Water Global Practice report Performance of Water Utilities in Africa Caroline van den Berg and Alexander Danilenko About the Water Global Practice Launched in 2014, the Word Bank Group's Water Global Practice brings together financing, knowledge, and implementation in one platform. By combining the Bank's global knowledge with country investments, this model generates more firepower for transformational solutions to help countries grow sustainably. Please visit us at www.worldbank.org/water or follow us on Twitter at @WorldBankWater. Performance of Water Utilities in Africa Caroline van den Berg and Alexander Danilenko © 2017 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Please cite the work as follows: van den Berg, Caroline, and Alexander Danilenko. 2017. “Performance of Water Utilities in Africa.” World Bank, Washington, DC. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights​@­worldbank.org. Cover photo: Alexander Danilenko/World Bank. Further permission required for reuse. Cover design: Jean Franz, Franz & Company, Inc. Contents Foreword xi Acknowledgments xiii Abbreviations xv Overview 1 Notes 4 Chapter 1  Objective of the Study 5 Chapter 2  Scope and Methodology 9 Performance Analysis 11 Institutional Performance Analysis 13 Drivers of Utility Performance Analysis 13 Case Studies 14 Notes 14 Reference 14 Chapter 3  Performance of Utilities in Africa: Trend Analysis 15 Operational Performance 15 Financial Performance 18 Customer Performance 23 Conclusions 30 Notes 31 References 31 Chapter 4  Performance of Utilities in Africa: Composite Performance Index 33 Calculation of the Composite Performance Index 33 Composite Performance Index 34 DEA Efficiency 45 Conclusions 50 Notes 51 Chapter 5  Performance of Utilities in Africa: Institutional Factors 53 Role of Economic Development 54 Role of Regulation 55 Role of Service Delivery Models 55 Performance of Water Utilities in Africa iii Economies of Scale and Scope 58 Conclusions 59 Notes 60 Reference 60 Chapter 6  Drivers of Utility Performance: Panel Data 61 Financial Performance 63 Customer Performance 65 Operational Performance 69 Overall Performance Index 70 Conclusions 71 Notes 73 Chapter 7  Drivers of Utility Performance: Case Study Perspective 75 Introduction 75 Performance of the Case Study Utilities 76 Lessons Learned from the Case Studies 79 Conclusions 89 Notes 90 References 91 Chapter 8  Lessons Learned 93 Lesson 1: Although Utilities in Africa in General Underperform, There Are Relatively Well-Performing Utilities Operating in the Continent 93 Lesson 2: Customer Performance Is Relatively Weak Even among the Best-Performing Utilities 94 Lesson 3: The Major Drivers of Water Utility Performance Are Linked to the Cost of Service, While an Enabling Environment as Reflected in Good Economic Management also Matters 94 Lesson 4: Improving Water Coverage in Africa Will Require Large Investments that Will Have to Be Mostly Paid for by Government Funds 95 Lesson 5: Specific Measures Are Needed to Ensure that Progress in Financial Cost Recovery Does Not Translate into Less Affordable Services 96 Lesson 6: Availability of Data is Critical to Assess Performance and Guide Sector Planning 96 Note 97 Appendix A  Customer Performance and Water Coverage 99 iv Performance of Water Utilities in Africa Appendix B  Case Studies 109 Burkina Faso, ONEA (L’Office National de l’Eau et de l’Assainissement) 109 Côte d’Ivoire, SODECI (la Société de Distribution d’Eau de la Côte d’Ivoire) 117 Kenya, NCWSC (Nairobi City Water and Sewerage Company) 122 Senegal, SDE and SONES 130 Uganda, NWSC (National Water and Sewerage Corporation) 138 Notes 146 References 147 Appendix C  Data Quality Procedures 149 Appendix D  Data Envelopment Analysis Methodology 151 References 152 Boxes 4.1. Pros and Cons of Composite Performance Indexes 35 4.2. Cross-Subsidies in SDE and ONEA 43 Figures 3.1. NRW as Percentage of Water Production 16 3.2. NRW, by Connection 16 3.3. NRW, by Size of Utility, 2010–13 17 3.4. Labor Efficiency and Annual Staff Cost, by Size of Utility, 2010–13 18 3.5. Median O&M Cost per m Sold, by Income Status, 2010–13 3 19 3.6. Median O&M Cost per m Sold, by Size of Utility, 2010–13 3 19 3.7. Median Annual Revenues per Capita, 2010–13 20 3.8. Median Annual Revenues per Connection, by Income Status, 2010–13 20 3.9. Operating Cost Recovery versus Cash Flow Operating Cost Recovery, 2010–13 21 3.10. Operating Cost Recovery versus Cash Flow Operating Cost Recovery, by Income Status, 2010–13 21 3.11. Median Water Coverage, 2010–13 23 3.12. Median Water Coverage, by Size of Utility 24 3.13. Median Water Consumption, by Income Status, 2010–13 25 3.14. Median Water Production, by Country, 2010–13 26 3.15. Population per Connection, 2010–13 27 3.16. Hours of Water Supplied per Day, by Utility Size, 2010–13 27 3.17. Median Affordability, 2010–13 28 3.18. Median Affordability, by Size of Utility, 2010–13 29 4.1. Performance Indicators 34 Performance of Water Utilities in Africa v 4.2. Operational Performance Index against African Benchmark, 2010–13 36 4.3. Operational Performance Index against Global Benchmark, 2010–13 36 4.4. Median Operational Performance, by Country, 2010–13 37 4.5. Financial Performance Index Using the Operating Cost Coverage Ratio against African Benchmark, 2010–13 38 4.6. Financial Performance Index Using the Operating Cost Coverage Ratio against Global Benchmark, 2010–13 38 4.7. Median Financial Performance Index, by Country, 2010–13 39 4.8. Median Financial Performance Index based on Cash-Based Operating Cost Coverage Ratio, 2010–13 40 4.9. Median Financial Performance Index based on Cash-Based Operating Cost Coverage Ratio, by Income Status, 2010–13 40 4.10. Customer Performance Index against African and Global Benchmarks, 2010–13 41 4.11. Customer Performance Index against Africa Benchmark, by Income Status, 2010–13 41 4.12. Customer Performance Index against African Benchmark, by Country, 2010–13 42 4.13. Water Coverage Index against African and Global Benchmarks, 2010–13 44 4.14. Water Coverage Index against Global Benchmark, 2010–13 45 4.15. Water Coverage Index, by Country, 2010–13 against African Benchmark 46 4.16. Overall Performance Index (Unweighted Average of Operational Performance, Customer Performance [Service Quality], and Financial Performance) against African Benchmark, 2010–13 47 4.17. Overall Performance Index (Unweighted Average of Operational Performance, Customer Performance [Service Quality], and Financial Performance) against African Benchmark, by Presence of a Regulator, 2010–13 48 4.18. DEA Frontier 49 4.19. DEA Assessment of Relative Efficiency 49 7.1. Impact of the Dimensions of Performance 79 7.2. Water Coverage, NWSC Uganda, 1998–2013 80 7.3. Reliability, NWSC Uganda, 1996–2013 80 7.4. Water Coverage, ONEA Burkina Faso, 2001–14 81 7.5. Reliability, ONEA Burkina Faso, 2006–14 81 7.6. Water Coverage, SDE/SONES Senegal, 1995–2013 82 7.7. Reliability, SDE/SONES Senegal, 1996–2013 82 7.8. NWSC, Uganda Tariff Structure 84 7.9. SDE/SONES, Senegal Tariff Structure 85 7.10. Access to Water, Dakar, 2000–14 87 7.11. Access to Water, Ouagadougou 87 B.1. Population Served Compared with Service Area Population, 2000–14 110 vi Performance of Water Utilities in Africa B.2. Water Coverage (Population Served Divided by Service Area Population), 2001–14 110 B.3. Access to Water Service, Ouagadougou Only 111 B.4. People Served per Connection, 2001–14 111 B.5. Number of Water Connections 112 B.6. Network Expansion 112 B.7. Hours of Supply per Day, 2006–14 113 B.8. Sufficiency of Residential Consumption, 2001–14 113 B.9. Nonrevenue Water, by Connection, 1995–2014 114 B.10. Nonrevenue Water as a Percentage of Production 114 B.11. Staff Numbers and Staff per 1,000 Water and Sewer Connections, 2000–14 115 B.12. Collection Ratio, 2001–13 115 B.13. Operating Cost Recovery Ratio, 2002–14 115 B.14. Average Tariffs, Average Costs per m , 2002–14 3 116 B.15. Sources and Amounts of Investment Financing, 2002–13 116 B.16. Population Served Compared with Service Area Population, 2000–14 119 B.17. Water Coverage (Population Served Divided by Service Area Population) 119 B.18. People Served per Connection, 2000–14 120 B.19. Sufficiency of Consumption, 2000–14 120 B.20. Hours of Water Supply per Day, 2000–14 121 B.21. Affordability, 2000–14 121 B.22. Nonrevenue Water as a Percentage of Production, 2000–14 122 B.23. Nonrevenue Water by Connection, 2000–14 122 B.24. Staff Numbers and Staff per 1,000 Water and Sewer Connections 123 B.25. Operating Cost Recovery 123 B.26. Average Tariffs, Average Costs per m 124 3 B.27. Collection Efficiency, 2000–11 124 B.28. Population Served Compared with Service Area Population, 2008–14 125 B.29. Water Coverage (Population Served Divided by Service Area Population) 125 B.30. Hours of Water Supply per Day, 2011–14 126 B.31. Sufficiency of Consumption, 2010–14 126 B.32. Nonrevenue Water, by Connection, 2010–14 127 B.33. Nonrevenue Water as a Percentage of Production, 2010–14 127 B.34. Staff Numbers and Staff per 1,000 Water and Sewer Connections, 2009–14 128 B.35. Collection Ratio, 2009–14 128 B.36. Operating Cost Recovery Ratio, 2009–14 128 B.37. Average Tariffs, Average Costs per m , 2010–14 3 129 B.38. Population Served Compared with Service Area Population, 1996–2014 131 B.39. Water Coverage (Population Served Divided by Service Area Population), 1995–2014 131 Performance of Water Utilities in Africa vii B.40. Access to Water Service, Dakar Only 132 B.41. People Served per Connection, 1995–2013 132 B.42. Network Expansion, 1995–2013 133 B.43. Number of Water Connections, 1995–2013 133 B.44. Hours of Supply per Day, 1995–2013 133 B.45. Sufficiency of Consumption, 2004–13 134 B.46. Nonrevenue Water, by Connection, 1995–2013 134 B.47. Nonrevenue Water as a Percentage of Production, 1995–2014 134 B.48. Staff Numbers and Staff per 1,000 Water and Sewer Connections, 1996–2013 135 B.49. Collection Ratio, 1996–2013 135 B.50. Operating Cost Recovery Ratio, 1998–2013 135 B.51. Average Tariffs, Average Costs per m , 1997–2013 3 136 B.52. Sources and Amounts of Investment Financing, 1996–2013 136 B.53. Population Served Compared with Service Area Population, 2003–14 139 B.54. Water Coverage (Population Served Divided by Service Area Population), 1998–2013 139 B.55. Network Length, 2002–13 140 B.56. Towns Served, 2003–13 140 B.57. Hours of Water Supply per Day, 1996–2013 141 B.58. Sufficiency of Residential Consumption, 2005–13 141 B.59. Nonrevenue Water, by Connection, 1998–2013 142 B.60. Nonrevenue Water as a Percentage of Production, 1998–2013 142 B.61. Staff Numbers and Staff per 1,000 Water and Sewer Connections, 1997–2013 143 B.62. Collection Ratio, 2001–14 143 B.63. Operating Cost Recovery Ratio, 2001–14 144 B.64. Average Tariffs, Average Costs per m , 2001–14 3 144 B.65. Capital Expenditure Financing, 2002–07, 2009–11 145 Tables 2.1. Characteristics of the Sample of Utilities 11 2.2. Operational Performance (Unweighted Average) 12 2.3. Customer Performance (Unweighted Average) 13 3.1. Cost Recovery among Bottom- and Top-Performing Utilities 22 3.2. Median Affordability, by Country, 2010–13 29 4.1. Setting Benchmarks 35 4.2. Best-Performing Utilities in 2013: Unweighted Average of Operational Performance, Customer Performance (Service Quality), and Financial Performance 47 4.3. Relationship between Utility Performance (Unweighted Average of Operational, Financial, and Customer Performance) and Utility Size 47 viii Performance of Water Utilities in Africa 4.4. Best Five Utilities’ Relative Performance, by Year, 2010–13 50 4.5. Correlation between DEA Relative Efficiency and the CPI for Each of the Years, 2010–13 50 5.1. Impact of Economic Development on Utility Performance 54 5.2. Impact of a Regulatory Agency on Utility Performance 55 5.3. Effect of District or Municipal Service Delivery on Utility Performance in Low-Income Countries 56 5.4. Labor Productivity and Labor Cost per Employee 57 5.5. Correlation between CPIA Indicator Clusters and Level of Decentralization 57 5.6. Correlation between CPIA Indicator Clusters and Utility Performance Indicators 57 5.7. Effect of Scale on Utility Performance 58 5.8. Economies of Scope on Utility Performance in Low-Income Countries 58 6.1. Explanatory Variables Used in the Performance Models 62 6.2. Correlation between Technical and Financial Performance for Water Utilities 62 6.3. Drivers of Financial Performance, Random-Effects Probit Model 64 6.4. Drivers of Water Coverage, Fixed-Effects Regression Model 65 6.5. Drivers of Water Coverage, Fixed-Effects Regression Model, by Size of Utility 67 6.6. Drivers of Customer Performance as Measured by Quality of Service, Fixed-Effects Regression Model 68 6.7. Drivers of Operational Performance, Fixed-Effects Regression Model 69 6.8. Drivers of Overall Performance (Measuring the Combined Scores of Financial, Operational, and Customer Performance), Fixed-Effects Regression Model 70 7.1. Country Data on the Growth in Access 77 7.2. Country Data on Access to Improved Water Sources by Country, According to the JMP 77 7.3. Summary of Performance of Case Study Utilities 78 7.4. Sources and Amounts of Capital Investment Financing 82 7.5. Sources of Financing in NWSC (US$,000) 83 7.6. Effect on Residents of Cross-subsidization in the Five Case Study Utilities 85 7.7. Effect of Cross-Subsidization on Consumption Patterns in the Five Case Study Utilities 86 A.1. Customer Performance and Water Coverage, National 100 A.2. Customer Performance and Water Coverage, Regional 101 A.3. Customer Performance and Water Coverage, Municipal 106 Performance of Water Utilities in Africa ix Foreword The Sustainable Development Goals (SDGs) aim to achieve universal and equitable access to safe and affordable drinking water and sanitation for all by 2030. The SDGs are also calling on more sustainable use of water resources through, amongst others, improving water quality by reducing pollution by halving the proportion of untreated wastewater and substantially increasing recycling and safe reuse. These goals and targets are very ambitious, especially since the region was not able to meet either the water or the sanitation Millennium Development Goals. Between 2000 and 2015, access to piped water in Africa increased, but the urban popu- lation grew even faster. This resulted in the decline of piped water on premises as a pri- mary source of water supply from 40 percent in 2000 to 33 percent in 2015. One of the reasons for the decline in piped water access is that many utilities are not able to cover their basic operation and maintenance (O&M) costs. This results in insufficient funds to keep up the service levels for existing consumers, let alone for new consumers. Government funds in such environments often have to pay not only for the expansion of access, but also to cover part of the O&M costs, which crowds out investments to connect new customers. This report tries to look into how African utilities are performing. The study used a panel data of 120 utilities throughout low- and lower-middle income countries in Africa utilizing the Bank supported IBNET toolkit and database. The study also includes a set of case studies of the best performing utilities in the continent. The report also investigates the drivers of utility performance. This will help governments in the design and implementation of water projects and sector reforms in urban environments as achieving the SDGs in the fast growing cities of Sub-Saharan Africa will require a massive effort to ensure that more people get access to sustainable and affordable services. The key findings of the report were shared with the ministers of Finance of Africa during the Joint Bank-Fund Annual Meetings in October 2016. These key findings included that util- ities in SSA show overall weak performance, especially with regard to customer perfor- mance. The case studies, however, show that there are well performing utilities. At the same time, using larger datasets, it is clear that sector reforms in combination with changes in the economic environment in which utilities are operating (referring to for instance civil service reform, energy policies, and land use planning, improvements in public investment plan- ning) can help to improve the efficiency of water utilities in Africa. But as there is large gap in service coverage, especially with regard to wastewater collection and treatment (as few utilities in the sample provided these services), the region would need to invest significantly in the sector if the SDGs are to be met. Performance of Water Utilities in Africa xi We hope that this publication will add information to the debate on how to most efficiently and effectively increase access to affordable water and sanitation services in the cities of Sub-Saharan Africa, and provide the reader with an informed view on the workings of the water utility sector to synergize international efforts in the region. Guangzhe Chen Senior Director, Global Water Practice, World Bank Group xii Performance of Water Utilities in Africa Acknowledgments The preparation of the report was funded by the Water Partnership Program (WPP) and International Network of Water and Sanitation Utilities (IBNET) of the World Bank. The main authors of the report are Caroline van den Berg (Lead Water Economist) and Alexander Danilenko (Senior Water and Sanitation Specialist). The authors appreciate very much help and support from of the World Bank colleagues who helped collect and analyses the data and other information through the study development. Matar Fall and Pierre Francois-Xavier Boulenger organized and managed information collection in Francophone Africa, Michael Webster collected data in Malawi. Yitbarek Tessema coordinated data collec- tion efforts in Ethiopia and Tanzania. Our external partners Peter Ramsden managed the consultation and data collection in South Africa and Leshoto, Peter Njaggah provided data from the Kenyan Regulator (WASREB) and Kelvin Chitumbo, director of NWASCO, provided data from the water and sanitation utilities in Zambia. Berta Macheve and Aroha Bahuguna assisted in the data aggregation and analysis. The contribution of Celine Nauges was instru- mental in the preparation of Chapter 3. L. Joe Moffitt advised the team in preparation of the Chapter 4. Christiaan Heymans coordinated the evaluation of performance of the individual utilities that are discussed in Chapter 6. The team benefited greatly from the peer reviewers who advised the team, including Richard Damania, Maria Vagliasindi, Luis Tavares, Pedro Antmann, William Kingdom, Glenn Pearce-Oroz, Luis Andres and Josses Mugabi, and many other colleagues, professionals and specialists who were guiding and contributing to the report.  Their review and constructive comments helped to streamline the report. Alexander Bakalian and Jonathan Kamkwalala supervised the production of this report. The team valued the guidance and advice provided by World Bank management, in particular, Jamal Saghir, Jennifer Sara and Jyoti Shukla. Performance of Water Utilities in Africa xiii Abbreviations CPIA Country Policy and Institutional Assessment DEA data envelopment analysis DHS Demographic and Health Survey GNI gross national income GoS government of Senegal IBNET International Benchmarking Network for Water and Sanitation Utilities IDAMC internally delegated management contract JMP UNICEF—WHO Joint Monitoring Programme lcd liter per capita per day MICS multiple indicator cluster survey NRW nonrevenue water NCWSC Nairobi City Water and Sewerage Company (Kenya) NWSC National Water and Sewerage Corporation (Uganda) OCCR operating cost coverage ratio O&M operation and maintenance ONAS Office National de l’Assainissement du Sénégal ONEA Office National de l’Eau et de l’Assainissement (Burkina Faso) PPP public–private partnership SDE Sénégalaise des Eaux (Senegal) SODECI Société de Distribution d’Eau de la Côte d’Ivoire SONES Société Nationale des Eaux du Sénégal TTL task team leader WASREB Water Services Regulatory Board (Kenya) WSB water services board WSP water service provider Performance of Water Utilities in Africa xv © Alexander Danilenko/World Bank. Further permission required for reuse. Overview Africa’s urban population is growing rapidly. Between 2000 and 2015, the urban population increased by more than 80 percent to 373 million people. Although access to piped water increased (from 82 million urban dwellers with piped water in 2000 to 124 million in 2015), African utilities were not able to keep up with the rapid urbanization as reflected in the decline of piped water on the premises as a primary source of water supply in percentage terms. The urban population served with piped water on the premises declined from percent in 2000 to 33 percent in 2015.1 The total population with improved services 40  ­ increased, but most of that increase came from an increase in the access to piped water off premises and self-supply. Government funds often have to pay not only for expansion of access, but also to cover part of the operation and maintenance (O&M) costs, which crowds out expansion investments. These government transfers also provide utilities with few incentives to improve their financial performance. This report looks into how African utilities are doing using a data panel of about 120 utili- ties in low- and middle-income countries in Africa, which represent about 53 percent of the urban population served by piped network services and covered 14 countries in different parts of Africa. The most notable exceptions in the analysis were Ethiopia and Nigeria— which had only one or two years of data available at the time of analysis. Utility performance in Africa is in general weak, although there are well-performing utili- ties. The performance of African utilities in the data panel has shown some improvements between 2010 and 2013. Utilities in the sample were able to slowly improve water coverage, but overall coverage stood at only 60 percent. Access to sewerage services is in its infancy in Performance of Water Utilities in Africa 1 Africa with very few utilities providing such services. The O&M costs are highly variable in Africa—ranging from less than US$0.23 per cubic meter of water sold to US$2.07. The vari- ance of utilities’ performance within and between countries is very large. This is, for the most part, because water and wastewater services are affected by local factors. These factors can vary widely between utilities and include factors such as the distance to the water source and the effect on the cost to store and transport water, the quality of the water source and the need for treatment, and design standards, among others. Yet, more general policies in the country (for example, energy subsidies and labor policies) also affect the cost of O&M. As virtually everywhere water tariffs are set on the basis of the O&M cost of water, controlling costs is a major element in making the service more affordable. A data envelopment analysis (DEA) showed that although there are some relatively effi- cient utilities in Africa, they make up a small minority of utilities. The majority of utilities register an efficiency of 0.30 (which is far below the highest score of 1), showing significant options for improvement. As the fixed cost component in water (and wastewater) service provision is very large, the design (but also environmental, water quality, public health, and other) standards under which water infrastructure is constructed will determine the cost of the services for decades to come. Hence, it is important to undertake a proper least-cost analysis when investment decisions are made to ensure that the benefits and costs of such investments are properly analyzed because the financial, organizational, and social implica- tions of such investment decisions will be felt for many decades. DEA tests also show that governance may be a significant driver of water utility performance in Africa and that atten- tion to improving governance may be key to improving performance in water utilities. The increase in O&M costs of water services between 2010 and 2013 has been accompanied by an increase in the affordability of water services. This suggests that in many utilities, even in Sub-Saharan Africa, there is some scope for balancing the goals of revenue sufficiency and affordability more in favor of the former as government subsidies will otherwise need to increase rapidly in some countries. This is especially important because in 2013 a little less than half the utilities in the sample were not able to cover their O&M costs with their reve- nues. The high dependence on subsidies has major consequences. It crowds out investments in the sector, but it also results in serious equity concerns as those with piped water services tend to be more likely to be households with higher incomes. The case studies undertaken show that there are relatively well-performing utilities, but that even those that are perform- ing well with regard to operational and financial performance show weaknesses in providing customers with high-quality services (as measured in type of service level, reliability, and water consumption levels), especially when compared with global benchmarks. The context in which utilities operate matters. Collecting data on the institutional and socioeconomic context in which utilities operate matters. Regulation is often seen as a tool to ensure better governance in the sector.2 In low-income countries a regulator results in better customer performance, but does not extend to other forms of performance, such as operational performance or water coverage. This may be expected, because the objective of 2 Performance of Water Utilities in Africa a regulator is to provide “customer protection” to existing customers (and hence the focus is on providing minimum levels of customer service) but does not necessarily focus on improvements in financial and operational sustainability or improvements in coverage com- pared with utilities without a regulator.3 Service provision at the local level will increase accountability and improve utility performance. Yet, the financial and operational perfor- mance of utilities is not statistically different from the performance of other forms of service delivery. As for customer performance, there are statistically significant differences: district- or municipal-based service delivery shows better results. Yet, coverage lags behind in municipal- or district-based service delivery compared with other levels of service delivery. This may be partially explained by the impact of economic development on a utility’s perfor- mance. If customers have access to more piped water to consume, the benefits of piped water are more apparent compared with other water supply sources.4 This focus on improv- ing the infrastructure will require, in the short term, more investments in the sector to improve water coverage and a level of water consumption that sets piped water apart from alternative water sources. The organization of water services affects utility performance. Economies of scale and scope matter. Analysis of the large data sample shows that size matters, but that being too big has a negative impact on performance. Economies of scope are evident. Utilities that provide water and wastewater services in low-income countries tend to show slightly higher levels of water coverage (as can be expected as sewerage coverage is provided sequen- tially after a certain level of water coverage has been