THE WORLD BANK GROUP. Output Based Aid. A Compilation of Lessons Learned and Best Practice Guidance. Final Draft. September 2009

Transcription

1 THE WORLD BANK GROUP Output Based Aid A Compilation of Lessons Learned and Best Practice Guidance Final Draft September 2009 GPOBA and IDA-IFC Secretariat

2 Acknowledgments This paper has been prepared by a team led by Yogita Mumssen (FEU), comprising Lars Johannes and Geeta Kumar (GPOBA) and Vyjayanti Desai (IDA-IFC Secretariat), with the support of Inga Murariu, Daniel Coila, and Mark Njore (GPOBA), under the supervision of GPOBA Program Manager Patricia Veevers-Carter and IDA-IFC Secretariat Director Nigel Twose. GPOBA and the IDA/IFC Secretariat would like to extend special thanks to the peer reviewers: Amie Batson (MDW), Andreas Schliessler (ECSSD), Dana Rysankova (AFTEG), Juan-Navas Sabater (CITPO), Logan Brenzel (HDNHE), and Luiz Claudio Martins Tavares (AFTU1). Additional comments and contributions were provided by the following OBA practitioners: Carmen Nonay, Catherine Russell, and Esther Loening (GPOBA), Cledan Mandri-Perrot, Iain Menzies and Mustafa Hussain (FEU), Xavier Chauvot de Beauchene (MNSSD), Dirk Sommer (CIADR), and Martin Schmidt (KfW). Special thanks to the comments and guidance provided by the OBA Advisory Committee that met in February 2009, comprising: Alejandro Jadresic and Irving Kuczynski (GPOBA Independent Panel of Experts), Adriana Aguinaga (CINPW), Amie Batson (MDW), Doyle Gallegos (CITPO), Gaiv Tata (CFP), Hartwig Schafer (SDNVP), Magda Lovei (EASOP), Marc Juhel (ETWTR), Neil Gregory (FPDVP), Penelope Brook (GIADR), Tjaarda Storm Van Leeuwen (AFTEG), and, Vijay Jagannathan (EASIN). 2

3 Acronyms and Abbreviations ACGF AFR AusAID BPHS CBO CCT CGDEV CITPO COD CREMA DFID DPL DPO DRC EAP ECA ECLAC EMW ESCO ESMAP EU FDT FDT FITEL FONDETEL FPD FSSAP GEF GPOBA HH IADB IBRD ICB ICR ICT IDA IDB IDCOL IDTR IFC IMF INDH JHU KfW LCR MCC MDG MENA MFI Africa Catalytic Grown Fund Sub-Saharan Africa Region Australian Agency for International Development Basic Package of Health Services Community-Based Organization Conditional Cash Transfers Center for Global Development Policy Division (IBRD Telecommunications and Informatics) Cash-on-delivery Aid Contrato de Recuperación y Mantenimiento UK Department for International Development Development Policy Lending Development Policy Operation Democratic Republic of the Congo East Asia and Pacific Region Europe and Central Asia Region United Nations Economic Commission for Latin America and the Caribbean East-Meets West Energy Service Company Energy Sector Management Assistance Program European Union Fund for the Telecommunications Development Telecommunications Development Fund Peru s Fund for Telecommunications Investment Fund for the Telephony Development Finance and Private Sector Development Bangladesh Female Secondary School Assistance Project Global Environment Facility Global Partnership on Output-Based Aid Household Inter-American Development Bank International Bank for Reconstruction and Development International Competitive Bidding Implementation Completion Report Information and Communication Technology International Development Association Islamic Development Bank Infrastructure Development Company Limited Decentralized Infrastructure for Rural Transformation International Finance Corporation International Monetary Fund National Initiative for Human Development Johns Hopkins University Kreditanstalt für Wiederaufbau Latin America and Caribbean Region Millennium Development Corporation Millennium Development Goal Middle East and North Africa Region Microfinance Institution 3

4 MIGA MMR MPOH MSC MTR MWC NGO NHAI NHDP NHSPT NPV NWDP O&M OBA OECD OPCS PBS PER PERMER PHSPT PMMR PNFP PPA PPER PPIAF PPP PRSC PSD PV QEII RBF REB REF SANRAL SAR SDN SHS SIAS SIDA SMME SOE SPUG STD SWAp UAF UNDP USAID USF WASIS WBG WSP Multilateral Investment Guarantee Agency Manila Metropolitan Region Ministry of Public Health Medium-Term Service Contract Mid-Term Review Manila Water Company Non-governmental organization National Highways Authority of India National Highways Development Project National Health Services Purchasing Team Net Present Value National Water Development Project Operating and Maintaining Output Based Aid Organization for Economic Cooperation and Development Operations Policy and Country Services Rural Electric Cooperatives Guatemala Rural Electrification Scheme Renewable Energy for Rural Markets Project (Argentina) Provincial Health Services Purchasing Team Performance-Based Maintenance and Management of Roads Private Not-For-Profit Performance-Based Partnership Agreements Senegal Rural Electrification Priority Programs Public Private Infrastructure Advisory Facility Public Private Partnership Poverty Reduction Support Credit Private Sector Development Photo-Voltaic Queen Elizabeth II Hospital Results-Based Financing Rural Electrification Board Rural Electrification Funds South African National Road Agency South Asia Region Sustainable Development Network Solar Home Systems Guatemala s Integrated Healthcare System Swedish International Development Cooperation Agency Small, Medium, and Micro Enterprises State Owned Enterprises Small Power Utility Group Sexually Transmitted Disease Sector Wide Approach Universal Access Funds United Nations Development Program United States Agency for International Development Universal Service Funds Water Services and Institutional Support Project World Bank Group Water and Sanitation Program 4

5 Table of Contents Executive Summary Introduction Background Objective and Audience Execution of Review/Methodology Structure of Report Context Definitions Traditional vs. OBA approaches Contracting out service delivery: the spectrum Applications of the OBA approach OBA as part of the Results-Based Agenda Snapshot of OBA in the WBG and Beyond Universe of OBA under review Funding OBA Sector focus Lessons Learned: Cross-cutting Challenges and Best Practice Transparency: Explicit targeting of subsidies for the poor Accountability: Shifting performance risk to providers Private sector capital and expertise Innovation and efficiency Sustainability, Tariffs and the Enabling Environment Monitoring of results Key Considerations and Next Steps External challenges and possible responses by the WBG Internal challenges and possible responses by WBG Annex 1: Table of OBA projects in the WBG (as of April 30, 2009) Annex II: Sector Specific Lessons A. ICT B. Roads C. Energy D. Water E. Health F. Education Annex III: Comparison of OBA portfolio to total WBG portfolio FY

6 Figures, Tables, and Boxes Figures Figure 1: Volume of OBA Subsidy by Sector and Region in the WBG... 9 Figure 2: Number of Projects by Project Status... 9 Figure 3: Contrast of a Traditional Input-based Approach to an Output-based Approach Figure 4: Contracting Spectrum Figure 5: Volume of OBA Subsidy by Sector and Region in the WBG Figure 6: Comparison of WBG OBA Portfolio to Total WBG Portfolio Figure 7: Comparison Portfolio vs. Number of Projects in IDA, IBRD and Blend Countries Figure 8: Distribution of Subsidy for All GPOBA Projects (Including Design Stage) Figure 9: Distribution of OBA Portfolio by Sector and Region Figure 10: Number of Projects by Project Status Figure 11: Comparison of Performance of OBA and Traditional Projects Figure 12: Distribution of OBA ICT Projects Figure 13: Global Application of Performance-Based Contracting for Roads Figure 14: Distribution of WBG OBA Projects in Transport Figure 15: Distribution of WBG OBA Energy Projects Figure 16: Distribution of WBG OBA Schemes in Water and Sanitation Figure 17: The Universe of Results-Based Financing Figure 18: Distribution of WBG OBA Projects in Health Tables Table 1: Benchmarks/Criteria and Cross-cutting Lessons from OBA Portfolio Table 2: Distribution of OBA Projects Table 3: Distribution of OBA Portfolio by Sector and Region Table 4: Cost and Effectiveness of Targeting Mechanisms Table 5: Universe of Output-Based Aid in ICT Table 6: Defining Performance in the Provision of ICT Services Table 7: Performance Indicators for Transport Projects Table 8: Health Service Providers in Contracts to Provide Health Services Table 9: Parties Responsible for M&E in Contracts to Provide Health Services Boxes Box 1: Applications of OBA Subsidy Design Mechanisms Box 2: Nepal Biogas Support Program Geographic Plus Self-selection Targeting Box 3: CREMA Phases I and II Box 4: Limitations to Shifting Risks to Service Providers Box 5: Colombia Natural Gas Distribution for Low Income Families in the Caribbean Coast OBA and the Private Sector Box 6: Bank Procurement in OBA Projects Box 7: Fund for Telephony Development in Chile Box 8: Development of Telephony Fund in Guatemala Box 9: Argentina: Pilot Project for Performance-Based Maintenance: Roads Maintenance and Sector Rehabilitation Project Box 10: Successful Performance Based Rehabilitation and Maintenance Contracts in Chad Box 11: Annuity Concessions in India Box 12: Bangladesh -- Rural Electrification and Renewable Energy Development Project ( ). 89 Box 13: The ESCO Model: Argentina PERMER Concession Box 14: Senegal Rural Electrification Priority Programs (PPER)

7 Box 15: Mozambique Water: OBA and IDA Box 16: Manila Water Supply Project Box 17: Phasing in Payments Due to Access-to-finance Constraints in Uganda s Water Sector Box 18: Kenya Microfinance for Small Water Schemes Box 19: Expansion of Water Services in Low-income Areas of Jakarta Box 20: Morocco Urban Water and Sanitation Project Box 21: New Hospital Public-Private Partnership in Lesotho Box 22: Cash-on-delivery Aid (COD) Box 23: Bangladesh Female Secondary School Assistance Project Box 24: Conditional Cash Transfers in Education and OBA Box 25: The Impact of PSP in the Concession Schools Program in Bogota

8 Executive Summary Objectives and Execution: The objective of this review is to provide a more definitive and practical understanding of lessons and best practice related to output-based aid ( OBA ), a results based financing (RBF) instrument that uses public-private partnerships and aims to more effectively target access to basic services. 1 The audience for this review will primarily be task teams and managers working on project design and supervision in the relevant sectors, as well as donors and governments who are interested in lessons in light of potential scale-up and mainstreaming of OBA approaches. The review has been undertaken jointly by the Global Partnership on Output-Based Aid (GPOBA) and the IDA-IFC Secretariat. What is OBA? OBA ties the disbursement of public funding in the form of subsidies to the achievement of clearly specified results that directly support improved access to basic services. Basic services include improved water supply, energy access, health care and education, communications services (ICT), and transport (mainly roads). In the case of OBA, outputs are defined as close to the desired outcome or impact as is contractually feasible. For example, an output might be the installation of a functioning household connection to the electricity network. In some cases, an output might also include a specified period of electricity delivery demonstrated through billing or collection records. Subsidies in OBA schemes are defined as public funding used to fill the gap between the total cost of providing a service to a user and the user fees charged for that service, justified by the need to improve basic living conditions or the existence of positive externalities. 2 Universe of OBA. When OBA was launched in the WBG with the 2002 PSD Strategy, 22 OBA projects with a total estimated value of about $100 million had been identified. Now, nearly 130 OBA projects with a total subsidy value of about $3.3 billion have been identified in the WBG 3 (figure 1). 1 Results-Based Financing (RBF) refers to a range of mechanisms designed to enhance the performance of aid through incentivebased payments. RBF has been used most extensively in the area of health systems. RBF is an umbrella term that includes inter alia output-based aid, performance-based inter-fiscal transfers, and conditional cash transfers. What these mechanisms have in common is that a principal entity provides a financial or in-kind reward, conditional on the recipient undertaking a set of predetermined actions or achieving a pre-determined performance goal ( 2 In some cases, for example for public goods such as roads, user fees may be zero. 3 The $3.3 billion figure is skewed upwards by a few large health and road projects. This excludes related government subsidy co-financing of $2.7 billion. Therefore the total value of WBG OBA schemes including co-financing is $4.9 billion. All dollar values are US dollars unless designated otherwise and figures are as of April (Note: Since the PSD Strategy, some additional projects of fairly large subsidy value have been identified that were approved before 2002, while some of the other projects identified at the time did not materialize or the OBA component was dropped.) 8

9 Figure 1: Volume of OBA Subsidy by Sector and Region in the WBG WBG OBA Portfolio by Region (Total = US$ 3.3 bn) WBG OBA Portfolio by Sector (Total = US $ 3.3 bn) MENA 1% SAR 9% AFR 29% Water & Sanitation 6% Education 5% Energy 6% EAP 3% Health 20% LCR 56% ECA 2% Transport 61% Telecom 2% Of those projects, 29 have closed, 83 are under implementation and for the most part delivering outputs, and 16 are in design stage (figure 2). Another 66 OBA schemes have been identified outside the WBG mostly in the rural energy, ICT and roads sectors 4. Figure 2: Number of Projects by Project Status Number of Projects by Project Status (Total = 128) Closed 23% Implementation 65% Design 12% Funding for OBA has come from IBRD, IDA, GPOBA, other donors such as KfW, and governments themselves through, for example, tax revenue and the collection of explicit (including cross) subsidies from users. Those OBA schemes in developing countries that do not involve donor support are mainly found in middle-income (IBRD) countries. 4 GPOBA continues to identify OBA projects within and outside of the Bank; the figures here reflect projects identified up to April 30,

10 Piloting and mainstreaming. The piloting phase of OBA has provided a wide array of lessons and in many cases has been a success. The 128 OBA projects that have been implemented or are underway in the WBG reach at least 58.9 million beneficiaries worldwide 5. The first OBA pilots were launched in the ICT and roads sectors in the 1990s in Latin America (LCR). These pilots were replicated and scaled-up first in that region, and eventually lessons learned were transferred from IBRD countries to IDA countries. It could be reasonably claimed that OBA has become mainstreamed as one of the mechanisms used to expand access in these two sectors in many parts of the world and in fact many non-wbg OBA projects identified are in ICT and roads. In the health and off-grid energy sectors, OBA has become an important results-based financing instrument enabling improved access for the poor, and OBA schemes are increasingly found outside the WBG as well. OBA is still at pilot stage in the water, education, and grid-based energy sectors, although lessons for scale-up in some cases are now starting to become available. Although OBA is increasingly being used as a tool to increase access for basic services, the percentage of the OBA portfolio as compared to overall IBRD and IDA activities is small. About 2.7% of the World Bank project portfolio in the transport, ICT, health, water & sanitation, energy and education sectors, approved between fiscal year 2000 and 2009 used an OBA approach 6. ICT was the sector using OBA most commonly 7, with 11% its portfolio using OBA, followed by health (5.7%) and transport (3.7%). So, although OBA is gaining ground and recognition, there is still some ways to go in terms of scaling-up so that real strides can be made toward improving access to basic services for the poor. Enabling environment. The reasons for OBA s predominance in some sectors, as well as origination in Latin America in the 1990s, are to a large extent related to sector and country circumstances. More specifically, in order for an OBA approach to be viable, service providers must be able to take on performance risk, and in particular, pre-finance investments until subsidies are disbursed based on output verification. Although there are several cases of OBA with public sector providers, private sector operators traditionally are better structured to respond to performance-based incentives 8 and are usually better able to pre-finance outputs. Thus there appears to be a correlation with the prevalence of OBA and the sector and regional experience with publicprivate partnerships (PPP). This would imply that OBA will take stronger root where contractual and regulatory practices are traditionally more supportive of the private sector taking risks. At the same time, OBA can be an important mechanism through which efficiency gains from sector 5 Data on the number of beneficiaries is not readily available for public access ICT and transport projects. Beneficiary information is particularly limited in the case of transport sector (available only in two of the 23 World Bank implemented projects) due to the difficulty in identifying number of beneficiaries for non-exclusive road projects. 6 There are several contributing factors to this low percentage besides that OBA has not been fully mainstreamed. While the WBG OBA portfolio only includes projects that aim at increasing household access to basic services, the overall WBG portfolio also includes projects financing large upstream investments, wider sector reform programs, AAA, etc and the overall WBG portfolio obtained from the WBG Business Warehouse includes sub-sectors such as mining, railways, ports or nutrition for which no OBA projects have been identified. 7 However, the overall ICT portfolio is relatively small ($657 million), compared for example to transport (over $30 billion), which helps explain the small percentage of OBA subsidies in ICT even though OBA is most mainstreamed in that sector. Additionally, OBA subsidies in the ICT sector mainly was financed through levies on service providers, not donor funds, so that the exact subsidy amounts depend on service provider revenues and are unknown at the start of the project. The review identified at least $6.2 billion in levies that were used to fund universal access funds. 8 As a consequence of ownership change from public to private, increases in labor productivity, service quality and investment have been reported in competitive markets. Gains in productivity and profitability associated with privatization have for example been demonstrated by Megginson, Nash, and van Randenborgh (1994), Frydman et al (1997); La Porta and Lopez-de-Silanes (1999), and Brown et al (2006) (Gassner et al, June 2007, p.3, footnote 6). 10

11 reform are shared with users through improved access, and thereby can help underscore the benefits of PPP. Cross-cutting lessons. The OBA portfolio as a whole has been analyzed against the benchmarks/criteria that were set out at the early stages of piloting as purported advantages of OBA compared to traditional approaches. These benchmarks/criteria and the cross-cutting lessons in relation to best practices and challenges are provided below. These lessons are gleaned from the 194 Bank and non-bank OBA projects reviewed. Most of the lessons are taken from the projects that have closed or are under implementation, although in some cases there were important and interesting lessons to be learned from projects at design stage (table 1). Table 1: Benchmarks/Criteria and Cross-cutting Lessons from OBA Portfolio Benchmarks/ criteria Increased transparency through the explicit targeting of subsidies Increased accountability by shifting performance risk to service providers Increased engagement of private sector capital and expertise Cross-cutting lessons from OBA portfolio OBA provides a good platform for targeting infrastructure and social services subsidies. The focus on subsidies for access is inherently propoor: the poorest segments of the population often cannot afford initial access (e.g. cost of connection, health insurance) and therefore often do not benefit from subsidies for on-going service provision. Further, by explicitly defining outputs, targeting can be made more precise. The process of output verification can also provide an additional check on the targeting of subsidies, and is helping provide early evidence that OBA schemes are reaching the poor. 9 Compared to similar input-based schemes, OBA shifts performance risk to service providers by virtue of the fact that payments to providers are made after delivery of verifiable access and service. But the degree of performance risk shifted depends on the ability of the service provider to pre-finance investments/services until output-based payments are disbursed. Ultimately access to finance will determine how much performance risk is reasonably shifted to the provider. OBA does leverage private funding, but because of its generally pro-poor nature, private financing leveraged will be limited by the extent user fees (e.g., tariffs) can incorporate investment costs while remaining affordable. What is particularly noteworthy is the ability, through relatively small amounts of OBA subsidy, to mobilize private sector expertise to serve customer segments the private sector might otherwise not serve. Ultimately, the effective use of private sector participation is dependent on the enabling environment, for example the depth and quality of experience with PPP contracts, regulation, and access to finance. 9 Explicit targeting of subsidies for specific users and uses is common across all the sectors where OBA is prevalent, except for the roads sector (and a limited extent ICT), where the public goods (access for all) nature makes it difficult to exclusively target specific beneficiaries. 11

12 Benchmarks/ criteria Encouraging innovation and efficiency Increased sustainability of public funding Enhancing monitoring of results Cross-cutting lessons from OBA portfolio There is some evidence that output-based payments have led to improvements in operational efficiency and the delivery of innovative (often propoor) access to service solutions. Further, OBA has demonstrated efficiency gains through competition in most sectors when competitive pressures have been applied in the selection of the OBA service provider (noting that competitive tendering processes can take time). The focus on outputs as opposed to inputs should lead to innovations which translate into future efficiency gains, as has been seen in ICT and to some extent in roads. It is too early to analyze whether OBA schemes have provided long-term sustainable solutions. There is no evidence to date to suggest that schemes that involve OBA subsidies are less sustainable than their input-based counterparts, and in fact, the design of OBA schemes (for example, greater degree of demand risk shifted to service providers given the link between outputs and uptake, which in turn incentivizes efforts at stakeholder participation and education through community organizations, NGOs, etc) can enhance longer-term sustainability. By paying on verified outputs, OBA internalizes the monitoring of results. Best practice would also use the monitoring platform of OBA beyond just the verification of outputs to also check other aspects of service delivery. With OBA schemes, accountability also increases for donors and governments: public funding is linked to the fact that preidentified outputs are (were) delivered and therefore waste or inappropriate allocation of such funding should be minimized. Next steps. OBA is not a panacea. It is one mechanism within the spectrum of results-based financing mechanisms which is showing clear promise. But cross-cutting challenges for successful OBA schemes remain even for the sectors where OBA is mainstreamed. Although some of these challenges are not specific to OBA, they must be addressed in order for OBA to continue to make a mark on aid effectiveness, especially in light of new challenges brought forth with the on-going financial crisis: access to finance to pre-finance outputs/secure and sustainable sources of subsidy funding; capacity of implementing agencies and service providers; donor coordination and harmonization; and, limitations to the use of geographic targeting as scale increases. Further, as most OBA schemes involve PPP arrangements, the most basic aspects for the relevant enabling environment should be in place. Not all these elements may be required at the time of piloting, but attempting to short-cut key issues related to, for example, tariff setting and adjustment, quality of service standards, and contract adjustment and monitoring, may lead to delays for the pilot and will need to be addressed in detail for scaling-up. Understanding the basis for 12

