19683
1999
Decentralization                                WORLD
Briefing Notes                                   BANK
Edited b!                                        I NSTITUTE
Jen zie Litvack and
Jessica Seddoni                                 WBI Worknl Papers
Contributors
Junaid Aimtad, -Harry Blaii; Talib Esmail,       n colrlabatLon with
Janzes Ford, Bert Hofinan, Graham Kerr,
Elizabeth- King, Riitta-Liissa Kolehmainen-
Aitken, Ernst Llltz, Keith McLeani, Deninis
Rondinelli, David Seiwell, An war Shah, Gi lioc
De Toininaso, and Dano Wiesi                       *    1t        |
_j  -J*<o7-! J  Si ?    :1 -_ '
a                ', '''Si                                                      ,'';
SW~~~~~~~~~~~   f
p4 ~~~~~~~~~'
A-                             .; .-4   e.
J :fr;l,           ,                     r
:                  r                                           z~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~7






DECENTRALIZATION
BRIEFING NOTES
Edited by
Jennie Litvack
and
Jessica Seddon
Contributors
Junaid Ahmad, Harry Blair, Talib Esmail, James Ford,
Bert Hoffman, Graham Kerr, Elizabeth k:ing, Riitta-
Liissa Kolehmainen-Aitken, Ernst Lutz, Keith McLean,
Dennis Rondinelli, David Sewell, Anwar Shah, Giulio
De Tommaso, and Dana Weist
World Bank Institute
Washington, D.C.



Copyright © 1999
The International Bank for Reconstruction
and Development/The World Bank
1818 H Street, N.W.
Washington, D.C. 20433, U.S.A.
The World Bank enjoys copyright under protocol 2 of the Universal Copyright Convention. This
material may nonetheless be copied for research, educational, or scholarly purposes only in the
member countries of The World Bank. Material in this series is subject to revision. The findings,
interpretations, and conclusions expressed in this document are entirely those of the author(s) and
should not be attributed in any manner to the World Bank, to its affiliated organizations, or the
members of its Board of Executive Directors or the countries they represent.



Contents
Foreword  .................................................v
Acknowledgments  ................................................. vii
Contributors .................................................                                                    ix
1.   What and Why .................................................1
What Is Decentralization? ..................................................2
Dennis Rondinelli
Rationale for Decentralization ..................................................6
James Ford
2.   Project Design: Political, Fiscal, and Administrative Decentralization .............                          9
Constitutional, Legal, and  Regulatory Framework for Decentralization  ......... 11
James Ford
Participation, Civil Society, and  Decentralization .......................................... 15
Jessica Seddon
Expenditure Assignment ................................................. 19
Anwar Shah
Issues in Tax Assignment ................................................. 23
Anwar Shah
Intergovernmental Transfers and  Grants .................................................   27
Anwar Shah
Decentralizing Borrowing Powers ......................                         ........................... 32
JunaidAhmad
Civil Service Reform  and  Decentralization  .................................................  39
Jessica Seddon and Gtulio De Tommaso
Information and  Monitoring in Decentralized  Systems ................................. 43
Dana Weist
Local Technical and  Managerial Capacity .................................................                     45
Graham Kerr
3.   Service Delivery .................................................                                           51
Decentralization of the Education  Sector .................................................  52
Keith McLean and Elizabeth King
Decentralization of the Health Sector ......................................                    ........... 57
Riitta-Liissa Kolehmainen-Aitken
Decentralization and  Safety Nets ....................                      ............................. 62
Jennie Litvack
Decentralization of Infrastructure .....................                     ............................ 66
Jessica Seddon
Decentralization  of the Water Supply ......................................                    ........... 72
Keith McLean
Natural Resources Management and  Decentralization .................................. 77
Ernst Lutz and Talib Esmail
jjjI 



Decentralization Briefing Notes
4.   Potential Impacts of Decentralization  ....................................................   83
Equity and Multileveled Government ....................................................   83
David Sewell
Macroeconomic Impact of Decentralization ................................................  89
Bert Hoffman
Decentralization and Economic Growth  .................................................... 93
Jessica Seddon
Accountability, Transparency, and Corruption in Decentralized Systems  .... 96
Hany Blair
Boxes
2.1  Principles and Better Practices in Grant Design ............................................ 30
2.2  Criteria for the Design of Intergovernmental Fiscal Arrangements ......... ....... 31
2.3  Evolution of Capital Markets and Decentralization of Borrowing Powers .... 35
3.1  Alternative Ways to Produce or Provide Local Public Services ..........                     .......... 67
Tables
2.1  A  Representative Assignment of Expenditure Responsibilities ..........                     .......... 21
3.1  Main Effects of Decentralization on Roads, Electricity, and Water Supply ... 71
3.2  Division of Responsibilities between Agency and Water Users ..........                       ......... 75
- Iv-



Foreword
Decentralization, a set of policies that encompasses fiscal, political, and adminis-
trative changes, can impact virtually all aspects of development. The structure of
intergovernmental relations affects everything from the efficiency and equity of
service delivery, the social safety net, and poverty alleviation programs to the de-
velopment of the financial sector and macroeconomic stability. The division of power
and responsibility between levels of govenmment also raises issues of institutional
capacity, as well as corruption and governance. Unfortunately, the far-reaching
implications of decentralization are often overlooked, as the literature and spe-
cialists tend to focus on specific dimensions. Top policymakers in decentraliz-
ing countries are often the only ones who address the full range of issues.
The notes in this volume are designed to highlight the broad range of issues
that need to be considered with regard to decentralization. They are intended to
provide brief overviews of the many different aspects of decentralization and sum-
marize key issues that need to be considered by practitioners. The notes were pre-
pared originally for a World Bank intemal web site to help Bank staff understand
the multidimensional aspects of decentralization. Each note is not meant to ad-
dress its topic in great depth, but rather to help readers consider which questions to
ask in their countries. Most importantly, by bringing these individual notes together
in this volume, readers are encouraged to consider the cross-cutting nature of de-
centralization and the importance of a comprehensive approach.
A key lesson emerging from this approach is that the impact of decentralization
will depend greatly on the many factors related to design. Simply put, with decen-
tralization, as with many complicated policy issues, the "devil is in the details."
The outcome will depend on myriad individual political, fiscal, and administrative
policies and institutions as well as their interaction within a given country.'
The volume is divided into four chapters. The first chapter, What and Why, pro-
vides an overview and explains the multidimensional nature as well as the ratio-
nale for decentralization. It also sets the stage for understanding that the impact of
decentralization will depend greatly on many specific policy and institutional is-
sues described in the second chapter, Project Design: Political, Fiscal, and Administra-
tive Decentralization. This chapter addresses key political factors (the constitutional,
legal, and regulatory framework and local participation), fiscal dimensions (ex-,
penditure assignment, tax assignment, intergovernmental transfers, and subnational
borrowing), and administrative considerations (civil service reform, information
and monitoring, and local technical and managerial capacity. The third chapter,
1. For a comprehensive discussion of these issues and their impact in developing coun-
tries, see Jennie Litvack, Junaid Ahmad, and Richard Bird, "Rethinking Decentralization in
Developing Countries," Sector Studies Series, PREM, World Bank, 1998.
-v -



Decentralization Briefing Notes
Service Delivery, focuses on issues pertaining to decentralization of specific sectors
and the delivery of different services (education, health care, safety nets, infrastruc-
ture, water, natural resources management). Finally, the fourth chapter, Potential
Impacts of Decentralization, discusses potential impacts on equity objectives, macro-
economic stability, growth, and accountability and corruption.
-vi -



Acknowledgments
This volume is a collective effort of the World Bank's Decentralization Thematic
Group. The notes that make up the volume were drafted, reviewed, and edited by
many insightful, experienced people.
We appreciate the contributions of Richard Bird and Keith McLean, who re-
viewed and contributed to the editing of several of the notes. In addition, the fol-
lowing individuals reviewed at least one of the notes: Harold Alderman, William
Ascher, Naazneen Barma, Joel Brakan, Greg Browder, Robert Ebel, Peter Fallon,
Ariel Fiszbein, Sandra Giltner, Graham Kerr, Christine Kessides, Deepa Narayan,
Barbara Nunberg, Jesse Ribot, Dennis Rondinelli, Eluned Schweitzer, David Sewell,
Dominique van de Walle, and Debbie Wetzel.
-vii -






Contributors
Junaid Ahmad is a principal economist in the Africa Region Resident Mission:
South Africa at the World Bank
Harry Blair is a democracy specialist in the Democracy and Governance Program
at USAID
Giulio De T ommaso is a public sector management specialist in the Middle East
and North Africa Social and Economic Development Group at the World Bank
Talib Esmail is a sector economist in theAfrica Technical Families: Rural Develop-
ment 2 at the World Bank
James Ford is a principal urban sector specialist in the Transportation, Water, and
Urban Development Department at the World Bank
Bert Hofman is a senior country economist in the Africa Technical Families: Mac-
roeconomics 1 at the World Bank
Graham Kerr is a senior associate at the consulting firm Associates in Rural Devel -
opment, Inc.
Elisabeth King is a principal economist in the Development Research Group at
the World Bank
Riitta-Liisa Kolehmainen-Aitken is a senior program associate in the Technical
Unit of the Family Planning Management Development Project at Management
Sciences for Health
Jennie Litvack is a senior economist in the Poverty Reduction & Economic Man-
agement Public Sector Division at the World Bank
Ernst Lutz is a senior economist in the Environment Department at the World Bank
Keith McLean is a consultant in the Rural Development Department at the World
Bank
Dennis Rondinelli is the Galaxo Distinguished International Professor of Man-
agement at the Kenan-Flagler Business School and director of the Center for Global
Business Research at the Frank Hawkins Kenan Institute of Private Enterprise,
University of North Carolina at Chapel Hill
Jessica Seddon is a consultant in the Office of the Chief Economist at the Inter-
American Development Bank
David Sewell is a fiscal economist in the Middle East and North Africa Social and
Economic Development Group at the World Bank
- ix -



Decentralization Briefing Notes
Anwar Shah is a principal evaluations officer in the Operations Evaluation De-
partment Country Evaluation and Regional Relations at the World Bank
Dana Weist is a senior financial economist in the East Asia and Pacific Urban De-
velopment Sector Unit at the World Bank
- x-



1
What and Why
This chapter contains two notes that describe the different forms of decentraliza-
tion and the rationale for decentralization. They provide a broad overview to the
several types of decentralization that can occur across countries, and even within
the same country and sector. Distinguishing among different types of decentraliza-
tion facilitates the discussion of policy design, and particularly of impact. For ex-
ample, whether a country chooses to "deconcentrate," 'delegate," or "devolve"
certain functions-and the impact of those decisions-will depend on the
policymakers' objectives, as well as on many factors related to the political, admin-
istrative, and fiscal structure of that country. It is important to introduce consis-
tency in any discussion of decentralization to avoid comparing apples and oranges
and to ensure that lessons can be elicited where appropriate.
-1 -



Decentralization Briefing Notes
What Is Decentralization?
Decentralization-the transfer of authority and responsibility for public functions
from the central government to subordinate or quasi-independent government or-
ganizations or the private sector-covers a broad range of concepts. Each type of
decentralization-political, administrative, fiscal, and market-has different char-
acteristics, policy implications, and conditions for success. All these factors need to
be carefully considered before deciding whether projects or programs should sup-
port reorganization of financial, administrative, or service delivery systems.
While distinguishing among the different types of decentralization is useful
for highlighting its many dimensions and the need for coordination, these con-
cepts overlap considerably. Political, administrative, fiscal, and market decen-
tralization can appear in different forms and combinations across countries, within
countries, and even within sectors. Precise definitions are less important than
ensuring a comprehensive approach.
Political Decentralization
Political decentralization aims to give citizens and their elected representatives more
power in public decisionmaking. It is often associated with pluralistic politics and
representative government, but it can also support democratization by giving citi-
zens or their representatives more influence in formulating and implementing poli-
cies. Advocates of political decentralization assume that decisions made with greater
participation will be better informed and more relevant to diverse interests in soci-
ety than those made only by national political authorities. The concept implies that
the selection of representatives from local electoral jurisdictions allows citizens to
better know their political representatives and allows elected officials to better know
the needs and desires of their constituents.
Political decentralization often requires constitutional or statutory reforms, de-
velopment of pluralistic political parties, strengthening of legislatures, creation of
local political units, and encouragement of effective public interest groups.
Administrative Decentralization
Administrative decentralization seeks to redistribute authority, responsibility, and
financial resources for providing public services among different levels of govern-
ment. It is the transfer of responsibility for planning, financing, and managing cer-
tain public functions from the central government and its agencies to field units of
government agencies, subordinate units or levels of government, semi-autonomous
public authorities or corporations, or areawide, regional, or functional authorities.
Administrative decentralization has three major forms-deconcentration, del-
egation, and devolution-each with different characteristics.
DEcoNcENTRAON. Deconcentration, the redistribution of decisionmaking author-
ity and financial and management responsibilities among different levels of the cen-
tral government, is often considered the weakest form of decentralization and is
used most frequently in unitary states. Within this category, however, policies and
- 2 -



What and Why
opportunities for local input vary: deconcentration can merely shift responsibilities
from central government officials in the capital city to those working in regions,
provinces, or districts, or it can create strong field administration or local adminis-
trative capacity under the supervision of central government ministries.
DELEGATION. Delegation is a more extensive form of decentralization. Through
delegation central governments transfer responsibility for decisionmaking and
administration of public functions to semi-autonomous organizations not wholly
controlled by the central government, but ultimately accountable to it. Govern-
ments delegate responsibilities when they create public enterprises or corpora-
tions, housing authorities, transportation authorities, special service districts, semi-
autonomous school districts, regional development corporations, or special project
implementation units. Usually these organizations have a great deal of discretion
in decisionmaking. They may be exempt from constraints on regular civil service
personnel and may be able to charge users directly for services.
IDEvoLurnoN. Devolution is the transfer of authority for decisionmaking, finance,
and management to quasi-autonomous units of local government with corporate
status. Devolution usually transfers responsibilities for services to municipalities that
elect their own mayors and councils, raise their own revenues, and have indepen-
dent authority to make investment decisions. In a devolved system, local govern-
ments have clear and legally recognized geographical boundaries over which they
exercise authority and within which they perform public functions. It is this type of
administrative decentralization that underlies most political decentralization.
Fiscal Decentralization
Financial responsibility is a core component of decentralization. If local govern-
ments and private organizations are to carry out decentralized functions effectively,
they must have adequate revenues-raised locally or transferred from the central
government-as well as the authority to make expenditure decisions. Fiscal decen-
tralization can take many forms, including:
* Self-financing or cost recovery through user charges
* Cofinancing or coproduction, in which users participate in providing services
and infrastructure through monetary or labor contributions
* Expansion of local revenues through property or sales taxes or indirect charges
* Intergovernmental transfers of general revenues from taxes collected by the cen-
tral government to local governments for general or specific uses
* Authorization of municipal borrowing and mobilization of national or local gov-
ernment resources through loan guarantees.
In many developing countries local governments or administrative units pos-
sess the legal authority to impose taxes, but the tax base is so weak and the de-
pendence on central government subsidies so ingrained that no attempt is made
to exercise that authority.
-3 -



Decentralization Briefing Notes
Economic or Market Decentralization
The most complete forms of decentralization from a government's perspective are
privatization and deregulation; they shift responsibility for functions from the public
to the private sector. They allow functions that had been primarily or exclusively
the responsibility of government to be carried out by businesses, community groups,
cooperatives, private voluntary associations, and other nongovernmental organi-
zations. Privatization and deregulation are usually accompanied by economic lib-
eralization and market development policies.
* Privatization-Privatization can range in scope from the provision of goods
and services based entirely on the free operation of the market to public-private
partnerships in which government and the private sector cooperate to provide
services or infrastructure. Privatization can mean allowing private enterprises
to perform functions that had previously been monopolized by government. It
can also mean contracting out the provision or management of public services
or facilities to commercial enterprises. There is a wide range of public-private
institutional forms and of ways in which such functions can be organized, par-
ticularly in infrastructure. Privatization can also include financing public sec-
tor programs through the capital market, with adequate regulation or measures
to ensure that the central government does not bear the risk for this borrowing,
and allowing private organizations to participate. And finally, it can mean trans-
ferring responsibility for providing services from the public to the private sec-
tor through the divestiture of state-owned enterprises.
- Deregulation-Deregulation reduces the legal constraints on private participa-
tion in service provision or allows competition among private suppliers for
services previously provided by the government or by regulated monopolies.
In recent years privatization and deregulation have become more attractive al-
ternatives to government provision in developing countries. Local governments
are also privatizing by contracting out service provision or administration.
Choosing the Most Appropriate Form of Decentralization
Under appropriate conditions all of these forms of decentralization can help broaden
participation in political, economic, and social activities in developing countries.
Where it works effectively, decentralization helps alleviate the decisionmaking bottle-
necks that are caused by central government planning and control of important
economic and social activities. Decentralization can help to simplify complex bu-
reaucratic procedures, and it can increase government officials' sensitivity to local
conditions and needs. Moreover, decentralization can help national government
ministries reach larger numbers of local areas with services; allow greater political
representation for diverse political, ethnic, religious, and cultural groups in
decisionmaking; and relieve top managers in central ministries of routine tasks, al-
lowing them to concentrate on policy. In some countries decentralization may cre-
ate a geographical focus at the local level, coordinating national, state, provincial,
-4 -



What and Why
district, and local programs more effectively, and can provide better opportunities
for local residents to participate in decisionmaking. Decentralization may lead to
more creative, innovative, and responsive programs by allowing local experimenta-
tion. It can also increase political stability and national unity by allowing citizens to
better control public programs at the local level.
But decentralization is not a panacea, and it does have potential disadvantages.
Decentralization may not always be efficient, especially for standardized, routine,
network-based services. It can result in the loss of economies of scale and of control
over scarce financial resources by the central government. Weak administrative or
technical capacity at local levels may result in services being delivered less effi-
ciently and effectively in some areas of the country. Administrative responsibilities
may be transferred to local levels without adequate financial resources, making
equitable distribution or provision of services more difficult. Decentralization can
sometimes make coordination of national policies more complex and may allow
functions to be captured by local elites. Also, distrust between public and private
sectors may undermine cooperation at the local level.
Project and program planners must be able to assess the strengths and weaknesses
of public and private sector organizations in performing different types of functions.
Before developing elaborate plans for decentralization, they must assess the lowest
organizational level of government that performs functions efficiently and effectively
and, for functions that do not have to be provided by government, the most appropri-
ate forms of privatization. Even program planners who do not see decentralization as
their primary motive must carefully analyze the types of decentralization alreadypresent
in a country in order to tailor policy plans to existing structures. The success of decen-
tralization also frequently depends on proper training for both national and local offi-
cials in decentralized administration. Technical assistance is often required for local
governments, private enterprises, and local nongovernmental groups in the planning,
financing, and management of decentralized functions.
Centralization and decentralization are not either-or conditions. In most coun-
tries an appropriate balance of centralization and decentralization is essential to
the effective and efficient functioning of government. Not all functions can or should
be financed and managed in a decentralized fashion. And even when national gov-
ernments decentralize responsibilities, they often retain important policy and su-
pervisory roles. They must create or maintain the enabling conditions that allow
local units of administration or nongovernmental organizations to take on moreW
responsibilities. Central ministries often have crucial roles in promoting and sus--
taining decentralization by developing appropriate and effective national policies
and regulations for decentralization and strengthening local institutional capacity
to assume responsibility for new functions.
- 5 -



Decentralization Briefing Notes
Rationale for Decentralization
Much of the decentralization that has taken place in the past decade has been
motivated by political concerns. For example, in Latin America decentraliza-
tion has been an essential part of the democratization process as discredited
autocratic central regimes are replaced by elected governments operating un-
der new constitutions. In Africa, the spread of multiparty political systems is
creating demand for more local voice in decisionmaking. In some countries,
such as Ethiopia, decentralization has come in response to pressures from re-
gional or ethnic groups for more control or participation in the political pro-
cess. In the extreme, decentralization represents a desperate attempt to keep
the country together in the face of these pressures by granting more autonomy
to all localities or by forging "asymmetrical federations." A variation on this
theme has been decentralization as an outcome of long civil wars, as in
Mozambique or Uganda, where opening political opportunities at the local level
has allowed for greater participation by all former warring factions in the gov-
ernance of the country. The transition economies of the former socialist states
have also massively decentralized as the old central apparatus crumbled. In
many countries decentralization has simply happened in the absence of any
meaningful alternative governance structure to provide local government ser-
vices. In some cases (particularly in East Asia) decentralization appears to be
motivated by the need to improve service delivery to large populations and the
recognition of the limitations of central administration.
Although in many cases decentralization seems to be happening without much
planning or design, there are a multitude of design issues that affect the impact of
decentralization on efficiency, equity, and macrostability. The success of decentrali-
zation is closely related to adherence to several of these design principles: finance
following on the clear assignment of functions, informed decisionmaking, adher-
ence to local priorities, and accountability. Because country circumstances differ,
often in subtle and complex ways, the policy and institutional instruments that
establish decentralization have to be shaped to specific circumstances.
Economic Arguments for Decentralization
A growing body of literature is examining the economic rationale for decentraliza-
tion. Economists justify decentralization on the grounds of allocative efficiency.
Their rationale is that decisions about public expenditure that are made by a level
of government that is closer and more responsive to a local constituency are more
likely to reflect the demand for local services than decisions made by a remote
central government. A second economic rationale for decentralization is to irnprove
the "competitiveness" of governments and enhance innovation-and hence the
likelihood that governments will act to satisfy the wishes of citizens. Another po-
tential benefit for decentralization is that people are more willing to pay for ser-
vices that respond to their priorities, especially if they have been involved in the
decisionmaking process for the delivery of these services.
-6 -



What and Why
A number of theoretical and practical issues complicate the view of whether de-
centralization is a good or undesirable economic strategy. At the macroeconomic level,
an important concern has been that decentralization may make stabilization policies
more difficult to implement and may even lead to destabilizing levels and composi-
tion of overall public expenditures and public debt. However, decentralized systems
can be designed to avoid destabilizing effects and to ensure correct incentives. Some
of the decentralization of the 1980s, for example, was actually an offloading of fiscal
imbalances by central governments to subnational governments. Under these cir-
cumstances, it is not surprising to see a strong association between decentralization
and fiscal imbalances at lower levels. Another macro level controversy for which
evidence is inconclusive is whether decentralization retards economic growth.
Concerns about equity-interjurisdictional and interpersonal-have been cen-
tral to the discussion of decentralization. Some jurisdictions are better endowed
with resources than others, perhaps because of size or location. In addition, histori-
cal circumstances (such as apartheid) may have created local disparities. Thus an
intergovernmental fiscal program may be designed to shift resources to disadvan-
taged areas to ensure that all citizens enjoy a minimum level of service, regardless
of location, or receive enhanced assistance to accelerate amelioration of deficits,
because of location. The allocation of poverty program grants to subnational levels;
should be analyzed carefully within a particular country context since lack of trans--
parency, or inadequate specificity in transfer design, sometimes results in wealthier
areas receiving more resources than poorer areas.
It is usually argued that central governments are ultimately responsible for ensur-
ing interpersonal equity. Where local economies are intrinsically open and many re-
sources, especially key human resources, are mobile, only limited success should be
expected from jurisdictionallyfocused distributional programs. Still, local governments
can and do play very important roles in implementing central distributional programs
and in determining a host of tax, expenditure, and intralocality transfer schemes.
The specific services to be decentralized and the type of decentralization will
depend on economies of scale affecting technical efficiency and the degree of
spillover effects beyond jurisdictional boundaries. These issues need to be taken
into account in the design of a decentralized system. In practice, all services do not
need to be decentralized in the same way or to the same degree.
In an important economic sense, the market is the ultimate form of decentrali-
zation in that consumers can acquire a tailored product from a choice of suppliers.
The nature of most local public services limits the full play of this option and estab-
lishes a government role in ensuring the provision of these services, but it does not
automatically require the public sector to be responsible for the delivery of all ser-
vices. Where it is possible to structure competition either in the delivery of a ser-
vice or for the right to deliver the service, the evidence indicates that the service
will be delivered more efficiently. Although uncommon in practice, local govern-
ments have successfully competed for the right to provide certain local services. In
an array of local public services in any particular country, a mix of solutions frorn
deconcentration to managed competition or privatization is likely to coexist.
- 7 -



Decentralization Briefing Notes
Conditions for Successful Decentralization
Although politics is the driving force behind decentralization in most countries,
decentralization may be one of those happy instances in which good politics and
good economics serve the same end. The political objectives of increased political
responsiveness and participation at the local level can coincide with the economic
objectives of better decisions about the use of public resources and increased will-
ingness to pay for local services. At least five conditions are important for success-
ful decentralization:
* The decentralization framework must link, at the margin, local financing and
fiscal authority to the service provision responsibilities and functions of the
local government, so that local politicians can deliver on their promises and
bear the costs of their decisions.
* Local communities must be informed about the costs of services and delivery
options and the resource envelope and its sources, so that the decisions they
make are meaningful. Participatory budgeting, as used in Porto Alegre, Brazil,
is one way to create this condition.
• Communities need a mechanism for expressing their preferences in a way that
is binding on politicians, so that there is a credible incentive for people to par-
ticipate.
• There must be a system of accountability based on public and transparent infor-
mation that enables communities to monitor the performance of the local gov-
ernnment effectively and to react appropriately to that performance, so that
politicians and local officials have an incentive to be responsive.
* The instruments of decentralization-the legal and institutional framework, the
structure of service delivery responsibilities, and the intergovernmental fiscal
system-must be designed to support the political objectives.
Meeting these conditions, or at least having local governments improve on the
central government's record, is a tall order, but achievable.
- 8 -



2
Project Design: Political, Fiscal, and
Administrative Decentralization
This chapter contains nine notes that address different aspects of political, fiscal,
and administrative decentralization. Although these issues are often not consid-
ered together, they are all important components of successful decentralization.
For example, the "right" fiscal arrangements absent the appropriate regulatory struc-
ture or information and monitoring can lead to very different outcomes. All the
issues presented in the notes of this chapter interact and thus all must be consid-
ered when designing decentralization.
The first two notes in this chapter pertain to political decentralization. The first,
note focuses on the constitutional, legal, and regulatory framework for decentrali-
zation and the second on the participation of society in decentralization.
The parameters within which decentralization can be designed depend on the
constitutional structure of the country, in particular whether it is a federal or unitary
government. In practice, the extent of autonomy provided to subnational levels var-
ies greatly; some federal countries resemble unitary countries (for example, Mexico),
and other unitary countries, de facto, operate as federal countries (China). Neverthe-
less, the legal opportunities and constraints for decentralization must be kept in mind.
These issues are explored in this note and are relevant to all the notes in this volume.
The extent to which decentralization is able to achieve benefits of allocative
efficiency depends in large part on the nature of local participation. Local leaders
can only reflect local needs and desires if channels for such input exist and if lead-
ers are accountable to their people. This note looks at the important relationship
between participation and decentralization.
The next four notes address important areas of fiscal federalism: expenditure
assignment, revenue assignment, intergovernmental transfer design, and
subnational borrowing. These four components make up the system of intergov-
ernmental finances. The impact of intergovernmental finances on the main eco-
nomic objectives of government-equity, efficiency, and macroeconomic
stability-depend on the system rather than any one component. For example,
significant decentralization of expenditures and revenues can lead to greater effi-
ciency and accountability in the wealthier parts of a country, but can lead to de-
clines in equity in the poorer parts unless intergovernmental transfers are used to
compensate the poorer areas. Similarly, expenditures and revenues need to be
matched when planning decentralization. If the center decentralizes more rev-
enues than expenditures, it can be starved of resources necessary to meet its na-
tional obligations; conversely, over decentralization of expenditures without
adequate revenues can lead to starving subnational levels of resources. The pres-
sure for unsustainable methods of finance increases in both these situations and
-9-