achieved). But customer performance (or  ­ service quality), operational performance, and financial performance are also higher when sewerage services are provided in tandem with water services. Using large utility performance datasets to explain the drivers of utility performance shows that the various aspects of what is considered good utility performance are very much interrelated with feedback. The different aspects of performance affect one another. Higher customer service quality has a positive impact on water coverage, whereas financial perfor- mance may affect operational performance and water coverage, suggesting that funding is necessary to improve access and measures to improve operations in the utility. In the case of improving water access, the level of economic development matters, but also the quality of economic management, especially for smaller utilities that may depend more than the large ones on external funding to increase access, and hence enhance their revenue base. The most interesting findings are that water coverage is directly influenced by customer perfor- mance (that is, service quality). Better economic management and higher gross national income (GNI) growth have a positive impact. In most cases there is no link between a utility’s financial performance and water coverage—with the exception of large utilities in low-­ income countries (although the effect is small). This suggests that most utilities are not able to improve access through improved financial performance but depend on external funds to do so. This finding was confirmed by the case study utilities as most of these utilities needed some external funding to make progress in improving access. Finally, as many of the Performance of Water Utilities in Africa 3 performance indicators are composite indexes, using a single indicator, the most important may be the O&M cost per cubic meter sold. We found that tariffs essentially are driven by the O&M costs and to a lesser extent by the collection efficiency. Hence, minimizing the cost of services and improving the efficiency with which utilities collect their revenues will explain how well they are able to manage their financial performance. Another interesting finding is the impact of operational performance and customer perfor- mance on financial performance. Better operational performance tends to have a positive impact on financial performance; but the opposite is true for customer performance. The better the customer performance, the lower the financial performance. This may link to the cost associated with improvements in the different aspects of performance. Financial per- formance has a positive impact on operational performance too, although the effect seems weaker than the other way around. There is evidence that unobserved utility-specific heterogeneity explains a large portion of  the total variance, which would call for the large-scale collection of additional utility-­ specific variables, for example, information on local conditions (topography, distance to the raw water source, whether the utility gets water primarily from groundwater or surface water, quality of the raw source, age of the infrastructure, access to alternative water source, and so on) and organizational and institutional data.5 Hence, much more and detailed infor- mation on utility operations and the context in which the utilities operate is required to explain with more clarity what drives individual utility performance. However, such data collection comes at a price with regard to the cost of collection and the willingness of utili- ties, regulators, and other stakeholders to provide such information; there is a trade-off to be made between data to be collected and analyzed and the cost of doing so. Notes 1. The United Nations Children’s Fund (UNICEF) and World Health Organization (WHO) Joint Monitoring Programme has served with piped water a total population of 153.4 million in 2015 (compared with 74.5 million in 1990), as there is also a small part of the rural population being served with piped water. The rural population with piped water increased in absolute numbers from 15 million to 29 million between 1990 and 2015, and piped water coverage remained more or less constant over the 1990 to 2015 period, at around 15 percent. 2. Countries that have a regulatory agency in place show higher indexes for public sector management and institutions (as measured by the Country Policy and Institutional Assessment [CPIA]) as there is reference in the definitions of CPIA to the existence and functioning of regulatory agencies in the quality of public administration definition (part of the public sector management and institutions overall index). 3. It is unclear whether this is the result of the higher service levels, higher labor costs (as labor efficiencies coincide with higher staff cost per employee), the existence of higher transaction costs for the utility operators, or a multitude of other factors. 4. In places where there are abundant alternative water resources, demand for piped water will always have to compete with these alternatives. Hence, when planning investments in piped water, this should be considered and a detailed demand analysis should be undertaken to investigate the demand for piped water in such environments. 5. The IBNET Toolkit includes organizational data, but in this round of data collection in Africa, this information was not ­ consistently collected by the task teams, and such information was not available in many of the regulatory reports. Hence, this information is not included in the analysis. 4 Performance of Water Utilities in Africa © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 1 Objective of the Study Africa’s urban population is growing rapidly. Between 2000 and 2015, the urban population increased by more than 80 percent from 206 million to 373 million people. Although access to piped water increased over the period (from 82 million urban dwellers with piped water in 2000 to 124 million in 2015), African utilities were not able to keep up with the rapid urban- ization as reflected in the decline of piped water as a primary source of water supply in per- centage terms. The urban population served with piped water on the premises declined from 40 percent in 2000 to 33 percent in 2015. The total population with improved services increased, but most of that increase came from an increase in the access to piped water off premises and self-supply. This means that the performance of water utilities has been seri- ously lagging behind as there seems to be no lack of demand for improved water supplies. One of the reasons for the decline in piped water access is that many utilities are not able to cover their basic operation and maintenance (O&M) costs and hence are not able to generate sufficient funds to expand access. Government funds often have to pay not only for expan- sion of access, but also to cover part of the O&M costs, which crowds out expansion invest- ments. This dependence on government transfers at the same time does not provide utilities with many incentives to improve their financial performance. The overall decline in performance has not been investigated in much detail. As a result, the drivers of success in utility performance are still rather elusive for two major reasons. The first is a lack of agreement on what constitutes good performance. Utilities are supposed to provide services that are efficient, affordable, and sustainable. Yet, to simultaneously be able to provide affordable and sustainable water services has proven to be difficult as these different goals often result in conflicts and trade-offs. Having financially sustainable water Performance of Water Utilities in Africa 5 services that cannot be afforded by the majority of the population and as such are only accessible to the most well-off may not be desirable for a service that has important public health benefits. Yet, a service that is affordable but not financially sustainable tends to set utilities on a path of inadequate maintenance that results rather rapidly in low service qual- ity for which consumers are not willing to pay and hence will stall the willingness of poten- tial customers to connect and the willingness of existing customers to pay for piped water services. Hence, different emphasis on different aspects of what constitutes good perfor- mance may result in widely varying performance assessments. Second, because of lack of empirical work, there is little clarity on what drives performance in utilities. Water utilities in Africa are diverse. Utilities differ in their institutional setup (ranging from national utilities to decentralized companies working at municipal levels), organization, and reporting requirements. In view of the large number of utilities all over the world, the wide variety in geographic, hydrological, economic, social, institutional, politi- cal, and cultural differences, and changes over time and space, this lack of empirical work often means that sector professionals apply results from one utility or one country (often utilities in developed countries) to utilities in other countries with, often, very different institutional, political, and economic environments. The objective of this assessment is to inform Bank and government policies and projects on the drivers of utility performance by • assessing the performance of a sample of African utilities and benchmarking their per- formance against one another; • investigating the drivers of utility performance and determining how this affects the way that the government and Bank staff design and implement water and wastewater projects and policies; and • helping to increase the monitoring and evaluation capacity in utilities using the data available and benchmarks for performance in Africa as a baseline for water utility perfor- mance in Africa. The report describes the main outcomes and lessons learned from the assessment that identified and analyzed the main features of water utility performance in Africa. The report includes the following chapters. Following this introduction, chapter 2 describes the methodology used in the study, including details on the data collection process. In  chapter 3, the study team undertook a trend analysis of utility performance of the ­ sector. The team had to weigh the advantages of providing shorter time series that cover larger groups of utilities or longer time series with a limited number of utilities. Chapter 4 examines the efficiency of utilities using a DEA while also using an absolute performance approach. Chapter 5 investigates the effect of institutional factors on utility perfor- mance. This is followed in chapter 6 by an econometric analysis of the drivers of utility performance, using various definitions of utility performance. The results from the econometric models are then triangulated with a set of case studies of five utilities 6 Performance of Water Utilities in Africa (Burkina Faso’s l’Office National de l’Eau et de l’Assainissement [ONEA], Côte d’Ivoire’s la Société de Distribution d’Eau de la Côte d’Ivoire [SODECI], Kenya’s Nairobi City Water and Sewerage Company [NCWSC], Senegal’s Sénégalaise des Eaux [SDE], and Uganda’s National Water and Sewerage Corporation [NWSC]), similar to those that the Electricity Study Team undertook, which are presented in chapter 7. The report concludes in chap- ter 8 with the lessons learned from the assessment. Performance of Water Utilities in Africa 7 © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 2 Scope and Methodology When looking into the performance of water utilities in Africa, one has to bear in mind that Africa is a large continent with significant differences in institutional setup of water sectors, access to and quality of water resources, and levels of economic development across and within countries. Much of that variation will be lost in aggregating information. Yet, taking into account all these utility-specific differences will result in a set of case studies that make it hard to compile any of the findings into more general lessons learned. This study is part of a larger study that also looks into the performance of electricity utilities. The electricity study used a case study approach and together with this water assessment will allow for a better understanding of the advantages and disadvantages of the different study approaches. A well-performing utility is a utility that is able to provide high-quality water and/or waste- water services to its customers in a sustainable manner. This definition of a well-performing utility includes elements of good financial and operational performance, but also universal access to water and wastewater services that are affordable to all. The analysis will look at three elements to define a well-performing utility: financial performance, customer per- formance (an index that covers the quality of access to water services), and operational performance. Water coverage is looked at separately because in many cases investments to support an increase in access to water and wastewater services are mostly funded by gov- ernment and hence not fully under the control of the utility. Another point to consider is that water coverage has increased rapidly over time in Africa. Between 1990 and 2015, Africa’s population grew from 510 million to 989 million—an increase of 94 percent. Population growth in urban areas was even higher—increasing by more than Performance of Water Utilities in Africa 9 169 percent between 1990 and 2015. Even though the proportion of access to improved ser- vices increased from 48 percent in 1990 to 68 percent in 2015, access to piped water barely changed and still stood at 15 percent in 2015 according to the UNICEF–WHO Joint Monitoring Programme (JMP). This means that access to piped water services on the premises in urban areas decreased from 43 percent in 1990 to 33 percent in 2015. At the same time, the absolute number of people with access to piped water services on the premises increased from 74 million to 153 million between 1990 and 2015 in Sub-Saharan Africa.1 Simultaneously, the number of utilities in Africa has increased significantly as more and smaller towns estab- lished utilities and a larger number of these smaller towns were included in the International Benchmarking Network for Water and Sanitation Utilities (IBNET) sample. The team combined a desk study using existing data from IBNET with fieldwork under- taken by task team leaders (TTLs) in Africa to collect additional operational and financial utility data. This ensured a sufficiently large panel to undertake analysis to determine (a) the drivers of good utility performance and (b) the scope of efficiency improvements in these utilities that could be achieved. The data collection by the TTLs was most effective in western Africa where data was col- lected for the period 2010 to 2013 for Benin, Burkina Faso, Côte d’Ivoire, Guinea-Bissau, Niger, and Senegal. In addition, data for the same time period was collected for the Democratic Republic of Congo, Kenya, Malawi, Mozambique, Tanzania, Uganda, and Zambia mostly by mining reports and databases from regulators, the IBNET team data col- lection efforts, and, in the case of Mozambique, through TTL project reporting. Data from Ethiopia and Nigeria were collected previously by the IBNET team managed by the Water and Sanitation Program; but data collection efforts have just started in these countries and hence only include one or two years of observations. The team used the IBNET Toolkit to ensure that the data collected was comparable. By the end of the data collection period, the team had access to 306 utilities from 41 countries covering a period of up to 20 years of observations. However, the data panel is unbalanced, with some utilities represented in the database with only one year of observations and others with up to 20 years. Some utilities provide very detailed information while others provide much less information. The data was provided voluntarily by the utilities and this—in combination with differences in the quality of the data collected—results in variations in what data is being submitted and can actually be used by the study team. The data was collected by TTLs from the utilities with a first quality check from the TTLs and a second quality check from the IBNET team. The quality assurance methods are discussed in appendix C. For the analysis, the team used a balanced panel (that is, a set of utilities that remains unchanged over the period under review). The team used datasets that included a subset of the dataset covering a shorter period (2010 to 2013). This balanced panel will, however, exclude large parts of Africa (especially utilities in countries like Ethiopia and Nigeria where data collection has just started and where there are only one or two years of data  available). The analysis also excluded utilities from Namibia and South Africa2 as 10 Performance of Water Utilities in Africa these are upper-middle-income countries whose utilities perform at a different level than most, if not all, other utilities in the region. Where possible, and as far as data availability allowed, the team also analyzed institutional factors. Despite the exclusion of many coun- tries, the panel data sample represents a large part of the African population served with piped water; partially because access to piped water in countries like Ethiopia and Nigeria is still rather low. The JMP estimated that 149 million people in Sub-Saharan Africa had access to piped water in 2013; when excluding upper-middle-income countries like Botswana, Mauritius, Namibia, and South Africa, the population with access to piped water on the premises drops to 108 million. The database between 2010 and 2013, excluding the upper-middle-income countries, covers a total population served with piped water of 58 million (equivalent to 55 percent of the people served by piped water according to the JMP), of which 41 million use piped water through house connections (including sharing of these house connections). Hence, the people covered in the database cover a very large part of the total population in Africa with access to piped water. Table 2.1 below describes the sample of utilities in detail. Performance Analysis The analysis starts with a sector status report looking at the performance of the balanced panel of 119 utilities in Africa. The sector status report provides data on how these 119 utili- ties have been performing between 2010 and 2013 on different elements of performance. The team undertook a performance analysis by defining three different indicators: operational performance, financial performance, and customer performance. Operational performance was defined as the unweighted average of three indicators: metering, nonrevenue water (NRW, as measured in cubic meters per connection per day), and staff efficiency (which measures how much revenues are collected for each U.S. dollar spent on staff costs). Table 2.2 shows the benchmarks that were used to calculate how far utilities Table 2.1. Characteristics of the Sample of Utilities Subsample 2010–13 (without Indicators Total sample UMIC) in 2013 Number of people served with piped water (million) 88 58 Number of people served with wastewater services (million) 24 5 Number of people in service area of utilities (million) 134 103 Number of towns served with piped water 1,754 1,574 Number of towns served with wastewater 347 183 Annual utility turnover (US$, billion) 3.7 1.1 Annual volume of water sold (billion m3) 3.0 1.3 Total staff employed in utilities 39,392 28,560 Note: UMIC = upper-middle-income country. Performance of Water Utilities in Africa 11 Table 2.2. Operational Performance (Unweighted Average) Africa benchmark Global benchmark Indicator (best 25 percent of sample) (best 25 percent of sample) Metering (%) 100 100 NRW per connection per day (in m ) 3 0.205 0.121 Staff efficiency 4.21 4.27 Note: Staff efficiency = Total revenue per employee/Total (labor) cost per employee. deviate from those benchmarks set at the best 25 percent of the sample being used. Scores are between 0 and 1. If the utility is achieving the benchmark value, a value of 1 is assigned; in case the utility achieves below the benchmark the utility achieves a score of below 1—the further away from the benchmark, the lower the value. A global benchmark was also calcu- lated using the data analysis from the second IBNET Blue Book (Danilenko et al. 2014) using the best 25 percent of a sample covering 2006–11. For financial performance, the operating cost coverage ratio (OCCR) was used. The African benchmark of the best performing (25 percent) utilities was 1.19; whereas the global bench- mark stands at 1.38. Customer performance was defined as the unweighted average of three indicators: popu- lation per connection, reliability, and affordability. The population per connection is looked at as a proxy for service levels. When utilities provide only household connections, the pop- ulation per connection tends to be relatively low (slightly above the average household size). Yet, sharing of connections is common in Africa through the provision of standposts, the use of water kiosks, and sharing of house connections with several households. Table 2.3 shows the benchmarks that were used to calculate how far utilities deviate from those benchmarks set at the best 25 percent of the sample being used. The same methodology is used as described above in the operational performance measurement. Finally, a water coverage performance indicator that measures how the utilities are performing in providing access to water services was included. Although there are utilities in the sample that provide wastewater, the number of utilities that provide wastewater ser- vices is limited and moreover, those that provide wastewater services provide services to only a very small population. Hence, this service has not been included. Yet, if it had been included, the performance would have lagged significantly behind global benchmarks. Even without taking wastewater coverage into consideration, Africa’s top utilities provide water to 77 percent or more of the population in its service area, compared with a global bench- mark of 100 percent. In addition, a data envelopment analysis (DEA) was undertaken to measure the relative efficiency of utilities. The DEA creates a performance index from indicators—referred to as inputs and outputs in the DEA literature—that can be related to other factors that drive performance. Under basic DEA, a water utility is regarded as a relatively efficient utility if its observed inputs can be scaled to yield outputs that equal or exceed any combination or 12 Performance of Water Utilities in Africa Table 2.3. Customer Performance (Unweighted Average) Africa benchmark Global benchmark Indicator (best 25 percent of sample) (best 25 percent of sample) Population per connection (proxy for 8.3 3.0 service level) Reliability (hours of supply) 21.6 24 Affordability (%) 1.22 0.5 scaling of what other utilities’ observed inputs yield. Productive efficiency was assessed through DEA. Water billed was considered as the major output while number of staff and number of connections are considered as inputs. Institutional Performance Analysis The institutional assessment used the data panel for 2010 and 2013 because institutional performance is a variable that does not vary too widely over time. Collecting organizational data from utilities turned out to be difficult (that is, the response rate was very low). So a set of more general institutional data was used, such as type of service delivery (national, regional, or municipal service delivery), the presence of an (independent) regulatory agency, and the scope of services (that is, utility provides only water or provides multiple services). A simple t-test analysis was used to test several institutional factors of the utilities’ institutional context and their effect on a number of performance indicators. Drivers of Utility Performance Analysis The focus in this analysis is on three indicators of performance: the first one assesses utilities’ financial performance and is defined as a binary variable. More precisely, a utility is considered to be financially well-performing if its OCCR is equal to, or greater than, 1.19.3 The second indicator measures the quality of service. The so-called customer performance indicator is the average of three indicators as laid out above: population per connection (as proxy for service level with a level of less than 8.3 set as the African benchmark), reliabil- ity as measured by hours of water supplied (African benchmark of 21.6 hours per day or more), and affordability (with water costing consumers less than 1.22 percent of GNI). The customer performance index is measured on a 0–1 scale, with a higher value of the indicator indicating better performance. The third indicator, assessing utilities’ operational perfor- mance, is calculated as the average over three indicators: staff efficiency (takes the value 1 if equal to or higher than 4.21), metering (takes the value 1 if equal to 100 percent), and NRW (takes the value 1 if equal to or lower than 0.205 m3 per connection per day). Finally, the team used water coverage where a top-performing utility has a coverage of 77 percent or higher. The team found that the correlations between the four indicators are positive but very small. The lack of, or rather small, correlation between the four indexes indicates, Performance of Water Utilities in Africa 13 for instance, that being financially a good performer does not necessarily correlate with the fact, or imply, that the utility provides a good service to its customers. The performance indicators thus have to be analyzed separately because the drivers of utility perfor- mance  are  not closely correlated; hence, improvements in financial performance do not automatically translate to improvements in customer performance and vice versa. The team then tried to estimate the drivers of performance using econometric techniques. Case Studies In chapter 3, the team analyzed five case studies to determine whether the findings in the econometric analysis could be validated by the individual case studies. To ensure some sim- ilarity in case studies, the team was asked to undertake the same case studies as those undertaken by a similar study conducted by the Electricity Team—looking into the drivers of performance of electricity utilities in Africa. The case studies were those of Burkina Faso, Côte d’Ivoire, Kenya (Nairobi), Senegal, and Uganda. The selection of the case studies was not at random, and hence there are certain characteristics that differ from the sample as used in the econometric analysis. All these utilities are large, and serve more than million people (compared with the typical utility in the econometric database of about 1 ­ 115,000). The five utilities are all—with the exception of Nairobi—national utilities. National utilities are not very common in Africa and they tend to be mostly located in western Africa. Yet, all five case studies refer to large utilities. Because all these utilities provide services to the capital city, they tend to have a longer history than the typical utility in Africa. The case studies show how these five utilities are doing in terms of providing services to clients (to make the results comparable with those of the Electricity Study), while also triangulat- ing the results from the econometric models presented in chapter 6. Notes 1. The increase in access to improved water sources (which includes not only piped water, but also other forms of improved water services such as access to protected wells and springs, rainwater) as defined by UNICEF-WHO Joint Monitoring Programme saw a much sharper increase from 242 million in 1990 to 669 million in 2015. 