredaround Guangzhou, Shenzhen, and Hong Kong),the Lower Yangzi Delta region (centered aroundShanghai), and the Beijing-Tianjin region. Coordi-nated development has started to emerge, asmunicipalities have begun to see the benefits ofregional integration. For example, a pilot exercise inthe Pearl River Delta region, led by the Ministry ofConstruction, aims to establish a metropolitanplanning model to be replicated in other regions.An even more ambitious project is the plan todevelop a Pan-Pearl River Delta Regional Coopera-tion and Development Area, which would encom-pass almost one-fourth of China’s territory, includ-ing nine provinces, Hong Kong, and Macau. Themain goal of this regional initiative is to facili-tate the management of highway and railway proj-ects, which are expected to generate significant

externalities. The trend toward regional integrationis therefore an important step toward more effi-cient infrastructure service provision.

On the other hand, examples of interjurisdic-tional management of shared water resources,where spillover effects also call for the involvementof higher tiers of government, are still rare in China.The main exception is the recent attempt to pro-mote shared environmental infrastructure in thePearl River Delta (PRD), one of the most complexurban systems in Asia. Many sections of the PRDhave extremely poor water quality. The municipali-ties of the Guangdong province are the highest con-tributors to PRD pollution, and the provincialgovernment—through its Environmental Protec-tion Bureau—has recently announced an eight-year,US$5 billion program to invest in wastewater treat-ment facilities. The plan is based on the recognitionthat investment in environmental infrastructureshould be guided by a regional development strat-egy that reflects sound environmental managementand fiscal sustainability. A pilot project will promotedevelopment of environmental infrastructure forthree groups of two or more municipalities, dis-tricts, and towns. A key parallel activity is the PRDCleanup Campaign, which has set phased targetsfor meeting water quality standards. One of thegoals of the campaign is to enhance intermunicipalcollaboration.

In China, more effective regional coordinationin setting investment priorities and allocatingresources is also needed to help avoid excessivecompetition among municipalities in the provisionof infrastructure services. Amid economic transi-tion and decentralization, local investment policiesare indeed driven primarily by a growth and com-petitiveness agenda. As a result, municipalities tendto compete excessively to attract outside invest-ment in businesses and infrastructure projects.Their competitive tools are mainly preferentialpolicies such as tax holidays, free land, and dis-counted land concessions. In this context, lackingcoordination at higher tiers of government, such adhoc policies may unduly distort resource allocationbetween municipalities, as well as between stake-holders within a municipality.

The Missing Middle: The Case of Indonesia

In Indonesia, Law 22 of 1999 accords provincestwo roles: as deconcentrated representatives of the


east_209-236.qxd 5/17/05 11:44 AM Page 229

center, and as autonomous regions. While provincescan officially coordinate regional policies andperform joint tasks on behalf of local governments,the legislative framework provides no hierarchicalrelationship between provinces and local govern-ments. This has jeopardized the ability of provincesto facilitate cooperation among local governmentsand establish their authority in regional functions.As a result, sectors with large externalities and sig-nificant economies of scale, such as watershed man-agement, have consistently underperformed. More-over, decentralization has resulted in a multiplicityof standards at the municipal level, which may dis-tort trade across jurisdictions (see box 10.3).

The resulting efficiency losses from the “missingmiddle” are compounded by the small size of somelocal entities, which suggests diseconomies of scaleand points to consolidation of regions.10 Thenarrow administrative boundaries of local govern-ments, combined with the limited role of pro-vinces, have led to suboptimal investment decisionsfrom a regional and national perspective. The argu-ment for strengthening the role of provinces inmanaging road networks, whose benefits accrue tomore than one local jurisdiction, is especially com-pelling. Secondary road networks have been largelyunderfunded compared with the need (the countryconfronted an estimated 15–20 percent fundingshortfall in 2000) (World Bank 2004). In this con-text, giving provinces greater financial authoritycan broaden their influence on local governmentsand thus increase investment efficiency. Oneapproach is to link provincial and national roadfunds, and assign the provincial government strate-gic oversight of all roads (provincial and kapub-aten) in the province, as well as responsibilityfor assessing investment needs and allocatingresources. The provincial road funds would financemaintenance and rehabilitation of local networks,provided that the kapubaten adopt sound roadmanagement practices.