Decentralizaion Briefing Notes
can lead to macroeconomic instability. Thus, component parts of fiscal federal-
ism must be examined together as a system.
The final three notes in this chapter address administrative decentralization:
civil service, information and monitoring, and technical and managerial capacity.
In any decentralized systems, both local and higher-level authorities will play
complementary roles to enable local leaders to respond to the needs of their popu-
lations and ensure national interests. The benefits of decentralization (that is, greater
allocative efficiency) can only be realized if local leaders have flexibility to respond
to the local needs and desires of their constituents; if they have the financial and
human resources capacity to respond effectively; if they are accountable to their
populations as well as to the central government (for services delegated to local
governments); and if local and central governments can obtain information and
monitor services for which they are ultimately responsible.
Civil service issues are particularly important for decentralized social services
because they represent the highest share of recurrent costs. Depending on wage
and employment policy and sectoral administrative norms, they can impact greatly
the degree of flexibility provided to local levels. Although the literature on civil
service reform is well developed, little has been written on civil service reform and
decentralization. The World Bank is beginning the process of developing its knowl-
edge in this area. We are particularly interested in collecting more country experi-
ences (successes and failures) with decentralization and civil service issues.
- 10 -



Project Design: Political, Fiscal, and Administrative Decentralization
Constitutional, Legal, and Regulatory Framework
for Decentralization
Constitutions, laws, and regulations codify the formal parameters in which decen-
tralized systems are supposed to function. Structurally, the desirable architecture
of these rules is straightforward:
* The constitutions should be used to enshrine the broad principles on which
decentralization is to operate, including the rights and responsibilities of all
levels of government, the description and role of key institutions at central and
local levels, and the basis on which detailed rules maybe established or changed.
* One or more laws should define the specific parameters of the intergovernmental
fiscal system and the institutional details of the local governnent structure, includ-
ing key structures, procedures (including elections), accountabilities, and remedies.
* A series of regulations associated with each law should interpret and detail the
practices and measures by which the related law will operate. Laws that deal
with tasks that are shared between national and subnational governments should
include sections on intergovernmental relations.
Substantially greater detail and specificity accompany the move down this three-
platform architecture from constitution to regulations, whereas greater difficulty
and a higher degree of authority (minister, parliament, and constitutional assem-
bly) are required to move up from regulations to the constitution. The levels of
difficulty and the locus of authority in effecting changes are important factors in
determining where in the architecture particular aspects of the decentralization
system are defined, as well as the relative specificity of the definitions. The rigidi-
ties and flexibility established in this structure have important implications for the
management of a decentralized system.
The placement of an item within this architecture may be the result of consen-
sus, but it is often the outcome of sometimes difficult negotiations between com-
peting interests. Those concerned with macroeconomic stability, for example, may
want intergovernmental fiscal rules to be a matter for regulation under the Minis-
ter of Finance, to give that ministry maximum flexibility in public expenditure
management. Local government advocates, by contrast, may argue (as they did.
successfully in Brazil) for these fiscal distributional rules to be enshrined in the
constitution. In Uganda, the purposes and mechanisms for transfers are specifiedL
in the constitution along with a formula for determining the minimum size of the
pool from which block grants are to be distributed; details of the distributional
formulas are the subject of regulations.
Because decentralization is a complex social experiment, a good case may be
made for allowing more flexibility to change the specificity of implementation in-
struments, while enshrining the political and philosophical principles in the con-
stitution and the operating structures in the laws.
- 1 1 -



Decentralization Briefing Notes
Interactions between Legal and Regulatory Framework
and Decentralization
In addition to substantive law, a country's procedural laws can have profound ef-
fects on the success of decentralization efforts. For example, when local expenditures
must be pre-audited by a central authority, rigidities are introduced that make the
benefits of decentralization more difficult to achieve. When reviewing the legal frame-
work for decentralization, it is not sufficient to examine laws specifically dealing
with decentralization. Other laws that mandate aspects of service delivery, civil ser-
vice, budgeting, and so on must also be considered to ensure a consistent approach.
Treatment of key issues in the legal and regulatory framework will be shaped by
the government's structure as a unitary or a federal system. In some federal systems
(for example, Canada and India) local governments are completely under the au-
thority of state or provincial governrnents. The federal government is thereby lim-
ited in the relationships it may establish with the local level and must seek to affect
local behavior and outcomes through the states or provinces. A decentralization policy
such as India is trying to establish is significantly complicated by this factor.
Some unitary systems may exercise extremely centralized control over local
governments. In Indonesia, the Ministry of Home Affairs has had the authority to
appoint (and remove) mayors and even village heads. The structural impediments
in designing a decentralized system in this context are few, but that does not mean
that the process of instituting such a system is without critical hurdles. Indonesia
has had decentralization legislation on its books since 1974; the process there re-
mains far from completion.
As with other key aspects of decentralization, the legal and regulatory frame-
work will be tailored to country circumstances. Nevertheless, there are issues this
framework may be expected to address. Those issues of potential interest to the
work of the World Bank include the following:
* Classification of local governments within the tiers established under the con-
stitution
* Broad organizational structures and their roles and responsibilities
* Terms of office, operating powers, procedures, and limitations of the political
leadership, as distinct from the civil service
* The degree of autonomy of personnel policies and administration of local
governments
* The taxing and fiscal administration authority of local governments
* The borrowing authority and capacities of local governments
* The distribution of budgeting, expenditure management, accounting, auditing,
and reporting requirements
* Service provision and delivery authority
* The mechanisms for citizen participation and voice.
The legal and regulatory framework also should be designed to recognize dif-
ferences in management capacity. The assignment of functional responsibilities-
-12 -



Project Design: Political, Fiscal, and Administrative Decentralization
for example, provincial capital or designated growth center-often implicitly rec-
ognizes the varying capabilities of municipalities, but a more dynamic framework
that recognizes capacity based on performance over time would be more desirable
in the long run. Matching degree of autonomy and privileges to a set of perfor-
mance indicators, which might include total expenditure, degree of revenue self-
sufficiency, budget management performance (absence of deficits), and service
delivery performance (client surveys), would allow the legal and regulatory frame-
work to adjust for changes in local capacity. The appropriate time period for reas-
sessments and indicators would need to be linked to country circumstances as well
as to the specific details of the decentralization framework.
Warning Flags
Five warning flags, selected from a potentially long list of downside risks, may
deserve special attention. First, local governments at the same nominal level may
vary considerably in their capacities. West Bank and Gaza, for example, have mu-
nicipalities that vary in population size from about 10,000 people to over 1 million,
with management capacities to match. Differences in fiscal capacity may be recog-
nized in the equity component of the intergovernmental fiscal system, but the fact
that management and administrative capacities also may vary substantially is rarely
accounted for. It is useful to have the legal and regulatory system recognize signifi-
cant differences in management capacities through a classification of local govern-
ment within levels. Policies and strategies to address these differences may then be
coherently considered.
Second, local governments should have the ability to borrow when they have
the capacity to repay. However, for moral hazard reasons (discussed in greater de-
tail in the section on borrowing) every effort must be made to promote the perspec-
tive that local government loans are internal obligations of local governments and
not of higher levels of government unless explicitly stated otherwise. The impor-
tance as well as the difficulties of doing this is illustrated by the circumstances of
subnational debt in Brazil. The legal and regulatory framework can support this
message by specifying the conditions under which local governments may borrow,
the limits of those borrowings, the reporting requirements for debt and debt ser-
vice, and the penalties for violating the rules.
Third, local government laws have not always anticipated the options, including
private participation and managed competition, that maybe pursued in the delivery
of local public services. As a consequence, legal barriers may inappropriately restrain
the ability of local authorities to select the most desirable options for the delivery of
decentralized services. China's cities have been imaginative in innovating and deliv-
ering some services not anticipated in the legal and regulatory framework within
which they operate; nevertheless, even under these circumstances rationalization is
desirable. Inappropriate barriers and constraints should be avoided or corrected in.
the design and detailing of the legal and regulatory framework for decentralization.
Fourth, a voting democracy is often believed to satisfy the conditions for citi--
zen participation and voice in the design of decentralized systems, but in prac-
tice this may not be sufficient. Meaningful participation requires that citizens be
- 13 -



Decentralization Briefing Notes
informed and that their voices have impact where consequences are immediate.
At a minimum, the legal and regulatory system needs to provide for full, timely,
and easily accessible public disclosure of resource allocation decisions-in bud-
gets, procurement, and expenditure programs. An output and outcome orienta-
tion to expenditure management would be even more desirable. Uganda is
designing readily accessible budgets for all levels of government as part of an
expenditure management reform program that emphasizes an output and out-
come orientation. In addition, citizens must have reliable, secure access to the
means to enforce appropriate penalties for violations of rules.
Fifth, the terms of office for local political leaders are closely related to issues
of authority and accountability. Mayors need incentives to focus beyond the short
term to at least the medium term. This requires a long enough tenure or potential
tenure to be able to be seen to be accomplishing meaningful objectives. Mexico's
three-year, nonrenewable mayoral terms, for example, have been associated with
a very short-term focus in local officials' governance strategies. Where multiple
terms are allowed, three- to four-year terms are desirable. Where only single terms
are permitted, then five to six years would be appropriate. The detailed design of
authority, powers, accountability systems, and procedures must be related to lo-
cal circumstances, including issues that may range from cultural traditions to the
state of accounting and auditing systems. Considerations also include the bal-
ance to be struck between leaving room for aggressive leadership and protecting
the community against excesses, a choice that is a matter of political taste and is
often a consequence of historical experience.
The discrepancy between formal rules and actual practice regularly observed
in many countries is itself a commentary on the importance of certain character-
istics of the design and implementation of legal and regulatory systems. Ambi-
guity and complexity create openings for conflicting interpretation and confusion.
A single agreed interpretation is essential. Particular efforts to prepare and dis-
seminate popularized versions of the legal and regulatory system, as Uganda has
done, must be a key part of the decentralization strategy. Complexity is often
unavoidable, especially at the level of instruments for implementation. It helps,
however, if one instrument is not asked to do too much. This facilitates commu-
nication and implementation of the policy that the instrument is intended to sup-
port, better monitoring of the effectiveness of the instrument in that role, and
adjustment to the instrument or the policy.
- 14 -



Project Design: Political, Fiscal, and Administrative Decentralization
Participation, Civil Society, and Decentralization
Participation and decentralization have a symbiotic relationship. On the one hand,
successful decentralization requires some degree of local participation. Subnational
governments' proximity to their constituents will enable them to respond better
to local needs and efficiently match public spending to private needs only if in-
formation flows between citizens and local governments. On the other hand, the
process of decentralization can itself enhance the opportunities for participation
by placing more power and resources at a closer, more familiar, more easily influ-
enced level of government. In environments with poor traditions of citizen par-
ticipation, decentralization can be an important first step in creating regular,
predictable opportunities for citizen-state interaction.
This symbiotic relationship between decentralization and participation can lead
to somewhat contradictory policy guidelines. Mechanisms for citizen participation
could be considered a helpful precondition when evaluating the prospects for suc-
cessful decentralization. Accordingly, the design of decentralization should take
into account the opportunities and limitations imposed by existing channels of lo-
cal participation. Lack of participatory mechanisms, however, could be considered
a motivation for decentralization and can help create local demand for more par-
ticipatory channels to voice preferences.
The following paragraphs discuss each of these dimensions: participation as a
means to successful decentralization and as a goal of decentralization. First, the
broad mechanisms for citizen input that are best seen as parameters of decentrali-
zation policy are considered. These types of institutional structures, embedded in
the national political environment and developed over a long period of time, can-
not be altered quickly by a simple regulation. Then the smaller, more specific av-
enues for citizen participation that can be created in the process of decentralization
are reviewed. Through such incremental changes, broader opportunities for citi-
zen participation and democratic governance may eventually evolve.
Participation as a Precondition: Broad Decisionmaking Influence
The existing institutionalized channels for participation and the ability of people to
use them should be taken into account in the design of decentralization programs
in developing countries. Fair and regular local elections and high levels of social
capital (community cohesion and history of working together) enable citizens to
signal their preferences efficiently and enforce leaders' compliance with their wishes.
Studies have shown that broad, ongoing citizen control over leaders can im-
prove the quality of government action. The continuous struggle to stay in office
might lead to broad efforts to satisfy consumers on all fronts. Results from stud-
ies on Colombia, for example, show that public officials who fear for their jobs are
more likely to pick better staff to carry out the day-to-day work of government.1
Research on regional government in Italy found that governments that were more
1. Ariel Fiszbein, "Emergence of Local Capacity: Lessons from Colombia," World De-
velopment 25:7(1997):1029-43.
- 15 -



Decentralization Briefing Notes
open to constituent pressure were more successful at managing resources and
creating innovative programs to distribute services effectively.2 A recent study of
villages in Tanzania shows a positive association between the quality of local
schooling and the level of social capital and between economic well-being and
high levels of social capital.3 Another study of local governments in Mexico sug-
gests that Oaxaca province's long history of participatory mechanisms may help
it provide services more effectively than socioeconomically similar, but less par-
ticipatory Chiapas.4 Several studies of decentralization and government size have
posited that more participatory local governments tend to be larger because their
constituents trust them with more resources.5
In some cases, however, broad participation can be disruptive. Local elec-
toral cycles can lead to periodic fiscal indiscipline as local leaders try to attract
more votes. Evidence on institutions and fiscal responsibility in Latin America
shows that two-party governments with hierarchical budget processes tend to
spend less than multiparty coalition governments with more collegial, partici-
patory budgeting processes.6
Assessing how much citizen input affects local government's actions provides
a starting point for designing decentralization policies. Such initial conditions help
determine the extent to which decentralization will increase the overall respon-
siveness of government to citizens and provide a guideline for the inclusion of
participation-enhancing measures in decentralization policy.
Regular elections, local referendums, permanent public-private councils, and
other institutional structures are other easily identifiable conditions that may im-
prove the ability of local governments to identify and act on citizen preferences.
Levels of social capital, which determine how well citizens are able to take advan-
tage of the institutional arrangements for participation, are slower to develop and
harder to determine. The presence and activities of nongoverrunental organiza-
tions (NGOs) and other citizen groups can be a revealing indicator, but it is also
important to determine whom these organizations really represent. Disparities be-
tween those who are represented by an NGO and those who remain outside the
system can lead to the capture of local goverrnent by one group and preclude
efficient matching of public spending to constituent needs.
2. Robert Putnam, Making Democracy Work: Civic Traditions in Modern Italy (Princeton,
New Jersey: Princeton University Press, 1993).
3. Deepa Narayan and Lant Pritchett, "Cents and Sociability: Household Income and
Social Capital in Rural Tanzania," Policy Research Working Paper no. 1796 (Washington,
D.C.: World Bank, 1997).
4. J. Fox and J. Aranda, Decentralization and Rural Development in Mexico: Community
Participation in Oaxaca's Municipal Funds Program, Monograph Series 42 (San Diego, Califor-
nia: Center for U.S.-Mexican Studies, University of California, 1996).
5. W. Oates, "Searching for Leviathan: An Empirical Study," American Economic Review
75:748-57.
6. E. Stein, E. Talvi, and A. Grisanti, "Institutional Arrangements and Fiscal Perfor-
mance: The Latin American Experience," Working Paper no. 6358 (Cambridge, Massachu-
setts: National Bureau for Economic Research, 1998).
- 16 -



Project Design: Political, Fiscal, and Administrative Decentralization
Participation as a Goal: The More Attainable Beginning Reforms
Decentralization relies on participation to improve the allocation of services, but it
does not require the kinds of broad citizen input mentioned earlier. In cases where
local governments are not elected, where the electoral process favors a small group
of elites, or where low levels of social capital impede active citizen-state interchange,
the process of decentralization can be designed to build up more limited types of
participation. Issue- and project-specific mechanisms for enhancing the flow of in-
formation between government and citizens can often be implemented more quickly
and easily at the local level than in central governments. Demand-driven projects
are one of the key tools (often supported by the World Bank and other donors) to
deepen the decentralization process.
Local governments use a wide variety of techniques to gather information from
their constituents. Surveys are a useful tool for refining service delivery and decid-
ing on appropriate user fees. A survey of water supply users in Baku, Azerbaijan,
for example, revealed that users were willing to pay more for better quality of ser-
vice and pinpointed the most important problems to address.7 Bangalore and sev-
eral other Indian districts use report cards to evaluate service delivery, Uganda
requires beneficiary feedback on some government services, and Nicaragua used a
series of surveys to reform the bus system and adjust fares. Issue-specific ad hoc
councils can also be a quick, easy way to determine citizen preferences and draw
on private sector expertise. Colombia's municipalities, for example, have supple-
mented the public sector's weak technical capacity by involving more private ex-
perts on a short-term basis. Inviting citizen participation in implementing
decentralized programs can also be cost-effective. World Bank-funded projects have
shown that, although initial training costs can be high, the long-run savings are
substantial and projects tend to be better maintained. Nicaragua's municipal de-
velopment program, for example, which used beneficiary participation in barrio
upgrading projects, completed projects 20 percent faster than predicted and had a
50 percent higher rate of return than expected.
In the short run, these mechanisms can help decentralization achieve improve-
ments in allocative efficiency if local leaders base their decisions on information
provided through these participatory channels. In the long run, these limited mecha-
nisms for participation can evolve into closer and more meaningful interaction be-
tween citizens and their local governments only if leaders are democratically elected
and accountable to their constituents.
Importance of Information Flows
Citizen participation in some form is an essential part of successful decentraliza-
tion. It is becoming a more common element in developing country political envi-
ronments-13,000 units of local government in Latin America are now elected, up
7. World Bank, "Azerbaijan: Baku Water Supply Rehabilitation Project," Envirom-nent
Department Papers Series, Assessment Series Paper 17 (Washington, D.C., 1995).
- 17 -



Decentralization Briefing Notes
from 3,000 in 1973-but the flow of information is by no means undistorted. De-
centralization policies should take informational imperfections into account and
attempt to improve the depth and degree of citizen participation in local govern-
ment action. Local government responsiveness, one of the main rationales for de-
centralizing, cannot be improved when there are no mechanisms for transferring
information between the local government and its constituents.
-18-



Project Design: Political, Fiscal, and Administrative Decentralization
Expenditure Assignment
Fiscal decentralization involves shifting some responsibilities for expenditures
and revenues to lower levels of government. The degree of autonomy of
subnational entities in the allocation of their expenditures is one important fac-
tor in determining the type of fiscal decentralization. (The other, their ability to
raise revenue, is addressed later.) This section outlines principles and best prac-
tice and highlights how country circumstances are ultimately the best determi-
nant of expenditure assignments.8
It is important to distinguish whether local governments determine the alloca-
tion of expenditures themselves or whether the center mandates expenditures and
local levels simply execute those expenditures. Statistics on subnational finance
(most notably those in the IMF's Government Finance Statistics) usually aggregate
these two types of expenditures and present them in one figure as the "percentage
of subnational expenditures." Thus, analysts must be very careful when using sta-
tistics on subnational finance as an indicator of local autonomy. Autonomy will not
be enhanced by fiscal decentralization if funds are tied by the center. (Of course,
some countries do derive the benefits of local expenditure autonomy, because local
governments find ways to access the funds and circumvent the central mandates!)
Unitary and federal governments provide different opportunities for fiscal de-
centralization. In systems of government, unitary subnational governments are not
constitutionally empowered to make decisions over a specified range of govern-
ment functions and services; rather, there are multiple subordinate levels of the
same government (central, provincial, district). Federal governments, on the other
hand, have constitutionally protected subnational levels of government. Thus the
possibilities for independent decisiomnaking are clearly stronger under federal
systems. While subnational governments (provinces or states) always enjoy consti-
tutional protection under federal systems, however, local governments (county,
municipal, township) may not. In practice, the extent and nature of decisionmaking
power exercised by lower tiers vary widely from country to country in both federal
and unitary countries and may change from time to time. At the extremes, some
nominally federal countries (for example, Venezuela) may be considerably less
decentralized than nominally unitary countries (for example, Colombia).
Despite such complexities, both theory and experience strongly suggest that ex-
penditure responsibilities need to be stated as clearly as possible to enhance account-
ability and reduce unproductive duplication of authority and legal challenges. Many
would argue that decisionmaking should follow the principle of subsidiarity-that
is, it should take place at the lowest level of government consistent with allocative
efficiency, the geographic area that internalizes the benefits and costs of
8. For further elaboration on the theory of expenditure assignments, see, for example,
Robin Boadway, The Constitutional Division of Powers: An Economic Perspective (Ottawa: Eco-
nornic Council of Canada, 1992); Wallace Oates, Fiscal Federalism (New York: Harcourt, Brace,
Jovanovich, 1972); Anwar Shah, The Reform of Intergovernmental Fiscal Relations in Developing
and Emerging Market Economies (Washington, D.C.: World Bank, 1994).
- 19 -



Decentralization Briefing Notes
decisionmaking for a particular public service. The optimal size of jurisdiction for
each service could differ theoretically, but in practice economies of administration
and transactions costs lead to the grouping of roughly congruent services at each
level-local (street lighting, refuse removal), regional (rural-urban roads, refuse dis-
posal), and national (intercity highways, environmental policy). Decentralized
decisionmaking enlarges possibilities for local participation in development. In ad-
dition, national allocative objectives may be carried out by local governments re-
sponding to incentives created by national grants and regulations as well as through
interlocal or interregional agreements. National governments have obvious roles with
respect to stabilization and distribution, and due attention must be paid to possible
local conflicts with these policies.
In most cases, it is desirable that the national government assume responsibil-
ity for national public services, international affairs, monetary policy, regulation,
transfers to persons and businesses, fiscal policy coordination, regional equity, re-
distribution (in which all levels of government may play a role), and preservation
of an internal common market. Some central functions, such as regulation of the
financial sector or the environment, may be effectively shared with subnational
governments. State governments may have significant responsibility for educa-
tion, health, social insurance, intermunicipal issues, and oversight of local govern-
ments. (Particularly for the social sectors, however, the central government continues
to play a critical role by ensuring a minimum level of some services for reasons of
externalities or equity.) All local services should be assigned to local governments.
In areas of shared responsibilities, the roles should be clarified. Generally, the cen-
tral government should be involved with overall policy, setting standards, and
auditing; state governments should have an oversight function; and local govern-
ments should be involved in the provision of infrastructure and services.
Assignment of public services to local or regional governments can be based on
considerations such as economies of scale, economies of scope (appropriate bundling
of public services to improve efficiency through information and coordination econo-
mies and enhanced accountability through voter participation and cost recovery),
cost-benefit spillovers, proximity to beneficiaries, consumer preferences, and flex-
ibility in budgetary choices on composition of public spending. Assignment of re-
sponsibilities to various local governments could be asymmetric, based on population
size, rural-urban classification, and fiscal capacity. Thus large cities may have re-
sponsibilities for some services that are provided directly by the center in other areas.
An illustrative representative assignment of expenditure responsibilities is de-
picted in table 2.1, which takes into account the desirability in some instances of
decentralization beyond formal government to civil society. An additional impor-
tant consideration is that accountability is often best promoted by establishing clear
and close linkages between the costs and benefits of public services, so that the
amount of expenditure responsibility assigned to a particular level of government
corresponds to the amount of revenues that level has at its potential command.
Table 2.1 provides guidelines on expenditure assignments based on the
subsidiarity principle. In practice, it is unlikely that any country would or should
precisely follow the divisions set out in the table. In many instances some functions
- 20 -



Project Design: Political, Fiscal, and Administrative Decentralization
Table 2.1. A Representative Assignment of Expenditure Responsibilities
Policy and
standards       Provisionl      Production!
Function                      oversight    administration   distribution         Comments
Interregional and intemational    U                U               N,P        Benefits and costs
conflict resolution                                                         international in scope
External trade                    U              U,NS               P         Benefits and costs
international in scope
Telecommunications               U, N              P                P         National regulation
not feasible
Financial transactions           U,N               P                P         National regulation
not feasible
Environment                     U.NI,SL          U,NSL           NS,L,P       Externalities of
global, national,
state and local scope
Foreign direct investment        N,L               L                P         Local infrastructure
is critical
Defense                           N                N               N,P        Benefits and costs
national in scope
Foreign affairs                   N                N               N          Benefits and costs
national in scope
Monetary policy, currency,      U, ICB            ICB             ICB, P      Independence from all
banking                                                                     levels essential, some
international role for
common discipline
Interstate commerce          Constitution,         N                P         Constitutional
N                                           safeguards important
for factors and goods
mobility
Immigration                      U,N               N                N         U due to forced exit
Transfer payments                 N                N                N         Redistribution
Criminal and dvil law             N                N                N         Rule of law, a national
concem
Industrial policy                 N                N                P         To avoid beggar-thy-
neighbor polices
Regulation                        N               N,SL           N,S,L,P      Intemal common
warket
Fiscal policy                     N               NSL            NS,L,P        Coordination is
possible
Natural resources                 N               N,SL           N,S,L,P      Promotes regional
equity and intemal
common market
Education, health and           N,S,L              SL             S,L,P       Transfers in kind
sodal welfare
Highways                        NS,L              N,S,L           S,L,P       Benefits and costs of
various roads vary in
scope
Parks and recreation            NS,L              N,S,L          N,S,L,P      Benefits and costs of
various roads vary in
scope
Police                           S, L              S,L             S,L        Primarily local benefits
Water, sewer, refuse,             L                L               L,P        Primarily local benefits
fire protection
U     Supranational responsibility.
ICB   Independent central bank.
N     National government.
S     State/provincial government.
L     Local government.
P     Nongovemmental sectors/dvil society.
Source: Anwar Shah, The Reform of Intergovernmental Fiscal Relations in Developing and EmergingMarket Economies (Wash-
ington, D.C.: World Bank, 1994), Table 2.
- 21 -



Decentralization Briefing Notes
should be shared between levels of government; higher levels of government might
exercise a regulatory or policy role, while lower levels of government are respon-
sible for service delivery. Even in service delivery there are aspects best shared
between different levels of government. For example, the technical specifications
for bridge construction might come from a higher level of government, while con-
struction and maintenance are handled at the local level. In health care, the center
may continue to provide technical training for staff, to procure pharmaceuticals to
benefit from economies of scale and ensure quality, and to fund public health ser-
vices. Intermediate levels might supervise local-level personnel, provide refresher
training courses, and together with the local level, decide the appropriate mix of
curative services to offer and ensure adequate maintenance of facilities and satis-
faction with the personnel.
Assigning responsibility for the provision of service to a specific level does not
imply that the same level of government should be directly engaged in its produc-
tion. Collection, transport, and treatment services in solid waste management, for
example, can be assigned to different public and private entities depending on econo-
mies of scale, commercial viability, and extemalities. Many other services can also
be unbundled. Production decisions should result from an evaluation of alterna-
tives using efficiency and equity criteria.
Experience has clearly shown that effective decentralization requires comple-
mentary adaptations in institutional arrangements for intergovernmental coor-
dination, planning, budgeting, financial reporting, and implementation. Such
arrangements may encompass both specific rules (for example, in the design of
fiscal transfers) and provisions for regular intergovernmental meetings and peri-
odic reviews of intergovernmental arrangements. Detailed central control over
local use of funds is seldom appropriate. Instead, what is needed is transparency
and accountability to local constituencies supported by strengthened higher level
monitoring and reporting of local fiscal performance (see section on accountabil-
ity, transparency, and corruption).
- 22 -



