22928 August 2001 io04 6 ?P t3S\6 06 6 0 69~~~~~~~~~ s 7H~~~~~~~IC ;5tUCxO ' V >1,0,f t Contracting for public services Output-based aid and its applications Copyright 2001 by the International Bank for Reconstruction and Deve opment/ THE WORLD BANK 1818 H Street, NW, Washington, DC 20433, USA All rights reserved Manufactured in the United States of America First printing August 2001 The findings, interpretations, and conclusions expressed in this report are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organiza- tions, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data nc uded in this publication and accepts no responsibility whatsoever for any consequence of their use. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address in the copyright notice above. The World Bank encourages dissemi- nation of its work and will norma ly give permission promptly and, when reproduction is for non- commercial purposes, without asking a fee. Permission to photocopy portions for classroom use s granted through the Copyright Center, Inc., Suite 910, 222 Rosewood Drve, Danvers, Massachusetts 01923, USA. ISBN 0-8213-5007-2 Cover and design: Grundy & Northedge, London. Copyediting and production: Alison Strong and Wendy Guyette, Communications Development Inc., Washington, DC. Contracting for public services Output-based aid and its applications Edited by Penelope J. Brook and Suzanne M. Smith VWORLD BANK s 4I . i! Contents Acknowledgments vii Foreword ix Part 1 Output-based aid: precedents, promises, 3 and challenges Part 2 Chapter 1 Expanding rural telephony 15 Output-based contracts for pay phones in Peru Chapter 2 Making water affordable 23 Output-based consumption subsidies in Chile Chapter 3 Easing tariff increases 31 Financing the transition to cost-covering water tariffs in Guinea Chapter 4 Maintaining roads 39 Experience with output-based contracts in Argentina Chapter 5 Extending rural electrification 47 A survey of innovative schemes Chapter 6 Educating mothers for health 57 Output-based incentives for teaching oral rehydration in Bangladesh Chapter 7 Promoting preventive health care 65 Paying for performance in Haiti Chapter 8 Improving primary health care 73 Output-based contracting in Romania v Chapter 9 Pursuing output-based education 8 The evolution of contracts for schools in the United Kingdom Part 3 Designing output-based aid schemes: 91 a checklist From strategy to implementation 91 Clarifying the role and sustainability of public funding Deciding who will be eligible to receive services that attract public funding 95 Deciding who will be eligible to provide services 96 Choosing the market environment 9 V Defining performance 101 Linking payment to performance 06 Shaping other aspects of the contract 0- Structuring the administration of the scheme 0 Leveraging experience 1 Suggested readings 119 vi z Acknowledgments 0 The preparation of this publication has been funded in part by the m Public-Pr vate Infrastructure Aov sory Facility, a multidonor technical A m assistance program (http.//www.ppiaf.org). In addition to the zI authors, the editors would like to thank the fo lowing people for research assistance, advice, and comments: Ian Alexander, Yao Badjo, Douglas Barnes, Lorenzo Berto ini, S mon Blair, James Brumby, William Bulmer, George C arke, Philippe Dongier, Michael Engelschalk, Charles Feinstein, Loraine Hawkins, Ada Karina Izaguirre, Michael Klein, Huiwen Leo, Ruth Levine, Nei Roger, Rishi Sawhney, Robert Taylor, Alan Townsend, Richard Verspyck, Bjorn Wellenius, and participants in the cinic series on output-based aid sponsored by the World Bank's Private Sector Advisory Services. The editors would also like to thank Stella Franco and Gracia Sorensen for assistance with project management. vii Foreword m Access to good, reliable public services is critical for the poor in 0 developing countries if they are to rise out of poverty. Safe water a and sanitation, modern energy and communications, good- quality basic education and health services-all contribute directly to individual well-being, and all improve economic opportunities for low-income households. This is not news. In recent decades developing country gov- ernments and bilateral and multilateral donors have focused mas- sive efforts on improving infrastructure and social services. But these efforts have all too often ended in frustration-with invest- ments in capital wvorks failing to produce sustained flows of reli- able services, and with subsidized service provision failing to translate into improved access for the poorest. The primary challenge now is to ensure that aid- and tax- funded spending reaches the poor, that the services this money finances respond to their needs and preferences, that these ser- vices are delivered efficiently, and that public funds are used in a way that leverages private financing of service delivery. For developing countries the stakes are high. Better deliv- ery of services is critical to the well-being and economic poten- tial of their citizens-and to economic growth more generally. Better use of the funds spent on public services is critical to fis- cal viability. Governments seeking to recover from indebtedness, and to avoid it in the future, need to ensure that projects increase productivity and are financially sustainable. Better delivery of public services is also critical to improving gover- nance. Governments that can run efficient, transparently financed programs for delivering public services reinforce their credibility with their own citizens and Wxith the international community-both private companies that might invest in ix a service expansion and donors wearied by skepticism about the effectiveness of aid. The cases gathered in this book tell of efforts to improve the delivery of services by contracting out their provision and link- ing the payment of subsidies to the delivery of services to tar- geted groups-illustrating aspects of an approach that we call "output-based aid." The cases highlight varied attempts to sharpen the focus on the objectives of aid and public spending, to improve incentives for efficiency and innovation, to expand opportunities for mobilizing private financing to meet social objectives, and to enhance accountability in the use of public resources. It is our hope that these cases, and the checklist on designing output-based aid schemes, will help advance the debate on how to improve service delivery to the poor in devel- oping countries and afford practical insights for all those com- mitted to this goal. Michael Klein Director, Private Sector Advisory Services The World Bank x I'. II ''''a 'a'"' a  a I aa 'a'a'aa'aa- 'a aa 'a a aa a a a/SI a'1;aj'1' a a a a 1, a1 i'i/'a' a'aa lllaaaaa'aa'aa'a I I a a a < /1 a a a aaa aa aa a1 //aa 'a-' a a/aa aaala' 1 a ala a aa11 'aa'aa aaI'll aaa'aaaal 7' aa a,1 I a aa a I laY aa'l111al a a aaa 'a i-a 4'a;"a-  a aa a'a  y/ al, a a, aa 4 a a / / aa a laal aaa a alaaa a / a a ,  ,,al1a '/l a/al a' -'a' a 'a' al11'/ 1a'a'l Aaaaa1a1'a'a1'aa a..aaallallllal a1aaaaaal all 1a'a  a' ala a1 4'al alali 'a'a a'al/Aa. ,a 'a aa a a a aaa I 41a1 'a ala-a'l'a- Ia a a ,aal aaa a I 'a alKlaa;"a a''a'aaa 'a" l'aal'a>lk al'aa'4 l.iaaa a a f al aSaa la a I aaaa a1 a1/' a 'a ja/ala aaa a a Il7 a', 1a 1/41/I 4' aal a a '7 1 a I a'a! 1 ,a/41a/l,llV1aI aA aa'/ aa/4Ii/la//a4a/al'a/a/l'all% a / a1l'aal$'a aa a 'II a' a 'ala a' 1, aal al 'a/a al/al41l./l-aaa .,a'al1aaal a ,a Ia / aa a,11/,11  -.\ 1 1 all a 'aaa a a/a a, a la' l -- al"a a a a -aa'a;lal/IL' ll a aaa a aaa'aall'a'alaal I lal a ,1a a a a a I II a ''a Il,If a' ' aa Ia a / a Ia a aa,aa aa 'a'aa aIlI'a1,all/'a4a'a X" a a1l a' a a/111 /laaa a'a' /'a// 'ala,) a aaa a' a, a1 ai I a 1' 11111 a 'i'aa'a-fa1a a a alll/Ialaa/al/ a'aa ala   'aa'a  I aa a 111111/ /a'aK,a/laaaa lllll'atll7llaIiif,1171/7aa7//ll i'll a a a a' aalaall'a/4a/1'a (¸) (Z) 'a '1111 aa1a aaaaa/ 11111 a 'al 'a ll1, 11a/a1/aaaala1ala 'a'aaa'aa 'a " a aa'aIlaaa/a1a/'a'aa'a4li I a aa aVaIlll1 l,a ,l Il",/a''4' ala /1 aI!aal% alla, ;$i4¾I- lLa/1lI1aiJ'a "a'a ailaaa a, aaaaa () 'a a' '11a  I  l'al '¼ lal/IlIt. ,1lIia I a, a, lll,a I ll'lal aI',a 'l4' < 'a Ia a a a a 1,1' U- aaI/I a 'll'4'l' " la'a a/ a I' a ala/ala a 'H A''  Ill//I Il' /J/l,h'li - U/ 'a' a a/a a 7 7 'a all -I/al ll a a  aa'/la'a Ia d 1/all ' aaa /l 1111  I1'1K7/a"laj'la'ai. ll,a'aaa al, a'a I' ''a aaa,aa lla/lla'aIlIiaialIl/lali'alllalla)a,l4ja/lallali'alialIllll'ail/laHla a ' I a', a'4  'a a''a 'aa a 111 'a I a ala/a lal'a'/,  a'a'a/4Ia a 11111'a1 alil/IaaiI'a'a'a$,4*aiIa/, 1/11 f/al/I >lAa'll/aI/aa , "a; aaa,l  '' a H a / -fa-l',la lal'aaal-/ .'a. aaaaa a,'l 1 a 'a I l'a a ' /a/, a 'ia'a' 'aI'aa  IaAillf/ja'/ / I'll a') 'a' ,lalal'aa'al'l'a lal'alaa 141/Ia aa/'a l""a'aa' alllll'lllalal,1'aaall la,al'a/ll /41 'aaalaaadllala'aa"aaf lllall'aala/la1/alaaa 41'iaa lf 11/ ,,/ 'I'll'la' l' 1, lall a 11, ; I,l'a4 p'a-I,,1 r 1a '-ilail /'a/lkl/  I 'a4 / / I 'a -'aa-.illa/-i'a   157;   llaS'a/4 1 /a4:ai4$l71a41laali'a'aa 'alA//7a'K Across a wide range of countries and sectors, reforms that transfer significant o commercial risk to the private sector have typically improved efficiency in > delivering services and expanded their availability.2 Analyses comparing the actual effects of reforms with reasonable coun- terfactuals show that in most cases private sector reforms have yielded sig- nificant welfare gains-even in countries starting out with weak institutions. But they also suggest that the benefits of private participation and its effects on distribution are sensitive to the design of contracts and the related mar- ket structure and regulatory reforms.3 Time and again, sound contract design, attention to facilitating competition, transparent project procure- ment, regulatory independence and accountability, and sound strategies for allocating and mitigating risk have proved to be essential for sustainable, ben- eficial projects. Drawing on experience to improve the delivery of aid Output-based aid draws on the experience and tools of public sector per- formance contracting and private infrastructure schemes. Service delivery is delegated to third-party providers under contracts designed to provide incentives for efficient, well-targeted service delivery, in part by tying a sig- nificant part of the compensation to delivery of specified outputs or results. 4 Public funds from external donors or domestic tax revenues may comple- ment user fees-or serve as proxies for user fees for services that are largely public in nature. Output-based aid extends some of the core features of traditional gov- ernment contracting and private infrastructure reforms. It goes beyond con- tracting by mobilizing commercial financing of service provision. And it differs from many private infrastructure schemes by complementing user fees 0 with carefully targeted subsidy payments. Both these characteristics increase -o the potential for niobilizing private funding for critical public services while ensuring a high level of accountability for the use of public finds. > m Many developing countries-and many organizations in these countries- have adopted aid schemes with output-based elements: C * In Chile subsidies for water services to low-income households flow to m 0 providers only when a qualifying household has received the service andd paid its share of the bill (see chapter 2). Guinea has also used an output- based subsidy scheme for water: an International Development Association credit was used to ease the transition to cost-covering tariffs 0 co) (see chapter 3). c * In Haiti and Romania primary health care providers in rural areas > receive compensation based on their delivery of defined basic services, n with an emphasis on preventive care (see chapters 7 and 8). > * In Peru telecommunications companies compete to expand and sustain services in rural areas on the basis of the smallest subsidy required (see v, chapter 1). These examples stand in stark contrast to aid projects focusing on financ- ing facilities (building a health clinic, a telecommunications network, or a xvater treatment and distribution system) or other inputs (such as books and medicines; figure 1).They also stand in contrast to traditional pubhc delivery of services in many developing countries, which too often has produced dis- appointing results in terms of both efficiency and access. Designing output-based aid schemes: the challenges In designing output-based aid schemes, the overarching goal is to ensure that they channel public funds in a way that provides incentives for improving the delivery of services to the intended recipients. That requires designing schemes that help mobilize commercial funding of service providers, that expand nongovernmental supply capacity, and that ensure accountability for results. It is also important to design schemes in ways that build alnd reinforce 5 X. S S@ S S SSS* Traditional approach Output-based approach (such as materials) 0 Private financing mobilized by m Public funding C mn Service _linked to service delivery C EZ D good governance, particularly in awarding and monitoring contracts, and that 7 ensure cost-effective administration. ,,, There are many ways to design such schemes, and the approach that is most appropriate will vary across sectors and countries. (For a detailed dis- cussion of the issues to be addressed and the factors driving kev design deci- sions see the checklist in part 3.) Defining interded results Clear definition of the intended results is one of the hallm-arks of output- based schemes. This involves deciding who the recipients should be-all con- sumers, or onlv those meeting certain eligibility requirenients.?-and what the service should be. In an ideal world schedles would focus on broadlv defined outcomes, such as wvell-educated children or households with adequate access to potable water. In practice, however, developing effective contracts requires much more precision, and contracts need to account for the fact that some factors affecting outconmes wvill be beyond the contractor's control. Contracts usually need to focus on specific outputs-such as an education 6 service or household connections to a water system-that meet specified requirements. Correctly identifying indicators for these outputs is critical: misspecified or incomplete indicators can lead to counterproductive or biased behavior by service providers. In some cases defining and verifying an acceptable level of quality for even specifically identified outputs may be so difficult that the contract will also have to address inputs, such as the choice of technologv, the quality of the materials, or the qualifications or reputation of the service provider. Even so, schemes that can focus primar- ily on outputs or outcomes offer the greatest promise, as they create oppor- r tunities and incentives for contractors to discover new and better ways of D achieving the intended results. Defining the service to be delivered often involves tradeoffs between price and quality. Some schemes can be designed to allow intended recipients m to choose a bundle of services based on their preferences and abilty to pay. m Others wvill need to nail down such choices in the design of contracts. In these C Cf) schemes especially there will be benefits from getting the input of the intended ,, recipients on the design of services and possibly from using different ° approaches in different areas, to match local preferences and circumstances. C The definition of the subsidized service also needs to take into account > budgetary constraints and the sustainability of the subsidies.These considera- a tions may lead to a narrower definition of eligible recipients or to a decision > m to direct public funds to the one-time costs of service connection-typically the main impediment to expanded access to services-rather than the ongo- m ing costs of consumption. Choosing the environment for service delivery A basic design choice is whether to provide the service in a competitive mar- ket or through monopolistic supply arrangements. For many services, schemes can use vouchers or similar instruments to give consumers choice over their service provider, creating incentives for efficiency and responsiveness to clients. For other services, concerns about potential market failures or subdued supply response might lead to a decision to grant a tmoinopoly in a particular area.This choice requires decisions about boxy long the monopoly should last, hoxy far it should extend, and hoxy best to ensure that the service provider faces incentives for efficiency. Competitive bidding for time-bound conces- sions can provide a useful discipline over such suppliers and may also help in gauging the level of public funding required. 7 Deciding which service providers are eligible Experience with public sector contracting schemes and with infrastructure reforms reinforces the importance of ensuring that service providers are in a position to respond to incentives and that they operate at arm's length from regulators and the funding source. Providers under output-based schemes need to pass the same tests. Most public agencies fail both. That does not mean that output-based schemes should be limited to the international private sector. Depending on the service, small-scale local entre- preneurs could be important suppliers-and so could community groups or > nongovernmental organizations.When considering whether to include these m groups, however, an additional concern might be whether it is possible to establish a "level playing field" between suppliers, so as to reap the benefits of :n competition. m rn Choosing the form, level, and structure of payment Yp The form, level, and structure of the subsidy payment to service providers are 0 crucial in determining the incentives they face and the possibility of mobiliz- 0) rn ing private financing. Subsidies usually take the form of cash payments. The level of the payment depends largely on the expected cost of delivering the ser- o vice, net of expected revenues from other sources, such as user fees. For schemes >. that involve concessions, one strategy for determining the appropriate subsidy z is to award the concession to the firm willing to provide the service for the smallest amount. In other cases the subsidy might be based on estimates of costs and of recipients' willingness and ability to pay for the service. Determining the structure of the payment involves two main issues. First, how tightly should payment be linked to performance? Should most of the payment take the form of a fixed fee (which might still be voided if per- formance falls below some minimum level), or should most be linked directly to performance against specific indicators? Linking pay to performance indi- cators provides stronger incentives, but increases the risk that the supplier will engage in undesirable behavior (such as cutting costs by reducing quality) where quality is hard to specifyz and monitor. Second, should the service provider receive sorme up-front payment (which might be subject to deduc- tions in the case of nonperformance), or should the provider be paid only after satisfactory delivery of services has been verified? Here again, payment linked to performance indicators (in this case, satisfactory delivery of services) provides sharper incentives, but it may need to be adapted if service providers face difficulties in mobilizing financing for service delivery. Hybrid schemes 8 combining fixed or up-front payment with payment linked to performance indicators are usually feasible only when the service provider has been awarded a monopoly. Designing effective administration Designing etfective administrative arrangements for output-based schemes requires resolving a host of issues. A threshold question is their scope. Should schemes be established for each narrowly defined sector (rural electrification) c or encompass a broad range of services (rural infrastructure)? Should they cover small subnational jurisdictions or have national or even international D coverage? And should a separate scheme be established for each funding source, or could schemes pool funding from multiple sources (for example, a P portfolio of donor funds)? Broader schemes promise lower administrative I 0 costs and may help pool expertise, but they may also be cumbersome and m m time-consuming to implement. Once the scope of a scheme is defined, who should be entrusted with r management-managing funds, designing and awarding contracts, monitor- 0 Zl) ing and verifying service delivery, and paying the service providers-a private firm, a nongovernmental organization, or a public agency? An important con- > sideration in this choice is which will be most effective in establishing credi- n ble arrangements that win the confidence of service providers and funding > sources while also helping to develop good governance practices. m Finally, should one entity be responsible for all aspects of the schemes m administration, or should some functions be delegated or contracted to oth- ers? In some cases there may be advantages in having regulatory bodies, non- governmental organizations, or communities monitor and verify the delivery of services. And local microfinance institutions or other intermediaries could play a role in the payment scheme. Conclusion Experience with contracting for service delivery, through public sector per- formance contracting and private infrastructure schemes, suggests cause for both optimism and caution for output-based aid initiatives. The case for optimism: public sector contracting and private infrastructure schemes appear to have improved outcomes and raised welfare, even in countries starting out with weak institutions. The case for caution: there is increasing evidence that the size of the benefits, and the extent to which they reach the poor, depend crucially on the incentives created for service providers- 9 through the details of contracts, through competitive pressure, and through regulatory oversight. This book is aimed at stimulating thought and debate about potential applications of output-based aid schemes to improve the delivery of basic ser- vices to the poor in developing countries. The cases it presents are not intended to be perfect examples of such schemes. Instead, they illustrate some of the key challenges in channeling tax and donor funds toward specific c results and creating incentive structures that ensure efficient achievement of those results-and thus reduce the subsidies needed. Together with the pre- w c> liniinary checklist of design issues that follows these cases, they will, it is m hoped, cast light on both good practices and potential pitfalls in contract P design and implementation for output-based aid schemes. m C) m m CD m -1 'U > m m0 1) 1 0 0 Notes access to potable water, or a Research Working Paper Pene ope J. Brook reduction in disease as a 2407. World Bank, World > mC, (pbrook@worldbank.org) is result of effectve preventive Bank Institute, Governance, U program manager of the heath care. Regu ation, and F nance ( -o Rapid Response Unit, 2. The ong time hon- Division; and Latin America a) a Private Sector Advisory zons of large-scale infra- and the Caribbean Region, r CD Services, World Bank, and structure projects, however, Finance, Private Sector, and Z Murray Petrie (mpetrie@ mean that full evaluations Infrastructure Sector Unit. -u 01 ihug.co.nz) is a principal are possib e only after a Washington. D.C ° 0) with the Economics aiid long period. OECD (Organisation for m U) Strategy Group, Ir New 3. For a good survey of Econormic Co-operation and >D Zealand. The authors wish the theoretica and empirical Development). 1999 to thank Warrick Smith for literature on the impact of Performance Contracting. > his comments and advice. pr vatizat on, see Shir ey and Lessons from Performance z 1. Inputs are the Walsh (2000). For an Contracting Case Studies no resources-such as overview of current data on and a Framework for Public employee t me. buildings. the distributiona effects of Sector Performance and equipment-used to these reforms that focuses Contracting. PUMA/PAC produce goods and ser- on Latin America. see (99)2. OECD Programme on v ces. Outputs are the goods Estache, Gomez-Lobo, and Publ c Management and and services themselves, Leipziger (2000). Governance (PUMA). Paris. such as drinking water sup- Sh rley, Mary M., and plied or health services References Patr ck Walsh. 2000. 'Public de vered. And outcomes are Estache, Antonio Andres versus Private Ownersh p: the consequences for con- G6mez-Lobo, and Danny The Current State of the sumers or the community Leipziger. 2000 'Utility Debate." Pol cy Research result ng from the outputs Pr vatization and the Needs Working Paper 2420. World suppl ed-such as an of the Poor in Latin America. Bank, Deve opment increase in productive time Have We Learned Enough Research Group. as a result of improved to Get It Right?" Policy Washington, D C. 11 N~~~~~~~~~~~~~~~~ )v ~ w~ Expanding rural telephony Output-based contracts for pay phones in Peru Geoffrey Cannock Geoffrey Cannock Innovations in technology, combined with (gtc@apoyo1.com,pe) was pro-competitive reforms, are rapidly general manager at Osiptel expanding access to telecommunications from 1997 to 2000. He,s in many developing countries. But now project director fora extending basic telephone access to the private economic rural poor can remain a stubborn problem. consultancyservice in Peru, To tackle this problem, Peru is using a ApoyoConsultoria. This "least subsidy" bidding approach. Private case study was funded by telecommunications operators bid for the the Public-Private minimum government subsidy they require Infrastructure Advisory to provide pay phone service in targeted Facility amultidonor rural areas. Part of the subsidy is paid on technicalassistance award, part once the equipment is program. installed, and the rest in semiannual installments for several years, contingent on compliance with performance standards. Winning bidders get a nonexclusive concession defining their rights and obligations. Early pilot results show that the private investment mobilized is twice the subsidy provided. Peru began reforming its telecommunications sector in 1992, privatizing the state telecommunications companies, establishing a regulatory authority (Organismo Supervisor de la Inversion Privada en Telecomunicaciones, or Osiptel), and gradually opening the market to competition. These reforms were expected to accelerate growth in service, but to leave high-cost rural areas-home to about 30 percent of Peru's population and 70 percent of its m extreme poor-largely excluded from a mostly commercial operation. So in 1992 the government also created a fund (Fondo de Inversion en C Telecomunicaciones, or Fitel) with a mandate to improve rural access to z :) telecomnmunications services by promoting private participation. Fitel's goal was to provide, by 2003, pay phone service in 5,000 rural towns and public access to the Internet in all 554 district capitals. r b Fitel's funding is assured by an earmarked 1 percent levy on the gross o operating revenues of telecommunications companies. Fitel is legally distinct 1 -< from Osiptel, but Osiptel provides technical and administrative services to 0 Fitel and approves policies and projects. Osiptel defined the target population as unserved poor rural localities with 500-3,000 inhabitants. Osiptel also con- co :CD> ducted policy, market, and engineering studies; set up a geographic informa- o) tion system; and defined the project cycle and procedures, including those for C) z identifying target localities, tendering projects, and monitoring performance >, against targets. " Deciding on competitive bidding > Osiptel selected potential localities on the basis of expressed local demand and I project analysis. The final choices are made during field visits, when local m authorities, who have far better knowledge of local trade and transportation z patterns, decide which towns should be served. Combining a demand-driven approach with top-down studies, rather than using a pure demand-driven approach, allowed network economies. (The Fitel rule now allows a pure demand-driven approach, since network facilities have been extended to most localities.) Government officials debated whether to hold just one tender for all towns (to promote economies of scale) or to encourage the entry of several operators to foster competition. The final decision was to partition the country into six regions, each with more than 700 towns, and then hold two tenders. The winning bidder is granted a nonexclusive 20-year renewable con- cession.The concession requires the operator to install at least one public pay phone in each rural locality listed in the tender, providing access to local and 16 long-distance voice and narrow-band data communications, and one point of public access to the Internet in each district capital. The operator is obliged to provide service over the entire 20-year concession, though the subsidy pay- ments extend only over the first five years. The operator niay use its facilities to provide additional services to individual subscribers, such as Internet and long-distance telephony. Osiptel expects that the service in rural towns will be fully commercial after five years. m Setting pricing, subsidies, and incentives C Retail prices for rural services are regulated by Osiptel under a price cap ° regime similar to that in urban areas-though the cost to the operator for C rural calls is higher (often because of geographic isolation or extremes in m altitude and climate). Interconnection charges, also regulated, should result m in a net payment to the rural operator. But since most calls originate in 0 C urban areas, the provisional sender-keeps-all agreement between the opera- < 0 tors has prevented the entrant rural operator from benefiting from this net payment. C A financial contract between Osiptel and the operator establishes the > m terms and conditions under which Fitel will provide funds, tying the dis- 0 bursement of the subsidy to project implementation and service quality: 35 ° -9 percent is paid at the start of the project, 25 percent once the facilities are > installed, and the remaining 40 percent in semiannual installments over five c, years, subject to compliance with service performance targets.The semiannual ° installments are reduced by US$ 1,000 a day for pay phone and network mon- > itoring system outages, and by 10 percent per locality per week of delay in initiating service for up to one month, at which time the balance of the sub- z sidy is canceled. 