2. Including Namibia and South Africa may skew the performance to such levels that are not necessarily realistic measures of good performance for utilities in other parts of Africa. 3. To avoid possible errors due to misreporting and the presence of outliers, we removed from the sample observations below the 2.5 percentile and above the 97.5 percentile in the OCCR distribution. Reference Danilenko, Alexander, Caroline van den Berg, Berta Macheve, and L. Joe Moffitt. 2014. The IBNET Water Supply and Sanitation Blue Book 2014: The International Benchmarking Network for Water and Sanitation Utilities Databook. Washington, DC: World Bank. 14 Performance of Water Utilities in Africa © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 3 Performance of Utilities in Africa: Trend Analysis The team undertook the analysis for the panel with the shorter time series but larger number of utilities. The trend analysis excludes utilities from Namibia and South Africa as these are upper-middle-income countries whose utilities perform at a different level than most, if not all, other utilities in Africa. When analyzing the data, three dimensions of performance will be examined: opera- tional, financial, and customer performance. Operational performance will look into how well the utility manages its operations. The financial performance is measured in terms of how effective the utility is in generating revenues from its operations, and using these rev- enues to cover its operation and maintenance (O&M) costs. Finally, customer performance is assessed. The objective of a utility is to provide customers with high-quality water ser- vices and (to a much lesser extent) wastewater services. The quality of its water services is measured by its ability to provide access to water users, but also the level of services it can provide to its customers. Operational Performance NRW. Operational performance of utilities is often measured using nonrevenue water (NRW). As can be seen in figure 3.1, NRW as a percentage of water production has declined between 2010 and 2013. Nevertheless, the NRW stands high compared with global benchmarks. but the figure shows the variation between utilities in keeping NRW under control. The NRW as Performance of Water Utilities in Africa 15 Figure 3.1. NRW as Percentage of Water Production a percentage of water production indicator has its draw- backs (Alegre 2006), so looking at another NRW indicator may be useful. This indicator measures NRW per connec- 0.8 tion per day; the typical utility experienced a similar decline that was mostly registered between 2010 and 2011 (see Nonrevenue water (%) 0.6 figure 3.2). Reducing NRW tends to be difficult (van den Berg 2015), 0.4 partially because of the distinct cost structure of water services. In the water sector, on average, production costs 0.2 tend to be low, while the costs of distribution are rela- tively high. Whittington et al. (2009) estimate that about 0 70 percent of the total costs of water supply are related to 2010 2011 2012 2013 the cost of distributing the water. Hence, every cubic meter not lost in the distribution system often has a rela- tively low opportunity cost; which may explain the stub- Figure 3.2. NRW, by Connection bornness of high levels of NRW in many utilities. Based on this finding, it is to be expected that utilities with high 3 operating costs or high water scarcity will be more likely to have incentives to decrease NRW than those that do not. And indeed, utilities that have high operating costs m3 per connection per day 2 per cubic meter of water sold have statistically signifi- cantly lower levels of NRW per connection per day than those that do not. 1 In general, larger utilities tend to have higher NRW losses than smaller utilities. Utilities serving more than 500,000 people showed an average NRW per connection of 0.49 m3 0 per day, compared with 0.31 m3 per day for the smallest utilities (see figure 3.3). This may reflect the age of the net- 2010 2011 2012 2013 work system; utilities started to develop in the capital cit- ies and then moved into secondary cities, hence the larger utilities tend to have been in existence longer and have more aged infrastructure. Using the dataset, it was found that in general utilities with 100 percent metering tend to have lower levels of NRW per connection per day. Equally important is the effect of continuous supply: 24 hours of supply tends to be associated with significantly lower levels of NRW per con- nection per day. Staff productivity (measured as number of staff per 1,000 connections) is low in Africa, although it is improving. In 2010, the typical utility registered 11 staff per 1,000 connections, which dropped to 8.7 in 2013. The staff productivity is especially low in the smallest utilities, with a median of almost 13 staff per 1,000 connections in 2013 compared with 6.4 ­ in the largest utilities. However, because many utilities in Africa provide low levels of service 16 Performance of Water Utilities in Africa Figure 3.3. NRW, by Size of Utility, 2010–13 a. Small utilities b. Medium utilities 3 3 m3 per connection per day m3 per connection per day 2 2 1 1 0 0 2010 2011 2012 2013 2010 2011 2012 2013 c. Large utilities 3 m3 per connection per day 2 1 0 2010 2011 2012 2013 and the number of people served per connection is very high, we also used staff productivity measured as number of staff per 1,000 people served so that utilities which provide different levels of service were not disadvantaged. The median staff per 1,000 people served stood1 at 0.72 in 2013, inching up from 2010 (when the ratio was 0.68). This is better than the global benchmark of 1.04 in 2010 (Danilenko et al. 2014). At the same time, the median annual staff costs are increasing. The data show the rapid increase in annual staff cost per employee from US$4,246 in 2010 to US$5,865 in 2013. In 2013 year, the bottom 10 percent of the sample spent US$2,305 or less per employee, but in the top 10 percent of the sample, utilities spent more than US$14,438 per employee. As a result, the share of labor in the total cost structure of the utilities is increasing. Figure 3.4 summarizes labor efficiency and annual staff cost by size of utility. Staff efficiency is measuring how much every U.S. dollar of staff costs translates into utility revenues generated. In 2010, the typical utility generated revenues at a ratio of 3.47. In 2013, the ratio had dropped to 3.04. Performance of Water Utilities in Africa 17 Figure 3.4. Labor Efficiency and Annual Staff Cost, by Size of Utility, 2010–13 a. Sta cost per employee b. Sta per 1,000 people served 14,000 1.40 12,000 1.20 Number of sta per 10,000 1.00 1,000 people US$ per year 8,000 0.80 6,000 0.60 4,000 0.40 2,000 0.20 0 0 00 0 00 0 0 0 00 0 00 0 0 0 00 00 00 00 00 00 00 00 ,0 ,0 ,0 ,0 0, 0, 0, 0, 0, 0, 0, 0, 10 00 10 00 –5 50 00 00 –5 50 00 00 < –1 < –1 10 10 0– 0– 1, 1, 1, 1, 50 50 0– 0– > 10 > 10 50 50 Number of residents in the area of service Number of residents in the area of service Financial Performance O&M Cost per Cubic Meter Sold. The median O&M costs per cubic meter sold increased from US$0.76 in 2010 to US$0.86 in 2013. This rapid increase seems to be mainly the result of higher labor costs, resulting from both a decline in staff productivity and an increase in staff cost per employee. The O&M cost per cubic meter sold is significantly higher in low-income countries in the sample, which may be the result of the smaller sample of reporting utilities in middle-income countries and/or the result of economies of scale in utilities in low-income countries (figure 3.5). When looking at the size of the utility and O&M cost, some evidence of economies of scale was found. The smallest utilities have the highest O&M cost per cubic meter sold. In the period between 2010 and 2013, the typical small utility had an O&M cost of US$0.96 per cubic meter, compared with US$0.71 for medium-size utilities and US$0.79 for large utilities (figure 3.6). Average Revenues per Cubic Meter of Water Sold. The increase in O&M cost per cubic meter of water sold is coinciding with rising average revenues per cubic meter sold from US$0.71 in 2010 to US$0.85 in 2013. Hence, utilities charge more when the O&M costs increase but not necessarily the full increase in O&M costs—hence, passing costs to consumers is not standard practice in Africa. Average revenues per cubic meter sold (proxy for tariff) is strongly correlated with O&M cost per cubic meter sold. Yet, average revenues per cubic meter sold is highest in small util- ities, but the trend over the last four years also shows that the typical utility saw minor 18 Performance of Water Utilities in Africa Figure 3.5. Median O&M Cost per m3 Sold, by Income Status, 2010–13 a. Low-income countries b. Middle-income countries 4 4 3 3 O&M cost (US$) O&M cost (US$) 2 2 1 1 0 0 2010 2011 2012 2013 2010 2011 2012 2013 Figure 3.6. Median O&M Cost per m3 Sold, by Size of Utility, 2010–13 a. Small utilities b. Medium utilities 4 4 3 3 O&M cost (US$) O&M cost (US$) 2 2 1 1 0 0 2010 2011 2012 2013 2010 2011 2012 2013 c. Large utilities 4 3 O&M cost (US$) 2 1 0 2010 2011 2012 2013 Performance of Water Utilities in Africa 19 changes in the average revenues per cubic meter. Although the tariffs in the typical medium and large utilities have increased more significantly, they are below the tariff levels that the smallest utilities charge their customers. Although tariffs can be high, the effect of high tariffs can be partially mitigated by lower lev- els of consumption (and collection). In addition, most utilities use cross-subsidies—­sometimes quite high levels of cross-subsidies—where nonresidential water users usually have a more price elastic demand for water and pay significantly higher tariffs than residential water con- sumers. Hence, to understand how well the utility is capable Figure 3.7. Median Annual Revenues per Capita, 2010–13 of generating revenues, the annual revenues generated per person served were examined. As can be seen from figure 3.7, the average annual revenues per capita are less than US$15. 100 This is a very small revenue base for a utility. Figure 3.8 show a significant variation in the revenue per connection in low-­ income countries due to structure of connections, when up Revenue (US$) to 200 people can be served from one connection. 50 OCCR. The typical utility in the sample is able to cover its O&M costs with its operating revenues. Yet, the “cash flow” operating cost coverage ratio (OCCR) (defined here as the dif- 0 ference between collected revenue and O&M cost) is signifi- 2010 2011 2012 2013 cantly lower because utilities can only collect part of their Figure 3.8. Median Annual Revenues per Connection, by Income Status, 2010–13 a. Low-income countries b. Middle-income countries 1,000 1,000 800 800 Revenue (US$) Revenue (US$) 600 600 400 400 200 200 0 0 2010 2011 2012 2013 2010 2011 2012 2013 20 Performance of Water Utilities in Africa Figure 3.9. Operating Cost Recovery versus Cash Flow Operating Cost Recovery, revenues, as can be seen in 2010–13 figure  3.9. Operating (O&M) cost recovery and cash flow operating 2.5 cost recovery are more compli- cated in smaller utilities than 2.0 in larger utilities. Smaller utilities typically charge higher water rates than larger utilities, and water in 1.5 such utilities tends to be much OCCR less affordable than in larger utili- 1.0 ties. The actual financial problems are, hence, rather serious as col- 0.5 lection efficiencies  were around 80 percent of total revenues, meaning that the utilities’ cash 0 inflow is negative and hence the 2010 2011 2012 2013 utility has to (a) decrease service Revenue-based OCCR Cash-based OCCR ­ ostpone maintenance; levels; (b) p (c) postpone ­payments to suppliers; and/or (d) if available, increase its dependence on govern- ment subsidies. The figure 3.10 presents that information for countries by income status. Figure 3.10. Operating Cost Recovery versus Cash Flow Operating Cost Recovery, by Income Status, 2010–13 a. Low-income countries b. Middle-income countries 2.5 2.5 2.0 2.0 1.5 1.5 OCCR OCCR 1.0 1.0 0.5 0.5 0 0 2010 2011 2012 2013 2010 2011 2012 2013 Revenue-based OCCR Cash-based OCCR Performance of Water Utilities in Africa 21 Table 3.1. Cost Recovery among Bottom- and Top-Performing Utilities Variable Bottom 10 percent Bottom 25 percent Median Top 25 percent Top 10 percent Operating cost recovery (billed revenues as % of O&M costs) Low income 0.73 0.88 1.00 1.15 1.36 Middle income 0.79 0.83 1.05 1.36 1.44 Cash-flow-based operating cost recovery (collected revenues as % of O&M costs) Low income 0.42 0.59 0.79 0.95 1.102 Middle income 0.60 0.76 0.92 1.13 1.33 Table 3.1 shows how the bottom and top performers among utilities are doing. There are two important observations. The first is that operating cost recovery rates do not really change when income levels increase. Even the best-performing utilities do not cover more than about 140 percent of total O&M costs. This is standard around the world—whether ana- lyzing utilities in developing or developed countries. The Blue Book 2014 confirmed this trend of more or less stagnant levels of cost recovery also at the global level. This means that most utilities cover only basic O&M costs plus a little extra. The second observation is that this operating cost recovery ratio is—even when looking at the top performers—very low in relation to the cost structure of water and wastewater services. Whittington et al. (2009) estimate that the economic cost of conventional water and wastewater infrastructure is US$2.50 per cubic meter2 (at a discount rate of 10 percent3 in 2006 prices), assuming very low opportunity cost of raw water and very limited external- ities associated with the discharge of treated wastewater,4 which would translate to more than US$3 in 2013 prices. Hence, with typical O&M cost of water and wastewater of US$0.86, operating cost recovery is likely to cover only a fraction of the actual economic costs of water services. Even the best performers are very far from being able to cover the financial and economic cost of the services. This is not just a trend happening in Sub-Saharan Africa; it is common around the world. Water supply services are provided in a complex environment in which various objec- tives in the provision of water and wastewater services are at play. With low levels of per capita income, people are often too poor to actually consume much piped water (let alone pay for wastewater collection) and this lack of capacity results in a very small revenue basis for the utility. This affects financial performance and makes utilities highly depen- dent on government funding to pay for part of the O&M costs, crowding out investments and slowing the growth in access to (quality) water (and wastewater) services. As incomes rise, more can be spent on water (in absolute terms), which improves the financial footing for utilities. However, this is not what we see in our database, possibly because of the lim- ited number of utilities in middle-income countries. Even though water consumption lev- els are going up when countries’ gross domestic product is increasing, the total revenues per capita do not see an increase despite these higher water consumption levels. 22 Performance of Water Utilities in Africa Yet, higher incomes result in better collection of revenues as the difference between the OCCR measured in billed and collected revenues is smaller in middle-income countries than in low-income countries. At the same time, as economic development accelerates and water consumption increases, the nature of the service changes. In low-income countries, water utilities mainly provide water supply services with little emphasis on dealing with the wastewater flows that accompany the provision of water supply services. Yet, in middle-income countries, wastewater collection becomes a more important additional service as higher incomes mostly coincide with higher per capita water consumption. When countries grow even richer, this collected wastewater is to be treated and disposed of against increasingly high environmental standards. Danilenko et al. (2014) found that the OCCR—a sign of financial health of utilities—in a global sample (of more than 1,800 utilities) barely changed between 2000 and 2011 (even though countries’ economies grew rapidly every- where in the world). Customer Performance A rapid trend analysis over the period 2010 to 2013 shows that utilities in Africa saw an increase in water coverage (see figure 3.11). This is not in line with the data provided by the JMP that saw at best a stagnation of piped water coverage. Yet, it should be taken into account that the service area of a utility may not coincide with the urban areas as defined Figure 3.11. Median Water Coverage, 2010–13 a. Low-income countries b. Middle-income countries 1.0 1.0 0.8 0.8 Water coverage (ratio) Water coverage (ratio) 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2010 2011 2012 2013 2010 2011 2012 2013 Performance of Water Utilities in Africa 23 Figure 3.12. Median Water Coverage, by Size of Utility by the household surveys. Service area boundaries may— adjust much especially in a period of rapid urbanization—­ 90 slower than the pace of urbanization. In addition, utilities 80 manage connections, without often having a good insight 70 into the use of these connections. Hence, the household 60 surveys and utility data often differ. Coverage (%) 50 Although the utilities in the sample have seen an 40 increase in coverage, this increase in uptake has been 30 rather uneven between utilities and between countries in 20 the sample. 10 As can be seen in figure 3.12, the smaller the utility, the 0 lower the water coverage—with the exception of the larg- est utilities. This phenomenon is linked to the fact that 00 00 00 0 00 0 00 00 ,0 ,0 ,0 ,0 0, 0, smaller utilities tend to be more recently established and 00 00 10 00 –5 50 1,0 ,0 < –1 10 0– 50 >1 0– hence tend to have lower coverage rates. The increase in 10 50 Population of the service area water coverage has been accompanied by a slow decline in wastewater coverage for those utilities that provide the service (which is only a small number of utilities). This is a challenge because wastewater coverage already lagged far behind that of water coverage—with possible adverse effects on water quality in the region as the lack of wastewater coverage increases public health risks and the occurrence of water pollution. The slow increase in water coverage also coincided with very modest progress in the quality of services provided. The decline in wastewater coverage, noted above, is one man- ifestation of the deficits in service quality, but the quality of service challenges are also reflected in (a) the low levels of per capita water consumption (at around 56 liters per cap- ita per day [lcd]),5 (b) the drop in hours of water supplied, and (c) the very modest progress in improvement in service levels (the number of people per connection declined slowly to 10.6). The low levels of water consumption are especially prevalent in low-income coun- tries and are linked to the common practice of sharing water connections. This sharing of connections is the result of both supply and demand constraints. The prevalence of increasing block rates in tariff setting will not provide consumers with much of an incen- tive to consume large quantities of water (even if the water services are available). As water metering is fairly widespread in the sample of utilities, consumers are able to have some control over their water use and, hence, may restrict consumption to avoid paying the higher block tariffs. Figure 3.13 shows the very low levels of total consumption in utilities in low-income countries. Even in the upper 10 percent (top decile) of utilities, consumers use only 100 lcd. The World Health Organization assumes that a consumption of at least 50 to 100 lcd is needed for all residential water uses. In 2013, slightly under 50 percent of the households 24 Performance of Water Utilities in Africa Figure 3.13. Median Water Consumption, by Income Status, 2010–13 a. Low-income countries b. Middle-income countries 250 250 200 200 Water consumption (lcd) Water consumption (lcd) 150 150 100 100 50 50 0 0 2010 2011 2012 2013 2010 2011 2012 2013 using utility services used 50 lcd of water or less and 25 percent did not exceed consumption levels of 77 lcd. This low level of consumption means that the revenue-generating potential of many utilities is small, especially as overall water coverage is low. The typical consumer (both residential and nonresidential) in the sample of utilities pays less than US$15 per year for the service. Yet, the overall cost of producing water is mostly fixed, leaving the utilities often cash-strapped. Supply constraints also play a role: most utilities have only limited production capacity— assuming that water production is a reflection of the capacity to produce water (in 2013, median water production was 88 lcd compared with 96 lcd in 2010). As can be seen in ­ figure 3.14, production in low-income countries is in general very low—below 100 lcd—and is slowly declining. In middle-income countries, the production is much higher: closer to 200 lcd. The production levels in utilities in low-income countries are very low in compari- son with other countries and utilities, and explain partially the low consumption levels in many utilities in the sample. Service Levels. The population per connection is high in Africa; much higher than the global benchmark of three people per connection. The median number of people per connection was around 11 in 2013. But there are large differences—with the best-performing utilities Performance of Water Utilities in Africa 25 Figure 3.14. Median Water Production, by Country, 2010–13 a. Benin b. Burkina Faso c. Congo, Dem. Rep. Water production (lcd) Water production (lcd) Water production (lcd) 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 d. Côte d’Ivoire e. Guinea-Bissau f. Kenya Water production (lcd) Water production (lcd) Water production (lcd) 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 g. Malawi h. Mozambique i. Niger Water production (lcd) Water production (lcd) Water production (lcd) 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 j. Senegal k. Tanzania l. Togo Water production (lcd) Water production (lcd) Water production (lcd) 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 m. Uganda n. Zambia Water production (lcd) Water production (lcd) 400 400 300 300 200 200 100 100 0 0 2010 2011 2012 2013 2010 2011 2012 2013 26 Performance of Water Utilities in Africa Figure 3.15. Population per Connection, 2010–13 providing services to about six people per connection and the worst mainly providing services through standposts 80 and kiosks as can be seen in figure 3.15. Number of people per connection 60 Hours of Supply. Many utilities in Africa ration water. A typ- ical utility only provides water supply for 15 hours per day 40 (compared with 18 in 2010). The smallest utilities provide the least hours of water supply and the largest utilities pro- vide the most (figure 3.16). The progress is rather uneven 20 between utilities; the largest utilities show improvements, whereas the medium-size utilities show a reduction in the 0 median hours of water supplied. 2010 2011 2012 2013 Figure 3.16. Hours of Water Supplied per Day, by Utility Size, 2010–13 a. Small utilities b. Medium utilities 25 25 Hours of water per day Hours of water per day 20 20 15 15 10 10 5 5 0 0 2010 2011 2012 2013 2010 2011 2012 2013 c. Large utilities 25 Hours of water per day 20 15 10 5 0 2010 2011 2012 2013 Performance of Water Utilities in Africa 27 Affordability. In recent years, affordability improved and people spent proportionally less on water (and wastewater) services. This trend also occurred in Sub-Saharan Africa. In 2013, users served by utilities in low-income countries spent 1.8 percent of income compared with 0.9 percent in middle-income countries. As noted by the Africa Country Diagnostic Surveys, service provision in Sub-Saharan Africa tends to be significantly more expensive than else- where in the world. In 2011, median affordability in low-income countries (including Africa) was 0.82 percent compared with 2.0 percent in Sub-Saharan Africa; in that same year (2011), the median O&M cost in Africa stood at US$0.83 compared with US$0.68 for all low-income countries. A t-test analysis shows that utilities with good financial performance tend to have lower levels of affordability. Utilities with good financial performance registered that on aver- age customers spent 3.2 percent of GNI per capita on water supply services compared with 2.3 percent for those that had weak financial performance (figure 3.17). This effect was noticeable in small- and medium-size utilities: consumers are paying a larger share of  their income on services. The opposite held true for the largest utilities; the better-­ performing utilities are able to charge customers on average smaller shares of their income on water services (figure 3.18). Median affordability for different countries presented in table 3.2. ­ Figure 3.17. Median Affordability, 2010–13 a. Low-income countries b. Middle-income countries 0.15 0.15 A ordability, ratio of water bill to GNI (Atlas) A ordability, ratio of water bill to GNI (Atlas) 0.10 0.10 0.05 0.05 0 0 2010 2011 2012 2013 2010 2011 2012 2013 28 Performance of Water Utilities in Africa Figure 3.18. Median Affordability, by Size of Utility, 2010–13 a. Small utilities b. Medium utilities Ratio of water bill to GNI (Atlas) Ratio of water bill to GNI (Atlas) 0.15 0.15 0.10 0.10 0.05 0.05 0 0 2010 2011 2012 2013 2010 2011 2012 2013 c. Large utilities Ratio of water bill to GNI (Atlas) 0.15 0.10 0.05 0 2010 2011 2012 2013 Table 3.2. Median Affordability, by Country, 2010–13 Median revenues per cubic meter Median annual revenues Median affordability Country sold (US$), proxy for tariff Per connection (% of country’s GNI per capita) Benin 1.20 201 2.4 Burkina Faso 1.10 269 3.2 Congo, Dem. Rep. 0.75 440 5.2 Côte d’Ivoire 0.82 171 0.9 Kenya 1.01 173 1.7 Malawi 0.83 238 9.2 Mali 0.64 279 2.6 Mozambique 0.70 122 3.9 Niger 0.60 234 1.8 Senegal 1.09 266 2.3 Tanzania 0.43 98 1.7 Togo 0.64 215 2.2 Uganda 1.14 207 4.7 Zambia 0.46 185 1.1 Performance of Water Utilities in Africa 29 Conclusions Coverage has increased slowly between 2010 and 2013. Overall coverage rates are slightly above 60 percent in the service areas of the utilities covered in the sample, suggesting that the utilities are not able to serve large parts of the population in the areas they are responsi- ble for. The combination of relatively high O&M costs of water (and even more so wastewater) services and the increasing affordability of water services suggest that, in many utilities, even in Sub-Saharan Africa, there is some scope for balancing the goals of revenue suffi- ciency and affordability more in favor of the former as government subsidies will otherwise need to increase rapidly in some countries. A little less than half the utilities in the sample are not able to cover their operating and maintenance costs through their revenues. The high dependence on subsidies raises ques- tions about the equity of such subsidies that tend to disadvantage those not yet connected to piped network services.6 The variance of utilities’ performance within and between countries is very large. This is, for the most part, because water and wastewater services are locally provided. These local factors can vary widely between utilities and include factors such as the distance to the water source and the effect on the cost to store and transport water, the quality of the water source and the need for treatment, and design standards, among others. Yet, more general policies in the country (for example, energy subsidies and labor policies) also affect the cost of O&M. The O&M costs vary widely in the sample, from less than US$0.23 to more than US$2.07 per m3 sold. The actual O&M costs can vary even more when utilities with less than four years of data are included in the sample. In that case, O&M costs vary from US$0.12 per m3 of water sold in Nigeria to more than US$4.75 in Cape Verde (because of dependence on desalinated water). The implication of this large variance on performance is that specific local circum- stances have a major impact on revenue sufficiency and affordability. Policy makers will need to understand how the variation in the costs of water and wastewater services affects the balance between the objectives of revenue sufficiency and affordability7 in their pric- ing policies. This will require much more information and research than is currently available. Finally, as the fixed cost component in water and wastewater service provision is very large, the design (and other) standards under which water infrastructure is constructed will determine the cost of the services for decades to come. Hence, it is important to undertake a proper least-cost analysis when investment decisions are made to ensure that the benefits and costs of such investments are properly analyzed. The financial, orga- nizational, and social implications of such investment decisions will be felt for many decades. 