Project Design: Political, Fiscal, and Administrative Decentralization
Issues in Tax Assignment
Governments rely on a wide variety of direct, indirect, general, specific, business,
and individual taxes to meet their revenue needs. The question addressed here is
which types of tax instruments are most suitable for use by each level of government.
Principles of Tax Assignment
The assignment of taxes by jurisdiction depends partly on the mix of taxes used in
the country overall. Public finance theory has not yet fully resolved the issue of the
ideal tax mix, even in the unitary state. Governments almost universally employ
balanced tax systems, meaning that different taxes apply to basically the same bases.
For example, general sales taxes, payroll taxes, and income taxes have broadly over-
lapping bases. On efficiency and equity grounds, a single general tax base should
suffice. Yet no government behaves that way, in large part, to minimize tax evasion.
The usual explanation is that administrative considerations play an important
role. A mix of taxes keeps the rate on any individual tax low, thereby reducing the
incentive to evade or avoid the tax. Furthermore, a mix of taxes means that taxpay-
ers who would otherwise be able to avoid taxation of one type are caught in the net
of another, making the tax system fairer. The importance of the various taxes in the
overall mix remains, however, a matter of judgment rather than something that
can be deduced from principles.
These same general considerations apply to assigning taxes in a federal system
of government. Efficiency and equity arguments have to be tempered by adminis-
trative considerations, and the exact assignment depends on informed judgment.
We can, however, outline the economic principles that come into play in deciding
which taxes to assign to lower levels of government.
Efficiency of the Internal Common Market
The internal common market is functioning efficiently if all resources (labor, capi-
tal, goods, and services) are free to move from one region to another without im-
pediments or distortions imposed by policy. Decentralized tax systems can interfere
with the efficiency of the economic union in two ways. First, the uncoordinated
setting of taxes is likely to lead to distortions in markets for resources that are mo-
bile across states or provinces, especially capital and tradable goods. State govern-
ments that recognize this mobility may shape their tax policies in order to attract
resources to their own states. If all jurisdictions engage in such socially wasteful
beggar-thy-neighbor policies, the end result will simply be inefficiently low taxes
(or high subsidies) on mobile factors.
National Equity
The tax transfer system is one of the main instruments of achieving redistributive
equity. The argument for making equity a federal objective is simply that all people
in the country ought to enter into society's "social welfare function" on an equal
basis, and the federal government is presumably the only level that can ensure that
residents in different regions are treated equitably. This view may be tempered if
- 23 -



Decentralization Briefing Notes
states have different tastes for redistribution, or if centralized decisionmaking is
not guided by normative criteria. To the extent that equity is viewed as a federal
policy objective, decentralized taxes can interfere with the achievement of that ob-
jective. As in the efficiency case, uncoordinated subnational tax policies may un-
wittingly induce arbitrary differences in redistributive consequences for residents
of different jurisdictions. States may use both taxes and transfers to attract people
with high incomes and to repel those with low incomes. Such policies are likely to
be offsetting with respect to resource allocation, but they will result in less redistri-
bution than in their absence. (Of course, those who are opposed to redistribution
through government will prefer decentralized policies for precisely that reason.)
This is obviously likely to be more of a problem for taxes that are redistributive in
nature, as well as for transfers.
Administrative Costs
The decentralization of tax administration can also increase the cost of collection
and compliance for both public and private sectors. There are fixed costs associ-
ated with collecting any tax that have to be borne for each type of tax used by the
states. Taxpayers also incur compliance costs for all taxes levied. For some types of
taxes, in particular where the tax base is mobile or straddles more than one juris-
diction, the possibilities for evasion and avoidance increase with decentralization.
In the case of multidistrict tax bases, rules are needed to allocate tax revenues among
jurisdictions, for in their absence, some tax bases may face either double taxation or
no taxation at all. Auditing procedures may also be more difficult for tax bases that
involve transactions across state boundaries.
Fiscal Need
To ensure accountability, revenue means should be matched as closely as possible
at the margin to revenue needs (that is, expenditure responsibility). Thus tax in-
struments intended to further specific policy objectives should be assigned to the
level of government with responsibility for the related service. Progressive redis-
tributive taxes, stabilization instruments, and resource rent taxes would be suitable
for assignment to the national government, while tolls on intermunicipal roads are
suitably assigned to subnational governments. In countries with a federal value
added tax (VAT), subnational sales taxes may be too cumbersome. In such circum-
stances, the fiscal need criterion would suggest allowing subnational governments
access to taxes that are traditionally regarded as more suitable for national admin-
istration, such as personal income taxes.
Reconciling the Difference: Making Subnational Governments
Accountable while Avoiding National Distortions
The main problem with the tax assignment that emerges from the preceding pre-
scriptions, as illustrated in table 2.1, is that it generally does not provide sufficient
revenues for lower-tier governments. In part for this reason, local and especially in-
termediate-level governments in many countries levy a variety of specific (excise)
- 24 -



Project Design: Political, Fiscal, and Administrative Decentralization
taxes on gambling, motor vehicles, and so on. Again, however, such levies seldom
produce the revenue needed to finance a significant part of such major expenditures
as education and health that are often assigned to subnational governments.
There are two main options for resolving the problem:
* The first way is to supplement subnational revenues with intergovernmental
fiscal transfers, but policies must be designed to avoid creating incentives for
local governments to reduce their own tax-collection efforts (see the intergov-
ernmental transfers section).
* The second option is to permit subnational governments to levy their own
broadbased taxes for local goods, as long as they burden local beneficiaries only. In
principle, a retail sales tax or a tax on personal income would be possible. In prac-
tice, however, an efficient, desirable broadbased subnational tax that seems fea-
sible is a flat-rate surtax (often called "piggybacking") on anationalpersonal income
tax. Retail sales taxes are seldom feasible in developing or transitional countries.
Whether with respect to such a surtax, a local property tax, or local taxes in
general, the critical elements required to ensure local accountability without effi-
ciency costs are restricting local governments as much as possible from exporting
taxes and permitting them to set their own tax rates. For efficiency, it may be desir-
able to assess the base of a tax centrally and even to have it collected by the central
government, but for accountability, it is critical that subnational authorities be re-
sponsible (within limits) for setting the tax rate.
The Transition Economies: Redefining Local Powers
A special situation exists in a few transition economies, including those in Eastern
Europe, Central Asia and China, and Vietnam, where subnational governments
have traditionally had a larger role in collecting national taxes. In these countries
the central government may not have full control over its tax bases because of such
local administration. For example, in China and Russia, revenues were collected at
the subnational level (through a tax contracting system in China) and shared up-
ward. This structure created incentives at the subnational level to put more collec-
tion effort into taxes fully retained at that level and less effort into taxes that were
largely transferred upward. Subnational governments preferred to receive trans-
fers in kind or contributions from their own enterprises rather than collect corpo-
rate taxes that had to be shared with higher levels. Revenue sharing on a tax-by-tax
basis led to highly varied levels of efficiency in tax administration because of the
disconnect between tax administration and the user of the proceeds.
Further, in countries with conflict between levels of government, subnational
administration of national taxes is not advisable, because an unhappy subnational
entity can refuse to submit national taxes (for example, Tatarstan in Russia). China
recently strengthened its central tax administration to collect revenues from central
and shared taxes. Problems are also caused by overlapping, uncoordinated admin-
istration, especially for sales and excise taxes.
- 25 -



Decentralization Briefing Notes
The Balance between Enough and Too Much
Decentralization can reduce accountability by breaking the links between the lev-
els of taxation and expenditure, because decentralized expenditures may well ex-
ceed local revenues (notwithstanding user fee revenues). Major expenditure
responsibilities are being transferred to subnational governments in an effort to
improve service delivery, but there are few high-revenue taxes that can be assigned
to subnational governments without creating national economic distortions. Effi-
ciency in tax administration suggests that subnational governments should levy
taxes on imnmobile factors (for example, property taxes) and fiscal need criteria sug-
gest that they should also levy cost-recovery user charges such as frontage taxes
(tax per linear front foot of property), tolls on local roads, and poll taxes. These tax
revenues are unlikely to be sufficient in many localities, and intergovernmental
transfers are required to mitigate this imbalance. While taxation increases can cre-
ate constituent pressure for good local performance, some grant designs can create
central government pressure for local performance.
- 26 -



Project Design: Political, Fiscal, and Administrative Decentralization
Intergovernmental Transfers and Grants
Intergovernmental transfers are the dominant source of revenues for subnational
governments in most developing countries.' The design of these transfers affects
the efficiency and equity of local service provision and the fiscal health of
subnational governments.
Taxonomy of Grants
For the purpose of economic analysis, grants can be broadly classified as
nonmatching and matching.
Nonmatching transfers may be either conditional (selective) or unconditional
(general). Conditional nonmatching transfers offer a set amount of funds without
local fund matching, provided the funds are spent for an established purpose. Such
transfers ensure that the recipient government's spending on the specified category
will be at least equal to the amount of the grant monies. If the recipient is already
spending an amount equal to the grant funds, some or all of the grant funds may
be diverted to other uses. In theory, because of the fungibility of funds, increases in
expenditures on the specified category would only be the marginal difference be-
tween the grant and the prior local expenditures. In practice, because of the lumpi-
ness of investments in areas such as infrastructure as well as other factors,
expenditures often do exceed the amount of grants.
General or unconditional nonmatching grants attach no constraints on how
the grants are spent and, unlike the case with conditional grants, no minimum
expenditure in any area is expected. Because the grant can be spent on any combi-
nation of public goods or services or to provide tax relief to residents, general
nonmatching assistance does not modify relative prices and is the least stimula-
tive of local spending.
Some empirical studies have observed that the increase in local expenditure
due to grant transfers is larger than would have resulted from a similar increase in
the population's income. This is referred to as the flypaper effect: grant money
tends to stick where it first lands.10 The implication is that for political, technical,
and bureaucratic reasons, grants to local governments tend to result in more local
spending than if the same transfers were made directly to local residents.
Matching grants or cost-sharing programs are conditional transfers that require funds
be spent for specific purposes and that the recipient match the funds to some degree.
Such transfers have two effects: an income effect, as the subsidy gives the community
more resources, some of which may go to acquire more of the assisted service, and a
price or substitution effect, as the subsidy reduces the relative price of the assisted ser-
vice, allowing the community to acquire more for a given budget. Both effects stimulate
expenditures on the assisted category. Such transfers can be open-ended (no limit on
9. For a broad set of essays on transfers, see Ahmad Ehtisham, ed., Financing Decentral-
ized Expenditures: An International Comparison of Grants, Studies in Fiscal Federalism and State-
Local Finance (Cheltenham, U.K., and Lyme, New Hampshire: Elgar, 1997).
10. J. Hines and R. Thaler discuss possible reasons for this observable but largely unex-
plained phenomenon in "The Flypaper Effect," Journal of Economic Perspectives 9:4(1995).
- 27 -



Decentralization Briefing Notes
matching funds) or closed-ended. Matching transfers may distort local priorities and be
considered inequitable in that richer jurisdictions can raise matching funds more easily.
But the second effect can be offset, if desired, by adjusting matching rates to jurisdic-
tional wealth. The first effect-distorting local priorities-may be the desired outcome
when the transfer is intended to internalize spillovers or achieve overriding national
policy objectives (for example, family planning programs).
Economic Rationale for Transfers and Implications for Grant Design
There are five broad economic arguments for central-state transfers, each based on
efficiency or equity, and each applying in varying degrees in different countries.
THE FISCAL GAP. An imbalance between the revenue-raising ability of subnational
governments and their expenditure responsibilities (the "vertical imbalance") might
arise for two reasons. First, there may be an (often inappropriate) assignment of
taxing and spending responsibilities that puts the expenditure needs of subnational
governments above their revenue means. Second, many taxes are more efficiently
collected at the central level, to avoid tax competition and interstate tax distortions,
so transfers are necessary to enable local levels to carry out their expenditure re-
sponsibilities.
FISCAL INEQUIIY. A country that values horizontal equity (the equal treatment of
all citizens nationwide) will need to correct the fiscal inequity that naturally arises
in a decentralized system. Subnational governments with their own expenditure
and taxation responsibilities will be able to provide different levels of services to
their residents for the same fiscal effort owing to their differing fiscal capacities. If
desired, these differences may be reduced or eliminated by tying the transfers to
each jurisdiction to relative tax capacity and to the relative need for and cost of
providing public services.
FISCAL INEFFICIENCY. The argument for such transfers is reinforced by the fact that
the same differentials that give rise to fiscal inequity also cause fiscal inefficiency.
The regional differences in levels of service received for a given fiscal effort can
influence business and investment decisions, possibly leading to a less inefficient
allocation of resources.
INTERSTATE SPILLOVERS. Normally, subnational governments will not have the
proper incentive to provide the correct levels of services that yield spillovers across
jurisdictions. In theory, a system of matching grants based on expenditures that
give rise to spillovers will provide the incentive to increase expenditures. In prac-
tice, the extent of the spillover will be difficult to measure, so the matching rate
selected will be somewhat arbitrary.
FISCAL HARMONIZATION. To the extent that redistribution is a goal of the central
government, there is a national interest in the redistribution that occurs as a re-
sult of the provision of public services by subnational governments. Expenditure
- 28 -



Project Design: Political, Fiscal, and Administrative Decentralization
harmonization can be accomplished by the use of (nonmatching) conditional
grants, provided that the conditions reflect national efficiency and equity con-
cerns and that a financial penalty is associated with failure to comply. In choos-
ing such policies there will always be a tradeoff between uniformity, which may
encourage the free flow of goods and factors, and decentralization, which may
encourage innovation, efficiency, and accountability.
The most appropriate form of a transfer depends in large part on its objective
(box 2.1). Regardless of the design, however, experience demonstrates that good
intergovernmental transfer programs have certain characteristics in common:
* Transfers are determined as objectively and openly as possible, ideally by some
well-established formula. They are not subject to hidden political negotiation.
The transfer system maybe decided by the central government alone, by a quasi-
independent expert body (a grants commission), or by some formal system of
central-local committees.
* Transfers are relatively stable from year to year to permit rational subnational
budgeting, but at the same time they are sufficiently flexible to ensure that na-
tional stabilization objectives are not thwarted by subnational finances. One
system that appears to achieve this dual objective is to set the total level of
transfers as a fixed proportion of total central revenues, subject to periodic re-
negotiation (say, every three to five years).
* The formulas are transparent, are based on credible factors, and are as simple as
possible. Complex formulas are unlikely to prove feasible or credible in devel-
oping countries because there are often serious disputes on fundamental issues
such as regional population size.
* If several objectives are applicable-for example, if some degree of equalization
is desired and if there are also clear national policy objectives, on the provision
of minimal standards of education, but perhaps variable degrees of national
support for certain local infrastructure activities-having separate transfers tar-
geted to each objective will generally assist with both clarity and effectiveness.
Revenue Sharing
Many countries attempt to achieve the objectives of transfers through "tax shar--
ing" or 'revenue sharing" systems. While there are many such systems, most of
them-perhaps most markedly in the transition economies-suffer from several
common problems. First, if they apply only to a subset of national taxes, they may
bias national tax policy. Second, if, as is often the case, the systems share revenues
from origin-based (production) taxes to the jurisdictions from which the revenues
are collected, they break the desirable link between benefits and costs at the local
level and hence reduce accountability and the efficiency of decentralization. Third,
because tax rates in such systems are invariably set by the central government, ancl
because the sharing rate is often applied uniformly throughout the country, the
accountability link is broken and subnational governments have no incentive to
ensure that the amount and pattern of their spending are efficient. In addition, if, as
- 29 -



Decentralization Briefing Notes
Box 2.1.  Principles and Better Practices in Grant Design
Grants are one of several tools available to bridge the fiscal gap between subnational
responsibilities and revenues. For other possible methods, such as reassignment of
responsibilities, tax abatement, or tax-base sharing between levels of government,
see notes on expenditure assigmnent and tax assignment.
Grant objective       Grant design       Better practices   Practices to avoid
To reduce regional   * General        * Fiscal equalization  * General revenue
fiscal disparities  nonmatching       programs of         sharing with
* Fiscal capadty     Australia, Canada,   multiple factors
equalization       and Germany
transfers
To compensate for   * Open-ended      * RSA grant for
benefit spillovers    matching transfers   teaching hospitals
with matching rate
consistent with
spillout of benefits
To set national    * Conditional      * Indonesia roads    * Conditional
minimum            nonmatching block   and primary        transfers with
standards          transfers with     education grants    conditions on
conditions on    * Chile, Colomnbia,   spending alone
standards of service   and South Africa   * Ad hoc grants
and access         education transfers
To influence local    * Open-ended    * Matching transfers  * Ad hoc grants
priorities in areas    matching transfers   for social assistance
of high national   (preferably with   as in Canada
but low local      matching rate to
priority           vary inversely with
fiscal capacity)
To stabilize the   * Capital grants   * Limit use of capital * Stabilization grants
economy            provided           grants and          with no future
maintenance        encourage private    upkeep
possible           sector participation   requirements
by providing
political and policy
risk guarantee
Source: World Bank, World Development Report (New York: Oxford Press, 1997), Box 7.4.
in some transitional economies, such taxes are collected by local governments and
then shared with national governments, in this case either an undesirable disincen-
tive for collection effort is created or, more usually, the temptation to "cook the
books" is likely to be overwhelming (perhaps especially if richer areas are assigned
higher sharing rates).
Practical guidance on the design of transfers is summarized in box 2.2.
- 30 -



Project Design: Political, Fiscal, and Administrative Decentralization
Box 2.2.  Criteria for the Design of Intergovernmental Fiscal
Arrangements
The criteria specified here may conflict with each other, and policymakers designing
grants may have to assign priorities to various factors in comparing policy alternatives.
* Autonomy-Subnational governments should have complete independence and
flexibility in setting priorities and should not be constrained by the categorical
structure of programs or uncertainty associated with decisionmaking at the cen-
ter. Tax-base sharing, allowing subnational governments to introduce their own
tax rates on central bases, formula-based revenue sharing, or block grants, is con-
sistent with this objective.
* Revenue adequacy-Subnational governments should have adequate revenues
to discharge designated responsibilities.
* Equity-Allocated funds should vary directly with fiscal need and inversely with
the taxable capacity of each province.
* Predictability-The grant mechanism should ensure predictability of subnational
governments' shares by publishing five-year projections of funding availability.
* Efficiency-The grant design should be neutral with respect to subnational gov-
ernment choices of resource allocation to different sectors or different types of
activity. The current system of transfers in countries such as Indonesia and Sri
Lanka to finance lower-level public sector wages contravenes this criterion.
* Simplicity-The subnational government's allocation should be based on objec-
tive factors over which individual units have little control. The formula should
be easy to comprehend so that "grantsmanship" is not rewarded, as appears to
occur with plan assistance in India and Pakistan.
* Incentive-The proposed design should provide incentives for sound fiscal man-
agement and discourage inefficient practices. There should be no specific trans-
fers to finance the deficits of subnational governments.
* Safeguarding grantor's objectives-The grant design should ensure that certain
well-defined objectives of the grantor are adhered to by grant recipients. This is
accomplished by proper monitoring, joint progress reviews, and technical assis-
tance or by the use of a selective matching transfer program.
Source: Anwar Shah, The Reform of Intergovernmental Fiscal Relations in Developing and Emerging
Market Economies, Policy and Research Series 23 (Washington, D.C.: World Bank, 1994).
-31 -



Decentralization Briefing Notes
Decentralizing Borrowing Powers
The debt crisis of subnational governments in Brazil, the inflationary impact of
subnational financing in Argentina, and city-level bankruptcies in the United States
have often been used to illustrate the possible macro concerns of decentralizing
borrowing powers. The argument centers on a moral hazard problem, namely, the
proposition that access to financial markets by subnational governments may in-
volve implicit or explicit central government guarantees that lead to imprudent
action by lenders and subnational governments, and ultimately create unplanned
fiscal liabilities for central government.
The academic literature and country experiences, however, do not suggest an a
priori adverse link between decentralizingborrowingpowers and central government's
ability to maintain fiscal discipline and macroeconomic stability. Rather, the key seems
to lie in the design of fiscal decentralization in general and, in particular, the design of
the regulatory framework under which borrowing powers are decentralized.11 The
difficulty in designing and implementing this framework should not be underesti-
mated. Nevertheless, where the necessary preconditions are in place, subnational bor-
rowing is both feasible and desirable. Where it is not in place, borrowing should be
restricted and institutions developed to support it in the longer run.
Why Is Access to Financial Markets Important?
There are three primary reasons why access to financial markets is considered im-
portant for subnational governments.
* Financing capital expenditure-Subnational governments often have respon-
sibility for public investments that are lumpy in nature. Financing such capital
investment through increases in current taxes would be inefficient. In addition,
because the benefits of such public investments often last over several decades,
equity considerations would suggest that future generations participate in the
financing. Capital markets provide this inter-temporal link.
* Matching expenditure and tax flows-Within a particular fiscal year, expendi-
tures incurred and tax intake may not be fully sychronized. Access to financial
markets offers an opportunity to smooth out these mismatches.
* Fostering political accountability-The pricing of capital by markets may pro-
vide an independent mechanism to foster political accountability-markets may
signal the poor performance of subnational governments through increases in
interest rates or simply by blocking access.
11. For empirical evidence of the link between decentralization and subnational deficits
see F. Fomasari, S. Webb, and H. Zou, "Decentralized Spending and Central Government
Deficits: International Evidence," draft Discussion Paper (Washington, D.C.: World Bank,
1998). David Wildasin, ed., Fiscal Aspects of Evolving Federations (Cambridge, U.K.: Cambridge
University Press, 1997) and Jennie Litvack, J. Ahmad, and R. Bird, "Rethinking Decentraliza-
tion in Developing Countries," Discussion Paper (Washington, D.C.: World Bank, 1998) sug-
gest that this linkage is directly related to the design of intergovernmental fiscal relations.
- 32 -



Project Design: Political, Fiscal, and Administrabve Decentralization
What Are the Mechanisms for Accessing Capital Markets
by Subnational Tiers of Government?
There are at least four channels through which subnational governments can ac-
cess capital markets:"2
* Direct borrowing by central government and on-lending to subnational tiers
* Through a public intermediary such as a financial parastatal
* Direct borrowing from capital markets
* Through market decentralization of public services, where possible.
In ascertaining the merits of one channel over the other, policymakers need to
consider several factors. The channel selected should minimize or eliminate the
inefficient political allocation of credit. Any implicit liability by an upper-tier gov-
ernment should be explicitly recognized and reduced or eliminated. Finally, selec-
tion should be made with a view toward strengthening capital markets.
While borrowing through the central government certainly guarantees access
by subnational governments to long-term finance, international experience sug-
gests that the allocation of credit through this manner may eventually get embroiled
in a political process.'3 Under such conditions, capital does not necessarily flow to
the most productive, but those who are politically the most astute, with the result
that government borrowing is inefficient and subsequent investments are unprc-
ductive. Intermediation by a public financial intermediary may also suffer fronm
these drawbacks, with the additional disadvantage that the debt of the public fi-
nancial intermediaries is an implicit obligation of the central government.
Additional complications may occur when multiple channels are used. For ex-
ample, in South Africa, the government has created a public infrastructure bank
that operates alongside a well-developed private capital market.'4 Unless the regu-
latory framework for municipal borrowing clearly delineates a complementary role
for the public and private financial system, there may be a risk that the former
either displaces the latter or that the public sector is saddled with the high risk, lowv
creditworthy clients. Finally, experience suggests that central government control
of borrowing creates the incentives for subnational governments to elude these
12. For an alternative taxonomy based on different forms of controls on subnational
borrowing, see T. Ter-Minassin and J. Craig, "Control of Subnational Government Borrow-
ing,' in T. Ter-Minassin, ed., Fiscal Federalism in Theory and Practice (Washington, D.C.: Inter-
national Monetary Fund, 1997).
13. World Bank, "Financial Systems and Development," Policy and Research Series no.
15 (Washington, D.C.: World Bank, 1990).
14. For more information, see Junaid Ahmad, "Structure of Urban Governance in South
African Cities," International Taxation and Public Finance 3:2 (1996) and Chris Heymarns,
"Parastatals and Development Finance," in Patrick FitzGerald, Anne McLennan, and Barry
Munslow, eds., Managing Sustainable Development in South Africa (New York: Oxford Univer-
sity Press, 1995), pp. 432-57.
- 33 -



Decentralization Briefing Notes
restrictions through innovative off-budget schemes."5 Centralization of borrowing
powers may, therefore, not necessarily increase the ability of central government to
control subnational liabilities.
Direct access to capital markets, by contrast, offers the potential for a more trans-
parent, market-based relationship to develop and for a greater chance of enforcing
a hard budget constraint (see box 2.3). Increasingly, an important aspect of market-
based systems includes market decentralization to finance infrastructure. This in-
cludes private sector participation in the financing and delivery of government
services, corporatization, and unbundling of service delivery."6 In countries with
limited local capital markets, market decentralization may enable access to inter-
national capital markets directly by the service provider, thereby facilitating the
deepening of capital markets.
The macro concern of moral hazard-the assumption by capital markets that
borrowing by subnational governments is ultimately backed by the central govern-
ment-may also apply in the case of direct borrowing from private financial mar-
kets. Managing such risks through an appropriate regulatory framework while
allowing market discipline to guide the allocation of capital may be the most effi-
cient mechanism for allowing subnational governments access to financial markets.
How Should the Regulatory Framework for Subnational
Governments Be Designed?"7
A well-designed regulatory framework is necessary to ensure that decentralizing
borrowing does not provide perverse incentives for excessive lending by markets
and excessive borrowing by subnational governments, excesses that eventually may
end up as liabilities of central authorities.
Such a framework requires transparency, preferably through better information
systems. This includes standardized accounting procedures for subnational govern-
ments and better public information of their liabilities and their repayment capacity.
However, more public information and better disclosure by itself will not be suffi-
cient to curb moral hazard problems. Penalties would need to be associated with
excessive borrowing. In particular, explicit bankruptcy procedures, for instance, the
U.S. type of financial control boards or the New Zealand system of court-appointed
receiverships, may need to be legislated.'8 A key in designing such processes is to
15. See T. Ter-Minassin and J. Craig, "Control of Subnational Government Borrowing,"
in T. Ter-Minassin, ed., Fiscal Federalism in Theory and Practice (Washington, D.C.: Interna-
tional Monetary Fund, 1997).
16. For example, organizing the delivery of water or electricity in a company structure-
whether a private or public corporation-may improve access to capital markets if such
organizations are better able to shield themselves from political interference than subnational
governments.
17. The discussion in this section focuses on the demand side of the market: regulation
of borrowing by the public sector. It does not discuss the regulatory aspects of the supply
side: the framework for regulating the financial system itself.
18. In addition to bankruptcy procedures, the central government could also legislate
intermediary monitoring steps with an associated penalty schedule, for example, legislating
debt thresholds and penalties for crossing them, such as requiring central approval for
- 34 -