7 The financial contract also specifies indicators of performance that are not linked to penalties (though Osiptel can impose penalties for noncompli- ance): grade of service (network congestion in peak hours), time to get dial tone, and overall quality of service as measured by mean opinion scores.These performance indicators, from International Telecommunication Union rec- ommendations, are readily available and understood by operators. Osiptel staff supervise project implementation.They use a network man- agement system to oversee system operations (traffic levels, continuity of ser- vice) in real time and a required dedicated data circuit in the operator's headquarters to monitor billing, failure reports, and the calls placed and received by the rural pay phones. In a semiannual report Osiptel assesses com- 17 pliance with performance targets and indicators and makes recommendations on Fitel payments. Getting started Osiptel started to collect funds after the privatization in 1994 and had collected enough funds and done enough studies to call for a tender by 1996. But the m tenders got off to a slow start.Technically, Fitel had everything to get under wvay: a clear mandate in the 1992 telecommunications law, strong support from ben- z o eficiaries and local authorities, money, technical support, and private operators. z o But institutional problems and lack of widespread political support at the rn national level delayed implementation.The 1992 law did not specify policies or procedures, so they had to be designed later by Osiptel, itself a start-up opera- tion.The Fitel model had to compete with different visions of the government's o role in delivering assets to the poor, and private interests lobbied against the -< Fitel mechanism as too transparent. The minister of transportation and com- 0 munications was reluctant to take political responsibility for approving the pro- -t jects. The approval process was further hampered by institutional conflicts with Osiptel, high turnover of ministers, and a centralized decisionmiaking process. o Still, by March 2001 three competitive tenders had been conducted for 0 z six projects covering all 5,000 rural towns due to be connected by 2003. Six > bidders competed for a pilot project, and four or more in each of the next CD two tenders. Winning bidders bid for all regions in the tender. New opera- °D tors, both foreign and domestic, entered the market. -u Reviewing results from the pilot project z0 m For the pilot project, covering 193 localities, the competitive bidding resulted z in a much smaller subsidy than expected.The winning bid requested a sub- , sidy 41 percent lower than Osiptel's estimate and 74 percent lower than a pre- vious offer by the incumbent operator. Results from the first year of operations (ending December 2000) are encouraging. Pay phones have typi- cally been located on the premises of a small business or local authority. Retailers provide space and security for the pay phones in return for a per- centage of the price of the prepaid cards. In addition, they may charge users for an informal messenger service to alert them to incoming calls.They also benefit because the phones help to cross-sell other products. The pilot project has reduced the average distance to the nearest pay phone to less than a tenth of xvhat it had been, and nearly doubled the share of the population living in localities with pay phones (table 1). In response 18 to user needs, the operator introduced service innovations, such as prepaid calling cards, and is providing dedicated Internet access and long-distance services. The operator met the deadline for initiating service in all 193 localities and also installed additional pay phones and individual telephone lines.Traffic exceeded Osiptel's forecasts by 7 percent in the first six months and 32 per- cent in the next six. The operator met targets for network management and average service reliability, but failed to meet service reliability targets in five z localities. That resulted in a fine of US$27,000, equivalent to 1.6 months' rev- r enue or a sixth of the semiannual subsidy payment. 0 During the first six months the operator also failed to meet the target for c grade of service, failed to supply enough prepaid cards, and had operational prob- lems. Osiptel delayed the first semiannual payment until these problems were corrected. It also postponed the second payment, because the operator failed to 0 0 act on a minor observation in the first supervision report. If uncorrected, minor < 0 observations become major observations in the next review and may delay pay- e inents. The delaved payments were equivalent to 1.8 months' revenue. Several : co performance indicators xvere not reported because of technical difficulties. > Surveys of users in June and December 2000 showed that a growing number wvere satisfied with overall service (up from 57 percent to 75 percent) ° and had access to prepaid cards (up from 35 percent to 50 percent). The sur- > veys also showed modest progress on service outages, hours of service, and H customer knowledge of how to use the facilities. ° T TAELE Access to telephones in the pilot project by department, 0 December 2000 rn CD -n Indicator Amazonas Cajamarca Piura Tumbes Total nn Rural towns served 57 54 54 28 193 0 Beneficiariesa 39,086 45,359 46,370 13,707 144,522 Distance to the nearest phone (kilometers) Without the project 251.4 26.1 26.1 9.0 n a. With the project 6.2 49 4.2 3.0 n.a. Penetration (percent) b Without the project 10.0 20.0 16 0 91.0 48.3 With the project 90.0 85.0 71.0 99.0 88.5 n a Not app icab e a nclides both direct benef c aries t nhab tants of the towns serveo) and ndirect beref c aries (those I v rg w th , 5 k lometers of the toens served) o Share af the population n the project area w th te ephone access Source Fite, 1998 19 Assessing Fitel as a policy instrument Initial results confirm that Fitel is an effective means for extending telecom- munications services to rural populations. Fitel attracts and leverages private participation and investment. And it enhances sustainability by spreading the subsidy over five years, which helps maintain a positive cash flow until rev- enues build up from growing traffic.The pilot project required a subsidy of m only US$11 per inhabitant while mobilizing private investment estimated at US$22 per inhabitant. Subsidy administration costs are low: according to r Osiptel's operating plan for 2000, after start-up costs (US$1.7 million) Fitel's o administrative costs have averaged less than 2 percent of the funds collected. Improvements are needed, however, in the links between performance and subsidies. First, performance targets and indicators should evolve over time. Since those used are typical for mature networks rather than start-ups, o a one-year grace period without penalties might be appropriate. Targets and indicators should also become more demanding over time, putting pressure 0 on the operators to continually improve service. Second, performance targets should reflect use, not just access. Where call charges are below incremental variable costs-as they may well be, since the regulated rural and urban tariffs m are similar despite the higher cost of rural service-the operator has no incen- .0 z tive to encourage traffic growth.Yet much of the benefit for the rural popu- lation comes from using the facilities, not just having access to them. m Moreover, the business case for investors and equipment suppliers may be ° determined largely by initial capital outlays and subsidies rather than recur- rent costs and revenues. A performance target that ties recurrent subsidies to I traffic may better align the interests of operators, pay phone retailers, and the economy. But it would be much more complicated to administer and would require Osiptel to monitor financial results, which it does not now do. 2 The tariff caps are another problem: there are high policing costs for Osiptel, cost shifting from the operator to the pay phone administrator, and reduced incentives for operators to generate traffic. Given these three problems, a more effective approach might involve fewer performance targets and regulatory controls, less use of penalties, and more emphasis on customer service. Project supervision, for example, which now not only verifies compliance with concessions, contracts, and performance targets and indicators but also seeks to influence project management, xvould be focused on a few key parameters and stripped of discretionary powers. This alternative approach would require more collegial relationships between oper- ators, civil society, and Osiptel than envisaged in Fitel's original design. 20 There are three ongoing challenges: First, the risk that private operators will underbid for subsidies and later default on their commitments, even though Osiptel has required operators to post performance bonds.' Second, the need to shift Fitel's support to smaller and less accessible localities as tech- nological innovations and network growth reduce the cost of reaching rural areas and as the private sector becomes increasingly willing to provide services k on commercial terms. And third, the exemption of cable television and Internet service providers from the levy, which raises concerns about the fair- ness and economic efficiency of Fitel's funding. z0 01 co 01 0 -r Z: 01 0 < -K C C( m NQote Referencec 1 The operators are Fitel (Fondo de Inversibn required to provide three en Telecomunicaciones). financial guarantees a guar- 1998. 1998 Annual Report. antee ernsuring the serious- Lima. ness of their offer (to prevent the "winner's curse"), an insta lation guarantee, and a guarantee against default on their contractual obligations. 21 Making water affordable Output-based consumption subsidies in Chile Andres G6mez-Lobo Andres G6mez-Lobo Chile began reforming the provision of (agomezlo@econ.facea. water and sanitation services in the late uchile.cl) is an assistant 1 980s. It first commercialized-then, in the professor n the Department late 1990s, privatized-most urban service of Economics at the provision. Before reform, tariffs were well University of Chile below cost. After reform, despite Previouslyhewasa substantial efficiency gains, concerns consultant at Oxford remained about the affordability of water Economic Research and sanitation services. To guarantee Associates in the United adequate and affordable services for low- Kingdom. He has income households, Chile introduced undertaken a number of individual means-tested water consumption research projects on subsidies in the early 1990s. Although the infrastructure subsidy public authorities determine how the design and performance, subsidy is applied, the now mostly private includtng in Chile, Colombia, companies deliver the service under a Panama, and the United scheme with built-in incentives to ensure Kingdom. This case study cost-effective service delivery by the was funded by the Public- companies and low wastage by the Private Infrastructure customers. Advisory Facility a multidonor technical assistance program In the late 198Os Chile began an overhaul of the legal, economic, and institu- tional structure of its water and sanitation sector. An important part of the reform was a new tariff setting methodology aimed at raising water prices to the true economic cost of the service. Before the reform water tariffs covered less than 50 percent of this cost-and only 20 percent in regions where production costs were high. Steady tariff increases in the 1990s doubled real charges for Empresa Metropolitana de Obras Sanitarias (EMOS), the service provider in the Santiago D metropolitan area and the country's largest water company. Rates increased even o more for the smaller companies, especially those operating in high-cost regions. D> As a result of these large rate hikes, a new mechanism was required to m protect vulnerable households. Since connection is almost universal in Chile's 5 urban areas, the affordability of consumption-not of connection-was the C3 > main issue, and a consumption subsidy was the obvious solution. Chile chose m a means-tested subsidy targeted to individual customers rather than a tradi- ° tional geographic or universal subsidy. c The subsidy program, introduced in the early 1990s, relies on the water D.j companies to deliver the service. The government reimburses them for the m subsidies on the basis of the actual amount of water consumed by each ben- o, eficiary rather than a preestablished amount, a method used in some coun- '5 tries. With the most important water companies having been privatized since 0 1998, private companies now serve 73 percent of urban clients. So the sub- ° sidy scheme is essentially being implemented by private companies on behalf co of the government. c/) How the subsidy works c, By law, the subsidy can cover 25-85 percent of a household's water and sew- erage bill for up to 20 cubic meters a month (though the limit now used is 15 cubic meters a month), with the client paying the rest. All consumption above the limit is charged at the full tariff. Each year the Ministry of Planning (Mideplan) determines, for each region, how many subsidies are to be granted and how they are to be applied, following several general principles:The subsidy is based on the willingness to pay for water services among low-income households. Only households that xvould be unable to purchase what is considered to be a subsistence level of consumption should benefit. And the subsidy should cover only the shorffall between actual charges and willingness to pay.' As a crude proxy for willingness to pay, Mideplan uses the benchmark set by the Pan-American Health Organization-that no house- hold should pay more than 5 percent of its monthly income in xvater and sew- 24 erage charges. It is unclear whether vulnerable households in Chile would be willing to pay more or less than this 5 percent. The subsidy scheme is funded entirely from the central government's bud- get. Using household survey information for each region and each company's published tariffs, Mideplan can determine how many households need a sub- sidy and how large benefits need to be to meet the benchmark for each region. To obtain a subsidy, a household must apply to its municipality, which determines its eligibility mainly on the basis of a scoring system called CAS > (box 1).2 Another important criterion is that households must not have pay- n ment arrears with the service provider. The municipality must award subsidies in the order of the applicants' CAS scores. Subsidies are normally renewed yearly for up to three years before a household must reapply. But if a municipality has distributed all the subsidies assigned to it and a new applicant has a lower CAS score than the last beneficiary, the municipality must withdraw the benefit from this last ben- o Cx eficiarv and assign it to the more deserving applicant. Building in incentives m The subsidy scheme has several incentive-based features. One centers on the 0 z fact that the amount of subsidy a beneficiary receives depends on the level of < 1 Determining eligibility for subsidies c CD An eligibility scoring system called CAS is the main targeting instrument used in Chile for distributing means-tested subsidies, It produces a score for z each household wishing to be evaluated based on a personal interview at its dwelling. The questionnaire used includes 50 questions on general informa- tion, identification of household members, living conditions, crowding condi- tions, health conditions, comfort, occupation and income, ownership of durable goods, and other socioeconomic indicators. Once the interview is conducted and the CAS score calculated, the score is valid for two years, and the household can use it to apply tor many different subsidies. Besides the water subsidy, eligibility for pension payments, family subsidy, free health benefits, and other subsidies is determined on the basis of the CAS score. Many municipalities outsource the interviews to private survey companies, but still calculate the CAS score. That lowers the risk of collusion between interviewers and households, since interviewers do not know the exact rela- tionship between the households' answers and their CAS score. 25 consumption, and results from two aspects of the program's design. First, the subsidy is expressed as a percentage of the household's bill. It is therefore a price reduction per cubic meter consumed, and no benefits are given if there is no consumption or delivery of service. Second, the household must pay the full tariff for consumption above the limit of 15 cubic meters a month. 2 This consumption limit reconciles the need to provide income support to low-income households for basic water consumption with the need to pre- serve financial incentives for efficient resource use. In essence, the Chilean o water subsidy can be thought of as a rising block tariff, where only means- -H tested households have access to the lower priced initial consumption block.3 > The fact that the subsidy scheme requires households to pay a fraction 5 of the bill even when their consumption does not exceed 15 cubic meters >n helps to maintain good payment habits among clients. It also preserves service r providers' incentive to improve commercial efficiency, since their income ° depends in part on the payment of this remaining charge. And the additional eligibility requirement of not having payment arrears has led to an improve- OD iiiment in clients' payment record. :S Another incentive-based feature of the scheme rests on the relationship 0 between the government and the service provider.This relationship is medi- 5 ated by the subsidy law and its accompanying regulations.4 The law requires 0 the companies to bill beneficiary clients net of the subsidy amount and then 0 z bill the municipality for the subsidies granted.The municipality is thus a client C of the service provider, meaning that it can be charged interest for late pay- En ment and that the service can be discontinued as a result of nonpayment (so En that in the next payment period the service provider can charge beneficiary households the full amount of the bill). This setup gives municipalities a strong incentive to transfer payments quickly to operators. The central government funds for the program are ear- marked, so municipalities do not stand to benefit financially from withhold- ing payment to the water companies. And the political wrath that could arise if they failed to pay the service providers-and thus lost the benefit for households-is potentially costly. The financial flows and control of the program are concentrated in the Undersecretariat for Regional Development of the Ministry of the Interior (fig- ure 1). The process requires that the company and municipality have synchro- nized lists of beneficiary households and that the interior ninistry verify that the regional invoice is consistent with the number and value of subsidies for the region approved in the annual budget. The arrangement is clearly bureaucratic, 26 F RE Financial control and procedures for the subsidy scheme Presents an invoice to Water municipality pany Validates invoice and sends it uncptyPswaecoay to governor M Pays water company t ~~~~~~~~~z Aggregates invoices of all 0 municipalities and presents Regional Transfers funds to regional invoice to Ministry governor municipalities m of the Interior D m Ministry of : Validates regional invoices the Interior and presents consolidated Undersecretariat m invoice to Budget Office for Regional m Development Ministry Approves expenditure c of Finance and deposits funds in 7 Budget Office each regional treasury's > account a C: a and municipalities are often unable to pay the companies' invoices on time. Some - companies charge the municipalities interest for the payment delay.The niunic- 0 ipalities must bear the interest and debt costs resulting from late payment, since a there is no provision in the national water subsidy budget for these charges. Despite this problem, the fact that companies receive a reimbursement ar for services and subsidies already delivered has several benefits. The arrange- a ment gives the companies full incentives for providing efficient and reliable r service. The subsidies accrue to households, not companies, and the amount of resources distributed is independent of the service provider's operational efficiency. Companies should be indifferent with respect to the subsidy scheme and receive no financial benefit from the program, except perhaps through the reduction of payment arrears by poor households. Who benefits? In 1998 nearly 45(,000 subsidies were distributed nationally, benefiting almlost 13 percent of households by an average US$Il 0 a month.The total cost was US$33.6 million. In some regions where incomes are low and wvater charges high, close to a third of households received the subsidy. On average, 27 52 percent of benefits in each region accrue to the three lowest income groups, and only 23 percent leak to the five highest income groups (figure 2).5 Subsidies represent a larger share of income for poorer households, nearly 8 percent for the lowest income group. 2 What are the lessons? The introduction of the subsidy-and especially the targeting results achieved- have been key to Chile's ability to raise water tariffs to levels reflecting costs C) xwithout compromising its social and distributional goals. And the costs to the government of doing so have been low.Tbe cost of the subsidy in 1998, US$33.6 m million, was well below the cost of the previous universal subsidy scheme. Before the reforms in 1988 the water and sewerage sector had a financial deficit of 2 percent of assets. But in 1998 this situation was reversed. Companies reported a surplus of close to 4 percent of assets and net profits of US$107 million, more o than three times the cost of the subsidy scheme (excluding administrative costs). Despite the successes of the subsidy program, several issues need to be considered if such a scheme is to be replicated in other countries. First, meter- u 0 * Share of total subsidy expenditure o * Average subsidy as a share of household income m Income decile Percent C ' 0 5 10 15 20 co 11; 2 C~~~~~ r 4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 5_ 7 Note: Income deciles are based on per capita hueodincome. Data rc ude cnly households with shared or own water connect ons Source: M deplan (Ministry of P anning), Encuesta de Caracterizacibr Soc oecon6rnica Nac onal (Sant ago, 1998). 28 ing is a prerequisite for this type of output-based consumption subsidy. For countries with low coverage of meters among low-income households, such a scheme may not be viable. In these countries, however, the most pressing social issue usually is not ensuring that xvater is affordable, but increasing the number of connections. For that purpose, a means-tested subsidy analogous to the one in Chile would be easier to apply, since metering is not required. Service providers could charge new clients the true cost of connection-perhaps pro- viding some credit by permitting payment in installments-minus a subsidy to > eligible households. The government could then reimburse the service r providers on the basis of the number of eligible households connected. > Second, the means-tested targeting used in Chile requires a certain amount of institutional capacity, especially at the municipal level. Even in s Chile, some municipalities still lack sufficient capacity to adequately adminis- a ¶2 ter and control the subsidy scheme. For countries with less institutional capac- ity such a complex system may not be viable. These countries could adopt o simpler targeting mechanisms, for example, a scheme based on a geographic poverty map, like that used in Colombia. Moreover, a scheme using a con- nection rather than a consumption subsidy will require less institutional m capacity, since households' eligibility must be evaluated only once. C z0 Third, an individual means-tested subsidy may be expensive to apply. c C Chile uses the same targeting instrument to distribute several welfare bene- a fits. lowering the administrative costs significantly. Applying such a schetne for 0 only one subsidy program may be too expensive. Again, however, the admin- c CD istrative costs for a connection subsidy are much lower than those for a con- c sumption subsidy. m Notes Social Comuna (Communal almost universal in urban m 1. Th s is not to say that Social Assistance areas. no further benefits should Committees), wh ch pro- 4. These laws and regula- accrue to low-income house- moted the idea of us ng a tions are ava lable in Spanish holds for purely d str butiona standard measure for a lo- at http://www.siss.cl reasons, only that the best cating social resources 5 By comparison, in way to meet such goals may among the poor. Colombia, the only other be through general welfare 3. A consumption-based Lat n Amer can country with programs, not a sectoral subs dy requires that each an important formal water consumption subsioy. household have a meter. subsidy scheme, 37 percent 2. CAS comes from This is not much of an issue of subsidies reach the five Comites de As stencia in Chile, where metering is h ghest income groups 29 Easing tariff increases Financing the transition to cost-covering water tariffs in Guinea Penelope J. Brook and Alain Locussol Penelope J Brook Guinea entered into a lease contract for (pbrook@worldbank.org) is water services in its major towns and cities program manager of the in 1989. The government was committed to Rapid Response Unit, cost recovery for the services, but wanted Private Sector Advisory to avoid a major tariff shock at the Services at the World Bank. beginning of the contract. So, for the first She has worked extensively six years of the contract an International on the design of Development Association credit subsidized infrastructure sector reforms a declining share of the private operator's to benefit the poor Alain verified supply costs while the water tariff Locussol (alocussol@ was raised until it covered costs. This worldbank.org), lead water arrangement jump-started the move toward and sanitation specialist in cost recovery and more sustainable water the Africa Region at the services-giving credibility to reform in a World Bank, was in charge region and during a time in which there was of the World Bank's support little experience with private provision of tor the Guinean water water services-while also setting a time projectin 1985 89. limit on subsidy commitments. When a public sector water utility does not recover the costs of providing ser- vice, it is often unable to extend the system-leaving poorer, marginal areas unconnected to the xvater grid. Consumers who are connected often receive poor-quality water and intermittent service because there is not enough rev- enue for treatment and maintenance. Unconnected poor consumers generally pay much more for water than do consumers with access. m Guinea faced this situation in the late 1 980s. Its urban water supply sys- tem was one of West Africa's least developed. Less than 40 percent of urban dwellers had access to piped water through household connections or D standpipes. Where connections existed, service was often interrupted and vwater treatment inadequate. To improve this situation, in 1989 the govern- ment entered into a lease arrangement for private operation of xvater ser- ''m vices in the capital city, Conakry, and 10 other cities and towns. Under a lease contract, because the lessor effectively buys the rights to the income >Z stream from the utility's operations (minus the lease payment), it assumes 0 much of the commercial risk of operations.The lessor's profitability depends on howx much it can cut costs while still meeting the quality standards in m the lease contract. Thus the lessor has incentives to make operations more Dz efficient. -< When the reform was implemented, the water tariff charged to house- 0 z holds was far below cost recovery levels. The government was committed to 0 ° seeing tariffs rise to cover costs, to ensure the financial viability of providing vwater services and enable their expansion over time. But it xvanted to phase o in tariff increases gradually, both to ease the burden on consumers and to Di exhibit improvements in performance before the full costs of those improve- ments showed up on consumer bills. > The challenge was finding a way to subsidize services while tariffs were M raised to cost recovery levels without undermining the private provider's _; incentives to make service provision, billing, and collection more efficient. In this context an International Development Association (IDA) credit that also Ct financed extension of the water supply system introduced a transparent, lim- z ited duration subsidy of the consumer tariff. This subsidy was designed to > achieve thvo objectives. First, the subsidy sought to preserve the operator's incentives to improve performance-by increasing staff productivity, reducing unaccounted-for water, and increasing connections and collection ratios. Second, in the early stages of the contract the subsidy xvas meant to protect the private operator against foreign exchange risk-a common obstacle to private participation in the water sector. 