In Indonesia, the central government retainssignificant control in allocating resources betweenjurisdictions. Limited decentralization of revenue-raising powers can help reduce regional inequality,given redistributive mechanisms. The intergovern-mental transfer system includes two equalizationgrants (the DAK and the DAU, or dana alokasiumum) to fund investment in infrastructure, espe-cially services that generate externalities. However,

in practice, these grants have tended to exacerbaterather than reduce regional inequality.11

The Philippines: Another Example of the Missing Middle

In the Philippines, the national government isresponsible for providing primary infrastructure,including backbone transmission grids in the powersector and primary road networks, while cities andmunicipalities are responsible for tertiary infrastruc-ture such as roads and water. However, it is unclearwhich tier of government is responsible for planninginvestment and coordinating development of sec-ondary networks that serve more than one local gov-ernment and involve common resources such asriver basins.While in principle provinces play a coor-dinating role among cities and municipalities, theylack the technical and financial resources to do so.

As a result, no intermediate tier is capable ofmanaging water resources shared by several localgovernments. Without a regional body to coordi-nate investment, local governments often argueover river basin planning and management, alloca-tion of water rights, and pollution control, andwater-stressed municipalities have had troublenegotiating water rights outside administrativeboundaries. For example, Cebu City reportedly hadgreat difficulty convincing Bohol province to sup-ply the water-starved city even though the marginalvalue of water consumption in the city was high.

A similar situation arises in managing secondaryroad networks. While road density in the Philippinesis among the highest in the region, only 20 percent ofthe road network is paved. Provincial roads accountfor only 13 percent of total roads, with more than 70percent consisting of city, municipal, or barangayroads. Moreover, in 2000, only 21 percent of provin-cial roads were paved—a rate much lower than thatfor national roads (62 percent), city roads (77 per-cent) and even municipal roads (34 percent)(Department of Public Work and Highways 2003).Under the institutional framework, volunteer coop-eration among local governments is the only mecha-nism for coordinating management of the secondaryroad network (see box 10.4).

Provinces also suffer from an acute lack of finan-cial resources. This is due largely to the fact thatdecentralization has significantly shifted own-sourcerevenue from provinces to cities and municipalities.


east_209-236.qxd 5/17/05 11:44 AM Page 230

Not only are the taxing powers of provincial govern-ments inferior to those of city and municipal govern-ments, but cities do not have to share their tax rev-enues with provinces. As a result, provinces dependmostly on the Internal Revenue Allotment (IRA)—amechanism for transferring funds from the center.

Regional Development Councils (RDCs) couldplay a role in integrating regional and local infra-structure plans with the country’s overall infra-structure plans.12 However, the majority of RDCsare weak and ineffective in planning and coordinat-ing infrastructure projects. The perception is thatRDCs merely act as endorsers of projects initiatedby regional offices of line ministries or local gov-ernments, which require international funding or aguarantee from the national government. Thus,rather than playing a coordinating role, RDCs areseen as more concerned with monitoring nationalprojects implemented at the local level (Llanto andLasam 2003).

In the Philippines, as in Indonesia, the IRA is sup-posed to correct the mismatch between revenue andexpenditure assignments across different levels ofgovernment. Although local governments have be-come more and more dependent on the IRA, thetransfer system has not contributed to increaseequity. On the contrary, the IRA formula favors big-ger and richer local governments at the expense ofpoorer and smaller ones, as it is based on land areaand population. Hence, cities and municipalities thatare more populous and have larger land areas enjoy astrong advantage, and richer local governments withlarger tax bases receive a bigger share of the IRA.

Lessons Learned

This section has discussed the role of higher tiersof government in decentralized infrastructure by

drawing on the experiences of China, Indonesia,and the Philippines. The need to strengthen therole of higher levels of government in providinginfrastructure services appears particularly com-pelling in Indonesia and the Philippines, wherethere is clear evidence of a missing middle in thearchitecture of decentralization. The experiences ofall three countries suggest the following lessons onhow far to decentralize infrastructure services:

Partnership between national, provincial, andmunicipal governments is crucial to maximizing theefficiency gains from decentralization. Infrastructureservices entail a broad set of functions. The extentto which higher levels of government performsome of these functions often depends on the char-acteristics of a particular industry. For example,when the main economic argument for involvingprovincial authorities is the presence of spillovereffects, higher levels of government need to retain aplanning and coordinating role, while municipali-ties may be better positioned to build and operatefacilities. When the main concern is excessive frag-mentation, decentralization of service provisiononly to the regional level can retain economies ofscale. Clearly defining responsibilities and provid-ing mechanisms for coordinating all tiers of gov-ernment are essential to maximizing the benefits ofdecentralization.