Project Design: Political, Fiscal, and Administrative Decentralization
Box 2.3. Evolution of Capital Markets and Decentralization
of Borrowing Powers
In countries without a domestic capital market, it may be preferable to have central
fiscal transfers (from central government borrowing from international sources) pro-
vide additional resources to subnational governments, while focusing central gov-
ernment policy efforts on developing a private commercial banking system. This
approach avoids the creation of public financial intermediaries, keeps the fiscal and
financial sectors separate, and potentially offers a better starting point for a munici-
pal finance system. The reform of the financial sector in Eastern Europe is a case in
point. As the private banking system emerges and establishes itself, long-term fi-
nance can be provided through discount facilities funded by the central govern-
ment. In this system, the discount facility only takes the maturity risk by stretching
the term of commercial bank lending to municipalities. The commercial banks, how-
ever, continue to retain the credit risk for their lending at the retail level, thereby
preserving the separation of risks between the public and private sector. Colombia
provides an interesting example of such an arrangement.
As capital markets emerge with financial instruments offering long-term fi-
nance, the need for a discount facility would disappear, and as long as the fiscal
system was well designed, subnational governments would now be able to access
longer maturities. Even in this case, however, for fiscally weaker municipalities, for
instance, local governments, the transfer system would remain an important ve-
hicle for access-to funds. In addition, in relatively more developed capital markets,
the use of private credit rating agencies and bond insurance agencies offers a mar-
ket-based mechanism to monitor and regulate subnational borrowing, but the pub-
lic sector also needs to monitor and regulate these institutions.
ensure that the bankruptcy mechanism adopted enables the delivery of basic ser-
vices to continue, even if on a reduced level, while management and financial re-
structuring are proceeding, otherwise the pressure for central bailouts will be great.
Ensuring that subnational governments have access to their own sources of rev-
enues that can be pledged as collateral is also necessary to reduce possible moral
hazard. Without such direct fiscal backing, markets might view any capital market
borrowing by subnational governnents as implicitly backed by the central govern-
ment. Having their own fiscal base is therefore an important prerequisite for
subnational governments' access to finance and for limiting moral hazard problems.
Equally important in this context is keeping the fiscal and financial systems
separate in the design of intergovernmental systems. Indeed, one can hypoth-
esize that it is the lack of separation between the fiscal and financial systems that
future borrowing or even the loss of borrowing powers altogether. There is a danger that too
many centrally imposed controls may, in fact, make central authorities implicitly liable for
subnational borrowing. For this reason, simply requiring full disclosure and legislate bank--
ruptcy procedures without introducing intermediate controls may be preferable.
- 35 -



Decentralization Briefing Notes
has led to macro instability in some countries-as the web of implicit obligations
suddenly appears as explicit budget commitments-rather than the decentrali-
zation process itself.
For example, in Argentina, provincial government deficits have been financed
in part by provincial banks, many of which have gone bankrupt. As Wildasin points
out, the central bank's policy of managing these banks and absorbing their losses
provided provincial governments with a circuitous mechanism of inflationary fi-
nance.i9 It also weakened incentives for fiscal discipline at the provincial level. In
the case of Brazil, evidence suggests that a soft budget constraint between tiers of
government can exist through the banking system notwithstanding a sound inter-
governmental fiscal system. In commenting on the situation in Brazil, Dillinger
points out that the hard budget constraint imposed on the fiscal side was circum-
vented as states have-wittingly or not-used the financial system to obtain fed-
eral resources.20 Similarly, in China the weak revenue base of the center has created
political pressures on the People's Bank of China to offer credit to lower-level gov-
ernments. As Lall and Hoffman argue, such "policy lending" can have a negative
impact on macroeconomic price stability and, presumably, on fiscal discipline at
the subnational level.21
These examples suggest both a cautious approach and one that emphasizes the
separation of the fiscal and financial systems. Several mechanisms are possible.
First, the independence of the central bank is a key element of the hard budget
constraint. Second, where possible, the public sector must get out of the financial
system. This includes the privatization of the banking system, but privatization is
by itself not sufficient to enforce the separation. As the U.S. savings and loan crisis
and elements of the recent Asian financial crisis suggest, private banks need to be
supervised with a clear prudential, regulatory framework that monitors the level
and nature of their aggregate liabilities. Finally, one lesson from worldwide experi-
ences is clear: subnational public financial institutions are not desirable. The incen-
tives for subnational governments to abuse their relationship with their own
financial institutions at the expense of macrostability is too great.2' The relation-
ship between provincial governments and provincial banks in Argentina stands
out as a classic example of this situation.
Market decentralization may also offer an interesting mechanism for reduc-
ing moral hazard. Privatization of public infrastructure may enable the applica-
tion of private sector bankruptcy laws and allow other private sector companies
19. David Wildasin, ed., Fiscal Aspects of Evolving Federations (Cambridge, U.K.: Cam-
bridge University Press, 1997).
20. William Dillinger, "Brazil's State Debt Crisis: Lessons Learned," Country Depart-
ment 1, Latin America and the Caribbean Region Economic Notes (Washington, D.C.: World
Bank, 1997).
21. R. Lall and B. Hoffman, "Decentralization and Governent Deficits in China," in J.
Roy, ed., Macroeconomic Management and Fiscal Decentralization (Washington, D.C.: World
Bank, 1995).
22. Of course, simnilar risks exist with national financial institutions. That is why the first
option of getting the public sector out of the banking business may be the preferable policy
route to follow.
- 36 -



Project Design: Political, Fiscal, and Administrative Decentralization
to bid for the assets in case of financial difficulties, an option unavailable with
public ownership. To implement such an option, however, governments need to
ensure that there are multiple deliverers in any one sector. Private participation
creates options for accessing financial equity in addition to debt creating an in-
centive for "equity to monitor debt" that is not available in public financing sys-
tems. In addition, unbundling may reduce the size of the entity concerned, thereby
avoiding the "too big to fail" syndrome that is often associated with bailouts.
Finally, the back-door channels to creating financial liabilities need to be moni-
tored, and where possible, closed. In particular, legislation must ensure that dip-
ping into pension funds or using subnational corporations to borrow on behalf of
subnational governments is not permitted or is explicitly included in subnational
debt. In addition, balanced budget requirements for subnational governments may
ensure that current accounts are balanced by the end of each fiscal year, so that
borrowing to match expenditures and revenue streams does not lead to financing
of current account deficits over time.
In sum, the combined use of information systems, access to own revenues, public
legislation, bankruptcy laws, and market decentralization offers an overall institu-
tional incentive for a hard budget constraint at all tiers of government and permits
borrowing to be decentralized.
Should Foreign Borrowing Be Allowed?
The issue of direct access to international capital markets by subnational govern-
ments is further complicated by issues of macroeconomic management, general
capital account liberalization, and the nature of the foreign exchange regime. This
overall context will determine whether direct borrowing by subnational govern-
ments in intemational markets should be permitted. Given the limited consensus
on this issue, there seems to be a bias toward not granting direct access to interna-
tional markets by subnational govemments. In any case, the issues raised in this
note about the regulatory framework for decentralizing borrowing apply equally
to local and foreign borrowing.
Conclusion: Implementing Financial Decentralization
Financial decentralization-allowing subnational governments to access capital
markets directly-is an important complement to the devolution of fiscal powers
to regional and local authorities. If properly designed, decentralization of borrow-
ing powers can add to the gains in efficiency and govemance expected from fiscal
decentralization. Implementing financial decentralization, therefore, should receive
equal priority in the design of intergovermmental fiscal relations.
However, as this note suggests, for financial decentralization to be properly
implemented, the regulatory framework will need to be well designed. This cau-
tion does not imply an all or nothing scenario: a phased and asymmetric approach
may be adopted to implement financial decentralization. Policies may include
sequencing fiscal decentralization prior to allowing subnational governments
access to capital markets; enabling larger urban centers, which are usually more
self-reliant in terms of fiscal resources and management capacity than other local
- 37 -



Decentralization Briefing Notes
governments, quicker access to financial markets and, perhaps, even controlled
and limited access to foreign markets; implementing market decentralization as
a step toward financial decentralization; and replacing the implicit resource trans-
fers through public financial institutions with explicit fiscal transfers. In sum, the
challenge is to design the regulatory framework for financial decentralization
and its process of implementation. Debating whether financial decentralization
is good or bad may be a moot point in an era of emerging local capital markets
and a greater political push in many countries to decentralize powers to
subnational governments.
- 38 -



Project Design: Political, Fiscal, and Administrative Decentralization
Civil Service Reform and Decentralization
Decentralization can be a way of improving access to services, tailoring government
actions to private needs, and increasing the opportunities for state-society interactions.
Subnational governments will only be effective, however, when they have access to the
human and financial resources needed to undertake the services assigned to them.
Civil service reform-both capacity building and adjusting to decentraliza-
tion-addresses the first of these requirements. There is fairly widespread agree-
ment that capacity building at all government levels is an essential component of
decentralization. There is less agreement over how to deliver the appropriate
human resources package to the appropriate levels of government and how to
coordinate human resources management across and between levels of govern-
ment. Adjusting to decentralization often requires organizational reforms in the
civil service in order to create appropriate incentives for workers at all levels. At
the center of this debate about human resources management and incentive struc-
tures is the discussion of what level of government should have the power to
determine pay scales, terms of service, and civil service structures. Policymakers
must respond to the demands of various constituencies-civil servants, unions,
service users, for example-as well as balance the objective of allowing subnational
governments to adapt to local conditions with the objective of maintaining na-
tional standards and an equitable distribution of human resources.
How Does Decentralization Affect the Civil Service?
Decentralization creates a new set of opportunities for increasing the information
that civil servants have to work with as well as strengthening their incentives to
perform well. Closer contact with their customers gives bureaucrats the tools to
tailor their activities to local conditions and preferences, while increased account-
ability to local politicians and their (vigilant) constituents creates the incentives
for them to do so.
Adjusting to decentralization and taking advantage of these opportunities can
be difficult. In the short term, the mechanics of rearranging the civil service-redis-
tributing personnel, rewriting job descriptions, and so on-can be daunting. In the
longer term, creating the ability and incentives for subnational governments to
surpass minimum national standards for service delivery presents several chal-
lenges. First, new responsibilities fall on people with varying experience and skills.
Economies of scale, such as the number of people that one skilled engineer serves,
can be lost as technical duties are devolved to more and smaller jurisdictions. Sec-
ond, decentralization can also exacerbate inequality: variations in living conditions
(rural areas versus urban centers, for example) can lead to a skewed distribution of
human resources as the best civil servants choose to live in the more developed
areas. Also, some regions may be better able to attract the best candidates with
higher pay. Third, local politicians and civil servants gain more power to dispute
central mandates. Although this local flexibility is often seen as a benefit of decen-
tralization, there is no guarantee that local input will be better for service delivery
than nationwide policies. Civil service reform in decentralizing countries should
- 39 -



Decentralization Briefing Notes
involve not only the building of human resources, but also institutions to ensure
use and equitable distribution of these resources.
RiSING TO THE CHALLENGES. Decentralization requires rearranging, and occasion-
ally expanding, the bureaucracy. Local governments need trained, competent people
in order to carry out new responsibilities, while central governments need to shift
the skills mix of their workers as their role changes (in most cases) from implementor
to technical consultant and overall coordinator. Several factors prevent simply re-
distributing the human resources and skills present in a centralized system across
tiers of government and subnational jurisdictions.
* There may not be enough skilled people to go around. Pre-existing shortages of
human resources can be exacerbated when responsibilities are divided up among
smaller jurisdictions.
* The skilled people may not want to relocate where they are needed. Certain
areas of the country are inevitably more pleasant to live in than others and are
thus better able to attract skilled workers.
The degree of local capacity determines the kind of human resources manage-
ment strategies that will be feasible and desirable. Decentralization of human re-
sources management is more likely to succeed where lower-level authorities have
the financial and managerial ability to set competitive compensation packages and
salary levels that will attract local talent. In these cases, the flexibility advantages of
allowing local governments to set hiring levels might outweigh the risk of increas-
ing interregional inequalities. Where talent and skills are lacking at the local level,
a unitary hiring system might be preferred to ensure that the necessary skills are
present locally in all regions. In these cases, where the center retains more control
over human resources, care should be taken to ensure that the management op-
tions of local stakeholders are not curtailed. Also, care must be taken to ensure the
national standards do not place unrealistic financial burdens on local governments.
SHORT-TERM MECHANICS: DEFNING THE GOAL. All levels of government should be
encouraged to define and plan for the types of workers they will need to carry out
new responsibilities. In the short term these sorts of rough plans substitute for the
human resources management staff and computerized personnel system that so
many countries lack and can help eliminate duplicate workers, unnecessary hires,
and other expensive mistakes. At the very least, such plans can be an exercise in
longer-term planning and role definition.
BUILDING LOCAL CAPACrrY. Local (or at least subnational) capacity is one of the most
important factors in creating a well-functioning, decentralized civil service. The start-
ing point for reforms varies from country to country. In countries where local govern-
ment institutions are embryonic or exist only at an informal level, the institutional and
legal framework will have to be created before any administrative reform or training
is undertaken. (See the section on local technical and managerial capacity.)
-40 -



Project Design: Political, Fiscal, and Administrative Decentralization
In countries where local institutions already exist, the challenge is to reinforce
their technical and managerial capacity. The more experienced central government
has an important role in coordinating and funding training through universities
and seminars. Joint projects, in which more experienced central government civil
servants work with local officials to transfer skills through practice, can also be an
effective way to train subnational civil servants.- More than training, however,
incentives for self-improvement can have an important impact on local govern-
ment performance. In Colombia, for example, accountability to constituents pushed
local mayors to concentrate more on training and hiring effective civil servants.24 In
Northeast Brazil, community oversight and fear of job loss helped motivate civil
servants to comply with work regulations and do the best job possible.15
In more advanced countries, the primnary challenge is to use existing resources
efficiently and preserve economies of scale. The need for specialized personnel is re-
lated in part to the size of the territory covered by the entity and it might be counter-
productive or cost inefficient to have specialists or technical personnel in jurisdictions
below a certain size. One way to address this problem is to allow local governments to
form associations and pool their resources to cover activities requiring specialized
personnel. Another is to create a central consultancy service that the local govern-
ments can use as a source of technical and managerial advice when needed.
PREvENTrG INEQUALuIEs. Decentralization creates more opportunities for local aui-
tonomy and responsiveness to more specialized constituencies, but it also gives
subnational governments more room to fail if specific steps are not taken to ensure a
nationwide minimum level of local technical and managerial capacity. Devolution of
power to subnational governments highlights variations in regions' ability to pay an
adequate number of civil servants and attract skilled people who are capable of imple-
menting the budget efficiently. The poorer regions often lose these competitions, for
their budgets do not allow high salaries, and their level of infrastructure and services
do not create pleasant living conditions. As more skilled and experienced civil ser-
vants migrate to comparatively well-off areas, the gap between regions can expan.d.
Countries have addressed these interregional variations in a variety of ways.
India, for example, has created a cadre system in which highly skilled civil servarnts
rotate between more and less developed regions. Ethiopia takes a less institutional-
ized approach: the Prime Minister's Office sends skilled civil servants from the
central government or more developed regions to the less developed areas of the
23. The international donor community can also get involved in capacity building with
these sorts of joint projects. Creating more opportunities for interaction between local civil
servants and international consultants can transfer much needed technical skills.
24. Tim E. J. Campbell, "Innovations and Risk Taking: The Engine of Reform in Local
Government in Latin America and the Caribbean," Discussion Paper no. 357 (Washington,
D.C.: World Bank, 1997).
25. Judith Tendler, Good Government in the Tropics (Baltimore, Maryland: The Johns
Hopkins University Press, 1997). The state government implemented a community moni-
toring system whereby candidates who had applied but not gotten civil service jobs were
supposed to watch those who had been hired and report infractions such as absenteeism
and lack of responsiveness to the community.
-41 -



Decentralization Briefing Notes
country for one to two years. Another possible option for addressing inequalities is
to emphasize capacity building of civil servants from poorer areas.
In general, some room for variation in conditions of service, pay scales, and so
on allows subnational governments the flexibility to hire a civil service that matches
a community's needs and budget constraints. National salary, eligibility, and per-
formance standards can ensure consistent quality, but they can also lead to person-
nel expenditures (especially for locally administered education and health sectors)
that exceed some local capacities.26 Flexibility also allows regions to develop unique,
cost-effective ways of attracting qualified civil servants: one region might attract
teachers, for example, by providing housing near schools. This could be built with
community labor and would place less strain on the budget than a high salary scale
would. On the other hand, decentralization of the civil service might magnify the
effects of regional budget constraints-a poorer region might be forced to hire fewer
civil servants than another. Allowing regional variation in civil service codes also
raises the possibility of bidding wars for the most skilled civil servants.
PRESERVING NATIONAL POUCES WHILE TAXING ADVANTAGE OF NEW INFORMATION. Increased
citizen control and augmented information for policymakers are two of the most
commonly recognized benefits of decentralization. The challenge is to balance these
bottom-up pressures with some sort of accountability to higher levels of goverm-nent
in order to ensure that local policies are compatible with larger-scale national efforts.
Decentralization, especially political decentralization, creates a class of govern-
ment workers that, based on the specific information that they receive (feedback
from their clients and constituencies), may have different preferences than workers
at the next higher level. This divergence in views and convictions can create con-
flict within the civil service that will require mechanisms to manage effectively.
Benchmarks, essentially lists of responsibilities or standards to meet, can effectively
balance these bottom-up influences.
Clear paths for information flow are the first step in ensuring coordination be-
tween different levels of government and citizens. Any reform effort to ensure sound
utilization of public resources and minimize corruption should include the creation
of a strong legal framework to address issues related to financing and reporting, to
determine what type of control mechanisms (especially financial) are necessary and
who is accountable for them, to evaluate hiring practices and compensation schemes,
and to address issues related to the procurement of public works. Clear reporting
procedures are needed from lower to higher levels of government (central govern-
ment, in the case of regional administrations, for example) and across govermnent
agencies at the same level. In the medium and long term, audit courts can be a useful
regulatory mechanism. Transitions have to be carefully planned to avoid conflict be-
tween new reporting arrangements and the mechanisms of the previous system.
26. Grant transfer systems (see section on intergovermmental grants) will need to take
different financing capacities into account in these and other types of mandated expenditures.
Indonesia, Pakistan, and the Philippines are examples of decentralized states with essentially
uniform terms and conditions of service for government employees in different regions.
- 42 -



Project Design: Political, Fiscal, and Administrative Decentralization
Information and Monitoring in Decentralized Systems
Accountability is a prerequisite for improving public sector performance, and in-
formation is the key to accountability. The systematic collection, analysis, and re-
porting of information are critical elements of decentralization programs because
information can be used to verify compliance with policy goals, to analyze alterna-
tive outcomes, and to guide future decisions. Information on financial flows (that
is, budgeting and expenditure reporting) and other inputs, outputs, and outcomes
is essential both at the local level-to inform local constituents and to encourage
public participation in the political process-and at the central level-to monitor
and supervise local activities funded (at least partially) by central sources.
Monitoring Mechanisms
Unless the public knows what public goods and services are provided, how well
they are provided, who the beneficiaries are, how much the goods and services
cost, and who paid for them, local constituencies will not encourage effective gov-
ernment. Central monitoring and evaluation of local performance has much the
same effect, except that national interests replace narrower local interests. Without
some central monitoring, there can be no assurance that functions of national im-
portance are adequately performed once they have been decentralized, that the
macroeconomic implications of decentralization are understood, or that the effects
of proposed changes in intergovernmental fiscal or administrative relations are
adequately analyzed.
Many decentralizing countries have weak or inadequate mechanisms for
citizens and higher levels of government to monitor, evaluate, and support de-
centralization. While this does not prevent decentralization from achieving some
of its goals, it does limit its ability to create large efficiency gains. Improve-
ments in financial accounting and reporting and the establishment of analytical
capabilities for monitoring and evaluation can strengthen monitoring and as-
sessment of subnational finances.
But the need for careful monitoring goes beyond finance. It depends on the
service delivery objective. For example, different aspects of decentralization may
have different effects on the construction and maintenance of various types of in-
frastructure, health care programs, or education system. If the objective is for safety
nets to reach the poor, information is required on who the poor are and where they
are located, and how much of the benefits from the program are reaching this tar-
get group. In most cases, countries spend significant resources on safety nets but
fail to collect data and monitor who receives the benefits and how they were af-
fected by them. (A notable exception is the valuable evaluation of the poverty tar-
geting of an Argentina safety net program.)27
27. Martin Ravallion, "Reaching Poor Areas in a Federal System," Policy Research Work-
ing Paper 1901 (Washington, D.C.: World Bank, 1998).
- 43 -



Decentralization Briefing Notes
Financial Accountability and Reporting
To improve financial accounting and reporting, detailed fiscal data should be regu-
larly collected and reported for subnational goverrunents. Ideally, these data would
be derived from uniform financial and reporting systems. At a minimum, these
data should be collected and processed on a regular and timely basis and should
exhibit the following characteristics:
* Breadth of coverage-Information should be broadly representative, across units
of subnational government and across financial classifications (revenues, ex-
penditures, debt).
* Consistency-Information should have minimal reporting and classifica-
tion errors.
* Comparability-The same types of activities should be reflected across units of
government.
* Clarity-Actual activities should be reported rather than budgeted activities,
and ideally, data should be audited.
The development and imnplementation of financial reporting and information
systems often require substantial technical assistance, training, time, and resources.
Implementation of these systems may also require that central institutions be es-
tablished to develop and maintain the reporting systems, to train and support local
officials, and to monitor and analyze developments in subnational finance.
Establishing a census of governments-similar in coverage to censuses of popu-
lation, industry, and employment in many countries-and publishing subnational
financial data would establish the foundation for a monitoring system and provide
important inputs for revenue estimating, economic research, and assessments of
creditworthiness. The U.S. Advisory Commission on Intergovernmental Relations,
for example, compiled comprehensive data on subnational finance and institutions
from U.S. census data, surveys of local government professional associations, and
other sources and published them annually in Significant Features of Fiscal Federal-
ism. Reductions in federal funding led to discontinuation of the publication in 1995.
It is difficult to monitor decentralization consistently across countries. For ex-
ample, fiscal information such as that reported in the IMF's Government Finance
Statistics is often used to track trends in decentralization. The problem with these
data, however, is that analysts often use the percentage of total public expenditures
undertaken at the local level as an indicator of decentralization. In some countries,
local governments decide on the allocation of these resources, but in others these
expenditures are mandated at the central level and only implemented at the local
level. The first arrangement is consistent with decentralization; the second is not.
For now, however, these are often the best data available for cross-country com-
parisons. In 1998, the Decentralization Thematic Group at the World Bank initiated
a project to coordinate several cross-country data collection efforts across the Bank
as well as with several outside partners to form one comprehensive, detailed data-
base of fiscal and institutional variables at the national and subnational levels.
- 44 -



Project Design: Political, Fiscal, and Administrative Decentralization
Local Technical and Managerial Capacity
The recent international trend toward decentralization has provoked a lively de-
bate about the capacity of local governments and communities to plan, finance,
and manage their new responsibilities. Assessing, improving, and accommodating
varying degrees of local capacity have become more and more important as decen-
tralization policies transfer larger responsibilities, as well as budgets, from national
governments to local governments and communities.
While a common rationale for decentrahization is that local governments' prox-
imity to their constituents enables them to better manage resources and match their
constituents' preferences, it is not clear that local governments and communities have
the capacity to translate this information advantage into an efficiency advantage.
Small, inexperienced, local governments may not have the technical capacity to imple-
ment and maintain projects or the training to effectively manage larger budgets.
This section discusses two main issues in the local capacity debate: what local
capacity is and what to do about varying degrees of local capacity once it has
been identified. "Local government" is taken to mean the level of government at
which some degree of everyday face-to-face interaction between citizens and
government is possible.
Assessing Local Capacity
Decentralization planners have used the general guideline that central agencies
should focus on creating and sustaining the enabling environment and strategic
priorities, while local organizations should concentrate on tailoring the specific
mechanisms of service delivery and public expenditure packages to fit local needs
and circumstances. In reality, however, the varying degrees of local capacity-both
in local government and in civil society and the private sector-obviously affect
decisions about which levels of government can best perform which tasks. In most
cases decentralization of basic services does not mean the wholesale transfer to
local agencies of all tasks associated with those services. An assessment of local
capacity is an integral part of designing decentralization.
What Is Capacity?
Measuring local capacity can be difficult, and the debate over quantifying it has
often been motivated by political concerns as well as technical considerations about
the local government's ability to provide services. Central governments have used
the "lack of capacity" excuse for refusing to transfer their authority, financial re-
sources, and accompanying privileges to local units. For example, Fiszbein found
in Colombia that "what was being characterized [by national agencies] as poor
planning in municipios was in fact a genuine disagreement between local and na,-
tional priorities."28 The municipios were actually demonstrating considerable local
28. Ariel Fiszbein, "Emergence of Local Capacity: Lessons from Colombia," World De-
velopment 25:7(1997):1029-43.
-45 -



Decentralization Briefing Notes
capacity by turning down conditional matching grants from central agencies and
borrowing funds locally at market rates in order to implement their own priorities.
For this reason, it is useful to set out some of the relevant issues in objectively
measuring local capacity. There are four general skill areas: identifying and analyz-
ing local problems in order to plan appropriate responses, mobilizing and manag-
ing resources, communicating and coordinating policy implementation, and
resolving local conflicts.39The fact that a community and its government exist indi-
cates the presence of at least some of these skills. The challenge is to identify the
latent capacity in the local government, civil society, and private sector and bring it
into the development programs.
The first task is to identify the specific tasks that local governments and citizens
will need to carry out. In the area of policy implementation, for example, a check-
list of responsibilities might include writing specifications for the technical ele-
ments of programs, evaluating the impact of the program on the local environment,
and contracting for services or buying equipment.
The second task is to create appropriate, comprehensive indicators to measure
local groups' ability to carry out the required functions. There are several issues to
consider in measuring capacity:
O Outputs or outcomes-Output measures can be defined as the number of ser-
vices delivered (for example, school attendance, vaccination coverage, kilome-
ters of road built or vehicle traffic) while outcomes refer to the ultimate objective
(for example, literacy, morbidity or mortality, access to markets). There are pros
and cons of using these different types of measurements. While both outputs
and outcomes reflect supply and demand factors, output indicators are easier
to measure and reflect more directly the service delivery efforts undertaken.
On the other hand, they are only an intermediate objective and if they fail to
achieve the ultimate objective, a great deal of money, which could have been
spent on other interventions, can be wasted.
* Variable ability-One of the most important tasks in planning decentralization
is to identify and account for the differing skill levels of local governments.
Variations in the ability to plan and execute projects, for example, can lead to
differing implementation rates and ability to use revenue, which in turn can
exacerbate inequalities.
*  People or institutions-Does local capacity consist of individuals who may or
may not continue to play a role in the government or are there institutionalized
mechanisms (such as competitive pay, prestige, contracting arrangements, or
training procedures) for ensuring a continuous supply of technical and mana-
gerial expertise? Important factors to look at in assessing the community's ca-
pacity are depth and history of civil society organizations, number of private
contractors, and concentration of skills. (Is there only one contractor, who could
move at any time, or are there several firms that could provide technical help?)
29. Norman Uphoff, "Institutional Capacity and Decentralization for Rural Develop-
ment," paper presented at the Technical Consultation on Decentralization, Food and Agri-
cultural Organization, Rome, 16-18 December, 1997, suggests these four skill areas.
- 46 -