32 Bidding for the lease contract was based on the lowest rate that the oper- ator would receive for each cubic meter of water billed and collected, sepa- rated into local currency and foreign exchange components, and on the lowest price it would charge customers for new connections. The winning bid, by a consortium of Compagnie G&nerale des Eaux and SAUR, was 30 percent below consultant estimates and 15 percent below the second-lowest bid. 3 Two organizations were central to the lease arrangement: a state-owned water authority, Societe Nationale des Eaux de Guinee (SONEG), and a water management company. Societe d'Exploitation des Eaux de Guinee C (SEEG). SONEG owned the water supply facilities in the cities and towns covered by the lease and was responsible for sector development, including IL servicing debt and planning, financing, and implementing new investments. SONEG was also responsible for setting tariffs, subject to ministerial approval. SEEG xvas owned by the state (49 percent) and the winning consortium " of Compagnie Gen&rale des Eaux and SAUR (51 percent). SEEG held the Z 1 0-year lease contract with SONEG, xvhich made SEEG responsible for oper- ating and maintaining urban water supply facilities, billing customers, and col- lecting charges. The private consortium provided management services to m SEEG through a separate technical assistance contract. > z Making the move to cost-covering tariffs z Before the reform Guinean households paid US$0.12 a cubic meter (1989 ° U.S. dollars) for water provided through the grid. Initial projections were that 2 the average charge to consumers would need to rise to US$0.76 a cubic 0 meter in 1995. and then fall back to US$0.68 (Triche 1990).This implied a , tariff increase of up to 630 percent. a As noted, SEEG was remunerated based on the rate set in the lease con- tract.That rate was intended to cover SEEG's operating expenses-in both for- eign exchange and local currencv-and depreciation on its assets, and to provide a return on equity. Over the course of the contract the goal was to raise the tariff paid by consumers to cover the lease contractor rate and to provide an C asset rental fee to SONEG, calculated to cover operating expenses, service debt, z and finance investment. In other xvords, the goal was to make water services > financially viable and so sustainable. At the same time, tariffs would need to be kept in check through continuing pressure on the operator to reduce costs. The cost of phasing in increased tariffs was funded by the IDA credit (US$16.9 million), calibrated to finance SEEG's foreign exchange costs by 100 percent for the first four years, and on a declining basis over the next two 33 years. The credit was combined with a government commitment to finance SONEG's debt service on a declining basis over these first six years.Through this mechanism SEEG obtained the lease contractor rate it had bid, but its financing was shared between consumers and the credit (figure 1). At the start of the contract SEEG received an advance installment from Wr; IDA calculated to be 30 percent of estimated first-year water sales. This advance was compensated by deductions from disbursements of the credit Do over the next five years. Following this initial payment, the credit was dis- obursed by the Guinean government based on independently audited state- iments of collections by SEEG, issued every four months. Access to the credit vwas thus directly linked to service delivery and collections performance. °u At the same time, SEEG made monthly asset rental payments to SONEG. In the first two years the rental fee was set to cover SONEG's oper- -n ating costs. Over the following four years it rose to cover an increasing por- > tion of SONEG's debt service obligations, with full coverage of these costs by oz the end of the sixth year. Two years into the contract, SEEG for a time dis- 31 counted its rental payments to SONEG because of long arrears from public m sector clients. II z - 0 O ~~~~Full cost of water (constant 1989 U.S. dollars per cubic meter) C) < 1r0 Lease contractor rate: 31 Other financial needs: A / external financing for foreign GO support from the national spending under lease contract > 0.8 budget for debt service 'U 0.6 vi o ) 0.4 m \00 itSi:0 Otherfin~~~ancial needs: X F -.; 0 2 i \ ~~~consumers' contribution to <:j0 :.E00:V$ 0.2 SONEG's rental fee 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Source: Wor d Bank data 34 Under the lease contract the lease contractor rate could be adjusted quarterly to reflect changes in foreign exchange costs and biannually to reflect changes in local costs. In addition, provision was made to renegotiate the lease contractor rate after four years-when a new, cheaper supply source came on- line. (This renegotiation did not occur, however.) Consumer prices wvere adjusted based on a cost-plus formula, intended to reflect changes in service costs for SEEG and SONEG. Based on this cost adjustment process, the tariff continued to increase after the subsidy was phased out, reaching US$0.83 in 1996, then holding constant in local currency for the rest of the lease contract. By late 1997 the minimum bimonthly payment for service was about US$13 per customer- very high given household incomes.The result was a steep fall in collections 2 and a rise in inactive connections. S Why did tariffs rise so high? One reason was costs that appear high by 1~~~~~~~~~~~~~~~~~ regional standards-costs driven by low labor productivity; a large continuing presence of expatriate staff, high debt servicing costs, and considerable bad debt. Moreover, regulatory pressures to control these costs xvere weak-as reflected, for example, in the government's failure to renegotiate a reduction of the lease contractor rate or revise the cost indexation formula after four years of opera- tion. Complicating matters were disputes between SEEG and SONEG over 0 definitions of xvater losses-and hence responsibility for actions to reduce them. 2 S Improving services and operations c The contract led to many service improvements. Investments in new capac- ity (external to the lease but financed by IDA and other donors), combined with rehabilitation and maintenance, increased the share of the population 0 with access to safe water from 38 percent in 1989 to 47 percent in 1996. By 1994 piped xvater in Conakry was in compliance with World Health 2 Organization norms (G61inas and others 1996. cited in Menard and Clarke D 2000). Household connections increased (if less than originally hoped), and metering increased from1 5 percent to 98 percent for private customers, and ) to 100 percent for government customers. Customer service also improved, with shorter delays for new connections and for repairs to the network. > But the contract did a poor job of reducing physical and commercial losses and achieving big improvements in access.These failures probably con- tributed to the high tariffs, because production inefficiencies could be passed through to the tariff, and slow progress on connections meant that capital costs xvere covered by a small customer base. New connections were proba- 35 bly inhibited by the absence of financing for small social service connections (an approach used successfully in Benin and Cote d'Ivoire) and by abundant alternative water sources, at least during the rainy season. Conakry's annual average of more than 4 meters of rainfall can yield sufficient roof catchment for low-income households even if roofs are small. J With the high price of water from the network, many residents could not or would not pay for it: in 1994 nearly 12,000 connections were inactive because of nonpayment. The record on billing and collection was patchy. In 1987 the G) public water utility collected on only 12.5 percent of its billings (Menard and Clarke 2000). Bill collection from private customers improved to about 75 per- I cent with the implementation of the lease contract. But as tariffs rose, collections 0 ' fell to around 50 percent in 1991-92, rising to about 60 percent in 1993-96. Collecting from the public sector proved even more problematic, particularly C!) after the early years of the contract, with collections falling to 50 percent in 1991 z> and lO percent in 1993. SEEG sought legislation to penalize illegal connections _' and facilitate recovery of arrears, but the legislation failed to receive political sup- port. Following this, litde attempt was made to eliminate illegal connections. m > Conclusion -i A recent analysis of the reform estimates that between 1989 and 1998 the z contract generated a net welfare gain in the region of US$33 million (1996 o U.S. dollars). Most of this gain went to domestic players, including consumers C-) 0 (M&nard, Clarke, and Zuluaga 2000). o While the subsidy scheme worked smoothly, operating and regulatory 77 performance were mixed. Guinea's experience illustrates the challenges of cre- 2 o) ating effective performance incentives for private operators when regulations >s and monitoring are weak and the operator is not fully subject to comrnmercial 11 I risk. The lease contract in Guinea, which expired in 1999, was not renewed, and the international partners left the country in early 2001. Given the bad publicity surrounding these events, it may be difficult for SONEG to attract a o) new private operator through the competitive bidding now being prepared. The declining, output-based subsidy used to smooth the introduction of m > Guinea's lease contract holds promise for other developing countries where tariffs are well below cost recovery levels and the long-term capacity to sub- sidize is limited. But the resulting arrangements are likely to be efficient and sustainable only if the tariff-smoothing process is combined with credible contractual and regulatory mechanisms for achieving cost savings, and these savings are passed on to customers. 36 m Note and Jean-Pierre Schmidt 477 World Bank, r This chapter draws exten- 1996 "Well Water in Two Washington, D C. 0 sively on Brook (1999) and Districts of Conakry z M6nard and Clarke (2000). (Repub ic of Guinea), and 0 m Details on the des gn of the Comparisons with the Piped U)m subsidy arrangement are City Water." Water drawn from Triche (1990) Resources 30(9): 2017-26. Evidence on results and on Menard, Claude, and z the welfare impact of the George Clarke. 2000. "A I lease arrangement is drawn Trans tory Regime Water m from M6nard, Clarke, and Supp y n Conakry, Guinea" > 0 Zuluaga (2000) The authors Po icy Research Working are gratefui to George Paper 2362. Wormd Bank, z Clarke and Richard Deve opment Research 0 Verspyck for comments and Group, Washington, D C. 0 to Lorenzo Bertolini for Menard. Claude. George 0 assistance In preparing fig- Clarke. and Ana Maria m ure 1. Zuluaga. 2000. "The We fare Effects of Private Sector > References Participation in Urban Water Brook. Penelope J. 1999. Supply in Guinea." Policy D "Lessons from the Guinea Research Working Paper m Water Lease." Pub ic Po icy 2361. World Bank, r C- for the Private Sector Note Development Research z 78. World Bank, F nance, Group, Washington, D.C. > Private Sector, and Triche, The ma. 1990 Infrastructure Network, 'Private Participation in the Washington, D.C Del very of Guinea's Water Gelinas, Yves, Harold Supply Services." Pal cy Randall, Luc.e Robidoux, Research Work ng Paper 37 Maintaining roads Experience with output-based contracts in Argentina Gerard Liautaud Gerard Liautaud (gliautaud@ Creating capacity and incentives for worldbank org) works in the ongoing maintenance of capital Latin America and the investments is a challenge across all Caribbean Region of the infrastructure sectors, but especially for World Bank. He has services that cannot attract cost-covering managed the Bank's funding user fees, such as low-volume roads. The for the Argentine roads Argentine government has responded to project and advised the this challenge in its nonconcessioned road Argentine government over network by using output-based contracts the duration of the Banks with the private sector for rehabilitation and support for the project. The maintenance. The multiyear, lump sum Bank loans total US$435 contracts, funded by the government and million and partially fund the the World Bank, specify required road project over the period service outputs and use incentive-based 1997 2002. a loan of payment schedules to ensure the quality of US$370 million (75 percent the work. After three years of operation the of the funding) for phase 1 60 contracts (averaging US$10 million) in andaloan of US$65 million the first phase are working well. By 2002 (50 percent) for phase 2. around 75 percent of Argentina's The Argentine government nonconcessioned roads should be funds the remaimng cost of operating under output-based contracts. US$215 million from general tax revenue This case study was funded by the Public- Private Infrastructure Advisory Facility a multidonor technical assistance program. Road maintenance and rehabilitation have traditionally been procured by the Argentine government through input-based contracts with the private sector. Several goals underlie its shift to output-based contracts: To cut the administrative costs associated with input-based contracts-in particular, the costs arising from the frequent requests for payments to cover necessary increases in inputs.To encourage innovation and cost-effectiveness by giving contractors more responsibility. To develop more stable funding for road maintenance (under traditional arrangements national funding dried up dur- > ing a fiscal crisis). And to better meet road users' needs. The contracting is run at the national level by the highways authority. o The first step was a nationwide road survey to estimate traffic, define the min- imum (rather than optimum) road standards, define the rehabilitation and x maintenance required, and identify the size and shape of the subnetworks for aJ contracting out. Roads with traffic in the range of 300-3,000 vehicles a day zm C, are eligible for output-based contracting. (Roads with traffic exceeding 3,000 vehicles a day are considered concessionable.) On the basis of the survey information the government set uniform national output indicators for the 0 contracts.To help define the indicators, road users (who spend US$10 billion annually operating vehicles) were surveved to find out what they consider an :,,> acceptable level of service. Contracts were awarded to the lowest lump sum O bidder, and a share of the payments to contractors is based on how well they 0 ° perform against these indicators. > In their initial application, the output-based contracts covered mainte- U) nance of paved roads, with payout schedules based on kilometers per month. > The next stage covered rehabilitation and maintenance, with contracts requir- GDl ing lumpy up-front payments to cover rehabilitation costs.A third stage under 2 consideration would cover new construction of low-volume roads (less than > 250 vehicles a day). Starting out with maintenance contracts Introduced in August 1995, the first output-based contracts are kilometer per month contracts spanning four years (and recently renewed) and covering a network of about 3,600 kilometers of paved roads. The 11 contracts cover roads that were in good to fair condition and expected to require only rou- tine maintenance to remain in that condition over the next few years. Contractors are paid equal monthly installments for specified services, as long as the quality of outputs complies with the technical specifications. If the outputs do not comply with standards, daily penalties are imposed (and sub- 40 tracted from future payments) until the necessary repairs are carried out. The penalties are based on deficiencies noted during monthly inspections. No penalties are imposed for the first two or three months following the award of a contract, giving the contractor time to repair any preexisting deficiency. The contracts are working well. Routine maintenance is costing an aver- age of about US$175 per kilometer a month. About 600 certificates of non- compliance have been issued, giving rise to penalties amounting to only 1 percent (US$300,000) of the total amount of the contracts. Given the satis- factory outcome, the contracts were renewed for four more years with the D same contractors. z 0 Adding rehabilitation On the basis of the experience with the maintenance contracts, a contract was x -u designed for combined rehabilitation and maintenance of paved roads. This II contract, called coinrrato dc recuperaci6n y n:anteni;iziento (CREMA), requires the z contractor to rehabilitate and then maintain a network of roads for five years for a lump sum amount. Each contract covers a network comprising contiguous or area-specific road sections ranging in length from 100 to 300 kilometers. The c° contract specifies the sections that need rehabilitation and the minimum solu- C CD tion required to ensure a positive net present value for the investment. D Bidding 0 In designing the contract, different rehabilitation strategies xvere tested, involv- C ing spreading the rehabilitation across the life of the contract to avoid funding - spikes, using contracts of different durations, and adjusting the payment sched- 2 ules to reduce financing costs. The highways authority settled on contracts a requiring rehabilitation works to be carried out during the first year, and rou- P 2 tine maintenance activities throughout the five-year contract period. > Bidding is done through international competitive tenders. In early bid- ding rounds the payment schedule called for a 5 percent advance followed by two equal payments of 1(1 percent, with the rest in equal installments over the next four years. But the bids exceeded official estimates by nearly 100 percent because of high financial costs (since most expenditures would have been in the first year, contractors would have had to borroxv).These early rounds were canceled, and a schedule with larger up-front payments was chosen to reduce contractors' financial costs. Only after the contract is awarded does the contractor prepare a detailed engineering design. On the basis of its own risk assessment, the contractor is 41 free to propose any rehabilitation solution above the minimum threshold defined in the contract. (This involves judgments about how much up-front rehabilitation is required to get the roads to a level at which they can be cost- effectively maintained.) But the contractor is not allowed to change the agreed financial bid. The decision to postpone the detailed designs until after contract award xvas made to expedite the bid proposals (and thus the reduc- tion in the rehabilitation and maintenance backlog) and to cut the bidding costs. z Payments and monitoring c The payment schedule is designed to provide incentives for the contractor to C/) iiimaintain the network for the full length of the contract. The contractor x receives an advance payment of 5-10 percent, followed by 15-25 percent at XD the end of the first six months, when specified activities have been executed, m and 25 percent at the end of the first year, xvhen rehabilitation works have been completed. Thus up to 60 percent is paid by the end of the first year; the remaining payments are made in 48 equal monthly amounts. In addition, c0 A the contract requires a performance guarantee of 20 percent. e:i The contract allows reimbursement of cost overruns in certain circum- CD ,D stances beyond the control of the contractor, such as earthquakes, hurricanes, m O and bitumen shortages. The government uses the contractor's schedule of 0 input prices submitted in the bid as a baseline for overrun estimates.The risk > of excessive cost overrun is contained by a 25 percent cushion on these m prices. If the contractor's estimate exceeds the baseline by more than 25 per- z > cent, the contract can be rebid. In contrast with input-based contracts, under the CREMA payments are z A made when the contractor achieves a specified level of service. Performance is > assessed during monthly on-site inspections by the government engineer and the contractor.Throughout the contract period the rehabilitation works must com- ply with the specified minimum and maximum standards (box 1). The comph- ance with maintenance standards is inspected visually on a monthly basis. Penalties for noncompliance are set for each indicator. For example, a pothole left unrepaired beyond the authorized time limit will cost the contractor US$400 a day until it is patched. Penalties are deducted from the monthly payments. Road users can also monitor performance, voicing concerns about the quality of service in a claim book available at the contractor's site office. Entries in the claim book are publicized in the local media. Contractors must signpost each network with information about how they can be contacted. 42 R!J Rehabilitation and maintenance indicators Throughout the contract period rehabilitation works must: * Meet or exceed the minimum thickness of overlay. * Not exceed the maximum level of roughness, rut depth, cracking, or raveling. Regular visual inspections of maintenance activities focus on a few essential items in ensuring compliance with the specifications: Z * Potholes, cracking, and rutting. > * The condition of shoulders, culverts and drains, and the roadside environment. * Guardrails and vertical and horizontal signs. 3 0 And a representative of the user community is periodically allowed to partic- ipate in monthly inspections. , Bidding for phases 1 and2 2 The CREMA program was designed to be implemented in two phases.The first phase involved a network totaling 11,700 kilometers in length, 55 per- I 0 cent of the nonconcessioned national paved network.That network was gen- c erally in good to fair condition, with 25 percent in poor condition, and had daily traffic averaging about 750 vehicles. In a process following international > 33 competitive bidding practices, 60 contracts were let in 1997, covering sub- networks averaging about 181 kilometers in length. 0 The contracts were awarded to mostly local construction companies for > a total of US$650 million, equivalent to US$11,000 per kilometer a year.With the larger up-front payments, the lowest bids exceeded the budget estimate by about 24 percent. Rehabilitation works accounted for 74 percent of the total o bid amount, and routine maintenance for 26 percent. (The costs-US$66,000 per kilometer for rehabilitation over the 8.25-year life of the works and an D estimated US$3,000 per kilometer each year for maintenance-are roughly in line with those in other parts of Latin America.) Private sector participa- tion was high, with each contract attracting 5-20 bid proposals. The average contract price was US$10 million. The second phase involves a network of 4,000 kilometers and 2( con- tracts. Bidding of the first two subnetworks, initiated in August 2(0)00, received a positive response from the private sector: the lowest evaluated bidders offered financial proposals 5 percent below official estimates. This outcome is probably the result of increasing private sector comfort with the contracting process, a higher share of up-front paynments (and thus lower financing costs), 43 and better estimating by the highways authority. The remaining subnetwvorks are expected to be tendered in 2001. Assessing results for phase I contracts The first phase was successful in many respects during its first three years: 4 * By requiring contractors to perform their own quality control, the sys- tem has cut the government's cost of supervising the network. * The fixed price contracts reduced the risk of cost overruns. The only z z ~~~cost increases so far have been due to natural disasters or force majeure o, events (mostly related to El Nino) and have amounted to about 3 per- o cent of the total contract price. cz 0* The requirement that contractors carry out detailed engineering designs m x before initiating the works has minimized delays in project implementa- U tion. In traditional programs such delays are due to lack of stocks of m o government-prepared subprojects. * By making the long-term payment obligations legally binding on the I government, the CREMA has deterred the Treasury from failing to pro- 0 H vide funding for road maintenance. * The performance indicators have been simple enough to apply and monitor, and they get the desired results. (Output indicators invariably m o involve a tradeoff between accurate measurement of the road service required and unambiguous and low-cost measurement.) > * The contractors' obligation to maintain the roads over a five-year C q period has reduced the risk of unsatisfactory quality in the rehabilita- D tion works. * The system has fostered some innovation in the programming and exe- cution of works, since payments are tied not to rigid specifications on > workmanship but to outputs. Nevertheless, as contractors get used to the new system, they are starting to question the appropriateness of uniform national standards and to ask that they be allowed to set the standards once the government has defined the quality of service. * Ex post financial and economic evaluations showed that the rehabilita- tion and maintenance funding yields an economic rate of return of 60 percent (at a 12 percent cost of capital). The contracts will reduce the need for capital investments by nearly 30 percent: after the five-year implementation period better quality roads will lead to a drop in on- going capital and maintenance expenditures from about US$11,300 per kilometer a year to US$8,000. 44 * Rates of return for the contractors have not been assessed. But the com- petition for the contracts and the fact that only one of the 60 contracts has had to be canceled because of a contractor's financial difficulties sug- gest that the contracts are financially attractive to the private sector. * The CREMA program has substantially improved the condition of the network, reducing the share of roads in poor condition from 25 percent 4 to less than 5 percent by the end of 1999. As a result, road users' costs have been reduced by more than 10 percent. > * Damage to roads caused by vehicle overloading is being addressed by > asking contractors to provide and operate devices for measuring axle D loads on-site and to report any excess load problems to the highways C authority. But the contractors still have to rely on the government for enforcement. 3l m ID Conclusion z m Argentina's approach to road maintenance offers an effective means to E improve efficiency and public accountability. By holding contractors account- C able for the future quality of the roads, output-based contracts keep themn more alert to quality during the execution of road works. And by passing C some monitoring functions on to contractors-and requiring the permanent > presence of their maintenance crews on-site-the contracts guarantee effi- cient molnitoring of pavenment and traffic conditions, leading to more timely 0 corrective actions. > In the long run this approach could both reduce the cost of maintenance A and improve its quality.The approach is one that could be transferred to other countries. Indeed, pilot initiatives with similar contracts are already under way in Brazil, Chile, Colombia, Guatemala, Paraguay, and Uruguay. 