13 an enabling environment for OBA schemes in any given sector is a prerequisite for sustainable OBA programs. The WBG, including its multi-donor programs, has a large role to play in addressing these challenges in terms of financial support and technical assistance (e.g., transaction advisory, capacity building, tariff/subsidy reform). The importance of WBG co-ordination can also demonstrated here. OBA is a platform that can effectively bring together the range of instruments across IDA and IFC in scaling up targeted infrastructure and social services. For example, while IFC investments can play a role in supporting the project with financing and bringing in other private capital, the IDA financing may be used for the OBA payment. Some strategic guidance has been provided through the Sustainable Infrastructure Action Plan (2008), Innovating Development Finance (Girishankar, 2009), the PSD Strategy Update, and the 2007 World Bank Strategy for Health, Nutrition and Population, to name a few. Efforts to integrate OBA within country assistance strategies and broader sector programs are underway, but more needs to be done. More work is needed to ensure OBA projects are able to maintain their output driven nature while continuing to meet Bank procurement and financial management requirements. The Investment Lending Reform (Concept Note, January 26, 2009) currently being considered by OPCS will be attempting to tackle these very issues, and OBA will provide some important lessons for this initiative as well. The ultimate decision on the success of OBA or of any other aid effectiveness tool rests with the WBG client governments and their interest, ownership, and commitment to design and sustain such approaches. The WBG has a key role to play in working with the broader development/donor community to demonstrate that OBA is an important mechanism to help improve access to basic services and reach the MDGs. The existing information and expertise on OBA, including this review, provide a solid underpinning for the successful design of pilots or programs that respond to client needs. 13

14 1. Introduction 1.1 Background Output-based aid (OBA) is being used as an innovative mechanism to deliver basic infrastructure and social services to the poor. The concept was introduced in the WBG in 2002 through the Private Sector Development (PSD) Strategy and more formally in January 2003 when the Global Partnership on Output-Based Aid (GPOBA) was launched as a World Bank-administered donorfunded pilot program to test the approach with a view to mainstreaming it within IDA as well as with other development partners. The IDA14 Mid Term Review (MTR) in November 2006 was an early opportunity to present the emerging lessons and preliminary findings of the OBA approach. The IDA14 MTR discussion highlighted that although it was too early to draw definitive conclusions, the emerging findings were encouraging. One of the recommendations of the IDA14 MTR discussion was that there should be focused scaling-up of OBA within IDA operations, as well more collaboration across the WBG, and that the results of the scaling-up would be reported at the IDA15 MTR. This has therefore been the focus of OBA within the WBG over the last two years, led by GPOBA of which IFC is a major funding contributor. In early 2008, the IDA-IFC Secretariat was created to foster greater collaboration between IDA and IFC. OBA has been identified as one of the instruments that could be used in joint IDA-IFC operations. Therefore, in preparation of the Secretariat s first Annual Report for May 2009 (IDA- IFC, 2009), and as a precursor to the IDA15 MTR in the fall of 2009, GPOBA together with the IDA-IFC Secretariat has conducted this review of lessons learned and best practice. 1.2 Objective and Audience The objective of this review is to provide a more definitive and practical understanding of lessons and best practice related to OBA, and the use of OBA as one approach to enhance the effectiveness of donor funds, including through promoting public-private partnerships. The audience for this review will primarily be task teams and managers working on project design and supervision in the relevant sectors as well as donors and governments who are interested in lessons in light of potential scale-up and mainstreaming of OBA approaches. The guiding principles will be to better articulate the lessons learned from the various applications of OBA and best practices by sectors. 1.3 Execution of Review/Methodology The review has been conducted by GPOBA and the IDA-IFC Secretariat. Much of the work has been conducted in-house by the GPOBA monitoring and evaluation team (M&E team), which is tasked with documenting and disseminating lessons learned both best practice and challenges from OBA schemes in and outside the WBG. Guidance and peer review has been provided by sector experts who have worked on both OBA and non-oba projects in the Bank. The methodology broadly involved the following: a) Identify the universe of OBA projects including projects funded outside of GPOBA and the WBG. 14

15 b) Gather information on project design, implementation and results. c) Compile sector specific lessons learned and best practices of OBA as well as lessons from the various applications of OBA such as one-off subsidies, transitional subsidies and ongoing subsidies. For this review, 194 OBA projects were identified and analyzed. They are classified as follows based on the OBA funding source (a few of which involve co-financing and some of which are now under consideration for scaling-up): GPOBA projects This includes 48 projects that have received either technical assistance and/or investment subsidy funding (or are in the process of receiving funding) from GPOBA. World Bank projects This includes 80 OBA projects that are funded by the World Bank Group, independent of GPOBA. Non-Bank Projects This includes 66 OBA projects that are funded either by other donors such as KfW and GtZ, or by governments themselves such as the OBA scheme for water and sanitation services in Chile. This also includes OBA projects in developed countries such as the special education voucher scheme in the United States. It is not possible to conduct an exhaustive search, but the review attempts to capture a representative sample. Further, the review has largely focused on developing countries. For each project, the following information was sought: Project design: Design elements captured include output definitions, payment triggers, financial sustainability/tariffs, targeting, total costs, funding source, role of private sector, transfer of performance risk to service providers, and administration and monitoring of the OBA scheme. Project implementation and results: This includes results of bidding (where applicable), efficiency gains, delivery of outputs and disbursement of funding, lessons learned and problems encountered during project preparation and implementation. Where possible, direct comparisons was made with input-based projects. OBA projects were identified with the help of OBA practitioners and experts both within and outside the WBG (e.g. the consulting firm Castalia for additional ICT and transport projects). The team also relied on several in-depth studies separate from this review exercise, such as the extensive Regulatel study (Stern and Townsend) undertaken by the WBG s Global ICT practice. For WBG (including GPOBA) projects, information on project design and results was obtained from the following standard documents. The review team also contacted task managers and other team members for any additional information and clarifications. Project Appraisal Document (PAD) Project Information Document (PID) Operations Manual (OM) Implementation Status Report (ISR) Unaudited Interim Financial Reports (IFR) 15

16 Implementation Completion and Results Report (ICR) In addition, the following documents that are specific to GPOBA were analyzed: GPOBA semi-annual Reports Independent Verification Agent Reports Post Project Report For non-bank projects, information is typically limited. The M&E team relied heavily on internet searches and discussions with sector experts to gather information on these projects. 1.4 Structure of Report Section 2 of this report starts by providing a clear definition of OBA, and then puts OBA in the context of other results-based arrangements and subsidy mechanisms. This section also provides a description of the various applications of OBA: one-off, transitional or ongoing subsidies. Section 3 of the report provides an overview of where OBA approaches are being implemented. Section 4 provides a cross-sector review of lessons learned from implementing OBA. The OBA schemes identified are analyzed against certain benchmarks/criteria which had been set out as purported advantages of OBA at the early stages of piloting. Challenges and best practices are analyzed across sector lines where appropriate, in particular if implementation in sectors where OBA is less prevalent can benefit from lessons where OBA is more mainstreamed. Section 5 of the report summarizes challenges to be addressed by the WBG and ends with a view to next steps. Annex I presents a table of all OBA projects identified in the WBG as of April 30, Annex II provides greater detail on the use of OBA in each of the sectors analyzed, on a sectorby-sector basis. Annex III provides data on the WBG OBA portfolio in context of the overall WBG portfolio for the relevant sectors for the same time period. 16

17 2. Context 2.1 Definitions OBA ties the disbursement of public funding in the form of subsidies to the achievement of clearly specified results that directly support improved access to basic services. Basic services include improved water supply, energy access, health care, education, communications services, and transport. In the case of OBA, outputs are defined as close to the desired outcome or impact as is contractually feasible. For example, an output might be the installation of a functioning household connection to the electricity network. In some cases, an output might also include a specified period of electricity delivery demonstrated through billing and collection records. The intended outcome of such an output-based scheme would be to improve the basic living conditions of the poor household, reducing indoor household pollution, increasing opportunities for education through better lighting or through information passed through radio and television, and the like. The intended development impact could include for example a reduction in morbidity or increased lifetime earnings. Subsidies are defined as public funding used to fill the gap between the total cost of providing a service to a user and the user fees charged for that service. 10 Policy concerns such as improving basic living conditions for the poor or the existence of positive externalities from a reduction in disease may justify the use of subsidies. Both the definition of outputs and the design of subsidy mechanisms are discussed in greater detail below. 2.2 Traditional vs. OBA approaches Neither performance arrangements nor subsidies are new to the developing world or the WBG. Performance contracts have been implemented for several decades, using both public and private operators. But outputs in OBA schemes are generally more narrowly defined than benchmarks in traditional performance arrangements, which in some cases may be more input-oriented. Subsidies have also existed in infrastructure and social services sectors. But OBA refines the targeting of subsidies by bringing together performance-based arrangements and subsidies through the explicit linking of the disbursement of subsidies to the achievement of agreed outputs. Figure 3 below provides a simple contrast of a traditional input-based approach to an outputbased approach. 10 In some cases, for example for public goods such as roads, user fees may be zero. 17

18 Figure 3: Contrast of a Traditional Input-based Approach to an Output-based Approach Traditional (inputbased) Approach Output-Based Approach Inputs (such as materials) Private Finance Inputs (such as materials) Private financing mobilized by service provider Service Provider Public Finance Service Provider Reimbursement for outputs delivered Service Recipient Service Recipient Source: Brook and Petrie, 2001 Because of the clear link between pre-identified outputs and ex post payment (or, subsidies ), the following advantages of OBA over traditional approaches are assumed: 1) Increased transparency through the explicit targeting of subsidies, tying these subsidies to defined outputs 2) Increased accountability by shifting performance risk to service providers by paying them only after they have delivered an agreed output 3) Increased engagement of private sector capital and expertise by encouraging the private sector to serve customers (usually the poor) they might otherwise disregard 4) Encouraging innovation and efficiency by leaving the service solutions partly up to the service provider and through least cost determination of subsidy required 5) Increased sustainability of public funding through the use of one-off subsidies and linking ongoing subsidies to sustainable service 6) Enhancing monitoring of results since payments are made against agreed outputs. Some of these advantages were postulated in the PSD Strategy of 2002 and were also the basis of analysis in the IDA14 MTR discussion paper on OBA. These are also the criteria/benchmarks 18

19 against which the current portfolio of OBA projects identified in the WBG and outside is analyzed in this review. 2.3 Contracting out service delivery: the spectrum Under OBA schemes, services are contracted out to a third party provider, and that contract or other official arrangement is the mechanism through which the output-based disbursement criteria are established. The third party in OBA schemes is typically a private enterprise, but could also be a public utility, NGO, community-based organization, or even a separate branch or institution of government from that entity providing the official public funds. Over the last two decades, schemes that harness private financing to deliver infrastructure services have expanded considerably. Under traditional procurement, private infrastructure services are contracted at the input end of the spectrum, with the government purchasing specific inputs and using these to build assets and provide services itself (figure 4). Figure 4: Contracting Spectrum OBA Outputs Independently verified Design Inputs Build, Operate Outputs (Intermediate) Outcomes Development Impacts -Output specification -Service provider selection OBA Outputs include -Water connection made & service provided -Solar Home System installed & maintained -Medical treatment provided Contracting closer to the input end (e.g. the construction of water treatment plants) does not guarantee that the inputs the government purchases actually lead to the outcomes (e.g. a reduction in waterborne diseases) or impacts (e.g. decreased morbidity) that the government actually wants. But because outcomes and impacts are a combined product of what the provider can influence and other factors outside the service provider s control, either governments seeking to pay on outcomes and impacts will not find a willing, credible service provider or the provider will charge a substantial premium for making its payment contingent on factors that it cannot control. What can be done, however, is to contract for an output that is as closely related as possible to the desired development outcome or impact, while performance risk is still largely under the service provider s control. This is the rationale behind output-based aid. Outputs would include, for example, contracting for functioning yard taps as part of a water supply program, a specified number of people to be vaccinated in the case of health programs or working public payphones or solar home systems to be installed in villages in the case of ICT or energy. But in order to ensure sustainability and for service providers to take on appropriate demand risk, OBA usually involves a mixture of payment on outputs and outcomes e.g. use of electricity or ICT services. However, the farther one goes along the output/outcome/impact spectrum, the greater the risk 19

20 borne by the service provider, and consideration must be given as to whether the provider is reasonably able to bear that risk, and at what cost. 2.4 Applications of the OBA approach OBA schemes can provide subsidies in three ways: one-off subsidies such as connection subsidies; transitional tariff subsidies which taper off as user contributions increase; or ongoing subsidies. The subsidy design chosen will depend on factors such as the sustainability of the funding source, the capacity for administering the subsidy scheme, the type of service to be subsidized, and the extent to which the service provider is willing and able to be paid over time (box 1). Box 1: Applications of OBA Subsidy Design Mechanisms One-off subsidies are the most common application of OBA approaches and usually involve capital subsidies for access to a given service. An example is when a large portion of the subsidy is paid after the targeted beneficiaries are connected to a network and connections are verified. Given that in OBA approaches the emphasis is on service delivery rather than on physical connections, even in the case of one-off subsidies a portion of the subsidy may be withheld and paid only after verification of a certain number of months of satisfactory service delivery (hence a mixture of outputs and intermediate outcomes against which one-off subsidies are disbursed to provide a measure of improved access). Transitional subsidies can be used to support tariff reforms, where a subsidy is used to fill the gap between what the user is deemed able and/or willing to pay and the cost-recovery level (e.g., long-run marginal cost) of the tariff. The subsidy is transitioned out after a specified period of time (e.g., months or years) as the user contribution increases (and possibly as tariff levels required for cost recovery decrease with efficiency gains). In these cases, the output against which the subsidy is paid is the service delivered and billed by the provider. The review identified only a handful of transitional OBA schemes and very few of those are still in place. Ongoing subsidies may be required in cases where there is a continuous gap between affordability and cost recovery including for consumption costs. As in the case of transitional schemes, ongoing subsidies should be paid out against pre-determined targeted outputs in order to be considered OBA. For utility services, most OBA schemes in water, energy, and telecommunications rely on oneoff subsidies enabling initial access, partly because OBA is targeted to the poor and the poor are usually not connected to network services in the first place so often cannot benefit from ongoing or transitional tariff subsidies. A one-off OBA subsidy may be used, for example, to help connect a poor household to the water or electricity network or to reduce a community s contribution for provision of pay phones or internet points of presence. Ongoing output-based subsidies in the utility sectors are seen more often in countries with higher rates of access. For example, in Chile an income-based targeting scheme channels an ongoing output-based subsidy through service providers to poor urban households for a life-line (minimum acceptable) amount of water consumed Most other commonly used quantity-based tariff subsidies such as increasing block tariffs, however, are not OBA. Such schemes usually charge tariffs below cost for low-consumption, because it is assumed that poor households consume small amounts. These subsidies are usually intended to be financed by cross-subsidies from higher-consuming customers (who are charged higher tariffs). However, the amount of cross-subsidy received by the operator is not related to the amount of subsidy provided to the low-consuming households, but rather, the subsidy collected from the high-consuming households. Therefore, the operator earns the subsidy from the high-consuming households whether he serves the targeted households or not. These issues are discussed in greater detail in Section 4. 20

21 Ongoing output-based subsidies normally fund the provision of basic services or maintenance in OBA projects in roads, health, and education. OBA road maintenance schemes require ongoing subsidies for the life of the road, often funded through road funds. OBA health schemes, to ensure continued access to care for the poor, often channel subsidies in an ongoing manner through health care providers as they deliver agreed services, such as well-child visits, over a period of time although some health projects may focus on interventions of a one-off nature such as safechild delivery. 2.5 OBA as part of the Results-Based Agenda OBA is an important part of the WBG s spectrum of results-based solutions of development finance. For example, OBA is one key element of a broader universe of results-based financing (RBF) in the health and education sectors, where RBF is a payment made to a provider, payer or consumer when measurable actions are taken or defined performance targets are achieved (Brenzel et al., 2009) Other RBF mechanisms include conditional cash transfers and cash on delivery (see health section in Annex II). This paper will not cover parts of the results-based spectrum that are not OBA. However, it is worth noting that some of different results-based mechanisms can play complementary roles. For example, conditional cash transfers (CCT) described in greater detail in the health and education annexes may be enhanced through the use of OBA-type mechanisms. CCTs can provide incentives for users to partake in health, education and other schemes (e.g. sanitation) that exhibit positive externalities when they otherwise might not choose to; OBA can help incentivize providers to ensure that the services are available when users demand such a service. This is an area to be explored further. 21

22 3. Snapshot of OBA in the WBG and Beyond 3.1 Universe of OBA under review At the time of the PSD Strategy Implementation Progress Report in 2003, 22 OBA projects with a total estimated value of about $100 million 12 had been identified although since then the number of projects identified that were approved before 2003 has risen to 33, with a total funding amount of $1.3 billion. Now, about 128 OBA projects with a total subsidy value of about $3.3 billion (excluding the $2.7 billion subsidy funded by recipient governments) have been identified in the WBG 13 (figure 5). Of those 128 projects identified in the WBG, 29 are closed, 83 are under implementation and for the most part delivering outputs, and 16 are in design stage. This review draws mostly from the closed projects and those under implementation, although there are also some important lessons to be learned from project design. Another 66 OBA schemes have been identified outside the WBG, mostly in the ICT, transport (mainly roads), and energy sectors. There may be more OBA schemes that were not discovered by this review. Figure 5: Volume of OBA Subsidy by Sector and Region in the WBG MENA 1% WBG OBA Portfolio by Region (Total = US$ 3.3 bn) SAR 9% AFR 29% Water & Sanitation 6% WBG OBA Portfolio by Sector (Total = US $ 3.3 bn) Education 5% Energy 6% Health 20% LCR 56% ECA 2% EAP 3% Transport 61% Telecom 2% To put this in context, OBA is only a small share of the World Bank portfolio, at 3% in total. Since 2000 between 0.56% and 7.9% of project volume approved each year used OBA. The largest share of OBA projects was 13.1% of funding volume in the ICT sector, followed by health (6.3%) and Transport (4.2%) (figure 6). There are several contributing factors to this low percentage, aside the fact that OBA has not yet been fully mainstreamed. While the WBG OBA portfolio only includes projects that aim at in- 12 All $ in this review represent US dollars unless designated otherwise. 13 GPOBA continues to identify OBA projects within and outside of the Bank; the figures here reflect projects identified up to April 30,

23 creasing household access to basic services, the overall portfolio also includes projects financing large upstream investments, wider sector reform programs, AAA, etc and the overall WBG portfolio obtained from the WBG Business Warehouse includes sub-sectors such as mining, railways, ports or nutrition for which no OBA projects have been identified. For a breakdown by sector, please refer to Annex III. Figure 6: Comparison of WBG OBA Portfolio to Total WBG Portfolio $12,000 $10,000 $8,000 $6,000 $4,000 $2,000 $- Total WBG funding approved (FY in m) Total $5,539 $6,705 $6,831 $7,333 $9,400 $8,214 $8,608 $11,136 $10,635 OBA $31 $83 $133 $68 $743 $247 $382 $132 $403 The regional and sector breakdown of these OBA projects is provided in table 2. As the table shows, most of the projects in the WBG are currently in AFR (due to recent piloting efforts by GPOBA) and LCR (which is where the first OBA pilots in each sector reviewed began). Projects outside of the WBG were overwhelmingly in the ICT and transport (i.e. roads) sector, and predominantly in LCR. Table 2: Distribution of OBA Projects AFR LCR EAP SAR ECA MENA Grand Total Water & Sanitation Energy Transport Health Telecom Education Grand Total The nearly four-fold increase in the number of OBA projects in the WBG within a period of five years is due to a variety of factors, in particular: 23

