__ The World Bank '' I '>js ^' J ~~~~~~~~~~~~~~~~~~~~U N E - -~~~~~ @ I.- 2001 I w < - - ;- nz s tn c NUMBER 55 - I ~no _ I etI Decentralization and govemance: does decentralization improve public service delivery? Decentralization holds great promise for improving the delivery of public ser- vices-but outcomes depend on its design and on the institutional arrange- ments governing its implementation. The Philippines and Since the 1980s many developing countries * Voice mechanisms for citizens to express Uganda show the have started to devolve central government their views to government bodies. functions to local jurisdictions. This shift * Exitmechanismsforcitizens to switch to conditions that must has occurred partly in response to politi- nonpublic service providers or to move cal pressures for local autonomy, but also to other localities. be met for to deliver public services more efficiently. * Central government laws, rules, budget It has been argued that decentralization constraints, and oversight over local gov- decentralization to improves governance and public service ernment operations, and channels for delivery by increasing: local governments to influence central improve public * Allocative efficienc-through better match- government decisions concerning them. ing of public services to local preferences. * Public sector management arrange- services * Productive efficiency-through increased ments that promote accountability- accountability of local governments to such as merit-based personnel policies citizens, fewer levels of bureaucracy, and and rules and arrangements promoting better knowledge of local costs. fiduciary accountability and constrain- But decentralization has had mixed ing corruption. effects on public service delivery. Limited A recent study by the University of Mary- evidence suggests that the effectiveness of land analyzed the validity of this argument decentralized service delivery depends on by assessing decentralized health and edu- the design of decentralization and on the cation services in the Philippines and institutional arrangements governing its Uganda. Using surveys of government offi- implementation. Specifically, the argument cials, households, health facilities, and that decentralization promotes allocative schools, the study explored whether local and productive efficiency assumes that the governments are aware of local preferences devolution of functions occurs within an and whether they adjust public service deliv- institutional environment that provides ery accordingly. The study also tested the political, administrative, and financial extent to which local governments contain authority to local governments, along with corruption, costs, and other waste. Thus the effective channels of local accountability study analyzed whether allocative and pro- and central oversight. These channels ductive efficiency are achieved, as stated by include: the theory. In addition, the study assessed FROM THE DEVELOPMENT ECONOMICS VICE PRESIDENCY AND POVERTY REDUCTION AND ECONOMIC MANAGEMENT NETWORK whether channels of local accountability and Local officials perceive local demands but oversight are effective in these countries, have limited authority to adjust services and the importance of different channels Decentralization may increase allocative effi- for decentralized service delivery. ciency if local governments have the authority and willingness to adjust resource Public service delivery has allocations and functions are devolved to been partly decentralized a low enough level. In both the Philippines In the Philippines and Uganda the devolu- and Uganda, officials of lower-tier govern- tion of responsibilities for public service deliv- ments (municipal and subcounty, respec- ery has been a complex, ongoing process tively) were more aware of local preferences involving retention of some powers by the cen- than officials of higher-tier governments tral government. Despite the election of local (province and district). The benefits of officials, both countries remain unitary states. For example, the top household priority In the Philippines the 1991 Local Gov- for incremental funds was water supply in decentralized ernment Code has increased the responsi- Uganda, roads in the Philippines. While both bilities of all subnational governments: 77 subcounty and district officials were aware of publIc services provinces, 69 cities, 1,538 municipalities, and these preferences in Uganda-subcounty offi- more than 40,000 communities. Most power cials more so-only municipal officials were do not always is vested in provinces, cities, and municipal- aware of the demand in the Philippines. ities. Most health spending has been devolved Philippine provincial officials knew little about materialize to subnational governments, though the cen- local demand. In both countries, however, tral government retains responsibility for resource allocations reflected local prefer- public goods like immunization and com- ences only weakly. This is partly due to the municable disease control. In education, the limited authority of local governments to central government is still responsible for adjust