4 45 Extending rural electrification A survey of innovative schemes Ray Tomkins Ray Tomkins (rayX Output-based contracting for rural eca-uk.com) is an energy electricity services-using private providers economist with wide and linking payments of subsidies to experience advising on outputs could deliver better results than power sector reform and traditional approaches to subsidized rural electrification in electrification. Even so, few rural developing countries in electrification schemes have explicitly (and Africa, Asia. and Eastern successfully) linked subsidies to output Europe. In 1997 he set up targets. This is probably because most Economic Consulting electricity sectors remain highly centralized, Associates Ltd., based in politicians find it hard to move to cost- London He was previously covering tariffs, standards are still tightly a director of London regulated, and entry is often illegal for small Economics, where he private operators. But recent initiatives managed the work of the provide useful lessons for developing full international utilities team. output-based contracting. Coupled with the examples in this book from other sectors, they suggest that by more precisely targeting subsidies, sharpening incentives for operational efficiency, and mobilizing private investment, output-based contracting could boost consumer access to services and cut funding costs and performance risks for taxpayers and donors. As a result of low population density, difficult terrain, and low consumption, rural electricity schemes are usually more costly to implement than urban schemes. In addition, low rural incomes can lead to problems of affordability (though where electricity replaces other commercial fuels, such as kerosene, households' energy costs may fall rather than rise). And the long distances mean greater electricity losses and more expensive customer support and rrl equipment maintenance. Thus rural electrification projects have often m required subsidies to make them financially viable. o In the past most subsidized rural electrification schemes have been run by high-cost, centralized public utilities charging tariffs that do not cover costs. Cash-constrained governments generally have not explicitly made up the difference. As a result, electrification rates remain extremely low in many o developing countries-as low as 10 percent in some African countries, for example. But the wave of electricity sector reform in the developing world- D~ breaking up public monopolies and privatizing generation and distribution- 6 ° provides an opportunity for a new approach, indeed, demands one. > One approach with much potential is output-based contracting. Its focus co c on outputs gives operators the flexibility and the incentive to innovate and to 1m _< respond to consumer preferences. Several countries are implementing or devel- 0 I oping rural electrification schemes using elements of output-based contracting. o This chapter reviews the lessons from these cases, focusing on decisions about who should receive the subsidy, how to link the subsidy to performance, how to select projects, and how to monitor contracts. I rnU Emerging approaches In Argentina a program of exclusive concessions pays private operators the low- est subsidy required to connect consumers in isolated rural areas far from the electricity grid to off-grid services based on renewable energy. Under this pro- gram concessionaires must provide service to all who ask for it within an exclu- sive area. The connection costs are partially subsidized by a World Bank loan, a Global Environment Facility grant, and a special electricity fund run by the Argentine government. Users must contribute at least 10 percent of the costs, with the share depending on capacity to pay in the province and on the size of the system. Connection subsidies are paid to the operator on proof of installa- tion (checked by random audits) and decline over time. One province has nego- tiated a pilot concession with an existing provincial distribution concessionaire, and three others will award new (negotiated) off-grid electrification conces- sions in 2001.Where there is no local concessionaire or the operator does not 48 want to provide service, the new concessions will be offered in a competitive tender for the smallest subsidyThe program, estimated to cost US$120 million, will cover about 70,000 households and at least 1,100 schools and clinics. A similar approach has been proposed in Cape Verde. Under competi- tively bid concessions, subsidies would cover part of the cost of installing wind generators or solar photovoltaic systems for houses, street lighting, or elec- tricity sales. Still in the design phase, the proposed scheme would disburse subsidy payments every six months on adequate proof of completed equip- x ment sales or installations. Z z Chile uses a rural electrification fund with a planned life of 10 years to offer one-time, competitively awarded subsidies to local operators bidding to provide service.The goal of the scheme is 100 percent electrification by 2004. m Local operators, often working with community groups, commit to a target C for newv connections. Their proposals are scored against a checklist of objec- tive criteria, including a cost-benefit analysis, the operator's investment com- mitment, and social impact. Although grid connections are preferred, ° renewable off-grid systems can also get support. Operators receive the subsidy > up front and must make a minimum contribution to project costs according to a formula set by the government. Operating since 1994 and using annual -< 0 tenders, the fund had increased rural electrification by 50 percent in 1999. The subsidy paid over this period amounts to US$112 million, while private C operators have invested US$60( million. Panama is applying a variant of this approach-"open season" competi- tion, in which blocks of money are offered to private companies bidding to construct the largest number of connections. Bidders can identify potential C connections from a database maintained by the rural electrification office. A social fund (using donor and government funds) disburses payments against agreed construction milestones, with random audits to check installations. Once construction is complete, the assets are transferred to the local private distribu- tion company, which commits to operate and maintain the infrastructure for 20 years.The distribution company calculates the subsidy required for the 20 years, and this is paid up front in a lump sum. In 1999 the program funded just over 100 projects, with an average size of 34 connections. Nonexclusive concessions are also allowed, but so far there have been no takers. In Asia countries have more commonly turned to cooperatives for rural electrification, applying performance-based subsidies with mixed success. In Bangladesh and the Philippines cooperatives have to meet specified targets- 21 in Bangladesh, S in the Philippines. Payout of the donor-provided subsi- 49 dies to the cooperatives depends in part on their achieving the annually nego- tiated targets, such as reducing system losses, increasing sales, meeting cus- tomer connection targets, improving collection rates, and repaying loans. In Bangladesh targets are also the basis for annual bonuses to cooperative employees.The cooperatives' performance has been mixed.While some have achieved the targets, others have suffered financial losses, requiring the diver- sion of subsidies to offset operating losses. m m D) Deciding what to subsidize o Schemes with output-based characteristics have delivered subsidies as capital D grants for extending or creating an isolated grid or for installing small gener- m ators powered by diesel or renewable energyv They have also delivered subsi- dies for connecting consumers. And some schemes have delivered subsidies for rehabilitating networks. n 2R Capital grants have been paid to operators as they begin the construc- 0 tion works, as they complete the works, or in phases as they reach milestones. co And sometimes when many areas are being electrified under a phased pro- C_ gram, grants are paid as a certain number of villages gain access to electricity -< or a certain number of consumers get connections. 0 C' Most grants have partially funded investments, not consumption. (Con- o sumption subsidies tend to erode operational efficiency and require long-term funding commitments.) Partial funding, requiring a supplementary contribution CD from users or operators, is a good test of user demand and preferences.The grants : can be provided as one-time payments following a grant competition (as in Chile) '] or paid for the completion of specific investments (as in Argentina and Panama and as planned in Cape Verde). A few schemes plan to pay grants per household, but these tend to be for solar photovoltaic systems for individual buildings and are paid to the suppher on installation or through equipment financing. In some rural electrification schemes that include off-grid systems (as in Argentina and as proposed in Cape Verde), grants are paid at a declining rate over the project period. The rationale is that the higher initial grants are needed to overcome "first cost" barriers, to help promote self-financing for later development, and to encourage a more rapid connection rate. Linking payment to performance Rural electrification schemes have used several output targets: * Connecting all consumers in an area who wish to be connected and are willing and able to pay the required (subsidized) costs. 50 * Connecting groups of households or villages in an area. * Installing systems (such as solar photovoltaics) in houses or buildings. Other performance targets include reducing electricity losses and improving revenue collection. Measuring performance in these areas is more complex, however, and no known schemes have directly linked the disburse- ment of aid to such targets. Where targets are insufficiently defined or monitored or the infrastruc- ture to support delivery of electrification services is missing, problems can m result. Grant funding targeted at equipment installation may end up support- ing better-off households rather than the poor, because the wealthier are bet- II ter able to pay their share of the costs. Moreover, even when installation ' targets are met, electricity services may not be delivered. In India the release m of subsidy funds is based on the number of villages electrified. But sometimes m the lines have been poorly constructed or household connections not made, so that little electricity flows to consumers.The cooperative schemes in Asia, 0 which use many different performance measures, give greater attention to ser- vice delivery, but the weighting of different factors dilutes the direct incentive D cC) effect of linking payment to performance. 0 Choosing projects m Rural electrification projects can be quite small, so that in many countries the C 0 number of potential projects could run into the hundreds. In selecting pro- jects, well-defined criteria are important to ensure that funds are allocated m DO efficiently and the program is financially sustainable. T Rural electrification progranms with output-based features have all devel- 7C oped a system for ranking and prioritizing projects, taking into account such factors as capital cost, number of consumers, and demand growth. In choos- ing areas to be electrified, programs need to avoid top-down selection and political patronage. The key issue is whether projects will be financially viable (with the subsidy)-whether the communities will have sufficient demand and whether they will be able to afford the electricity payments to cover operating costs. Projects that are not financially viable tend to drag the entire program into losses, resulting in the diversion of subsidies to sustaining the existing system rather than expanding access to electricity. A particularly effective approach to selecting projects is to emphasize the contribution-especially the financial contribution-to be made by the local organization sponsoring the project, whether the community, a cooperative, a 51 private company, a local authority, or a nongovernmental organization. In Panama the larger the share of financing to be provided by the community, the higher up the priority list a project moves. The Chilean scheme requires both the users and the distribution company to contribute to the financing. Users must pay the costs of connection from the distribution transformers to the house and of the wiring within the house, m roughly 10 percent of the cost of the project. This cost is initially financed by the operators and repaid by the user over time.While the distribution company O must contribute a minimum amount calculated by a formula, it may contribute more to increase its chances of receiving a grant.This demand-driven approach to choosing projects helps to ensure that those selected have local support and mr that there is sufficient willingness to pay for electricity. It also improves the o) probabihty that the forecast demand for new connections and electricity will materialize rapidly, thus helping to ensure projects' financial viability, and that the allocation of capital costs and subsidies is targeted toward maximizing the a ° desired output-the delivery of electricity services. Selecting operators _< For selecting operators, some recent schemes have introduced an element of 0 I competition, with the grant linked to the number of new connections. The o competition has been based on either the sniallest grant to supply a given number of consumers or the largest number of consumers for a given grant. FT In some cases the competition gives the winner a concession to supply all potential consumers in an area (an exclusive concession). In others, the competition gives the winner only the responsibility to supply specified vil- lages or households (open entry, in which other competitors could supply additional consumers). The Chilean scheme uses competition for grants but does not offer an exclusive concession. Companies prepare submissions specifying the areas to be electrified and the number of consumers to be connected.These subniis- sions compete for grants, but there is no direct competition between compa- nies to electrify a given area. A drawback of the Chilean approach is the imbalance of information between the companies and local authorities.While municipalities and regional planning agencies may exert some pressure on the companies at the planning stage to keep costs low, their lack of information limits their effectiveness: municipalities have little understanding of electricity distribution, and regional planning agencies little knowledge of geographic constraints to electrification in particular areas. 52 The proposed scheme in CapeVerde would channel aid through private companies that successfully tender for concessions-one to supply electricity services on Santiago Island and another for the eight other inhabited islands. The companies, which will not have exclusive concessions, will be obligated to sell off-grid electricity systems based on wind or solar photovoltaics or to install such systems and sell the electricity. The successful tenderers will be those asking the smallest fee (subsidy) to start up and manage the concession to provide these services. Other companies might enter the same market, ensuring some competition, but they would have to compete against the sub- sidized concessionaire. Also possible, however, is that only one company v would emerge, as the same company could bid for both concessions. A ten- der for the concessions xvas launched but did not attract sufficient interest, m largely because the costs are high, the ability to pay is low, and the grant was , insufficiently attractive. The scheme is being redesigned. Nonexclusive area concessioning is also used in mini-grid or off-grid schemes, such as for village mini-grids or household photovoltaic systems, with ° the rights granted to firms resembling a license or dealership. Such schemes > GD should aim to build the delivery capability of local private firms to sustain the a connection rate and ensure equipment maintenance.They can provide incentives < 0 based on installations achieved or on equipment subsidies. More advanced forms focus on energy service companies, which provide consumers both equipment and ongoing maintenance for a regular fee. The companies also provide the D financing for consumers, supported by a grant or equipment financing. The concession approach appears attractive because it offers instant market i aggregation, keeps transactions costs for the government relatively low, and sim- K plifies the flow of finance. But in isolated rural areas markets may be neither large nor viable enough to attract substantial private participation and competition for the market at any level of subsidy. An alternative is to build delivery capability among small enterprises through business advisory services and business devel- opment and working capital financing. This approach, though initially slower, may ultimately be more financially sustainable. One variation of this approach, the "dealer sales" model for delivering household solar systems, is being tried in a number of countries, particularly in Asia, such as Bangladesh, China, and India. Exclusive and nonexclusive concessions differ in the information rights and incentives they provide for identifying viable projects ("prospecting"). In an exclusive concession the concessionaire has the right to supply all consumers in a defined area but also the obligation to connect consumers. To determine whether extending the grid is economic and what it will cost, the concession- 53 aire needs to carry out preinvestment studies. In return for bearing the cost of these studies, the concessionaire gains exclusive right to the information. For nonexclusive concessions, the information about prospective connec- tions needs to be publicly available, so private companies would not be keen to carry out such studies. If public bodies provide the information (as the rural elec- trification office in Panama does), the costs of gathering the information could be subsidized, with the subsidy linked directly to the number or size of areas mapped. Fm x zm o Monitoring performance and ensuring accountability CD Concessions and other models for subsidized private participation in rural electrification create a contractual relationship for carrying out such obliga- tions as extending the grid at a specified rate, connecting new customers, o meeting investment targets, and installing off-grid equipment. These obliga- tions need to be monitored and enforced-by the local authority granting the > concession, a national regulator, a rural electrification agency, or even the util- ity (for technical standards, for example). One well-tested approach is to have > a dedicated rural electrification agency carry out the monitoring.The agency can also play a supporting role, acting as a catalyst for private sector and local participation and providing financial, technical, and managerial support. A m dedicated, independent agency can provide a focused supporting framework Z U o for the effective disbursement of performance-related subsidies. -4 Where a demand-based approach is adopted, as in Chile, the local com- Fm munity wil feel a close link to the scheme and can be expected to alert regula- tors or the government if promised programs are not dehvered. Panama's scheme, <,, which disburses subsidies in phases based on a schedule for connections, relies on random audits instead. But once a project is connected to the main grid, it becomes the responsibility of the distribution company, and the project area is regulated just as any other area xvould be. Isolated grids are regulated by contract with local communities. A similar arrangement is used in Chile. In CapeVerde it was originally hoped that using multiple concessionaires, and thus introducing competition, would mean that prices to users would not need to be regulated. But with the redesign aimed at making the concessions more attractive, in part by reducing the threat of entry by competitors, it is now expected that stronger regulation of prices and standards will be needed. Conclusion Why has output-based contracting been so limited in electricity? Several fac- tors play a role: 54 * Output-based contracting is best implemented by an organization in close touch with local conditions. But until recently most utilities have stayed under centralized control, and the stringent technical (safety) stan- dards required tends to keep centralized utilities involved even where local participation is strong. * The large funding required for rural electrification subsidies (including donor funding) has usually been administered centrally. In addition, utili- ties have found it hard to move away from using "hidden" cross-subsidies. m * The remoteness and small scale of individual connections have made it Z hard to put in place an effective monitoring system for measuring per- formance against output targets. The experience with contracts using output-based elements suggests several lessons. Delegating responsibility for evaluating and selecting schemes 2 to the lowest practicable administrative level (as in Chile and Panama) helps _ to ensure that the projects selected are those most likely to be financially viable. And combining demand-based project selection with community responsibility for a substantial share of project costs (by using tariffs that are > not too heavily subsidized, as in the Philippines, or by encouraging the coni- munity to bear sonme of the investment costs, as in Chile) helps to ensure that m projects achieve output targets. There is strong evidence that subsidies should be directed toward access 2 to electricity rather than consumption. Investment subsidies can directly link the disbursement of aid to the most important output indicator-the num- ber of newly electrified villages or new household connections. C: Competition for grants can lead to the lowest capital subsidy per new 4 connection and create incentives to minimize operating costs. State-owned companies serving rural communities and cooperatives face no direct pressure to keep costs low unless the regulator, government, or municipality is vigilant. Private companies, having won a competition at a fixed or regulated tariff, have their own incentives to minimize costs. Concessions, grant competitions, demand-based project selection, and dis- bursenient against milestones are all promising approaches. Many variants on these methods are possible, and indeed, approaches will need to be fine-tuned to fit each country's circumstances. Strong supporting measures can enhance success, including a sound investment climate and an agency dedicated to pro- moting efficient, financially sustainable schemes of rural electrification. 55 Educating mothers for health Output-based incentives for teaching oral rehydration in Bangladesh Sadia Chowdhury Sadia Chowdhury Reducing infant deaths from diarrheal (schowdhury@worldbank. diseases poses a major challenge in many org) isa senior health developing countries. Education of mothers specialist at the World Bank. can be an effective way to address this Before oining the Bank in challenge, but it is time-consuming and 1998. she worked as often difficult to monitor. In Bangladesh a director of the Health and nongovernmental organization launched an Population Division of the education program aimed at teaching Bangladesh Rural mothers how to prepare and use oral Advancement Committee rehydration solution to treat diarrhea. To (BRAC), a nongovernmental ensure that the teaching was effective, the development organization program relied on output-based incentives, with an annual budget of tying field-workers' pay to fast-cycle US$131 million and more feedback on performance against output than 25.000 full-time staff indicators. Monitoring results show that the Together with Oxfam. BRAC approach worked: the mothers learned funded and rana trial of the effectively. And over 10 years the program oral rehydration program. reached 12 million households. Bilateral donors funded the subsequent rollout of BRACe large-scale program, run in three phases over 10 years. Diarrhea was the leading cause of infant mortality in Bangladesh during the 1 980s. Diarrhea leads to the loss of fluids and electrolytes, resulting in mild to severe dehydration and, in some cases, death. Because it is the loss of fluids and electrolytes that causes illness and death, it is essential that they be replaced. Intravenous therapy is critical for treating cases of severe dehydration, but the 6 means are expensive and difficult to transport. A severely dehydrated person would require 5-10 bags of saline. One bag cost 100 taka (Tk) in Bangladeshi o villages in 1983, when the average income was Tk 1,500 a year (Chowdhury and Cash 1996; Shepard, Brenzel, and Kenneth 1985). Moreover, administer- oD ing this therapy requires sterile fluid, needles, tubing, and trained professionals. 0 Choosing the solution © Seeking a solution, the Bangladesh Rural Advancement Committee (BRAC), a nongovernmental organization, realized that what was needed was an inex- pensive therapy easily administered by nonprofessionals using no special C equipment. BRAC saw that the best option was a fluid that could be given C orally, which would eliminate the need for sterile equipment. An oral rehy- o dration solution was developed in Bangladesh, clinically tested, and perfected 'm so that it was effective in treating dehydration from all types of diarrhea in infants and children, even cases of severe dehydration, as long as the children ° were alert and could drink the fluid. To work, though, the solution needed to < be available in the community and used effectively as soon as an episode of C diarrhea began. --i_U At this stage BRAC weighed several options for a diarrhea prevention M > program: . Having trained personnel provide treatment at fixed government facilities. * Providing safe water and sanitation. > * Making packets of oral rehydration solution broadly available through °, commercial outlets. is u Teaching mothers how to make the solution at home with commonly >s available ingredients. A treatment program relying on trained personnel was rejected because w there were too few facilities and health workers. An effort focused only on improving water and sanitation was also rejected, because experience with tubewell programs to supplv safe drinking water showed that when the wells m E vwere broken, people resorted to unsafe sources of water. Marketing packets of oral rebydration solution had many advantages. The product was easily identified, it was a complete mix, and it could be eas- 58 ily distributed to all small shops. But it was impossible for Bangladesh to pro- duce the volume needed. Treating all episodes of diarrhea only in children under five-79 million in 1989-would require about 200 million packets a year.The government had the capacity to produce only about a third of that. Moreover, the annual cost of US$16 million was onerous. Commercial production and marketing of packets also appeared imprac- tical, as people had limited purchasing power and the distribution system was C) weak. Moreover, the printed instructions for mixing and administering the solution could not be read by the 80 percent or so of the population who A were illiterate. But the main flaxv of a commercial initiative was that it would G) not include efforts to build people's awareness of the deleterious effects of ° diarrhea and the effectiveness of oral rehydration solution in combating it- crucial in overcoming social and cultural barriers to the solution's use. T For all these reasons, the option chosen was to teach mothers to prepare I an oral rehydration solution and treat their children themselves. So in 1980 > BRAC launched a pilot program supported by its own funds and by Oxfam. I 0 After the pilot the teaching program was rolled out in three phases over 10 C years, with the US$9.3 million cost funded by different donors. The donor ' funding was disbursed every quarter as a reimbursement of expetnditures. The > m cost of teaching each household was estimated at US$0.75. m 2 Teaching mothers < m The teaching program had three aims: m * To reduce diarrhea-related illness and deaths, particularly among chil- dren under five. > E To teach at least one woman in each household to prepare the oral rehy- C dration solution. C * To raise axvareness in the community about diarrhea prevention. > Using mainly one-on-one and group communication techniques, this _ program taught a 1(0-point health message, including how to prepare the solu- tion using local ingredients and a simple but accurate measuring system (box D 3 1).The solution is prepared from a three-finger pinch of loboti (common table z salt) and one fistful of gur (unrefined brown sugar) in half a seer (467 cubic centinmeters) of water. The loboi-gur solution has almost all the essential prop- erties: it is simple, cheap, safe, effective, and readily available (BRAC 1980). > The female health workers teaching women how to prepare the solution I first had to overcome some suspicion and confusion. They had to convince villagers that the solution was not harmful. And because the last time health 59 e What diarrhea is and how it is transmitted. * How to prepare the rehydration solution. * What can go wrong if the quantities are not right (too much salt, too large a 6 dose). * When to consult a doctor. m 0 > o workers had appeared in villages was to teach family planning, they had to ° overcome an assumption that the solution was a contraceptive. T 2 In the program's first phase most health workers were from the district C vwhere the program xvas being implemented. This allowed the health workers access to their families, and it eased communication with mothers, since the health workers could talk to them in their own dialect. The health workers had to be 20-35 years old and have at least 10 years of schooling. 0 2 Teams consisting of 14-16 health workers and a team coordinator worked together in a village (unless a village was small). They visited all the 2 households, with each worker covering an average of 10 a day. Before leaving a household, they made sure that the mother had understood the messages by 2 asking questions about each of the 10 points. Most important, they had the mother make the solution. The health workers reviewed the accuracy of the m CO measurements and then asked the woman to repeat the process. If a health 0 o:1 worker was dissatisfied with the woman's level of understanding, she repeated > the entire session. Each session took about 20-30 minutes. By December 1990 the health wvorkers had reached two-thirds of the country's rural 0 households-around 12 million. 