24 an increased emphasis on results and accountability by donors and governments, including the WBG results agenda; an explicit recognition that well-designed subsidy schemes are an integral part of a propoor infrastructure and social services delivery strategy; and, a recognition that in order for PPP to be successful, specific attention needs to be paid to pro-poor service delivery. This explicit acknowledgment that subsidies are sometimes necessary, coupled with new evidence that many existing subsidy schemes such as quantity-based subsidies embedded in tariffs often have a regressive targeting incidence, have contributed to the appeal of more targeted subsidy schemes such as OBA (Komives et al., 2005). 3.2 Funding OBA Funding for OBA schemes has come from IBRD, IDA, GPOBA, other donors such as KfW, and governments themselves through, for example, tax revenue and the collection of cross-subsidies from users. The WBG is the biggest donor with over $3.3 billion committed to fund subsidies to approximately 80 projects 14. Many of the WBG s original OBA projects were in the Latin America region and in the roads and ICT sectors. Subsequent roads and ICT schemes have built on the lessons from these schemes with varying degrees of success and expanded into other regions so that there are now a substantial number of roads and ICT schemes in regions such as AFR. This is discussed in the sector sub-sections in Annex II as well as in Section 4. Of the $3.3 billion WBG portfolio, the majority of the projects (59%) are in IDA countries compared to 29% percent in IBRD countries. In terms of subsidy volume, IBRD countries account for the largest share at 56% of the total WBG subsidy portfolio compared to IDA s 39%. About 12% of projects, accounting for 5% of the subsidy volume, are located in IDA/IBRD blend countries (figure 7). Figure 7: Comparison Portfolio vs. Number of Projects in IDA, IBRD and Blend Countries Distribution of OBA subsidy (Total = US$3.3 bn) BLEND 5% Distribution of OBA projects (Total = 128 projects) BLEND 12% IDA 39% IBRD 56% IDA 59% IBRD 29% 14 Excludes projects with GPOBA subsidy funding or technical assistance. 24

25 It is worth noting that some of the WBG projects have received substantial amounts of complementary subsidy funding from the recipient governments worth a total of $2.7 billion. The bulk of this government funding has been in the transport and health sectors, accounting for 64% and 30% respectively. More than 76% of this funding came from IBRD governments, though they accounted for only 48% of the projects that received complementary subsidy funding from government. Therefore the total OBA subsidy portfolio for WBG projects (including government co-financing) is $6 billion. In addition to the 80 projects described above, another 48 projects have either received funding or are in the process of being funded by the World Bank-administered program, GPOBA. GPO- BA was created in 2003 by the UK s DFID and the World Bank. GPOBA was originally intended to help assist in the preparation of OBA projects and document and disseminate the lessons learned. In 2005, through an additional DFID contribution, GPOBA also became able to help fund actual subsidy schemes. This galvanized the development of over 28 GPOBA-funded projects, most of which are under implementation or for which grant agreements are imminent. New donors have since joined GPOBA, including the Netherlands government (DGIS), AusAID, Sida, and the IFC. GPOBA has to some extent focused on designing and developing OBA schemes in areas where OBA has been less tested, in particular the water and sanitation sector. Over 60% of the GPOBA projects are in IDA countries and they account for 72% of GPOBA funding volume. A majority of GPOBA projects are in the water and sanitation sectors (52%), followed by energy (25%), and together they account for over 80% of the GPOBA subsidy volume (figure 8). Further, as described in greater detail below, although OBA was originally envisioned as a tool to enhance private sector participation, GPOBA has also attempted to pilot OBA with commercially-viable state-owned enterprises (SOEs) in sectors where public utilities have continued to play a dominant role in service provision. Figure 8: Distribution of Subsidy for All GPOBA Projects (Including Design Stage) GPOBA Subsidy Distribution (Total = US$ 145 m) BLEND 8% IBRD 20% Water & Sanitation 51% GPOBA Subsidy Distribution (Total = US$ 145 m) Energy 30% IDA 72% Telecom 3% Health 16% 25

26 Millions OBA schemes have also been identified outside the WBG both in developed and developing countries. In some cases donors are playing an active role, such as KfW in the health and renewable energy sector or DGIS with the Energising Development program, implemented by GTZ in energy. More generally, for developing countries, OBA schemes that do not involve donor support are mainly found in middle-income (IBRD) countries which are able to fund subsidy schemes largely from cross-subsidies or tax revenue. 3.3 Sector focus Since OBA approaches can vary a great deal depending on the sector context, best practice and challenges encountered are easier to delineate by sector. Some of these lessons can be transferred across sectors, and where this is the case, such lessons are discussed in Section 4 of this review. But what is clear is that in order for OBA or any innovative mechanism to become relevant, it must be able to adapt to regional and sector circumstances and constraints. Figure 9 shows the key sectors under review and provides the amount of subsidy of OBA projects by sector and region identified in the WBG; table 3 shows the distribution of the OBA portfolio by sector and region. Annex II provides a more detailed description of lessons learned on a sector-bysector basis. Figure 9: Distribution of OBA Portfolio by Sector and Region $2,000 $1,800 $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $- Distribution of OBA Portfolio by Sector and Region AFR EAP ECA LCR MENA SAR Telecom Water & Sanitation Energy Education Health Transport Table 3: Distribution of OBA Portfolio by Sector and Region Transport Health Education Energy Water & Sanitation Telecom Grand Total AFR $ 635,588,000 $ 144,815,164 $ - $ 58,350,000 $ 97,861,433 $ 21,490,705 $ 958,105,302 EAP $ - $ 32,140,000 $ - $ 40,600,000 $ 23,275,640 $ 10,076,630 $ 106,092,270 ECA $ 55,000,000 $ - $ - $ 10,100,000 $ - $ - $ 65,100,000 LCR $ 1,303,026,000 $ 375,104,000 $ 41,140,000 $ 59,560,000 $ 47,380,880 $ 29,950,000 $ 1,856,160,880 MENA $ 40,000,000 $ 6,232,100 $ - $ - $ 8,400,000 $ - $ 54,632,100 SAR $ - $ 119,000,000 $ 138,007,143 $ 24,370,000 $ 2,264,743 $ 11,900,000 $ 295,541,886 Grand Total $ 2,033,614,000 $ 677,291,264 $ 179,147,143 $ 192,980,000 $ 179,182,696 $ 73,417,335 $ 3,335,632,438 26

27 4. Lessons Learned: Cross-cutting Challenges and Best Practice In this section, the OBA portfolio as a whole is analyzed against the benchmarks/criteria which were set out as purported advantages of OBA at the early stages of piloting. These benchmarks/criteria include: 1) Increased transparency through the explicit targeting of subsidies, tying these subsidies to defined outputs 2) Increased accountability by shifting performance risk to service providers by paying them after they have delivered an agreed output 3) Increased engagement of private sector capital and expertise by encouraging the private sector to serve customers (usually the poor) they might otherwise disregard 4) Encouraging innovation and efficiency by leaving the service solutions partly up to the service provider and through least cost determination of subsidy required 5) Increased sustainability of public funding by allowing one-off subsidies, and linking ongoing subsidies to sustainable service 6) Enhancing monitoring of results since payments are made against agreed outputs. Challenges and best practice are analyzed across sector lines where possible, in particular so that implementation in sectors where OBA is less prevalent can benefit from lessons where OBA is more mainstreamed. Most lessons are drawn from the WBG portfolio of 29 closed projects and 83 projects under implementation most of which are delivering outputs. Lessons are also drawn to some extent from the 16 projects at design stage and from non-wbg projects where sufficient information is available (figure 10). Figure 10: Number of Projects by Project Status Number of Projects by Project Status (Total = 128) Closed 23% Implementation 65% Design 12% 27

28 To judge the merits of OBA, this review aims to answer the following two questions: (i) to what extent do OBA projects meet the six criteria listed above, and, (ii) does OBA address these criteria better than traditional aid approaches? To answer the first question, the review mainly relies on information and lessons gathered on GPOBA projects that are under implementation and from Bank and GPOBA projects that have closed. Information from evaluation studies and other publications is used to the extent available. With regards to the second question, the main challenge is establishing a valid counterfactual. To allow conclusions as to the relative effectiveness of OBA, OBA projects need to be compared to projects with similar objectives, using other approaches. However, in reality no two projects are totally alike. For example, two water supply projects providing poor households with access to clean water may result in vastly different unit costs, depending on how much distribution infrastructure they involve, whether they involve water treatment infrastructure, etc 15. Nevertheless, some studies were found that can help put OBA in the context of other aid modalities. For example, on the targeting incidence of utility subsidies and the pro-poor benefits of connection subsidies such as those that are predominant with OBA, most notable is a 2005 World Bank publication by Kristin Komives et al on Water, Electricity and the Poor: Who Benefits from Utility Subsidies. And, a number of studies 16 have demonstrated cost savings/quality improvements, resulting from competitive selection of service providers in roads, health projects and ICT, compared to previous traditional provision of service. In summary, when reviewing the lessons learned for each of the six criteria listed above, the section draws both on the record of the OBA portfolio against these criteria, and also on general or specific comparisons with input-based schemes. Altogether, cross-cutting lessons and best practice guidance can be confidently derived. In future, this analysis will be supplemented with results from on-going impact evaluations or other studies, which will add to the increasing body of knowledge on results-based financing in general, and output-based aid approaches more specifically. Examples of OBA schemes are illustrated to help provide a clearer picture on how lessons are drawn. However, most detail on the OBA portfolio is provided in the annex both the table in Annex I, as well as the sector sections in Annex II for ICT, transport (namely roads), energy, water, health and education. 15 The GPOBA M&E team has also explored the option of using unit cost data from the World Bank Africa Infrastructure Country Diagnostic Study. However, the Country Diagnostic Study uses a disaggregated concept of unit costs that does not allow a systematic comparison. The country diagnostic study includes outputs such as meters of pipe of a certain quality and diameter laid or stand-posts installed, whereas the output definition for OBA includes necessary network extension and installation of final connections. 16 Add roads, health, etc. where this was confirmed 28

29 4.1 Transparency: Explicit targeting of subsidies for the poor Traditional approaches to subsidies versus OBA subsidies Traditional input-based schemes which subsidize specific investment projects such as power plants or more general budget support for utilities are often equivalent to across-the-board subsidies, as they decrease the tariffs needed to cover costs. Wealthier households tend to consume more utility services than poorer households so that the bulk of such subsidies benefit non-poor households. Further, a large percentage of the poor who are often not connected in the first place cannot benefit from these across-the-board subsidies. Quantity-based tariff subsidies that charge lower tariffs for lower quantities of water or electricity are a common way of attempting to target utility subsidies to the poor. 17 However, empirical evidence from the water and electricity sectors shows that such subsidies usually have a regressive targeting effect in that larger proportions of subsidies benefit richer households. This is largely because many of the poorest households do not have access to these services in the first place, but also because of inherent regressive characteristics of quantity-based tariffs such as: - fixed charges that negate subsidy benefits if they increase effective average tariffs of low consumption households - increasing block tariffs subsidizing first units consumed of all income strata alike whether rich or poor and regardless of total consumption behavior - in some cases, low social tariffs that result in incentives to actually increase the upfront connection costs for users as compensation for lost revenue through the tariff subsidy (Komives et al 2005, p 79ff) - the lack of clear correlation between income and consumption in the case of water (Komives et al 2005, p 82ff, p 167). The seminal study by Komives et al has demonstrated (largely through simulations given the lack of data) that connection subsidies to provide initial access to households in the first place are inherently more pro-poor than quantity-based tariff subsidies that are linked to the amount of service used. OBA in the infrastructure sectors mainly relies on one-off capital subsidies for increased access usually through connection, in the case of the network industries. These OBA subsidies aim to increase access to poor households in the first instance. Then, if there are quantity-based subsidies implicit in the system, poor households can also benefit from them if the quantity-based subsidies are properly designed to actually benefit the poor. Well designed OBA schemes in the utility sectors usually rely on the target population being able to afford sustainable tariffs that cover ongoing costs of service provision in other words, although access may be subsidized, tariffs or running/operations costs may not be. But this is often not a major hurdle: the poor usually are paying more for alternative services (HDR 2006, p. 52, p. 83). 18 In some cases, additional pro-poor mechanisms are required to ensure effective targeting. For example, tariff affordability issues for the very poorest can be partially mitigated by 17 Quantity-based subsidies through tariffs in ICT are less common as the sector has moved toward collecting explicit subsidies through universal access funds, but are still in place in some former Soviet bloc countries. 29

30 subsidizing pro-poor access (such as public water points) which have lower running costs per capita, and also by ensuring appropriate schemes which fit the payment patterns of the poor. In the social sectors, traditional funding mechanisms may not primarily benefit the poorer segments of the population. while governments devote about a third of their budgets to health and education, they spend very little of it on poor people that is, on the services poor people need to improve their health and education. Public spending on health and education is typically enjoyed by the non-poor (WDR 2004, p.3). For example, the poor frequently live in areas with little or no access to health care services such as rural areas and poor neighborhoods. They are less able to wield political influence to direct health care spending to the areas they live in and the basic services they need most urgently. Also, the usual form of funding hospitals and health centers benefits all patients and all types of interventions. When buildings, drugs, machines and salaries are subsidized, all users benefit, regardless of whether they require basic emergency care or less urgent up-scale procedures. OBA subsidies, on the other hand, often through on-going subsidy mechanisms, target the services the poor are more likely to use (see below). Several OBA schemes in the health and education sectors help mitigate the cost of access to poor households as well such as the quasiinsurance schemes in Latin America (see health sub-section in Annex II) Sharpening traditional forms of targeting with OBA All OBA schemes by definition must specify the outputs against which subsidies will be disbursed, and consequently beneficiaries can be identified more clearly than in traditional inputbased schemes. An analysis of the OBA portfolio to date describes the following forms of targeting used in OBA schemes. Most OBA projects use geographic targeting. This form of targeting is useful when intended beneficiaries are concentrated in certain areas and where few people outside the target group live. For projects in such areas, excluding unintended beneficiaries can be costlier than including them. For example, an OBA water project is channeling subsidies for connections in slum areas of Kampala, where most households are very poor and excluding the few non-poor households would be too costly. Geographic targeting is more complicated and less effective in cases where the poor and the non-poor live relatively interspersed. Self-selection targeting involves designing projects so that outputs chosen by poorer beneficiaries receive a higher share of subsidies. Subsidies can be targeted progressively by providing higher subsidies for more basic services for example, smaller solar home systems, as in a rural electrification project in Bolivia or by subsidizing services less attractive to the rich such as external yard taps in urban Mozambique. Self-selection targeting is also widely used in the health sector. Because wealthier patients tend to favor more sophisticated up-market facilities, OBA projects usually finance more basic health care services. Many OBA projects have used self-selection to complement geographic targeting (box 2). 18 Based on extensive research on the patterns of service delivery to and consumption by the poor, the poor are on the whole paying more per unit of service consumed and often of inferior quality. This indicates that tariffs that cover the ongoing cost of service delivery are likely not a major hurdle for the poor (Bardasi and Wodon, 2008). 30

31 Box 2: Nepal Biogas Support Program Geographic Plus Self-selection Targeting The Biogas Support Program in rural Nepal provides household-size biogas plants to families. Biogas plants use decomposition of organic waste, such as cow manure, to produce a flammable gas that can be used for cooking and lighting. The subsidies vary according to the plant s size and location. Smaller plants, used by poorer families, receive relatively higher subsidies than larger plants. Wealthier families, with more livestock to provide input, prefer larger plants with greater gas output. Plants in remote mountainous regions, where the population is poorer, receive a higher subsidy than plants in the Terai lowlands, where the population is richer. The higher subsidy for remote mountainous regions is also meant to offset their higher construction costs. The Biogas Support Program has successfully installed over 150,000 biogas plants funded by DGIS, KfW and the Community Development Carbon Fund. As of April 2009, 4,772 biogas plants have been installed and verified under a GPOBA-funded component. Means-tested targeting is used in several OBA schemes in middle-income countries. Means testing involves measuring a beneficiary s wealth to assess whether a subsidy is warranted. Such schemes require more advanced administrative systems. For this reason, OBA schemes that rely on means-testing usually piggyback on broader welfare programs that identify poor households for a variety of public services. This is being done in an urban gas service project in the Armenia Access to Gas and Heat Supply for Poor Urban Households project funded by IDA/GPOBA, for example. One approach used by some OBA projects includes proxy means testing, in which easily observable characteristics such as possession of indicative assets, for example a dwelling of a certain size, is used as a proxy for income. Marie Stopes International, an NGO that serves as management agency in the Reproductive Health Vouchers in Western Uganda, jointly funded by KfW and GPOBA, uses a simple questionnaire assessing assets owned, number and quality of meals per day, and other indicators of potential beneficiaries to determine eligibility. Community-based targeting relies on collaboration with the local community or its representatives to help identify the community members most in need of the service. Community involvement can increase ownership and reduce the risk of targeting criteria being rejected by the population in the service area. But community-based targeting may have drawbacks, such as the risk of being hijacked by special interests. Moreover, this form of targeting can be time consuming, as evidenced by the Water Access with Small-Scale Providers project in Cambodia (Navarro 2008) Role of output verification for targeting Eligibility criteria for beneficiary households are usually clearly defined and made a precondition for subsidy disbursements. The third-party verification that triggers disbursement of OBA funds can sometimes include verification that the poverty targeting criteria have been met. For example: One of the outputs in an IDA/GPOBA urban gas project in Armenia is defined as an individual gas-heater based solution for households living on an average per capita income of approximately $0.5 per day. Using the existing social protection program supporting 31

32 low-income households in Armenia, the output verification ensures that the output is delivered to eligible households. One of the primary output criteria in the Colombia Natural Gas Distribution for Low Income Families in the Caribbean Coast project is proof that each newly connected household belongs to one of Colombia s two poorest economic strata as officially classified with an average per capita income of less than $1 per day. To avoid excluding the poorest from benefiting from the service, the project also includes the provision of a basic gas stove. Voucher schemes (as used in health and education) can be used to market services specifically to the poor, for example by selling them in poor areas or targeting social marketing campaigns to high risk groups. Verification of targeting can be included in voucher schemes by making vouchers non-transferable, for example by registering biometric data of voucher recipients. In projects using geographic targeting, output verification and therefore subsidy disbursement only takes place in pre-identified low income areas. For projects using self-selection targeting, verification that the correct outputs are delivered (e.g., outdoor yard taps versus indoor water connections) usually implies verification of targeting. This does not imply that the output verification process inherent in OBA schemes leads to fool-proof targeting, but if verification is reliable and robust, it can enhance targeting substantially (see Section 4.6 on monitoring) Conclusions on explicit targeting of subsidies for the poor The review conducted to date concludes that OBA provides a stronger platform from which to target infrastructure and social services subsidies than traditional forms of subsidies in these sectors. Firstly, the access focus of OBA schemes can help ensure limited subsidies are reaching those who need them most. Secondly, explicit targeting linking subsidies to specific users and uses is common across all the sectors where OBA is prevalent, except for the roads sector, where the public goods (access for all) nature of roads makes it difficult to specifically or exclusively target the poor, and to some extent in the ICT sector, when the target is expanding coverage. Thirdly, the process of output verification also inherent to OBA schemes provides an additional check on the accurate targeting of subsidies, and is helping provide early evidence that OBA schemes are reaching the poor. Ultimately, the choice of targeting mechanism for any OBA scheme will depend on several factors, in particular: the nature of service delivery in the sector; cost-benefit (table 4); and the existing enabling environment, in particular the type of social welfare mechanisms already in place and the ability to appropriately monitor them. 32

33 Table 4: Cost and Effectiveness of Targeting Mechanisms Targeting approach Cost / admin. complexity Targeting effectiveness Geographic Low Low-Moderate Self-selection Low High Means testing High High Proxy means testing Moderate Moderate Community- based Moderate - High Moderate Source: GPOBA and Grosch et al. A combination of geographic and self-selection targeting seems to be the most promising methodology for OBA in lower income countries, while income and means-testing targeting seem more appropriate in the middle-income or lower-middle income (IBRD/blend) countries within which the WBG operates. 4.2 Accountability: Shifting performance risk to providers 19 By paying service providers on the delivery of pre-specified outputs, OBA schemes should shift performance risk to the service provider, the entity technically best able to bear and manage that risk. More specifically, the components of performance risk that are shifted to the service provider through OBA mechanisms to a greater extent than traditional input-based schemes include the following: Construction risk related to infrastructure and other investments made under the project, particularly the risk of cost overruns or benefit shortfalls due to non-delivery of outputs or delivery of inappropriate or insufficient outputs; Operational risk related to ongoing service delivery; and, Demand risk related to whether the intended beneficiaries request the service provided (at the price provided) in the first place (also known as uptake risk ). While shifting performance risk to service providers, OBA can raise some additional unintended risks, in particular, the payment risk that once outputs have been pre-financed and delivered as agreed, subsidy disbursements (or, payments to the provider) are substantially delayed or not made. OBA design must incorporate measures to mitigate for this potential risk Construction risk and one-off subsidies There are a large number of sources providing anecdotal evidence that cost overruns and benefit shortfalls occur fairly frequently in international aid projects and infrastructure projects in general. But very few systematic studies exist that can be used to compare results with the sample of OBA projects that are under implementation or already completed. A series of papers by Flyvbjerg et al, some of which were published by the World Bank, showed consistent cost overruns in transport projects. In his 2002 article Underestimating Costs in Public Works Projects: Error or 19 The enhanced accountability in the use of public funds and potential reduction in corruption is discussed in the section on monitoring in