A progressive approach to decentralization hasmerit. Developing countries may need to build orstrengthen institutions at intermediate and lowerlevels of government to coordinate responsibilities—institutions that are often the norm in mature infra-structure industries. This implies that countries mayneed to phase in decentralization while buildingcapacity. Another argument in favor of a progressiveapproach is that once functions are decentralized to


BOX 10.4 The Philippines: Toll Road Management

The construction of a circumferential road acrossCabanatuan City and adjacent municipalities is agood example of cooperation in planning invest-ment and implementing a project. CabanatuanCity signed a memorandum of agreement withthe municipalities of San Leonardo and Sta. Rosathat defined their contributions and obligationsto this project. The toll road, financed by contri-butions from the local governments, is expected

to raise revenues once it is operating. The coop-eration reflected strong leadership from the chiefexecutives of the three local governments andtheir understanding of the benefits of a jointapproach to combating rising urban congestion.

Source: Gilbert Llanto, mayor of CabanatuanCity, field interview.

east_209-236.qxd 5/17/05 11:44 AM 1

the lowest tiers, creating a role for higher tiers of gov-ernment can be very difficult, as municipalities maybe reluctant to relinquish decision-making andrevenue-raising powers. However, political consider-ations often play a critical role in designing a decen-tralization strategy. For example, while a more grad-ual phasing-in may have been warranted in thePhilippines and Indonesia, political imperativescalled for a Big Bang approach, under which localgovernments assumed responsibility for providingbasic infrastructure services almost overnight.

Given economies of scale, decentralizing infra-structure to the lowest tiers of government may leadto excessive fragmentation. The risk of excessivefragmentation is particularly high when decentral-ization is not conceived as a response to specificproblems but rather as a byproduct of widerreform. The result could be an industry structurethat is far from optimal from an economic pointof view.

Careful design of intergovernmental transfermechanisms is needed to meet equity objectives. Whilecentral intervention may be warranted to redressregional inequalities, experiences in Indonesia andthe Philippines show that limiting fiscal decentral-ization has not produced the expected results interms of income redistribution. On the contrary,intergovernmental transfers have exacerbatedregional difference in income, jeopardizing the abil-ity of the poorest regions to finance their infrastruc-ture needs. Countries need to improve the efficiencyof their equalization mechanisms to address fiscalimbalances across regions.

Key Issues for Policy Makers

Several key issues stand out from this review of EastAsia’s experience in decentralizing infrastructure.Perhaps the most important is the coherence andalignment of administrative, financing, perfor-mance measurement, and incentive policies andprograms. Where alignment exists, anticipatedresults will be forthcoming, as in the case of theprincipal-agent arrangements of China andVietnam. Of course, this can be a doubled-edgedbenefit. If the goals make sense, performance willyield desirable outcomes. On the other hand, doingthe wrong thing well is also a possibility. Thus, it isas important to ensure that coherence and align-ment favor appropriate local decision making,

control of results, and accountability for results.This lesson is elaborated in detail below.

Performance measurement opens the door to effi-ciency gains from decentralized infrastructurebecause it permits meaningful accountability. At thelocal level, standardized cost comparisons, such ascost per kilometer of road construction, can imme-diately translate into savings when used as aguide for competitive procurement. Performancecontracting—in which a municipality commits toauthorizing tariff increases if a utility meets well-defined performance goals—can upgrade perfor-mance while breaking the deadlock over settingtariffs high enough to recover service costs, whichhas handicapped local investment. NGOs and citi-zens gain the power of accountable oversight only ifthey can measure performance against quantifiedtargets.

The power of upward accountability is evidentin the case of China, where measurable perfor-mance against state-determined investment targetshas driven the infrastructure sector. On the otherhand, in politically decentralized systems, perfor-mance measures are almost totally lacking in theinfrastructure sector. This is in striking contrast tothe health and education sectors, where client sur-veys and output measures are far more common.Simple, transparently reported measures of infra-structure performance tracked locally and used forlocal management would enable countries to go farin realizing the potential of decentralization.