0 D 2 Creating incentives through the salary system 2 There was concern that a normal monthly salary system would make the DH health workers more task oriented than results oriented, particularly as the program began to expand. So an incentive system was developed that linked nD their earnings to results-mothers' knowledge of diarrhea and their ability to G) prepare the solution. > A monitor visited 10 percent of the mothers a health worker had taught m T over the past month, asked them questions about the 10-point message, and had them prepare the solution. To determine the health worker's pay, the mothers' responses were graded from A to D and the number of mothers receiving each 60 grade was multiplied by 10, since only 10 percent were interviewed. For each A the health worker received Tk 4, and for each D, nothing (table 1). This salary system put the emphasis squarely on ensuring that mothers learned correctly. During the pilot phase health workers received Tk 600 a month on average (US$40 at the 1980 exchange rate, a good income in rural areas). The range was Tk 400-700. With the minimum salary set at Tk 250, much of the pay was based on performance. (There was no incentive-based salary system for male xvorkers, and questions were raised about gender bias.) The incentive system improved teaching methods. Before it was intro- duced, only the health workers prepared the solution during a teaching ses- a sion. But the performance-based salary increased their incentives to ensure ° that the mothers had learned to prepare the solution correctly, so the health r workers began to ask the mothers to demonstrate. This also increased the s mothers' interest in the teaching session. I D Monitoring and evaluating results 0 The program's results were monitored both internally and externally in all -D phases. Monitoring costs were estimated at about 4 percent of project costs. AC 9~~~~~~~~~~~~~~~ m Internal monitoring 0 The salary system required an effective, built-in monitoring system. During n 2 the pilot phase each health worker gave a list of the mothers she had taught A that day to her supervisor, who randomly selected I 0 percent of the mothers En p~~~~~~~~~~~~~~~ 0 and assigned a monitor to interview them. Men were recruited to work as monitors, as the job required extensive travel. AI~~~~~~~~~~F ~~~~~~0 Rate paid for each grade Dt> m Mother's performance Grade Rate (taka) a Remembered all 10 points and prepared the solution correctly A 4 a Remembered 7-9 points and 2 prepared the solution correctly B 2 UJ Remembered fewer than 7 points but prepared the solution correctly C 1 Failed to prepare the solution correctly D 0 r Note In 1981 the rates for grades B and C were ncreased by Tk 1 Source. Chowdhury and Cash 1996, BRAC 1984 Bhat a, Cash, and Cornaz 1983 61 In the expanded program the selection of mothers had to be done at a higher level and the monitors kept separate from the health worker teams to prevent collusion. The lists were sent to Dhaka, where monitors were ran- dom-y assigned to visit 5-10 percent of the mothers a month. One technique used to ensure the veracity and accuracy of the monitors' reports was to ask w them to record the name of each mother's youngest child. This information m was then compared with that collected by the health worker to confirm that the monitor had visited the right mother. BRAC's Research and Evaluation Division played an invaluable role in o helping the program to continually increase its effectiveness. An emphasis on A evaluation as a means to improve results rather than to punish the workers I helped to create a positive attitude. The division encouraged field-workers 0 and supervisors to suggest ways to improve the system, and their ideas often LU T led to research that helped to verify field observations. m External monitoring C_ The monitoring also required an analysis of the solutions prepared by the 71 mothers. For this purpose, the International Center for Diarrheal Disease m >o Research helped BRAG set up field laboratories. As a quality control mea- sure, the center reanalyzed 10 percent of the samples. In addition, its staff ° nmade routine supervisory visits to the field laboratories (Chowdhury and A Cash 1996). During periodic visits a technical advisory committee of international 0 ° experts provided valuable technical assistance to the program.And the donors > conducted independent assessments of the program at the end of each phase. I Conclusion > Monitoring showed that 90 percent of mothers scored in the A and B cate- r gories. Two years later only 65 percent of these mothers scored A or B. To o increase retention, follow-up education was introduced on local radio and tele- > vision and in schools. Mortality rates fell after the program was implemented, 6 LU but it is hard to isolate the effects of the oral therapy from other factors. mU Analysis showed that the program had characteristics that allow scaling up: It dealt with a problem common to all of Bangladesh. The intervention >) was relatively simple. It was also inexpensive, requiring households to pur- chase only salt and gur. The training and messages built on existing skills and knowledge, such as cooking and child care, and were culturally acceptable. The health workers' performance could be measured through the knowledge 62 acquired by mothers. Though the program was large, it was possible to put in place an administrative structure of checks and balances and rigorous super- vision. Finally, the program had a clear goal and specific outcome indicators. BRAC's experience with the output-based salary system shows that the approach can be successful in certain circumstances. First, the outcomes sought must be tangible and quantifiable. Second, there must be an indepen- dent monitoring unit, something that many organizations may not have. And third, it must be possible to dismiss nonperforming employees rather than reassigning them to other positions, as required by labor laws in many coun- D tries. BRAC has only recently begun to use the technique again-in a rl community-based education program with four clearly defined HIV/AIDS D messages. 0? c ED GD l D GD References Advancement Committee." Program in Rural rl Abed, F H. 1999. Swiss Agency for Bangladesh.'' Doctora the-° "sDevelopment Cooperation Development and sis. University of Londonm on the Brink of a New Cooperation, Bern. Chowdhury, A. M. R., andI Century: What Role for BRAC (Barngladesh Rural R A. Cash. 1996. A Simple r C) -1 NGOs?" UvA Development Advancement Committee). Solution: Teaching Millions to > r- Lecture Amsterdam 1980. "Oral Therapy Treat Diarrhoea at Home. mro Research Institute tor G obal Extens on Program Phase I." Dhaka: University Press Issues and Development Dhaka. Shepard, D. S., L. E.D Studies --.1984. Report of Brenzel, and K. T Kenneth, -z Bhat a, S., R. A. Cash, Phase I of the Oral Therapy 1985 "Cost-Effectiveress ofr and 1. Cornaz. 1983. Extension Program. Oral Rehydratbin Therapy "Evaluation of the Ora Dhaka for Diarrheal Disease." > Therapy Extersion Chowdhury, A M. R Harvard School of Publicr Programme of the 1986. "Evaluation of a Health, Institute for Health Bangladesh Rural Commurnity-Based ORT Research, Boston. 63 Promoting preventive health care Paying for performance in Haiti Rena Eichler, Paul Auxila, and John Pollock Rena Eichler In Haiti nongovernmental organizations (reichler@msh.org) is a provide basic health services such as health economist at immunization and prenatal and maternal Management Sciences for care. In an effort to improve the Health (MSH), a US.-based effectiveness of some of these nongovernmental organizations, the U.S. Agency for organization working in International Development introduced developing countries. She is performance-based contracting in 1999. based in MSHs Washington, The providers had been operating under a O.C., office and has worked payment system that reimbursed their on performance-based expenses up to a ceiling. The new system payment schemes in Haiti set performance targets and withheld a and Kenya. Paul Auxila, also portion of their historical budget, allowing at MSH, leads the Haiti them to earn back the withheld amount prolect. John Pollock plus a bonus if they met the targets. A one- provides US.-based support year pilot involving three of the providers to the Haiti project from showed some marked improvements in MSH's central office in performance. Boston, Massachusetts. I Those paying for health care services in developing countries typically have not required the providers to guarantee their performance. Public payers tend to fund public institutions to maintain capacity (paying salaries and recurrent costs) rather than to ensure that consumers receive high-quality services. Any contracts with private providers generally have not held them accountable for 7 performance. Donors have tended to adopt similar practices, providing lump - sum grants or reimbursing public providers and nongovernmental organiza- c tions (NGOs) for documented expenditures. As a result, providers tend to 0 focus on securing funds rather than improving efficiency or the quality of 0 OD care. In this context, in 1995 the U.S. Agency for International Development (USAID) launched a 10-year project in Haiti aimed at strengthening the m< capacity of NGOs to deliver primary health care services. A key part of this effort was the introduction of a performance-based payment system. The I challenge was to develop a system based on attainment of goals without > imposing an excessive burden of monitoring and reporting requirements. T Following competitive tenders, USAID awarded funding for the two- D phase, US$92 million project to Management Sciences for Health (MSH), a o) U.S.-based NGO operating in developing countries. MSH manages and dis- 0 burses the funds. During the first five-year phase, beginning in 1995, the pro- S ject provided funding to 23 NGOs, an established group that had received ° USAID support in the past. For the second five-year phase, beginning in > 2000, the number of NGOs increased to 33. m When the project began, the immediate need was to develop rapid mech- anisms for funding NGOs so that they could provide critical basic health ser- t vices, including maternal and child health and family planning services. Initially, and in line with general practice, NGOs were reimbursed for expenses up to a ceiling that was essentially a negotiated budget. Under this expenditure-based financing NGOs submit a proposed annual budget and a plan showing how they intend to ensure the delivery of a basic package of services. Then each month they submit cost reports with detailed documentation of their expendi- tures for reimbursement. NGOs are free to set their own fees for services. Most charge patients for drugs, and some for consultations. Switching to performance-based contracts A 1997 population-based survey to review the existing system found that NGO performance was extremely uneven. In vaccinations a good performer reached 70 percent of the target population, while the worst performer 66 reached only 7 percent. One NGO made sure that 80 percent of women knew how to prepare an oral rehydration solution; another educated only 44 percent. Some NGOs provided the minimum two prenatal visits to 43 per- cent of pregnant women; others reached only 21 percent.These wide-ranging results were not correlated with costs (average costs per patient visit ranged from US$1.35 to US$51.93). So in 1999 MSH decided to test a new approach-performance-based payment. The new payment system was expected to lead to efficient delivery o of high-quality services in several ways: ° * Because institutions receive a bonus if they achieve performance targets, Z) they feel strong incentives to attain those targets. * Because institutions assume financial risk for improving performance, . they feel strong incentives to use resources efficiently and effectively. < I * Because institutions are paid on the basis of results, they face strong I incentives to improve management, motivate staff, and innovate. I Three NGOs, serving about 534,000 people, participated in a one-year pilot study. Under the performance-based system NGOs receive an up-front m payment and then a quarterly sunm rather than submitting their expenditures > z every month. At the end of a defined period-one year in this case- ° performance is measured and the size of the bonus determined. To ensure that the NGOs viewed the change as advantageous, MSH , used a collaborative approach in designing the new system. NGOs demon- ° strating the leadership and institutional capacity to respond to the system were > invited to meetings to express their views about the pilot. Because these m meetings occurred after NGOs had signed contracts for fiscal 1999 (October T 1998-September 1999), they were willing to renegotiate only if the new con- - tract could make them better off. The meeting,s led to agreement on a new contract that would pay 95 percent of the budget under the expenditure-based contract-but would also pay a bonus of as much as 10 percent of that budget.The NGOs thus assumed a financial risk: if they failed to attain performance targets, they would lose 5 percent of the budget under the original contract. But they were willing to do so because they also had the possibility of earning 5 percent more than the budget. Seven performance indicators were chosen, and a target was negoti- ated for each indicator and linked to a share of the bonus (table 1). (Negotiating with MSH, each NGO then translated the general targets into specific targets.) Five indicators related to improving health impact, 67 Share Indicator Target of bonus Women using oral rehydration therapy to 7 treat diarrhea in children 15% increase 10% Children ages 12-23 months receiving full :D vaccination coverage 10% increase 20% 0 < Pregnant women receiving at least 0 3 prenatal visits 20% increase 10% Z Discontinuation rate for oral and injectable contraceptives 25% reduction 20% m < Clinics with at least 4 modern methods 100% of clinics; m z of family planning; outreach points with 50% of outreach at least 3 points 20% Average waiting time for attention to a child 50% reduction 10% m > Participation in local health organizing 7 committee (UCS) and coordination with the C, > Ministry of Health UCS defined 10% r7 > Source: Authors' comp lation. 0 °D one to increasing consumer satisfaction by reducing waiting time, and one :33m to improving community participation and coordination with the Ministry ° of Health. > Another goal of the project was to improve institutional sustainability. m To facilitate learning and sharing, the project helped create a network of local NGOs. Regular meetings encouraged NGOs to share strategies that have succeeded or failed in the challenging Haitian environment. The pro- ject also provided technical assistance, to help NGOs review their pricing policies and develop a plan to generate revenue through sources unrelated to health services. CORE, a cost and revenue analysis tool, was used to help NGOs identify unit costs, revenues, and staff utilization (MSH 1998). The goal was to promote a culture of information-based decisionmaking to improve efficiency. Measuring performance Since payment is tied to performance, the NGOs agreed that reporting on their own performance would create incentive problems. MSH contracted a neutral third partv-l'lnstitut Haitien de l'Enfance (IHE), a local survey research firm-to measure baseline and end-of-pilot performance. 68 Using the standard cluster sampling methodology recommended by the World Health Organization (WHO 1991), IHE sampled households in each NGO'.s service area to measure immunization coverage, based on both immu- nization cards and reports from caretakers. IHE determined the percentage of women using oral rehydration solution to treat diarrhea through exit interviews at clinics with women who brought children in for other reasons. It reviewved a sample of medical records to find out what share of pregnant women had had three or more prenatal visits. Discontinuation rates for oral and injectable con- 3 traceptives were determined by reviewing family planning registers to identify women who had discontinued use, had not chosen another method, and had a not expressed a desire to have a child. And average waiting time was determined through measures in a sample of institutions at different intervals. m This survey needs to be done annually, first to set up a baseline and then < a to check performance against this baseline. The annual cost is US$40,000, less than US$1 per person benefiting from the project. E The results one year later m The most striking result was the increase in immunization coverage in all three NGO service areas (table 2). In two of the three areas the share of moth- C ers who reported using oral rehydration therapy increased-and so did the share who reported using it correctly. Performance was relatively weak in , meeting prenatal care and contraception targets, probably because of the need ° for ongoing counseling and behavioral change. The availability of modern > contraceptive methods increased substantially. " Waiting time was judged to be an invalid indicator of quality because people who have to travel long distances to obtain lab tests might wait an _ entire day for results rather than come back. A new indicator of client satis- faction is being developed for the next phase. And because no easily measur- able and verifiable indicator could be devised for community participation and collaboration with the Ministry of Health, the bonus linked to this per- formance indicator was given to all three NGOs. All the NGOs in the pilot received more revenue than they would have under the expenditure-based scheme-and all supported continuing performance-based payment.The shift from justifying expenditures to focus- ing on results inspired them to question their model of service delivery and experiment with changes. The NGOs' possibility of earning bonuses sharp- ened staff's focus on achieving goals and led to innovation, including greater efforts to involve the community. 69 NGO 1 Indicator Base Target Results Percentage of women using oral 7 rehydration therapy 43 50 47 Percentage of women using oral -I rehydration therapy correctly 71 80 81 o Immunization coverage (percentage of o children ages 12-23 months) 40 44 79 ZD Prenatal visits (percentage of pregnant -z women with at least 3) 32 38 36 m Contraceptive discontinuation rate (percent) 32 24 43 m z Clinics with 4+ modern family planning methods 6 9 9 m T Note: Base data refer to September 1999, results to April 2000. > Source Authors' comp at on. D m To motivate staff to focus on results, two of the three NGOs introduced D bonus schemes for staff. And one introduced a bonus scheme for community health agents, cutting their salary in half and reserving the rest for bonuses tied to performance. But when it found that transferring this much risk to rela- tively low-paid staff lessened their motivation, it increased the fixed share of payment and reduced the share from bonuses. Another NGO set up a bonus 0 Dscheme for local organizations with which it works, and all considered allo- m cating a share of any bonus they earn to clinics in their network on the basis of performance. - The performance-based payment also motivated NGOs to request assis- tance in strengthening their strategic planning, strategic pricing, cost and rev- enue analysis, human resource mnanagement, and measurement of client perceptions of quality. The future The results of the pilot suggest that performance-based payment is a power- ful way to hold NGOs accountable for results.The challenge is to define indi- cators that relate directly to health impact, client satisfaction, and institutional sustainability and to measure and monitor performance in a xvay that is not prohibitively expensive. MSH staff and the NGOs in the pilot will work together to develop new indicators and improve processes for measuring and verifying performance. 70 NGO 2 NGO 3 Base Target Results Base Target Results 56 64 50 56 64 86 53 59 26 61 67 74 , 0 49 54 69 35 38 73 o 49 59 44 18 21 16 - 43 32 30 26 20 12 m 2 5 5 0 5 5 r -r m I More NGOs will be incorporated into the performance-based payment sys- m Fl tem each year. And to increase performance incentives, a model being con- > sidered for fiscal 200)2 wvould reduce the share of payment based on historical r2 F- :0 Note healthflagship/journal/ WHO (World Health This chapter is based on index.htm. Organizat on). 1991 The EPI a longer paper by the same Coverage Survey WHO/EPI/ authors, NrPerformance-Based References MLM/91.10. Geneva. Payment to Improve the MSH (Management Impact of Health Services: Sciences for Health) 1998. Evidence from Haiti," avail- CORE: A Tool for Cost able at http //www. and Revenue Analysis, worldbank.org/wbi/ Boston . 71 Improving primary health care Output-based contracting in Romania Cristian Vladescu and Silviu Radulescu Cristian Vladescu In 1994 the Romanian government (cvladescu&buc.osf.ro) is introduced a pilot scheme of output-based executive president of the contracts for the provision of primary health Center for Health Policy and services. The contracts were aimed at Services in Bucharest and supporting efforts to develop the an associate professor in the independent provision of primary services, Faculty of Sociology and to increase the share of health spending Social Assistance of going to preventive care, and to improve Bucharest University Silviu access to health services, especially in Radulescu (sradulescu@ rural areas. To help achieve these woridbank org) isahealth objectives, the scheme relied on output- specialist with the Europe based financial incentives and competition and Central Asia Human between doctors for patients. The pilot Development Unit at the resulted in higher patient satisfaction, more World Bank. Both authors primary care services for patients, and were involved in the early higher incomes for doctors. But the weak design of health reforms in regulatory and monitoring capability of the some of the Romanian purchasing authorities diluted the benefits. districts piloting the new Now, with modifications based on lessons approach to primary health from the pilot, the scheme has been care The pilot was part of a extended across the entire country. major project to restructure health sector financing and management, Begun in 1991, the project was supported by a US$150 million World Bank loan. Under Romania's old health care system patients had httle confidence in pri- mary care services and often bypassed them, seeking care directly from special- ist physicians and hospitals.Treatment was free, in theory, at the point of service for all levels of care-though many paid under the table for better service, espe- cially for specialist care. Primary care doctors faced little incentive for giving preventive care or ensuring patient satisfaction. Their incomes (low compared wvith the average income, and much lower than doctors' relative incomes in OECD countries) were determined by seniority and length of service, with no 0 < link to the quality or volume of services provided. And patients were simply z assigned to practitioners on the basis of their residence or employment. Fragmentation led to both gaps and overlaps in primary care services. And as a result of resource scarcity (public health spending averaged around 3 percent of GDP in 1990-97) and a distribution of poxver in favor of hospitals, primary care centers had even less access to drugs and equipment than other levels of care. > In the early 1990s Romania started to reform its health sector-to shift JJ m * the balance from hospital services toward primary care, and from curative care 0 I toward preventive; to address inequities in access to basic services resulting from inadequate staffing (especially in rural areas) and funding; and to give patients choice over their doctor. Fr c As part of the reform a pilot scheme introduced output-based contracts '2 for primary care in 8 of the country's 40 districts, covering about 4 million > people, between 1994 and 1996.The contracts anticipated moves to set up an z autonomous national health insurance fund (financed by a 7 percent tax on p salaries and a 7 percent levy on employers) and develop an accreditation sys- o tem for health care providers to improve the quality of care. > Designing the payment system The new contracting system used competition backed by financial incentives as a mechanism for shifting resources. In designing the contracts policymak- ers had to strike a balance: to avoid under- or overspecifying service require- ments, to set workable and monitorable performance targets, to allow the flexibility to respond to demand, and to protect the funding system against abuse and budget blowouts. In the eight pilot districts one-year service contracts were concluded between doctors and district health authorities, ending the doctors' status as hospital employees. The contracts covered primary health services (curative and preventive) for the patients registered with each doctor, and the services continued to be free. Doctors had to be accredited, but the criteria were basic 74 and contracts were offered to all general practitioners with an open-ended employment contract with public hospitals. Patients were allowed to choose their family doctor and to switch after three months. To hold a contract, a family doctor had to have at least 500 reg- istered patients (fewer would suggest poor service and would result in an inef- ficient allocation of funds to cover the fixed costs of running a medical practice). The optimum number of patients was deemed to be 1,500. The payments to doctors combined a capitation adjusted for patients' 0 ages (about 60 percent of the total) and reimbursement for about 30 fee-for- < -z service items (about 40 percent). A capitation is a fixed payment for provid- 0 ing a list of services for a set period to an enrolled patient. Providers received the payment whether or not the patient used the services, so in principle they had an incentive to keep their patients healthy.The capitation was also meant I to give physicians in areas with a surplus of providers an incentive to move to areas with a deficit, where it would be easier to build up patient lists.The bal- ance between capitation and fee-for-service payments was calibrated so as to m 0 create incentives for both quantity and quality of service. C The capitation was increased for work in remote areas and difficult con- C ditions (by 20-60 percent) and for professional qualifications (5 percent for > specialists, 10 percent for senior specialists)-and reduced for not providing u coverage at night (5 percent) or on weekends (10 percent). (The contract ° required 24-hour availability for emergencies.) These adjustments were designed as incentives to increase access in underserved areas and improve the scope and quality of services. The contract provided no clear guidelines on what was expected for the 0 capitation, however, and defined fee-for-service requirements only in general > terms. The fee-for-service payments covered primary and secondary clinical > preventive care, including periodic checkups, immunizations, pregnancy and child developnient monitoring, cancer and tuberculosis detection, and public health activities.These payments were expected to create a strong incentive to provide more preventive services, which are in lower demand. To calculate payments to doctors, their outputs were converted into points reflecting relative values. To discourage doctors from registering too many patients, points were progressively reduced above two thresholds-by 50 percent above 1,500 registered patients and by 70 percent over 2,000. And to take into account the higher demand for services by infants and the elderly, these patients were assigned three to four times as many points as a young adult. The monetary value of a point was determined quarterly by dividing 75 the budget for paying family doctors in the eight districts by the points reported. The contract required doctors to submit monthly reports on changes in their patient list and on the number of fee-for-service items pro- vided. District health authorities had the right to audit the doctors' records. Assessing the results The small number of indicators collected for routine monitoring has limited analysis of the pilot's results. Most data were collected for calculating payments < and are in paper form rather than electronic, hampering their use. Moreover, most are aggregate, preventing before-and-after comparisons of such indica- tors as immunization rates and the share of patients getting pre- and postna- tal checkups. A one-time evaluation in most of the districts showed large I variations, reflecting inconsistent interpretation of what services are to be reimbursed and a lack of specificity in the guidelines. > The data that are available show steady growth in the share of the popu- lation registered with family doctors-from 72 percent in the first quarter to 82 percent after one year and 86 percent after two vears-with registration aver- - aging 8 percentage points higher in urban than in rural areas. (The rest of the people may be receiving no care or still seeking care at hospitals.) Few patients o have changed doctors, though the number differed significantly among districts. z Two years after the pilot started the average patient lists in urban and rural > areas were almost the same (about 1,740), although there were more people per physician in rural areas (2,125) than in urban (l,929).The smallest lists (the high- est density of physicians) occurred in the urban areas of a poorer district (about 1,200) and the rural areas of relatively well-off districts (1,300-1,400)-and the > largest lists in the rural areas of a poorer district (nearly 2,500).This pattern sug- > gests that in poorer districts family doctors prefer to practice in urban settings, even though that means a lower income, and that better-off districts attract more doctors, who have to build up rural practices even if they live in towns. Family doctors increased their output, providing 21 percent more con- sultations and 40 percent more home visits, and at the end of 1996,87 percent were providing emergency coverage at night or on weekends. The doctors identified higher patient expectations as a major factor in their increased work- load. Indeed, surveys revealed that family doctors had become more client ori- ented. And interviews of both doctors and patients provided some evidence of a reduction in informal payments (already relatively low at this level of care). In the first months there was a tendency toward "inflation" of fee-for- service items. Payments for these exceeded the 40 percent initially estimated 76 as a maximum, so payinent rules were changed and monitoring mechanisms strengthened. When fee-for-service points exceeded 65 percent of the capi- tation points, they were reimbursed at only 50 percent of their value.Auditing reports for reimbursement became a niajor task for the inspectors hired by district health authorities. Feedback from doctors suggests that the audits were uneven and erratic, reducing the credibility of the payment scheme. For an average family doctor with a specialist qualification, caring for about 1,700 patients, the capitation accounted for about 65-70 percent of income and I fee-for-service for the rest. Incomes increased by 1 5 percent on average (to about US$180-20() a month), with more than 80 percent of participating doctors experiencing an increase. The range of incomes also increased, with the top 5 percent earning more than four times as much as the bottom 5 percent. Technical efficiency has increased in primary care, with output growing I more than payments to service providers. But an effect on the overall effi- 0 ciency of health services is difficult to prove. Although referrals to hospitals and polyclinics reportedly declined (by 8 percent in 1996), there are doubts m 0 about the quality of the data.There seemed to be no effect on hospital admis- sions, which increased by 3.6 percent in 1996, in line with the national trend CD (a 4 percent increase). > m 'Neighing the prognosis° By introducing competition for patients, the schemne was expected to encour- age doctors to expand their services rather than refer patients to hospitals, reducing health care costs. Although family doctors do report fewer referrals, there has been no significant change in hospital admissions. In urban areas with an adequate supply of doctors, however, competition does seem to be > producing the desired results. > Still, a lack of influence on the utilization of hospitals and specialist care would not be surprising, since no disincentives to unnecessary referrals were built into the payment system. Moreover, most patients still expect to be eas- ily referred to specialists, and therefore would find family doctors responsive to that expectation more attractive. Giving family doctors budgets for pur- chasing specialist services seems an attractive solution, but it would require the purchasing agency to have technical competencies for monitoring purchases. A more realistic solution is to focus on developing monitoring capacity in the purchasing agency and professional associations, improving equipment and increasing the range of clinical skills offered by family doctors, and changing patients' expectations. 