Project Design: Political, Fiscal, and Administrative Decentralization
* Bureaucratic and technical infrastructure-The processes by which information
is received, processed, and stored underlie most local government functions. The
existence of appropriate databases and filing systems, for example, is essential to
the ability to collect taxes or user fees.
* The role of civil society-Nongovernmental organizations can often be a source
of trained, experienced personnel, and local construction, accounting, and other
firms can provide services on a case-by-case basis. The local government's rela-
tionship with the private sector and its demonstrated ability to contract out ser-
vices are an important, often overlooked part of capacity. In assessing the
community's longer term capacity, the depth and history of civil society organi-
zations (is there one skillful community leader or a network?) as well as the num-
ber of private contractors and concentration of skills should be considered.
Decentralize or Build Local Capacity? Which Comes First?
The traditional approach to decentralization has been to build capacity before trans-
ferring responsibilities or revenues. This cautious method was motivated by wor-
ries about irresponsible spending, local corruption, regional inequities, and service
collapse as well as by many central governments' reluctance to devolve authority.
Some analysts even argued that the lack of local capacity, among other factors,
made decentralization ineffective and even undesirable in developing countries. 3
This traditional approach is changing, however, as increasing evidence shows
that the capacities of all levels increase as decentralized service systems mature.
There is a growing appreciation that "management is a performance art" better
learned by doing than by listening. Some researchers report that local government
capacity increased slightly but perceptibly in the years following some modest de-
gree of decentralization in Indonesia, Morocco, Pakistan, and Thailand. How local
governments would have handled more complete decentralization is unknown.
Ongoing research in Bolivia shows that local governments' education investments
are more rational and more in line with local needs than all the national government's
expenditure.3" In general, much of the evidence indicates that decentralization has
increased local participation and hence local government leverage in gaining ac-
cess to national resources and has encouraged the development of public and pri-
vate planning and management skill.
Implementing the Answer: Doing both Simultaneously
Decentralization can, in itself, be the best way to build local capacity. Central
support can be important to maintain equity in spending across jurisdictions and
ensure proper attention to training. One researcher points out that effective de-
livery of local services rests on partnerships crossing levels of government and
30. R. Bahl and J. Linn, "Fiscal Decentralization and Intergovernmental Transfers in Less
Developed Countries," Publius: The Journal of Federalism 24 (Winter 1994).
31. Jean-Paul Faguet, "Decentralization and Local Government Performance," paper pre-
sented at the Technical Consultation on Decentralization, Food and Agricultural Organiza-
tion, Rome, 16-18 December, 1997.
- 47 -



Decentralization Briefing Notes
the public, private, and civil sectors. Nevertheless, capacity building should not
be a supply-driven endeavor that provides the same support package to widely
varying local jurisdictions.32 It is also not always clear that national capacity is
greater than local capacity. One study, for example, showed that Italians rate lo-
cal government effectiveness higher than national government capacity.33
DEMAND-DRIVEN CAPACITY-BUILDING PROGRAMS CAN STRENGTHEN LOCAL COMMUNITIES AND
NETWORKS OF SKELLS. One way to enhance local capacity is through training and prac-
tice. Local institutions can be allowed to use a portion of program funds, or their
own funds, to contract for the technical expertise that they feel is appropriate to
their specific requirements. This technical help can often be found locally and can
be acquired quicker and cheaper than from central or regional sources. Similarly,
communities (or regional groupings of communities) can be given block grants for
their own capacity-building training programs. They can purchase the training they
need to fill the gaps that they have identified in their own management and techni-
cal capacity. They can decide whether to buy the training from local, regional, or
central institutions. Using local sources enables a local network of technical exper-
tise to develop. This local network can be tapped more efficiently for maintenance
of existing and new programs in the future.
LOCAL PARTICIPATION CAN BE A STRONG MOTIVATOR FOR CHANGE. Recent evidence
from Bolivia and Colombia shows that citizen oversight can be an important
impetus for local governments to actively improve their capacity.34 Regular, fair
elections and citizen councils can increase the pressure on local leaders to turn
popular demands into outputs.
CLARIY IN RESPONSrBILITY ASSIGNMENT IS ESSENTIAL. India's technically and mana-
gerially over-ambitious Small Farmers' Development Agency and Sri Lanka's lack
of guidance for the appropriate uses of district budgets, for example, led to low
levels of success. The more successful, albeit modest, decentralization efforts in
Indonesia and Thailand, however, had clearer procedures for local budget alloca-
tion and responsibilities.
Conclusion: Building Capacity through Decentralization
The prevailing wisdom today can be summed up in the following statement made
at the Technical Consultation on Decentralization and Rural Development, FAO,
Rome, December 1997: "Rather than plan and make large upfront investments in
32. Judith Tendler, Good Government in the Tropics (Baltimore, Maryland: The Johns
Hopkins University Press, 1997).
33. R. Putnam, Making Democracy Work (Princeton, New Jersey: Princeton University
Press, 1993).
34. Ariel Fiszbein, "Emergence of Local Capacity: Lessons from Colombia," World De-
velopment 25:7(1997):1029-43; Jean-Paul Faguet, "Decentralization and Local Government
Performance," paper presented at the Technical Consultation on Decentralization, Food and
Agricultural Organization, Rome, 16-18 December, 1997.
- 48 -



Project Design: Political, Fiscal, and Administrative Decentralization
local capacity building as a prerequisite for devolution of responsibility, there was
a broad consensus that it would be quicker and more cost-effective to begin the
process of devolution, permit learning by doing and build up capacity through
practice." The evidence increasingly shows that local capacity can be built by the
process of decentralization, particularly when appropriate programs to increase
interaction with the private sector are included in decentralization design.
-49 -






3
Service Delivery
This chapter is divided into six notes, each focusing on a different sector: educa-
tion; health care; safety nets; infrastructure; irrigation, water supply, and sanita-
tion; and natural resources management and environment.
It should be stressed that the other notes in this volume are relevant and ex-
tremely important for those involved with the decentralization of a particular sec-
tor. Decentralization is often undertaken as part of a sectoral reform; yet its impact
on efficiency will depend on political, fiscal, and administrative accountability,
which is part of the broader framework of intergovernmental relations. Its impact
on equity will depend greatly on the design of intergovernmental transfers. Un-
dertaking a sectoral decentralization dialogue without examining the intergov-
ernmental landscape can lead to unintended consequences such as local corruption
or growing geographic disparities. In all sectors, much has yet to be learned about
the different outcomes derived from varying types of decentralization implemented
in disparate institutional settings. Nevertheless, a number of studies are under-
way within and outside the World Bank; the Bank will disseminate findings as
they become available.
-51 -



Decentralization Briefing Notes
Decentralization of the Education Sector
Today most countries are experimenting with or contemplating some form of edu-
cation decentralization. The process transfers decisionmaking powers from central
ministries of education to intermediate governments, local governments, commu-
nities, and schools. The extent of the transfer varies, however, from administrative
deconcentration to a much broader transfer of financial control to the regional or
local level.
While there are solid theoretical justifications for decentralizing education sys-
tems, the process requires strong political commitment and leadership in order to
succeed. The path, depth, and ultimately the outcome of decentralization reforms
depend on the motivations for reforms, the initial country and sector conditions,
and the interaction of various important coalitions within the sector.
Why Decentralize Education?
In a world where most governments have experienced the pitfalls of centralized
provision of education services-mainly opaque decisionmaking, administrative
and fiscal inefficiency, and poor quality and access to services-the theoretical ad-
vantages of decentralization are extremely appealing. Two sets of motivations are
often in play. On one hand, many people believe that the process of decentraliza-
tion will substantially improve efficiency, transparency, accountability, and respon-
siveness of service provision; better reflect local priorities and encourage
participation; and eventually improve coverage and quality. On the other hand,
governments with severe fiscal constraints are enticed by the potential of decen-
tralization to increase efficiency and lower costs. Beneficiary cost recovery schemes
such as community financing have emerged as a means for central governments to
offload some of the fiscal burden of education service provision. In practice, the
objectives often compete and the net outcome depends considerably on the extent
of accountability-the more that national and local leaders are accountable to their
populations, the more likely that decentralization will lead to better, more respon-
sive services and will not be used solely as a cost-cutting measure.
Deciding Who Controls What
There is ongoing debate about the appropriate locus of decisionmaking within the
education sector. The debate remains unresolved because the process requires that
policymakers rationalize and harnonize a complex set of complementary func-
tions that includes curriculum design, teaching methods, student evaluation, text-
book production and distribution, teacher recruitment and pay, school construction
and rehabilitation, education financing, and parent-teacher linkages. The choices
of who does what are further complicated because each of these functions has to be
evaluated for primary, secondary, and tertiary education, and often for preschools
and adult literacy as well. Some emerging areas of consensus are summarized in
the following paragraphs.
SrmNG srANDARDS. The central government is generally better equipped to address
equity issues and ensure a common minimum educational standard. Distribution of
- 52 -



Service Delivery
teachers and technical assistance to underperformiing areas are commonly central func-
tions. Some federal governments (for example, the United States) have successfully
delegated monitoring and standards to states or provinces.'
CURRICULUM DESIGN AND TEACHNG METHODS. The central government generally plans
the national curriculum within the national ministry of education. In heterogenous
societies such as Nigeria or Ethiopia, local governments are given some latitude to
localize curriculum, by introducing indigenous languages, for example.
STUDENT EVALUATION. National-level examination systems can shape the general
direction of local evaluation and monitoring, but teachers, schools, and districts
generally evaluate on a day-to-day basis. Also, developing and disseminating school-
based performance plans increases the information available so that communities
can take an active part in monitoring education quality.
TEXTBOOK PRODUCTION AND DISTRIBUTION. Textbook production and distribution is
usually centralized in order to harmonize textbooks with curriculum design and
development process as well as increase cost efficiency due to scale economies in
centralized procurement, production, and distribution.
TEACHER RECRUTrMENT AND PROMOTION. Teachers are central government civil ser-
vants in most developing countries, with conditions of service determined by the
public service commission or an independent teacher service commission. In a few
countries, mostly federal systems such as Brazil and India, where responsibility for
provision of education is fully devolved, some teachers, mainly in primary and
secondary education, are civil servants of intermediate governments.
In many cases, including federal states such as Pakistan, teachers have re-
sisted reforms that would have placed them under the jurisdiction of local or
intermediate governments, on the grounds that they lose interjurisdictional mo-
bility, comparable conditions of service, and prestige. Constructive dialogue with
the relevant teacher unions and groups is an important priority in education re-
form, because teacher resistance to reforms of conditions of service designed to
make them formally accountable to intermediate and local governments remains
a common impediment to system reform.
EDUCATION FINANCING. A large share of education services, particularly primary
and secondary schooling, are publicly provided and. financed in most countries;
though some degree of private financing and provision is increasingly common.
In devolved education systems where subnational governments have statutory or
constitutional responsibility for some education subsectors, central governments
either assign taxes or transfer resources sufficient to cover these and other de-
volved responsibilities. Some central and intermediate governments also provide
1. Eric A. Hanushek and Dale W. Jorgenson, eds., Improving America's Schools: The Role
of Incentives (Washington, D.C.: National Academy Press, 1996) contains a variety of essays
on the organizational structure and assignment of responsibilities in the U.S. school system.
- 53 -



Decentralization Briefing Notes
additional matching grants to local governments to increase local expenditures on
priority areas. In deconcentrated systems, field offices receive their budgets from
the central ministry through the next highest level.
Many governments have also implemented community management and cost-
recovery schemes in response to shrinking central fiscal resources and growing
demand for basic education. Partial community financing in the form of contin-
gency fees, reduced (subsidized) school fees, PTA contributions, and ad hoc com-
munity contributions in cash or kind are becoming quite common.2 In rare cases
communities are entirely responsible for school financing, usually for a subset of
schools. In China, for example, there are government schools (gonban) and com-
munity-run schools (minban).
On the management side, communities are increasingly given substantial con-
trol over the day-to-day operation of schools. In many countries, local school com-
mittees and district education boards and committees composed mainly of
community representatives have been set up for this purpose. In rural El Salvador,
the elected members of a community education association are vested with legal
responsibility for enrolling a specified number of students, establishing classrooms
or new schools, and hiring and supervising their teaching staff.3 Bhutan and Zam-
bia also provide good examples of this trend. Ireland, the United Kingdom, and the
United States also have highly developed school-level management systems. PTA
roles are highly variable across schools, community income levels, and countries.
CONSTRUCTlON AND MAINTENANCE. Although in a few cases, such as Tanzania, the
central government finances both construction and maintenance, this is traditionally
among the most decentralized education subfunctions. Responsibilities for construc-
tion and maintenance are often divided between government levels: for example, the
province provides resources for construction and major rehabilitation, and the local
government conducts routine maintenance; communities and their representatives
share costs. Responsibilities also tend to be divided according to the level of school-
ing: local governments are responsible for primary schools, mid-level governments
for secondary schools, and central governments for tertiary education.
TEAcERs' SALARrES. Salaries, even in highly centralized systems, are usually paid from
the recurrent budget of the administrative level to which teachers report. The true ex-
tent of local control over teachers depends on whether or not the teachers are members
of a unified national civil service with pay scales set by the relevant service commission.
Community involvement in financing teachers' salaries can help make schools
more responsive to community needs, but the practice is controversial with teachers'
2. For an evaluation of the incentive effects of even marginal community cofinancing,
see Emmanuel Jimenez and Vicente Paqueo, "Do Local Contributions Affect the Efficiency
of Public Primary Schools?" Economics of Education Review 15:4 (October 1996):377-86.
3. World Bank, Development Research Group, "El Salvador's Community-Managed
Schools (EDUCO) Program: A First Report," Working Paper Series on Impact Evaluation of
Education Reforms 4 (Washington, D.C., 1997); World Bank, Development Research Group,
"Do Community-Managed Schools Work? An Evaluation," Working Paper Series on Impact
Evaluation of Education Reforms 8 (Washington, D.C., 1998).
- 54 -



Service Delivery
unions because it can create inequities in salaries and student-teacher ratios among
poorer and richer communities.
Impacts of Decentralization
The limited available evidence about the impact of decentralization on education
services is mixed depending largely on why and how decentralization was under-
taken. In some cases, central governments have delegated responsibilities to local
governments and communities without providing adequate targeted support to
poorer areas. In Brazil the central government has increased overall access (enroll-
ments) but has done little to reverse persistent regional inequities in access to school-
ing, per capita expenditures, and quality. Chile's experience also suggests that
decentralization does not by itself remove inequalities between localities of vary-
ing incomes, and quality in poorer communities continues to lag. These results are
supported by experiences in New Zealand and Zimbabwe.4 However, the design
of these decentralized systems has been criticized.
The initial evidence suggests that decentralization to subnational governments
(that is, provinces and states) may not be sufficient and that increased autonomy for
communities and school actors may be necessary to improve schools and learning.
By increasing the participation of parents, community-managed schools in El Salva-
dor show significantly lower rates of student and teacher absenteeism.5 While this
type of management does not appear to have improved student performcance in tests
according to a recent evaluation, it may be just a matter of time before better student
attendance translates into higher student achievement. In Nicaragua, controlling for
similar household background and school inputs, students in schools that make more
of their own decisions about school functions perform better in tests.6 These results
4. For more information on New Zealand, see Liz Gordon's "Controlling Education:
Agency Theory and the Reformation of New Zealand Schools," Educational Policy 9:1 (March
1995):54-74 examines the central tenets of agency theory, describes how the approach has
been applied in the New Zealand school system, and through a case study report analyzes
the effects of this approach on the schools. Lyall Perris's "Implementing Education Reforms
in New Zealand: 1987-97," The Education Reform and Management Series 1:2 discusses the
economic and political context for the changes, describes the interplay of interests during
the implementation process, and concludes with a list of "lessons" to be learned from the
reforms' pitfalls and as-yet unresolved issues. For information on Chile, see T. R. Perry,
"Achieving Balance in Decentralization: A Case Study of Education Decentralization in Chile,"
World Development 25:2(1997):211-25.
5. World Bank, Development Research Group, "El Salvador's Community-Managed
Schools (EDUCO) Program: A First Report," Working Paper Series on Impact Evaluation of
Education Reforms 4 (Washington, D.C., 1997); World Bank, Development Research Group,
"Do Community-Managed Schools Work? An Evaluation," Working Paper Series on Impact
Evaluation of Education Reforms 8 (Washington, D.C., 1998).
6. Bruce Fuller, Magdalena Rivarola, Elizabeth King, and Berk Ozler, "Nicaragua's
School Autonomy Reform: A First Look," Working Paper Series on Impact Evaluation of
Education Reforms 1 (Washington, D.C.: World Bank, 1996); World Bank, "Views of Parents,
Teachers, and Directors," Working Paper Series on Impact Evaluation of Education Reforms
5 (Washington, D.C., 1998); World Bank, Development Research Group, "What's Decentrali--
zation Got to Do with Learning?" Draft paper presented at the Annual Meetings of the
American Educational Research Association, Washington, D.C., 1998.
- 55 -



Decentralization Briefing Notes
are derived from ongoing impact evaluations. Because the impact of management
reforms such as these may take time to unfold, further empirical analysis is needed.
Conclusion
Decentralization of education systems demands harmonization of a complex set of
functions at each level-primary, secondary, tertiary, and nonformal education. Is-
sues of how far to devolve decisionmaking in each case and to whom continue to be
debated. There are a number of ongoing experiments worldwide, ranging from devo-
lution of limited functions to intermediate governments and local governments to
community-based management and financing of schools. The current consensus is
that tertiary education and specific functions such as curriculum design and stan-
dards setting are best retained by the center, that secondary and primary education
should be devolved as far as possible, and that local participation in school manage-
ment improves accountability and responsiveness and fosters resource mobilization.
- 56 -



Service Delivery
Decentralization of the Health Sector
The general argument for decentralizing health care is that greater community par-
ticipation in local health activities can lead to improved service quality and cover-
age. Yet exactly how these benefits can be realized and just what impact different
kinds of reforms will have are not well understood. Several features of health care
services (the controversial nature of some services such as family planning, the
importance of formal training and supervision for personnel, the need for inte-
grated referral services) make health sector decentralization more complex and
potentially more difficult than in other service delivery areas. Since decentraliza-
fion of health care services is often politically driven, the theoretical benefits tend
to get more attention than results of actual experiences, which are mixed. One les-
son that does seem clear from existing experiences is that without proper planning
and acknowledgment of the lessons learned by other countries, decentralization of
health care can be disappointing at best and detrimental at worst.
What Is Decentralization?
The term "decentralization" is used to describe a wide variety of power transfer ar-
rangements and accountability systems. Policies range from the transfer of limited
powers to lower management levels within current health management structures
and financing mechanisms to extensive sector reform efforts designed to reconfigure
the provision of even the most basic services. In the first case, decentralization may
later become the driving force for health sector reform; in the second, it is driven by
the wider sectoral reform efforts. The parameters for decentralization-the speed,
pressures, and scope of issues to consider-vary considerably.
The Promise of Decentralization
Health sector decentralization has become appealing to many people because it
has several theoretical advantages.7 The following are sometimes cited as justifica-
tions for decentralizing health care:
* A more rational and unified health service that caters to local preferences
* Improved implementation of health programs
* Lessened duplication of services as the target populations are defined more
specifically
* Reduction of inequalities between rural and urban areas
* Cost containment from moving to streamlined, targeted programs
* Greater commnunity financing and involvement of local communities
* Greater integration of activities of different public and private agencies
* Improved intersectoral coordination, particularly in local government and ru-
ral development activities.
7. Anne Mills, "Decentralization and Accountability in the Health Sector from an Inter-
national Perspective: What Are the Choices?" Public Administration and Development 14
(1994):281-92.
- 57 -



Decentralization Briefing Notes
The Caveats
There is little concrete evidence to date to confirm that these potential benefits can
always be realized. Few developing countries have long-term experience with health
sector decentralization, and its impact on the management of the sector and on the
services it delivers has rarely been evaluated. The impact of health decentralization
will depend on which specific health services are decentralized in what kinds of
institutional environments and under which intergoverrLtnental fiscal policies. Yet,
the debate about whether decentralization does indeed improve equity, efficiency,
accountability, and quality in the health sector continues at a general level without
data to inform it. Although anecdotal and country-study evidence confirms that poorly
designed and hastily implemented decentralization has serious consequences for
health service delivery, we do not have a clear analytical framework to isolate or
generalize the factors behind successful and unsuccessful decentralization.8
General Guidelines
Experiences that have been documented show that achieving the benefits of decen-
tralization depends heavily on policy design. In general, careful attention must be
given to health service needs and priorities in deciding which functions and pro-
grams to transfer and which to retain under central control. If a function is critical
to the attainment of central-level goals and its sustainability at the local level can-
not be guaranteed, it should not be decentralized. With this in mind, the following
paragraphs summarize general guidelines for assigning responsibilities to central
and local levels and outline a framework for planning decentralization.
PROGRAM DESIGN. The central government often has a comparative advantage
in creating and disseminating information and establishing parameters for prior-
ity national programs, such as family planning and vaccination programs, with
externalities that reach beyond individual communities. Local governments, on
the other hand, are better positioned to select the details of service delivery mecha-
nisms that suit local conditions.
FINANCING. The high cost of health services generally requires central govern-
ment transfers to ensure local government's ability to carry out responsibilities.
Conditions, such as minimum levels of expenditure on maintenance and training,
are often attached to these resources to ensure consistent quality and sustainability.
Subnational governments are better able to set user charges and other cost-recov-
ery mechanisms and address their communities' needs by planning expenditure
allocations within the parameters set by national standards.
8. Lucy Gilson, Peter Kilima, and Marcel Tanner, "Local Government Decentralization
and the Health Sector in Tanzania," Public Administration and Development 15 (1994):451-77;
Riitta-Liisa Kolehmainen-Aitken and William Newbrander, Decentralizing the Management of
Health and Family Planning Programs, Lessons from FPMD Monograph Series (Boston: Man-
agement Sciences for Health, 1997).
- 58 -



Service Delivery
STANDARDS SETTING. Central governments usually have the sole mandate for set-
ting standards for licensing health care professionals, for registration and quality
control of drugs, and similar functions. The capacity for licensure rarely exists at
the local level, and in countries (such as India) that have decentralized this func-
tion, pharmaceuticals of questionable quality and safety freely cross from states
with a poorly managed drug registration system to other, better managed states.
SERVICE DELIVERY. Although one can say, as a rough approximation, that most health
care services should be decentralized to the lowest level of government whose juris-
diction includes the beneficiaries, the complex linkages between different services
require careful planning of appropriate organizational roles, relationships, and orga-
nizational structures. Serious distortions and hindrances to health service delivery
are created if functional roles and organizational structures are devised without full
participation of health managers. In the Philippines, the integrated district health
office concept that had taken years to develop was destroyed when provinces were
handed responsibility for managing hospitals and municipalities became respon-
sible for primary health care. No linkages were created between these two levels of
health service, even though the patient load and quality of service in one are highly
dependent on the other. In Bolivia and Zambia, the small size of many local govern-
ments has similarly separated primary care from its secondary referral facilities. In
Papua New Guinea, the failure to specify clearly the new roles and responsibilities
for subsidy payments to church-run health services resulted in a severe funding cri-
sis, which forced the churches to close many of their health facilities for a time. Be-
cause churches provide about half of all rural health care in Papua New Guinea, and
their services are well integrated with the public sector, large numbers of rural resi-
dents were deprived of any access to health services.
LEGAL REspoNsIBILrry. The legal and regulatory implications of decentralization
for the work of health personnel must be considered to avoid placing health staff at
risk. In the Philippines, the legal protection of local government health staff under
national malpractice legislation has been questioned. After decentralization in Papua
New Guinea, provincial health inspectors could no longer legally enforce national
legislation concerning the cleanliness of markets, restaurants, and workplaces.
A First Attempt at a Framework for Assigning Expenditure
Responsibility
Local governments' freedom to adapt to local conditions must be balanced by a
common vision about the goals of the health sector and the purpose of decentrali-
zation in furthering those goals. Decentralization policy should include some coor-
dinating mechanism.
The prominence of local political interests increases as decentralization trans-
fers more responsibility to this level. While responsiveness to local demands is a
benefit of decentralization, it brings two main disadvantages. First, local officiaLs
frequently change and may therefore be uninformed about key national health
policies. Second, local groups may also oppose national policies. One provincial
- 59-



Decentralization Briefing Notes
governor in the Philippines banned a donor-funded project in support of family
planning services. (It is certainly acceptable and indeed desirable if decentraliza-
tion enables local governments to design programs according to local preferences;
however, services of nationai priority, such as family planning, should be man-
dated and funded by the central government.)
Adequate financing and clear delineation of new financial flow mechanisms
are essential. In the preoccupation with defining an essential package of basic health
services or a new decentralized health service model, the crucial issue of financing
the decentralized health system may be overlooked. A significant financial gap
between what is available and what is being planned can compromise the health
sector's ability to provide equitable, efficient, and good quality services under de-
centralization. South Africa and Zambia are both facing this issue in their current
decentralization efforts. Several guidelines emerge from experience:9
* Revenue allocation must take existing local expenditure responsibilities and
own resources into account. In Bolivia and the Philippines, for example, the
fixed funding formulas used to allocate national revenue between local gov-
ernments failed to take into account the existing level of health facilities that
these governments inherited and the services they were expected to provide.
Local governments that inherited expensive new responsibilities, such as hos-
pitals, often were not able to maintain the level of service previously provided.
* Local freedom to allocate funding should be tempered by nationally set mini-
mum requirements. The degree of freedom that the local level has in deciding
what resources to devote to health services, together with local budgetary reali-
ties and financial procedures, greatly influences the operation of the health sys-
tem. In Papua New Guinea, tight budgets and weak planning capacity led several
provincial governments to stop paying for nurse aide training, which had been
decentralized to them. Within four years the training capacity for this important
staff category fell from 13 govemment schools with 135 annual graduates to 3
schools with 13 graduates resulting in a severe shortage of skilled workers.1"
* Any central policies must consider local conditions and capacities."' In the
Philippines, the discrepancy between the cost of health staff benefits prom-
ised under centrally negotiated labor agreements and the local government
capacity to pay has seriously eroded staff morale, while cumbersome local
9. For more information on these and other examples, see C. Collins and A. Green,
"Decentralization and Primary Health Care: Some Negative Implications in Developing
Countries," International Journal of Health Services 24:3(1994):459-75.
10. J. Thomason, 'Disbursement, Decentralization, and Development: Lessons from the
First Rural Health Services Project in Papua New Guinea," Public Administration and Devel-
opment 8(1998):391-99; D. Campos-Outcalt, K. Kewa, and Jane Thomason, "Decentralization
of Health Services in Western Highlands Province, Papua New Guinea: An Attempt to Ad-
minister Health Service at the Subdistrict Level," Social Science and Medicine 40:8
(1995):1091-98.
11. The Special Edition of The Family Planning Manager IV:2 (March/April 1995) con-
tains multiple case studies.
- 60-



Service Delivery
financial procedures have jeopardized the timely availability of essential medi-
cal supplies. Some local governments require up to 40 separate signatures
before a purchase order can be sent to a supplier!
Capacity constraints cannot be ignored in either central or decentralized man-
agement levels. Ignoring capacity constraints at either central or local levels and
giving inadequate or delayed attention to training staff for their new roles are very
serious omissions with predictable effects on health services. Decentralizationbrings
a heavy new management burden, especially to the lower levels. Qualified health
managers are in short supply in many countries, and regional variations in the
supply of health workers can exacerbate existing inequities." Furthermore, man-
agement training capacity may be insufficient to meet rapidly expanding training
needs. Madagascar, for example, has begun to transfer planning, management, and
budgetary authority to 111 health districts without having a sufficient number of
qualified health managers to serve all these districts.
Decentralization changes the roles of the central ministry staff from line man-
agement to policy formulation, technical advice, and program monitoring. The
central-level managers also require systematic retraining and reorientation, which
many countries have overlooked. In other cases, staff cuts at the central ministry
of health have been so severe that the center's capacity to function effectively is
in question. In Nepal, for example, initial staff cuts at the central level paralyzed
the expanded program on immunization.
Conclusion: The Essentials
hn summary, decentralization creates major challenges for health service provision.
Active involvement of health managers in the decentralization design, clear national
resources allocation standards and health service norms, and an ongoing system for
monitoring are essential for guarding equity and quality and for improving efficiency.
12. M. Gonzalez-Block, R. Leyva, 0. Zapata, R. Loewe, and J. Alagon, "Health Services
Decentralization in Mexico: Formulation, Implementation, and Results of Policy," Health
Policy and Planning 4:4(1989):301-15, gives an example of how differing local capacity can
create new inequities in health service availability.
- 61 -