34 Lie (Flyvbjerg 2002) the authors find that of 258 transport projects, 86% exceeded cost expectations with an average overrun of 28%. The authors also reject with overwhelming significance the thesis that the error of overestimating costs is as common as the error of underestimating costs (Flyvbjerg 2002, p. 282). OBA can help mitigate some of the risk of cost overruns (or benefit shortfalls) related to project investments through one-off OBA subsidies for access. OBA subsidies are fixed before project implementation but paid only after outputs have been delivered. This presents a credible cap on public funding available so that service providers are aware that they have to bear any cost overruns. Further, the explicit nature of output and subsidy design should clearly identify the risks being taken. And, OBA disbursements are usually not tied to the completion of project input milestones, such as the completion of a telecom tower, but rather to fund connections such as pay phones and tele-centers up and running. The service provider only gets paid for the parts of the system that are actually being used. Thus there is a disincentive for creating excess capacity, and instead an incentive for increasing access. However, it is important that relevant running costs will be covered by tariffs or fees paid by the new customer (although this is not usually the constraining factor for access for poor households see Section 4.1 on targeting above). Given the lack of a comprehensive body of literature on benefit shortfalls/cost overruns in development projects, this review compares the results of closed OBA projects with comparable closed World Bank projects. The GPOBA M&E team analyzed a sample of all 37 available ICRs for World Bank-funded water, energy and health projects closed in fiscal year This sample was reviewed to identify all projects that have quantifiable outputs as project development objectives (PDOs). 20 This sample was compared with 13 completed OBA projects in water, energy and health for which results information was available 21. The review showed that a total of 85% of OBA projects achieved or over achieved the desired results within or below budget, compared to 49% of traditional projects. Nearly 70% of OBA projects were completed under budget, compared to slightly more than half of the traditional projects. While some of the traditional projects recorded cost overruns, none of the OBA projects did. Similarly, only two OBA projects (15% of sample) did not achieve the intended results, compared to at least 30% of the traditional projects reviewed. The two OBA projects that did not deliver all outputs projected disbursed only a small proportion of the funds related to outputs actually delivered. 22 This provides some indication that OBA can help mitigating risks to the project sponsor of disbursing substantial amounts of funding for projects that do not produce the desired outputs (figure 11). 20 Projects with PDOs relating to policy formulation/institutional strengthening or consisting of PDOs with higher level project impacts not exclusively under the control of the project implementer were excluded from the analysis due to the inability to compare against OBA. 21 The analysis is based on information from ICRs for IDA/IBRD-funded projects and on information from the GEF website for projects funded but the GEF. 22 Funds disbursed included some fixed costs related to setting up the project. 34

35 Over-achieved Achieved Not (fully) achieved Unclear Under run Within Budget Overrun Figure 11: Comparison of Performance of OBA and Traditional Projects 80% 70% 60% 50% 40% 30% 20% 10% 0% Comparison of Performance of OBA and traditional Projects OBA Non-OBA Results Budget These results are consistent with the presupposition that OBA shifts performance risk to service providers and helps to counteract cost overruns and benefit short falls Operational risk transfer through on-going and one-off OBA subsidies Are service providers bearing sufficient risk for on-going service provision, after construction is complete? In many PPP contracts whether OBA or not service providers do bear operational risk. With the addition of an OBA mechanism whereby payments for investments made are actually with-held until pre-identified outputs are delivered, OBA schemes can provide an additional hard incentive for performance. With ongoing OBA subsidy schemes, particularly those in health and roads, performance-based payments to service providers are based on continuous service delivery of a stipulated quality. Performance contracts can define the minimum level of service to be delivered at an agreed payment. This shifts the performance risk of the project entirely to the service provider who is to some extent free to decide how to reach performance targets, and can therefore probably better manage operational risks related to service delivery. As discussed in more detail in Annex II, performance-based roads contracts have shifted more ongoing service delivery risk to roads contractors as compared to traditional forced accounts or contracting. Such contracts have, for example, reduced the share of roads in a poor condition in Argentina from 25% to 5% and increased the average road maintenance rating in Florida, USA, from 51 to 87, well above the agreed minimum standard defined in the performance contract, while at the same time allowing the government funding agency to save money (Segal et al 2003) (FHWA 2005). 35

36 On the other hand, one-off subsidies for access do not necessarily shift performance risk to service providers for the entire duration of their service contracts, unless the project involves significant investment by the service provider to be recouped through the tariff. In order to ensure more sustainable services, OBA projects involving one-off subsidies should take into account the nature of the longer-term service contract, including clear standards of performance, related penalties, and regulatory rules and practices under which the OBA scheme is operating. The link between sustainability, subsidy levels and tariffs is discussed in more detail in section 4.5 below. For example, although most of the ICT contracts identified do not disburse OBA subsidies based upon the provision of ongoing service throughout the duration of the contract, the contracts do include longer-term service provision requirements for about 5 to 10 years. Service providers that are in breach of their contracts can also lose their operating licenses, which tend to be for 15 years or more. Although they do not make reference to the OBA scheme, the ICT licenses provide additional incentive for meeting service quality obligations. In contrast, in the case of the dealer model for off-grid energy solutions, there are no long-term contracts for service provision. Therefore the issue of sufficient performance risk transfer and sustainability comes into question. To address this issue of long term sustainability, recent projects involve capacity building to train users and local technicians to fix non-functioning units, to develop product standardization and certification, and to negotiate longer term warranty schemes from manufacturers. The creation of the medium-term service contract as in the Bolivia IDTR project and the phased subsidy payments in the case of the Ghana dealer model to ensure proper maintenance in the first few years after installation are some mitigating steps that have been taken to help shift performance risk further out to service providers Managing demand risk in OBA schemes When payments (for investments made) to service providers are made on outputs delivered, and those outputs involve user applications and down payments, the service provider bears the risk of uptake. Although this risk can be partially mitigated through willingness/ability to pay studies and is not completely new to PPP schemes, for example concession arrangements which involve expansion of services, low-uptake is particularly relevant for infrastructure and social services to the poor since the poor are often neither familiar with the services (e.g., sanitation or medical interventions) nor are certain aspects of the services (e.g., payment schemes). The demand risk component of OBA schemes is substantial. This can prolong the time required for project roll-out. But it is important for demand risk to be shifted to the service provider, both to help address the issue of low uptake and because the service provider should manage investments and operations most efficiently to meet required demand. Both the Rural Community Water Project in Andhra Pradesh (GPOBA) and the Senegal On-Site Sanitation Project (IDA and GPOBA) utilize NGOs to promote community participation to improve uptake. The implementing agency of the Colombia Natural Gas Distribution for Low Income Families project reports that natural gas is the first utility service that some of the poorest beneficiaries will receive. Some households were reluctant to access the subsidy, even if 36

37 the use of gas resulted in a saving compared to traditionally used fuels such as wood and kerosene, because they were reluctant to commit to paying a monthly bill. To mitigate this, the implementing agency started outreach campaigns and gave beneficiaries savings boxes to create a habit of saving up for the gas bills, and in some cases distribution companies even initially over-subscribed the program given the degree of attrition between customer registration and actual service delivery Definition of outputs and access to finance determine nature of performance risk The degree of performance risk shifted to the service provider through OBA schemes is dependent on how outputs on which subsidies are disbursed are actually defined. As discussed in Section 2 of the review, traditional procurement of private infrastructure services that contract at the input end of the spectrum, whereby the government may purchase specific inputs or even assets, often does not guarantee that the government purchases actually will lead to desired outcomes. What OBA attempts to do, therefore, is contract for an output that is as closely related as possible to the desired outcome, while performance risk is still largely under the provider s control. For example, compared to traditional schemes in the health sector where public funding would be disbursed along with the building of a hospital or the procurement of special medical equipment, with OBA, payments to service providers would be dependent on the delivery of well-child visits or the administration of vaccinations. In energy projects, instead of payment for the delivery of transmission lines, an OBA scheme will disburse payment for working connections made and/or actual electricity delivered. The definition of outputs has often evolved as OBA has taken root in a sector and the degree of performance risk that providers are able to bear has in some cases increased. For example, OBA in the ICT sector often used construction or installation milestones as outputs. Through time, outputs have been refined. For example, some of the more effective contracts disburse a portion of subsidies upon installation of the phones and disburse the rest on a regular basis provided the phones are maintained to standard. But there can be wide differences in shifting performance risk based on output definition from sector to sector, and even within projects themselves. The contrast between the small-town component and the green-field component of the Uganda Water Supply in Small Towns project is explained in more detail in Annex II, and is such an example. In short, a more pure OBA approach is used where a sub-project mainly involves extensions from existing systems. But for the green-field operations, output-based payments are phased such that 60% of the subsidies are disbursed during construction and only 40% of the subsidies are disbursed with final connections and water delivery. The difference is attributed to the perception of the private operators ability to reasonably pre-finance the larger construction costs of the green-field schemes. In order to effectively shift performance risk to service providers through OBA, service providers need to be able to pre-finance investments and services. This pre-financing can be funded by the service provider through own cash flow, supplier credit and other aspects of working capital, and/or equity and debt financing. This finance must be available at reasonable cost in order 23 Colombia Natural Gas Project, Implementation Completion and Results Report, April,

38 to be affordable for the provider and to minimize the impact of financing costs on the tariff charged to the household. Access to finance has been a major issue in the design of OBA schemes, particularly when involving small and local providers, as well as when considering OBA for public providers. For example, even though ICT has moved closer to an ideal OBA model, a portion of OBA subsidies are still often paid up-front, given the cost of capital for prefinancing outputs. An interesting example of how OBA design has attempted to adapt to the access to finance constraint is provided in box 3 in relation to the CREMA road contracts in Argentina, which were among the first performance-based rehabilitation and maintenance contracts outside the OECD. Box 3: CREMA Phases I and II Based on positive experience with performance-based routine maintenance contracts (see Box 9:), in 1997 Argentina introduced a new contract combining rehabilitation and maintenance. The CREMA (Contrato de Recuperación y Mantenimiento) requires the Contractor to rehabilitate and subsequently maintain a subnetwork of roads for five years for a lump sum contract. Payments are made when a specified level of service has been achieved. The CREMA contracts implemented between 1997 and 1999 specified that rehabilitation works should be carried out during the first year of the contract. The Contractor received 5-10 percent of the contract price as an advance payment, and additional payments at the end of the first year when rehabilitation works have been completed. But the largest percentage of the contract price, about 50 percent, was paid in 48 equal monthly installments spread over the remaining 4-year contract period. This front-loading of rehabilitation and delayed payment schedule resulted in contractors having to finance much of the rehabilitation themselves. The rehabilitation costs sometimes exceeded 50 percent of the contract value. In the new generation of Phase II CREMA contracts, the contractor now receives full payment for rehabilitation works executed, proportionately to the outputs achieved during the first eighteen months of execution. While this approach has helped resolve contractors financing difficulties, it has given rise to another problem. With some contractors receiving up to 80 percent of the contract value for rehabilitation works in the early years of the contract, contractors incentive to perform their ongoing maintenance obligations across the multi-year life of the contract is reduced. Some contractors have tried to renege on their contracts once they have completed rehabilitation and not fulfill their maintenance obligations. Nevertheless, the first two phases of Argentina s CREMA program, covering nearly 14,000 km, resulted in significant improvement in the proportion of roads in good condition from 70% in 1998 to 85% in Further, the percentage of roads in poor condition decreased from 8% in 1998 to 4.2% in 2005 (Implementation completion report of National Highways Rehabilitation and Maintenance project in Argentina 2006, p.4). Source: Liautaud 2001; Highways Toolkit; Cabana 1999; Transport Note TN 27 The access to finance constraint seems most binding for sectors or sub-sectors that rely on small and local/regional providers as in the case of the off-grid energy schemes. But these constraints may impact the public sector as well. The predominance of public utilities in the water sector may be one explanation of the limited number of OBA schemes identified/implemented. Many of these public utilities or local municipalities are not in the position to pre-finance output delivery. However, there are exceptions. For example, in Morocco, the public utility of Meknes has taken on commercial debt in order to pre-finance output delivery and allow households to pay their connection costs in installments over time. In the health sector, projects such as the Yemen Safe Motherhood Program or the Lesotho New Hospital PPP (see Annex II) work with up-market service providers, which may be less constrained by access to finance, to reach out to 38

39 the poor. Access to finance can become more of an issue for projects that involve contracting out exclusive service provision in a poor area, particularly if this requires significant investments. As a result, most of these projects include some element of up-front grants Payment risk mitigation OBA may shift performance risk to the entity best able to manage that risk, but OBA can also lead to additional unintended risks. A further consideration in OBA schemes is payment risk. Even if outputs are delivered, what assurance is there that service providers will be paid, and paid on time, even after verification of output delivery? The structure of the funds flow for an OBA scheme can determine if (in the case of competitive schemes) the transaction can attract bidders, or in any scheme, whether financial institutions will be comfortable lending to the providers. In some GPOBA-funded schemes, private fiduciary agents such as banks or well-known multi-service accounting firms have been used for transparent funds flow. But when funds are channeled through the finance ministries and/or the national (central) banks, this may add time and therefore cost. One OBA scheme involving a privatized electricity company in Guatemala used the privatization proceeds to fund an OBA facility. The payment risk was guaranteed by a breach of contract coverage from the Multilateral Investment Guarantee Agency (MIGA), which is part of the WBG. In the roads sector, funding from and involvement of bilateral and multilateral development institutions in the CREMA contracts has provided greater assurance to contractors that they will be paid: By making the long-term payment obligation legally binding on the government, the CREMA has deterred the Treasury from failing to provide funding for road maintenance; and experience during implementation showed that at time of fiscal constraints, the budget process respected the CREMA contracts and funds were allocated to them in priority, as if they were considered as non-discretionary expenditures Conclusion on performance risk OBA does shift more performance risk to providers of infrastructure and social services as compared to similar input-based schemes by virtue of the fact that payments to providers are made after delivery of measureable and verifiable access and service. How much performance risk is borne by the OBA service provider depends on the extent that the output-based subsidy is with-held (and, like other PPP schemes, to the extent the service provider invests money to be recouped through tariffs or fees) (box 4). One-off subsidies for access do not necessarily guarantee the degree of performance risk shifted to service providers in future years; this would depend on factors such as the enabling environment, for example contractual and regulatory arrangements that are in place (see below). Also, how much performance risk is shifted to the provider depends on the definition of outputs and the extent of phasing-in of payments. This in turn depends on the ability of the service provider to pre-finance the investments and services until output-based payments are disbursed to it. Therefore, access to finance will to a large extent determine how much performance risk is reasonably shifted to the provider, and is an important area for donor and government co-ordination. 24 Performance Based Contracting for Preservation and Improvement of Road Assets. Resource Guide. Updated April World Bank. Available at: 39

40 To date, there is limited experience on mitigating the access to finance constraint with formal financial instruments such as guarantees. More work needs to be done in order to work with small and local providers, as these are the most likely providers of services in rural and periurban areas where access to services are often most in need. Some examples to date are described below and are relevant not only for pre-financing, but also for project financing in general, since access to affordable finance will impact the cost of the entire project and affect the level of required subsidy (see next section on leveraging private sector finance). Provision of guarantees to banks lending to output-based service providers. For example, K-Rep Bank in the Kenya water project has purchased a USAID Development Credit Authority partial credit guarantee to reduce the collateral required from borrowers. In many of the countries where OBA is operating, central banks discourage unsecured lending and so where banks are pre-financing works that will be subsidized with output based grants they still require the borrower to post collateral for the subsidized asset (or portion of asset). In Honduras for example, the issue of access to finance is being dealt with in two ways: a) for private providers including NGOs, limited commercial debt is possible (albeit with very short tenors/debt repayment periods), ultimately secured against municipal assets but with commercial lenders drawing comfort from a grant mechanism payable by the World Bank; and b) for public implementers, bridge loans are possible, i.e. government loans (issued effectively at zero interest) secured against future sector transfers from central to municipal governments. In the event of default by a public operator, the Municipality would lose its central government transfer. 40

41 Box 4: Limitations to Shifting Risks to Service Providers Although OBA in general can mitigate the risks to governments/donors/users of cost over-run and benefit short-falls as described in the review, there is still a need to take into account factors outside the control of the service provider as in the case of any well-designed intervention. In the case of the GPOBA-funded Morocco Urban Water and Sanitation with unit subsidies fixed in Moroccan Dirhams, the output projections were lowered by 14% in the first quarter of 2008 as a result of a 14% depreciation of the US dollar against the Dirham. The East-Meets West Foundation, an NGO providing output-based connections in the GPOBAfunded Vietnam water scheme, has received an increase in unit subsidy as a result of unit cost increases partially due to the recent increase in commodity prices. This is especially relevant for projects implemented by small and medium service providers who do not have means to hedge against price increases. In the Colombia Natural Gas project, the Grant Agreement specified the unit cost of $141 to be payable in Colombian pesos. Given the depreciation of the US dollar against the Colombian peso, this meant that the actual subsidy payable to the distribution companies was reduced significantly. At the time of grant agreement signing, the exchange rate was 2300 Colombian pesos to the US dollar, while in 2008 it was 1705 pesos to the US dollar. In the case of the project, the distribution companies effectively absorbed the depreciation. With the onset of the global financial and economic crisis in late 2008, it is difficult to predict which way projects will be affected on the whole. By October, 2008, the global economic landscape changed dramatically with the unraveling of the credit markets, yet inflationary pressures have eased considerably due to a crash in commodity and energy prices, with the dollar strengthening against most developing country currencies. It is important to take these issues into account when structuring an OBA project and have some flexibility to adjust subsidy amounts if the sustainability of the project is at risk. 41

42 4.3 Private sector capital and expertise Leveraging private sector debt and equity The ratio for leveraging private sector debt and equity in OBA projects is about 1 to 1.66, i.e. for every dollar of subsidy raised, about $1.66 of private sector financing was mobilized 25. These estimates only refer to longer-term private investments made in addition to any investment prefinanced and reimbursed through the OBA subsidy. The possibility of mobilizing private finance varies from sector to sector, with ICT and energy mobilizing more than health and water. For example, in Guatemala s FONDETEL projects, each $1 of subsidy leveraged between $2 and $4 of private investment. In Peru s FITEL projects, an average of $2 of private capital was raised for each $1 of subsidy. In off-grid energy, the average private sector financing leveraged is about 1 to 1.4. Dealers are typically small and medium operators with limited capacity to take credit risk for extending loans to rural households and also lack experience in credit-facility management. On the contrary, for the Senegal rural electrification concession, the winning bidder has proposed to more than double the minimum number of connections set in the tender from 8,500 to 21,800 by bringing in $9.6 million in private financing. This constitutes about 60 percent of the total financing, compared to the 20 percent minimum private financing requirement under the tender. IFC has recently been requested to participate, possibly with an equity stake (see energy sub-section in Annex II). For subsidies that partially finance ongoing service provision, the amount of private capital mobilized is difficult to identify, because the amount of subsidy is not determined as a share of the investment cost. In such projects, the service provider also has to pre-finance investments for a much longer period of time. One lesson is that for network/utility services, private finance leveraging is wholly related to tariff reform: ultimately, the service provider must be able to recoup these costs through the tariff. If the aim is to have a smaller amount of subsidy with more of the investment recouped through private financing, the tariff would need to be able to absorb these costs. Often this would be feasible only if contracts were of a longer nature than is normally accepted. Because OBA schemes target the poor, who often are charged social tariffs or who consume small amounts, the possibilities of leveraging in the traditional sense are limited compared to non-oba schemes which do not target the poor. The link between tariffs and output-based subsidies is critical not only for the issue of leveraging private finance, but also for ensuring that service providers are not over-compensated through OBA subsidies yet at the same time are incentivized to make the output-based connections. This is discussed in Section 4.5 Sustainability and Enabling Environment. 25 The ratio is derived from 33 private sector projects for which the amount of private financing could be identified. 42