77 The new delivery and payment arrangements had a positive effect on the quality of services. But the effect was limited by doctors' lack of control over resource management-which prevented them from influencing nurses' per- formance and improving the physical conditions of practice-and by the lack of incentives and conditions for professional development. Although the reg- ulations for the pilot provided for a practice budget that would have allowed doctors to make autonomous decisions about expenditures on equipment or X3 maintenance, in reality they had no possibilities for doing so because of bud- getary constraints. Attendance by family doctors at training courses was poor, in part because of the lack of financial resources and a good system for ensur- ing that a substitute would be available. The "optimum" list size of 1,500 patients proved too small to improve the distribution of family doctors, providing insufficient competition, and the disin- centives over the threshold of 1,500 proved too steep. Further undermining >4 competition, district health authorities turned out to be unwilling to make the "hard" decision of denying contracts to family doctors with small hsts. Moreover, no mechanisms were designed to provide financial support for physicians mov- ing to underserved areas, and the incentives from bonuses proved too weak. Dn> The underspecification of requirements for the capitation led doctors to m ,) see it as an entitlement. Among fee-for-service items, the technical quality of 0 z some was questionable and difficult to monitor (such as cervical screening, > routine examinations, and detection of breast and prostate cancer), and the frequency with which they were provided varied widely.This is unsurprising, given the lack of clear guidelines for these services, the inadequate equip- 0 ment, and the uneven training of physicians. > After a new health insurance laxv came into effect in 1998, payment > mechanisms similar to those piloted in the eight districts were introduced for general practitioners throughout the country for services contracted by newly established district health insurance funds. Lessons from the pilot led to some changes, however: * To reduce the complexity of the capitation, the number of age groups and rates was reduced. * Regulations on the capitation describe more clearly the services that need to be provided. * The threshold above which the capitation is reduced was set at 2,000 (3,000 in localities with insufficient family doctors). * A 100 percent bonus was added to the capitation for family doctors practicing in remote or low-income areas. 78 * For vaccinations, fee-for-service payments were simplified (awarding all vaccinations the same number of points), and fanily doctors can claim payments for vaccinating patients not on their lists (such as children tem- porarily residing or attending schools in the doctor's practice area). * Requirements for routine checkups are more clearly defined. * To reward effective intervention (rather than mere reporting of clinical activities of uncertain quality), more points are awarded for screening and detection of cancer and tuberculosis, but only after confirmation by a specialist physician. < * An allowance based on the number of registered patients was introduced C _U to cover all practice costs.This allowance is managed by the doctors, who have gained significant discretion over spending on staff and maintenance. * Rather than setting a cap on individual doctors' fee-for-service pay- I ments, the new system splits the primary care budget into allocations for I capitation, fee-for-service, and practice budgets. * The College of Physicians, established in 1997, has started to develop practice guidelines and requires doctors to participate in continuous C medical education as a condition for periodic recertification. District- C level accreditation committees have been established, with joint repre- > sentation from the College of Physicians and district insurance funds. C The new health insurance system led to a 25 percent (real) increase in ' health spending in 1999 compared with the mid-1990s.This increase allowed > a significant rise in doctors' incomes, the main element in primary care costs. But by the end of 2000 pressures from the increasing costs of drugs and hos- pital care were reversing the initial shift in resources toward primary care. 0 Conclusion > Romania's output-based payment scheme for primary care physicians, along with free choice of provider for patients, has the potential to improve the quality of services and increase doctors' productivity. But purchasing author- ities operating in a weak regulatory environment, with insufficient capacity and experience, have faced serious difficulties in monitoring both the num- ber and the quality of services reported (billed) by providers. The changes have not yet significantly reduced the use of hospital services or redistributed providers to improve access in rural areas. The system may need to establish more credibility before it can encourage patients to change their behavior and doctors to move to underserved areas. 79 Pursuing output-based education The evolution of contracts for schools in the United Kingdom David Goldstone David Goldstone The private sector has provided investment (david.goldstone@ funding and services for U.K. public partnershipsuk.org.uk) is schools under output-based contracts project director at since 1997. Under these contracts firms Partnerships U.K. He has provide accommodation and related been involved in projects in services, while teachers, still employed by the school sector and the public sector, provide the core currently leads a joint education services. Firms bid their lowest venture with the National price, and payments to the winning bidder Health Service that will begin only when services become available deliver new investment in at defined standards. The contracts have primary care facilities and evolved from building new schools, to services. Partnerships U.K. providing maintenance and rehabilitation is a joint venture between across many schools, to setting up and the public and private supporting information technology systems sectors but works while incorporating learning targets for exclusively for the public students. Contracts focused on facilities sector to develop, procure, can improve education outcomes, but only and implement public- to a limited extent-raising questions about private partnerships. whether output-based contracting could be extended to education services. I New schools have traditionally been procured by local governments through input-based construction contracts specifying the inputs-or the entire design-in great detail. Once a school was built, the local authority and the school were responsible for managing and maintaining the asset and deliver- ing such services as cleaning, catering, security, and utilities. But the input-based approach has had a rather inglorious history of lead- ing to delays in construction, significant cost overruns, and maintenance costs that became steadily less affordable. A key attraction of the output-based con- z tract, pioneered in the United Kingdom by the Private Finance Initiative in the early 1990s for asset-intensive transport and government accommodation services, is its focus on outputs-what is required rather than how it is to be m7t delivered (box 1). That transfers the risk of cost and time overruns from the E school and local authority to the contractor, which is paid only when the out- F7 puts are delivered. In most cases the private contractor must raise finance in D-) the private sector to cover the initial construction works. o Under this approach the contract goes beyond just the provision of the asset-the school. By transferring responsibility and risk for the ongoing nmanagement, maintenance, and operation of the asset to the contractor, it gives the contractor the incentive to develop design solutions that take into account the costs of managing the asset over its entire life. This ensures that the contractor does not design an asset that is cheap to build but unaffordably 0 o>; expensive to manage and maintain over the rest of its life-as the public sec- tor often has in the past. > This approach offers important advantages to the government, not only more efficient risk allocation and access to private sector expertise but also the delivery of new assets without pressure on government resources. The 0 m I c ~~~~Key features of the output-based approach: C + *~~~ A competitive bidding process (reputable bidder with lowest price and best U ~~~~~value for money wins). z *~~~ Specification of the outputs required, not the method of delivery. 9 * Procurement of a service, not the underlying asset. GD 02 g * Allocation of risks to the party best able to manage them. * Value for money as the overriding objective, K Value for money assessed over the whole life of the asset. • Payment for service as received, not as asset delivered. 82 approach started later in education than in other asset-intensive service sec- tors, largely because local authority regulations constrained it until 1996. Moving from single to bundled contracts The first stage of the Private Finance Initiative in U.K. schools was an output- based contract for a single school in the Dorset area of southwest England (table 1). This first contract set a precedent in defining outputs that has been C_ widely followed as school projects have evolved. The contract defined output c requirements in terms of the conditions that would make a room or space "available" for use, such as the level of lighting, heating, and ventilation. It also D defined standards for ongoing services, such as cleaning, catering, and build- ing and grounds maintenance. The contractor's incentive to meet the requirements was established mainly through the fact that payment would occur only when the output stan- dard was met. If a space (or the entire school) was not available for use, no pay- 0 ment relating to that space would be made. Deductions for nonavailability were 0 weighted to reflect the operational importance of different areas-with faculty offices, for example, weighted less than classrooms. Deductions could also be made if rooms were available but the performance standards had not been met. c 0 Th.V Selected output-based schools contracts 0 I in the United Kingdom n 0 C z Local Value authority and (millions of o contract date Schools Requirement Pupils U.S. dollars) 0 Dorset, 1997 1 Replace secondary 1,000 35.6 0 school 00 Stoke, 2000 122 Refurbish or replace 38,000 185.0 0 all schools C Dudley, 1999 104 Provide managed infor- 44,000 49.8 mation and commun- ications technology c C service for all schools Z Glasgow, 2000 29 Refurbish or replace 30,000 540.6 secondary schools and provide managed a information and com- c munications technology service Source Author's compl at on. 83 Payments are made in roughly equal installments (minus any penalties) from the date of first availability over the life of the contract.They are partially indexed to reflect the fact that some of the contractor's costs (such as running costs) will rise with inflation, while its capital costs will be incurred early in the contract, financed by borrowing at a fixed rate, and therefore do not require indexation. The next step was to bundle groups of schools into one project. This approach offered a range of benefits, including lower transaction costs for -D both the public and the private sector. Several such projects are now in oper- C z ation.The new facilities have generally been delivered on time and to cost and o have met-and in some cases exceeded-expectations. But the projects have X involved little innovation. XD The next stage of contracting began to address this issue. The Stoke m schools project, covering all 122 schools in Stoke on Trent, defined its require- m ments in terms of standards that should apply to categories of accommodation o in any school, rather than school by school as in previous contracts. The spec- 6 ification document was brief-much briefer than those for many single-school A schemes, which have typically given a lot of detail about such issues as site con- m straints and design aesthetics. Covering, as before, only such areas as thermal o comfort, ventilation, lighting, and fixtures, the specifications defined for each 6 standard a performance level that would make the accommodation available, a Z level at which it would be available but performance deductions could be 0 made, and a level at which it was nonavailable and no payment would be made. A By focusing purely on accommodation standards, the local authority gave the ,> contractor complete discretion over how to meet the standards. As a result of the Stoke project's focus on high-level outputs, the con- o tractor's solution has shown much more innovation than those in many single- site schemes. For example, many schools that the local council expected to be refurbished or repaired are being demolished and replaced, because this solu- cD tion is more cost effective over the 25-year life of the contract. I The paynment arrangements in the Stoke scheme have higher-powered C: incentives too. If nonavailability continues for a long time or recurs often, the m payment deductions can exceed the total payment for the affected space and start to eat into payments for other parts of the contract.Thus the contractor's o incentive to meet the output requirements increases as its exposure increases. Extending standards to learning Contracts developed from the focus on school buildings to encompass the installation and maintenance of information and communications technology 84 systems across many school properties.These contracts have extended output requirements to education performance. Output standards and incentive structures centering on education per- formance began with the Dudley Grid for Learning scheme. In this project the contractor is providing a managed information and communications technology service-an integrated network, with hardware (computer labs with personal computers and printers, a personal computer in each classroom) and software (word processing, spreadsheets), access to email and the Internet c for everyone, school administration systems, user support such as data backup z and help desks, and maintenance. The project includes refreshing the tech- C nology over the 10-year life of the contract and extending use of the facili- ties to the wider community to support lifelong learning and strengthen community links. Payment for the service is based on performance against availability, usage, and learning gain targets. Availability payments are based on the avail- ability of the required services and facilities for use at the required standards, H and usage payments on the number of pupil log-ins during each school term. H The usage payments are designed to give the contractor an incentive to make m I the service useful to pupils in doing their work: the more they use it, the more o the contractor will be paid. 0 The learning gain payments increase over the life of the contract, after z 0 performance benchmarks are established in the early years. In the second half -n C 0 of the contract period, in years 5-10, 15 percent of the payment will be based z on the impact on education outcomes. This impact will be assessed by an > independent third party on the basis of such criteria as pupils' and teachers' attitudes about using technology in learning, attendance levels, and educa- m tional attainment, including performance on national achievement tests. n 0 By including payments linked to education outcomes, the Dudley con- o tract begins to match the contractor's objectives much more closely with m those of the teachers. Equally important from the contractor's point of view, I m the risk related to education outcomes is a relatively small part of the total c project risk and so is proportionate to the level of influence that the contrac- E tor's performance can have on education outcomes. A more recent project, the Glasgow Secondary Schools Project, is 0 unique in many respects, but its key feature is its strategic focus on achieving rapid improvement in education performance across the whole secondary school population in a big city. This project is the only one that has combined a whole system approach to accommodation requirements with a require- 85 ment for fully managed information and communications technology service. The Glasgow schools contract was signed at the end ofJuly 2000 by the local council and the 3ED consortium.The principal service providers under 3ED's contract are a construction company, a business services firm, and several information and communications technology firms. The technology agree- ment will run for 12 years, and the accommodation requirements for 30 years. With the new technology installed almost immediately after the contract was signed, education benefits are already being achieved. p As in the Stoke project, the focus on outputs allows the contractor to opt C o for new construction as a way to meet the council's requirements for many _J of the schools in the scheme. The council had expected 2 of the 29 schools to be rebuilt, but the contractor is rebuilding 11. ') The specifications and payments relating to acconmmiodation require- m ments are similar to those in earlier school projects. But as in the Dudley pro- o ject, the technology elements of the agreement require the contractor to 5 enhance education performance, and payment is at risk if it does not. The main emphasis is on availability of the technology service. A sliding scale of penalties tied to the impact of the problems on the teaching process creates o an incentive to ensure continuity of service. The Glasgow contract also features a learning gain element, with deduc- 0 tions made if achievements in reading, writing, and mathematics (in national C , tests for pupils age 13) do not improve by agreed margins each year.This ele- z nment accounts for only 5 percent of the payment linked to the information > and communications technology, but it is significant in aligning the interests cn of the contractor with those of the teachers. As in the Dudley scheme, the 12 0 o contractor's risk relating to education outcomes is supposed to be propor- o) tionate to the influence it can bring to bear. o The contractor's current rate of progress suggests that all 29 Glasgow secondary schools will have been replaced, expanded, or refurbished-with E1- fully managed information and communications technology services installed 1n and operational-within two years of the contract signing. Moreover, accord- ing to the published comparator, the contractor will do all this for 9 percent less cost than the council could have if it had carried out the project itself. cc 0 Could the scope be widened? Better school facilities and up-to-date technology are widely recognized as beneficial in education-in boosting students' and teachers' morale and moti- vation, in making modern education methods possible, and in enabling teach- 86 ers to focus on teaching rather than dealing with failures in facilities or sys- tems. In this way output-based schools contracts can have an impact on edu- cation outcomes-but only a limited one. Much more important than facilities and technology are the quality of teaching, the management of the school, and the nature of the school's intake. Could the scope of education contracts be widened? To do so would require specifications and performance and payment regimes for contractors that focus on education outcomes for the pupils. In drawing up such regimes, C local authorities would have to decide whether to measure performiance z against an absolute standard, based on, say, national targets. or against a target 0 retlecting the schools' intake and historical performance. These would be dif- ficult issues to resolve both technically-where to pitch the standards?-and commercially-how would the contractor respond to, and price, the risks C involved? m Difficult issues touching on education policy would be raised-for C example, relating to a contractor's ability to influence a school's intake or to c exclude pupils, and to its employment policies for teachers. And contractors exposed to risk linked to education performance would want full control over the inputs that affect outcomes and so might challenge existing public sector < approaches to delivering education services-a political hot potato. 0 z 0 C 0 0 (D 0 m C: m 0 8o 87 4-< 0 (DI, cw ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~,Yr.4,.. Designing output-based aid schemes: a checklist Warrick Smith Any discussion about designing output-based aid schemes must begin with an understanding of the underlying strategy-in particular, how it differs from traditional ways of funding and delivering public services. Traditional approaches to providing health, education, infrastructure, and other public services channel public funding (whether sourced from domes- tic taxpayers or international development assistance) to the labor, materials, and other inputs consumed by state-owned monopolies, with at best indirect links between funding and the delivery of services. In the developing world especially, results have often been disappointing. Incentives for efficiency and innovation have been weak. Accountability for performance has been dismal. And opportunities for leveraging scarce public resources through private financing have been limited. Output-based aid seeks to address these xveaknesses by delegating service delivery to a third party under contracts that link payment to the outputs or results delivered. It thus has the potential to improve incentives and account- ability while also expanding opportunities for mobilizing private financing. The focus shifts not only from inputs to outputs, but also toward the Holy Grail of development outcomes. From strategy to implementation The extent to which the potential benefits of output-based aid can be real- ized depends critically on the design of individual schemes. There are no standard models or blueprints, and approaches need to be adapted to the characteristics of the service and to the environment in which it will be delivered. As with any policy intervention, the design of output-based aid schemes often involves tradeoffs between goals-such as providing high- powered incentives for efficiency, attracting sufficient interest from service providers, ensuring effective monitoring of performance, and minimizing administrative costs. Further complicating design is that many issues are interrelated. For example, the definition of the service to be provided may be influenced by budgetary constraints-and in turn may influence the def- inition of eligible service providers, the amount and structure of payments, the duration of the contract, and the design of appropriate monitoring arrangements. For these and other reasons, designing effective output-based aid _ schemes is neither simple nor mechanical. But while many of the issues are challenging, they are not altogether novel. Indeed, important insights can be o' gained from the solutions to similar issues in other areas, particularly in pri- vate infrastructure arrangements.' c t To navigate the key issues in designing output-based aid schemes, it may QJ be helpful to group them under eight headings-a checklist of major design 0 > questions: * Clarifying the role and sustainability of public funding. What is the rationale I for public funding? How might budgetary constraints and sustainabilitv m issues influence design? Co > * Deciding Wilo will be eligible to receive services that attract publicffunding. Who are the intended recipients? How will they be targeted? _ Deciding who will be eligible to provide services. What criteria should govern eligibility? * Choosing the market environment. Will services be provided in a competi- tive Or monopolistic market? * Defining performance. What should the service package include? How should key performance standards be defined? n Linking payment to pe formance. What should be the form and size of pay- ment? How will payments be structured? * Shaping other aspects of the contract. What should be the form and duration of the contract? HoW will issues of contractual adaptation and dispute settlement be addressed? * Structuring the administration of the scheme. What should be the scope of the scheme? Who should be responsible for administering the scheme? Clarifying the role and sustainability of public funding One of the hallmarks of output-based aid is a clear focus on the intended results. This means that policymakers and their advisers must clarify the rea- sons for public funding and consider the feasibility and sustainability of that funding from a budgetary standpoint. 92 Establishing the policy rationale for public funding Many public services can be financed through user fees, without recourse to taxpayer or donor funding.This is increasingly the case for infrastructure ser- vices such as electricity, telecommunications, water supply, and many modes of transportation, and for many health and education services. But reliance on user fees alone can give rise to efficiency and equity concerns. Efficiency concerns will arise if a service has characteristics that make individual preferences as expressed through user fees a poor measure of social welfare. For services that have "merit good" features-as some education ser- z vices do, for example-consumption creates benefits for society in addition ° 0 to those captured by individual consumers.2 For these services, reliance on C C user fees alone would lead to underconsumption relative to the socially opti- mal level. For services that have "public good" features-such as public >, C defense and public health interventions-the benefits from consumption are not depleted by additional users, and it is difficult or impossible to exclude C people from the benefits even if they are unwilling to pay for them. For this reason, user tees usually are not a feasible financing strategy for services with a the characteristics of pure public goods. > Equity concerns xvill arise if reliance on user fees limits the ability of the I poorest members of society to meet their basic needs. This is a particularly pressing challenge in the developing world, where 2.8 billion people live on - less than $2 a day (in purchasing power parity terms).And these concerns can be particularly acute in rural areas, where providing services is often more expensive than in urban settings. Complementing user fees with subsidies can be an important part of strategies for promoting universal access to public ser- vices. The need for such support will have to be assessed case by case, how- ever, taking into account the costs of providing the service and recipients' willingness and ability to pay for it (box 1). Questions to consider * Can the service be adequately funded from user fees alone? * When public funding is proposed on the basis of "merit good" considera- tions, has the gap between recipients' willingness to pay and the expected social benefits been assessed? * When public funding is proposed on the basis of "public good" considera- tions, has the basis for this claim been assessed? * When public funding is proposed on the basis of affordability concerns, have recipients' willingness and ability to pay been assessed? 