Community participation at the project level iscritical. A large body of evidence supports theimportance of community-scale participation ininfrastructure choices and management. Commu-nity water supply and latrine projects have provenmore sustainable—longer lasting, more fully used,and more financially self-sufficient—when designedin partnership with the community and managedby community organizations. The participation of across-section of users in management has beenfound to be particularly important. Women and thepoor are most likely to be excluded from manage-ment, undermining project sustainability.

The economic literature on decentralized fiscalchoice has emphasized the gains from allowingclients to choose among an array of projectoptions. On the other hand, some East Asian coun-tries offer communities a much smaller range ofchoices. Rural water projects in China, for example,


east_209-236.qxd 5/17/05 11:44 AM Page 232

give communities the opportunity to sign on tostandardized arrangements. Community consulta-tion consists of explaining to residents the servicesthey will receive and the tariff costs they will beresponsible for, if they participate. Such projectshave high sustainability in China despite the closednature of the initial choice.

Scaling up community participation in municipal-level capital planning is difficult. Decentralization inthe infrastructure sector has been premised on thevalue of community choice. Both Indonesia and thePhilippines have attempted to incorporate commu-nity participation in higher-level choices by includ-ing NGOs in efforts to set municipal investment pri-orities. The results of this experiment have beenmixed, at best. The presumption that NGOs repre-sent community consensus rather than advocateparticular points of view has often broken down inpractice. Urban forums open to all have proved valu-able in stimulating public debate over investmentpriorities, but a difficult vehicle for actually estab-lishing capital budgets. Whether direct communityparticipation in investment decision making over-comes skepticism about local representative democ-racy remains to be seen. Substantial differences inthe information available to stakeholders remain asignificant problem, as such asymmetry contributesto the greater influence of elites in decision making.

Efficiency in capital investment is important.Paradoxically, at a time of great pressure on infra-structure investment budgets, large portions ofexisting capital remain unused. In China, waste-water treatment plants in some provinces are notfunctioning for lack of operating and maintenancefunds while expensive new treatment plants arebeing built. In Indonesia, some water utilities havemuch excess capacity because they failed to takeinto account ample free water from householdwells. In politically decentralized systems, commu-nity participation in initial project design andrequired local cost sharing can reduce excess capac-ity. In upwardly accountable systems, performancetargets need to measure relevant outputs ratherthan merely the capacity of capital facilities.

A well-functioning local credit market is anessential ingredient of a decentralized infrastructuresector. The supply of credit for local investmenthas become a bottleneck to decentralizing infra-structure. In both Indonesia and the Philippines,central authorities retain control over the channels

for local credit, frustrating development of a self-sustaining domestic credit market that meetsmunicipal needs for financing infrastructure.Government financial institutions can encouragethe emergence of a municipal credit market byremoving regulations that give them preferenceover commercial banks and other lenders. Onlend-ing of international funds to municipalities viacommercial banks, rather than government finan-cial monopolies, would likely speed developmentof such a market.

Higher tiers of government play a critical role inthe architecture of decentralization. A partnershipbetween national, provincial, and municipal gov-ernments is crucial to maximizing the benefits ofdecentralization, even in infrastructure industriesthat are essentially local. For example, empower-ing provincial governments to perform a planningrole is essential to correct for interjurisdictionalspillovers, while limiting fiscal decentralizationmay be warranted on equity grounds to allow forcross-subsidies between geographic groups. Theneed to strengthen the role of intermediate tiersof government is compelling in Indonesia andthe Philippines, where the missing middle hasresulted in poor coordination between jurisdic-tions. The consequences are particularly evidentin the transport sector, where secondary roadnetworks have suffered from severe maintenancebacklogs as a result of poor interjurisdictionalcoordination.

Annex: Comparing Water Coverage in Different Countriesand Regions

Truly comparable data across countries on infra-structure coverage, investment levels, and tariffs aredifficult to produce and generally not available.Global Water Supply and Sanitation Assessment2000, prepared by the World Health Organizationand the United Nations Children’s Fund, providesprobably the most standardized reporting, but eventhe data in that volume are imperfect. Coveragerates reflect access to “improved” water and waste-water systems. Tariff and investment ratios arereported as averages for the decade 1990–2000 andtherefore do not take into account recent changes.Nonetheless, the data do provide a general compar-ative baseline for East Asian countries:


east_209-236.qxd 5/17/05 11:44 AM Page 233

Through the Millennium Development Goals, allthe East Asian countries have set far more ambitioustargets for coverage during the period 2000–5.Because of decentralization, local governments willbe the primary instruments for implementing andfinancing this accelerated coverage.