allocations. budgeting and hiring, but local governments are responsible for operating and main- Corruption is less pronounced at local taining schools and are often involved in hir- than at national levels ing teachers. Decentralization may increase productive In Uganda the 1995 Constitution and the efficiency by limiting the leakage of funds 1997 Local Government Act specify five lev- and other resources. In both countries cor- els of local government-district, county, sub- ruption was seen as less pronounced at local county, parish, and village-among which the levels than at higher levels (province or dis- 56 districts and 800 subcounties have the most trict and national). Still, corruption was political authority and resources. The deliv- perceived as substantial at all levels. ery of primary health care and primary edu- cation are decentralized. In health, staffing What impedes increased decisions are made at the district level, but efficiency? district funding comes largely from the cen- Several factors have kept decentralized ser- ter in the form of conditional grants with vice delivery from achieving its efficiency goals. explicitly identified uses. In education, the curriculum and most funding for primary Local governments have limited authority education flow from the center, but decisions In both the Philippines and Uganda, cen- about personnel, school construction, and tral governments place tight constraints on operation and maintenance are made locally. the authority of local governments. For example, in Uganda local governments have Does decentralization promote limited authority to adjust resource alloca- efficient public services? tions because most of their funding comes The study found that decentralizing service from the center in the form of conditional delivery offers benefits, but these benefits grants. This setup is considered necessary have not always materialized. because of weak financial management at PREMNOTE 55 JUNE 2001 local levels. Among other things, most dan households use private health care more health units, schools, and subcounty offices often than do Philippine households. The reported that there are no regular audits. quality of care is one of the main reasons for While these constraints limit the real- exit: private services were reported to be ization of allocative efficiency, central gov- of much higher quality than those in public ernments in both countries consider them health units. But competition from the pri- necessary at the current stage of decen- vate sector has not induced significant tralization to ensure that funds are used improvements in public health care deliv- properly. In addition, both central govern- ery, partly because many doctors in public ments emphasize that decentralization is health units also run private practices. ongoing and gradual, and that further devo- The argument that decentralization Further devolution lution of authority over resources needs to increases allocative efficiency presumes that be preceded by increased accountability citizens "vote with their feet"-that is, move of authority needs at local levels. to anotherjurisdiction if public services are unsatisfactory. It is argued that this threat to be preceded by Citizen influence at the local level is ham- makes governments more likely to satisfy the pered by limited information wishes of their citizens, to avoid risking the increased account- Citizens in both countries have a number loss of tax revenue. But the study found that of ways to express their views to local mobility acrossjurisdictions was rarely dri- ability at local levels governments. These include voting, local ven by public service delivery. Thus the threat committees, and other civic action. Both of migration does not appear to discipline countries have had high voter turnouts in local governments in these two countries. elections, and the survey found that vot- ing patterns and rationales are similar Local governments have weak capacity between local and national elections. Local Local governments suffer from severe capac- committees were considered of limited use ity constraints-especially in Uganda- in Uganda, but in both countries civic action limiting the effectiveness of decentralized was common at the community level. service delivery. Materials and equipment But limited information on local politics needed to carry out work were in short sup- and events is constraining the effectiveness ply. In addition, a shortage of skilled staff of these efforts. Citizens in both countries and lack of training were hampering deliv- were less informed about local government ery. Capacity constraints did not appear to than national government. While citizens in be as severe in the Philippines. both countries rely on the media for infor- In both countries local governments mation about national politics and corrup- reported following merit-based personnel poli- tion, they rely largely on community leaders cies. Merit and quality of services were rated (in Uganda) and local officials and personal as the main criteria for promotion. But there contacts (in the Philippines) for such infor- was evidence of corruption