30 Performance of Water Utilities in Africa Notes 1. Owing to the high incidence of shared connections, through house connections and public taps, the often-used indicator of staff per 1,000 connections is not used. In addition, as the number of connections was not collected systematically, the actual number of utilities providing this information results in biased results. 2. It should be noted that these estimates are highly variable. For instance, in places where water has to be hauled over long distances or where water scarcity is a reality, the costs can be significantly higher. 3. This assumes that, for utilities, financing is available against international market rates, whereas it is also assumed that no particular country risks are included. 4. There is some trade-off between wastewater treatment costs and externalities associated with discharge of treated waste- water. The higher the levels of wastewater treatment, the lower the externalities associated with the discharge of treated wastewater. 5. Note that this refers to total water consumption (including consumption from all types of consumers). Residential water consumption data, where available, suggest that consumption is significantly lower than the 56 lcd. 6. In most developing countries, those not connected to the piped network services are disproportionally poorer citizens or disadvantaged future generations (Komives et al. 2005). 7. And any other policy objectives that are to be included, such as economic efficiency, transparency, and so on. References Alegre, Helen. 2006. Performance Indicators for Water Supply Services. London: IWA Publishing. Danilenko, Alexander, Caroline van den Berg, Berta Macheve, and L. Joe Moffitt. 2014. The IBNET Water Supply and Sanitation Blue Book 2014: The International Benchmarking Network for Water and Sanitation Utilities Databook. Washington, DC: World Bank. Komives, Kristin, Vivien Foster, Jonathan Halpern, Quentin Wodon, and Roohi Abdullah. 2005. Water, Electricity and the Poor: Who Benefits from Utility Subsidies? Washington, DC: World Bank. van den Berg, Caroline. 2015. “Drivers of Non-Revenue Water: A Cross-National Analysis.” Utilities Policy 36 (C): 71–78. Whittington, D., W. M. Hanemann, C. Sadoff, and M. Jeuland. 2009. “The Challenge of Improving Water and Sanitation Services in Less Developed Countries.” Foundations and Trends in Microeconomics 4 (6–7): 469–609. Performance of Water Utilities in Africa 31 © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 4 Performance of Utilities in Africa: Composite Performance Index This analysis used the same sample as the previous analysis.1 Three composite performance indicators that, respectively, measure different aspects of customer performance, financial performance, and operational performance are used. The theory suggests that good opera- tional performance translates into better financial performance because of a reduction of the operation and maintenance (O&M) cost. Good financial performance allows the utility to generate excess funds that could—once the utility generates sufficient cash to pay for O&M costs, depreciation, and debt service—be used for improvements to service levels or water coverage (see figure 4.1). The hypothesis is that once the customer experiences improved service level, the utility will be able to generate more revenues that then can fund further improvements in operational and/or customer performance. Calculation of the Composite Performance Index For each of the two performance indexes measuring operational and customer performance, we calculated a composite index that provides some insight into the various aspects of performance. The major reason for setting up a composite index is to gain a wider perspec- ­ tive of the situation, while being aware of the pros and cons of using composite performance indexes (for more details, see box 4.1). The criteria for a well-performing utility with regard to operational performance relate to behaviors that are under management control, including metering, nonrevenue water (NRW) Performance of Water Utilities in Africa 33 Figure 4.1. Performance Indicators (as measured by NRW per connection per day), and staff efficiency (measuring the revenue generated as a pro- portion of cost per employee). Customer performance is  measured by service level quality (population per con- nection: the higher the number of people per connection, the lower the service level as there is more dependence on sharing connections, standposts, and kiosks), reliability Operational (number of hours that water is supplied), and affordability,2 performance which is mostly under management control. We did not include water coverage (or water consumption) as these indicators are highly dependent on investment infrastruc- ture that may or may not be funded by government. For each variable, the value was calculated for the best-performing quartile of utilities, and then the variance with this threshold was calculated (see table 4.1). The larger Financial the deviance toward the well-performing threshold, the performance lower the value. No specific weights were given to the dif- ferent indicators. In case of missing information on any one of the underlying indicators, the utility’s performance is not reported; only utilities that have a complete set of data to calculate the indicators were included. The maximum value that a utility can achieve on the indicator is 1. In theory, the lowest value is zero, but in practice utilities will always pro- Customer duce some level of service. Yet, as can as can be seen in performance ­ figure 4.2, the minimum operational performance using African benchmarks is set at only about 0.10, with a maxi- mum value of 1.00. Hence, there is a wide variation in oper- ational performance between the utilities in the sample. Composite Performance Index Operational Performance As can be seen in figures 4.2 and 4.3, the typical utility saw an improvement in operational performance. The variation between utilities is large. This improvement coincides with a slight decrease in standard deviation, with worse-performing utilities seeing operational performance increase more than better-performing utilities A full list of operational perfor- mance and the underlying indicators is provided in appendix C. As can be seen in figures 4.2 and 4.3, African utilities have a reasonably good opera- tional performance but when using a global benchmark, African utilities perform slightly worse. The typical utility in the Africa sample had an operational performance score of 0.71 (African benchmark) and 0.64 (global benchmark) in 2013, which means that most 34 Performance of Water Utilities in Africa Box 4.1. Pros and Cons of Composite Performance Indexes Pros Cons They can summarize complex, multi-dimensional They may be misinterpreted realities with a view to supporting decision makers They can provide the “big picture” They may require more data They are easier to interpret than a large set of The selection of indicators and weights could individual indicators be the subject of political debate They reduce the size of indicators to analyze They will require judgment (and hence can without dropping the underlying information bring in some subjectivity) They can assess performance over time Table 4.1. Setting Benchmarks Best-performing quartile Best-performing quartile Indicator of African utilities in of global utilities as per sample Blue Book Operational performance Metering (%) 100 100 NRW per connection per day (in m3) 0.205 0.121 Staff efficiency (revenues per staff employee as ratio of 4.21 4.27 cost per staff employee) Financial performance OCCR 1.19 1.38 Customer performance Population per connection (as proxy for service levels) 8.3 3.0 Reliability (hours of supply) 21.6 24 Affordability (%) 1.22 0.5 Other Water coverage (%) 77 100 Water consumption (lcd) 76.6 220 utilities are working on their operational performance. However, the best African utilities are able to meet global benchmarks in operational performance. In general, African benchmarks are about 10 percent below global benchmarks when evaluating operational performance. When disaggregating operational performance trends by country (again using the subset of utilities with data from 2010 to 2013), 116 utilities in 14 countries can provide operational performance data. As can be expected, national utilities show much less variation than utilities that are using regional or district service delivered approaches; these utilities tend ­ Performance of Water Utilities in Africa 35 Figure 4.2. Operational Performance Index against African to give very high operational performance—with values Benchmark, 2010–13 close to 1.000. The countries with regional utilities (Kenya, Malawi, Mozambique, Tanzania, and Zambia) show much 1.0 more variation in operational performance. They may register very high values (with the exception of Zambia) ­ Operational performance index 0.8 but  they also register utilities with the worst operational performance. 0.6 As can be seen in figure 4.4 (for a more detailed list, see appendix A), well-performing utilities can be found in dif- 0.4 ferent countries, working under different types of regula- tory frameworks and at different levels of economic 0.2 development. For those utilities that consistently provide data between 2010 and 2013, the change in the composite 0 operational performance index varies significantly. The 2010 2011 2012 2013 fastest-improving utilities over that period are found in Kenya, but this country also shows the largest variation in Figure 4.3. Operational Performance Index against Global operational performance. Most national utilities—with the Benchmark, 2010–13 exception of Benin—show a very stable performance over the observation period. 1.0 Financial Performance Operational performance index 0.8 figures 4.5 Operating Cost Coverage Ratio. As can be seen in ­ and 4.6, the typical utility saw a slow decline in finan- 0.6 cial performance. The variation between utilities is large. A  full list of financial performance is provided in 0.4 appendix A. Yet, the top 25 percent of utilities are able to ­ obtain a perfect score of 1.000 (equivalent to an operating 0.2 cost coverage ratio [OCCR] of 1.19). It is interesting to note that the majority of the utilities are clustered. This may 0 point to the fact that most utilities are located in countries 2010 2011 2012 2013 that are poor. Using a global benchmark of 1.38, the top performers in Africa can meet global benchmarks. Yet, as was shown in the composite operational performance index, African top performers can meet global benchmarks, but the typical African utility is not performing as well as its global counterparts. When disaggregating financial performance trends by country (again using the subset of utilities with data from 2010 to 2013), data from 118 utilities in 13 countries are available. Different countries register variations in financial performance improvements. In some 36 Performance of Water Utilities in Africa Figure 4.4. Median Operational Performance, by Country, 2010–13 a. Benin b. Burkina Faso c. Congo, Dem. Rep. 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Operational Operational Operational 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 d. Côte d’Ivoire e. Guinea-Bissau f. Kenya 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Operational Operational Operational 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 g. Malawi h. Mozambique i. Niger 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Operational Operational Operational 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 j. Senegal k. Tanzania l. Togo 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Operational Operational Operational 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 m. Uganda n. Zambia 1.0 1.0 performance index performance index 0.8 0.8 Operational Operational 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2010 2011 2012 2013 2010 2011 2012 2013 Performance of Water Utilities in Africa 37 Figure 4.5. Financial Performance Index Using the countries, utilities see an improvement in financial perfor- Operating Cost Coverage Ratio against African Benchmark, mance, in others a decline in financial performance, and 2010–13 again in others performance is rather stable (figure 4.7). 1.0 Cash-Based Operating Cost Coverage Ratio. This ratio is a variation on the normally used OCCR, and instead of 0.8 Operating cost coverage ratio using billed revenues as a percentage of O&M costs, it uses 0.6 collected revenues as a percentage of O&M costs (see ­ figure  4.8). Utilities in middle-income countries show ­ 0.4 ­ better performance than utilities in low-income countries ­(figure 4.9): they tend to have overall higher levels of cash- 0.2 based financial performance, but also much less variation within the sample suggesting that affordability can be a 0 challenge at times. 2010 2011 2012 2013 Customer Performance Customer performance is the ultimate goal of each utility. Figure 4.6.Financial Performance Index Using the As can be seen in figure 4.10 the typical utility did not see Operating Cost Coverage Ratio against Global Benchmark, 2010–13 much change in this indicator between 2010 and 2013. The variation between utilities is large. A full list of customer 1.0 performance, including its underlying indicators, is pro- vided in appendix C. 0.8 quality African utilities lag behind in the provision of high-­ Operating cost coverage ratio water services (see figure 4.10). The type of services 0.6 provided with regard to hours of supply, affordability, and ­ service levels (high dependence on shared connections) 0.4 is  in stark contrast with global benchmarks. Even the better-performing utilities in Africa are far from achieving ­ 0.2 global benchmarks. Break-down by the country income is presented in figure 4.11. 0 When disaggregating absolute performance trends by 2010 2011 2012 2013 country (again using the subset of utilities with data from 2010 to 2013), it is clear that the well-performing utilities on this aspect of performance tend to concentrate in countries with a larger number of utilities, most notably Kenya, Tanzania, and Zambia. The national utilities tend to showcase lower levels of customer performance (figure 4.12). In general, there are more utilities that provide better financial and operational perfor- mance than those that provide good customer performance. This means that the translation of better operational and financial performance into better customer performance is not automatic. 38 Performance of Water Utilities in Africa Figure 4.7. Median Financial Performance Index, by Country, 2010–13 a. Benin b. Burkina Faso c. Congo, Dem. Rep. 1.0 1.0 1.0 cost coverage ratio cost coverage ratio cost coverage ratio 0.8 0.8 0.8 Operating Operating Operating 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 d. Côte d’Ivoire e. Kenya f. Malawi 1.0 1.0 1.0 cost coverage ratio cost coverage ratio cost coverage ratio 0.8 0.8 0.8 Operating Operating Operating 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 g. Mozambique h. Niger i. Senegal 1.0 1.0 1.0 cost coverage ratio cost coverage ratio cost coverage ratio 0.8 0.8 0.8 Operating Operating Operating 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 j. Tanzania k. Togo l. Uganda 1.0 1.0 1.0 cost coverage ratio cost coverage ratio cost coverage ratio 0.8 0.8 0.8 Operating Operating Operating 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 m. Zambia 1.0 cost coverage ratio 0.8 Operating 0.6 0.4 0.2 0 2010 2011 2012 2013 Performance of Water Utilities in Africa 39 Figure 4.8. Median Financial Performance Index based on Some analysis using t-tests between good financial per- Cash-Based Operating Cost Coverage Ratio, 2010–13 formance and customer performance shows that most of the proxies for service level (that is, hours of supply and 1.0 people per connection) do not differ between utilities with Cash-based operating cost coverage ratio good financial performance and those that are operating 0.8 with lower OCCRs. Water coverage is not higher when utili- ties are better able to cover their O&M costs through their revenues. Yet, utilities with better financial performance 0.6 are able to generate more revenues because of significantly higher water tariffs; this may be linked to demand patterns 0.4 because utilities in Africa show large variations in the dependence on different types of consumers and the use of cross-subsidies in their tariff structures (see box 4.2). 0.2 Although higher water tariff revenues are not always 2010 2011 2012 2013 ­ collected—as collection efficiencies tend to decline with Figure 4.9. Median Financial Performance Index based on Cash-Based Operating Cost Coverage Ratio, by Income Status, 2010–13 a. Low-income countries b. Middle-income countries 1.0 1.0 Cash-based operating cost coverage ratio Cash-based operating cost coverage ratio 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 40 Performance of Water Utilities in Africa Figure 4.10. Customer Performance Index against African and Global Benchmarks, tariff levels—the effective water 2010–13 rates are still significantly above those of the utilities that apply 1.0 lower water tariffs. Hence, using not only tariff levels but also tariff 0.8 structures as a tool to kick-start Customer performance index financial performance against a 0.6 backdrop of poor customer per- formance may be useful depend- ing on the types of customers 0.4 utilities are serving. Many utili- ties do not provide much detail 0.2 about their clients and their demand patterns, but this is key 0 information that a utility should 2010 2011 2012 2013 have to better plan and manage African benchmark Global benchmark its performance. Figure 4.11. Customer Performance Index against Africa Benchmark, by Income Status, 2010–13 a. Low-income countries b. Middle-income countries 1.0 1.0 0.8 0.8 Customer performance index Customer performance index 0.6 0.6 0.4 0.4 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 Performance of Water Utilities in Africa 41 Figure 4.12. Customer Performance Index against African Benchmark, by Country, 2010–13 a. Benin b. Burkina Faso c. Congo, Dem. Rep. 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Customer Customer Customer 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 d. Côte d’Ivoire e. Guinea-Bissau f. Kenya 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Customer Customer Customer 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 g. Malawi h. Mozambique i. Niger 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Customer Customer Customer 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 j. Senegal k. Tanzania l. Togo 1.0 1.0 1.0 performance index performance index performance index 0.8 0.8 0.8 Customer Customer Customer 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 m. Uganda n. Zambia 1.0 1.0 performance index performance index 0.8 0.8 Customer Customer 0.6 0.6 0.4 0.4 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 42 Performance of Water Utilities in Africa Box 4.2. Cross-Subsidies in SDE and ONEA Senegal’s increasing block tariff structure has a subsidized social tariff for levels of consumption below 20 m3 (CFAF 202; US$0.40) per two months. There is also a regu- lar tariff for consumption from 21 m3 to 40 m3 (CFAF 697.97; US$1.39), and a “dissua- sive” or discouraging tariff for consumption above 40 m3 (CFAF 878.35; US$1.75). The dissuasive tariff is designed to be a disincentive for excessive water use. It can be seen that the tariff for household consumption of less than 20 m3 per 60 days is less than a third of the regular tariff, and less than a quarter of the tariff for consumption in the top block. Bills are sent every two months based on meter readings, and Sénégalaise des Eaux (Senegal) (SDE) can cut off water supply for nonpayment.a Nonresidential, nongovernmental customers must pay the higher tariff regardless of amount consumed. As of 2013, just 7 percent of SDE customers were classified as non- residential, down from 33 percent in 2004. High tariffs could be a contributing factor to this trend. Government customers pay more than twice the high tariff—their tariff is CFAF 1,868.88 per m3 (US$3.72 per m3). The structure has been like this since 2007. In that year, the government agreed to raise tariffs for government customers by 70 per- cent, while keeping tariffs for other customers constant. This was introduced as a way to keep domestic tariffs from rising while still ensuring cost recovery for Société Nationale des Eaux du Sénégal (SONES) and SDE. In 2015, domestic tariffs were also raised—the lowest tariff block was raised by 4 percent and other rates were increased by 9 percent. Office National de l’Eau et de l’Assainissement (Burkina Faso) (ONEA) subsidizes consumption for basic needs by charging much higher tariffs for what it regards to be excessive consumption. ONEA’s tariff for the first consumption block (up to 8 m3 per month) is just 18 percent of the tariff for consumption in excess of 30 m3. This subsi- dized tariff is US$0.39 per m3, compared with US$2.16 per m3 for consumption above 30 m3. The latter is designed to be a disincentive for excessive water use. There are two other residential tariff blocks—above 8 m3 and up to 15 m3 (US$0.89), and above 15 m3 and up to 30 m3 (US$1.06). The standpipe tariff is equal to the basic needs tariff (US$0.39 per m3).b Nonresidential tariffs help cross-subsidize. They are set at US$2.16 per m3 regardless of amount consumed. Industrial customers account for 5 percent of consumption by volume. a. Every 2 months. http://sde.sn/Pages/Votre-facture-a-la-loupe.aspx. b. ONEA, “Les tarifs,” http://oneabf.com/les-tarifs/. Water Coverage When water coverage is also taken into account, the picture does not change significantly. Because of lower access to piped water coverage, Africa is lagging behind global benchmarks (figure 4.13). Progress is made—faster than in other dimensions of performance. It can be seen again that the variation within countries can be very large as a result of more decentral- ized service delivery and the inclusion of many less-established utilities in the sample Performance of Water Utilities in Africa 43 Figure 4.13. Water Coverage Index against African and Global Benchmarks, 2010–13 a. Low-income countries b. Middle-income countries 1.0 1.0 0.8 0.8 Water coverage index Water coverage index 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2010 2011 2012 2013 2010 2011 2012 2013 Coverage against African benchmark Global benchmark (see  figures 4.14 and 4.15). Newly established, often smaller utilities tend to have much lower coverage rates than utilities that have been in business for decades, which shows essentially that the development of utilities and the provision of piped water started later in Africa than in other parts of the world. The overall performance index shows that performance has improved marginally between 2010 and 2013 (figure 4.16). In tables 4.2 and 4.3, the best-performing utilities with regard to operational, financial, and customer performance are shown for the sample for 2013 by type of service delivery. The results show that some of the best-performing utilities in Africa are the ones you have probably never heard of. It should be noted that different definitions of performance will result in very different lists of well-performing utilities. Different definitions of performance will result in different rankings whereas the use of  weightings for different forms of performance will also result in different rankings. Yet, many utilities—even among the best-performing—show some variation in perfor- mance, not only between utilities but also within the same utility over time. This points to utilities being relatively vulnerable to factors that can affect their performance quite dramatically. As can be seen in table 4.3 utility performance also varies between smaller and larger util- ities. Smaller utilities typically have lower scores than larger utilities, but once utilities are much bigger that benefit becomes less obvious. We will look into the importance of econo- mies of scale in water supply service delivery in the chapter 5. 