Decentralization Briefing Notes
Decentralization and Safety Nets
Safety nets protect a person or household in three types of situations:
* When there is chronic incapacity to work and earn (the severely disabled, eld-
erly, young orphans)
* When there is an unpredictable "idiosyncratic" shock (sudden death or serious
illness of breadwinner)
* When there is an unpredictable community "covariate" shock (economic reces-
sion, bad harvest, flood).
Hence safety net programs serve two important roles: redistribution in situa-
tions of chronic incapacity (income transfers, food supplement programs) and in-
surance in times of unpredictable shocks (public works programs, drought relief).
A key challenge in the design of safety nets is to maximnize the benefits to the needy
and vulnerable while minirnizing the administrative and political costs as well as
the disincentive effects. The high cost of obtaining accurate information about the
incomes and needs of the poor, for example, creates a significant barrier to improv-
ing the targeting of services and transfers to the poor.3
Experience with Decentralized Safety Nets
Some have argued that decentralizing poverty monitoring and management of
antipoverty programs from the national to the local level can reduce costs and im-
prove targeting. Local governments and administrators may be better informed
about members of their community and thus better able to identify their poor.
There is some empirical evidence about this, but experiences seem to differ de-
pending on which level of local government is involved. For example, in Albania,
where social assistance allocations are deternined at the commune level (population
of about 2,500), local authorities appear to have more information than is available
through formal monitoring systems. This enables them to target the poor more effec-
tively. District officials in Karnataka State, India, have attributed a tenfold increase in
information flow from communities to decentralization. This has helped to increase
the warning time before natural disasters and has improved the government's abil-
ity to respond to and fend off potential disasters. In contrast, where decentralization
has focused at the provincial or state level (rather than the district or commune level),
targeting has not been as successful; for example, in Vietnam, despite more centrally
funded transfers going to the poorer provinces, the poorer areas and people within
13. Subbaro and others, Safety Net Programs and Poverty Reduction: Lessons from Cross-
Country Experience, Directions in Development Series (Washington, D.C.: World Bank, 1997);
T. Besley and R. Kanbur, "Principles in Targeting," in Michael Lipton and Jacques van der
Gaag, eds., Including the Poor, Regional and Sectoral Studies Series (Washington, D.C.: World
Bank, 1993); D. van de Walle, "Incidence and Targeting: An Overview of Implications for
Research and Policy," in D. van de Walle and K. Nead, eds., Public Spending and the Poor
(Baltimore and London: The Johns Hopkins University Press, 1995).
- 62 -



Service Delivery
the provinces do not necessarily receive additional resources. This is because the
provinces use different criteria for distributing funds within their jurisdiction than
the center does."4 In Argentina wealthier provinces appear to be more effective at
targeting federal safety net subsidies to their poorer areas than do poorer provinces.15
Many of the benefits of decentralization seem most apparent at the village
level and least apparent when decisionmaking rests at the provincial level. De-
centralization at the village level can lead to greater participation and more effec-
tive local development strategies and thus can enhance the delivery of services
targeted to the poor. But the evidence is mixed. In some cases, there is also evi-
dence of corruption at the local level.
In some countries, particularly the transition economies of the former Soviet
Union, national governments have decentralized safety nets with the goal of low-
ering the national deficit. But often expenditure responsibilities have been decen-
tralized without adequate provisions for revenue sources. Such unfunded mandates
lead to a deterioration in the quality and quantity of services.
Ways to Decentralize Safety Nets
There are many different ways to decentralize a safety net program. It is difficult to
generalize about the potential advantages or disadvantages of safety net decen-
tralization, because these will differ according to the aspect of the program being
decentralized. Different levels of goverrunent can be given responsibility for differ-
ent aspects of program design and implementation (for example, eligibility crite-
ria, financing, and administration). The design of intergovernmental grants is also
key to determining how each level of government will participate in the program.
ELIGIBILrrY cRarcRA. Eligibility criteria can be set at the national level or decentral-
ized to the subnational level. The main advantages of nationally consistent targeting
criteria are that central funds can be distributed to the greatest number of nationally
defined absolute poor or vulnerable people, regardless of where they live. Centrally
established criteria also make it possible to evaluate programs consistently across
areas to assess their effectiveness, and they may limit the ability of local elites to
monopolize benefits. The main disadvantage to central eligibility criteria is that the
identification of vulnerable groups may not reflect local values and preferences. This,
need not be a problem if the locality can raise additional local funds to supplement
the central allocations, but if the locality is very poor, it may not be able to do so.
However, depending on local power dynamics, local eligibility criteria may actually
not be desirable. Local criteria may be appropriate for distributing locally raised rev -
enues, but nationally unifonn eligibility criteria are generally recommended for the
distribution of national funds for poverty alleviation or safety nets.
14. Govinda Rao, Richard Bird, and Jennie Litvack, "Intergovernnental Fiscal Relations
and Poverty Alleviation in Vietnam," Policy Research Working Paper 1430 (Washington,
D.C.: World Bank, East Asia and Pacific, Country Operations Division, 1995).
15. M. Ravallion, "Reaching Poor Areas in a Federal System," Policy Research Working
Paper 1901 (Washington, D.C.: World Bank, 1998).
- 63 -



Decentralization Briefing Notes
APPLYING THE cRrTERIA. Based on criteria set by the center or a subnational level,
individuals or households can be identified for participation by central adminis-
trators in the locality; local officials from the provincial, district, or village level; or
members of local nongovermmental organizations (NGOs). Generally, project imple-
mentation is best undertaken at the lowest possible level, with monitoring by higher
levels to ensure that standard eligibility criteria are being followed.
FINANCING. A well-functioning social safety net is an important element of a
broad-based, inclusive growth strategy. Given the objectives of a safety net identi-
fied earlier, the central government has an important role in financing it. For ex-
ample, areas that require the greatest assistance often have the least local revenues
with which to support safety nets; subnational levels are often unable to deal with
covariate risk. Nevertheless, subnational governments are also often concerned
about household vulnerability and the need for safety nets and sometimes are able
to raise substantial revenues to supplement central funds. Such cofinancing of safety
nets (between levels of government) can be encouraged through the design of match-
ing intergovernmental transfers. For example, the centrally sponsored safety net
schemes in India are generally matched by the states.
Intergovernmental transfers canbe either general or conditional. For general trans-
fers, a sum of funding is provided to the local government to deliver a range of ser-
vices, but actual expenditures are determined at the local level (Hungary, Latvia).
Because a minimum level of services is desired by the central government, it is usu-
ally recommended that safety nets not be funded through general grants. Conditional
transfers that require the subnational level to spend the designated funds on safety
nets are preferable. Conditional transfers can be matching or nonmatching grants.
The central government can encourage subnatioral governments to raise local funds
for safety nets by using matching transfers, and it can vary the matching rates de-
pending on local fiscal capacity. For example, a wealthier province might receive one
central dollar for every three dollars it raises and spends on safety nets, while a poorer
province might receive three central dollars for every one dollar it raises and spends
on safety nets (see the section on intergovernmental transfers and grants).
While these methods, combined with centrally determined eligibility criteria,
can encourage local governments to deliver safety nets, another option is for the
central government to administer the program itself through a representative in
the locality. This may be desirable if the central government cares strongly about
access to these services and has concerns about performance at the local level.
In short, decentralization of safety net financing can involve several different
arrangements. The central government can devolve resources to the province to
cover a range of expenditures, including safety nets. Each province then determines
how much it will spend on safety nets. Or the central government can set the crite-
ria by which the province selects the poor areas to receive funds and create incen-
tives that will reward provinces for allocating the funds to the poorer areas within
their jurisdiction. A third option is for the center to simply ask the provinces to pass
on the central funds to the villages and households that it has identified as vulner-
able. The second option is likely to be most effective.
- 64 -



Service Delivery
MON1TORING. To determine if funds are being targeted to the appropriate indi-
viduals and areas, it is important to have information on program expenditures
and household and village characteristics. Although there are many different de-
sign possibilities for decentralized safety net programs, ultimately the central gov-
ernment, whether it is involved in financing safety nets or not, should monitor the
programs. Information is best collected regularly and maintained at the subnational
level. (See the section on information and monitoring in decentralized systems.)
- 65 -



Decentralization Briefing Notes
Decentralization of Infrastructure
Technological developments that diminish economies of scale and political pres-
sures for decentralization have led to increased subnational government participa-
tion in providing and financing roads, water and sewerage systems, transit, power,
and telecommunications networks, as well as other public infrastructure. The ad-
vantages of this decentralization are becoming more apparent: subnational gov-
ernments are better positioned to identify local preferences for infrastructure
technology or service quality, accountability is enhanced with local decisionmaking,
and voters have more information on the price and quality of services, thereby
increasing competition in the sector. Theoretically, decentralization can also im-
prove equity in the distribution of infrastructure as smaller governments away from
the political center gain more latitude and funding to serve their constituents. Still,
decentralized provision and financing of infrastructure does not guarantee improve-
ments in the quality or distribution of infrastructure. Local performance depends
on the incentives facing decisionmakers, which in turn depend on the financial,
institutional, and political environments in which decentralization occurs.
This section outlines some general guidelines for assigning responsibility for
public infrastructure to central, subnational, and local governments and to com-
munity groups and the private sector. It describes the institutional settings that can
create the right incentives for decisionmakers. It also examines some of the evi-
dence about decentralization's effects on infrastructure provision.
Assigning Responsibility in Infrastructure Policy
Services may be provided, financed, or both by many different public sector insti-
tutional structures; they can be provided by the private sector under various con-
tracting arrangements; or they can both be provided and financed by the private
sector alone (see box 3.1).
Each type of infrastructure investment has particular characteristics-econo-
mies of scale, exclusivity, and spatial concentration of beneficiaries, for example-
that affect the desirable degree of decentralization. Provision of infrastructure
can be broken up into a number of decisions and separable responsibilities: plan-
ning networks, choosing locations, setting standards, constructing and operating
facilities, regulating, and maintaining systems."6 The basic logic behind assigning
these tasks is that the central government should have primary responsibility for
ensuring the quality and consistency of infrastructure networks that affect the
well-being of citizens in multiple jurisdictions, while local or subnational gov-
ernments should provide services that can vary between jurisdictions to suit lo-
cal needs without disrupting national development policies. The private sector
can play an important role in building, operating, and maintaining different types
of infrastructure. The assignment of responsibilities should ensure that the ben-
efits and costs of decisions are borne by those who make the decisions.
16. World Bank, World Development Report (New York: Oxford University Press, 1994).
- 66 -



Service Delivery
Box 3.1.  Altemative Ways to Produce or Provide Local Public
Services
Services can be produced (delivered) or provided (financed) by many different pub-
lic institutional structures, including:
* Central government (department, decentralized agency, or enterprise)
* Regional government (department, decentralized agency, or enterprise)
* Local government (department, decentralized agency, or enterprise)
* Central-regional arrangement
* Central-local arrangement
* Central-regional-local arrangement
* Regional-local arrangement
* Association of local governments
* Special-purpose local authority (encompassing more than one local government,
coterminous with a local government, or covering less area than a local govem-
ment).
In each of these cases the service may be produced by one structure, such as a
regional enterprise, and provided by another, such as a specific local govern-
ment department.
Services can also be provided in whole or part by the public sector but produced
by the private sector. Again, there are many possible forms:
* Build-operate-transfer (BOT) or build-own-operate (BOO) arrangements
* Other forms of public-private partnerships
* Development charges, exactions, and similar schemes
* Franchise arrangements
* Service contracts.
Finally, public services can also be both produced and provided by private agents,
sometimes in response to coercive legal requirements and sometimes even voluntarily:
* Compulsory provision by developers
* Compulsory provision by individuals through vouchers or self-financed
* Voluntary provision through formal or informal arrangements
* Provision by nongovernmental organizations (churches, enterprises).
Even this extensive list is incomplete. Many of the arrangements listed have a
number of possible variants, and there are ways to combine all these organizational
structures. Moreover, different structures might apply for policymaking, regulation,
financing, production, and so on. Finally, different structures may apply for differ-
ent services and for local governments with different characteristics (size, financial
capacity, and so on).
Source: Richard M. Bird, "Fiscal Federalism and Federal Finance," in Jornadas de Finanzas Publicas
(Cordoba, Argentina: National University of Cordoba, Faculty of Economic Sciences, 1995).
- 67 -



Decentralization Briefing Notes
DESIGN. Each level of govermment has a comparative advantage in designing infra-
structure: the central government has the nationwide perspective and jurisdiction to
ensure compatibility across the entire network, while the lower-level governments face
lower transactions costs in gathering and aggregating place-specific information. Local
initiative can also lead to innovative, more efficient designs. Private contractors can
often serve as a source of technical expertise in design.
Construction is one of the most commonly decentralized functions, for it is rela-
tively simple and place-specific. The private sector is often involved in this area as
well. In the Republic of Korea, local governments have acted as developers-con-
solidating large parcels of land, building infrastructure, then returning part of the
land to the original owners and selling the rest to recoup development costs.
FiNANaNG. Central governments usually provide the bulk of finance for infra-
structure investment, while lower-level governments and the private sector partici-
pate more often in financing small improvements, modifications, and local additions
to the national networks. Many central governments also provide transfers to local
governments for local infrastructure investmnent or maintenance in order to ensure a
minimum level of infrastructure for both reasons of efficiency (externalities) and eq-
uity. These transfers can be quite a large percentage of infrastructure funding.
Local governments can support capital investment and maintenance through
several channels including: (1) general transfers from the center, (2) specific pur-
pose transfers from the center, (3) local taxes, (4) earmarked user fees for specific
types of infrastructure (for example, water charges), and (5) subnational borrow-
ing (see separate notes on intergovernmental transfers, taxation, and subnational
borrowing). Their use of these different channels will depend on the system of
intergovernmental finance and the legal and regulatory system that sets the rules
for local finance (see constitutional, legal, and regulatory framework note). User
fees are a particularly important source for local infrastructure finance because lo-
cal governments can identify users and collect user fees in their jurisdiction quite
easily. (Note that overreliance on earmarked user fees for infrastructure finance
can starve local governments of an important source of finance with which to fi-
nance other public goods such as social services and address equity concerns.) Valo-
rization, a system of local taxation in which the cost of public works is allocated
proportionally to affected properties, has become more common. 17
In developing countries, in particular, maintenance tends to be neglected at all
levels, because the results of money spent are not as immediately apparent and the
pressures to extend the infrastructure network to underserved areas tend to be
strong. Central government transfers are often earmarked for specific capital ex-
penditures for infrastructure, but maintenance is provided under a general block
grant, which can then be used for many competing needs.
17. This method of taxation has varying success rates, however, and studies in Colom-
bia indicate that participation and careful planning are essential and that central govern-
ment funds may be required for initial financing.
- 68 -



Service Delivery
Informal financing of infrastructure, whereby private citizens contribute time,
money, and technical expertise to build and maintain community infrastructure
where governments have failed, is attracting increasing attention. Residents of the
Orangi district of Karachi, Pakistan, have invested money and labor to install a
sewerage system and build schools, health care posts, and other facilities. In Lima,
Peru, local community associations have financed such public works through in-
formal taxation. Throughout Vietnam, commune residents have assumed the re-
sponsibility for maintaining local infrastructure through household contributions
and in kind payments (most notably labor).
REGULATION. The central government usually plays a coordinating role in set-
ting standards and monitoring compliance. This ensures compatibility and consis-
tent quality across the national network and enables a wide variety of potential
service providers. The outcome of decentralizing regulation depends on the politi-
cal circumstances. On the one hand, turning regulatory authority over to local gov-
ernments will generally facilitate the adaptation of regulation to local conditions
and preferences. On the other hand, devolution can lead to interest group capture
if local governments do not have adequate regulatory capacity and broad-based
accountability mechanisms are not in place.
An Enabling Environment for Effective Local Provision
Decentralization's potential impact on infrastructure performnance is quite similar
to its impact on other sectors: it depends largely on the degree of subnational politi-
cal and financial autonomy, whether accountability is well established, and whether
the correct incentives are offered for a long-term view of infrastructure investment,
maintenance, and use. Because decentralization increases the number of actors in-
volved in infrastructure provision, the institutional settings and range of incentives
for subnational governments increase and more attention must be paid to the sig-
nals implicit in financing arrangements, regulatory structures, and other interac-
tions between levels of government and citizens. Some core elements of successful
institutions for decentralization include accountability and technical support.
ACCOUNTABILITY. The links between the central government, local government, the
private sector, and citizens must be designed to ensure that providers of infrastructure
are accountable to those who pay for the services as well as those who benefit from the
services. Participatory mechanisms should be structured so that the entire community
can participate in infrastructure decisions, particularly regarding location and financ-
ing issues, which have substantial distributive implications. Enhancing the availability
of public information regarding budgetary and procurement decisionmaking is im-
portant for community participation and accountability. The World Bank's Economic
Development Institute is currently conducting workshops with local governments
worldwide to enhance participatory budgeting and open procurement processes.
TECHNICAL SUPPORT. Inexperienced subnational and local governments often re-
quire some form of technical support as they take on new responsibilities. Local
- 69 -



Decentralization Briefing Notes
governments' technical capacity varies widely across jurisdictions, and national
assistance may be required to maintain consistent infrastructure quality.
A variety of public-private partnerships can be useful in building up local capacity.
Colombia, for example, met these technical needs in several ways when it devolved
primary responsibility for the provision of drinking water and sanitation services to
local governments in 1987. It contracted training out to nongovernmental organiza-
tions, private consultants, and universities in addition to drawing on public sector spe-
cialists from the Ministry of Public Works and Army Corps of Engineers.'8
Technical support is particularly important in maintaining regulatory quality
during decentralization. Among possible arrangements are training programs run
by national agencies, mutual assistance through regulatory associations (such as
the U.S. National Association of Regulatory Utility Commissioners), and twin-
ning relationships that pair inexperienced regulators with more experienced do-
mestic or international regulators.
Decentralization and Infrastructure Performance
Available data and cross-country studies indicate that decentralization can have
varied effects on the infrastructure sector (table 3.1). Choices about the optimal
level of decentralization are sector- and country-specific.
Estache and Sinha show that both aggregate and subnational infrastructure
expenditure increase as decentralization proceeds, particularly in developing coun-
tries. This could be an indicator that local governments prefer more infrastructure
than would have been provided by the central government, but it could also be a
sign that infrastructure expenditure is not as efficient.'9
Humplick and Estache found that performance indicators generally improve
slightly or stay the same when infrastructure sectors are decentralized, although
they do observe a few negative effects (the strongest being lower labor efficiency in
the electricity sector). They also show that decentralization can result in more vari-
able performance across jurisdictions, but it is not clear whether this is a symptom
of varying local capacity or of varying local preferences.20
18. World Bank, Colombia: Decentralizing Revenues and the Provision of Services; A Review of
Recent Experience, Report no. 7870-CO (Washington, D.C.: 1989).
19. A. Estache and S. Sinha, "Does Decentralization Increase Public Infrastructure Ex-
penditure?" In A. Estache, ed., Decentralizing Infrastructure: Advantages and Limitations, Dis-
cussion Papers 290 (Washington, D.C.: World Bank, 1995).
20. F. Humplick and A. Estache, "Does Decentralization Improve Infrastructure Perfor-
mance?" in A. Estache, ed., Decentralizing Infrastructure: Advantages and Limitations, Discus-
sion Papers 290 (Washington, D.C.: World Bank, 1995).
- 70 -



Service Delivery
Table 3.1. Main Effects of Decentralization on Roads, Electricity, and
Water Supply
Sector                   Desirable effects       Undesirable effects     Neutral effects
Roads                 * Condition of unpaved                           * Condition of paved
roads improves.                                 roads is
* Overall performance                             unchanged.
of roads improves.                            * Share of paved
roads in total
network is
unchanged.
Electricity            * Generation capacity    * Numnber of employees  * Customers served
improves.               per gigawatt/hour is    per employee
* Tariffs are lower.      high if there is no     unchanged.
vertical unbundling.   * System losses
unaffected (but
spatial decentrali-
zation preferred to
functional
decentralization).
Water supply           * Percentage of water                           * Production costs
loss decreases.                                 unchanged.
* Operation costs
unchanged.
* Percent access to
service unchanged.
* Incidence of water
borne diseases
unchanged.
Source: F. Humplick and A. Estache, "Does Decentralization Improve Infrastructure Performance?" in
A. Estache, ed., Decentralizing Infrastructure: Advantages and Limitations, Discussion Papers 290 (Wash-
ington, D.C.: World Bank, 1995).
- 71 -



Decentralization Briefing Notes
Decentralization of the Water Supply
Water is increasingly being managed as an economic, rather than a social good,
and various forms of decentralization are being used to support this new approach.
Govermments and other reformers are now trying to link service levels and costs,
provide incentives that increase the efficiency of water use, reduce costs, and in-
crease the sustainability of water service systems. This section briefly summarizes
some of the lessons that are emerging from these varied attempts at decentralizing
water services. In particular, it will focus on some of the lessons from decentralized
water supply, sanitation, and small-scale irrigation.
The Rationale for Decentralization: Water as an Economic Good
In theory, decentralized water services should improve governments' ability to treat
water as an economic good and to assess user charges that will create incentives for
efficient water use as well as finance improved service delivery. Lower-level gov-
ernments, closer to beneficiary populations, have an informational advantage in
identifying citizens' preferences as well as the flexibility to respond to local condi-
tions. As local govermments act on this information to improve quality, reliability,
and variety of services, consumers will be willing to pay more for services. These
increased user charges can, in turn, be used to finance expansion, improvement,
and maintenance of the existing network.
Evidence from new decentralized approaches confirms that users are willing
to pay for water services that are tailored to their needs. A 1993 World Bank study
found this to be true across income levels. It showed that even poor rural farmers
and households are willing to pay for levels of service above the minimum usu-
ally supplied under centralized systems. However, the key is proper incentives
and demand responsiveness. Users will pay more when assured of choice and
voice in system design and operation.
However, decentralization alone will not resolve all the problems associated
with water services provision. Supporting institutional reforms are important.
For example, the quality and representativeness of institutions to which powers
and responsibilities are devolved matter greatly as elite capture is often a prob-
lem. Reformers must ensure that subnational entities such as provincial and dis-
trict governments and community-based user groups to which new powers are
given make decisions that reflect the will of a majority of their constituents.
Strengthening the decisionmaking processes within decentralized institutions
ensures genuine choice and voice. These in turn are necessary conditions for real-
izing the full benefits of decentralization.
The Evolution of Decentralized Water Services
Water was treated primarily as a social good under the centralized, supply-driven
approach. Its positive health, environmental, and social externalities seemed to jus-
tify the view of both government and communities that free water was a funda-
mental right of the people. However, inadequate resources constrained
governments' ability to fulfill this goal. They struggled to provide a minimum level
- 72 -



Service Delivery
of services with low-cost technologies, and with the center absorbing both capital
and recurrent costs, the approach was unsustainable. Both coverage and quality of
services suffered greatly.2,
Under the centralized approach, water services were usually delivered through
a deconcentrated system, that is, through field offices established within interme-
diate and local jurisdictions, staffed with civil servants (usually engineers) fromrr
the central ministry responsible for water services. These staff were responsible for
delivering water. Engineers devised schemes based mainly on technical consider-
ations such as viability of the water source and area or population to be served;
they did not normally seek the active participation from intended users, and staff
managed systems with little effort to identify or address users' preferences. Not
surprisingly, this approach created few incentives for users to assist government inl
maintaining or financing water services.
Worldwide, the failures of centralization and increasing water scarcity in some
regions created considerable pressure for decentralization. Three new principles,
The Dublin Principles, were issued by the International Conference on Water and
the Environment, held in Dublin in 1992:1
* The Ecological Principle, requiring holistic water management
* The Institutional Principle, requiring participatory water management includ-
ing devolution of responsibility "to the lowest appropriate level" and greater
involvement of NGOs, the private sector, and women
* The Instrument Principle, requiring that water be managed as an economic
resource.
Decentralizing Water Supply, Sanitation, and Irrigation
Three main trends in decentralization of water services have emerged: private sec-
tor participation (PSP), delegation, and devolution.
PRIVATE SECTOR PARTICIPATON. PSP is a spectrum ranging from full privatization to
contracting out for services. In PSP governments hand over all or part of water
service provision to private companies. The approach is used in both irrigation and
water supply and sanitation. In the latter, PSP is increasingly common in the urban
sector and has led to better quality services and higher operational efficiency in
countries such as Argentina and Chile. PSP does not work as well for rural water
supply because private, profit-seeking companies tend to neglect harder-to-reaLch
segments of the population. Rural needs often go unaddressed.
21. Mike Garn, "Managing Water as an Economic Good," in Proceedings of the Commnu-
nity Water Supply and Sanitation Conference, World Bank/UNDP-World Bank Water and Sani-
tation Program, Washington, D.C., May 5-8,1998.
22. J. Briscoe, "Managing Water as an Economic Good: Rules For Reformers," Water Supply
15:4(1997):153-72.
- 73 -



Decentralization Briefing Notes
DELEGATION. Under the delegation model, govermnents transfer water manage-
ment to public or semiprivate water companies.13 These companies are responsible
for providing services within a specified region. As in the case of Tunisia, this ap-
proach has often worked well. However, as in the pure private sector, there are strong
incentives to underserve rural areas. Typically, public and semiprivate companies are
also plagued by the same inefficiencies and incentive problems of other government
agencies, partly because of strong worker rights and few performance incentives.
DEvOLU nON. Two imnportant levels of devolution have evolved in water services
management: Devolution to local governments, and devolution to community-based
user groups. The latter is more common and, depending on the country, is often
incorporated into the first type.
Small-scale irrigation and rural and urban water supply and sanitation are of-
ten devolved to local governments. Ideally, responsibilities vary with capacity: the
strongest local governmnents can undertake activities ranging from interaction with
communities to technical planning to supervising construction. Other local gov-
ernments might focus more on interacting with the communities while relying on
staff from central or intermediate governments for technical support. Sometimes
stronger urban municipalities provide services to neighboring rural areas.
The new push toward participatory management processes has enabled decen-
tralization to user groups. These groups comprise the intended beneficiaries, who
weigh all technically feasible options, consider capital and recurrent cost implica-
tions, make choices, and then manage systems. The approach pays dividends for
both governments and communities: communities get what they need, and gov-
ernments are relieved of the long-term operation and maintenance (O&M) burden.
User groups are common to irrigation and rural water supply and sanitation. Gen-
erally they are referred to as water users' associations (WUAs) in the former and
water and sanitation committees (WSCs) in the latter.24
Dividing Responsibilities
A number of management options ranging from full agency control to full WUA or
WSC control are used in water systems. Forms of joint agency-user management
are the most popular.
INTERMEDIATE AND CENTRAL GOVERNMENT FUNCTIONS. Table 3.2 shows that, except
for complete devolution to user groups, regulatory functions remain the domain of
central agencies. However, in four of the six management options, an agency also
owns the structures and water. As a general rule, the regulatory and operating
functions should always be assigned to different agencies to avoid creating self-
regulating monopolies. Whatever the management arrangement, central agencies
have an important role in establishing the legal and regulatory frameworks with
23. Delegation to the semiprivate utilities can also be considered as a form of PSP.
24. World Bank, "Water Resources Management: A World Bank Policy Paper" (Wash-
ington, D.C., 1993).
- 74-