43 Mobilizing private sector expertise Another aspect of mobilizing the private sector is encouraging private service providers to connect and serve poor customers who the private operator would otherwise not serve. This is achieved by OBA schemes extending existing or newly created assets by providing relatively small amounts of subsidy to incentivize the private operator to reach these customers. For example, previous infrastructure investments may have been made and have excess capacity, but there is no real incentive to serve additional mainly poor customers. This is the case in several urban water projects including those in Manila and Jakarta (described in boxes 16 and 19 respectively, in Annex II). OBA interventions in these cases lead to very efficient subsidy per capita results, whereby small dollar amounts can connect poor households to a network otherwise unreachable. For example, in the case of the Manila Water project, the Manila Water Company (MWC) is investing some $14 million in new water supply infrastructure in these areas, but the low-income households cannot afford the connection charges set by MWC and the Regulator. Under the proposed OBA scheme, the household would contribute PhP 1,620 ($36) towards the connection charge and GPOBA would provide a subsidy for the remainder (PhP 5, or $131). This results in a subsidy per capita of $29.6. Box 5 presents private sector mobilization in the Colombia natural gas project. Box 5: Colombia Natural Gas Distribution for Low Income Families in the Caribbean Coast OBA and the Private Sector In 2006 GPOBA signed a grant agreement with Fundacion Promigas, a charitable foundation established by the Colombian gas transportation and distribution company Promigas S.A. The project has connected 35,000 poor households from the two lowest of five socioeconomic strata. Gas connection prices in Colombia are regulated by the national regulator and in order to ensure equity gas companies are not allowed to offer connection below the regulated price. This precludes many of the poorest Colombians from accessing natural gas. The subsidized connections were made by regional gas companies owned by Promigas, who marketed the project to the poor target groups, provided payment plans to beneficiaries that allowed them to pay the remaining connection fees over a time of up to 5 years and documented connections and consumption of beneficiaries for verification by an independent auditing firm. Subsidies were disbursed only for connections made to households in strata 1 and 2 who also were provided with a basic stove and who had completed three months of successful billing of services. The project successfully achieved its target of connecting 35,000 poor households within the estimated time. Promigas and the gas companies absorbed shortfalls resulting from USD depreciations during the project. Fundacion Promigas conducted a successful community outreach campaign, creating demand for the subsidized connections. This campaign included measures to convince beneficiaries for many of whom natural gas is the first utility service that they receive, to commit to paying monthly bills. Private sector expertise and discipline brought through OBA schemes can bring benefits to the delivery of social services as well, as described in more detail in the health and education sector annexes. In the Bogota Concession Schools scheme, the freedom to choose the teaching and administrative staff led to a better quality of education than in the public schools where the 43

44 teachers union made it difficult to implement staff changes. On average, 55 percent of the subsidy amount was allocated to human resources, well below the 90 percent in the public school system, freeing up 33 percent for nutritional support and education materials. Also, the private sector providers took the initiative to partner with parents and the community, which helped reduce the dropout rates and improve educational attainment, compared with regular public schools (Barrera-Osorio 2007). More examples of efficiency gains through OBA schemes usually involving the private sector are provided in section 4.4 below. However, working with the private sector is not always a panacea in and of itself and does require consideration of a host of issues, for example in relation to information asymmetry and principle-agent problems that have been analyzed through a vast array of studies on PPP in infrastructure and social services for the last couple of decades. Capacity can also be an issue when working with the private sector in OBA schemes. That is clearly the case with small and local private providers as seen with off-grid energy or rural water projects, and even in the roads and ICT sectors where OBA is more mainstreamed. Successful projects tend to involve capacity building elements for the private sector, including learning how to bid for a contract, how much to bid, how to self-monitor against outputs, how to mitigate against payment risk, and so on Conclusion on private sector involvement This OBA review analyzes several forms of leveraging, and the record of OBA varies depending on the type of leveraging considered. Private financing requires recouping costs through tariffs in the case of network industries, and it is often not possible for tariffs to incorporate a large element of investment costs while remaining affordable for the poor. Further, in sectors with even clearer positive externalities such as health, education and sanitation, user charges should not be set at levels that discourage participation. But the real success of OBA is the ability of relatively small amounts of OBA subsidy to mobilize private sector expertise to poor areas where the private sector would otherwise not go. There are success stories here from all sectors. Where private sector experience is bringing efficiency gains and market discipline to sectors, such as in the health and energy sectors, more needs to be done to encourage and strengthen potential private sector providers, especially small and local ones. For example, increased capacity building including on billing, marketing etc, as well as greater partnerships with local organizations can enhance PPP with OBA further. Key to enhancing private sector finance and expertise will be tackling the access to finance constraints for medium to long-term financing which would enable greater participation of the private sector not only in OBA, but in PPP in general. OBA schemes can provide some lessons, especially from the rural energy sector where tackling access to finance seems to have been a priority over the last few years: In poor, off-grid areas, rural affordability increases substantially with micro-credit and longer term fee-for-service arrangements. Typically 2-3 % of residents can afford cash payment for the service, but with microcredit the customer base can increase up to percent of residents. Longer term, fee for service arrangements could increase the customer base even further (Terrado 2008). To address the limited liquidity of MFIs which 44

45 hampers the more widespread adoption of SHS, an IDA line of credit worth $11.4 million is made available for the Bangladesh RERED project to provide long-term credit refinance to eligible MFIs to finance households' or individuals' purchases of SHS. With time, the private operators have reduced the interest charged to consumers and already the largest volume supplier (Gramen Shakti) has reduced the interest to 6% flat. (Source: RERED case study from RE toolkit website). An IDA line of credit is also being used in Ghana SHS to provide long-term liquidity to participating private rural banks. Repayment of the line-of-credit will be retained in a revolving fund to provide consumer credit to additional customers to purchase Solar-PV systems. Guarantees to mitigate risk and increase loan tenure are being used in a Papua New Guinea rural electrification scheme. The program allows credit terms to be extended from three to five years (REtoolkit note, Nov 2008). 45

46 4.4 Innovation and efficiency Fiscal benefits: Efficiency gains from OBA Using competition to determine the amount of subsidy required is one of the more tested ways to ensure maximum value for money as long as transaction costs do not prove prohibitive, and if there is not already an existing provider that can reap clear economies of scale. OBA lends itself readily and transparently to competitive processes, since the bidding variable is often set at the lowest subsidy required to meet expected outputs given fixed user charges. 26 Here are some examples (all described in greater detail in the sector sub-sections in Annex II) of how OBA projects have resulted in efficiency gains, usually using competitive tendering processes based on lowest subsidy required or greatest numbers of beneficiaries reached: In a Mongolia ICT project, competition resulted in 28% savings in the total subsidy required for the original areas/beneficiaries to be served. The savings were used to fund the Chulut Soum wireless center, which is estimated to have expanded the project to 1,000 more beneficiaries. The IDA-funded Bolivia IDTR project for rural electrification led to 25% more beneficiaries for the fixed subsidy than the minimum required under the tender; and a 40% reduction in SHS prices compared with the Bolivia UNDP (2004) project. As of February 2009, the IDTR has installed 6,154 individual systems, benefiting over 30,000 people in rural Bolivia. In addition, 87 social systems were installed in schools and clinics. In the case of the Sri Lankan rural electrification OBA, one dealer introduced 15-percent price discounting and its own consumer finance, seeking to capture market share. Most recently, the competitive bidding process in Water Supply in Uganda s Small Towns project has resulted in an average efficiency gain in the ten towns of about 20% -- although the project is still in early stages of implementation. Demonstration effects have also led to leveraging. The bidding for the GPOBA-funded OBA scheme expanding mobile phone services in Mongolian towns has demonstrated that mobile phone services can be provided on a commercial basis without subsidies in some cases although not for internet services nor for facilities such as schools. Many other examples of zero subsidy bids exist in the ICT sector, for example in Chile and recently in Nicaragua where the incumbent offered to pay rather than receive monies from the Universal Access/Service Fund (UASF) in order to gain market share. Similarly, the bidding for piped water supply in Uganda s Small Towns has demonstrated that in some cases, extensions can be made on a commercial basis, with the private sector estimating that it can recoup cost related to new connections through the tariff: at least two towns received zero subsidies requirements through competitive tendering. 26 Based on experience, including from the Paraguay water OBA which attempted both forms of bidding, it is recommended that it would be better to fix tariff levels ex ante and bid on the minimum subsidy per connection (or maximum beneficiaries served, given a set subsidy). This would prevent any resentment from the regulator/users about inequity of tariffs across the country and would also allow for inclusion of a lifeline tariff if the national policy requires it. 46

47 However, competitive processes take time and can require extensive capacity building, including in relation to bidding or in obtaining access to finance. The transaction process can be costly both in terms of administrative costs and in terms of time required for the process, especially if capacity is low or the situation is highly political. Further, there is a risk of under-bidding during the tender process, followed by financial problems down the line, especially if the growth in demand does not materialize as expected. And in particular for small-scale projects, oversophisticated (often donor-led) systems with a wide array of checks and balances may prove costly, cumbersome and lead to inaction. These costs should be weighed against the many obvious advantages of competition to drive down costs through efficiency gains Improvements in operational efficiency Other demonstrations of efficiency gains, which may not be as easily quantified, are for example quality enhancements or improvement in service delivery. The disbursement of funds after service delivery can create strong incentives to deliver outputs in a timely manner. For the Armenian Access to Gas and Heat Supply for Poor Households, there is evidence that timeliness of service delivery and quality of work has led to increased customer satisfaction. The provision of telecommunications services to nomadic herder communities in Mongolia required innovative solutions related to difficulties with determining the adequate size of solar cells and the prediction of mobility patterns of beneficiaries. These difficulties were overcome within the first six months of project operations and output-based payments most likely helped trigger such a speedy resolution. The health sector in post-conflict countries, for example in the Democratic Republic of Congo and Afghanistan, has shown that contracting out services to NGOs can lead to quicker and more comprehensive coverage than building up an input-based health system. Other examples of ongoing output-based subsidies that perform better than comparable inputbased approaches are performance-based roads management contracts. In Argentina the CRE- MA [Contrato de Recuperación y Mantenimiento] program has substantially improved the condition of the network, reducing the share of roads in poor condition from 25 percent to less than 5 percent by the end of As a result, road users costs have been reduced by more than 10 percent (Liautaud 2004). Similar positive experiences of cost savings and increased quality have been reported in the United States. In Florida, Virginia, the District of Columbia and Texas the use of maintenance contracts resulted in cost reductions and significant improvements in the quality scores of maintained roads (Segal et al 2005) Output-based innovations OBA is clearly an innovative mechanism which enables a variety of service solutions to reach the poor. The portfolio of OBA schemes analyzed shows a wide array of service providers and a variety of technical solutions used. Further, one assumption when OBA was initiated was that the focus on outputs would itself enable increased innovation. This has certainly been the case in the ICT sector, where service providers have to some extent been free to provide the most appropriate and affordable technology to respond to demand. The general trend in ICT with or without OBA is a clear reduction in costs. What OBA has brought in addition is the recognition 47

48 that there are other markets to be tapped, and that more pro-poor models of technology such as prepaid services can help reach these markets. In the energy sectors, many OBA schemes are bid out as technology neutral, although additional subsidies (e.g., from GEF) might be provided for renewable energy technologies. In the off-grid Bolivia IDTR project, service providers were required to provide credit to users, leaving the terms up to them. Different solutions and arrangements for example with micro-finance institutions were developed, based on the technologies, customers, etc which could not have been presumed by the project team ex ante. The achievement of cost reduction through OBA relies on the flexibility allowed to service providers to bring their commercial and operational practices into the structuring of OBA schemes. However, service quality and delivery requirements are often affected by the weight of the procurement policies and procedures of governments and donor agencies imposed on the service providers. These requirements and procedures rightly aimed at ensuring transparency and competition in the award of public-funded subsidies, are generally designed for structuring inputbased projects. This may create bias on how the service providers procure the inputs necessary to deliver the outputs, and limit freedom to proceed as innovatively as they would wish. This is especially true for smaller service providers, and may translate into higher cost. Lessons learned have demonstrated that the key is to specify those essential inputs that cannot be left out, but to leave some degree of discretion beyond that. This is critical in the infrastructure sectors where outputs should be supported by long-lived assets, but also in the health care sector where poor quality can have a detrimental impact. Finding a balance between specifying key input requirements to ensure long-term sustainability with providing space to innovate and lower costs can be difficult. Finally, considerations related to the scope and scale of the project must be taken into account. Competitive processes take time and can require extensive capacity building, including in relation to bidding and access to finance concerns. Lessons from ICT indicate that determining the appropriate level of subsidy often requires projections of demand and necessary investments particularly difficult in dynamic sectors such as ICT and energy. While large projects may justify this expense, smaller projects may not. In the case of off-grid projects, finding bidders can be difficult as these projects are perceived as risky and there is often a lack of technical capacity to undertake these projects in remote locations. There is a clear need for road shows, awareness programs, workshops and capacity building exercises including business development services, technical training, market surveys, databases on renewable energy resource availability and institutional building Conclusions on innovation and efficiency OBA has demonstrated efficiency gains through competition in most sectors when competitive pressures have been applied in the selection of the OBA service provider. There is also anecdotal evidence that the output-based nature of payments has led to improvements in operational efficiency. Further, although evidence as to a direct correlation is not available, it is possible that the use of output-based arrangements over a period of time in the ICT and roads sectors has led to increased innovation and a reduction in costs. OBA s impact on improving efficiency and encouraging innovation is largely driven by the ability to procure service provision competitively, and, a balance is required between specifying key input requirements to ensure long-term sustainability while providing space to innovate and low- 48

49 er costs from which future users can benefit. A Guidance Note to procurement staff has been issued (box 6), but more is needed to be done, including individual training and a general shift of mind-set. This may fit well within the broader push in the Bank to streamline procedures to meet client s needs more effectively and efficiently, as is proposed in the recent concept note on Investment Lending (IL) reform by OPCS dated January 26, The concept note describes the current disconnect between results focus and inputs-based wiring of IL and mentions that the Bank needs flexible and agile instruments that can easily adapt and respond to changes in the global development business (OPCS, 2009, pp 5 and 9). Box 6: Bank Procurement in OBA Projects 27 A Guidance Note to Procurement Staff on Procurement Issues in Structuring Output-Based Aid (OBA) Operations Financed by the World Bank was issued on April 11, The note, a joint effort by staff from the GPOBA unit of the Finance, Economics and Urban Development Department (FEU) and the Procurement Policy and Services Department (OPCPR), follows up on the November 17, 2005, Guidance Note for Staff on Structuring Output-Based Aid (OBA) Approaches in World Bank Group Operations and the OPCS November 7, 2005, Operational Memorandum on the Application of Paragraph 3.13 of the Procurement Guidelines. It identifies different scenarios and actions required by specialists to facilitate compliance with the Procurement Guidelines in the design of procurement arrangements for the two types of OBA schemes, namely: Projects where there is no existing service provider; or projects where there is already an existing service provider (incumbent concessionaire or equivalent arrangement). The Guidance Note provides illustrative practices and incorporates the lessons learned from procurement assessments conducted under OBA pilot operations across Bank regions. 27 The note is available in the OPCPR intranet website. 49

50 4.5 Sustainability, Tariffs and the Enabling Environment OBA design and sustainability Sustainability of infrastructure and social services schemes implies that an intervention has a long-lasting positive impact rather than having short-lived and easily reversible results. Some OBA schemes have been running for many years in ICT and roads in particular and have been scaled-up and replicated elsewhere in the respective regions and even into other regions. This is to some extent a testament of the sustainability (and of course replicability) of the model in these sectors. The evidence on long term sustainability for off-grid energy projects is limited to a few projects such as Sri Lanka and Bangladesh RERED, which have been operational for over a decade. The results are encouraging. In Sri Lanka for example, as of June 2008, some 120,000 HH were using SHSs, with 750 installations occurring monthly (Terrado, 2008). The results in Bangladesh are similar where by mid 2008, the SHS sales reached 8,000 units per month with total sales of about 180,000 (REtoolkit issue note, June 2008, pp. 132, 135). Both projects continue to use grants, but increasingly to reach the poorest segments of the population. Two characteristics of OBA in particular help address the issue of sustainability: a) nature of the subsidy design, and b) performance risk shifted to the providers. Nature of subsidy design: As already discussed in detail, OBA schemes are predominantly subsidies for access, which are by nature one-off capital subsidies. One-off subsidies do not rely on an ongoing source of subsidy funding: once a household receives the subsidy to connect to a network, for example, there is no longer a subsidy requirement in relation to access for that household. 28 Output-based performance risk: As discussed in Section 4.2, by shifting performance-risk to service providers, OBA requires service providers to plan and implement schemes to ensure that performance expectations are met in order to be fully compensated. More careful planning in terms of capacity (given lack of incentives to over-size with OBA) and final output delivery help enhance sustainability. Further, because demand risk of uptake is to a large extent shifted to service providers, service providers are taking more care in providing appropriate solutions to the targeted customers. By involving households and communities in the service expansion process, a greater sense of ownership can be developed, which in turn enhances sustainability. For example, the IDA-funded Senegal On-site Sanitation Project involves an extensive network of NGOs and community-based organizations (CBOs) working together to ensure not only demand for the on-site sanitation systems, but also proper use and maintenance. The project resulted in 63,000 household sanitation facilities built between 2003 and 2005 against the initial target of 60,000 and was two years ahead of schedule (OBA learning event presentation, June 21, 2007). 28 However, to enhance the sustainability of services, one-off OBA schemes may with-hold a portion of the subsidy disbursement after, for example, several months of bills paid and/or collected. Also, it is important to note that this assumes that the household can afford ongoing operations and maintenance costs for example in the form of tariffs. This is on the whole not a major hurdle. See targeting section above for more related discussion. 50

51 Setting tariffs and subsidy amounts The sustainability of any OBA scheme most specifically for network industries is dependent on the relationship between the subsidy provided and the tariff charged to consumers for ongoing service provision. For example, when the poor who are being connected through OBA schemes benefit from subsidized social tariffs, this can create an additional burden on providers, who will in turn need to be able to cross-subsidize these new poor users with consumers that are not benefiting from social tariffs schemes. This has been the experience with connecting the urban poor to the water network in Morocco, and will likely be the experience with any scheme that connects many new customers benefiting from social tariff schemes. On the other hand, it is important to differentiate between costs that are covered by the tariff regime and those that are not. At the given tariff levels (including regulated connection charges), even if the utility is not recovering costs for the specific low income connections under consideration, it may be covering its costs for its regulated jurisdiction as a whole - i.e. the tariff as a whole may be set to ensure that the utility recovers efficient cost of serving its entire area including the low income area. Design of OBA project should consider this and in this instance not over-compensate the utility as a whole by providing an additional subsidy for this low income area. There are likely to also be costs associated with the connection that are not covered by the tariff, and these may be most appropriate for the OBA subsidy to focus on -- depending on consumer's ability and willingness to pay. These issues were encountered in the case of the OBA design for an electricity scheme for slum areas of Mumbai (see Annex II for more details). The distribution company in this case is obligated to provide a metered working connection to anyone that applies. The regulated connection charge covers these costs (and upstream costs of providing the connection). The costs of connections from the meter to the house and internal wiring are not addressed by the utility or the regulator and need to be faced fully by the slum dweller and thus are creating the bottleneck to take up. Once a subsidy is provided here, the distribution company had the incentive to carry out upstream investment without any need for subsidy support (as this is compensated for by the tariff regime). Further, the utility may already have made certain performance promises to the regulator (e.g. connecting low income communities) as part of the last tariff review, and it is important that the OBA design takes account of such regulatory obligations that the utility may have. The OBA project should not be helping the utility if it is already being compensated by the regulator to meet these obligations. The OBA design should consider what barriers may be affecting take up even after the regulator's/policymaker's efforts have been taken into account. For example, in the case of a multi-year tariffs regime set by the regulator, the utility may be in a position to earn extra returns from reducing technical and commercial losses (as the tariffs are set for the multi-year regulatory period and include the cost of losses). In such an instance, connections for low income consumers that help to reduce theft - and therefore reduce commercial losses - may increase returns to the utility. The utility may therefore have adequate commercial 51

52 incentive (and returns) to undertake a certain amount of connection activity in low income areas without subsidy support. Designers of an OBA project for supporting connections would need to be cognizant of this and the level of incentive the utility already faces to ensure that the subsidy payments to the utility do not over compensate it for connections made in order to design more economic and politically sustainable solutions Sustainability of funding source The sustainability of the funding source needs to be considered based on the subsidy design mechanism. For example, in the case of roads and health which rely on ongoing subsidy streams, the sustainability of any given scheme is dependent on a constant flow of subsidies. Road maintenance funds in developed countries may provide a certain degree of security and sustainability, but road maintenance funds in developing countries have a more mixed record. The health sector often does not have such an ear-marked subsidy pool and therefore relies mainly on budget made available by the government. This makes it more difficult to ring-fence the health budget allocated to performance-based schemes although this is a risk that also applies to traditional ways of providing health care. The ICT sector seems to have a more robust form of subsidy funding. In ICT, all of the methods levies, auctions, government budgets -- for funding USF have been reliable in providing the required funding. Levies on telecommunications operators revenues may be preferable since they may be seen as more reliable and sustainable from a private sector perspective, in contrast to government budget funding or spectrum auctions. However, although the funding source has generally been reliable in ICT, the problem has been with the political sustainability of the approach due to large sums being accumulated but not being disbursed in a timely manner. (See Annex II for more details.) Some of the solutions proposed for improving the efficiency of the UASF include appointing private entities to manage the UASF, as in the USA and Pakistan. Other solutions include Pay or Play there is usually no tendering, but rather an allocation of obligations among the service providers. Players are free to contract each other to fulfill their share of obligations, if paying someone else to do so makes economic sense. Another option considered is the Virtual Fund. No country has ever implemented a virtual UASF. The advantage seen in virtual funds is mainly that the actual money of operator levies does not need to move into - and then later out of - a fund to the recipient, eliminating the need for actual fund management, but this would simply be an accounting system that records each operator s annual UASF levy. Further, a scheme is only sustainable in terms of subsidy required if the subsidy amounts are reasonable, reflecting the cost and benefit to society. Efficiency gains through innovation and competition therefore contribute to sustainability. Nevertheless, subsidy per capita can vary significantly between regions and sectors, also depending on the scope of service to be subsidized. Connecting the poor to grid-based infrastructure services can be relatively cheap, where trunk infrastructure exists and only the connection itself needs to be subsidized. The same service in a similar geographic area can require much higher subsidies if distribution and production infrastructure needs to be subsidized. For example the subsidy for natural gas connections (including a gas heater), in Armenia is $160, while the connection subsidy to the electricity grid in densely populated rural Ethiopia with its relatively larger families is $6.25. Those figures compare to $115 per capita for the installation 52