and favoritism mation at the local level. Media coverage of in personnel management in both countries. local politics and events is limited. This is noteworthy because the analysis confirms the Decentralization can inhibit the delivery of importance of the media by indicating that some public goods better access to the media is associated with Evidence from the Philippines and Uganda better education and health care. highlights the need to preserve the delivery of public goods with interjurisdictional Exit mechanisms have limited effects on spillovers-such as immunization and com- public service delivery municable disease control-as vertical pro- Citizens in both countries can opt out of the grams under central government control. In government system in favor of nonpublic Uganda immunization rates have been falling. service providers or move to another loca- One reason for the fall is thatvaccination ser- tion in search of better public services. Ugan- vices have been decentralized. Although the PREMNOTE 55 JUNE 2001 central government still delivers vaccines to countries radio is especially important for health facilities, local governments have been disseminating information about govern- made responsible for funding outreach pro- ment. The media, however, tend to focus grams to immunize children and pregnant on national events and politics. In a decen- women. But local governments do not always tralized environment, adequate coverage of view immunization as a priority, and the study local events and politics is also important. found that households ranked immunization as a low priority for additional government Further reading spending. As a result program funding and Azfar, Omar, Satu Kahkonen, and Patrick immunization rates have fallen. Meagher. 2001. "Conditions for Effective Decentralized Governance: A Synthesis Policy implications of Research Findings." University of Mary- Decentralization Decentralization holds a lot of promise, but land, Center for Institutional Reform and whether it improves public service delivery the Informal Sector, College Park, Md. needs to be depends on the institutional arrangements Azfar, Omar, Tugrul Gurgur, Satu Kahkonen, governing its implementation. Several con- Anthony Lanyi, and Patrick Meagher. 2000. accompanied by ditions must be met before the full benefits "Decentralization and Governance: An of decentralization can be reaped. Empirical Investigation of Public Service reforms that First, for decentralization to increase alloca- Delivery in the Philippines." University tive and productive efficiency, local govern- of Maryland, Center for Institutional increase the ments need to have the authority to respond Reform and the Informal Sector, College to local demand as well as adequate mecha- Park, Md. transparency and nisms for accountability. Because granting Azfar, Omar, Satu Kahkonen, Anthony Lanyi, authority without accountability can lead to Patrick Meagher, and Diana Rutherford. accountability of corruption and lower productive efficiency, 1999. "Decentralization, Governance and decentralization needs to be accompanied Public Services: The Impact of Institu- local government byreforms thatincrease the transparencyand tional Arrangements-A Review of the accountability of local government. Literature." University of Maryland, Cen- Second, functions need to be devolved ter for Institutional Reform and the Infor- to a low enough level of government for mal Sector, College Park, Md. allocative efficiency to increase as a result of decentralization. Low-level governments This note was written by Satu Kahkonen (Senior are likely to be aware of local preferences Economist, Macroeconomics 2 TechnicalFamily, and, if able to do so, are likely to adjust ser- Africa Region) and Anthony Lanyi (Director of vice delivery accordingly. EconomicPolicy, CenterforInstitutional Reform Third, citizens should have channels to and the Informal Sector, University of Maryland) communicate their preferences and get based onAzfar, Kahkonen, and Meagher (2001). their voices heard in local governments. But The broader study on decentralization and gov- the existence of such channels is not ernance was supported byagrantfrom the World enough. To effectively influence public poli- Bank's Netherlands Trust Fund. cies and oversee local governments, citizens Ifyou are interested in similar topics, consider need to have information about govern- joining the Decentralization Thematic Group. ment policies and activities. The media play ContactJennie Litvack, x8051 9, or click on The- a crucial role in this area. In developing matic Groups on PREMnet. => ; This note series is intended to summarize good practice and key policy find- ings on PREM-related topics. The views expressed in these notes are those of the authors and do not necessarily reflect the views of the World Bank. PREM- notes are distributed widely to Bank staff and are also available on the PREM website (http://prem). If you are interested in writing a PREMnote, email your idea to Sarah Nedolast. For additional copies of this PREMnote please contact WddRadkonamkM=m the PREM Advisory Service at x87736. Prepared for World Bank staff