44 Performance of Water Utilities in Africa Figure 4.14. Water Coverage Index against Global Benchmark, 2010–13 a. Small utilities b. Medium utilities 1.0 1.0 Water coverage index Water coverage index 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2010 2011 2012 2013 2010 2011 2012 2013 c. Large utilities 1.0 Water coverage index 0.8 0.6 0.4 0.2 0 2010 2011 2012 2013 Utilities under a regulator do not show better performance than those that are not under a regulatory regime, but these utilities tend to show less variance in performance than those utilities that are not subject to regulation (figure 4.17). The difference in standard deviation in the overall performance index is significantly lower with regulation, suggest- ing that regulation may have a positive impact on reducing vulnerability in utility performance. DEA Efficiency A data envelopment analysis (DEA) was conducted to measure the relative efficiency of util- ities. The DEA creates a performance index from indicators—referred to as inputs and out- puts in the DEA literature—that can be related to other factors that drive performance. Under basic DEA, a water utility is regarded as a relatively efficient utility if its observed inputs can be scaled to yield outputs that equal or exceed any combination or scaling of what other utilities’ observed. Productive efficiency was assessed through a DEA. Water billed was con- sidered as the major output while number of staff and number of connections are considered as inputs. Performance of Water Utilities in Africa 45 Figure 4.15. Water Coverage Index, by Country, 2010–13 against African Benchmark a. Benin b. Burkina Faso c. Congo, Dem. Rep. 1.0 1.0 1.0 Water coverage Water coverage Water coverage 0.8 0.8 0.8 0.6 0.6 0.6 index index index 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 d. Côte d’Ivoire e. Guinea-Bissau f. Kenya 1.0 1.0 1.0 Water coverage Water coverage Water coverage 0.8 0.8 0.8 0.6 0.6 0.6 index index index 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 g. Malawi h. Mozambique i. Niger 1.0 1.0 1.0 Water coverage Water coverage Water coverage 0.8 0.8 0.8 0.6 0.6 0.6 index index index 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 j. Senegal k. Tanzania l. Togo 1.0 1.0 1.0 Water coverage Water coverage Water coverage 0.8 0.8 0.8 0.6 0.6 0.6 index index index 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 2010 2011 2012 2013 2010 2011 2012 2013 2010 2011 2012 2013 m. Uganda n. Zambia 1.0 1.0 Water coverage Water coverage 0.8 0.8 0.6 0.6 index index 0.4 0.4 0.2 0.2 0 0 2010 2011 2012 2013 2010 2011 2012 2013 46 Performance of Water Utilities in Africa Figure 4.16. Overall Performance Index (Unweighted Average of Operational Performance, Customer Performance [Service Quality], and Financial Performance) against African Benchmark, 2010–13 a. Low-income countries b. Middle-income countries 1.0 1.0 Overall composite performance index Overall composite performance index 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 Table 4.2. Best-Performing Utilities in 2013: Unweighted Average of Operational Performance, Customer Performance (Service Quality), and Financial Performance National Regional Municipal/District Senegal, SDE 0.94 Malawi, SRWB 0.88 Kenya, Kiamumbi 0.93 Uganda, NWSC 0.92 Zambia, North Western WSC 0.85 Kenya, Nyeri 0.90 Côte d’Ivoire, SODECI 0.89 Zambia, Southern WSC 0.83 Tanzania, Tanga 0.87 Burkina Faso, ONEA 0.89 Malawi, NRWB 0.83 Kenya, Lodwar 0.88 Benin, SONEB 0.83 Malawi, CRWB 0.82 Kenya, Olkalou 0.88 Table 4.3. Relationship between Utility Performance (Unweighted Average of Operational, Financial, and Customer Performance) and Utility Size Band size by population served Standard deviation Bottom 25 percent Median Top 25 percent Maximum <50,000 0.129 0.624 0.697 0.697 0.959 50,000−500,000 0.112 0.691 0.768 0.768 0.941 >500,000 0.134 0.705 0.750 0.886 0.940 Performance of Water Utilities in Africa 47 Figure 4.17. Overall Performance Index (Unweighted Average of Operational Performance, Customer Performance [Service Quality], and Financial Performance) against African Benchmark, by Presence of a Regulator, 2010–13 a. No regulator b. Regulator 1.0 1.0 Overall composite performance index Overall composite performance index 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 2010 2011 2012 2013 2010 2011 2012 2013 This study assessed the relative performance of water utilities in terms of relative produc- tive efficiency using DEA and investigated the role of governance as a  driver of relative ­performance. Analysis of relative performance was conducted for water and wastewater pro- viders from 17 countries in the Africa region, providing a sample representing 130 million people living in the service area of  the utilities. The sample utilities served more than 85 ­million ­people during 2013. All available observations from these utilities for each year for the period 2010–13 were used in the analysis. The detailed methodology is provided in appendix D. DEA employed International Benchmarking Network for Water and Sanitation Utilities (IBNET) data. Water billed was considered as the major output while number of staff and  number of connections were considered as inputs representing labor and capital, respectively. DEA assigns a number between 0 and 1 to each water utility which describes how efficient the utility is in transforming capital and labor inputs into water output relative to other utilities in the sample. In this scheme, 1 signifies that a utility is efficient when compared with the other utilities in the sample. Hence, DEA efficiency is relative to the  other utilities in the sample and may not indicate superior performance in a broader context. A water utility is regarded as a relatively efficient utility if its observed inputs can be scaled to yield outputs that equal or exceed any combination or scaling of what other sample 48 Performance of Water Utilities in Africa Figure 4.18. DEA Frontier utilities’ observed. Scaling of effi- cient utilities describes a surface s which shows the maximum out- ction ne con f 600,000 put achievable for every capital ro m be 400,000 Nu and labor combination as far as al: 200,000 pit Ca can be gleaned from the sample 0 of utilities. Figure 4.18 shows a 6 10 DEA efficiency frontier which is 3 7 llions of m formed by scaling efficient utili- 4 107 ties in the sample during 2010. Water billed: mi Inefficient utilities are depicted 2 107 as points below the surface in the figure. 0 3,000 Summary of the results pre- 2,000 Labor: Number 1,000 sented in the Figure 4.19: of sta 0 Only two or three of the utili- ties in the sample maintain a Figure 4.19. DEA Assessment of Relative Efficiency higher than average level of performance, ­ while the vast 40 majority of  utilities do not per- 35 form efficiently. This can be seen 30 from year-to-year assessments, Percentage of utilities 25 where the same set of utilities maintain  the highest perfor- 20 mance ­ status (table 4.4). These 15 are utilities from Zambia and 10 Mali while in certain years utili- 5 ties from Senegal and Kenya 0 were also included, reflecting 0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 those utilities that need the low- DEA efficiency est level of inputs to achieve the highest output of water pro- duced. However the vast majority of the remaining utilities perform far below the most efficient utilities. Transparency International’s Corruption Perceptions Index (CPI) was tested against DEA for correlation. The hypothesis is that a positive correlation between DEA relative efficiency and the CPI indicates that DEA relative efficiency is associated with good governance. This outcome, including the magnitude of the correlations, suggests that governance may be a driver of efficiency in the water sector. Correlation, of course, does not prove causation. Performance of Water Utilities in Africa 49 Table 4.4. Best Five Utilities’ Relative Performance, by Even so, the correlations are positive for each year and large Year, 2010–13 enough to cast serious doubt about their nature being coin- Country Utility name DEA score cidental (see table 4.5). 2010 Zambia Mulonga WSC 1.0000 Conclusions Kenya Nol Turesh 1.0000 The overall composite performance index measures sev- Zambia Lukanga WSC 0.9189 eral features of good performance defined as operational, Zambia Nkana WSC 0.9117 financial, and customer performance. As this is an aggre- Senegal SDE 0.8465 gated index, the number of utilities that provide data on all 2011 dimensions of performance is not always even. A sample Zambia Mulonga WSC 1.0000 of about 120 utilities shows that there are well-performing Kenya Gulf 1.0000 utilities in Africa. African utilities tend to be better Mali SOMAPEG 1.0000 performing on aspects such as financial and operational ­ Zambia Lusaka WSC 0.8544 performance. In these dimensions of performance, the Zambia Lukanga WSC 0.7782 best of Africa’s water utilities are among the best globally 2012 (as measured in the IBNET database). But as far as cus- Mali SOMAPEG 1.0000 tomer performance is concerned, measuring the quality of Senegal SDE 1.0000 access, there are still utilities that show good performance, Zambia Lukanga WSC 1.0000 but in general African utilities tend to lag significantly Zambia Luapula WSC 0.8544 Zambia Mulonga WSC 0.7782 behind global benchmarks. The same is true for water cov- 2013 erage, where African utilities in general do not meet global Zambia Mulonga WSC 1.0000 benchmarks. Zambia Luapula WSC 1.0000 When the data are disaggregated at the utility level, Zambia Lusaka WSC 0.9149 huge variations in utility performance are detected—both Zambia Lukanga WSC 0.8395 positive and negative. Hence, utilities can improve their Mali SOMAPEG 0.7384 performance dramatically over a span of four years (that is, the time period over which it was possible to analyze the data while maintaining a balanced data panel) but Table 4.5. Correlation between DEA Relative Efficiency and the CPI for Each of the Years, 2010–13 ­ performance can deteriorate equally rapidly, which shows Year Correlation coefficient that the performance of utilities in the sector is vulnera- 2010 0.6707 ble. This vulnerability is especially evident in the smallest 2011 0.4225 and the largest utilities where the standard deviation 2012 0.4369 in the performance index is about 30 percent higher than 2013 0.5023 in utilities that are more medium-size. Utilities that are being regulated in some formtend to show less variance in performance than those utilities that are not subject to a more or less formal set of regulations. The DEA efficiency analysis showed that although there are some relatively efficient utilities in Africa they make up a small minority. The majority of utilities register an effi- ­ ciency of 0.30 (which is far below the highest score of 1), showing significant options for 50 Performance of Water Utilities in Africa improvement. This is not unexpected as most African utilities are not yet mature in terms of coverage, service levels, and consumption levels—often experiencing rationing—and hence inputs less easily translate into outputs. Zambian utilities tend to perform the best in terms of both absolute performance and DEA, and at the same time Zambia is a middle-income country with relatively high coverage and consumption levels compared with many other utilities in Africa. It is important to notice that Mali Water (SOMAGEP) also keeps a high level of performance, despite its recent (2008) split with the National Electricity Company, Electricité-de-Mali, EdM, that apparently provided some indirect support to water opera- tions in earlier years. Senegal SDE also shows up in the list of more efficient utilities, but not as systematically as the utilities in Mali and Zambia. DEA tests also show that governance may be a significant driver of water utility performance in Africa and that attention to improving governance may be key to improving performance in water utilities. Notes 1. The countries included in this sample are Benin, Burkina Faso, Côte d’Ivoire, Kenya, Malawi, Niger, Tanzania, and Zambia. The more complex the indicator becomes, the more likely it is that not all utilities report the data. Hence, the sample loses more than half of its national utilities, whereas the only municipal utilities in this sample come from Tanzania. 2. Affordability is measured as the revenues per capita per year as a percentage of the GNI per capita in the country. The higher the ratio, the more people have to pay for water and the less affordable the service is. It should be noted that piped water access is rarely equally distributed over a country, and most piped water is provided to urban areas, and within these urban areas, more into richer than lower income neighborhoods. Affordability may have an upward bias and actual affordability may be lower than the calculation provided here. Performance of Water Utilities in Africa 51 © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 5 Performance of Utilities in Africa: Institutional Factors The IBNET Toolkit includes organizational data, including data on human resources poli- cies, business planning, and so on. Yet, collecting this data turned out to be more challeng- ing than collecting operational and financial data. We have some data on the use of private sector participation in utilities, but the response rate with data on HR policies, planning pro- cesses, and customer service was in general very poor, which can result in serious sample biases. Hence, as theory suggests that institutional and organizational factors play an important role in the delivery of water services, we will use the case study approach (in chapter 7) to take a look into these organizational data where possible. In addition, the team collected data—outside the primary data collection effort from secondary sources—that included information on politics and governance at the country level, the role of regulation, the role of different service models (national versus subnational levels of service delivery), and the scope of service delivery (for example, utilities providing only water services).1 The impact of economic development on utility performance was also tested. The team tested several indicators from the Polity Project, but none of these indicators showed any statistical significance. This may be linked to the fact the utilities and the countries in which they are located are not a random sample, and hence may show too little variance with respect to many of these indicators to generate statistically significant results. Performance of Water Utilities in Africa 53 Role of Economic Development Economic development has a positive impact on customer performance indicators as can be seen in table 5.1. In middle-income countries, water coverage is higher than in low-income countries, and so is the quality of the service provided. Interestingly, operational perfor- mance is better in low-income countries and significantly so. This result may be linked to the relatively small number of utilities in middle-income countries included in the sample. Financial performance when measured with the operating cost coverage ratio (OCCR) is not significantly different between low- and middle-income countries; the variation within countries with respect to this indicator tends to be large. Economic development translates into more consumers (that is, higher water coverage) using more water (that is, higher water consumption), and as such the annual revenues per capita increase rapidly, giving the utilities a more significant revenue flow. In the sample of African utilities, the median per capita total water consumption is only 52 lcd in low-income countries compared with 92 lcd in middle-income countries. When look- ing at residential water consumption—with a much smaller sample as not all utilities dis- aggregate consumption data—total residential water consumption is only 32 lcd in low-income countries, compared with 78 lcd in middle-income countries. This low resi- dential consumption provides utilities with a very small revenue basis, but the very low consumption also may make it less compelling for water consumers to use water and/or connect to the supply compared with other water sources at least with regard to service delivered. Hence, increasing coverage and residential consumption may be needed to improve customer and financial performance in the long run. If customers have access to more piped water to consume, the benefits of piped water are more apparent compared with other water supply sources. This focus on improving the infrastructure will require, in the short term, more investments in the sector to improve water coverage and a level of residential water consumption that sets piped water apart from alternative water sources (and in line with WHO public health standards of residential water consumption of at least 50–100 lcd). This increase in investments requires major improvements to ensure that the sector can absorb these investments and use the capital invested more efficiently, including but not limited to (a) detailed analysis of demand for piped water to Table 5.1. Impact of Economic Development on Utility Performance Indicator Low-income countries Middle-income countries t-test Significance Customer performance Customer performance as measured by 0.67 0.77 −5.12 0.000 quality of service Water coverage 0.56 0.73 −4.57 0.000 Financial performance (measured by OCCR) 1.06 1.08 −0.44 0.329 Operational performance 0.71 0.51 8.09 0.000 54 Performance of Water Utilities in Africa ensure that investments pay off; (b) selecting water supply investments whose benefits exceed their full life-cycle costs; and (c) transparent financing policies that underpin bet- ter investment selection. Role of Regulation Regulation has been promoted as a tool for improved utility performance while protecting customers. The sample contains three forms of regulation: (a) through government ministries or departments; (b) regulation by contract; and (c) a regulatory authority or agency. The presence of a regulatory agency is not necessarily randomly assigned across countries. In the sample, regulators were present in Niger (western Africa) and Kenya, Mozambique, Tanzania, and Zambia (eastern and southern Africa). As most regulating agen- cies are active in countries where regional- or district-level utilities are active, the sample contains many utilities under the regime of a regulatory agency and very few under any of the other types of regulation. A t-test analysis shows that having a regulatory agency in place does not automatically result in better performance. With regard to customer protection, regulation is effective in Africa. As can be seen in table 5.2, service quality is higher in utilities under a regulatory regime than in those that are not. Yet, with regard to water coverage, which includes potential customers, utilities under a regulatory regime show lower water coverage than those that do not have such a system in place. With regard to financial and operational performance, utili- ties with regulation perform poorer than those without a regulator in place. This is linked with the fact that the average size of a utility (as measured in population served) under a regulatory regime is about 204,000 compared with 2.5 million for those utilities whose performance is not supervised by a regulatory agency. The difference in scale may be part of the explanation why having a regulator in place results in lower financial and operational performance. Role of Service Delivery Models The Dublin principles defined in the 1990s made the point that water utility services should be delivered at the lowest appropriate level of government. The sample contains three types of utilities: (a) national; (b) regional; and (c) municipal. National utilities are mostly Table 5.2. Impact of a Regulatory Agency on Utility Performance Indicator With a regulatory agency Without a regulatory agency t-test Significance Customer performance Customer performance as 0.69 0.63 −3.25 0.0006 measured by quality of service Water coverage 0.57 0.65 1.75 0.040 Financial performance 1.06 1.05 −0.28 0.612 (measured by OCCR) Operational performance 0.68 0.76 3.01 0.001 Performance of Water Utilities in Africa 55 concentrated in western and central Africa. The median size of these utilities shows a large variance. The typical municipal utility serves 87,000 people, the typical regional utility serves 261,000 people, and the typical national utility 2.7 million people. Hence, there is a huge difference in size and likely economies of scale between these different utilities. All the municipal service providers in the sample are located in low-income countries and regional utilities are present in both low- and middle-income countries. The effect of decentralization will be examined only in low-income countries. The results of the effect of decentralization on utility performance are shown in table 5.3. It should be noted that only three countries2 in the sample use municipal or district service delivery and they are all located in low-income countries. The results show the same ambi- guity in results as found in Estache (1995). As far as financial performance is concerned as measured by the OCCR, the effect of municipal service delivery is similar to that of utilities that deliver services through regional or national utilities. Customers from municipal utili- ties spend significantly less on water than consumers served by other types of utilities. This lower spending comes at a price as customers consume less water. As tariffs are not signifi- cantly different from other types of utilities, water services provided by municipal utilities are much more affordable, which explains most of the better customer performance linked to municipal utilities. Similar to Estache (1995), the team found that decentralization tends to have a negative impact on labor productivity. Yet, even though national utilities tend to have higher labor productivity (fewer staff per 1,000 people served), the employee costs are significantly higher, with the result that the share of labor costs in total operating expenses does not nec- essarily decline with improvements in labor productivity (table 5.4). Levels of decentralization are highly correlated with certain Country Policy and Institutional Assessment (CPIA) indicators3 as can be seen in table 5.5. Municipal service delivery is highly correlated with four CPIA cluster indexes, which themselves are also highly correlated. The question arises whether municipal service delivery is more likely to occur when the overall economic environment is more conducive or municipal service delivery builds a more conducive economic environment. Table 5.3. Effect of District or Municipal Service Delivery on Utility Performance in Low-Income Countries Non-municipal Indicator Municipal service delivery t-test Significance service delivery Customer performance Customer performance as measured by quality 0.68 0.61 −3.63 0.000 of service Water coverage 0.55 0.63 1.90 0.029 Financial performance 1.06 1.03 −0.42 0.373 As measured by OCCR Operational performance 0.71 0.72 0.36 0.643 56 Performance of Water Utilities in Africa Table 5.4. Labor Productivity and Labor Cost per Employee Median labor productivity measured as staff per Type of utility Median annual staff cost per employee (US$) 1,000 peoplea National 0.31 12,578 Regional 0.86 9,201 Municipal 0.92 4,563 a. We used staff per 1,000 people for two reasons: (a) the number of missing observations on staff per 1,000 connections is relatively high and especially prevalent in western Africa and (b) sharing of connections is relatively high in many utilities— either the sharing of house connections or the dependence on public taps. This results in high staff numbers per connection, as there are relatively few connections compared to an environment where most households do not share connections. Table 5.5. Correlation between CPIA Indicator Clusters and Level of Decentralization District/ Public sector Policies Economic Structural Variable Municipal management for social management policies service delivery and institutions inclusion Municipal service delivery 1.0000 Economic management 0.8157 1.0000 Structural policies 0.6155 0.6212 1.0000 Public sector management and institutions 0.5259 0.7003 0.4756 1.0000 Policies for social inclusion 0.6113 0.5898 0.6690 0.7257 1.0000 Table 5.6. Correlation between CPIA Indicator Clusters and Utility Performance Indicators Customer performance Financial performance Operational Variable Water Customer OCCR OCCR dummy performance coverage performance Economic management −0.1017** −0.0010 0.0631 −0.0418 0.2446*** (0.027) (0.984) (0.177) (0.365) (0.000) Structural policies −0.1562*** 0.0742 0.1685*** 0.0178 0.0929** (0.000) (0.108) (0.000) (0.699) (0.045) Public sector management and institutions 0.1000** 0.1440** 0.1010** −0.0389 0.3333*** (0.030) (0.017) (0.030) (0.399) (0.000) Policies for social inclusion −0.0820* −0.0026 0.1173** −0.1063** 0.2594*** (0.076) (0.956) (0.012) (0.021) (0.000) *p < 0.10, **p < 0.05, and ***p < 0.01. Customer performance is especially weakly correlated to the quality of the country’s polit- ical and institutional environment, as can be seen in table 5.6. Higher scores for public sector management and institutions are correlated with better customer performance (and hence higher service quality). Water coverage—mostly the result of large investment programs—is correlated (albeit weakly) to better economic management and structural policies. Yet, the effect is negative; higher scores for economic management and structural policies are Performance of Water Utilities in Africa 57 correlated with lower levels of water coverage. Yet, higher scores of public sector manage- ment are correlated with higher water coverage. Financial performance is weakly correlated with higher scores of structural policies, and public sector management. Operational performance is correlated with higher scores of all CPIA cluster indicators, but the correlations are still rather weak. More research is needed to understand how the political and institutional country environment affects utility perfor- mance, beyond these highly aggregated country indicators. Economies of Scale and Scope In many countries, there is a trend to agglomeration, to essentially enable utilities to benefit from economies of scale in service provision. Hence, the performance of larger utilities (util- ities with more than 1 million people served) and of those that serve smaller populations have been compared. Tables 5.7 and 5.8 show that the effect of scale on utility performance by size of utility is not showing a straightforward picture. Customer performance (that is, service quality) is lower in large utilities than in smaller utilities, but medium-size utilities do better than the smallest utilities. For operational performance, a similar pattern of a concave relationship can be seen. Yet, water coverage is not significantly different between smaller and larger utilities. The effect of scale on utility performance is much more difficult to detect. The size of the utility Table 5.7. Effect of Scale on Utility Performance Indicator Small utilities Large utilities t-test Significance Customer performance Customer performance as measured by quality of service 0.703 0.648 2.57 0.010 Water coverage 0.669 0.667 0.18 0.468 Financial performance 1.10 1.03 −1.60 −0.109 As measured by OCCR Operational performance 0.723 0.696 −0.92 −0.179 Table 5.8. Economies of Scope on Utility Performance in Low-Income Countries Water and Indicator Water utilities only t-test Significance wastewater utilities Customer performance Customer performance as measured by quality of 0.705 0.670 −2.16 0.016 service Water coverage 0.700 0.671 −3.28 0.000 Financial performance 1.107 0.971 −4.65 0.000 (as measured by the OCCR) Operational performance 0.764 0.700 −3.38 0.000 58 Performance of Water Utilities in Africa seems to have little effect on the median OCCR (that is, financial performance) or operational performance. The effect of the importance of economies of scale will be tested in chapter 6. Conclusions A utility operates in a particular local context. Although an attempt was made to collect organizational data that reflect the characteristics of the particular utility, these data were difficult to collect as the response rate was low (possibly because the team collected a rela- tively large set