53 of a solar home system in very remote areas of rural Bolivia. Similarly per capita subsidies for water projects range from $125 in Morocco, where subsidies include sanitation connections, to $6.2 in a project in Kampala that makes extensive use of public water points and shared yard taps. Those numbers, while different due to different project contexts, have all showed sufficient economic rate of returns in cases where economic rates of return information was analyzed for this review (where funding is coming from GPOBA) Enabling environment Any OBA scheme is only as sustainable as the environment within which it operates. A pilot, by definition, may seem to some extent to be isolated from broader sector reform issues in the shortterm, but for greater impact with scaling-up and mainstreaming, a supportive enabling environment is critical. The following aspects of an enabling environment seem to play a particularly important role in the success of OBA in any given region or sector, most of which are interrelated and many of which are common requirements for successful PPP in the sector: Extent of experience with the private sector in service provision. Market structure and experience with competitive processes to encourage efficiency. Regulatory or legal/contractual framework for the sector, including tariff setting and adjustment policies. Capacity of implementing agencies (e.g. to handle procurement and transaction processes, monitoring and verification, funds flow but also in relation to understanding and willingness to work with performance-based arrangements.). For example, output based grid and mini-grid projects cannot bypass the need for a strong regulatory regime or at least robust contractual (and supporting legal) arrangements. This could partially explain the limited use of OBA in grid and mini-grids as many of the developing countries have weak regulatory capacity to enforce/oversee provider performance and added to it, the costs of regulation in remote areas is very high (Reiche). At the same time, a light-handed approach to regulation may be required with respect to service standards: in order to provide services that the poor can afford, and for example to more effectively utilize self-selection targeting methods, service standards may sometimes need to be relaxed although always meeting a minimum standard (Baker and Tremolet). The importance of an appropriate legal and regulatory environment is also demonstrated through the discussion of tariff and regulation above. Without proper regulatory accounting, or at the minimum capacity to understand the issues at hand, tariff and subsidy policies could lead to misalignment of incentives and/or waste of limited resources. Less optimal and sustainable solutions are likely to result Conclusions on sustainability, tariffs and enabling environment It is too early to analyze whether OBA schemes have proven sustainable on the whole. There is no evidence to date to suggest that schemes that involve OBA subsidies are less sustainable than their input-based counterparts, and on the contrary, the nature of OBA subsidy design and shift- 53

54 ing of performance-risk to service providers provide the platform for relatively sustainable interventions. Evidence from ICT and some other sectors demonstrate that OBA can provide sustainable results if OBA schemes are linked to contracts with adequate provisions for ongoing service delivery and if sources of funding are both reliable and available. One-off subsidies provide a clear and discrete one-time (although possibly phased) subsidy requirement for access. But to be designed well, costs for use and maintenance (or in the case of utilities tariffs) should be affordable to the poor, and, tariffs charged by the service provider should cover all reasonable and efficient running costs. In the case where tariffs are lower than running costs, connecting new customers will result in operators incurring losses and having insufficient funds to maintain the system. In such cases, there would be no incentive for the service provider to serve customers that are charged below running costs, and thus service provision to these customers will not be sustainable. Unfortunately, given the lack of political will to raise tariffs to sustainable levels (even though the poorest segments of the population are often paying more for alternative services), one-off OBA subsidies can be sustainably applied in limited circumstances. Transition tariffs to help raise tariffs to cover running costs could be considered, but the political will to raise tariffs often comes into question, putting the transitional nature of the subsidy at risk. Thus the scale-up and mainstreaming of OBA in certain sectors such as grid-based water and energy would need to go hand in hand with tariff reform. In some cases, OBA schemes have brought these issues to the fore, as in the Uganda and Kenya water projects. Subsidy amounts, especially in the case of network industries, should be set in line with (regulated) tariff levels to ensure proper alignment of incentives as well as to ensure that service providers are not over-compensated. For example, due consideration should be given to costs already incorporated into the regulatory asset base. Ongoing subsidies help shift performance risk out further into the future as in the case of the roads schemes described in more detail in Annex II and therefore ensure a relatively more sustainable service of a specified quality. But ongoing subsidies require a clear source of continuous future funding. Road maintenance funds and ICT UASFs provide a relatively transparent and predictable source of funding. Both types of schemes have experienced difficulties, but lessons are being learned in each sector for example, to possibly short-cut the problems related to slow disbursement from current UASFs. The energy sector is also in some places moving to universal access type funds through rural energy funds. In all OBA schemes, capacity of the government to regulate and monitor will be decisive in the sustainability of the scheme. Realistic programs based on existing enabling environment should be considered in sectors or regions where there is less experience with OBA. For example, complex long-term concession contracts in the water sector (where there have been few if any PPP concessions since the early part of this decade) may be less realistic, and the 54

55 focus should be on interventions that transfer less risk to service providers or require less sophisticated regulatory and institutional arrangements. 4.6 Monitoring of results A forthcoming review by the World Bank s Independent Evaluation Group (IEG) highlights results monitoring as an area of concern (The World Bank, forthcoming): IEG s assessment of M&E quality, which started in 2006, rates the quality over the past two years as modest in three fifths of cases. Issues cited are indicators that are not clearly defined or not measurable. At the same time the review shows a clear correlation between the quality of M&E systems and overall project outcomes. The main reasons identified for the poor performance of M&E systems of many projects is a lack of staff incentives to focus on M&E and a culture of approval resulting from a drive to get projects launched. OBA schemes can help toward mitigating this as they require output verification prior to disbursement of subsidies. This helps internalize monitoring and makes it a key element in the project design process. However, the degree and quality of monitoring and verification varies from scheme to scheme, rather than from sector to sector. The effectiveness of monitoring is related to several factors: The definition of the output to be monitored, and the ease of measuring and verifying the delivery of the output. The availability of trained local consultants and engineers to effectively monitor the delivery of outputs: capacity becomes a greater issue the more complicated the outputs to be verified. The chance of capture of monitoring agents by service providers: capture is greater the smaller the verification agent and the more powerful the service provider. In some cases, projects may have to look outside of the country for a viable verification agent. The capacity of the local government institutions to accept and interpret the monitoring results and to use the results for intended purposes. Even if the verification agent does his job correctly and on time, is the government (if involved) able and appropriately incentivized to quickly and effectively process that information for timely subsidy disbursement? However, whether or not an OBA scheme will result in the appropriate output and disbursement information being provided for broader monitoring and evaluation purposes is dependent on the recording and information dissemination process. In the case of Bank OBA projects, this is dependent on the feedback loop between the governments or project implementation units on the ground and their relevant Bank task teams, who in turn would need to systematically record the data where it could be shared and evaluated. The experience through this review demonstrated that although GPOBA-funded projects required an output-oriented monitoring and information sharing system, many Bank-funded OBA projects did not involve such a feedback loop, sometimes because the OBA component of a Bank-funded project was only a small component of a much larger input-based program. 55

56 Best practice would also use the monitoring platform of OBA beyond the verification of outputs to check how outcomes and other aspects of service delivery are faring. But this is often the weakest link in the feedback loop between OBA planning, design and implementation. Although OBA internalizes the monitoring of outputs, the monitoring framework established is rarely used for purposes other than payment of subsidies; hence the benefit of the information collected for example to improve on quality of service is not always exploited in full. Or, in some cases, additional information is not collected for long-term gain due to short-term costs. All sectors might be able to draw some lessons from some performance-based roads schemes that are attempting to enable even users to report and monitor performance by making performance criteria easily measurable or discernable. Impact evaluations are a good way to supplement verification activities. Further, impact evaluations comparing output-based and input-based approaches should also be considered; a few such impact evaluations are being attempted. These enhancements to monitoring and verification under OBA schemes require capacity and resources. Verification agents must be appropriately trained, and also incentivized. GPOBA uses for the most part third party verification agents to help enhance transparency and also improve on effectiveness so that outputs can be verified relatively quickly and the disbursements can be made. The key is to balance the independence of output verification with the broader sector monitoring needs, as well as ensure ownership on the part of the regulatory or relevant government agency. A recent case of over-reporting of outputs achieved and resulting in excess performance payments under performance contracts between GAVI and several recipient countries provides an example that without proper monitoring, the advantages of results-based schemes are lost. An emphasis should be placed on independent verification agents, who are also experts in the sector. With OBA schemes, accountability also increases for donors and governments: public funding is linked to the fact that pre-identified outputs are (were) delivered, and therefore waste or inappropriate allocation should be minimized. Thus OBA can play an important role in the fight to improve governance and reduce corruption. OBA projects should make full use of the requirement to monitor for outputs, through physical audit, surveys of beneficiaries, and oversight by civil society. And to support broad monitoring, OBA projects should include an active communication strategy that advertises what services are to be delivered to whom and at what price as is currently being done in Morocco s urban water sector. But it is noted that while project-specific outputs may be easy to monitor, compliance with general regulations or laws governing the sector may be less transparent. Detecting poor construction or below-standard delivery can be difficult. This requires resources and capacity, and to some extent a relaxation of the completely pure OBA since some key inputs will need to be specified, and the ability to monitor them will be required. 56

57 5. Key Considerations and Next Steps The section summarizes the challenges that OBA faces in terms of scaling-up or mainstreaming, and recommends possible responses. The first section focuses on challenges external to the WBG, and the second focuses on internal challenges. 5.1 External challenges and possible responses by the WBG The piloting phase of OBA has in general been a success. About 128 OBA projects have been implemented or are underway in the WBG. These projects are expected to reach at a minimum almost 60 million beneficiaries worldwide. 29 The pilots in the Chilean ICT sector and the Argentine roads sector in the 1990s led to scale-up in their respective countries and eventual replication worldwide. It is now safe to say that OBA has become mainstreamed as one of the key modus operandi for interventions in these two sectors in most parts of the world 30. In the health and off-grid energy sectors, OBA is recognized as one of the key financing mechanisms to expand targeted access to the poor and is being widely used. In the water, education, and grid-based energy sectors, OBA is still in the pilot stage. The lessons learned regarding the use of OBA compared to traditional approaches to contracting have demonstrated that there are clear advantages of an OBA approach in regards to efficiently targeting subsidies and mobilizing the private sector to serve poor households that would otherwise go without an improved service. OBA has also demonstrated that monitoring for results is possible if appropriate systems for capturing and transmitting systems are put in place. At the same time, the review recognizes that OBA is not a substitute for sector reform. The experience of government contracting with the private sector, and to some extent the existence of legal or regulatory practices that are more supportive of private sector risk-taking, are part of the environment that enables OBA to be more successful in some contexts than others. In turn, OBA is one of the main mechanisms through which efficiency gains from sector reform have been shared with users through improved access and improved standards of service. OBA therefore not only relies on a supporting enabling environment but can itself help underscore the benefits. Cross-cutting challenges for successful OBA schemes remain even for the sectors where OBA is mainstreamed. Although some of these challenges are not specific to OBA, in order for OBA to continue to have an impact, they must be addressed. 29 Data on the number of beneficiaries is not readily available for public access ICT and transport projects. Beneficiary information is particularly limited in the case of transport sector (available only in two of the 23 World Bank implemented projects) due to the non-exclusivity of roads projects. 30 Although the recent Regulatel study demonstrates that ICT technology evolves so fast, there is a constant need for innovation to keep up with the market (Stern and Townsend). 57

58 Access to finance: Security and sustainability of funding: Capacity Access to finance can be a hurdle for OBA schemes in all sectors even in ICT. The current financial crisis is likely to exacerbate this problem. Access to finance can present a hurdle for both providers and users, resulting in the following challenges: a) difficulty in shifting sufficient performance risk to service providers under an OBA arrangement if the cost of pre-financing the outputs would put an undue burden on the provider and/or if the resulting fees to users (e.g., tariffs) would be unaffordable; b) limitations on the development of a vibrant private sector that can afford to take risks and invest in business expansion, even with the availability of targeted subsidies to help defray the costs. A few guarantees and lines of credit to the banking sector are being tested. Where such financing instruments are not as readily available, OBA schemes may need to phase-in payments against reasonable milestones as long as performance risk for output delivery for the most part remains with the service provider. A secure source of funding and an administrative framework that allows for swift disbursement when outputs have been achieved are required. Further, for the sustainability of OBA in any given sector, the source of funding also needs to be reliable and consistent. Lessons are being learned benefits and challenges from on-going funds (e.g. ICT, roads, energy). Capacity to implement and monitor OBA schemes is limited, particularly in those countries/sectors where OBA is needed most although some limitations are not OBA-specific. Capacity limitations are most obvious in regards to transaction design and implementation, output monitoring and verification implementation, and demand management. Targeted training, hiring of independent verification agents, involvement of NGOs, and private administrators to manage universal access funds are all part of solutions being implemented to mitigate capacity constraints. envi- Enabling ronment: As with most PPP arrangements, in order for OBA schemes to be sustainable and of scale, there should be basic elements of an environment that supports the development, monitoring and adjustment of contracts, which often is demonstrated by a successful track record of private sector participation. Relatively transparent legal or regulatory arrangements for example for tariff setting and adjustment would be amongst such enabling factors, and ultimately aim to reduce perceived risk to providers and to some extent the cost to final users. 58

59 Targeting: Competing interests: As the size of pilots and programs increases, and as technologies change, more refined targeting mechanisms will be required. These can be costly to administer and require additional capacity, but these costs may outweigh the leakage from large geographically targeted schemes. A blend of targeting mechanisms is proving effective. Ultimately, all else being equal, service providers will opt for inputbased schemes because they transfer less risk to the provider. But donors and governments should want the contrary, i.e. to transfer risk to the provider to hold them accountable (as long as providers are equipped to take on that risk). If OBA is to be the mechanism chosen for a given intervention, donor/government coordination is required. There is clearly a role for the WBG in enhancing the effectiveness of OBA schemes to improve the reach of basic services to the poor. For example, the Bank and IFC should work together to provide financial solutions to help mitigate the access-to-finance constraint by encouraging banks to improve lending conditions to service providers both for pre-financing of outputs and for longer-term project finance. Capacity building and technical assistance (e.g., for transaction support, tariff design and subsidy policy, monitoring and evaluation) can be provided by the WBG and the multidonor programs such as GPOBA, PPIAF, ESMAP, and WSP. Donor funding and discipline are key to secure and sustain sources of funding for OBA. The Bank is well placed to transfer lessons from county to country and between sectors for example between universal access and service funds in ICT, road maintenance funds, and rural electrification funds. This will help OBA practitioners across sectors and regions to benefit from the lessons learned over the past decade, while tailoring them to specific contexts. The Bank should continue its important work on regulatory reform for example, through the newly created Regulatory Thematic Group and on pushing the agenda for sustainable tariffs and subsidy policies that are pro-poor. OBA can help provide the starting point for these discussions in some cases. The WBG also plays a pivotal role in donor coordination through its work on sector-wide approaches (SWAPs) and development policy lending (DPL). Mainstreaming OBA would require working closely with WBG clients to better understand how performancebased arrangements such as OBA could best work in their specific context, and would also require close co-ordination with other donor initiatives to ensure additionality and harmonization. Finally, GPOBA needs to step up its efforts to facilitate sharing of experiences and best practices in OBA, and to provide WBG staff and other development partners with the 59

60 practical knowledge they need to assess when OBA is suitable and to design and implement OBA schemes. This is in line with GPOBA s strategy to evolve within the coming two to four years from providing subsidy funding to acting primarily as a center of OBA expertise, which includes activities such as training events for WBG staff, development of online resources and an E-learning course on OBA, and development of a diagnostic tool which would provide more guidance to projects teams on issues such as project design and relevant characteristics of an enabling environment for OBA schemes. By taking these steps, the WBG can help make OBA more effective where it is already being used more regularly i.e. the ICT, roads, health, and off-grid energy sectors. These actions can also help provide a stronger platform in those sectors such as grid-based energy, water, and education where OBA is still in the pilot phase but in some cases ready for scale-up as important lessons for improved design and implementation are already available. 5.2 Internal challenges and possible responses by WBG Although mainstreaming implies that OBA is one of the key modus operandi for interventions in the sector, effective mainstreaming in the WBG would imply that policy decisions in the WBG have been made and/or that operational guidelines have been in place to help and encourage Bank staff to actively design and develop OBA schemes in the sector. In order for Bank staff to seriously consider and implement OBA schemes on a larger scale, policy and strategy decisions need to be taken by senior management. Some steps have already been taken: The SDN Sustainable Infrastructure Action Plan, which has been endorsed by the Board, states that the WBG should pilot and ramp up innovative financing options for the private sector, including OBA (SIAP, 2008). The CFP-led Innovating Development Finance paper (Girishankar, 2009) discusses the role of OBA within the context of results-based financing, and endorses mainstreaming OBA in the WBG. The forthcoming PSD Strategy Update, 2009, put together by FPD, endorses mainstreaming OBA in the WBG. The 2007 World Bank strategy for Health Nutrition and Population calls for an increase in the "proportion of output-based lending in health. But more needs to be done to translate these strategies into policies on the ground, for example by incorporating OBA into Country Assistance Strategies. Further, although OBA is used extensively within Standard Investment Loans (SIL), there is limited experience of OBA as part of other instruments. For example, the Morocco Urban Water OBA project (funded by GPOBA) and the first sector reform DPL funded by IBRD together helped galvanize the government to reach for OBA-type approaches: the Government of Morocco has expressed interest in requesting a Bank loan to scale-up OBA in the sector. Output-based schemes that involve performance-based intergovernmental transfers are being piloted in Latin America with the support of IBRD loans (e.g., Guanajuato, Mexico) and an 60

61 OBA facility is being piloted in the water sector in Honduras. These are important steps to take OBA from project intervention to wider sector programs. The nexus between OBA and Bank fiduciary operations and policies needs to be considered further. Here too important steps have been taken. For example, in 2003 the Bank adopted a Sample Bidding Document for OBA road contracts and in 2008, GPOBA and OPCS released a joint Guidance Note to staff on procurement procedures and OBA 31. But more needs to be done. Bank staff should be given incentives to innovate and greater emphasis should be placed on disbursing for outputs as opposed to inputs, including more amenable fiduciary procedures. The Investment Lending Reform currently being considered by OPCS will be tackling these very issues (OPCS, 2009). OBA will provide important lessons for this initiative as well. The ultimate decision on the success of OBA or any other aid effectiveness tool to improve access to the poor and enhance accountability rests with the WBG client governments, and their interest, ownership and commitment to design and sustain such approaches. The WBG has a key role to play in demonstrating that OBA, a key part of the results-based financing approach, can help improve access to basic services and reach the MDGs. The WBG also has a role in ensuring that the donor community speaks with one common voice on these issues related transparency and efficient use of resources to reach the poorest. The existing information and expertise on OBA, including this review, provide a solid underpinning for the successful design of pilots or programs that respond to client needs