Investing in the Water Supply and Sanitation Sector

During the decade 1990–2000, Asia lagged behindother developing regions in the share of govern-

mental investment devoted to water supply andsanitation. This probably implies, as Global WaterSupply and Sanitation Assessment 2000 concludes,that Asian countries gave the water and sanitationsector lower priority, but the situation also reflectsthe higher shares of public sector budgets devotedto investment in Asia.

Investment shares in East Asia in this sector haveclimbed recently as countries have focused on meet-ing their coverage targets. China has also set ambi-tious targets for treating wastewater before discharge.


TABLE 10A.1 Water Supply Coverage Rates (percent of population covered)

Country Year Urban coverage Rural coverage Total

Cambodia 1990 — — —2000 53% 25% 30%

China 1990 99% 60% 71%2000 94% 66% 75%

Indonesia 1990 90% 60% 76%2000 91% 65% 87%

Philippines 1990 94% 81% 87%2000 92% 80% 87%

Vietnam 1990 81% 40% 48%2000 81% 50% 56%

Source: WHO and UNICEF 2000.Note: Rapid rates of urban population growth mean that even where urban coverage rates declined, largenumbers of households gained access to a municipal water supply. (—) � not available.

TABLE 10A.2 Sanitation Coverage Rates (percentage of population covered)

Country Year Urban coverage Rural coverage Total

Cambodia 1990 — — —2000 58% 10% 18%

China 1990 57% 18% 29%2000 68% 24% 38%

Indonesia 1990 76% 44% 54%2000 87% 52% 66%

Philippines 1990 85% 64% 74%2000 92% 71% 83%

Vietnam 1990 — — —2000 86% 70% 73%

Source: WHO and UNICEF 2000.Note: Given the high differentials in coverage between urban and rural areas, one of the most statisticallysignificant ways of expanding national coverage is through rural to urban migration and other sources ofurban population growth. (—) � not available.

east_209-236.qxd 5/17/05 11:44 AM Page 234

Tariff Rates

Median urban tariffs for water and sewerage from1990–2000 were lower in Asia than in otherregions, although Asia’s lower rates partly reflectlower production costs. Asia faces a particularlychallenging task in raising tariffs to commerciallevels.

Endnotes

1. As of 2001, only 200 of China’s 667 cities treated any waste-water before discharge (Murray 2003).

2. Local officials are evaluated based on their contribution toeconomic growth, which is often interpreted as reachinginvestment targets.

3. The city of Naga in the Philippines makes similar use ofcomparative disclosure at the local level. The city publishesall per unit costs from different bidders for local construc-tion contracts on its website.

4. In the case of the Kecamatan Development Program,30,000 villagers were hired and trained in project develop-ment and 2,000 community facilitators were also hired andtrained.

5. This discussion follows Usui and Alisjahbana 2003.6. A city development strategy is an action plan for equitable

growth in cities and their surrounding regions, developedand sustained through participation, to improve the qualityof life for all citizens. See Chreod Ltd. 2002.

7. For example, Tynan and Kingdom 2004 proved economet-rically that smaller water utilities, particularly those serving

a population of 125,000 or less, could reduce per customeroperating costs by increasing their scale of operation.

8. However, there is often a trade-off between efficiency gainsand loss of local accountability, which efforts to identify theoptimal area of service provision would need to take intoaccount.

9. This argument is made, for example, in Smith 2000.10. Local governments range in population from 24,000 to 4.1

million. The per capita wage bill of local governmentsseems to suggest that efficiency falls sharply at the level ofabout 500,000 people.

11. In Indonesia, revenue disparities are significant at all levelsof government. These inequalities are most extreme at thelocal level, where the richest region accounts for 46 timesthe revenues of the poorest region. The richest province alsohas 32 times the per capita revenues of the poorest province.

12. There are nine RDCs, including those in the CordilleraAdministrative Region and the Autonomous Region ofMuslim Mindanao.

Bibliography

Alonzo, Ruperto P. 1998. “Local Governance and Poverty Allevi-ation.” In Growth, Poverty, and Income Inequality in thePhilippines, eds. A.M. Balisacan and S. Fujisaki. Tokyo: Insti-tute of Developing Economies.