of data). Hence, the team used a set of national-level factors that describe the general environment in which utilities operate at the country level. In follow-up work, it is recommended that more attention is paid to collecting organizational data from utilities, as this may help improve the understanding of how institutional and organizational factors affect utility performance. As seen in previous chapters, the quality of governance seems to matter for good utility performance. Institutional factors play a role in driving utility performance but not necessarily always in the way that is predicted. Regulation is often seen as a shortcut for ensuring better governance in the sector.4 Yet, the analysis shows that having a regulatory agency in place does not automatically produce better results for customers. In low-income coun- tries, the presence of a regulator results in better customer performance, but having a regulator in place does not extend to other forms of performance, such as operational performance or water coverage. This is as expected because the objective of a regulator is to provide “protection” of existing customers (and hence focuses on providing minimum levels of customer service) but does not necessarily focus on improvements in financial and operational sustainability or improvements in coverage compared with utilities without a regulator.5 Decentralization’s normally held assumption that service provision at local level tends to increase accountability and improves utility performance is only partially borne out by the analysis of the utilities. The financial and operational performance of utilities is not statisti- cally significantly different from other forms of service delivery. As for customer perfor- mance, there are statistically significant differences: district- or municipal-based service delivery shows better results. Yet, coverage lags behind in municipal- or district-based ser- vice delivery compared with other levels of service delivery. The team also looked into the presence of economies of scale and scope. The effect of scale was not easy to detect in the sample, but that might have to do with the fact that there were few very large utilities in the sample and these tend to be mostly concentrated in the cate- gory of national utilities. However, economies of scope are evident. Utilities that provide water and wastewater services in low-income countries tend to show slightly higher levels of water coverage (as can be expected because sewerage coverage is provided sequentially after a certain level of water coverage has been achieved). But customer performance (or service quality), operational performance, and financial performance are also higher when sewerage is also provided. Performance of Water Utilities in Africa 59 Finally, the level of economic development matters. Economic development translates into more consumers using more water, and as such the annual revenues per capita increase rapidly, giving the utilities a more significant revenue flow. If customers have access to more piped water to consume, the benefits of piped water are more apparent compared with other water supply sources. This focus on improving the infrastructure will require, in the short term, more investments in the sector to improve water coverage and a level of water con- sumption that sets piped water apart from alternative water sources. At the same time, as economic development accelerates and water consumption increases, the nature of the ser- vice changes with more emphasis on wastewater collection and treatment, which will add additional expensive investments to be funded. In the provision of water and wastewater infrastructure, lock-in effects of infrastructure, technology, and product design are a major challenge. In Africa, where a major part of the infrastructure necessary to meet water and wastewater demand is still to be built, the range of alternatives is considerably larger than elsewhere, but will require the adoption of better urban and regional planning, more effi- cient water, wastewater, and drainage systems, and improvements in investment planning. However, to move in this direction, major barriers, especially institutional and cultural bar- riers, must be overcome. Notes The team collected data on private sector participation, but this indicator largely overlapped as most national utilities in 1.  the sample also use some form of private sector participation. In the period under review, Kenya, Mozambique, and Tanzania had district-level service delivery. Yet, since then 2.  Mozambique has merged its utilities into four regional utilities. CPIA database at www.databank.worldbank.org. 3.  Countries that have a regulatory agency in place show higher indexes for public sector management and institutions 4.  (as measured by the CPIA) as there is reference in the definitions of CPIA to the existence and functioning of regulatory agencies in the quality of public administration definition (part of the public sector management and institutions overall index). It is unclear whether this is the result of the higher service levels, higher labor costs (as labor efficiencies coincide with 5.  higher staff cost per employee), the existence of higher transaction costs for the utility operators, or a multitude of other factors. Reference Estache, Antonio, ed. 1995. “Decentralizing Infrastructure: Advantages and Limitations.” Discussion Paper 290, World Bank, Washington, DC. 60 Performance of Water Utilities in Africa © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 6 Drivers of Utility Performance: Panel Data The panel data of 119 utilities used in the previous chapters is used in this analysis. As mentioned above these utilities represent a significant proportion of the population that was served by piped water in 2013: 58 million out of 108 million people. As possible drivers of performance, a number of country-specific and utility-specific characteristics were tried. The variables that proved to be significant and that were kept in the final version of the models are shown in table 6.1. The different forms of utility performance are weakly correlated, as can be seen in table 6.2. This may be a surprising result because the hypothesis laid out in figure 4.1 translating good operational performance into better financial performance is hence not borne out by the econometric analysis. Such an improvement allows the utility to gener- ate excess funds that could—once the utility generates sufficient cash to pay for operation and maintenance (O&M) costs, depreciation, and debt service—be used for improvements to service levels or water coverage. The hypothesis is that once the customer experiences improved service level, the utility will be able to generate more revenues that then can fund further improvements in operational and/or customer performance. This virtuous cycle seems to be far from robust when analyzing the data of close to 120 utilities. Table 6.2 shows that there is a weak, albeit statistically significant, correlation between financial performance and operational performance. There is also a weak correlation between financial and customer performance, but it is negative, suggesting that better customer performance (in terms of better quality of service) is correlated with lower levels of finan- cial performance. Access to water services is not correlated to financial performance. Performance of Water Utilities in Africa 61 Table 6.1. Explanatory Variables Used in the Performance Models Variable name Variable definition Level of observation Measurement unit Popserved_water Population served with water Utility 1,000,000 Share_laborcost Share of labor cost in total Utility % operational costs Operational_performance Operational performance as defined Utility 0–1 (continuous) indicator in chapter 4 Customer_performance indicator Customer performance (service Utility 0–1 (continuous) quality) as defined in chapter 4 Average revenues per cubic Total operating revenues per Utility US$ meter sold volume water sold in U.S. dollars Water consumption Water consumption per person Utility served (in lcd) Sewerage coverage Presence of sewerage coverage Utility 0/1 (dummy) GNI per capita GNI per capita (Atlas method) Country US$ (current) Share-urban Proportion of urban population Country % Large-utility Takes the value 1 if water services Country 0/1 (dummy) are managed by large utility CPIA_economic_management The higher the value, the better the Country 1–6 economic management CPIA_transparency The higher the value, the more Country 1–6 transparency Trend Time trend n.a. 1 (in 2010) to 4 (in 2013) Note: n.a. = not applicable. Table 6.2. Correlation between Technical and Financial Performance for Water Utilities Financial Operational Customer Variable Water coverage performance performance performance Financial performance 1.0000 (OCCR) Operational performance 0.1830 1.0000 (0.0001) Customer performance −0.1365 0.0573 1.0000 (service level) (0.0034) (0.2469) Water coverage −0.0445 0.1933 0.3844 1.0000 (0.3423) (0.0000) (0.0000) Yet, water coverage and quality of service levels are correlated; hence customers seem to be more willing to connect when service levels are higher. The relatively weak correlation between different performance indicators may be sur- prising when assuming there is a linear relationship between them, in which better opera- tional performance results in better financial performance which in turn can result in better 62 Performance of Water Utilities in Africa service delivery to the customers. Yet, this linear approach does not always work. For instance, if the utility increases the level of metering, this should, in theory, result in more revenues for the utility and hence improve the ability of the utility to cover its costs. Yet, if the cost of the metering program (in terms of the O&M costs to manage such a metering program) exceeds the benefits of the metering program, the effect on financial performance of the utility can be negative, even with an increase in revenues. At the same time, the metering program will reduce consumption, which may result in consumers being less willing to pay for the service—resulting in lower consumption levels. In the case of Africa, where many connections are shared, it might also affect the willingness to share connec- tions and could even affect water coverage. Hence, the specific outcomes of different measures to improve utility performance are not given but are highly dependent on the context in which they take place. In the next sections, the drivers of different aspects of performance will be examined— specifically the drivers of financial performance, water coverage, and customer performance (as measured by the quality of service levels). Financial Performance Several models have been used to explain what drives financial performance. As discussed in chapter 3, the typical utility has an operating cost coverage ratio (OCCR) of about 1 in the sample. As such, there are many utilities that have a minimum financial performance. The best-fitting model was the one where a dependent variable that takes only two values was used: 1 if the utility has an OCCR of more than 1.19 (found in the best performing 25 percent of utilities) and 0 if it does not. The appropriate econometric method to use when the dependent variable is binary is a Probit or Logit model, which describes the probability that the dependent variable takes the value 1. Here a Probit approach was used to be able to control for utility-specific unobserved heterogeneity. The latter may include unobserved variables such as the skills of the utility’s manager; the physical location of the area serviced by the utility (topography, distance to and quality of the raw water sources); the technology of the infrastructure; and any other unobserved specific conditions that the utility may face and that could affect its financial performance. Controlling for unobserved heterogeneity is crucial in the presence of panel data, to avoid biased estimated coefficients. The model has been estimated using 427 observations corresponding to 116 distinct utili- ties (see table 6.3). The sample size is slightly reduced compared with the original sample because of missing observations for some of the utilities. The Wald test indicates that the model is globally significant. The value for the parameter rho, at the bottom of the table, indicates that 64 percent of the total variance is contributed by the panel-level variance component that is the unobserved utility-specific heterogeneity. The results show that both utility-specific and country-specific characteristics influence the probability that utilities are performing well financially. When utilities achieve a better Performance of Water Utilities in Africa 63 Table 6.3. Drivers of Financial Performance, Random-Effects Probit Model Financial performance (0/1) Coef. T-test P>z Utility-specific variables Operational performance 1.817* 1.72 0.085 Customer performance (quality of −2.405** −2.11 0.035 service levels) Popserved_water 1.448*** 3.14 0.002 Popserved_water ^2 −0.186** −1.61 0.022 Staff efficiency 0.3157*** 3.71 0.000 Operation and maintenance cost per −1.285*** −2.80 0/005 cubic meter of water produced Country-specific variables GNI_per capita 0.002*** 2.61 0.009 CPIA_transparency 0.3149 0.47 0.640 Trend −0.170 −1.63 0.103 Constant −3.523 −1.64 0.101 Number of observations 427 n.a. n.a. Number of utilities 116 n.a. n.a. Wald chi test (p-value) 2 35.82*** (0.000) n.a. rho (p-value of test: rho = 0) 0.64* (0.083) n.a. Note: n.a. = not applicable. *p < 0.10, **p < 0.05, and ***p < 0.01 operational performance, this reflects positively on the utility’s financial performance. Higher service quality is negatively correlated with the probability of achieving better financial performance. This suggests that achieving higher service quality results in costs, whereas lower service quality results in lower revenues, either one affecting the OCCR; albeit that the effect of better operational performance is stronger than the effect of better customer performance. Utilities of larger size (as measured by the size of the population served with water) are more likely to be good financial performers, but once the utility reaches a certain critical size, the likelihood of being a financially well-performing utility decreases (the coefficient of the square of the variable measuring population served is negative), signifying the existence of economies of scale. Utilities for which staff efficiency (the revenues generated by each employee as a percentage of their costs) is higher tend to have better financial performance than those that have lower efficiency. Utilities with higher O&M cost per cubic meter of water produced tend to perform worse than those with lower O&M cost. Utilities that operate in poorer countries (as measured by the gross national income [GNI] per capita) are less likely to be financially good performers. Finally, the coefficient of the trend variable is negative, which indicates an overall decrease in financial performance over time. 64 Performance of Water Utilities in Africa Customer Performance Water Coverage The water coverage index can be used as a continuous variable. The model uses 468 obser- vations covering 119 distinct utilities. It is controlled for utility-specific heterogeneity and utility-specific unobserved effects are specified as fixed effects in the model. Estimation results are shown in table 6.4. The Fisher test indicates global significance of the model and overall significance of the utility-specific effects. The R-square is 0.47, which indicates that 47 percent of the total variance in customer performance is explained by the covariates. As such, even though this model is the best performing of all the models tested, the result shows that explaining the drivers of customer performance requires more information. It also shows that it is not always possible to capture all the variance in Africa in panel data regression models as it is likely that variables are omitted, most notably local information on the physical and organizational environment in which utilities are working. It was found that the quality of service levels matters. Higher service-level quality results in higher water coverage. The price of the service was not found to be statistically significant. More interesting is the effect of good financial performance (as measured by an OCCR of at least 1.19) on water coverage. The relationship shows that good financial performance is Table 6.4. Drivers of Water Coverage, Fixed-Effects Regression Model Water coverage (0 to 1 scale) Coef. T-test P>t Utility-specific variables Customer_performance indicator (proxy for 0.7935*** 11.76 0.000 service quality) Average revenues per cubic meter sold −0.0124 −1.25 0.212 Popserved 0.1525*** 4.13 0.000 popserved_water ^2 −0.0223*** −4.09 0.000 OCCR_DUMMY by incomestatus = 1 0.0229 1.57 0.118 OCCR_DUMMY by incomestatus = 2 0.0935** 2.43 0.016 Country-specific variables GNI per capita 0.0001 1.57 0.118 CPIA_economic_management cluster 0.1719*** 6.28 0.000 Trend 0.0057 0.94 0.349 Constant −0.8301*** −6.33 0.000 Number of observations 468 n.a. n.a. Number of utilities 119 n.a. n.a. Fisher test of global significance (p-value) 33.31*** (0.000) n.a. Fisher test (fixed effects are equal to 0) 9.97*** (0.000) n.a. R-square 0.47 n.a. n.a. Note: n.a. = not applicable. *p < .10, **p < .05, and ***p < .01. Performance of Water Utilities in Africa 65 linked to higher water coverage. Yet, this effect does not hold for utilities in low-income countries. In view of the high capital intensity of the water infrastructure, investments in this type of infrastructure tend to be large and usually utilities are unable to pay for the initial investments. The effect holds in middle-income countries, but it should be noted that the effect of good financial performance—even in these countries—is rather limited in increasing access to water supply services. Hence, utilities may be able to fund some of the investments but are unlikely to pay for most of the up-front investments, as shown in the case studies described in chapter 7. The model shows a concave relationship between utility size (as measured by population size) and water coverage: water coverage increases more rapidly when utilities are larger and then decreases afterwards—suggesting economies of scale in providing water access. Yet, it is also likely that countries provide funds to expand coverage to larger utilities first before providing funds to utilities in smaller towns and localities. Utilities that operate in richer countries (as measured by GNI per capita) demonstrate higher water coverage in general, but only at the 10 percent level. Utilities that operate in countries that are associated with better economic management (as measured by the Country Policy and Institutional Assessment [CPIA] economic management cluster) tend to be associated with higher water coverage. The effect of better economic management is rather significant and could point to the fact that better economic management ensures that more resources are available to increase water coverage. Finally, the trend is found to be statistically insignificant. The model was run on water coverage again and distinguished by large utilities and those that are smaller—with large utilities classified as those serving more than 1 million customers. Although similar factors play a role, the importance of the variables differs significantly for the two models, showing that the behavior of utilities is far from similar. In general, utilities started to be established in capital cities; hence large utilities tend to have been in place much longer and by definition tend to have higher coverage than smaller utilities. When comparing what drives coverage in large and smaller utilities, some similarities and some important differences (see table 6.5) were seen. For both types of utilities, service qual- ity matters. Both models show a concave relationship between utility size (as measured by population size) and water coverage: water coverage increases more rapidly when utilities are larger and then decreases afterwards. However, the role of tariffs is much less significant as a driver for performance for large utilities than for smaller utilities. This may be linked to the fact that large utilities have a much larger customer base and, because of the relative anonymity of the individual client, it is much easier to not pay the bill. This is reflected by the generally lower collection efficiencies in larger utilities. The effect of good financial per- formance (as measured by an OCCR of at least 1.19) on water coverage shows that good finan- cial performance is not linked to higher water coverage in smaller utilities—independent of  the level of economic development of the country. For large utilities in low-income 66 Performance of Water Utilities in Africa Table 6.5. Drivers of Water Coverage, Fixed-Effects Regression Model, by Size of Utility Smaller utilities Large utilities Water coverage (0 to 1 scale) Coef. T-test P>t Coef. T-test P>t Utility-specific variables Customer_performance indicator 0.634*** 9.86 0.000 0.2447** 2.57 0.018 (proxy for service quality) Average revenues per cubic meter −0.007 −0.83 0.408 0.023 0.45 0.657 sold by incomestatus = 1 Average revenues per cubic meter −0.062 −1.18 0.240 0.108 0.99 0.333 sold by incomestatus = 2 Popserved 1.6909*** 10.88 0.000 0.2099*** 6.27 0.000 Popserved_water ^2 −1.5416*** −8.07 0.000 −0.011*** −6.17 0.000 OCCR_DUMMY by 0.0151 1.13 0.260 0.0258** 2.13 0.046 incomestatus = 1 OCCR_DUMMY by −0.0004 −0.01 0.991 −0.075 −0.58 0.568 incomestatus = 2 Country-specific variables GNI per capita 0.0001** 1.17 0.245 0.0001 1.57 0.132 CPIA_economic_management 0.1582*** 5.68 0.000 0.0182 0.89 0.386 cluster Trend 0.0028 0.41 0.686 −0.0156*** −3.21 0.004 Constant −0.8133*** −5.44 0.000 −0.194*** −3.05 0.006 Number of observations 426 n.a. n.a. 42 n.a. n.a. Number of utilities 111 n.a. n.a. 12 n.a. n.a. Fisher test of global significance 30.95*** (0.000) n.a. 60.92*** n.a. n.a. (p-value) Fisher test (fixed effects are equal 10.15*** (0.000) n.a. 30.99*** n.a. n.a. to 0) R-square 0.62 n.a. n.a. 0.97 n.a. n.a. Note: n.a. = not applicable. *p < .10, **p < .05, and ***p < .01. countries, good financial performance has a small, but positive effect on water coverage. Yet, this effect disappears for utilities in middle-income countries. Whether utilities operate in richer countries (as measured by GNI per capita) is statistically insignificant in both low-income and middle-income countries. Utilities that operate in countries that are associated with better economic management (as measured by the CPIA economic management cluster) tend to be associated with higher water coverage in the case of smaller utilities. Yet, the effect of the quality of economic management on large utilities is insignificant. As larger utilities are mostly located in the capital or larger cities, the quality of economic management may be less important as governments may have most likely already Performance of Water Utilities in Africa 67 provided funds to these larger utilities. Finally, the trend is found to be statistically insignifi- cant in smaller utilities, but negative and significant in larger utilities. Service Quality The customer performance index can be used as a continuous variable with data observa- tions ranging between 0 and 1. The model uses 468 observations covering 119 distinct utili- ties. It is controlled for utility-specific heterogeneity and utility-specific unobserved effects are specified as fixed effects in the model. Estimation results are shown in table 6.6. The Fisher test indicates global significance of the model and overall significance of the utility-specific effects. The R-square is 0.14, which indicates that only 14 percent of the total ­ variance in customer performance is explained by the covariates. As such, even though this model is the best performing of all the models tested, the result shows that explaining the drivers of customer performance requires much more information. It also shows that it is not always possible to capture all the variance in Africa in panel data regression models as it is more than likely that variables are omitted, most notably local information on the physical and organizational environment in which utilities are working. The higher service quality for consumers (as measured with the indicator defined in chapter 4) is linked to the cost of the service. The more people spend on water services, the lower the service quality tends to be. One of the drivers of service quality is the depth of rationing. The more the water consumed per person per day and hence the less rationing Table 6.6. Drivers of Customer Performance as Measured by Quality of Service, Fixed-Effects Regression Model Customer performance (0 to 1) Coef. T-test. P>t Utility-specific variables Average revenues per capita −0.0037*** −7.00 0.000 Sewerage coverage present (dummy) 0.883*** 4.01 0.000 Operational performance 0.0982*** 2.78 0.006 Country-specific variables GNI per capita 0.0001** 2.08 0.038 share_urban −0.006*** −2.60 0.010 CPIA_transparency −0.011 0.45 0.656 Trend −0.0057 −1.37 0.172 Constant 0.6900*** 7.46 0.000 Number of observations 462 n.a. n.a. Number of utilities 119 n.a. n.a. Fisher test of global significance (p-value) 12.64*** (0.000) n.a. Fisher test (fixed effects are equal to 0) 11.93*** (0.000) n.a. R-square 0.21 n.a. n.a. Note: n.a. = not applicable. *p < .10, **p < .05, and ***p < .01. 