93 -~,fne-* -_ - 0 Traditional approaches to pricing public services have often relied on untested assumptions about the willingness and ability of users to con- tribute to the financing of the services through user fees. For example, pub- licly provided infrastructure services in developing countries have typically been priced substantially below the full economic costs of supply, ostensi- U bly to help the poor. But the main beneficiaries have tended to be the more ;, affluent, with the poorest lacking access to any service or paying consider- Z ably more for poor-quality substitutes (see World Bank 1994). cz 0 In reviewing the need for subsidies for a public service, a first step is to c assess users' willingness and ability to pay. One way to do this is through C7 surveys, for which there are established techniques. Another is to look at D~> actual expenditure on substitute services. In many developing countries, for m example, the poor often pay prices for water from tankers or other informal ZwD providers that are 10-15 times the price for services provided through U household connections by an efficient supplier. U I Many societies espouse a goal that expenditure on certain basic services > should not exceed a fixed percentage of household income. Chile's subsidy c) policy, for example, is based on a view that subsistence-level water and san- T U itation services should account for no more than 5 percent of a household's '; income. Governments considering such policies need to determine whether they can mobilize sufficient funding for the required subsidies in the near term. Considering the sustainability of financing Budgetary and donor resources are subject to many competing demands, plac- ing practical limits on public funding for services no matter how compelling the policy rationale. The volume and sustainability of the funding available are therefore key considerations in designing output-based aid schemes, particu- larly for those involving multiyear commitments. These considerations may influence the size and focus of the scheme, including the services eligible for public funding, the amount of subsidies that can be paid, and the pool of eli- gible recipients. (For simplicity, the term subsidy is used here for public pay- ments designed to address either efficiency or equity concerns.) Sustainability considerations may be particularly important in choosing the action that will attract public funding. Subsidies for consumption typically require a long-term commitment unless they are directed to one-time events, such as immunizations, or are part of a transition strategy for moving to user fees that fully cover costs, as in Guinea's water sector (see chapter 3). In some 94 cases it might be more feasible to direct subsidies to household or commu- nitv connections to such services as water and electricity, leaving consump- tion to be financed from user fees. Well-designed output-based aid schemes can help ease budgetary con- straints. Improving the targeting of subsidies and strengthening the incentives for efficiency can increase the impact of a given level of public funding. Schemes can also help mobilize private financing in support of service provision.3 In Peru the rural telecommunications scheme required a subsidy of US$11 per inhabitant, a but mobilized an estimated US$22 per inhabitant in additional private invest- z ment (see chapter 1). Indeed, in many cases even a relatively modest subsidy pay- 0 0 ment may be enough to make larger projects attractive to private providers. C -u C These factors, coupled with improved accountability for results, can also help i CD reduce resistance to funding assistance prograims anmong taxpayers and donors. >U) r Questions to consider CD * What is the budget envelope for financing subsidies for the service, and T how long will that funding be secure? i * How will budgetary constraints influence the size and focus of the scheme, > including the amount of subsidies that can be paid and the pool of eligible C recipients? m * How will sustainability considerations influence the choice of action that will attract public funding? '- Deciding who will be eligible to receive services that attract public funding Both the policy rationale for public funding and budgetary constraints will influ- ence the designation of people eligible to receive subsidized services. Where public funding is justified by merit good or public good considerations, the pool of intended recipients may include most members of society.Where subsidies are aimed at making services more affordable, the pool is usually smaller. Precisely targeting the people to receive subsidies can be information intensive and costly. In some cases existing census or other data may provide reli- able information for means testing of households. Where such information is lacking, some pragmatism may be required, with targeting strategies taking into account the costs of generating reliable information and the potential for errors of inclusion (including an unintended recipient) and of exclusion (excluding an intended recipient). In some cases area of residence may be a reasonable proxy for household income.This criterion determines eligibility for some subsidies in 95 Colombia, for example. And many rural electrification schemes simply assume that those living in rural areas are relatively poor (see chapter 5). Subsidies can sometimes be designed so that recipients signal their eligi- bilitr. For example, if a subsidy is directed to a low-cost service option, more affluent citizens will be less likely to choose that option. Peru's rural telecom- munications scheme directs subsidies to pay phones, in the knowledge that more affluent people will prefer the convenience of their own household or co mobile phones. Similarly, subsidies directed to new household connections to F infrastructure services will exclude those already connected. Recipients might o also be required to take some positive action to qualify for subsidized services, 0 such as registering with local authorities. And they might be required to meet some additional condition, such as paying a share of the bill, as in the water Dn supply scheme in Chile (see chapter 2). > When the pool of potential recipients is large, some mechanism may be co needed for determining priorities. In many parts of Africa less than 10 per- I cent of the population has access to electricity. Where do you begin? One :,, practical approach is to allow service providers to address the pool of poten- > tial recipients progressively, on the basis of cost and recipients' willingness and rl ability to pay user fees to complement the available subsidies. The intended recipients-and the broader conununity-can also play other important roles in the design and implementation of output-based aid schemes (box 2). Questions to consider : When subsidies are justified by affordability concerns, how will intended recipients be identified? • When the pool of potential recipients is large, how will priorities be determined? * How will the perspectives of recipients and communities be incorporated into the design of the scheme? Deciding who will be eligible to provide services Output-based aid schemes need to mobilize competent service providers that respond to incentives for performance.What criteria should apply? For some services, the eligible providers may need to be limited to those with certain technical qualifications-for example, adequate training to provide specialized health services. The definition of qualifications needs to take into account policy concerns about the nature of the service to be delivered, trade- 96 F, I,' 2 Involving recipients and communities Traditional approaches to delivering public services have been dominated by the perspectives of public sector suppliers. Output-based aid creates opportunities for increasing the input and involvement of the intended recip- ients and the broader community in three ways: * Designing and evaluating schemes. Recipients and communities can help clarify the need for subsidies and ensure that service specifications take En into account local needs and preferences. They can also help evaluate the z operation of schemes, especially important in pilot projects. o) * Providing services. Small-scale entrepreneurs and community groups can 0 c in some cases play the role of service providers. H * Monitoring services. Where communities are not involved in providing ser- -M vices, they might play roles in monitoring and verifying the performance of > service providers. m Cl En offs between cost and quality, and the effect on the supply response. Requiring ° unnecessarily high qualifications can increase the cost of service provision and m reduce the number of potential providers. In some cases it may be necessary or > desirable to ofTer training to expand the pool of potential providers. C~ The ownership and governance structure of potential service providers- may be another important consideration in determining eligibility. Output- H based aid schemes rely on high-powered financial incentives for performance, which means that service providers need to be motivated by such incentives, and also have the flexibility to adapt delivery processes and experiment with innovative financing arrangements. Etfective contractual discipline and over- sight require the service provider to be at arm's length from the schemes administrators and from regulatory authorities. And effective competition- whether in the market for providing the services or in the bidding for monopoly franchises that attract public funding-requires a reasonably level playing field between potential providers. Including incumbent public enterprises among eligible providers usually creates difficulties on all these criteria. Moreover, any private financing mobilized by public enterprises generally involves at least an implicit government guaran- tee, reducing the potential benefits of output-based aid. Private entrepreneurs and firnms-local and foreign-more cleailv meet these tests. Nongovernmental organizations or community groups might also be potential suppliers. A range of other design issues also affect success in mobilizilig effective supplier interest (box 3). 97 Questions to consider * Is it necessary to impose technical qualifications for eligible service providers? If so, what is the appropriate level and form? Is it necessary to offer training? * Does the definition of eligibility take into account the ownership and gover- nance structures of potential service providers? * How will other features of the scheme influence the ability to mobilize a sup- 0 ply response? m co Co r C Choosing the market environment 0 Traditional approaches to delivering public services have relied on monopolistic supply arrangements. Output-based aid schemes allow more competitive >o approaches.The choice of approach has important implications for the incentives m > faced by service providers and the intensity of regulator,y oversight required. co Advances in technology and in economic thinking are expanding the a I scope for competitive delivery of many services traditionally reserved for m imonopolies. Competition has become the norm in telecommunications and Co a To be effective, output-based aid schemes need to attract competent ser- vice providers and, ideally, to mobilize private financing to complement public resources. These considerations need to be taken into account in designing all aspects of a scheme. Particularly for new and innovative schemes, early efforts may be required to identify potential service providers and to obtain their input on the design of the scheme. Knowing their views on the feasible level of risk sharing, as reflected in such matters as the performance standards and the level and structure of payments. will often be critical to mobilizing an adequate supply response. The scale of the scheme can also influence the level and source of supplier interest. Schemes requiring substantial investment and service commit- ments may be too large for local entrepreneurs in many developing coun- tries, while small schemes will be less likely to attract interest from major international firms. Finally, the credibility of the payment scheme may be critical to mobilizing private financing. Ensuring credibility requires thinking about the security provided by contracts, the choice of scheme administrator, and, in some cases, additional measures such as escrow accounts. 98 increasingly common in electricity and gas supply and in many modes of transportation. Competition is also being used to spur efficient delivery of a range of other public services, including education and health care.When fea- sible, competition between service providers improves incentives for effi- ciency, innovation, and consumer responsiveness while reducing the regulatory burden. Under competitive approaches, service providers are compensated on 0 evidence that they have served an eligible recipient. This evidence might CO take many forms, including vouchers that targeted recipients have z 0 4 Using competition to improve health services in Nicaragua Nicaragua has used a voucher scheme to support the delivery of health ser- >cI vices to female sex workers since 1995 (before then, government clinics ED provided services), Targeted participants receive vouchers (see sample below) entitling them to a specified set of diagnostic and counseling ser- vices from their choice of preidentified clinics, which include both private (for profit) firms and nongovernmental organizations. The clinics redeem the 01 vouchers for payment from the voucher agency. The voucher scheme is credited with expanding the use of the services, improving their quality, and a T reducing costs (Harper and others 2000). r How the voucher scheme works Voucheraagency Voucher Cash Voucher Nongovernmental payment Cash 0 \'Nnoennna n en her paymens t organizations / \ / \ | ~~~~~~Voucher Donor or government Sex workers Clinics ~-Voucher/ Con este BONO Ud recibfre: t rr ,Se C-Ilta Grm aed romg HGRATIS r 0 r ,Examenes de Laboratorlo GRATISI IatC ,,Papamcolaou GRATIS! r, ° ,M.t.r al.s Ed.catfvs iGRATIS, _ r Reconsu t GRATIS' 1 rc Med,cae,,tos iGRATIS' o INSTITUTO CENTROAMERICANo DE LA SALUD W So,rce Dtagram adapted from Harper and, others (2000), vouc&her from Instnu.to CentroAnTeencano de la Salud. 99 exchanged for services from their chosen provider.Vouchers are being used for a variety of public services, including health services in Nicaragua (box 4). Effective competition may require relaxing regulatory barriers to market entry and in some cases educating recipients or requiring service providers to publish information so that the recipients can exercise well-informed choice.4 k Competition between providers may not be feasible in very small mar- uf) kets or where service delivery depends on investments with characteristics of z a natural monopoly, which may be the case for some infrastructure services, a C for example. In other cases competition night be technically feasible but a deliberately constrained-for example, where a monopoly structure is chosen -t to help mobilize private financing or to facilitate cross-subsidization between dmD users or services.3 Restricting competition when it is technically feasible is a > costly, however, particularly when access to services is low. The tradeoffs .0 '/: between competitive and monopolistic approaches thus need to be carefully I evaluated in each case. u7 Where monopolistic service delivery is chosen, efforts are needed to > ensure adequate regulatory oversight and to involve intended recipients more closely in the design of the scheme.The scope of the monopoly also needs to be carefully defined. Should there be a single national monopoly or a num- ber of smaller, local monopolies? While a single national monopoly may offer potential economies of scale, this advantage needs to be weighed against the potential benefits of having several local monopoly franchises: a smaller financing burden on service providers, the ability to adapt approaches to local conditions and preferences, and the potential for using "yardstick" competi- tion between providers. In choosing betveen a single monopoly and several, the size of the market and the significance of scale economies will often be decisive.Where contracts for monopolistic service delivery are being awarded, some of the benefits of competition can be obtained by tapping competition for the nmarket (box 5). Questions to consider * Is competitive delivery of the service feasible? If so, what measures are needed to ensure that competition is effective? * If monopolistic delivery is chosen, what will be the scope of the monopoly and what measures will be taken to ensure effective regulatory oversight and opportunities for input by recipients? 100 D, ]>s 5 Forms of competition in public services Output-based aid schemes might tap one of three forms of competition: * Competition in the market-where service providers compete directly with one another to supply intended recipients, Where feasible, this form of com- petition usually offers the greatest advantages. * Competition for the market-where potential service providers bid for the right to supply a monopolistic market under a time-bound franchise. Well- m designed bidding processes can serve as a market test of the required level of funding. Rebidding franchises at the end of the contract period can help z to sharpen incentives for performance. Z * "Yardstick" competition-where comparative data on the performance of ° C: different service providers are used to simulate some features of competi- tion in the market. For example, the relative performance of suppliers in dif- ferent service areas might influence performance standards or be tied to the > reward system to strengthen incentives for performance. Experience with m this approach in infrastructure is potentially transferable to a wider range of D public services. C I m C Defining performance > The contract between the scheme administrator and the service provider must clearly define the services to be provided and the key standards of per- formance. The approach taken has important consequences for the costs of -D the scheme, the incentives faced by service providers, and the strategy for monitoring and verifying performance. Setting the scope of the service package The service agreement might cover a package of related services or focus on a single service. For example, health contracts might cover the full range of services available from primary care providers (as in Romania); a package of related services, such as immunization and prenatal and maternal care (as in Haiti); or a single service, such as training in oral rehydration therapy (as in Bangladesh). Education contracts might cover the delivery of education ser- vices or focus on the provision and maintenance of facilities used in deliver- ing education (as in early schemes in the United Kingdom). In the roads sector in Argentina initial contracts included only maintenance but were later extended to rehabilitation as well. The coverage of the service agreement will depend on the policy objec- tive, budgetary constraints, and potential economies of scale and scope in putting together an integrated service package. It will also depend on the 101 implications for attracting supplier interest, as there may be tradeoffs between the breadth or depth of the service package and the amount of performance risk that can be transferred to service providers. For example, schemes seek- ing to shift significant performance risk to service providers may need to start with a more narrowly defined service package, expanding the scope as sup- pliers and their creditors become more familiar with the arrangement. This was the approach taken in the roads sector in Argentina and the education cc sector in the United Kingdom, for example. Even when the service agreement covers a range of related services, it may C be possible to focus the subsidy payment on a subset of services or actions when C the service provider xvill have access to user fees.An infrastructure service agree- ment, for example, could cover a range of actions required in delivering ser- vices, but target subsidies to new household or community connections. o Questions to consider m . What services will be included in the service agreement? c, * How will the breadth of the service package affect the amount of perfor- > mance risk that can be transferred to service providers? s* When the agreement covers a package of services and the service o provider will have access to user fees, is it desirable to focus subsidy pay- ments on a subset of services or actions? GD Specifying performance The services covered by the service agreement can be defined in different ways, with implications for the complexity of the contract, the incentives and performance risk faced by service providers, the cost of subsidies, and the ease of monitoring and verifying performance. The definition of some services might involve only a few key parameters, such as inoculation wvith a specified vaccine through a clean syringe. For oth- ers, there might be a larger set of parameters. For many health services, for example, effective delivery might require diagnosis, testing, and counseling. Similarly, water supply services might involve consideration of the quality of xvater provided, the hours per day of availability, the pressure of delivery, respon- siveness to service disruptions, billing arrangements, and a host of other matters. The number and nature of the service characteristics that potentially require definition will vary according to the service as well as the approach taken to incentives in the contract. In structuring incentives, it is important to distinguish between measures that focus on inputs, outputs, and outcomes (box 6). 102 F.g: ' 6 Inputs, outputs, and outcomes Performance under output-based aid schemes may be defined by ref- erence to inputs, outputs, or outcomes. * Inputs are the resources consumed in producing and delivering a service- such as labor, technology, or physical materials. * Outputs are the immediate results of the service provider's activities-such as the number and quality of the health, education, or infrastructure services u m provided. * Outcomes are the ultimate effects of services on the community-such as z improvements in health or in educational attainment. Outcomes may be D influenced by factors other than the activities of the service provider, which ° distinguishes them from outputs. X c >~~~~~~~~ TD m Other CzF potent al influences m m Inputs S~~ervice provide B-. ~~~~~~~~~~~~~~~~~~C)31C Measures focusing on outcomes and outputs create opportunities and incentives for service providers to discover new and better ways of achieving the desired results. A focus on education outcomes, for example, allows ser- vice providers to experiment with a range of approaches that achieve better results than traditional ones. A focus on outputs, such as the delivery of elec- tricity services in rural areas, would give service providers opportunities and incentives to try new models of service delivery. Outcome-related measures are also attractive because they directly target the larger social objective. But outcomes are often difficult to measure with precision and are usually subject to influences beyond the control of the ser- vice provider. Success in achieving an intended outcome of eliminating a par- ticular disease, for example, might be influenced not only by the health services provided, but also by the actions of service recipients, the perfor- mance of providers of other services, and a range of other factors. And achievement of an intended outcome of improving educational attainment 103 might be influenced by such factors as differences between individuals and the support provided by families. In these cases making the service provider's compensation completely dependent on outcomes would be inappropriate, and the high level of per- formance risk involved might deter potential service providers or lead to demands for larger payments. Nevertheless, some schemes link at least a small part of the service provider's compensation to outcomes beyond the n provider's control to align its interests with those of the scheme adminis- trator. For example, contracts for providing and maintaining school facili- C ties in the United Kingdom link part of the payment to the impact on C education outcomes, including attendance levels and educational attain- CD A ment as measured by performance on national achievement rests (see chap- (D ter 9). C Measures focusing on outputs-such as the quantity and quality of ser- Cf) vices provided-are more clearly limited to matters within the service C T provider's control. Monitoring and verifying the quantity of services pro- m vided is relatively easy. But it can sometimes be difficult to assess the qual- > ity of those services without at least some reference to the inputs consumed. I For example, it is often hard to evaluate the quality of professional services without reference to the qualifications of the provider. Similarly, when ser- vice delivery requires the construction or rehabilitation of long-lived assets that are intended to be transferred to the state at the end of the contract period, the scheme administrator may not be indifferent about the con- struction materials used. Of course, heavy reliance on input-related mea- sures requires much more detailed contracts and limits opportunities and incentives for innovation. The tradeoffs between outcome-, output-, and input-related measures of performance thus need to be carefully evaluated in each case. Reducing the number of defined service parameters can simplify con- tracting and monitoring. But the performance measures need to be chosen with care. In Haiti's health care scheme, for example, the waiting time in health clinics was originally included as a proxy for quality. But this indica- tor was found to be unreliable in a setting where people often must travel long distances for services and might wait an entire day for test results rather than come back (see chapter 7). When schemes focus on a subset of mea- sures, they also need to ensure that this does not lead service providers to neglect other important matters or that it does not otherwise produce per- verse results. In a scheme intended to promote recycling of household waste 104 7 Establishing minimum quality standards Performance specifications need to incorporate judgments about the mini- mum level of service quality Quality has a cost, however, and budgetary constraints may require careful evaluation of the tradeoffs, particularly when the alternative for intended recipients might be no service at all. Schemes can sometimes be designed to allow recipients to choose a mix o m of price and quality that reflects their individual preferences and budgetary 's priorities. For example, a voucher scheme might allow recipients to choose z from a range of service packages-and to "top up" the minimum entitlement a by paying additional user fees. In other cases schemes may need to establish a uniform set of minimum C performance standards for a large number of recipients. When this is nec- essary, it is important to ensure that the performance specifications take into account the views of intended recipients. It might also be feasible to define > 0 different service packages for different service areas, to match local condi- tions and preferences. - m CZ) in the United Kingdom, for example, focusing on the amount of recyclable a material collected rather than the amount recycled resulted in carefully seg- regated vaste beinig incinerated (U.K. National Audit Office 2001). And when contracts specify multiple service parameters, it may be appropriate to signal the relative importance placed on each, as health care contracts in Haiti do (see chapter 7). Embedded in any set of performance specifications are judgments about minimum acceptable qualityThere are important tradeoffs betwveen quality and cost, however, and these need to be carefully evaluated in each case (box 7). Questions to consider * Will service requirements be defined in terms of outcomes, outputs, or inputs-or some combination of these? * Can the number of defined service parameters be reduced without risking the neglect of other important matters or otherwise leading to perverse results? * When service requirements are defined in terms of multiple parameters, how will performance against each be weighted? * Do minimum quality standards reflect appropriate tradeoffs between qual- ity and cost? Do they reflect the priorities of intended recipients? 