Azfar, Omar. 2002. “Decentralization in Indonesia: A Reportfrom a Household Survey in West Java.” Paper presentedat the Regional University Consortium Conference onRegional Economic Development in a DecentralizingIndonesia.

Azfar, Omar, Tugrul Gurgur, Satu Kahkonen, Anthony Lanyi,and Patrick Meagher. 2000. Decentralization and Gover-nance: An Empirical Investigation of Public Service Delivery inthe Philippines. College Park, MD: University of Maryland,Department of Economics, Center for Institutional Reformand the Informal Sector.

Bellier, Michel, and Yue Maggie Zhou. 2003. “Private Participa-tion in Infrastructure in China.” Working Paper No. 2.Washington, DC: World Bank.

Chanthaphone, Soutsakhone, and Santanu Lahiri. 2003. “Look-ing Backward to See Forward.” In Sustainability Planningand Monitoring in Community Water Supply and Sanitation,eds. Mukherjee and van Wijk Washington, DC: World Bank.

Chreod Ltd., with the Urban Institute. 2002. China City Develop-ment Strategies: Changsha, Zhuzhou, Xiangtan. Ottawa:Chreod Ltd.

———. 2003. China City Development Strategies II: InceptionMissions—Lanzhou, Chengdu, Zhengzhou, Xinxiang, Erdos.Ottawa.

Department of Public Work and Highways, Philippines. 2003.“Roads in the Philippines.” Manila: Department of PublicWork and Highways.

Dupar, Mairi, and Nathan Badenoch. 2002. “Environment,Livelihoods, and Local Institutions: Decentralization inMainland Southeast Asia.” Washington, DC: WorldResources Institute.

Fontanelle, Jean-Philippe. 2000. “Water Management Decentral-ization in the Red River Delta, Vietnam: An UncompletedTransition Process toward Local Governance.” Paper pre-sented to the eighth biennial conference of the InternationalAssociation for the Study of Common Property, Blooming-ton, Indiana.

Fontanelle, Jean-Philippe, and François Molle. 2002. “The Intri-cacies of Water Pricing in the Red River Delta, Vietnam.”


TABLE 10A.3 Median Investment in WaterSupply and Sanitation, 1990–2000(percentage of overall government investment)

Region Percentage

Africa 5.3%

Asia 3.5%

Latin America and Caribbean 8.3%

Source: WHO and UNICEF 2000.

TABLE 10A.4 Median Urban Tariff Rate (US$ per cubic meter)

Region Water Sewerage

Africa 0.35 0.12

Asia 0.22 0.14

Europe 0.67 0.59

Latin America 0.44 0.21

North America 0.48 0.41

Source: WHO and UNICEF 2000.

east_209-236.qxd 5/17/05 11:44 AM Page 235

Paper presented to International Conference on IrrigationWater Policies, Agadir, Morocco.

Goodpaster, Gary, and David Ray. 2000. “Trade and CitizenshipBarriers and Decentralization.” Indonesian Quarterly 3.

Government of China. 2000. City Dirty Water Treatment andPollution Prevention and Treatment Policy. Beijing: Ministryof Science and Technology, Ministry of Construction, StateEnvironment Protection Administration.

Gross, Bruce. 2003. “Do Project Rules Promote Sustainabilityand Equity?” In Sustainability Planning and Monitoring inCommunity Water Supply and Sanitation, eds. Mukherjeeand van Wijk. Washington, DC: World Bank, InternationalWater and Sanitation Centre.

Gross, Bruce, Christine van Wijk, and Nilanjana Mukherjee.2001. “Linking Sustainability with Demand, Gender, andPoverty: A Study in Community-Managed Water SupplyProjects in 15 Countries.” Delft, Netherlands: Water andSanitation Program and IRC International Water and Sani-tation Centre.

Hofman, Bert, and Kai Kaiser. 2002. “The Making of the BigBang and Its Aftermath: A Political Economy Perspective.”Paper presented at conference on Can Decentralization HelpRebuild Indonesia?, Atlanta, GA.

Hopkins, Richard. 2003. “How Well Did Those DevelopmentProjects in Flores Work?” In Sustainability Planning andMonitoring in Community Water Supply and Sanitation, eds.Mukherjee and van Wijk. Washington, DC: World Bank,International Water and Sanitation Centre.

Indonesian Partnership on Local Governance Initiatives. 2002.Various papers for the International Workshop on Participa-tory Planning Approaches for Local Governance, Bandung,Indonesia.