Service Delivery
Table 3.2. Division of Responsibilities between Agency and Water Users
Full    Agency                         WUA/WSC
agency    O&M       Shared    WUAIWSC  own (agency  Full WUA/
Activity        control  (user input) management   O&M    regulated)  WSC control
Regulation      Agency   Agency     Agency     Agency     Agency   WUA/WSC
Ownership of
structures,
water         Agency   Agency      Agency     Agency   WUA/WSC  WUA/WSC
O&M
responsibility   Agency   Agency    Both    WUA/WSC  WUA/WSC  WUA/WSC
User
representation   Agency WUA/WSC  WUA/WSC  WUA/WSC  WUA/WSC  WUA/WSC
Source: Ashok Subramanian and others, eds., "User Organizations for Sustainable Water Services," Tech-
nical Paper 354 (Washington, D.C.: World Bank, 1997).
respect to water rights, water pricing policies, and environmental standards. Set-
ting laws governing the status of user groups, coordinating various types of users,
and resolving disputes are also important roles which central or intermediate gov-
ernments should assume in order to enable the sector to run smoothly.
Central and intermediate governments also have a financial and technical com-
parative advantage in managing and financing capital-intensive trunk systems. End
users are better able to manage feeder systems that require local knowledge and
ingenuity, local resource mobilization, and enforcement of contracts among second-
ary and tertiary users (for example, irrigation channels, water and sewage networks).
RoLE OF WATER USER GROUPS. Water user groups are usually formed around a
group of potential users, farmers, or rural or urban households to access water
resources. Two types of groups are common, water users' associations (WUAs) in
irrigation, and water supply and sanitation committees (WSCs) in rural (or poor
urban) water supply and sanitation. These groups usually have some sort of le-
gal character (that is, they are legally registered, have elected leadership and a
constitution) and are held together by common interests of members, the public
good characteristics of the service, and the expected gains from collective action.
User groups work well and fully internalize costs and benefits of schemes when
they are forced to bear the cost of failure. Therefore, there must be a credible
threat that public agencies will not rescue user groups and their systems if groups
fail to mobilize the necessary funds for operation and maintenance.
FINANCING. Although high-level capital-intensive systems are usually financed
on a grant basis or with high levels of subsidy by central or intermediate agen-
cies, user charges are increasingly common for operations and maintenance of
feeder systems. Experience suggests that in order for feeder systems to be sus-
tainable, users must commit to paying the full cost of O&M, in cash, kind, or a
combination. User charges are also sometimes used for partial capital cost recov-
ery. Since cost recovery links choice with costs, when presented with all techni-
cally feasible options, users will reconcile their needs with levels and types of
services that they can afford to maintain.
- 75 -



Decentralization Briefing Notes
While O&M should be financed by users, there is considerable good experience
with capital subsidies for water supply, sanitation, and irrigation. Subsidies offset
some of the substantial startup costs associated with water services provision. Gov-
ernments often agree to subsidize capital costs to the value of a certain amount per
beneficiary, or establish a subsidy ceiling per water system. Users then provide any
additional funds required. The subsidy is borne by the level of government with
functional responsibility. Subsidies should be designed with care so as to avoid per-
verse incentives such overbuilding, or inertia while waiting for government funds.
The structure of tariffs used to support operations and maintenance are impor-
tant for determining equity and efficiency. For example, in developing countries,
an increasing block tariff (IBT) is often applied to improve access and discourage
wasteful water use. This is a price structure in which a commodity is priced at a
relatively low initial rate up to a specified volume of use, then at a higher rate for
additional consumption. However, IBTs are usually set too high and subsidize those
who do not need the subsidy. The poor may face a lower rate than the nonpoor, but
utility companies have little incentive to provide the lower-level service." Also, in
many developing countries, households do not have individual water meters (that
is, several households share one meter). Thus the overall impact of the IBT is actu-
ally to worsen equity.26 Informational constraints make it advisable to leave tariff
setting and collection to municipalities because they can better monitor water use
and evaluate users' willingness and ability to pay.
Some Results
Successes of the decentralized approach have been recorded in water supply, sani-
tation, and irrigation. In Mexico, after government transferred the management of
irrigation systems to users' associations, recovery of O&M costs increased from 30
percent to about 80 percent. In Egypt, cropping intensity nearly doubled in farmer
managed systems that were introduced, and user associations also were able to
reduce some environmental irnpacts, such as salinity level of water applications.
There have also been many successes in water supply and sanitation, and some
have been recorded in rural areas, which are usually more difficult to reach. The
Ghana Community Water and Sanitation Project reports that 78 percent of the tar-
get population is now benefiting from improved water sources. Successful water
supply and sanitation programs are also under implementation in Benin, Bolivia,
Ecuador, India, and South Africa, to name a few cases.
25. An alternative approach is for government to subsidize the poor directly with vouch-
ers that they can redeem to the utility.
26. Dale Whittington, "Possible Adverse Effects of Increasing Block Water Tariffs in De-
veloping Countries," in Economic Development and Cultural Change (University of Chicago
Press, 1992).
- 76 -



Service Delivery
Natural Resources Management and Decentralization
Sound natural resources management is an essential part of sustainable develop-
ment. Yet in the past, growth of the rural sector has been associated with a variety of
environmental and resources problems including excessive deforestation and forest
degradation, water depletion, waterlogging and salinization, fish stock depletion,
soil degradation, health effects, and biodiversity losses. These problems are often
associated with inappropriate incentive systems and institutional settings, insuffi-
cient investment in resource-poor areas, inadequate attention to poverty and social
concerns, and political economy systems in which the rich and powerful extract rents.
Intensifying agriculture without damaging natural resources and the environ-
ment is a critical goal for rural development. A growing population requires more
productive agriculture in the short and medium term, while longer-term concerns
for future generations require that these productivity gains not occur at the ex-
pense of natural resource conservation.
The trend toward decentralization-the redistribution of property rights and
bargaining powers among levels of government and groups in society-in many
countries adds a new dimension to this long-standing concern. This section ad-
dresses the key question of how various forms and degrees of decentralization might
affect natural resources management.
Pros and Cons of Decentralization
Many past attempts at sound resources management and conservation have failed
because an overly centralized approach was pursued involving top-down plan--
ning by technicians and bureaucrats without concern for the opinions or well-be--
ing of the people affected by their decisions. Complete decentralization, on the
other hand, can also be counterproductive for sustainable resources management
and conservation. Communities then have no incentive to consider their actions'
effects on their neighbors or successors; under those conditions, local people, per-
haps in cooperation with outside entrepreneurs, may simply degrade and deplete
the resources faster and more efficiently.
Mobilizing local interests can improve a program's chances of success. Decenm-
tralization facilitates local involvement and participation, which are important for
project sustainability because they build local ownership and result in more user-
friendly projects. Decentralization could involve giving local groups the authority to
manage resources and to tax and spend in order to do so. But not all functions of
government can usefully be decentralized, and broader groups may be in a better
position than local ones to appreciate long-term or large-scale issues and to act as
disinterested arbiters of disputes that cannot be solved locally. This outside perspec-
tive is vital in resources management and conservation, which fundamentally conL-
cerns avoiding or managing conflicts of interest arising from externalities.27
27. Ernst Lutz and Julian Caldecott, eds., Decentralization and Biodiversity Consermation: A
World Bank Symposium (Washington, D.C.: World Bank, 1996).
- 77 -



Decentralization Briefing Notes
Empowerment of local groups should therefore be balanced by a continuing role for
central government to deal with market failures and to ensure both social equity and
environmental protection. The challenge is to find the type and degree of decentrali-
zation that is most appropriate for the resource in question.
The Levels of Intervention, Typology, the Unit of Account, and
Extemalities
Sound resources management is taken to mean improving the allocation and use
of natural resources. These resources are consumed at various levels of society,
which can be classified broadly as individual farmers and other local users, indi-
vidual communities, a group of communities (such as in a watershed), regions,
and the national and international community. Many different costs and benefits
are associated with managing resources in a particular way at each level of soci-
ety, and managing them to maximize benefits at one level may impose costs on
other levels. What is internal to one level may be external to another. Thus each
level of society has its own interests in how resources are used, and payments
from one to another could in principle be used to compensate losing parties so
that all sides are satisfied. Such transfer payments may take the form of a rural
development project offering alternative benefits through investments in infra-
structure, education, or other priority projects specified by the community. In re-
turn for such investments, the villagers agree to adhere to natural resources
management plans. Such approaches are pursued, for example, in the Laos Natu-
ral Resources Management Project and the India Eco-Development Project. Simi-
lar compensation arrangements via transfer payments are the rationale behind
many of the integrated conservation-development projects that have been under-
taken in recent years with mixed success.
The larger the extent and geographic spread of negative external effects, the
more there is a need to involve stakeholders and government units from higher
levels in finding better solutions to resources management problems.
FARM AND USER LEVEL. Local farmers and users are making decisions in light of
their own objectives, production possibilities, and constraints. Understanding the
incentives facing individual farmers and resources managers is necessary, there-
fore, if patterns of resources use are to be understood and appropriate responses
to problems formulated.28 Farmers generally seek to maximize the present value
of the stream of expected net returns to agricultural production and resources
exploitation. With regard to adoption of conservation measures, the issue is
whether returns under the optimal path of the new, more conserving system are
sufficiently greater than returns under the path of the current, more degrading
28. Ernst Lutz, Stefano Pagiola, and Carlos Reiche, eds., "Economic and Institutional Analy-
ses of Soil Conservation Projects in Central America and the Caribbean," Environment Paper
8 (Washington, D.C.: World Bank, 1994); Dean Current, Ernst Lutz, and Sara Scherr, eds., "Costs,
Benefits, and Farmer Adoption of Agroforestry: Project Experience in Central America and the
Caribbean," Environment Paper 14 (Washington, D.C.: World Bank, 1995).
- 78 -



Service Delivery
system to justify the cost of switching. Motivational factors other than strict pri-
vate cost-benefit considerations also play a role.
In the context of farm management, unless there are important offsite effects
or the price signals received by farmers are significantly distorted, subsidies to
encourage use of conservation measures would not bring increased economic
efficiency. Pesticides can have serious external effects; public policy should there-
fore encourage integrated pest management. This approach requires less reliance
on the intensive use of external inputs and greater dependence on management
skills and location-specific knowledge of agroecosystems.
Similar to agricultural production decisions, individuals and enterprises defor-
est because it is their most profitable alternative (without considering social costs).
To get these private actors not to deforest in situations where forest clearing is in-
appropriate, deforestation must be made less profitable, or other alternatives, based
on retaining forests or completely outside forest areas, must be made more profit-
able through appropriate incentives and regulations.29
One important element for getting incentives right at the farm and resource
user level is to provide access to land and security of tenure. This is expected to
increase the environmental sustainability of agricultural production and to reduce
rural violence. Therefore, even though land reform is politically sensitive and diffi-
cult to implement, it is important that these issues be tackled.`0
GRoUP, COMMUNrr, AND WATERSHED LEVEL. Conserving or improving natural re--
sources often requires collective action by groups of users or communities.3" Com-
munity-driven development is a process in which community groups initiate, or--
ganize, and take action to achieve common interests and goals. Cooperation is not
necessarily easy to initiate or sustain, even when local organizations are involved.
Multiple uses for any given resource create multiple stakeholders, often with com-
peting priorities. For this reason, conflict among stakeholders over the rights ancd
conditions for use of a given resource is a common feature of resources manage-
ment. Thus, one of the key functions of local organizations is to build the social
29. David Kaimowitz, Neil Byron, and William Sunderlin, "Public Policies to Reduce
Inappropriate Deforestation," in Ernst Lutz, ed., Agriculture and the Environment: Perspectives
on Sustainable Rural Development (Washington, D.C.: World Bank, 1998).
30. Land reform remains a hotly debated issue in a number of countries (Brazil, Colorn-
bia, El Salvador, Guatemala, South Africa, and Zimbabwe are some examples). In Brazil,
Colombia, and South Africa land transfers are made in a decentralized manner, based on
voluntary negotiation and agreement between buyers and sellers; the government's role is
restricted to the provision of a land purchase grant to eligible beneficiaries. (Klaus Deininger,
"Making Market-Assisted Land Reform Work: Initial Experience from Colombia, Brazil, and
South Africa," in Ernst Lutz, ed. , Agriculture and the Environment: Perspectives on Sustainable
Rural Development (Washington, D.C.: World Bank, 1998).
31. See case study for Colombian watershed in Jacqueline Ashby, Edwin Knapp, and
Helle Munk Ravnborg, "Involving Local Organizations in Watershed Management," in Ernst
Lutz, ed., Agriculture and the Environment: Perspectives on Sustainable Rural Development (Was]h-
ington, D.C.: World Bank, 1998).
- 79 -



Decentralization Briefing Notes
capital required for people to build consensus about and enforce the agreed use of
resources for diverse, multiple, and often conflicting purposes.32
Few local organizations at the watershed scale have arisen without some inter-
vention from outside agencies to catalyze their formation and action. At the local
level, outsiders usually play a significant role in enabling stakeholders to arrive at
a joint plan of action that takes into account effects not readily perceived or mea-
sured by those who are not directly affected. Another important point of interface
between local organizations and external agencies in watershed management is
the need for external intervention in identifying the relevant stakeholders and bring-
ing them to the table, whether this is literally a negotiating table or participation in
collective labor, monitoring, or enforcement of sanctions.
Getting local groups and organizations to become self-managing organizations
can extend over several years and does not happen without investment in capacity
building. Local elites often take leadership roles. Although not necessarily bad,
this can result in the hijacking of resouTces unless transparency and accountability
are somehow enforced." Case studies show that women, the landless, marginal
ethnic groups, and the laboring poor are unlikely to be represented in participatory
resources management. Therefore, natural resources development projects should
deliberately seek to overcome this bias not only for reasons of social equity but also
because of productivity.
Local organizations can often be effective in generating and securing compli-
ance with rules for the use of common property such as water, common grazing
land, or forest, and the management of buffer zones around conservation areas, all
of which may be important features of watershed management. Devolving respon-
sibility to local organizations can both externalize some costs of enforcing conser-
vation from the state to local communities and also reduce costs overall, by creating
conditions in which outside agencies become more efficient and effective through
their collaboration with local organizations.
NATIONAL LEVEL. Even in relatively decentralized settings, sound national level
policy, institutional frameworks, and enforcement of laws remain of great impor-
tance.34 An important element of sound national level policymaking is to give fiscal
as well as administrative authority to the lowest level of government that can effi-
ciently deal with a resources issue within its jurisdiction. This includes supporting
the principle of compensation for external damages caused. Where there remain
resources management problems involving issues between regions, national level
32. Jacqueline Ashby, Edwin Knapp, and Helle Munk Ravnborg, "Involving Local Or-
ganizations in Watershed Management," in Ernst Lutz, ed., Agriculture and the Environment:
Perspectives on Sustainable Rural Development (Washington, D.C.: World Bank, 1998).
33. Deepa Narayan, 'Participatory Rural Development," in Ernst Lutz, ed., Agriculture
and the Environment: Perspectives on Sustainable Rural Development (Washington, D.C.: World
Bank, 1998).
34. Dan Bromley, "Property Regimes in Economic Development: Lessons and Policy
Implications," in Ernst Lutz, ed., Agriculture and the Environment: Perspectives on Sustainable
Rural Development (Washington, D.C.: World Bank, 1998).
- 80 -



Service Delivery
institutions should assist by creating an atmosphere and venues for discussion be-
tween representatives. Finally, national level authorities often have an important
positive or negative influence on resources management through decisions regard-
ing the construction of new roads and maintaining existing ones. In the Amazon,
for example, improved infrastructure has decreased the costs of resources extrac-
tion, thus making more areas vulnerable to exploitation.
GLOBAL LEVEL. Some local actions cause global externalities. These include, for
example, loss of biodiversity from deforestation and global warming due to the
carbon released in forest fires. Mechanisms have been and are being developed to
provide incentives for local resources users to adopt changes. The Global Environ-
ment Facility and the World Bank's Carbon Initiative are such examples.
Conclusion: No Simple Recipes
Decentralization and natural resources management at various levels are complex
processes that interact with one another in many ways. Resource and environmen-
tal problems and management issues are quite different for the array of conditions
that one encounters, and this variability demands flexible institutions and actions.
Sustainable resources management is not a rigid model; rather, it is a process in-
volving widespread experimentation, learning by doing, sharing of information,
innovation, group problemsolving, and an enhanced capacity to adapt to unex-
pected changes and emerging uncertainties.
Although many lessons can be learned and principles derived from experience, the
last word on what works (and what does not work) is by no means decided. Indeed,
the dynamic nature of the process of involving various levels in resources manage-
ment is one of its most important characteristics and requires focusing as much on the
management of the process as on defining a blueprint for how to go about it.
Decentralizing by giving new responsibilities to regional, watershed, and local
government units, to communities, and to NGOs creates both opportunities and
potential problems for resources management. To take advantage of the former
while avoiding the latter, a cluster of arrangements must be made as a whole if
natural resources management is to work well in a moderately decentralized set-
ting. Of these, six institutional and incentive elements merit special attention:
* Local participation, especially in a way that allows local people to understand
and endorse the boundaries and management plans of forests and nature re-
serves and that promotes clear tenure over land and other resources.
* Capacity building, especially to increase skills and accountability, among local
government units and NGOs to enable them to work together to promote im.-
proved resources management, conservation, and rural development.
* incentive structures that properly address the externality issues at the appro-
priate levels; this could include conditional subsidies, especially where diver-
gent costs and benefits of conservation are experienced by local and nonlocal
groups, making it necessary for global and national society to bridge the gap
with livelihood investments or grants.
- 81 -



Decentralization Briefing Notes
* Appropriate enforcement, especially against powerful local or central interests
and always in the context of education and public relations activities.
*  Stakeholder forums, which need decisionmaking power and fiscal authority to
fulfill their three main roles of avoiding conflict through dialogue, authorizing
action, and requesting help from outside organizations to meet local develop-
ment priorities.
* Policies, laws, and institutions that create incentives at the local level for sound
resources management and reduce the need for costly monitoring and enforce-
ment action by the central government.
- 82 -



4
Potential Impacts of Decentralization
This chapter contains four notes that review the existing evidence on the impact of
different forms of decentralization on equity, macroeconomic stability, economic
growth, and accountability, transparency, and corruption.
Equity and Multileveled Government
Should governments redistribute to people or places? If so, which level of government
should be responsible for the redistribution, and under what conditions? And how can
governments attain these distributional goals while minimizing economic distortions?
What has been the experience with decentralization and its impact on equity?
These questions are priority policy areas in industrial, developing, and transition
economies. For example, in recent years there has been a substantial devolution in
distributional responsibilities for welfare payments from the federal to the state gov-
ernments in the United States. Economic integration and increased mobility in Eu-
rope raise questions about these national governments' ability to redistribute income
effectively within their boundaries. Developing countries are often concerned with
large variations in the relative revenues of subnational governments.
Assignment of Responsibility for Interpersonal Redistribution
of Income
THE THEORY. The conventional wisdom from classic sources in the literature im-
plies that redistribution has to be carried out by higher levels of govermment.1 Fac-
tor mobility (particularly labor) will make attempts by lower jurisdictions to change
the distribution of income self-defeating as the poor gravitate to areas of high re-
distribution, while the rich cluster in areas of low redistribution.
THE REALITY. The world does not fully correspond to this theoretical model, how-
ever. Many functions of subnational govermments do, in practice, have significant
distributional effects: regulatory policies on land use and rent controls have pro-
found distributional implications, for example. Some frequently subnational func-
tions-most notably health and public education-also affect distribution.
Infrastructure services are another area in which local govermment policies
can have important redistributive effects, both progressive and regressive.
Throughout the world, infrastructure services are one of the core areas of respon-
sibility of local governments, an indication of the important role of this level of
1. Wallace E. Oates, Fiscal Federalism (New York: Harcourt Brace Jovanovich, 1972), pp.
6-8; R. A. Musgrave, The Theory of Public Finance (New York: McGraw-Hill, 1959).
- 83 -



Decentralization Briefing Notes
government in attaining efficiency goals. One implication of such efficiency goals
is that the appropriate method of paying for these utility and infrastructure ser-
vices is by cost recovery through user fees, so that the price charged for these
services reflects their value to consumers.
Departures from the user cost pricing principle are generally an unsuitable
means of achieving other goals of government, such as the distributional objec-
tive of helping those with low incomes. For instance, subsidies for utility use
always benefit those who are better off more than others, because utility use rises
with income. Algeria's 1988 expenditure survey found that the richest 10 percent
of households spent four times as much on subsidized urban transport as the
poorest 10 percent. Subsidies on fuels and urban transport in Algeria were cited
in the Bank's World Development Report 1994 as an illustration of the regressive
effects of infrastructure subsidies. Better targeted and more effective transfer in-
struments are available to help low-income groups.
There is some evidence that the formal assignment of distributional authority to
a local government can work. In Switzerland, cantons have jurisdiction over health,
education, and welfare services and have priority in levying taxes on personal in-
come and wealth. In some nonfederal states such as Sweden, the constitution for-
mally gives local governments the right to levy income taxes. In fact, many attribute
the relatively large size of local governments in Scandinavian countries to their sub-
stantial responsibilities in distributional programs. In Denmark, for example, local
governments account for over half of general government expenditures and about a
third of gross national product, while social security and welfare payments account
for more than half of these large local government budgets ,2 Local distributional func-
tions are often substantially financed by local income tax rates that are piggybacked
on broadly redistributive tax rates of the central government.
There is also plenty of evidence that subnational jurisdictions systematically in-
corporate their own particular distributional preferences into choices on spending
decisions.3 One study shows that monthly average benefits for Aid to Families with
Dependent Children vary by a factor of five among U.S. states. Noting that such
differences have persisted over long time periods, the study asserts that "decentral-
ized redistribution is a fact of life that must be dealt with as a practical matter."
REcONCILING THEORY AND REALITY. Although debate on this subject continues, the
traditional view that central governments have the major role to play in functions
of income redistribution still prevails. Depending on how much a nation values
equity, the central government will need to play a redistributive role to raise wel-
fare for poor people in poor areas. Some of the most recent contributions to the
literature, however, recognize that local preferences for redistribution do matter.
2. Giancarlo Pola, George France, and Rosella Levaggi, eds., Developments in Local Gov-
ernment Finance: Theory and Policy (Cheltenham, U.K.: Elgar, 1996), contains several essays
on local governments' revenues and responsbilities in Scandinavian countries.
3. David E. Wildasin, "Income Distribution in a Common Labor Market," American
Economic Review 81(1991):757-74.
- 84 -



Potential Impacts of Decentralization
And while the traditional idea that the central government should control or at
least coordinate redistribution remains, there are some interesting new ideas for
approaches which incorporate subnational levels. In particular, Wildasin argues
that an optimal solution in multilevel systems of government is to provide grants
that, because of differing subnational preferences for redistribution, are nonuni-
form by subnational jurisdiction, but lead to identical levels of transfer payments.4
This researcher finds examples of behavior in multilevel systems of government
that support the analysis and policy prescription.
Factor mobility, the largest theoretical constraint on subnational redistribution
activities, may not be so large a concern in reality. First, while empirical studies
broadly indicate that interregional factor flows are influenced by the potential for
monetary gain, such flows are usually relatively small. Second, barriers to mobility
may be significant. For example, in many of the countries in transition, labor mo-
bility is limited because of lack of housing. Because housing is often the responsi-
bility of local governments in these countries, local redistributional policies (or the
absence thereof) acquire added importance.
A BENEFIT OF SuBNAnONAL REDISTRIBuTION. While the goal of reducing poverty is
explicit in the objectives of the World Bank and much other public policy, such
redistributional policies rest on value judgments that are not shared by all. Those
who place greater priority on goals of individual freedoms and fear the arbitrary
use of power see limits on the ability of predatory governments to carry out redis--
tribution as important benefits of migration and federal structures of government.5
Interregional Transfers: An Often Overlooked Issue
Interregional redistribution has received less attention at the World Bank. A long-
standing complaint in economic policy is that redistribution to poor regions may
interfere with natural adjustment mechanisms such as outmigration that will oth-
erwise rectify regional imbalances. Nevertheless, this view needs to be modified to
take into account the important variation in net fiscal benefits from the fiscal activi-
ties of subnational governments.
VARIATION IN NET FISCAL BENEFrIS. Variation in net fiscal benefits may arise for sev-
eral reasons. Two of the most common are:
* Residents with high per capita incomes-Subnational governments whose
residents have high per capita incomes may finance a given amount of public
services with less tax effort (such as lower tax rates on income) than govern-
ments of poorer regions.
4. David E. Wildasin, "Income Distribution in a Common Labor Market," American
Economic Review 81(1991):757-74.
5. Geoffrey Brennan and James M. Buchanan, The Power to Tax: Analytical Foundations of
a Fiscal Constitution (Cambridge, Mass.: Cambridge University Press, 1980); Charles E. McLure,
Jr., "Tax Competition: Is What's Good for the Private Goose Good for the Public Gander?"
National Tax Journal 39(1986):341-48.
- 85 -



Decentralization Brefing Notes
* Possession of or access to natural resources revenues-The potential for in-
come disparities from this source can be striking in some countries. The Tyumen
oblast in Russia, for example, accounts for two-thirds of Russia's oil but only 2
percent of its population.
EFFICIENCY ARGUMENTS. The efficiency argument centers on the presumption
of factor mobility. A classic example concerns fiscally induced migration, for
example, if people relocate next to provincial oil wells simply to share in the
jurisdiction's ownership rights. If such fiscally induced migration is significant,
it would be in the interests of jurisdictions where such net fiscal benefits are
substantial to make transfers to low-income jurisdictions so as to limit the
amount of such migration. There would also be a net benefit to all in the nation
because the real costs of migration would be avoided.
There is some doubt as to the significance of such efficiency losses, however. A
study in Canada, for instance, calculated that there would have been a welfare gain
of only $35 million from shifting billions of dollars around to equalize provincial
government fiscal capacities, a proxy for net provincial benefits from subnational
governments.6 As a consequence, the equitable case for redistribution of net fiscal
benefits arising from subnational government activities probably deserves more
weight than that arising from their efficiency costs.
EQUrrY ARGuMENTs. The equity problem for a central government's tax and trans-
fer system arises because of the horizontal equity goal of treating citizens alike
no matter where they reside. The problem is that central government attempts to
attain interpersonal equity are implemented through tax and transfer systems
that deal only with the market or taxable incomes of individuals, but the net fis-
cal benefits from subnational government activity may lead to substantial differ-
ences in the real or comprehensive incomes of citizens in different areas. It is
difficult to think of practical ways for a central goverrunent to deal with the fiscal
benefits from subnational govermments by adjusting its own personal taxes and
transfers. It would be impractical and politically unacceptable to have different
national personal income tax rates in different subnational jurisdictions, for in-
stance, and such a policy would also override legitimate subnational redistribu-
tion policies. One practical solution is to undertake intergovernmental transfers
to equalize subnational per capita fiscal capacities so as to make it financially
possible to achieve horizontal equity goals.
EXTENT OF EQUALIZATION PROGRAMS. This reasoning may help to explain why pro-
grams to equalize subnational per capita fiscal capacities are common throughout
the world. Programs of equalizing state and provincial fiscal capacities may be
more the rule than the exception in industrial countries with federal systems; they
6. William G. Watson, "An Estimate of the Welfare Gain from Fiscal Equalization," Ca-
nadian Journal of Economics 19(1986):298-308.
- 86 -