68 Performance of Water Utilities in Africa customers experience, the higher the service quality. Yet, this effect only shows up in low-income countries, where it is statistically significant. Then, rationing is much more widespread and pertinent in low-income countries than in middle-income countries. Sewerage coverage tends also to be linked to better service quality, which may hint toward economies of scope. Country-based factors matter but the effect is not very clear. When utilities also provide sewerage coverage, utilities that operate in low-income countries (as measured by GNI per capita) demonstrate higher service quality in general, but it depends on whether sewerage coverage is also provided. Where sewerage coverage is provided, the effect of income is not significant. Transparency (as measured by the CPIA index) is not significant. Finally, the trend is found to be negative, but statistically insignificant. Operational Performance This index is also measured on a 0–1 scale, so the same methodology as the one used for assessing the drivers of customer performance is adopted. The model is estimated using 415 observations for 116 distinct utilities (table 6.7). The model is significant overall and the R-square is 0.35, so the overall fit of the model is moderate. Operational performance is found to have a concave relationship with the size of the pop- ulation served with water when utilities are smaller: operational performance increases up Table 6.7. Drivers of Operational Performance, Fixed-Effects Regression Model Operational performance (0 to 1 scale) Coef. T-test P>t Utility-specific variables Popserved_water 0.0732** 2.30 0.022 Popserved_water ^2 −0.0060 −1.21 0.226 Customer performance 0.1830*** 2.86 0.004 Financial performance (as measured by the OCCR) 0.1097*** 6.30 0.000 Share_labor −0.3912*** −7.20 0.000 Country-specific variables GNI per Capita Atlas −0.0007 −0.83 0.406 CPIA_economic management 0.0872*** 2.86 0.005 Trend 0.0171*** 2.70 0.007 Constant 0.2266** 1.60 0.110 Number of observations 415 Number of utilities 116 Fisher test of global significance (p-value) 19.20*** (0.000) Fisher test (fixed effects are equal to 0) 7.61*** (0.000) Within R-square 0.35 Note: *p < 0.10, **p < 0.05, and ***p < 0.01. Performance of Water Utilities in Africa 69 to a threshold and then decreases: economies of scale matter.1 The better the financial performance of the utility, the higher the operational performance. The higher the customer performance, the higher the operational performance. Finally, the higher the share of labor costs, the lower the operational performance. This shows that utilities with a high share of labor in total costs tend to have lower staff efficiencies (that is, the revenues generated by each employee as a percentage of the cost of that employee are significantly lower) suggest- ing that these utilities tend to be less efficient in employing staff. This lack of efficiency will affect operational performance. Utilities operating in richer countries, as measured by the GNI per capita, do not show better operational performance. There is evidence of a positive, statistically significant trend. And, interestingly, the better the economic management in the country in which the utility is located, the higher the operational performance. Overall Performance Index We ran the models again using the overall composite index (combining customer, opera- tional, and financial performance). This index is also measured on a 0–1 scale, so the same methodology as the one used for assessing the drivers of customer performance is adopted. The model is estimated using 426 observations for 116 distinct utilities (table 6.8). The model is significant overall and the R-square is 0.52, so the overall fit of the model is good. The economies of scale as measured by the population served show that the larger the population served, the better the overall performance, but up to a threshold; too large a utility will generate diseconomies of scale. The threshold is, however, quite high since the proportion of utilities that are above the threshold is usually small. Utilities which also pro- vide sewerage coverage tend to be correlated with better performance. Economies of scope Table 6.8. Drivers of Overall Performance (Measuring the Combined Scores of Financial, Operational, and Customer Performance), Fixed-Effects Regression Model Operational performance (0 to 1 scale) Coef. T-test P>t Utility-specific variables Popserved_water for large utilities = 0 1.154*** 9.97 0.000 Popserved_water ^2 for large utilities = 0 −1.205*** −8.66 0.000 Popserved_water for large utilities = 1 0.196*** 7.56 0.000 Popserved_water ^2 for large utilities = 1 −0.011*** −2.97 0.003 Sewerage coverage (dummy) for low-income countries 0.0297 1.39 0.165 Sewerage coverage (dummy) for middle-income 0.237*** 4.06 0.000 countries Operation and maintenance costs per cubic meter of −0.034*** −5.65 0.000 water produced 0.006*** 3.17 0.002 table continues next page 70 Performance of Water Utilities in Africa Table 6.8. continued Operational performance (0 to 1 scale) Coef. T-test P>t Country-specific variables GNI per capita, Atlas Method 0.00003 0.44 0.662 CPIA_economic management 0.0802*** 3.68 0.000 Trend −0.0054 −1.20 0.231 Constant 0.2276*** 2.35 0.020 Number of observations 426 n.a. n.a. Number of utilities 116 n.a. n.a. Fisher test of global significance (p-value) 28.98*** (0.000) n.a. Fisher test (fixed effects are equal to 0) 10.80*** (0.000) n.a. Within R-square 0.52 n.a. n.a. Note: n.a. = not applicable. *p < .10, **p < .05, and ***p < .01 also play a role, but the effect is only statistically significant in middle-income countries because the number of utilities that provide sewerage services in low-income countries in Africa is relatively limited. Another finding is that costs (measured by the O&M costs per cubic meter of water produced) also play a critical role: the higher the O&M costs per cubic meter produced, the lower the overall performance. The level of O&M costs are the result of investment decisions; these investment decisions lock in costs for decades to come; hence assessing investment decisions properly is key. Staff efficiency also matters. The higher the staff efficiency, the better the overall performance. Finally, the quality of economic management matters. The better the eco- nomic management, the better the utility performance. Interestingly, the quality of the business environment also matters. If the quality of the business environment is high, util- ity performance benefits. Conclusions The performance models are globally significant but for many the predictive power is highly variable with the exception of the water coverage and the overall performance model. It was observed that the major drivers of utility performance are linked to their costs as reflected in the presence of economies of scale, economies of scope, the O&M costs per cubic meter of water produced, and staff-related efficiency. The economies of scale as measured by the population served shows that the larger the population served, the better the overall performance, but up to a threshold; too large a utility will generate diseconomies of scale. The threshold is, however, quite high since the propor- tion of utilities that are above the threshold is usually small. Utilities which also provide sewerage coverage tend to be correlated with better performance. Economies of scope also play a role. Another finding is the impact of costs Performance of Water Utilities in Africa 71 (measured by the O&M costs per cubic meter of water produced): the higher the O&M costs per cubic meter produced, the lower the overall performance. The level of O&M costs are the result of investment decisions; these investment decisions lock in costs for decades to come; hence assessing investment decisions properly is key. Staff efficiency and share of labor in total costs also matter. The higher the staff efficiency, the better the overall performance. Yet, staff efficiency is the result of a complex of factors, including labor regulations, but also O&M costs and tariff policies. Interestingly, some of the other drivers of utility performance are beyond the control of the utilities. The quality of economic management in the countries in which utilities operate affects utility performance: the higher the quality of the economic management, the higher the utility performance. The different aspects of performance affect one another. Higher customer service quality has a positive impact on water coverage whereas financial performance may affect opera- tional performance and water coverage, suggesting that funding is necessary to improve access and measures to improve operations in the utility. In the case of improving water access, the level of economic development matters, but also the quality of economic manage- ment, especially for smaller utilities that may depend more than the large ones on external funding to increase access. The additional findings are that water coverage is directly affected by customer perfor- mance (that is, service quality). Better economic management and higher GNI growth have a positive impact as they may be linked to the availability of investment funding. In most cases, there is no link between utilities’ financial performance and water coverage—with the exception of large utilities in low-income countries (although the effect is small). This suggests that most utilities are not able to improve access through improved financial performance but depend on external funds to do so. ­ The hypothesis posed in chapter 4, that better operational performance results in better financial performance and subsequently in better customer performance, is not borne out by the econometric results. Better operational performance tends to have a positive impact on financial performance; but the opposite is true for customer performance. The better the customer performance, the lower the financial performance. Hence, utilities can optimize financial performance by increasing operational performance and/or reduc- ing customer performance. In general, financial, operational, and customer performance affect each other. There is evidence that unobserved utility-specific heterogeneity explains a large portion of the total variance, which would call for the large-scale collection of additional utility-specific variables, for example, information on local conditions (topography, distance to the raw water source, whether the utility gets water primarily from groundwater or surface water, quality of the raw source, age of the infrastructure, access to alternative water source, and so on) and organizational and institutional data.2 Hence, much more and detailed information on utility operations and the context in which the utilities operate is required to explain with 72 Performance of Water Utilities in Africa more clarity what drives utility performance. However, such data collection comes at a price with regard to the cost of collection and the willingness of utilities, regulators, and other stakeholders to provide such information. Notes 1. A concave relationship (and hence the presence of a threshold) for large utilities is also expected. The fact that it does not show in the estimation results is probably explained by the characteristics of our sample as there is only a small number of large utilities in the sample. 2. The IBNET Toolkit includes organizational data, but in this round of data collection in Africa, this information was not consistently collected by the task teams, and such information was not available in many of the regulatory reports. Hence, this does not allow for including such information in the analysis. Performance of Water Utilities in Africa 73 © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 7 Drivers of Utility Performance: Case Study Perspective Introduction The purpose of the case studies is to complement the econometric analysis of the larger dataset. Unlike the econometric analysis, the case studies cannot show quantitative rela- tionships between indicators. However, the case studies can: • Track performance for a single utility over time • Assess why certain results were achieved • Analyze management techniques, organizational culture, and capital investment financing Five utilities were chosen for the case studies: Nairobi City Water and Sewerage Company (NCWSC) in Kenya; National Water and Sewerage Corporation (NWSC) in Uganda; Office National de l’Eau et de l’Assainissement (ONEA) in Burkina Faso; Sénégalaise des Eaux (SDE) in Senegal; and Société de Distribution d’Eau de la Côte d’Ivo- ire (SODECI) in Côte d’Ivoire. These utilities were selected to ensure consistency with the case studies written by the Electricity Study Team. The Electricity Study Team wrote case studies for electricity utilities for the same set of countries. Each of these countries has a national water utility except Kenya, so in four of five cases the national utility was selected. In Kenya, NCWSC was selected because it is the largest Kenyan utility, serving nearly 3 million people in the capital city, Nairobi. Performance of Water Utilities in Africa 75 The selection of the case studies was not at random, so it should be noted that the case study sample is not representative. For instance, all five utilities are large, serving close to or more than 3 million people. By contrast, the typical utility in the econometric database serves about 115,000 people. All five utilities were established decades ago, and hence have a longer track record and relatively more developed infrastructure. All of them are serving capital cities. Apart from Nairobi, none of the utilities provides wastewater services. In addi- tion, the five utilities are mostly well-performing. This allows the case studies to focus on the dynamics of well-performing utilities. Performance of the Case Study Utilities Data from the JMP regarding the performance of the five case study countries over the past 15 years are provided in table 7.1 as context. How each case study utility performs on key indicators is shown in table 7.2. These data do not always correspond with utility data partially because the services that utilities provide may be for areas that are smaller than the area that is classified as urban. In addition, utilities use different yardsticks to trans- late the number of (active) connections into coverage or access. The indicators included in the table are those used in the econometric analysis, plus three others—nonrevenue water (NRW) as a percentage of production, staff per 1,000 connections, and the collection ratio (table 7.3). These three indicators have been added because they are analyzed substantially in the case studies. Where available, the global benchmark (best quartile of all utilities in the International Benchmarking Network for Water and Sanitation Utilities [IBNET] database) and the African benchmark (best quartile of all utilities in the Africa-specific IBNET analysis) are provided for comparison. Each utility’s aggregate perfor- mance ranking out of 118 utilities is included. It should be noted that the focus of the case studies is how these five utilities have been performing in their service areas. This is an important point to make because, as can be seen in tables 7.1 and 7.2, even in national utilities the service areas are not necessarily overlap- ping with the population to be served, and even though some of the utilities have done excellent work, the increase in service areas has been very different between the various utilities. For the four national utilities for which we have data from 2000 to 2013, we find that the population without improved services has still not decreased in three out of the five countries. In table 7.1, we see the efforts that have been put in, with sharp increases in the service area and in the population served—significantly beyond the overall population growth rates over a similar period. This also shows the scale of the challenge in providing water supply services in Africa. Table 7.2 shows some interesting features of the demand for water supply. Based on household surveys in the five case study countries, the population has increased rapidly in 76 Performance of Water Utilities in Africa Table 7.1. Country Data on the Growth in Access Increase in service area Increase in population Increase in population Increase in urban Country 2000–13 (%) served 2000–13 (%) 2000–15 (%) population 2000–15 (%) Burkina Faso 197 226 154 252 Côte d’Ivoire 220 188 132 164 Senegal 150 172 152 163 Uganda 320 498 165 224 Table 7.2. Country Data on Access to Improved Water Sources by Country, According to the JMP Urban Urban Urban Urban Population Population Population Population population population population population with piped with other without with piped Urban Population with piped with other without with piped Country Year water on improved improved water on population (millions) water on improved improved water on premises services services premises (millions) premises services services premises (millions) (millions) (millions) (%) (millions) (millions) (millions) (%) Burkina 2000 11.60 0.40 6.60 4.70 3.4 2.10 0.40 1.40 0.30 19.0 Faso 2015 17.90 1.40 13.30 3.20 7.8 5.30 1.40 3.80 0.10 26.4 Côte 2000 16.10 4.80 7.70 3.60 29.8 7.00 4.00 2.50 0.60 57.1 d’Ivoire 2015 21.30 9.10 8.40 3.80 42.7 11.50 7.60 3.10 0.80 66.1 Kenya 2000 31.30 6.00 10.20 15.10 19.2 6.20 3.20 2.30 0.80 51.6 2015 46.70 10.10 19.40 17.20 21.6 12.00 5.40 4.40 2.20 45.0 Senegal 2000 9.90 3.00 3.60 3.20 30.3 4.00 2.40 1.20 0.40 60.0 2015 15.00 7.90 3.80 3.20 52.7 6.50 5.30 0.80 0.50 81.5 Uganda 2000 24.30 0.50 13.10 10.60 2.1 2.90 0.40 2.10 0.40 13.8 2015 40.10 2.00 29.70 8.40 5.0 6.50 1.50 4.70 0.30 23.1 Source: wssinfo.org. the past 15 years, but the increase has been especially fast in urban areas. Utilities that pro- vide piped water services have been able to improve access to piped water on the premises, but the largest number of people are serviced by other means: standposts, kiosks, but also point-sources such as dependence on groundwater. Despite massive efforts, only in Burkina Faso and Uganda did the population with access to unimproved services decline, including in urban areas. Between 2000 and 2015, 16.8 million people in the five case study countries got access to piped water on the premises (of which 10.8 million were in urban areas). But most of the growth took place for other improved services which increased by 31.9 million people (of which only 7.3 million were in urban areas). So a significant part of the growth in access to water supply even in urban areas is not in piped water on the premises, but depends on other sources of improved services. The role of small private sector providers Performance of Water Utilities in Africa 77 Table 7.3. Summary of Performance of Case Study Utilities Global Africa Kenya, NCWSC Uganda, NWSC Burkina Faso, Senegal, SDE/ Côte d’Ivoire Indicator benchmark benchmark (2014) (2013) ONEA (2014) SONES (2013) SODECI (2014) NRW (m3 per connection 0.121 0.205 0.697 0.265 0.135 0.159 0.174 per day) NRW (%) — 30.3 39 35 19 20 24 Metering (%) 100 100 94 100 97 96 98 Staff-efficiency 4.27 4.21 2.10 4.34 3.04 5.24 5.83 Staff per 1,000 — 5.0 5.0 5.4 3.2 2.4 2.9 connections OCCR* 1.38 1.19 1.01 1.30 1.13 1.33 1.06 Collection ratio (%) — 91.3 91 96 97 94 86 People per connection 3.0 8.3 9.6 9.4 12.9 10.8 14.9 Reliability (hours of supply) 24 21.6 18 20 23 23 20 Affordability (%) 0.5 1.22 2.14 3.40 2.53 2.16 0.96 Water consumption per — 77.1 110 52 47 59 39 capita per day Water coverage (%) 100 77 75 78 86 98 69 Water coverage in people n.a. n.a. 2.9 3.0 3.9 5.8 11.7 served in utility (2013), millions Population in service area n.a. n.a. 3.9 3.8 4.8 5.9 17.0 of utility (2013), millions Population according to n.a. n.a. 10.1 2.0 1.4 7.9 9.1 JMP with piped water on premises (2015), millions Ranking n.a. n.a. 42 4 5 1 12 Note: the OCCR from the data collected through the TTLs and the annual reports show some discrepancy for Burkina Faso and Uganda. This may be linked to a different interpretation of the costs (including some financing costs). — = not available; n.a. = not applicable can be very significant in different parts of the continent (for example in Uganda). This paper will not look into these small-scale providers but it is important to realize that utilities in Africa in many cases operate in an environment where there are multiple providers active—including in the service area of the utilities. Table 7.3 shows that the utilities perform reasonably well against African benchmarks. At least 91 percent of connections are metered at all five utilities. There are fewer than six staff per 1,000 connections. Water supply is available, on average, for 18 hours per day or more in all five utilities. There is considerable dispersion among the cases on some indicators. Coverage in the utilities’ service area ranges from 69 percent (SODECI) to 98  percent (SDE/Société Nationale des Eaux du Sénégal [SONES]). ONEA’s NRW level (0.135 m3 per connection per day) is less than one-fourth of NCWSC’s NRW (0.697 m3 per 78 Performance of Water Utilities in Africa Figure 7.1. Impact of the Dimensions of Performance connection per day). NCWSC’s revenues barely cover operating costs, whereas SDE/SONES has an operating Operating cost coverage ratio (OCCR) of 1.33. efficiency Lessons Learned from the Case Studies The case studies show that the three measures of perfor- Service to mance (financial, operational, and customer) are linked. customers Financial This relationship is shown in figure 7.1. performance As shown in figure 7.1, operating efficiency translates into better financial performance because O&M costs are controlled. In turn, good financial performance allows a utility to make operational improvements. Better financial Finance performance allows a utility to generate cash to finance network expansion, or other projects that will minimize costs and improve water supply service. With new customers and increased consumption, revenues can increase and financial performance will continue to improve. Thus, there is a feedback loop between the three measures of performance. Finance is another import- ant element, which leads to better operational performance and customer performance. How the relationship shown in figure 7.1 works in practice was analyzed through the five case studies. Six key findings are summarized in this section: • Good financial performance enables financing, which enables utilities to expand access to piped water and provide higher quality water supply service. • Operating efficiency and financial performance are linked. • Successful utilities combine cost recovery with affordable access to water supply services. • A variety of strategies have been successful in expanding water supply access. • Incentivized contracts are successful in generating performance. • National utilities can be a management model to expand water supply services to small towns. Financing through Good Financial Performance When discussing good financial performance, it should be noted that the use of the OCCR is the bare minimum of financial performance. When a utility is able to cover its operating costs through its revenues, a utility will be able to provide services to its existing customers in the short run. In case a utility wants to ensure service provision in the medium term (it should be able to cover its depreciation) and if it wants to expand services it will have to be able to generate even more cash flow. As mentioned in Whittington et al. (2009) the vari- able costs of water service provision are relatively small in comparison with the capital costs. Performance of Water Utilities in Africa 79 Over the last 20 years, three of the utilities—NWSC, ONEA, and SDE/SONES—have turned around their performance. Figures 7.2 to 7.7 show how water coverage and reliability improved substantially in each case.1 NWSC in Uganda has increased coverage in its service area from 47 percent in 1998 to 78 percent in 2013. Water is available 20 hours per day on average, compared with 8 hours on average in 1996. ONEA in Burkina Faso has achieved similar success, with coverage increasing from 50 percent (1998) to 86 percent (2014). Water supply, once intermittent, is now available almost 24–7. In Senegal, 98 percent of the service area’s population is now served, with water available 23 hours per day on average. In all three cases, improvements in access and service were achieved even as the service area pop- ulation was expanding. How did these utilities achieve these improvements in access to piped water and water supply service? Historical data on the sources of investment financing help answer this question. The utilities were on-lent donor finance on concessional terms, at low interest rates with long grace periods. The loans were repaid with operating cash, generated by improving operat- Figure 7.2. Water Coverage, NWSC Uganda, 1998–2013 ing efficiency (thus minimizing costs) and expanding access 100 (thus increasing revenue). Some 80 of the finance was given as an Water coverage (%) equity contribution—essentially 60 a grant or investment subsidy. Table 7.4 shows the sources and 40 amounts of capital expenditure 20 for the three utilities. It is important to note that Uganda’s 0 capital cost per person served 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 was much lower than in the two 20 20 20 20 20 20 20 20 20 20 19 19 20 20 20 20 20 other countries. Figure 7.3. Reliability, NWSC Uganda, 1996–2013 NWSC, Uganda. In the early 2000s, 24 NWSC did not repay its loans with 20 operating cash. The government of Uganda agreed to a morato- 16 Hours per day rium on debt service for a period, 12 which gave NWSC financial 8 breathing space. Then, in 2007, 4 the government converted the outstanding loan balance of 0 US$47 million into equity—­ 6 97 98 9 00 01 02 03 04 05 06 07 08 09 10 11 12 13 9 9 effectively forgiving the debt. 20 20 20 20 20 19 20 20 20 20 19 19 19 20 20 20 20 20 80 Performance of Water Utilities in Africa Figure 7.4. Water Coverage, ONEA Burkina Faso, 2001–14 Since then, NWSC has borrowed from commercial banks and is 100 servicing the debt with cash gen- erated from operations. 80 One project NWSC undertook Coverage (%) 60 during the 2002–11 period was the construction of the Gaba III 40 water treatment plant and associ- 20 ated transmission mains (U Sh 52.7 billion; US$28.8 million; 0 2006). This project increased 01 02 03 04 05 06 07 08 09 10 1 12 13 14 1 water production for Kampala 20 20 20 20 20 20 20 20 20 20 20 20 20 20 and nearby areas by 80,000 m3 Note: Data pre-2006 was not available; however, it is known that reliability was low in the late 1990s and early 2000s because of a water shortage in Ouagadougou. per day. In 2010, a commercial loan of US$2 million was obtained for financing the extension of the Figure 7.5. Reliability, ONEA Burkina Faso, 2006–14 Ggaba intake plant, which supplies water to Kampala city and the surrounding areas. This loan is being serviced from 24 the operating cash flow. 20 Increasing operating cash flow was key to achieving 16 expansion in service. NWSC increased collection efficiency Hours per day 12 (from 85 percent in 2001 to 95 percent in 2011), reduced NRW (from 43 percent in 2001 to 33 percent in 2011), and 8 increased labor productivity by limiting staff growth as 4 connections increased. Real tariffs increased at a modest 3 0 percent per year. Together, these factors provided an oper- ating cash surplus that was used repay debt. However, as 95 06 07 08 09 10 11 12 13 14 20 20 20 20 20 19 20 20 20 20 can be seen from the last four years of audit reports, NWSC Note: Data pre-2006 was not available; however, it is known that reliability remains dependent on grants for financing its investments was low in the late 1990s and early 2000s because of a water shortage in Ouagadougou. (table 7.5). ONEA, Burkina Faso and SDE/SONES, Senegal. In the 1990s, the cities of Ouagadougou and Dakar both faced water shortages. Supply was rationed and water coverage was low. ONEA and SDE/SONES—the utilities that serve these cities, respectively—both received large loans to finance bulk supply. The Bank was a major financier in both cases. For ONEA, a major program was the Ouagadougou Water Supply Project (US$269 million, 2001–07). The program included construction of the Ziga Dam, Boudtenga Reservoir (5,400 m3), a water treatment plant and pumping station, and extension of the distribution net- work. Increased water production led to near perfect water supply reliability (23 hours per day), whereas before the project, service was intermittent. Performance of Water Utilities in Africa 81 Figure 7.6. Water Coverage, SDE/SONES Senegal, 1995–2013 In Senegal, two major programs were implemented—the Senegal 100 Water Project (US$223 million, 1996–2004) and the Long Term 80 Water Project (US$255 million, Coverage (%) 60 2002–09).2 The first project focused on urgent investments 40 needed to increase water supply 20 in Dakar. This included addi- tional boreholes, expansion of a 0 treatment plant, and leakage 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 reduction works. The second 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 phase of reforms involved the construction of the much larger Figure 7.7. Reliability, SDE/SONES Senegal, 1996–2013 Keur Momar Sarr water treatment plant (in 2005, 65,000  m3 per 24 day; upgraded in 2008 to 130,000 20 m3 per day) along with a contin- ued expansion in the distribution 16 Hours per day network. 