62 Annex 1: Table of OBA Projects in the WBG (as of April 30, 2009) Project Name (P0 Number) 1 OBA Telecommunications. (P081250) 2 Competitiveness & Enterprise Development Project (P071443) 3 Rural Telecommunication Access Project (P102475) 4 Infrastructure for Territorial Development (P076807) 5 Power and Communications Sectors Modernization and Rural Services Project PROMEC (P063644) 6 OBA and Regulatory Frameworks for Rural and Peri-Urban Telecommunications (P094321) 7 Rural Telecommunications Development (P093925) 8 Extending Telecommunications in Rural Indonesia (P102476) 9 Rural Infrastructure Services Project (P057761) 10 OBA Pilot Project of Universal Access Strategy (P102488) 11 Information and Communications Infrastructure Development Project (P092965) 12 Mozambique Communication Sector Reform Project (P073479) 13 Telecommunications Sector Reform Project (P050671) 14 Telecommunications Reform Project (P055853) 15 Rural Telecommunications project (P089989) 16 Privatization Support Project (ICT) (P070293) 17 Telecommunications and Postal Sector Reform Project (P075739) 18 Telecommunications and ICT Development Project (P088448) 19 Energy for Rural Transformation Project (P069996) 20 Increased Access to Electricity and ICT Services Project (P077452) Country (Region) Funding Source Type of output World Bank Subsidy amount including GPOBA 32 Planned Number of beneficiaries Telecommunications Azerbaijan (ECA) TBC TBC TBC TBC Design Burkina Faso (AFR) IDA Payphones and POPs Project Status (Latest information on actual # Beneficiaries) 33 $1,039, ,146 Implementation Cambodia (EAP) GPOBA Beneficiaries $2,500, ,000 Implementation Chile (LCR) Ecuador (LCR) TBC Public Phones TBC TBC Implementation IBRD Payphones and Internet Stations $4,150,000 TBC Closed (0 34 ) Guatemala (LCR) IBRD Public Phones $16,000,000 3,500,000 Implementation India IDA Public Phones & TBC TBC Design (SAR) Internet Indonesia (EAP) GPOBA Direct Users $1,868, ,000 Implementation Malawi (AFR) Mongolia (EAP) Mongolia (EAP) Mozambique (AFR) IDA GPOBA IDA IDA Pay phones, telecenters, and Internet Stations Pay phones, telecenters, and Internet Stations Pay phones, telecenters, and Internet Stations Payphones and Internet PoPs $1,500,000 TBC Implementation $257,335 21,312 Closed (22,315) 35 $5,450,000 45,000 Implementation $3,000,000 2,600,000 Implementation Nepal IDA Access lines $11,900,000 4,000,000 Implementation (SAR) Nicaragua (LCR) IDA Payphones $900, ,000 Closed Nicaragua (LCR) IDA Payphones and Internet PoPs Nigeria IDA Payphones and (AFR) Internet Stations Samoa TBC Payphones and (EAP) Internet Stations St Lucia (LCR) Uganda (AFR) Zambia (AFR) IDA/ IBRD IDA IDA Access lines & Internet Internet POPs and Public phones Public telephones, POPs, Access Center, IXP $7,900, ,000 Implementation $6,130,000 2,657,422 Implementation TBC TBC Implementation $1,000,000 TBC Implementation $6,695,981 3,600,000 Implementation $3,125,000 TBC Implementation 32 This amount does not include government subsidy contribution, which sums up to nearly $1.7 billion. 33 Latest information on actual # of beneficiaries are only available for the projects that have received either technical assistance and/or investment subsidy funding from GPOBA, as well as for a few other WBG projects. 34 According to the implementation completion reports (ICR), the OBA component of the project was cancelled in June The operator failed to meet the technical specifications and comply with the deadlines for installation of the telecenters. The Project had paid $1 million of the $4.15 million committed as advances and for the reported installation of half of the telecenters. None of telecenters are in operation. 35 For Mongolia Telecom, the number of final beneficiaries increased due to competitive bidding. 62

63 Project Name (P0 Number) 1 Road Maintenance and Sector Rehabilitation Project (P006003) 2 National Highway Asset management (P088153) 3 Provincial Road Infrastructure Project (P070628) 4 National Highway Rehabilitation and Maintenance (P052590) 5 Federal Highways Project (P006532) 6 Rio Grande Do Sul Highway Management Project (P034578) 7 Transport Sector Project (P074030) 8 OBA in Road Network Management and Maintenance 2 (P087004) 9 Chad - National Transport Program Support Project (P035672) 10 Chad - Road Network Management & Maintenance (P079736) 11 India - Annuity Road Projects (TBC) 12 Kenya Northern Corridor Improvement Project (P082615) 13 Madagascar - Transport Infrastructure Investment Project (P082806) 14 Nigeria - Federal Roads Development Project (P090135) 15 Paraguay - Road Maintenance Project (P082026) 16 Uruguay - Transport Project II (P049267) 17 Second Rural Access Project (P085231) 18 Transport Sector Support Program (P055120) 19 Tanzania - Road Network Management & Maintenance P (GPOBA); P (Bank) 20 Peru - Regional Transport Infrastructure Decentralization Project (Provias Descentralizado) (P078813) 21 Peru - Second Rural Roads Project (P044601) 22 Peru - Rural Road Rehabilitation and Maintenance Project (P037047) 23 Serbia & Montenegro - Transport Rehabilitation Project (P075207) Country (Region) Argentina (LCR) Argentina (LCR) Argentina (LCR) Argentina (LCR) Brazil (LCR) Brazil (LCR) Burkina Faso (AFR) Cape Verde (AFR) Chad (AFR) Chad (AFR) India (SAR) Kenya (AFR) Madagascar (AFR) Funding Source IBRD & Govt Type of output World Bank Subsidy amount including GPOBA 32 Planned Number of beneficiaries Transport 11,667 km $248,300,000 N/A Closed Project Status (Latest information on actual # Beneficiaries) 33 IBRD & Govt 8,188 km $182,800,000 N/A Implementation IBRD & 2,204 km $96,400,000 N/A Implementation Govt IBRD & 19,885 km $295,775,000 N/A Closed Govt TBC 3,500 km 36 $247,700,000 N/A Closed IBRD & 2,200 km $70,000,000 N/A Closed Govt IDA 1,021 km TBC N/A Implementation IDA 225 km $6,900,000 N/A Implementation IDA 440 km $11,088,000 N/A Closed IDA 600 km $24,000,000 N/A Implementation TBC TBC TBC N/A Implementation IDA,ND 300 km $207,000,000 N/A Implementation F, Govt IDA TBC $27,600,000 N/A Implementation Nigeria IDA 1,800 km $330,000,000 N/A Implementation (AFR) Paraguay (LCR) IBRD 968 km $39,270,000 N/A Implementation Uruguay (LCR) Yemen (MNA) Tanzania (AFR) Tanzania (AFR) Peru (LCR) Peru (LCR) Peru (LCR) Serbia & Montenegro (ECA) IBRD & Govt 856 km $42,237,000 N/A Closed IDA & 950 km $40,000, ,000 Implementation Govt IDA $708 $3,000,000 TBC Design IDA 850 km $26,000,000 TBC Implementation IBRD & IADB & Govt IBRD & IADB & Govt IBRD & IADB & Govt IDA & Govt 4,906 km $50,000,000 TBC Implementation 14,950 km $23,154,000 3,500,000 Closed (3,500,000) 10,881 km $7,390,000 1,500,000 Closed (3,500,000) 1,200 km $55,000,000 TBC Implementation 36 In effect, over the period, about 4,000 km of rehabilitation works on the federal road network were undertaken under the World Bank funding, out of which about 3,500 km with a result-based approach. Additional 3,000 km of rehabilitation works were financed by the IADB loans, and about 1,300 by the Federal Government itself. 63

64 Project Name (P0 Number) 1 Renewable Energy in the Rural Market Project (P006043) 2 Access to Gas & Heat Supply for Poor Urban Households (P103071) 3 Heating and Gas (IDA Project) (P095329) 4 Rural Electrification and Renewable Energy Development (IDCOL SHS) (P071794) 5 Decentralized Infrastructure for Rural Transformation (P073367) 6 Bolivia Rural Access with Small- Scale Providers (P102479) 7 Rural Electrification and Transmission (P071591, P064844) 8 Renewable Energy Development (P046829) 9 Natural Gas Distribution for Low Income Families in the Caribbean Coast (P102095) Country (Region) Funding Source Type of output Energy Argentina (LCR) IBRD Household Connections Armenia (ECA) GPOBA Individual Gas connections Armenia (ECA) Bangladesh (SAR) Bolivia (LCR) Bolivia (LCR) World Bank Subsidy amount including GPOBA 32 Planned Number of beneficiaries Project Status (Latest information on actual # Beneficiaries) 33 $30,600, ,000 Implementation (48,000) $3,100,000 18,676 Implementation (11,120) IDA, Govt. Individual Gas connections $3,000,000 21,924 Implementation (7,000) GEF, SHS installations $8,200, ,960 Implementation IDA, (1,800,000) ADB, IDB, KfW, GTZ IDA SHS installations $10,000, ,746 Implementation (30,776) GPOBA SHS and Pico PV systems Cambodia (EAP) GEF Household Connections $5,175,000 45,000 Implementation $5,600, ,200 Implementation China (EAP) GEF Photovoltaic SHS $27,000,000 1,600,000 Closed (1,600,000) Colombia (LCR) GPOBA Household gas $5,085, , Closed connection and a (210,000) 38 gas stove 10 Rural Energy Access (P105651) Ethiopia (AFR) GPOBA Household Connections $8,000,000 1,142,857 Implementation 11 Solar PV Systems to Increase Ghana GPOBA SHS installations $4,350,000 90,000 Implementation Access to Electricity Services in (AFR) Ghana (P105617) 12 Rural Electrification Program (P112651) 13 Improved Electricity Access for Indian Slum Dwellers (P104649) 14 Home Solar Systems Project (P035544) 15 Southern Provinces Rural Electrification Project (P044973) 16 Liberia Electricity Access (P110723) 17 Household Energy and Universal Access Project (P073036) 18 Biogas Support Programme (P103979) Guatemala (LCR) GPOBA Individual Household connections $6,850, ,195 Design India GPOBA Household Connections $1,570, ,000 Design (SAR) Indonesia (EAP) GEF SHS installations 5,200,000 35,438 Closed Lao PDR (EAP) IDA Household Connections Liberia (AFR) Mali (AFR) Nepal (SAR) GPOBA IDA, GEF DGIS, KfW, GPOBA $1,000,000 50,000 Closed (51,805) Household Connections $5,000,000 91,241 Design Household Connections $19,300, ,700 Implementation and SHS (178,685) installations Biogas plants $5,000, ,000 Implementation (33,404) 19 Off-grid Rural Electrification (Perza) (P073246) Nicaragua (LCR) IDA Household Connections $1,850,000 42,000 Implementation (46,445) 20 Rural Power Project (P066397) Philippines GEF SHS installations $1,800,000 50,000 Implementation (EAP) 21 Rural non-grid Power Supply (P090238) Philippines (EAP) Local Govt. TA from GPOBA Electricity supplied (kwh) N/A 360,000 Implementation 37 The verified connections in Colombia serve 204,852 beneficiaries. The remaining 5,000 beneficiaries were connected but the connections were not verified as they were made after the deadline. 38 Includes beneficiaries from unverified connections 64

65 Project Name (P0 Number) 22 Electricity Services for Rural Areas Project (P085708) 23 Renewable Energy for Rural Economic Development (P076702) 24 Energy Services Delivery Project (P010498) Senegal (AFR) Country (Region) Funding Source IDA, GEF, AfDB, KfW Type of output Energy Household connections World Bank Subsidy amount including GPOBA 32 Planned Number of beneficiaries Project Status (Latest information on actual # Beneficiaries) 33 $18,000, ,622 Implementation Sri Lanka (SAR) GEF SHS installations $3,900, ,000 Implementation (500,000) Sri Lanka (SAR) IDA SHS installations $5,700,000 75,000 Closed (104,765) 25 Pamir Private Power Project (P075256) 26 Energy Development and Access project (TEDAP) (P101645) 27 Energy for Rural Transformation Phase I (P069996) 1 Multi-sector Project for Basic Services in Rural Areas (P053578) 2 Design of Innovative OBD Schemes for Water Supply and Sanitation Projects in Two Brazilian States (P114151) 3 Water Affermage contract - OBA for coverage expansion (P104526) 4 Access to Water for low income communities (P109102) 5 Second Water Supply Project (P001044) 6 Extension of Water Services (P102474) 7 Improved Rural Community water in Andhra Pradesh (P102472) 8 Jakarta PT Thames/Suez (P102529) 9 Expanding Piped Water Supply to Surabaya s Urban Poor (P105590) 10 Micro-finance for small water schemes (P104075) 11 Extension of Water and Sanitation in Low Income Areas in Kisumu (P098285) 12 Small Towns Water Supply (P099575) 13 Guanajuato Water Project (TBC) 14 National OBA Facility for Wastewater Sector (P111610) Tajikistan (ECA) IDA Electricity consumed (in kwh) Tanzania (AFR) GEF Rural Household Connections Uganda (AFR) Bangladesh (SAR) Brazil (LCR) GEF SHS Installations & Institutional systems Water and Sanitation IDA Water connections TBC Water connections Cameroon (AFR) GPOBA Water connections Ethiopia (AFR) GPOBA Water connections Guinea IDA Water connections (AFR) Honduras (LCR) GPOBA Water connections India GPOBA HH Water Connections (SAR) Indonesia (EAP) GPOBA Household water connections Indonesia (EAP) GPOBA Water connections and Master Meter installation Kenya GPOBA, HH Water Connections (AFR) EU Kenya GPOBA Water Connections (AFR) Lao PDR (EAP) Mexico (LCR) Mexico (LCR) GPOBA Water Connections Water and Sanitation IBRD water Connections TBC Water Connections $4,000, ,126 Implementation (178,126) $2,300,000 75,000 Implementation $1,400,000 37,500 Implementation (18,330) $314,743 26,000 Implementation (5,510) TBC TBC Design $5,250, ,000 Implementation (4,302) 39 $8,400, ,000 Design $16,900, ,000 Closed (138,000) $4,440, ,000 Implementation $850,000 75,000 Implementation (43,422) 40 $2,573,140 55,824 Implementation (15,955) 41 $2,407,500 77,500 Implementation $1,151,300 60,000 Implementation (2,232) 42 $350,000 72,000 Design $2,350, ,000 Design $38,006,000 90, TBC TBC Design 39 Includes beneficiaries from unverified connections 40 Includes beneficiaries from unverified connections 41 Includes beneficiaries from unverified connections 42 Includes beneficiaries from unverified connections 65

66 Project Name (P0 Number) 15 Rural Water Supply and Sanitation Project (P086877) 16 Urban Water and Sanitation (P102527) 17 Water Private Sector Contracts - OBA for coverage expansion (P104945) 18 Second National Urban Water Sector Reform Project (P071391) 19 Fourth Rural Water Supply and Sanitation Project (P039983) 20 National Project for Rural Water and Sanitation (P065256) Country (Region) Morocco (MNA) Morocco (MNA) Mozambique (AFR) Nigeria (AFR) Funding Source Type of output Water and Sanitation TA from Water Connections and Flush GPOBA Latrines GPOBA Water Connections GPOBA Household yard taps IDA Household Water Connections Paraguay (LCR) IBRD Household Water Connections Peru IBRD Household Water (LCR) Connections World Bank Subsidy amount including GPOBA 32 Planned Number of beneficiaries N/A 51,840 Design Project Status (Latest information on actual # Beneficiaries) 33 $7,000,000 55,704 Implementation (14,475) 43 $6,000, ,000 Implementation $13,350, ,000 Implementation $834,880 27,625 Closed $2,500,000 TBC Implementation 21 LGU Urban Water and Sanitation Project APL2 (P069491) Philippines (EAP) IBRD Household Water Connections 22 Manila Water (P106775) Philippines (EAP) GPOBA Water Connections 23 On-Site Sanitation Project Senegal GPOBA DP, Septic Tank, (P095587) (AFR) TCM, BALP 24 On-Site Sanitation Project (IDA Senegal IDA Household sanitation Project) (P041528) (AFR) connections 25 Colombo Wastewater Sri Lanka (SAR) GPOBA sewer connections (P111161) 26 National Water Sector Fund St. Lucia GPOBA Sanitation connections (P104335) (LCR) 27 Water Supply in Secondary Tanzania GPOBA Household Water Towns (P097290) (AFR) Connections 28 OBA in Water Supply in Uganda's Uganda GPOBA Public Water Small Towns and Rural (AFR) Points and HH Growth Centers (P102462) Yard Taps 29 Kampala - Water Connections for the Poor (P104943) Uganda (AFR) 30 Rural Water (EMW) (P104528) Vietnam (EAP) 31 Service Expansion and Water Loss Reduction (P106450) 32 Al Qabel Village Water Supply (P111757) 1 Health System Emergency Reconstruction and Development - Supplement (P098358) 2 Provincial Maternal-Child Health Invest (P072637) 3 Provincial Maternal-Child Health Invest. Loan - Phase II (P095515) 4 Contractual Approaches for Improving Health Services Delivery (P088751) Vietnam (EAP) Yemen (MNA) Afghanistan (SAR) Argentina (LCR) Argentina (LCR) GPOBA GPOBA GPOBA GPOBA IDA IBRD, Govt IBRD, Govt Public Water Points and Yard Taps Working House connection to network Individual Household water Connections Household sanitation connections Health Medical treatments Medical treatments for mothers and children Medical treatments for mothers and children Congo, DR (AFR) IDA Medical treatments $2,300,000 TBC Closed $2,900, ,463 Implementation (26,372) $5,764, ,900 Implementation (3,267) 44 $28,000, ,000 Implementation (567,000) $1,100,000 35,000 Design $1,600,000 25,600 Design $7,000, ,000 Design $3,169,001 55,511 Implementation $2,527, ,050 Implementation (9,150) 45 $3,000, ,000 Implementation (10,904) $7,745, ,561 Implementation $1,400,000 15,000 Implementation $30,000,000 2,250,000 Implementation $90,400, ,292 Closed (527,305) $277,400,000 1,700,000 Implementation (388,188) $5,000,000 1,500,000 Implementation 43 Includes beneficiaries from unverified connections 44 Includes beneficiaries from unverified connections 45 Includes beneficiaries from unverified connections 66

67 Project Name (P0 Number) 5 Health Zone Project: Health Zone Administration and Facilities Contracting Component (P057296) 6 Rajasthan Health Systems Development Project (P050655) 7 Lesotho New Hospital PPP (P104403) 8 Pre-paid Health Scheme Pilot in Nigeria (P104405) 9 Poverty Reduction Support Credit - PRSC (P078806) 10 Poverty Reduction Support Credit - PRSC II (P090690) 11 Mother and Child Basic Health Insurance Project (P082056) 12 Comparison of OBA Health Schemes (P092944) 13 Poverty Reduction Support Credit I (P085192) 14 Poverty Reduction Support Credit III (P098129) 15 Poverty Reduction Support Credit IV (P104990) 16 Health Sector Development Program (P058627) 17 Health Sector Development Project II - HSDP II (P105093) 18 Reproductive Health Vouchers in Western Uganda (P104527) 19 Mekong Regional Health Support Project (P079663) 20 Northern Upland Health Support Project (P082672) 21 Health Support to the Poor of the Northern Upland (P110251) 22 Safe Motherhood Program (P104946) 1 Female Secondary School Assistance Project - FSSAP I (P009555) 2 Female Secondary School Assistance Project - FSSAP II (P044876) 3 Lifelong Learning and Training Project (P068271) 4 Balochistan Education Support Project (P094086) Country (Region) Funding Source Type of output Health Congo, DR (AFR) IDA Immunization coverage India (SAR) IDA, Govt Medical treatments Lesotho GPOBA Medical treatments (AFR) Nigeria GPOBA Medical treatments (AFR) Pakistan IDA Medical treatments (SAR) Pakistan IDA Medical treatments (SAR) Paraguay (LCR) IBRD Mother-Baby Treatment Package Rwanda IDA, Medical treatments (AFR) Govt Rwanda IDA. Medical treatments (AFR) Govt Rwanda IDA, Medical treatments (AFR) Govt Rwanda IDA, Medical treatments (AFR) Govt Tanzania IDA Medical treatments (AFR) for mater- nal and childhood illnesses Tanzania (AFR) Uganda (AFR) Vietnam (EAP) Vietnam (EAP) Vietnam (EAP) Yemen (MNA) Bangladesh (SAR) Bangladesh (SAR) Chile (LCR) Pakistan (SAR) IDA GPOBA, KfW Medical treatments to prevent Malaria STD and Safe Delivery Vouchers World Bank Subsidy amount including GPOBA 32 Planned Number of beneficiaries Project Status (Latest information on actual # Beneficiaries) 33 $5,000,000 10,000,000 Implementation $89,000,000 3,034,000 Implementation $6,250, ,000 Implementation $6,015,165 22,500 Implementation TBC TBC Closed TBC TBC Closed $7,304, ,000 Implementation $3,600,000 1,070,000 Closed $13,000,000 TBC Closed $8,250,000 TBC Closed $8,400,000 TBC Closed $20,000,000 TBC Closed $65,000,000 TBC Implementation $4,300, ,912 Implementation (65)** IDA Health Insurance $8,000,000 TBC Implementation IDA IDA GPOBA IDA, Govt IDA, Govt IBRD, Govt IDA, Govt Medical treatments Medical treatments Mother-Baby Treatment Package Education Female Students Enrolled Female Students Enrolled $10,000,000 TBC Implementation $14,140,000 TBC Implementation $6,232,100 80,000 Implementation $68,100,000 1,600,000 Closed (1,540,000) $67,807,143 1,450,000 Closed (1,200,000 Students Enrolled $41,140, ,874 Implementation Students Enrolled $2,100,000 34,500 Implementation 67