Llanto, Gilberto M., and Joel Lasam. 2003. “Report on theForum on Planning, Programming, and Monitoring.” Sub-mitted to National Economic and Development Agency andDeutsche Gesellschaft für Technische Zusammenarbeit.

Loehr, William, and Rosario Manasan. 1999. “Fiscal Decentral-ization and Economic Efficiency: Measurement and Evalua-tion.” Cambridge, MA: Harvard Institute for InternationalDevelopment.

Meadley, Thomas. 2003. “Community Design of Hygiene Pro-motion IEC Materials.” Presentation to World Bank WaterWeek. World Bank, Washington, DC.

Murray, Bruce. 2003. Presentation to Workshop on NationalGuidelines for Urban Wastewater Tariffs and Management.Manila: Asian Development Bank.

Orial, Lydia N. 2002. “Local Government Finance and LGUBond Market Development: Philippines.” Manila: AsianDevelopment Bank.

Pratikno. 2002. “Initiating Citizen Participation in Local PolicyMaking in Indonesia: The Case of Surakata Municipality inthe Post Authoritarian Regime, 1999–2002.” Gadjah MadaUniversity.

Royal Government of Cambodia. 2003a. National PovertyReduction Strategy 2003–2005. Phnom Penh: Council forSocial Development.

———. 2003b. Water and Sanitation Resources: Sector Overviewand Market Structure. Phnom Penh: Ministry of Commerce.

Smith, Warrick. 2000. “Regulating Utilities: Thinking aboutLocation Questions.” World Bank Summer Workshop onMarket Institutions, Washington, DC.

Triche, Thelma, Abel Mejia, and Emanueal Idelovitch. 1993.“Arranging Concessions for Water Supply and SewerageServices: Lessons from Buenos Aires and Caracas.” WS-10.Washington, DC: World Bank, Transportation, Water, andUrban Development Department.

Tynan, Nicola, and Bill Kingdom. 2004. “Optimal Size forUtilities? Returns to Scale in Water—Evidence from Bench-marking.” Viewpoint note. rru.worldbank.org/documents/publicpolicyjournal/283tynan.pdf.

Urban Institute. 2003. “Local Government Water ServicesProject Final Report.” Washington, DC: Urban Institute.

Usui, Norio, and Armida Alisjahbana. 2003.“Local DevelopmentPlanning and Budgeting in Indonesia: Update.” Prepared forthe international symposium on Indonesia’s Decentraliza-tion Policy: Problems and Policy Directions. Jakarta.

World Bank. 2002a. “Selling Sanitation in Vietnam: WhatWorks?” Vietnam Country Report, WSP-EAP. Washington,DC: World Bank, Water and Sanitation Programme.

———. 2002b. “Benchmarking the Urban Water SectorVietnam.” Hanoi.

———. 2003a.“China Country Assistance Strategy.” Washington,DC: World Bank.

———. 2003b. “Cities in Transition: Urban Sector Review in anEra of Decentralization in Indonesia.” Working Paper No. 7.Washington, DC: World Bank.

———. 2003c. “Decentralizing Indonesia: Regional PublicExpenditure Review.” Report 26191-IND. Washington, DC:World Bank.

———. 2003d. “Indonesia: Selected Fiscal Issues in a New Era.”Report 25437-IND. Washington, DC: World Bank.

———. 2003e. “Project Appraisal Document for a ThirdKecamatan Development Project.” Washington, DC: WorldBank.

———. 2004. “Indonesia: Averting an Infrastructure Crisis—AFramework for Policy and Action.” Washington, DC: WorldBank, Recent Economic Development in Infrastructure.

World Bank and Asian Development Bank. 2003. “Philippines:Decentralization and Service Delivery—From Promise toPerformance.” Manila: World Bank and Asian DevelopmentBank.

WHO (World Health Organization) and UNICEF (UnitedNations Children’s Fund). 2000. Global Water Supply andSanitation Assessment 2000. www.wssinfo.org/pdf/GWSSA_2000_report.pdf.

Zhixiong, Luo. 2003. “Financing Arrangements for Rural WaterSupply and Sanitation Project in China: Experiencesfrom Guizhou Province.” Presentation to World Bank,Washington, DC.


east_209-236.qxd 5/17/05 11:44 AM Page 236

decentralizing basic infrastructure services

Documents