105 Linking payment to performance How an output-based aid scheme links payment to service delivery influences the cost of the scheme, the incentives and risks faced by providers, and the supply response mobilized, including private financing. Key questions relate to the form and amount of payment and to its structure. Determining the form and amount of payment Cash is the most flexible, most transparent, and, usually, most appropriate form 6 of payment for service providers in output-based aid schemes. Noncash forms 0 of compensation-such as tax credits or the conferral of special rights or privileges-are less flexible and transparent and can complicate the calibration of payment to results.6 oo In some cases governments might borrow from international financial , institutions to finance the provision of subsidies, as the government of Guinea o did for water consumption subsidies (see chapter 3). But using access to con- I cessional finance as the sole form of "payment" to service providers under c output-based aid schemes can be problematic. If the financing is made avail- > able only after the service has been delivered, the critical need for financing C I will usually have passed. If the financing is made available up front, incentives for performance are dulled and opportunities for mobilizing private financ- -A ing are forgone. The amount of payment required depends on the cost of providing the service to the specified standard, less any user fees from recipients. But the actual cost of providing the service is often unknown when a scheme is being designed. A first approximation could draw on benchmarks from incumbent suppliers. But by providing strong incentives for efficiency, well-designed output-based aid schemes lower costs. A recent survey of experience in con- tracting out services in OECD countries found cost savings averaging around 8-14 percent over a range of services, with savings of 50 percent or more in some cases (Hodge 1999).The actual savings depend on many factors, includ- ing the relative efficiency of the incumbent provider, the opportunities and incentives for innovation provided by the contract, the premium required by the service provider to compensate for payment or other risks, and the process used to award contracts. The process for awarding contracts can be used to reduce costs and pro- vide a market test of the amount of payment required. One strategy for doing so is to award contracts competitively to qualified service providers on the basis of the least payment required to provide the service.7 In the rural 106 telecommunications scheme in Peru this strategy resulted in bids 41 percent lower than official estimates and 74 percent lower than an earlier offer from the incumbent telecommunications operator (see chapter 1). Questions to consider * Is there any basis for considering noncash forms of payment? * What benchmarks will be used to estimate the likely costs of supplying the service to the required standard? m * Will contracts be awarded through a process that helps in determining (and a minimizing) the payment required? z a Structuring the payment CD Output-based aid schemes that link payments completely to performance, and D withhold payment until satisfactory delivery of services has been verified, cre- D ate high-powered incentives for performance. Incentives can also be fine-tuned C Co) C by deducting payments for subpar performance and including bonus payments ° for service that exceeds minimum requirements.Where multiple dimensions of m performance are specified, payments can also be calibrated to reflect the relative > wveight attached to each, as in the health service contracts in Haiti. In some cases it may be necessary to include some payment that is not dependent on performance, such as when the service provider is required to H bear significant performance risks. This might also be necessary as a transi- tional measure where suppliers have traditionally been compensated on the basis of expenditures rather than outputs. In the health scheme in Haiti, for example, initially only 10 percent of service providers' compensation depended on performance. As service providers become more familiar with the system, it should be possible to reduce the share of payments unrelated to performance. Even in schemes where payment depends fully on performance, it might be appropriate to share some risks between the service provider, the scheme, and even the service recipients.When the delivery of a service depends heav- ily on an input whose price tends to fluctuate, for example, a mechanism for adjusting payments or permitted user fees might be needed to reduce the risk faced by the service provider, particularly when longer-term contracts are involved. Payment might still depend fully on delivery, but changes in the cost of the key input could be reflected in the size of the payments or user fees. Care needs to be taken, however, to avoid reducing the service provider's incentives to negotiate hard with its suppliers. 107 A credible scheme assuring payment on delivery should help service providers mobilize the financing they need to deliver the services. But when significant investments are required and local capital markets are underdevel- oped, some part of the payment may need to be made up front to ensure suf- ficient supplier interest at reasonable prices. For example, Peru's rural telecommunications scheme paid 35 percent of the subsidy at the start of the project and 25 percent once the facilities were installed (see chapter 1). Argentina's road rehabilitation and maintenance contracts included advance z payments of 5-10 percent, followed by 15-25 percent at the end of the first G) six months, when specified activities had been completed, and 25 percent at 0 c the end of the first year, when rehabilitation works had been completed (see Ti chapter 4).YWhen up-front payments are made, it may be necessary to obtain Dmc performance bonds or similar guarantees from service providers.9 > The balance required between up-front and post-performance payment needs to be tested in each case, taking into account the investment required 7 and the feasible level of private financing. This reinforces the importance of consulting closely with potential service providers when designing the > scheme.The appropriate balance may change over time as suppliers and their creditors develop greater famiharity with and confidence in the scheme. Questions to consider * How will payments be tied to particular dimensions of performance, and will penalty and bonus payments be used to fine-tune incentives? * Is it necessary to include payments not dependent on performance? If so, what share of the total payment will they be? * Is it appropriate to share some risks with service providers, such as changes in the prices of essential inputs? * Is it necessary to include up-front payments? If so, what share of the total payment will they be, and how will the risk of subsequent nonperformance be addressed? Shaping other aspects of the contract Clear definition of the services to be provided and the payment arrangements is clearly central to the service agreement. But a range of other issues require attention, including the legal nature of the contract, its duration, and the mechanisms it provides for adjusting its terms during its life and for resolving potential disputes. Performance contracting between government entities often relies on "quasi-contracts" that are not legally binding. 10 In contrast, the enforceability of 108 contracts in output-based aid schemes is essential to eliciting a supply response, providing strong incentives for efficiency, and mobilizing private financing. Indeed, the existence of a binding contract may be the basis on which banks or other financial intermediaries extend credit to service providers, as is the case with similar agreements in private infrastructure projects. The appropriate duration of the contract depends in large part on the investments the service provider needs to make to deliver the service. The term of the contract usually needs to be long enough to allow the investment costs m to be amortized over its life. Argentina's road maintenance and rehabilitation z concessions are for 5 years, Guinea's xvater lease was for 1 0 years, and Peru's rural C telecommunications scheme involves (nonexclusive) licenses for 20 years.When the service provider will have access to user fees, it may be possible to limit the m commitment for subsidy payments to a period shorter than the service agree-> ment. Subsidy payments under Guinea's water lease were limited to the first six C years, and in Peru's rural telecommunications scheme to the first five, after which tinie the service providers were expected to rely on user fees alone. Longer-term contracts need to anticipate possible developments during i the life of the arrangement. Because anticipating all contingencies is impossi- > ble, it is usually prudent to include mechanisms for adjusting contractual C terms to unforeseen developments. Particularly when significant investments are required that depend heavily on access to subsidy payments, these mech- anisms need to take into account the risk investors may face from oppor- tunistic government action once the investment has been made.' In most cases disputes over the interpretation or application of a service contract will be subject to the jurisdiction of the courts in the host country. But it is usually important to establish mechanisms for the amicable settle- ment of disputes and-particularly when foreign firms are required to make significant investments-to consider international commercial arbitration. Questions to consider * What measures need to be taken to ensure that the contract is legally enforceable? * What is the optimal duration of the service contract? Does the commitment to provide public funding need to be of the same duration? * With longer-term contracts, have measures been put in place to facilitate adjustments of the contract to unforeseen developments? * Have adequate measures been put in place to deal with possible disputes? 109 Structuring the administration of the scheme Output-based aid schemes need to be administered competently and at rea- sonable cost. This requires consideration of the scope of the scheme, the nature of the scheme administrator, and the assignment of responsibilities. Defining the scope of the scheme The scope of output-based aid schemes has three main dimensions: co * Sectoral. Should the scheme cover a single sector, such as rural electrification, Z or a broader range of related sectors, such as rural infrastructure services? a Geographic. Should the scheme cover a single subnational jurisdiction, or C the country as a whole? c * Funding. Should the scheme rely on funding from only one source, or allow the pooling of resources from several sources? > Schemes with a broad sectoral and geographic scope offer potential economies of scale that can reduce administrative costs. For example, the I means testing of households in Chile is cost-effective only because the system ,, is used to determine the eligibility for subsidies for a wide range of services. > Broader approaches also allow the pooling of expertise, which may be scarce I in some countries, and can be helpful in sharing lessons across a range of pro- jects. Schemes that cover two or more similar sectors (such as rural water and electricity) may also alloxv the bundling of services, offering potential cost sav- ings in service delivery (see Sommer 2001). Schemes that are open to multiple funding sources can reduce adminis- trative costs and help to ensure that funding is sustainable. They can also aid the coordination of donor activities, which can be important in ensuring that the disciplines of output-based aid approaches are not undermined. The fea- sibility of this approach depends on the extent to which potential sources of funding share common objectives. Some pilot schemes might initially cover a relatively narrow set of sectors and geographic areas and rely on a single funding source. In these cases it may be possible to design the scheme in a way that facilitates broadening over time. Questions to consider * Are the scheme's sectoral and geographic scope appropriate? * Does the scheme allow the pooling of multiple funding sources? * If a pilot scheme is established initially with a relatively narrow sectoral, geographic, or funding scope, does the scheme's design facilitate broad- ening over time? 110 Selecting the scheme administrator and assigning responsibilities All schemes require an entity responsible for their overall management. This role might be performed by existing or new public agencies or contracted out to private firms or to nongovernmental organizations. The credibility and competence of the scheme administrator will be critical to winning the con- fidence of prospective service providers and thus helping to mobilize a sup- ply response and private financing. 0 The administration of output-based aid schemes can involve a range of distinct activities-mobilizing resources, designing contracts, awarding con- z z 0 r R- c 1 Administrative structures-two options - CD Option I CO Adrwlnlsorwith > paymet anW sevi CD t )m Payment Monitoring > I Service 0 = Oversight X Oversight t ~~~~~Authorization Payment Monitoring i i ~~~~~Service s , ! ! | | | | | l ~~~~~~~~~~~~~~~~~~~~~C E g , i | i~~~~~~~~~~~~~~~~~~~~~~~~~~c E E | E~~~~~~~~~~~~~a 1 1 tracts (including conducting bidding), distributing vouchers or similar instru- ments, monitoring and verifying service delivery, and paying service providers. There may also be ancillary activities related to marketing the scheme to prospective suppliers and providing capacity building services. These roles might be performed by the scheme administrator or allocated among several entities (figure 1). C3 The critical responsibility for monitoring and verifving service delivery 0 C umight be retained by the scheme administrator or contracted out to other 6 6 entities. Potential candidates include existing regulatory bodies (as in Peru's C rural telecommunications scheme), independent institutes (as in Haiti's health care scheme), private consulting firms, and local communities. The appropri- ,T ate choice will depend on such factors as the expertise required, the need to pmD ensure an arm's-length relationship with service providers, the need to ensure C > high standards of probity, the administrative costs (including potential C) economies in bundling responsibilities), and the benefits from involving com- I munities in implementing the scheme. m m Cfo I Questions to consider m . Is there an existing public agency with the competence and credibility to administer the scheme? If not, will a new public entity be created, or will administration be contracted out? . Will the scheme administrator undertake all responsibilities under the scheme, or will some of these be contracted out to other entities? Leveraging experience Output-based aid is a relatively novel approach to financing and delivering public services. But those designing and implementing output-based aid schemes can draw on a large body of relevant experience, particularly in pri- vate infrastructure arrangements. Moreover, many schemes are begun as pilots, to test and refine approaches. Building on the lessons from these pilots, output-based aid schemes usually expand progressively within countries, as with roads in Argentina, education in the United Kingdom, and health ser- vices in Haiti. There are also important opportunities to leverage lessons of experience across countries and sectors. The potential for leveraging lessons across countries for a particular service is clear.The design of a scheme for supplying water or education services in one country can provide important insights for schemes covering the same services 112 B,, 8 Measuring progress Output-based aid strategies can be implemented in a variety of ways. Progress in applying the key principles can be measured at the level of indi- vidual schemes against the following criteria: * Targeting of development outcomes-reflected in the approach to designating eligible recipients and in the role of outcome-related performance measures. m * Accountability for results-reflected in the extent to which payment co depends on achievement of the specified results. z * Incentives for efficiency-reflected in the form and extent of competition 0) and of contract-based incentives. ° * Opportunities for innovation-reflected in the balance between input-, - output-, and outcome-related performance measures. * Mobilization of private financing-reflected in the amount of private financ- > ing leveraged by the public resources. Cost-effectiveness is also an important consideration. Pilot schemes usually U) have relatively high design and implementation costs, reflecting the addi- j tional costs associated with pioneering new approaches and the small num- 4: m ber of service recipients over which costs are nominally allocated. Follow-on or expanded schemes usually improve cost-effectiveness because they can draw on the lessons of pilots and allocate costs over a larger pool of recipi- m ents. Assessments of cost-effectiveness also need to take account of bene- p fits that can spill over to services outside the scope of the service agreement. International experience shows, for example, that contracting out service delivery usually inspires improved performance by public sector providers of the same or similar services (see Hodge 1999). in other countries. Transferring lessons is particularly important for such mat- ters as specifving performance standards, but can extend to many other aspects of design and administration. While care needs to be taken to avoid cookie- cutter approaches, efforts to distill and disseminate lessons of experience Awill reduce the costs of designing schemes and avoid needless repetition of mistakes. There are also opportunities to leverage experience-and in some cases resources-across sectors. The unique characteristics of each public service cannot be ignored. But many common issues arise in designing output-based aid schemes, ranging from the design of contracts to the structure of admin- istrative arrangements. Within a single country there may be advantages in drawing on common approaches to targeting intended recipients, engaging with communities, mobilizing local financing, and dealing with contract design issues-and even in using common entities for administering at least 113 parts of different schemes. Insights and lessons that transcend sectors can also be shared across countries. Taking full advantage of this opportunity may require a change in mind-set for professionals unaccustomed to looking beyond traditional sector boundaries. Ultimately, output-based aid strategies and particular schemes will need to be evaluated by the same standard they seek to apply to providers of pub- lic services-on the basis of results (box 8). m z z c 0 C) H -U m D r.n m m C I on CA 11 114 CD m Notes tion, suggesting that public handed regulation is Z Warrick Smith is manager education could be an required to make the 0 of the World Banks Prvate important part of the policy arrangement work: as a Provision of Public Services response. result, this approach a not 7i m Group. This checklist bene- 3. While pr vate financ- feasib e or effective in all D fited greatly from many con- ng typ ca ly has a higher environments Indeed, D versations with Pene ope J. nomina cost, public financ- cross-subsidy systems C) co Brook as well as comments ing usually ref ects implicit ostensibly designed to help I on drafts from various mem- contingent claims against all the poor may act against m bers of the Private Provision taxpayers (see Klein 1997). their interests in many devel- > of Public Services Group All 4. For a review of ssues oping countries by both I 0) shortcomings remain the relating to voucher schemes destroying incentives to offer responsibility of the author. for pub ic serv ces see them services and restrict- 1 For a useful review of OECD (1999b) and Steurele ing their access to other ser- design issues associated and others (2000). vice options with infrastructure conces- 5 Cross-subsidy 6. Under monopolistic sions see Kerf and others approaches involve charg- schemes, exclusive access (1998). ing higher prices for some to user fees from more afflu- 2 There is often room for serv ces or categories of ent users in the service area debate about what serv ces users to allow lower prices may itself constitute a form have the characteristics of to be charged for others. of implicit subsidy. But merit goods and the extent They require monopoly pro- schemes that do not nvolve to which any social benefits vision to preverit those pay- an explicit payment are not exceed the benefits cap- ing the higher prices from considered to be output- tured by individual users. In defecting to other options. based aid schemes here, some cases underconsump- Because service providers even when retention of the tion may result from inade- have no natura incentive to monopoiy or the level of per- quate understanding by . prov de the underpriced ser- mitted tariffs is conditional users of the benef ts they vices (they lose money on on meet ng certa n perfor- can derive from consump- each transaction), heavy- mance standards 115 m 2 7. A variation on this contractual obligations (see Kerf, Miche, R. David approach is to specify the chapter 1). Gray, Timothy Irwin, Celine subsidy and to award fran- 10. For a review of Levesque, and Robert R. TA chises on the basis of the issues associated with per- Taylor. 1998. Concessions m c> user fee component, other formance contracting for Infrastructure. A Guide m >iJ obligations assumed by the between government entities to Their Design and Award. service provider under the see OECD (1999a). World Bank Technical co C-) I scheme, or some combina- 11. For a review of how Paper 399. Washington, m m tion of the two. private infrastructure D. C. C( > 8. The Argentine case schemes address issues of Kein, Michael. 1997. a: also reinforces the need for po it cai and regulatory r sk "The Risk Premium for m carefu market testing of the relevant to output-based a d Eva uating Public Projects." Co scheme: an initial bidding schemes see Smith (1998). Oxford Review of Economic round offering a smaller Policy 13(4): 29-42. share of up-front payments References OECD (Organisation for had to be canceled when Harper, Malco m, Jorge Econom c Co-operation and bids exceeded official esti- Arroyo, Tushar Deve opment). 1999a. mates by near y 100 percent Bhattacharya, and Tom Performance Contracting: because of high f nancing Bulman. 2000 Public Lessons from Performance costs. Services through Private Contracting Case Studies 9. In Peru firms Enterprise: Micro- and a Framework for Public awarded licenses for rural Privatisation for Improved Sector Performance te ecommunications ser- Detivery London: Contracting. PUMA/PAC vices were requ red to pro- Intermediate Technology (99)2. OECD Programme on vide three financial Publications. Public Management and guarantees: a guarantee Hodge, Graeme. 1999. Governance (PUMA). Paris. ensuring the seriousness of Privatization. An . 1999b. Voucher their offer, an installation Intemational Review of Programs and Their Role in guarantee, and a guaran- Performance. Boulder, Colo.: Distributing Public Services. tee aga nst default on their Westview Press. Paris. 116 m cr Sm th, Warrick. 1998. Infrastructure for z 'Covering Political and Development. New York: 0 Regulatory Risk in Prvate Oxford University Press. C H Infrastructure " In Timothy r Irwin, M chae Klein, >c GuiNlermo Perry, and Mateen 0 Thobani, eds , Dealing with U Public Risk in Private I Infrastructure. Washington, m D C: World Bank. > Sommer, Dirk. 2001 I "Multi-Uti it es: Policy" U) Viewpo nt 228. World Bank, Private Sector and Infrastructure Network, Washington, D.C. Steurele, C. Eugene, Van Doorn Ooms, George Peterson, and Robert D. Reischauer, eds. 2000. Vouchers and the Provision of Public Services. Washington, D.C.. Brookings Institution Press U K. National Audit Office. 2001 Measuring the Performance of Government Departments. London. World Bank. 1994. World Development Report 1994. 11 7 Suggested readings Baker, Bill, and Sophie Tremolet. 2000. "Utility Reform: Regulating Quality Standards to Improve Access for the Poor." Viewpoint 219. World Bank, Private Sector and Infrastructure Network,Washington, D.C. Determining the quality standards that should apply to service delivery is a critical element in identifying parameters anid indicators of perfornmance for output-based schemes. This paper, focusing otu infrastructure. notes the importance of thinking carefully about the inmplications of quality standards for costs, particularly wvhere the objective is to expand services for the poor. Domberger, Simon. 1998. The Contracting Oigatnization: A Strategic Guide to Oitsonrcinig. NewYork: Oxford University Press. This seminal book on contracting for services explains wheni it niiakes sense to contract out and what the key deterinilianats of success are. Chapter 9 focuses on the issues that arise in public sector contracting, reviewing the prospects and limitatioins of contracting out service delivery and noting crit- ical designi challenges. Foster, Vivien. 2000. "Measuring the Impact of Energy Reform: Practical Options." Viewpoint 210. World Bank, Private Sector and Infrastructure Network,Washington, D.C. To be effective, output-based elemenits of sectoral reform require good indi- cators of impact. Few reforms in infrastructure have included systematic efforts to assess the impact of reform on access to services for the poor. This paper suggests approaches to identifying meaningful, measurable indicators of the impact of energy reforms. Glennerster, Rachel, and Michael Kremer. 2000. "A Better Way to Spur Medical Research and Development.' Regulation 23(2): 34-39. This article looks at the possibilities for spurring research on effective vaccines against communuicable diseases-such as HIV, malaria, and tuberculosis-by complemiienting traditional funding approaches (direct funding and patent revenues) with purchase precommitments for future vaccines. Harper, Malcolm, Jorge Arroyo, Tushar Bhattacharya, and Tom Bulman. 2000. Public Services through Private Enterprise: M7licro-Privatisation Jor Improved Delivery. London: Intermediate Technology Publications. This book looks at the potential and options in developing countries for coIn- o tracting out the provision of infrastructure and social services to small-scale cr private and community enterprises. It includes 24 case studies from Asia, m o Africa, Europe, Latin America, and the United States. z -lHodge, Graeme. 1999. Privatization: Arn International Review of Peformance. "' Boulder, Colo.: Westview Press. Chapters 6 and 7 of this book review privatization reforms, focusing on the chal- lenges of performance-based contracts with private entities. The chapters iden- rift indicators against which the impact of these contracts mzight be measured and analyze the identifiable impacts of the contracting out that has been done. Hoxby, Caroline. [http://post.econornics.harvard.edu/faculty/hoxby/papers. html]. The Web site of Caroline Hoxby. an associate professor of economiiics at Harvard University, provides a series of papers on the use of vouchers for edu- cation services and their impact on school pertormance. Illinois Department of Children and Family Services. 2000. Performance Contracting in Illinois' Child Welfare System. Chicago. In 1997 the state of Illinois introduced performance contracting to increase the success with which children in foster care were placed in permanent homes.The program has transformed the state's child welfare system from one of the poorest performing in the United States to a national leader in per- nianent placements. Jadresic, Alejandro. 2000. "Promoting Private Investment in Rural Electri- fication." Viewpoint 214. World Bank, Private Sector and Infrastructure Network,Washington, D.C. One of the earliest and most thorough energy' reformers, Chile has also been one of the more innovative in restructuring its subsidy schemes. Its rural electrification program includes subsidies designed to be consistent with the 120 broad principles of energy reform-decentralizing decisions to the regional and community level. introducing competition (between technologies as wvell as suppliers), and requiring that all partners in the process (users and pri- vate companies as well as the state) contribute to the financing of expansion projects. Klein, Michael. 1998. "Bidding for Concessions: The Impact of Contract Design." Viexvpoint 158. World Bank, Finance, Private Sector, and Infrastructure Network,Washington, D.C. , m Infrastructure concession contracts set out the performance obligations and rights of concessionaires and the incentives and risks under which they oper- n ate, including pricing arrangements. This paper exanmines the design of infra- structure concession contracts in relation to bidding processes. ( Klein, Michael. 1998. "Rebidding for Concessions." Viewpoint 161. World Bank. Finance, Private Sector, and Infrastructure Network,Washington, D.C. This paper looks at the issues that come into play in a decision about whether to rebid a concession-issues relating to the incentives for investmient and maintenance that the incumbent faces. The paper also examines the case for rebididing. Levesque, Celine. Michel Kerf, David Gary, Timothy Irwin, and Robert Taylor. 1998. Concessiolns for Infrastnrcture: A Guide to Their Design and Avard. World Bank Technical Paper 399. Washington, D.C. Concessions give private companies the right to provide a particular infra- structure service. This book-length paper sheds light on the most importaiit and difficult issues involved in designing. axvarding. implenienting. and mon- itoring concessions in the inifrastructure sector. OECD (Organisation for Economic Co-operation and Development). 1999. Performance Contracting: Lessons from Perfornnance Contractig Case Studies and a Framewiorkfor Public Sector Pefonritance Contractinig. PUMA/PAC (99)2. OECD Programnie on Public Management and Governance (PUMA). Paris. This tvo-part report on public sector pertoriiiance contracting combines case studies from iiiie OECD countries (Australia, Belgium, Canada, Denmark, Finland, France. New Zealand, Norway, aiid Spain) with an analytical frame- work for determininig the relevance and design of performance contracts in different contexts. 121 Scott, Graham. 2001. Public M11anagement in New Zealand: Lessons and Challenges. Wellington: New Zealand Business Roundtable. New Zealand undertook a series of significant public sector reforms starting in 1 987. This book surveys these reformis and their impact. Chapter 7 focuses on the issues in setting and nmonitorinig performance specifications for core public services. o Vinson, Elisa. 1999. "Performance Contracting in Six State Human Services 0 :n Agencies:" Urban Institute,Washington, D.C. This paper looks at the activities of state agencies in Florida, Illinois, Maine. ,, Minnesota, North Carolina, and Oklahoma that have used performance coIn- tracting to improve adoption,job placement, and related services and reviews early evidence on1 the impact of alternative contracting arrangemelnts. Wellenius, Bjorn. 1997. "Extending Telecommunications Service to Rural Areas: The Chilean Experience-Awarding Subsidies through Competitive Bidding." Viewpoint 105. World Bank, Private Sector and Infrastructure Network, Washington, D.C. After Chile privatized its telecommtinications industry in the late 1 980s, ser- vice provision expanded rapidly-but around 10 percent of Chileans still lived in rural localities without even a public telephone. This paper describes the least-subsidy bidding scheme used to tiobilize the private sector to extend public telephone service to these areas. 122 Private Sector Advisory Services The World Bank 1818 HStreet NW Washington, DC 20433, USA Telephone: 202 473 3256 or 202 458 1 111 Web site: r:'.woridbart .o72g Email:osasousr@vroridheok.o orrru©vqoritl3sn.,K