68 Annex II: Sector Specific Lessons This Annex provides a detailed description of OBA in the following sectors: A. ICT B. Transport (mainly Roads sub-sector) C. Energy D. Water E. Health F. Education A. ICT OBA is largely mainstreamed in the information and communication technology (ICT) sector, where universal access and service funds rely on explicit subsidies from wealthier largely urban populations to help extend access to less wealthy and more costly to serve (usually) rural populations on a performance-basis. The subsidy is often determined by having private companies bid for roll-out of infrastructure and services on the lowest subsidy required. The subsidy is disbursed on outputs, or milestones, such as the installation of functioning hardware (pay phones, tele-centers or internet points of presence, amongst others), and in some cases continued service provision for a specified period of time. A.1. Universe of OBA in ICT Before countries liberalized their ICT sectors, public (monopoly) utilities generally had limited success in expanding ICT services. Cross-subsidy schemes were usually not successful, in part because state-owned enterprises were often not able to charge high enough tariffs to wealthier or less-costly to serve customers to help finance the cost of extension to more costly areas. Therefore access to the poor was limited. But since the mid-1990s, a substantial change has taken place in the sector. The sector went from largely monopolistic service provision to competition, which had a major impact on expanding access. In this time, output-based contracts for the provision of ICT services became an effective means of expanding ICT services to the poor (Stern and Townsend 2007). The first of these contracts was funded by Chile s Fund for the Telecommunications Development (FDT) in FDT has since funded the installation of more than 25,000 payphones in about 8,000 rural centers benefiting close to 2.7 million people. It is estimated that there are less than 150,000 people (1% of the population) without access to a basic telephone in Chile (Stern and Townsend 2007, p. 92). Chile s FDT was followed by other similar schemes throughout Latin America. In 2001, Peru was the first country to develop contracts for the provision of private internet connections. As in the case of OBA in roads (see next sub-section), the Latin America & Caribbean (LCR) region has been the forerunner in contracts for all ICT technologies that are represented and has continued using such contracts. Sub-Saharan Africa implemented its first ICT OBA contracts in 2002, South Asia in 2004 and East Asia & Pacific in Table 5 below presents the universe of OBA in the ICT sector. 68

69 Table 5: Universe of Output-Based Aid in ICT Public phones Tele-centers (facilities that offer use of ICTs in a publicly shared manner, with or without a fee) Internet Cellular network Private phone connections Common Outputs : Serviceable Asset Provision of public phone connections (examples: Chile, Indonesia, Guatemala, Mongolia, Uganda, Cambodia, India, Malawi, Nepal, Nigeria, Zambia) Provision of tele-centers (examples: Brazil, Chile, Indonesia, Mongolia) Internet coverage through points of presence (PoPs) and private connections (examples: Uganda, Organization of Eastern Caribbean States) Provision of cellular networks (examples: Bolivia, Mongolia) Provision of private phone connections (examples: Uganda, Pakistan) The review has identified about 20 World Bank projects 46 and 19 projects outside the World Bank that can be considered OBA in the ICT sector. The 16 projects for which there is information on contract value, all of which are World Bank-funded, have a total value of $73.4 million (figure 12). The 12 projects for which information on the number of expected beneficiaries is available (total volume $63.6 million) are expected to reach more than 17.3 million beneficiaries. The OBA projects involve a number of different ICT services, including: public phones, telecenters, private phone connections, internet service including private connections as well as wholesale facilities known as points of presence (PoPs), and cellular networks. OBA for public phones is the most common type of OBA identified, with tele-centers second. This is partly because of the more public and therefore pro-poor nature of these two services, although as discussed below, this is changing. Figure 12: Distribution of OBA ICT Projects Regional Distribution of WBG OBA Projects in ICT SAR Total: US$ 73m 16% AFR 29% Regional Distribution of WBG OBA Projects in ICT SAR Total: 20 projects 10% AFR 30% LCR 30% LCR 41% EAP 14% ECA 0% ECA 5% EAP 25% 46 Many of these projects involve more than one contract. 69

70 The volume of OBA projects is relatively small, both in absolute terms ($73 million) and relative to the World Bank ICT portfolio (11% of portfolio). This is due to a combination of factors. ICT Universal Access projects are mainly funded through universal access funds that usually receive their funds from other sources than development aid. As a result donors, such as the World Bank, focus on different aspects of ICT development, such as setting up regulatory regimes or financing international infrastructure to ensure interconnection. Most of the contracts for ICT services identified as OBA are funded through universal access funds (UAFs) and universal service funds (USFs). UAFs and USFs ( UASFs for short) are funded from one or a combination of the following three sources: Levies on telecommunications operators The most common method for funding universal access/services are governments levies on a percentage of telecommunications operators revenues. Levies identified range between one and four percent, and average about two percent. Spectrum auctions Governments or sector regulators hold auctions to distribute ICT/radio spectrum in order to raise funding for the country s UASF (e.g., Guatemala s Fund for the Development of Telephony). Government budgets In a limited number of cases, governments fund the UASFs out of their own budgets. This is the case in Chile s Fund for Telephony Development (box 7). The scale of subsidies varies between projects. Subsidies for the provision of public phone connections were lowest in Chile, where they averaged $996 per phone. Such subsidies were highest in Nigeria, where they averaged $5,511 per phone. The average subsidies per phone in projects for which we have information were $2,848. UASFs have proven an effective mechanism for mobilizing investment into challenging rural areas. In 1994, the Fund for the Telephony Development (FONDETEL) in Chile became the first to make finance available competitively under a reverse auction (lowest subsidy demanded wins) and generate new licenses for rural operators. Maximum subsidies to be paid by UASFs are typically calculated at the amount required to make projects commercially viable. In most countries, commercial viability is defined as a net present value of zero. For instance, in Bolivia s SITTEL project, SITTEL determined the maximum subsidy required to make the net present value of each public phone connection zero through projections of necessary investment and demand. 70

71 Box 7: Fund for Telephony Development in Chile The Government of Chile established the Fund for Telephony Development (FDT) in 1995 to attract private investment in payphone services for rural and urban areas with low incomes and telephone density. FDT has since funded the installation of more than 25,000 payphones in about 8,000 rural population centers benefiting close to 2.7 million people. It is estimated that there are less than 150,000 people (1% of the population) without access to a basic telephone. The subsidies awarded cost the government less than 0.3 percent of total telecommunications sector revenue during the funding period, and administering the FDT cost about three percent of the monies granted. Between 1995 and 2000, rural telecommunications operators invested $161 million in universal access projects. Of this amount, $22 million (13.6%) was provided by the FDT. The FDT succeeded because of: a highly competitive market prior to the launch of the FDT and therefore the ability to rely extensively on market forces to determine and allocate subsidies; less regulatory discretion; simple and relatively expeditious processing; and effective government leadership. Competition between existing and new operators for the rural market substantially reduced the cost of telephone services to Government (especially compared to earlier public sector investments in similar facilities). Some concerns remain, however, about the long-term sustainability of the services, the small residual rural population still excluded, and some urban areas. Source: Wellenius 2002, Stern and Townsend 2007 (pg 92) and Nuevos Modelos Para el Acceso Universal de Los Servicios de Telecomunicaciones en América Latina: Informe de Países e Informe Completo However, although the funding source in ICT OBA has become explicit and is considered sustainable, another problem has arisen: UASFs have been good at collecting revenue from explicit subsidy contributions, but they have not been successful at disbursing. A total of 15 funds in developing markets that are operational, had collected a total of approximately $6.2 billion from operators, beginning in the late 1990s (but mostly since 2001 and 2002). 78% of the total collections ($4.8 billion) came from two countries - India and Brazil. By 2006, these 15 funds redistributed approximately $1.62 billion to the sector for universal access and service (UAS) projects, which represents just 26% of the total collected. 47 Some of the reasons set forth as to why include the following (Stern and Townsend 2007): 1. Defining eligible programs too narrowly accepting only those proposals linked to public payphones and community Internet; 2. Overestimating the amount of subsidy that operators would request; 3. Requiring that programs be approved by two or more ministries, that they must comply with all public expenditure review and monitoring procedures, or that they must conclusively demonstrate that the subsidies are well designed; 4. Imposing significant legal, administrative and financial burdens that act as a barrier on operators participation in tenders, especially for smaller operators. 47 InfoDev and ITU. ICT Regulation Toolkit. 71

72 Alternative models which would allow for the more efficient and effective management of UASFs are being considered, such as pay or play or virtual funds (with no actual physical fund). Private management of UASFs is also a possibility, such as the one in the USA, and a new privately managed UASF in Pakistan. These are discussed in Section 4 and 5 of the review. A.2. Targeting Targeting for output-based contracts for the provision of ICT access is largely geographic. Government target areas are defined based on level of telephony coverage. Some projects target smaller communities on the basis that they are likely to be the least economic to provide service to or areas with a low income level. Most OBA schemes target public (shared or community) access to maximize use by the poor, but as the nature and cost of technology is rapidly changing, private (household) access through OBA UASFs are increasing. Uganda s Energy for Rural Transformation project 48 aims to provide at least one public telephone connection per 2,500 inhabitants throughout Uganda (equating to an average distance of about 3 kilometers). Another targeting objective is to provide rural multipurpose tele-centers at schools, hospital, associations of farmers and micro-entrepreneurs outside of district capitals in selected districts. Bolivia s SITTEL project targeted 4,160 rural communities with 10,000 or fewer inhabitants that did not have access to public phone connections, and went further to define access as at least one public phone within two kilometers (or a 24 minute walk) of each household (Stern and Townsend 2007, p25). Many UASFs set national targets for tele-center coverage. In these schemes, tele-centers will reach poor areas, but are not targeted specifically to poor people. For example, Mexico s universal access fund aims to provide at least one tele-center (called a Community Digital Center ) per municipality in the country. Seventy-one percent of the tele-centers established to date have been in schools and libraries (accessible to poor and staffed to help people learn to use the computers). Part of the tele-center program is aimed at developing local educational, health, economic, and government content (Stern and Townsend 2007). The Mongolia Information and Communications Infrastructure Development Project (ICIDP) targets poor, underserved areas. It is a follow-up to the three GPOBA funded pilots that are successfully operational since 2007, bringing phone service to about 20,000 beneficiaries in remote herder communities, and both mobile phone and Internet services to 2,315 people in rural villages. A user survey showed that the pilots have cut by more than half the distance herders must travel to reach a phone and more than doubled the frequency with which they use the telephone service (Dymond et al. 2008, p. 3). A.3. Performance-risk Contracts for the provision of public phone connections most commonly define performance as installation of working public phone connections, with subsidies disbursed when public phone connections have been provided and determined to be in good working order. Some of the more effective contracts disburse a portion of subsidies upon installation of the phones and disburse the rest on a regular basis provided the phones are maintained to standard. This was the case in 48 (P069996) 72

73 Peru s Fund for Telephony Investment (FITEL), which has been considered to be largely successful (table 6). Since 2000, Peru s FITEL has increased the access to public telephony for nearly 6.7 million people, reducing the average distance to reach a public payphone from 56 km to 5.7 km (Stern and Townsend 2007, p. 95). In the 39 projects identified, there are three main ways of defining performance: Construction completion milestones: Contracts that define performance as construction completion milestones disburse some subsidies prior to the delivery of a serviceable asset. Provision of serviceable assets: Contracts that define performance as provision of serviceable assets disburse all subsidies at the delivery of the serviceable asset. Provision of serviceable assets and continued service: Contracts that define performance as provision of serviceable assets and continued service disburse some subsidies upon the delivery of the asset and some upon the successful continuation of service. Table 6 below gives examples of OBA projects that use each type of scheme for defining performance. 73

74 Table 6: Defining Performance in the Provision of ICT Services Indicator Construction and Installation completion milestones Serviceable assets Serviceable asset and continued service Example In Nigeria s Privatization Support Project (Sub-Project 2), performance is defined by construction completion milestones over a 16 month period. Nigeria s Privatization Support Project (Sub-Project 2), subsidy paid in 6 tranches over 16 month period (10%, 20%, 25%, 25%, 10%, 10%). First payment of 10% paid upon contract signature, provision of certificates and guarantees. Remaining tranches paid based on installation goals Chile s Fund for Telecommunications Development (FDT) pays subsidies in a lump sum once the phones have been installed In contracts funded by Peru s FITEL, 40 percent of the subsidies are delivered based on continued service provision over the course of five years. In the Mongolia ICIDP disbursed 20% of the subsidies on contract signature, 60% on reaching several output targets, such as providing service in a number of herder communities or reaching a number of mobile phone subscribers and 20% after 9 months of successful operation. For the ICT component of Uganda s Energy for Rural Transformation Project 49, 10% of the subsidy is disbursed upon signature of the service agreement, followed by 30% upon completion of 50% of the public access objective in year 1 and another 30% when the remaining 50% public access objectives have been achieved, in year 2. Another 20% is paid upon completion of the individual private access objectives in year 2 and the remaining 10% is paid in year 3 upon verification of satisfactory continuous service operation. A performance bond is required at contract signature, and its amount is reduced at every milestone in proportion to the percentage of the subsidy paid 50 It is also important to note that, although most of the contracts identified do not disburse OBA subsidies based upon the provision of continued service, the contracts do include longer-term service provision requirements. These contracts are commonly for 5 to 10 years. Contractors are sometimes required to post performance bonds in order to incentivize continuing service provision. However, most ICT companies will have other incentives to continue service provision. For example, ICT companies are sometimes granted operating licenses for up to 15 years, some of which allow the use radio frequencies, and can be revoked in the case of a breach of contract. A.4. Private sector capital and expertise In all of the cases identified, private sector companies were contracted for the provision of ICT networks. Usually, government agencies or a UASF defined the area to be served and the minimum number of people to be served. Contracted companies were able to decide the technology that would most efficiently allow the provision of coverage. Companies were also able to decide 49 (P069996) 50 There is also a private access component to this project. Source: Navas-Sabater et al. (2007), p. 1 74

75 if they would serve more than the target community using the infrastructure installed under the UASF contract, giving a potential for economies of scale and thereby increasing private sector incentives. Leveraging of private sector capital varied in the contracts that were identified. In Guatemala s FONDETEL projects, each $1 of subsidy leveraged between $2 and $4 of private investment. In Peru s FITEL projects, an average of $2 of private capital was raised for each $1 of subsidy. Most of the universal access funds aimed to provide subsidies that would make the net present value (NPV) of providing ICT services equal to zero in other words, it was the expectation that without subsidies, the NPV would likely be less than zero and the project would not be commercially viable. The intensity of competition between bidders for ICT contracts can greatly affect the amount of private sector leveraging. For example, the Chilean Telecommunications Company bid 100 percent of the maximum available subsidy in localities close to its existing network where there were no other established local operating companies, and zero subsidy in areas where its competitors had a strong presence. OBA was originally piloted to mobilize private sector expertise to serve segments of the population that without a subsidy would most likely go un-served. For example in Uganda, when two commercial operators declared they could not serve almost 20 percent of sub-counties on a commercial basis, OBA contracts were designed to specifically target these un-served subcounties. In order to mitigate demand risk under the OBA contracts including the risk of surprise substitutes, these areas were taken out of the license obligations of the non-oba providers and awarded exclusively to the OBA providers. But because the ICT sector is very dynamic, there are also instances where operators have started to provide services in areas originally thought to be un-commercial. This has led to the redesign of an OBA scheme in Cambodia prior to grant signing to ensure that only those areas which are not attractive without a subsidy fall within the OBA grant agreement. A.5. Monitoring The sub-sections above describe in detail the outputs that are normally designated for OBA projects in the ICT sector: a) construction completion milestones for example for the roll-out of network; b) installation of serviceable assets such as working pay phones; and, c) completion of public access objectives and continued service, for example provision of a working asset plus demonstration of service over a course of several months or years. In this sub-section, we describe the processes, lessons and challenges of monitoring output provision in OBA schemes in the ICT sector. Under most contracts except Chile and recently Peru, most UASFs are managed by the regulatory authority for the sector. In some cases, the UASF performs monitoring and evaluation of the specific contracts as well. In projects funded by Peru s Fund for Telephony Investment, for instance, the regulator Osiptel uses a network management system to oversee system operations (traffic levels, continuity of service) in real time. Ospitel s monitoring and evaluation scheme also requires a 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 compliance with performance targets and indicators and makes recommendations on FITEL payments. 75

76 Box 8 provides an example from Guatemala where a lack of monitoring and evaluation meant providers were not held accountable to the outputs prescribed by their contract. Currently, the Bank is working with the government on new implementing measures to address some of the failures of the past. Box 8: Development of Telephony Fund in Guatemala The Government of Guatemala (the Government) established the Fund for the Development of Telephony (FONDETEL). The FONDETEL sought to expand the coverage of public and private phone lines to about 5,000 rural localities that represented more than half of all households without access to telephone services. The FONDETEL was funded by spectrum auctions. FONDETEL contracted private firms for the provision of public phones in or near localities with the least access to ICT services were targeted Since 1998, FONDETEL has subsidized the construction of more than 5,500 public phones benefiting about 1.49 million people. However, a World Bank-funded inspection of 220 public phones found that only 28 phones were adequately functioning. It is estimated that only 20 percent of targeted localities had adequate service. The Regulatel study commented that FONDETEL s methodology for selecting these underserved communities was flawed, as it did not consider cell phone coverage and usage. Disregard of these factors made urban areas with high cellular tele-density eligible for subsidized public phones. The delivery of serviceable public phones was hampered by two factors: i) payment was not sufficiently linked to performance (with all subsidies disbursed directly after the installation of the phones, but before a period of service was provided); and, ii) nearly no monitoring and evaluation was performed of the installation of the phones. Revisions to the FONDETEL arrangements are expected to improve on linking payments to performance as well as enhancing monitoring and evaluation measures. Source: Nuevos Modelos Para el Acceso Universal de Los Servicios de Telecomunicaciones en América Latina: Informe de Países e Informe Completo and Stern and Townsend 2007 (p.xvii, 94) 76

77 B. Roads OBA is becoming mainstreamed in the roads sector through various types of performance-based maintenance and rehabilitation contracts. For example, in 2003 the World Bank formally adopted (and since then improved) a Sample Bidding Document for Output and Performance Based Road Contracts (OPRC) which can be found on the Bank s procurement website. Under OBA schemes in the roads sector, private contractors enter into agreements of a longer nature than traditional roads contracts, and the outputs on which they are paid (for example monthly) relate to the quality of road service provided based on clearly identifiable and measurable parameters such as average speed obtainable. The nature of OBA in roads maintenance and rehabilitation is quite different than the nature of OBA in other infrastructure sectors in a variety of ways: the ongoing nature of service delivery and the related ongoing subsidy requirement; in relation, the subsidy (or payment to the service provider) is disbursed on a continuing service i.e. management and maintenance of roads as opposed to investments; and, the public goods aspect of roads whereby there is no user fee and the entire cost to the user is subsidized through public funding, as discussed below. 51 B.1. Universe of OBA in Roads Governments and donors alike recognized the drawbacks of managing roads in house and hence for many years have relied on outsourcing through various contractual arrangements for road maintenance works. Under such arrangements, private contractors are responsible for carrying out physical works defined by the government and are paid on the basis of the quantity of civil works executed. But contractors often have a vested interest in executing large, lumpy works and little or no incentive to carry out the many small activities needed to ensure that roads are in good condition over a long period. Also, governments usually have an incentive to allocate scarce public funds on projects that convey clearly visible benefits to their constituents instead of less visible maintenance expenditures, which tend to be the first budget items to be cut. This contracting approach and the general tendency of governments to neglect maintenance have generated a vicious circle of heavy rehabilitation works followed by long periods of neglect and thus rapid deterioration. It is estimated that $1 spent on timely maintenance can save more than $3 that would be required for road reconstruction (Queiroz 2008). Adding to the problem has been the limited institutional capacity to consistently plan and supervise effective road maintenance. In an attempt to counteract these limitations, road agencies have begun to adopt performancebased contracting for rehabilitation and maintenance. Performance-based contracts expressly link payments the contractor to clearly defined performance standards. 52 Using performance-based contracts offers several potential advantages to road agencies over more traditional approaches, including better road conditions with limited funding available through incentives to the private sector for innovation and higher productivity, and, greater certainty about road expenditures to allow for better sector-wide planning. 51 All OBA schemes in developing countries indentified in the transport sector were in the roads sub-sector. A few transport services schemes have been considered, but are not far along enough in design or implementation to be included in this review. 52 Performance based contracting is also known as Performance-based Management and Maintenance Contract, Output Based service contract, Performance Specified Maintenance Contract. These names are used interchangeably though they often describe different types of road contracts. 77

78 Figure 13 below shows the widespread use of performance-based contracts in the roads sector. First appearing in Canada in the late 1980s, this approach is now used in many countries and is being considered by others, particularly in Asia. Figure 13: Global Application of Performance-Based Contracting for Roads Source: World Bank Resource Guide (2008): Performance-based Contracting for Preservation and Improvement of Road Assets., The review has identified 23 projects that involve OBA road contracts within the World Bank 53, for a total value of $2 billion (excluding over $1.7 billion in government subsidy co-financing), and 11 projects outside the Bank (predominantly in developing countries and majority in Latin America) (see figure 14). These are grouped mainly into two archetypes: 1) performance-based contracts for road maintenance, which are maintained to a specified level of service; 2) performance-based contracts for rehabilitation and maintenance, which are rehabilitated and then maintained to a specified level of service. 53 The term OBA is rarely used for these contracts other than in the case of Output and Performance-Based Road Contracts, but more commonly performance-based alone is used. 78