Potential Impacts of Decentralization
exist, for example, in Australia, Canada, Germany, and Switzerland. Programs to
reduce tax-base disparities among local governments are also common in both
unitary and federal systems. In developing countries, equalization is a factor in the
general revenue sharing transfers to subnational governments that exist in such
countries as Brazil, Chile, Colombia, India, Mexico, Morocco, Nigeria, and Paki-
stan. Economies in transition have also realized that revenue sharing based on the
origin of revenues (the derivation principle) results in a very unequal distribution
of subnational govermnent resources. Several countries of the former Soviet Union,
the largest in population (Russia) to the smallest (Estonia), as well as Hungary,
Poland, and other Baltic countries, have instituted programs to reduce disparities
in the per capita revenues of subnational governments.7
TRANSFERS BETWEEN SUBNATIONAL GOVERNMENTS. It may be inappropriate to finance
such equalization transfers through central govern-ment revenues, if the revenues
are raised from different sources than those received by subnational governments.
The usual principles of tax assignment, whereby higher-level governments tax more
mobile resources and local governments tax more immobile resources such as prop-
erty, will automatically result in such differentiation of revenue sources by level of
govermment. Another example is when subnational rather than national govern-
ments own natural resources that yield substantial revenues. This problem assumed
serious proportions in Canada following oil price increases in the 1970s that raised
the revenues of oil-producing provinces. The government of Canada was respon-
sible for the equalization of provincial government fiscal capacities but did not
have access to these oil revenues. In effect, redistribution of revenues to poorer
provinces resulting from higher oil prices was being financed by transfers from
federal taxation raised in the richer nonoil-producing provinces.
The straightforward method of dealing with this problem is for direct redistribu-
tion to take place between subnational governments, where equalization to the de-
sired standard of fiscal capacity is financed by "have" subnational governments
contributing to a revenue pool that is then redistributed to "have not" subnational
governments. Two of the best-known examples are in the Federal Republic of Ger-
many and Switzerland. There are also examples of voluntary direct transfers between
local governments, as in the tax-base sharing among 201 municipalities in the Min-
neapolis-St. Paul area. Alternatively, some central governments require such equal-
ization transfers to be undertaken between local governments, as in Denmark and
Sweden. Chile also has a system of direct transfers from rich to poor municipalities.
Impact of Decentralization on Equity
Many nongovermental organizations (NGOs) and international donors encour-
age countries to decentralize because they believe it will lead to greater local par-.
ticipation, which will benefit communities, households, and particularly the poor.
7. For essays on equalization grants as well as other forms of transfers, see Ehtishamn
Ahmad, ed., Financing Decentralized Expenditures: An International Comparison of Grants, Stud-
ies in Fiscal Federalism and State-Local Finance (Cheltenham, U.K., and Lyme, N.H.: Elgar, 1997).
- 87 -



Decentralization Briefing Notes
This is not necessarily the case. This will depend on how much emphasis the
central and local govemments place on redistribution, how the system of inter-
governmental finance is designed, particularly transfers (see the sections on ex-
penditure, revenue, and transfers) and the extent to which participation in local
decisionmaking is truly broad-based (see the section on participation). Although
evidence is scant, decentralization seems to be more "pro-poor" when under-
taken at the local rather than intermediate level of government (see the section on
safety nets). This permits funds to reach the poorer areas and encourages greater
local accountability.
- 88 -



Potental Impacts of Decentralization
Macroeconomic Impact of Decentralization
The design of decentralization can significantly influence how the restructuring of
decisionmaking and responsibility will affect a country's macroeconomic conditions.
Some countries, including Brazil and China, experienced macroeconomic problems
when tax bases were decentralized without a clear assignment of expenditures to the
level of government that receives the revenues.8 Others, including Argentina and
Mexico, ran into problems, because subnational governments accrued unsustainable
debts and had to be bailed out by the central government. Decentralization efforts in
which governments solved their fiscal imbalances at the center by decentralizing
expenditure responsibilities without the matching revenues left essential services
unperformed and, in some cases, gave rise to unsustainable subnational deficits.
The design of intergovernmental fiscal relations needs to recognize the possibility
of these negative effects. Most important, it must ensure a match between expenditure
responsibilities and revenues at each level of government and create institutional mecha-
nisms that will enforce a hard budget constraint between levels of government.
Subnational Incentives
Compared with central governments, subnational governments have less incen-
tive to consider the macroeconomic impact of their policies. Such impacts tend to
leak away to other jurisdictions, giving subnational macroeconomic stability poli-
cies public goods characteristics. Central governments are therefore usually assigned
the task of maintaining stability, and they should have the tools to go with it, such
as control over monetary policy and at least some control over fiscal conditions.
That includes a share of revenues and expenditures sufficiently large and suffi-
ciently flexible to influence aggregate demand in a country. However, whether in-
tended or not, subnational policies can influence stability: even balanced-budget
spending increases by subnational governments can affect macroeconomic stabil-
ity, and subnational borrowing can become a major macroeconomic concern. Yet
because macroeconomic shocks can hit different jurisdictions differently, some
subnational influence on macroeconomic conditions is actually desirable. Finally
centralizing all expenditures and revenues with a potential macroeconomic impact
may have excessive efficiency costs.
Structuring Intergovemmental Fiscal Relations to Enhance Stability
REVENUE ASSIGNMENT AND TAx SHARING. Subnational goverunents should have a
fairly stable tax base, because highly income-elastic tax bases (such as the value-
added tax (VAT) and progressive income taxes) can induce procycical government
expenditures that aggravate macroeconomic imbalances. Central governments hav e
tended to be more responsible in this respect. Moreover, taxes such as the VAT and
income tax lend themselves well for use as macroeconomic policy tools (increasing
rates to cool down an overheating economy, for example).
8. William Dillinger, "Brazil's State Debt Crisis: Lessons Learned," Economic Notes 14,
Country Departnent 1, LatinAAmerica and the Caribbean (Washington, D.C.: World Bank, 1997).
- 89 -



Decentralization Briefing Notes
EXPENDITURE ASSIGNMENTS. Although its discussion of this topic is limited, the pub-
lic finance literature usually recommends assigning sufficiently stable expenditures
to subnational governments. For instance, many analysts argue that social security
is better assigned to the national level because of its procycical nature. See the
section on expenditure assignment.
Incentives, however, may moderate this standard conclusion. For instance, while
assigning unemployment compensation to central government may create an "au-
tomatic stabilizer," it also takes away the incentive for subnational governments to
hold down unemployment expenditures by, for instance, following a proactive
employment policy. Thus the centralization of social security may actually increase
macroeconomic imbalances, while making local governments responsible for a part
of these highly cyclical expenditures may reduce instability.
GRANTrs. Macroeconomic considerations-as well as tax administration and eq-
uity considerations-are likely to give central government the larger part of rev-
enues but not necessarily expenditures. Thus, a system of grants will be needed to
fill the vertical imbalances. Although efficiency and equity considerations are likely
to dominate the process of grant design, several features could be beneficial to
macroeconomic stability. See the note on intergovernmental transfers and grants.
* Grants should be designed to fill the ex ante gap between subnational revenues
and expenditures. Filling ex post gaps could give rise to excessive expenditures
at the local level, as central government will foot part of the bill.
• The size of the grant pool should be fairly independent of macroeconomic
conditions, to avoid the procyclical spending patterns noted earlier. Having a
broad revenue base that feeds the grant pool is therefore to be preferred over
single-tax-based grants.
* Central government needs some discretion to adjust the amount of grants trans-
ferred. Although this is often impossible for general grants mechanisms, which are
usually based in law, special-purpose grants offer more flexibility. Their expendi-
ture effects can even be leveraged, if the grants require subnational matching.
SUBNATIONAL BORROWvNG. Subnational borrowing can have a significant effect on
macroeconomic conditions in a country. Countries such as Brazil and Mexico saw
their restrictive central fiscal policies of the early 1990s undermined by subnational
deficits (either open or hidden), resulting in the end in a bailout by the national
government. However, intertemporal efficiency speaks strongly for allowing
subnational governments at least some access to borrowing. Countries have there-
fore taken a number of approaches in allowing subnational borrowing. See the sec-
tion on subnational borrowing.
Reliance on market discipline to limit borrowing requires that several strict con-
ditions be fulfilled, among them open capital markets, adequate information, re-
sponsiveness of the borrower to market signals, and a strict no-bailout policy by
the central government. Except for some federal countries (the United States and
Canada, for example), few countries rely solely on market discipline.
- 90 -



Potential Impacts of Decentralization
A number of countries (Germany, Switzerland, Spain, the Republic of Korea)
rely on legally binding rules to restrict borrowing. Some have used the golden rule
of allowing borrowing only for investment purposes. Others use rules that restrict
borrowing to some indicator of debt-servicing capacity.
Some countries directly control subnational borrowing by setting overall lim-
its, approving individual debenture issues, or allowing borrowing only through a
centrally controlled municipal bank. Local governments have ways to circumvent
these regulations, including reclassifying expenditures from current to capital, run-
ning arrears, creating off-budget activities, and using state enterprises for borrow-
ing or providing guarantees for enterprise borrowing. Thus, any rule should have
sufficient incentives and controls to enforce it.
COORDINATION MCHANISiMS. In recognition of the influence of subnational gov-
ernments on stability, several countries have designed intergovernmental macro-
economic coordination mechanisms. Germany, which probably has the most elabo-
rate one, coordinates expenditure and borrowing plans within the context of the
Stabilitaetsgesetz (stability law). The five-year fiscal plan is discussed and coor-
dinated with the states before submission to parliament. In Australia, the Loan
Council now coordinates the borrowing requirements of the national and
subnational public sector, after previously serving as a forum to limit subnational
borrowing. In China, the annual planning and finance conference is used to dis-
cuss and coordinate the plans of each level of government. And many countries
have informal coordination mechanisms, either through state or provincial rep-
resentation in a house of parliament or through special subcommittees of parlia--
ment. Whether or not coordination without binding rules actually delivers better
results is an open question.
Regulating Incentives by Design
In designing intergovernmental fiscal relations, the incentive effects on macroeco-
nomic stability should be explicitly considered. Large vertical imbalances in favor
of national governments are likely to lead to ex post gap filling by national govern-
ments, bailouts of subnational debt, or circumvention of national policies on taxes
or debt. Large imbalances in favor of subnational governments, on the other hand,
could lead to large national debts and to insufficient fiscal discipline at subnational
levels. Little tax autonomy of subnational govermments reduces the incentive to
behave responsibly, because they cannot be required to increase taxes to restore
balance on the budget. Thus, other things equal, a broad matching of expenditure
and revenue responsibilities, together with an explicit responsibility for each level
of government to live within its means and the autonomy to do so, is desirable.
Rules for budget management for subnational governments could address marLy
of the detrimental macroeconomic effects of decentralization. As long as subnational
governments are held to responsible management-including limits on deficit
spending-and provided the means and incentives to actually do so, the macro-
economic impact of their actions can be limited. Having harsh measures for use by
the central government in dealing with fiscal irresponsibility could help, such as
- 91 -



Decentralization Briefing Notes
bankruptcy procedures for local governments in New Zealand. Transparent fiscal
accounts that are regularly published and audited can add to discipline. Also, in
the end, one of the best incentives for responsible behavior is likely to be the pres-
sure from an electorate that would suffer from a fiscal crisis.
- 92 -



Potential Impacts of Decentralization
Decentralization and Economic Growth
We know very little about the relationship between decentralization and growth.
Empirical evidence for the way in which decentralization affects growth has been
contradictory and has had problems with measurement, specification, and analy-
sis. There is stronger evidence for a relationship in the other direction-from growth
to decentralization-but the interpretations of this correlation between high in-
come and decentralization have varied.
Decentralization and Growth: The Empirics
The few empirical studies that have directly examined the relationship between
decentralization and growth come up with mixed results. Zhang and Zou show
that decentralization has a positive and sometimes significant effect on regional
economic growth in India. Their work on China and Davoodi, Xie, and Zou's un-
published work on the United States, however, find fiscal decentralization to be
associated with lower growth. In a larger sample of developed and developing
countries, Davoodi and Zou found that decentralization has a negative relation-
ship to growth in developing countries and no discernible effect on growth in de-
veloped countries. Several methodological problems in these studies discount even
these mixed results and much more needs to be done to ensure that the measured
decentralization-growth relationship is robust.9
In the absence of strong, unambiguous empirical evidence, researchers have
put forward three hypotheses about the relationship between decentralization and
growth. In each hypothesis, growth has only a secondary relationship to decen-
tralization, and the nature of the connection, whether growth-enhancing, growth-
impeding, or growth-requiring, depends on what are considered to be the primary
effects of decentralization. These primary effects, in turn, have much to do with the
specific design of decentralization policy.
While other decentralization notes discuss the methods for reaching certain
primary effects, ideally, more efficient, responsive delivery of services, this sectiorn
outlines the ways in which the commonly discussed primary effects of decentrali-
zation can shape decentralization's effect on growth and desirability in developing
countries. Three alternative hypotheses for how decentralization can indirectly af-
fect growth are presented in the following paragraphs.
HYPoTHEsIS 1-DECENTRALIZATION INCREASES ECONOMIC EFFICIENCY IN PUBLIC SPEND-
ING; THEREFORE ITS DYNAMIC EFFECTS SHOULD BE GROWTH-ENHANCING. Local governments'
9. See Tao Zhang and Heng-fu Zou, "Fiscal Decentralization, Public Spending, and
Economnic Growth in China," Journal of Public Economics 67:2(1997):221-40; H. Davoodi, D.
Xie, and H. Zou, "Fiscal Decentralization and Economic Growth in the United States," 1995
(unpublished); T. Zhang and H. Zou, "Fiscal Decentralization and Economic Growth: A Cross-
Country Study," Journal of Urban Economics 43(1996). A critical summary of the evidence is
available in J. Martinez-Vazquez and R. McNab, "Fiscal Decentralization, Economic Growth,
and Democratic Governance," paper prepared for the U.S. Agency for International Devel-
opment Conference on Economic Growth and Democratic Governance, Washington, D.C.,
October 9-10, 1997.
- 93 -



Decentralization Briefing Notes
information advantage allows them to respond better than central governments
to constituents' preferences. Central governments would have to invest far more
resources to obtain the same information. Although constituents' preferences do
not always serve as a guide for productive investment, they can be useful as re-
views of program design and as guidelines for expenditure allocation.
Citizens in decentralized systems are able to move across jurisdictions to settle
where the available package of public goods matches their preferences. High mo-
bility may force subnational governments to be more competitive, thus more effi-
cient and possibly innovative, in providing services.
It is an intuitive leap from these stylized effects to the dynamic effect of in-
creasing growth. Several studies of public expenditure and growth show that
certain kinds of public expenditure have more significant growth-enhancing ef-
fects than others. Decentralization might improve the quality of proven growth-
enhancing inputs such as education.
HYPOTHEsIs 2-DECENTRALIZATION CAN LEAD TO MACROECONOMIC INSTABILT, WHICH
CAN INHiBIT GROWTH. Three arguments underlie this hypothesis. First, decentraliza-
tion reduces the tax and expenditure base that the central government can use to
carry out its stabilization function. Subnational governments are incapable of tak-
ing over this national macroeconomic stabilization because the borders of their in-
dividual jurisdictions are open to migration of people and goods. The practical
imnportance of this effect, however, depends on the extent to which the central gov-
ernment actually uses these bases for stabilization.
Second, poorly designed decentralization policies can create incentives for
subnational governments to spend irresponsibly and unsustainably. A hard bud-
get constraint between levels of government is essential. Loosely defined, unpre-
dictable transfer arrangements can lead to fluctuating and unsustainable subnational
government spending, particularly in cases where the fiscal gap can be filled by
subnational borrowing. Policymakers should explicitly consider ways to create or
reinforce institutions and incentive structures that prevent worst-case scenarios such
as deficit-increasing central government bailouts.'0
And third, decentralization of revenues and expenditures to subnational gov-
erunents can take scarce resources away from national projects, such as infrastruc-
ture, that could make greater contributions to growth.
The evidence concerning decentralization's destabilizing effect is mixed, how-
ever."' Research in China and Russia has clearly linked subnational government
10. David Wildasin, "Externalities and Bailouts: Hard and Soft Budget Constraints in
Intergovernmental Fiscal Relations," Policy Research Working Paper 1843 (Washington, D.C.:
World Bank Development Research Group, 1997), discusses the incentives operating at vari-
ous levels of government.
11. For further discussion of these themes and examples, see the note on macroeconomic
stability; Charles McClure, "Comment on Prud'homme," World Bank Research Observer
10(1995):221-26; Remy Prud'homme, "On the Dangers of Decentralization," World Bank Eco-
nomic Review 9(1995); David 0. Sewell, "The Dangers of Decentralization According to
Prud'homme: Some Further Aspects," World Bank Research Observer 11:1(1996):143-50; Vito
- 94-



Potential Impacts of Decentralization
autonomy to growing central government deficits, but these cases may not be gen-
eralizable because of their unique system of using local governments to collect cen-
tral government revenue. Other, more typical subnational governments in Argentina
and Brazil have followed the destabilizing path of expanding expenditures beyond
revenues and then turning to the central government for bailouts or debt relief.12
Switzerland, on the other hand, has been both highly decentralized and highly
stable in the post-war period. Research on Canada has shown that increasing
subnational governments' budgets may have actually had a stabilizing effect.
Subnational governnents have coordinated spending and may be better able to
offset region-specific business cycles.
HYPOTHEsIs 3-DEVELOPING COUNTRES HAVE SIGNICANTLY DIFFERENT INSTITUTIONAL AND
ECONOMIC ENVIRONMENTS THAN DO INDUSTRIAL COUNTRIES AND WILL NOT REAP THE BENEFITS
OR SUFFER THE CONSEQUENCES OF DECENTRALIZATION IN THE SAME WAYS.13 Institutional ar-
rangements in developing countries do not necessarily give subnational govern-
ments the incentive to use their information advantage in responding to their con-
stituents. Leaders may be appointed or elected on the basis of their social standing.
Further, subnational governments in developing countries may not have the eco-
nomic resources to attract sufficiently trained people or the human resources to
manage larger budgets. The decentralization of Peru's water supply and sanitation
sector, for example, reportedly reduced the quality of service because local govern-
ments did not have the technical ability to maintain or plan infrastructure. Never-
theless, Colombia's success in decentralizing some services suggests that these ability
constraints can be overcome by private-public partnerships."4
Conclusion: Design Matters
There is no clear, automatic relationship between decentralization and growth. Di-
rect empirical studies have not been satisfying, and a return to the literature on
decentralization's primary effects may be a more useful way to think about the rela-
tionship between growth and decentralization. From this point of view, the design of
decentralization becomes the key factor in determining whether policies will lead to
the efficiency linked to higher growth, exacerbate the deficits and instability con-
nected to lower growth, or simply become mired in institutional constraints.
Tanzi, "Fiscal Federalism and Decentralization: A Review of Some Efficiency and Macroeco-
nomic Aspects," in Proceedings of the Annual Bank Conference on Development Economics 1995,
(Washington, D.C.: World Bank, 1995).
12. William Dillinger, "Brazil's State Debt Crisis: Lessons Learned," Economic Notes 14,
Country Department 1, Latin America and the Caribbean (Washington, D.C.: World Bank;,
1997).
13. This hypothesis is particularly associated with Roy Bahl and Johannes Linn, "Fiscal
Decentralization and Intergovernmental Transfers in Less Developed Countries," Publius:
The Journal of Federalism 24(Winter 1994):1-19.
14. Ariel Fiszbein, "Emergence of Local Capacity: Lessons from Colombia," World De-
velopment 25:7(1997):1029-43.
- 95 -



Decentralization Brefing Notes
Accountability, Transparency, and Corruption
in Decentralized Systems
Decentralization, in its democratic political aspect, has two principal components:
participation and accountability. Participation is concerned chiefly with increasing
the role of citizens in choosing their local leaders and in telling those leaders what
to do-in other words, providing inputs into local governance. Accountability is
the other side of the process; it is the degree to which local govermnents have to
explain or justify what they have done or failed to do. Improved information about
local needs and preferences is one of the theoretical advantages of decentralization,
but there is no guarantee that leaders will actually act on these preferences unless
they feel some sort of accountability to citizens. Local elections are the most com-
mon and powerful form of accountability, but other mechanisms such as citizen
councils can have limited influence.
Accountability can be seen as the validation of participation. The test of whether
increased participation is worthwhile is if people can use the participatory chan-
nels to hold a local government responsible for its actions.
Types of Accountability
Accountability comes in two dimensions: accountability of government work-
ers to elected officials and accountability of the elected officials to the citizens
who elect them.
GovERNMENT WORKERS TO LOCAL OFPICIALS. The first type of accountability can prove
difficult to achieve since civil servants, particularly professionals in such fields as
health, education, and agriculture-the very sectors that are most commonly de-
centralized-often have considerable incentive to evade control by locally elected
officials. Such people generally have university training and sophisticated lifestyle
practices that are hard to maintain in small towns and villages, career ambitions
that transcend the local level, and goals for their children's education that local
schools cannot meet. They may also well fear that quality standards. for service
delivery will suffer if provision is localized. Finally, they often find opportunities
for corruption are greater if they are supervised by distant managers through long
chains of command than if they must report to superiors close at hand. For all these
reasons, they tend to prefer strongly maintaining ties with their parent ministries
in the central government and resisting decentralization initiatives. And under-
standably, their colleagues at the center have a parallel interest in maintaining these
ties, for they are much concerned about preserving national standards in service
delivery and often about avoiding opportunities for venality (many corruption
schemes provide for sharing ill-gotten gains upward through bureaucratic chan-
nels to the top). See the section on the civil service for more about this subject.
Given all these reasons both good and bad for opposition, it is scarcely surpris-
ing that decentralization initiatives so often run into heavy bureaucratic resistance
and that designers find themselves pressured to keep significant linkages between
the field and the central ministries, especially concerning such issues as postings,
-96 -



Potential Impacts of Decentralization
promotions, and salaries. Needless to say, such ties tend to undercut the capacity of
elected officials to supervise government servants who are supposedly working
for them. Some decentralized governance systems (for example, Karnataka State in
India) appear to have worked through these problems to establish popular control
over the bureaucracy, but it has taken many years to do so.
ELECTED LEADERS ro THE ClTIZENRY. The second type of accountability is that of
elected officials to the citizenry. Elections (provided they are free and fair) provide
the most obvious accountability, but this is a rather blunt tool, exercised only at
widespread intervals and offering only the broadest citizen control over govern-
ment. Voters can retain or reject their governors, a decision that can certainly have
salutary effects on governance, but these acts are summary judgments, generally
not reactions to particular acts or omissions. And when local elections do revolve
around a given issue, such as schools, they necessarily leave everything else out of
the picture. Citizens need more discriminating instruments to enforce accountabil-
ity. Fortunately, a number of these are available:
* Political parties can be a powerful tool for accountability when they are estab-
lished and vigorous at the local level, as in many Latin American countries.
They have a built-in incentive to uncover and publicize wrongdoing by the
party in power and to continuously present an alternative set of public policies
to the voters.
i Civil society and its precursor social capital enable citizens to articulate their
reaction to local government and to lobby officials to be responsive. These rep-
resentations generally come through NGOs (though spontaneous protests can
also be considered civil society), which, like political parties, often have parent
organizations at the provincial or national level.
- If citizens are to hold their government accountable, they must be able to find
out what it is doing. At the immediate neighborhood level, word of mouth is
perhaps sufficient for transmitting such information, but at any higher level
some form of media becomes essential. In some countries, print media can per-
form this function, but generally their coverage is minimal outside larger popu-
lation centers. A feasible substitute in many settings is low-wattage AM radio,
which is highly local, cheap to operate, and can offer news and talk shows ad-
dressing local issues.
* Public meetings can be an effective mechanism for encouraging citizens to ex-
press their views and obliging public officials to answer them. The cabildos abiertos
held in many Latin American countries are a good example. In some settings,
such meetings may be little more than briefing sessions, but in others they can
be effective in getting public officials to defend their actions. Participatory bud-
geting processes are also a powerful channel for communicating citizen prefer-
ences to local decisionmakers.
* Formal redress procedures have been included as an accountability mechanism
in some decentralization initiatives. Bolivia probably has the most elaborate
instrument along these lines with its municipal vigilance committees, which.
- 97 -



Decentralization Briefing Notes
are based on traditional local social structures, charged with monitoring elected
councils, and encouraged to file actionable complaints with higher levels if
needed. In other systems, formal recall procedures are available to citizens dis-
satisfied with their officials.
* Opinion surveys have generally been considered too complex and sophisticated
to use at the local level, but usable and affordable technologies are being devel-
oped in the Philippines, enabling local NGOs through polls to assess public
opinion about service provision.
A recent U.S. Agency for International Development assessment of democratic
local governance in six countries found that each country employed a different mix
of these mechanisms, while no country employed them all. No one instrument
proved effective in all six settings, but various combinations offered considerable
promise. Some of these instruments may be able to substitute at least in part for
others that are weak or absent. Civil society and the media, for example, might
together be able to make up for a feeble party system at the local level.
Transparency and Corruption
In theory transparency and corruption should be inversely related, such that more
transparency in local governance should mean less scope for corruption, since dis-
honest behavior would become more easily detectable, punished, and discouraged
in the future. The history of industrialized countries indicates that this tends to be
true in the longer term, but recent experience shows that this relationship is not
necessarily true at all in the short run. In the countries of the former Soviet Union,
for example, local governance institutions have become much more open to public
scrutiny in the 1990s, but at the same time there can be little doubt that corruption
at all levels has increased greatly. It is to be hoped that the local mechanisms of
accountability discussed here will in tandem with greater probity at the national
level improve the degree of honesty at all levels, but at best this will take time. The
message for the international development community is to press forward with as
many of these accountability mechanisms as is feasible.
Manor notes a second type of linkage between transparency and corruption.
He found that in India, greater transparency in local governance was not accompa-
nied by increased corruption, but it did lead to popular perceptions of greater pub-
lic malfeasance, simply because citizens became more aware of what was going
on.'5 This pattern has surely repeated itself in many other locales. Over time, to the
extent that accountability mechanisms begin to become effective and corruption
begins to decline, the citizenry should appreciate the improvement.
Conclusion: Slow but Sure Progress
The democratic local governance initiatives currently underway in many countries
hold much promise for developing effective systems of public accountability that
15. James Manor, "Democratic Decentralization in Africa and Asia," IDS Bulletin
26:2(1995):81-88.
- 98 -



Potential Impacts of Decentralization
will ensure that government servants are responsible to elected officials, and that
officials are in turn responsible to the public that elected them. In the process these
systems of accountability should increase the pressure for more transparent local
governance, in which corruption will be easier to bring to light and thus to curtail.
But just as it took many decades for such efforts to make much headway in the
industrial countries, so too quick results cannot be expected elsewhere.
99 -















The mission of W3BI is to help
World Bank clients and staff
acquire new development
knowledge and skills through a
variety of courses. sem:nars.
and other learning events. It
designs programs on topics
related to economic and social
development for governments,
nongovernmental organiza-
tions, and other stakeholders.
The institute oroduces and dis-
serninates publications and
electronic information products
that su pport these objectives.
For informafion onr WvUIBi
publications wirite to:
Publications
WB I
The World Bank
1818 H Street, NW.
Washington, D.C. 20433
Tel: (202) 473-6349
Fax: (202) 522-1492
Visit us orl the VVo,Ild VVide W.v'eb at.
httLo:Iwww.worldbank.org//wbi