12 Programs in both countries 8 introduced private sector partici- 4 pation, which led to operational efficiency, which in turn led to 0 increased operating cash flow. At 09 10 11 12 13 96 97 98 99 00 01 02 03 04 05 06 07 08 20 ONEA, Veolia was hired under a 20 20 20 20 19 20 20 20 20 19 19 19 20 20 20 20 20 Table 7.4. Sources and Amounts of Capital Investment Financing NWSC (2002–11) ONEA (2002–13) SDE/SONES (1996–2013) Incremental coverage between start and end 1,112,387 2,632,000 3,230,391 Estimated total capital investment (US$, millions) a 100 600 770 Percentage grant-financed 28 52 29 Percentage financed by internal cash flow 52 19 23 Percentage financed by loans 16 29 47 Capital investment per additional person 90 228 238 served (US$) a. For NWSC, total capital expenditure was estimated using cash flow statements. Cash outflow was summed from the following financing activities: capital work-in-progress; purchase of property; plant and equipment; and purchase of computer software. For ONEA and SDE/SONES, investment data was provided by the Bank. 82 Performance of Water Utilities in Africa Table 7.5. Sources of Financing in NWSC (US$,000) 2011 2012 2013 2014 Cash flow from operations 14,201 (15,735) 13,831 (7,635) Investment financing 33,933 31,451 45,988 51,408 Total cash flow from operations and investments 19,732 47,186 32,157 59,043 Financing Payments of borrowings (2,114) (3,894) 0 (4,056) Proceeds from grants 11,562 51,859 38,749 85,474 Increase in cash equivalents (10,344) 778 6,593 22,375 Note: Annual Reports from NWSC, Auditor General. performance-based service contract in 2001 to help manage the commercial function. As a result, collection efficiency rose from 78 percent in 2002 to 95 percent in 2006. This has been sustained and even improved, measuring 97 percent in 2013. Low levels of NRW were main- tained. In Senegal, the private operator (SDE) was brought in through the  Senegal Water Project. Signing a public–private partnership (PPP) contract satisfactory to the Bank was a condition of the Bank loan. SDE steadily reduced NRW from 1996 to 2003 (29 percent to 20 percent) because the affermage contract included strong incentives to do so. Collection rates were maintained at 95 percent or above. Correlation of Operational and Financial Performance The case studies show an association between good operating efficiency and financial per- formance. Higher levels of operating efficiency are correlated with higher levels of financial performance. Good financial performance can provide financial resources that allow the utility to invest in further operating efficiencies, for example by replacing leaky mains, worn-out meters, and inefficient, unreliable pumps. The relationship between operational and financial performance at NWSC, ONEA, and SDE/SONES is described in the previous section on financing investment. A similar rela- tionship is present at SODECI, whose peak performance on operational and financial indicators was in 2000. The OCCR was 1.09 and the collection ratio 92 percent. NRW was 131 liters per connection per day, or just 17 percent in percentage terms. There was an average of just three staff per 1,000 water connections. During the Ivorian Civil Wars (2002–11), however, financial and operational performance declined. Since 2011, perfor- mance has improved. As of 2014, SODECI is again a good performer. It ranks 12th of all utilities in the econometric analysis (though its performance is still below the level it reached before the war). Nairobi’s utility, NCWSC, provides a contrast to the other four utilities. Its OCCR is 1.01, which indicates that revenue barely covers operating costs. NRW is also relatively high, at 697 liters per connection per day—partially the result of inefficiencies and partially the Performance of Water Utilities in Africa 83 result of significantly higher levels of water consumption than are provided in the other four utilities. Although collection efficiency has improved, the lack of cash generated from operations limits resources available for fixing pipes and replacing meters. Combining Cost Recovery with Affordable Water Supply Service It is often assumed that there is a trade-off between cost recovery and affordable service. And indeed, all of the case study utilities—with the exception of Nairobi and to some extent SDE in Senegal—are providing low levels of water consumption (especially low levels of residential water consumption) and show lower than average levels of afford- ability compared with African benchmarks (table 7.3). Yet, the use of cross-subsidies in several of the case study utilities shows that this tool can to a large extent transcend this trade-off. There are two basic cross-subsidization strategies: • Charging nonresidential customers more than residential customers • Selling a basic needs quantity of water at a below average tariff and selling water above the basic needs quantity at an above average tariff NWSC’s tariff structure is an example of the first strategy (figure 7.8). Residents are charged less than nonresidents, at a rate below the average tariff, no matter how much they consume. Senegal’s tariff structure is an example of applying both strategies in parallel (figure 7.9). NWSC, Uganda. NWSC charges households US$0.77 per m3, a rate slightly below the weighted average tariff (US$0.82 Figure 7.8. NWSC, Uganda Tariff Structure per m3). The standpipe tariff is even lower at US$0.47 per m3. To compensate for the low rates charged to house- 4.0 holds, the tariff for institutional and commercial custom- 3.5 ers is above average, at US$0.91 per m3 and US$1.12 per m3, 3.0 respectively. However, to encourage large users to stay on Tariff (US$ per m3) 2.5 the system, the tariff for commercial consumption above 2.0 1,500 m3 per month (at US$0.95) is lower than the tariff for 1.5 consumption below that amount (US$1.12). 1.0 0.5 SDE and SONES, Senegal. Senegal’s increasing block tariff 0 0 50 100 1450 1500 1550 structure has a subsidized social tariff for levels of con- Monthly consumption (m3) sumption below 20 m3 (CFAF 202; US$0.40) in a two- Commercial Domestic month period. There is also a regular tariff for consumption Government Standpipe from 21 m3 to 40 m3 (CFAF 697.97; US$1.39), and a “dissua- Average sive” tariff for consumption above 40 m3 (CFAF 878.35; Source: NWSC, Annual Report 2012/2013. US$1.75). The dissuasive tariff is designed to be a 84 Performance of Water Utilities in Africa Figure 7.9. SDE/SONES, Senegal Tariff Structure disincentive for excessive water use. It can be seen that the tariff for household consumption of less than 20 m3 4.0 per 60 days is less than a third of the regular tariff, and less than a quarter of the tariff for consumption in the top 3.5 block. Only the social tariff and the standpipe tariff (CFAF 366; US$0.73) are below the average tariff (CFAF 494; 3.0 US$1.08). 2.5 Nonresidential, nongovernmental customers must pay Tari (US$ per m3) the dissuasive tariff regardless of the amount consumed. 2.0 Government customers pay more than twice the dissua- sive tariff—their tariff is CFAF 1,868.88 per m3 (US$3.72 1.5 per m3). The structure was established in 2007. In that year, the government agreed to raise tariffs for govern- 1.0 ment customers by 70 percent, while keeping tariffs for other customers constant.3 This cross-subsidy structure 0.5 was introduced as a way to keep domestic tariffs from ris- ing while still ensuring cost recovery for SONES/SDE. In 0 0 50 100 150 200 2015, domestic tariffs were again raised—the lowest tariff Bi-monthly consumption (m3) block was raised by 4 percent and other rates were Government Average Domestic increased by 9 percent. Commercial Standpipe As can be seen in table 7.6, the case study utilities have been using cross-subsidization to a large extent to facili- Source: SONES, “Grille tarifaire, 5éme Bimestre 2014.” tate water supply services (that is, not including any wastewater services); however, in three of the case study utilities, the residential water consumption is still significantly below the level that the World Health Organization (WHO) recommends for good health (that is, below 50 lcd). Share of Nonresidential Customers in Consumption by Volume. Both SDE and NWSC rely on nonresidential customers to cross-subsidize residential consumption. At NWSC, all domestic customers, regardless of amount consumed, are charged at a rate below the average tariff. This type of tariff structure has been in place at NWSC since at least 2005. Table 7.6. Effect on Residents of Cross-subsidization in the Five Case Study Utilities Burkina Faso, Senegal, SDE/ Côte d’Ivoire, Type of consumer Kenya, NCWSC Uganda, NWSC ONEA SONES SODECI Average affordability (%) 2.14 3.40 2.53 2.16 0.96 Affordability for residential 1.11 1.05 1.62 0.58 0.32 consumers (%) Residential consumption per 70 23 39 55 32 capita per day (lcd) Performance of Water Utilities in Africa 85 Since then, the share of commercial consumption has risen from 21 percent to 33 percent (by volume), as commercial consumption by volume has risen by an average of 11.1 per- cent per year. These trends indicate that charging commercial customers at rates above the average tariff has not resulted in reduced consumption. The decreasing block tariff for commercial customers could also be a reason the cross-subsidization mechanism contin- ues to work well. Senegal’s tariff structure also relies heavily on nonresidential consumption for cross-­ subsidies. However, unlike NWSC, nonresidential revenues have decreased over time, from 39 percent of all revenue billed (2004) to 29 percent (2013); over the same period, the volume of water sold to nonresidential water users dropped to only 8 percent in 2013. High nonresi- dential tariffs could be a contributing factor to this trend. These cross-subsidies are summa- rized in the table 7.7 The takeaway from these cases is that cross-subsidization can be helpful in keeping the service more affordable as long as certain principles are followed. The average tariff should be sufficient for cost recovery. To ensure affordability for households, residential tariffs can be set lower than the average tariff. Commercial tariffs or tariffs for high levels of consump- tion can be set higher than the average tariff, but not too high. To keep nonresidential users connected to the piped network, nonresidential tariffs must be set lower than the cost of alternate water sources. Strategies for Expanding Access Utilities have employed different strategies to expand access in their service area. SDE is an example for serving a large proportion of all customers with piped water to their premises (89 percent). Other utilities—such as ONEA (Ouagadougou) and NWSC (Kampala)—serve about half of the population with public taps. Household survey data for Dakar and Ouagadougou are shown in figure 7.10 and figure 7.11.4 ONEA’s strategy is to focus first on achieving good access for the poor through public taps, then to increase individual connections. Household survey data for Ouagadougou, Table 7.7. Effect of Cross-Subsidization on Consumption Patterns in the Five Case Study Utilities Burkina Faso, Senegal, SDE/ Côte d’Ivoire, Type of consumer Kenya, NCWSC Uganda, NWSC ONEA SONES SODECI Share in volume of water sold (%) Residential 65 44 85 92 76 Nonresidential 35 56 15 8 24 Share in billed revenue (%) Residential 60 n.a. 65 71 n.a. Nonresidential 40 n.a. 35 29 n.a. Note: n.a. = not applicable. 86 Performance of Water Utilities in Africa Figure 7.10. Access to Water, Dakar, 2000–14 the capital, show the utility is executing this strategy successfully (see figure 7.11). In 1993, 28 percent of peo- 100 ple in the city reported access to piped water to the 90 premises. This proportion rose to 37 percent in 2003 and Access, total population (%) 80 47 percent in 2010. ONEA plans to serve 80 percent of its 70 60 service area population with piped water to the premises 50 by 2030. 40 30 Generating Performance through Incentivized Contracts 20 Credible commitments through legally binding contracts 10 between public and private parties, incentivized contracts 0 with senior management teams, performance-based con- 2000 2005 2011 2014 tracts, and multi-stakeholder frameworks have worked to Public tap Piped to premise sustain successful governance and utility management models. The four well-performing utilities—NWSC, SDE/ SONES, ONEA, and SODECI—all have some form of incen- Figure 7.11. Access to Water, Ouagadougou tivized contract in place. 100 90 Affermage Contract. In 1995, the Government of Senegal 80 (GoS) split SONES, the existing national utility, into three Access, ouagadougou only (%) 70 entities: an asset-holding company for water service 60 assets (SONES), a private operator for water services 50 engaged via an affermage contract (SDE), and a public, 40 combined asset-owner and operator for wastewater ser- 30 vices (ONAS). SDE has an affermage contract with SONES and the government. SONES and ONAS have entered into 20 performance contracts with the GoS. SODECI in Côte 10 d’Ivoire operates under a similar contract with the govern- 0 1993 1995 2003 2010 ment. SDE’s affermage contract included strong incentives Public tap Piped to premise to reduce NRW and improve collection efficiency. The desired results were achieved—NRW fell from 29 percent in 1996 to 20 percent in 2003. Collection efficiency aver- aged 97 percent from 1996 to 2013. These performance improvements allowed the utility to generate operating cash that financed investment and service improvement. The affermage contract also helped increase the durability of service improvements. The contract defined the asset regime, service standards and conditions, regime governing the works, the remuneration regime for the operator, monitoring mechanisms, and sanctions. While a contract between a public utility and its government owner is generally easily amended or ignored if the government so wishes, a contract with a private party cannot be changed without the consent of both parties. Performance of Water Utilities in Africa 87 Incentivized Management Contract. NWSC’s internally delegated management contracts (IDAMCs) have been another successful initiative. In 2004, each town was established as a business unit and managers were held accountable for meeting set performance targets. Kampala was further divided into branches, each responsible for operational activities in its service area and incentivized to meet targets set in the branch performance contract. Those who do meet the targets can earn bonuses up to 120 percent of their gross salary, whereas those who do not can attract a penalty of up to 25 percent of their salary. This con- cept was derived in part from two initial management contracts with private operators (Gauff and Ondeo) from 1998 to 2004. Gauff was contracted to undertake a massive over- haul of water services, including deploying geographic information systems (GIS), improv- ing metering and billing procedures, achieving yearly connections targets, and reducing NRW. Ondeo’s contract had similar objectives. These contracts rapidly improved NRW, which fell from 50 percent to 38 percent from 1998 to 2004. The collection ratio and staff productivity also rose. When the second management contract with Ondeo ended, NWSC maintained these performance standards without private sector involvement. The IDAMCs, as described above, ensured sustained performance. Performance-Based Contract for the Commercial Function. When ONEA needed donor finance for a large bulk water scheme, the Bank suggested an affermage contract, citing Senegal’s recent success. However, the Burkina Faso government was committed to public sector control of the utility. An alternative model was developed involving strong multi-­ stakeholder accountability arrangements and a performance-based contract with a spe- cialist firm to boost ONEA’s commercial performance. Financing partners found this proposal credible, and the bulk water scheme and institutional reforms to ONEA pro- ceeded in parallel. ONEA had a performance-based service contract for commercial management with Veolia from 2001 to 2006 (Marin, Fall, and Ouibiga 2010). Veolia provided two deputy managers, plus other short-term advisers, for ONEA’s commercial and finance departments. They set up new accounting and customer management systems, and helped ONEA identify illegal customers, improve meter reading and meter repairs, and improve customer service. After an initial decline from 85 percent to 78 percent, collection efficiency rose to 95 percent by the end of the contract. ONEA has maintained high cash collection rates above 95 percent since the end of the contract. This includes collections from the government, which is obliged to settle its water bills according to the terms of the performance contract. Formal Supervision Involving Multiple Stakeholders. Another innovative model used by ONEA is the multi-stakeholder supervision committee at ONEA. ONEA’s management credits this committee with an important role in making sure that both the utility and the govern- ment play their agreed roles under the contract plan (performance contract with 88 Performance of Water Utilities in Africa operational  targets). The multi-stakeholder committee comprises representatives of cus- tomers, NGOs, and donors. The committee monitors performance of both the utility and the government under the contract, on the basis of independently audited financial and techni- cal reports. The committee’s monitoring role is centered on an annual meeting. Prior to the meeting, committee members receive not just a report from ONEA on its performance against the contract, but also the report of a financial auditor and a technical auditor, whose job it is to assure the quality of the information. The auditors’ reports indicate the degree of confidence they have in the information presented. The auditors appear in person before the committee, and explain their reports. National Utilities as a Model for Expansion of Water Supply Service There is a continuous debate about what the optimal scale for utilities is, and whether util- ities should be organized and controlled at the national, regional, or local level. All five utilities in this sample are large, serving close to or more than 3 million people. Four of the five are national utilities. Thus, based on the case study sample, it is not possible to address the optimal scale issue. However, the case studies do show that national utilities can be successful in expanding water supply service to small towns. In Uganda, the NWSC service area grew from 3 towns (1972) to 15 towns (2003) to 23 towns (2013).5 The original three were relatively large urban centers—Kampala, Jinja, and Entebbe. Since then, the utility has expanded to smaller towns with preexisting networks. This arrangement works well because the smaller towns benefit from the managerial and techni- cal know-how of NWSC. Also, the towns cross-subsidize each other. In 2013, 14 of 23 towns served had average unit production costs that were higher than the weighted average tariff. Kampala, which accounts for more than 60 percent of revenue, has the lowest average unit production cost. In Côte d’Ivoire, the concession contract awarded to SAUR in 1959 was exclusively for water services in Abidjan, the capital. Over the next 15 years, SODECI (the operator) gradu- ally signed contracts with local governments in 10 other towns. In 1974, SODECI became responsible for all water supply systems in the country. As of 2014, SODECI serves nearly 900 towns. Similar patterns are present in Senegal and Burkina Faso. The case studies cannot con- clude that national utilities are performing better than regional or local utilities in providing water supply service to smaller urban centers. However, they do show that national utilities can provide water supply services to smaller towns. Conclusions The case studies show that well-performing utilities do exist in Africa. The utilities in the case studies show that operating efficiency and financial performance are linked. Successful utilities are able to ensure that they cover their operation and maintenance Performance of Water Utilities in Africa 89 costs and generate additional funds to at least cover (part of the) depreciation of the existing infrastructure and possibly allow for some debt repayment. Nevertheless, even though the five case studies show that utilities in Africa can generate cash to pay for the  operation and maintenance, depreciation, and some debt repayments, many of these  utilities are not able to depend on only their own funds to pay for all of their investments. Affordability remains a challenge also in the case study utilities. Increasing water rates do not necessarily result in higher revenues if nonpayment increases. In some cases, top-performing utilities have used cross-subsidies to much effect to reduce the burden on the residential consumer, but this has at times resulted in a very small customer base, with nonresidential water users opting out. Of the five case study utilities, most were able—with varying degrees of cross-subsidization—to make the water supply services more affordable. Only two of the five case study utilities—Nairobi’s NCWSC and Senegal’s SDE—were providing piped water services at levels above the thresholds set by WHO as necessary for good health and hygiene (that is, at least 50 lcd provided to residential water consumers). Cross-subsidies that ensure long-term financial viability will require that the higher tariffs charged to nonresidential users are not set higher than the costs of alterna- tive water supplies (such as groundwater) so that they opt out of the piped network sys- tem. If these tariff prices are set too high, in the long run, the nonresidential basis on which to provide cross-subsidies to residential consumers will erode. Incentivized contracts can be successful in improving performance. Credible commit- ments through legally binding contracts between public and private parties, incentivized contracts with senior management teams, performance-based contracts, and multi-stake- holder frameworks have worked to sustain successful governance and utility management models. The four well-performing utilities—NWSC, SDE/SONES, ONEA, and SODECI—all have or had some form of incentivized contract in place. Notes 1. The focus of this section is how utilities have financed improvements in their water supply service. The case studies show that expanding a piped network has been successful in these utilities. For technical and financial reasons, the same does not apply for the sewerage network. Due to the complexity of increasing access to sanitation services, and because this responsibility is outside the mandate of some utilities in the sample, water supply is the focus of this section. 2. Implementation Completion Report, Senegal Water Project; Implementation Completion Report, Long Term Water Project. Note: These projects included sanitation components, which are overseen by ONAS, not SDE or SONES. 3. This very high water tariff for government customers essentially amounts to a government subsidy to the utility. 4. All data are from the Demographic and Health Survey (DHS) except Dakar, 2000, which is from the Multiple Indicator Cluster Survey (MICS). Dakar and Ouagadougou each account for about half of the total service area population of SDE and ONEA, respectively. This access data applies only to the capital cities and differs from that for the entire service area which in both utilities includes areas and towns outside the capital city. 5. NWSC operates in 146 towns as of March 2016. However, for the case studies, performance after 2014 was not studied. 90 Performance of Water Utilities in Africa References Marin, Philippe, Matar Fall, and Harouna Ouibiga. 2010. Corporatizing a Water Utility: A Successful Case Using a Performance Based Service Contract for ONEA in Burkina Faso. Gridlines Note No. 53. Washington, DC: PPIAF, World Bank. Whittington, D., W. M. Hanemann, C. Sadoff, and M. Jeuland. 2009. “The Challenge of Improving Water and Sanitation Services in Less Developed Countries.” Foundations and Trends in Microeconomics 4 (6–7): 469–609. Performance of Water Utilities in Africa 91 © Alexander Danilenko/World Bank. Further permission required for reuse. Chapter 8 Lessons Learned There is a lack of agreement on what constitutes good performance in utilities. In this report, good performance has been defined as utilities that provide water and wastewater services that performance are efficient, affordable, sustainable, and with a minimum service quality. Hence, ­ has many different dimensions. A definition that focuses on these different aspects of financial, operational, and customer performance—has been used. When disag- performance—­ gregating the different elements measuring financial performance, customer performance, and operational performance, while separately addressing water coverage, it was found that these four indicators are in general not very strongly correlated (with the exception of water coverage and customer performance), meaning that good performance in one aspect of performance does not automatically predict good performance in other aspects. Even in the case study util- ities, which were in general well-performing, some aspects of good performance—especially with regard to customer performance—were less well articulated. It was found that incentive structures within and outside the sector can distort aspects of performance. The findings of literature about economies of scale and scope were confirmed, while the overall economic management in the economy and governance also matter in how well utilities are performing. Lesson 1: Although Utilities in Africa in General Underperform, There Are Relatively Well-Performing Utilities Operating in the Continent Well-performing utilities are in general doing well in terms of operational and financial performance. It was found that good performance varies widely between countries and also ­ within countries. This suggests that water is indeed a local service and that local Performance of Water Utilities in Africa 93 circumstances have a big impact on how well utilities are performing. Unfortunately, it was not possible to capture all these local particularities in the data collection. Variation in local ­circumstances can include, among others, the distance to water sources, the quality of water resources available to the utilities and water users, the availability of other water resources, spatial patterns and levels of economic development that affect the cost of infrastructure and service delivery, the types of consumers, the willingness of water users to connect and pay for utility services, but also the quality of management. Lesson 2: Customer Performance Is Relatively Weak Even among the Best-Performing Utilities Customer performance is not necessarily very well developed in African utilities, especially not when compared with global benchmarks. Water tends to be supplied for fewer than 24 hours a day. In addition, many utilities provide relatively low levels of service compared with utilities elsewhere in the world in terms of a much heavier reliance on shared connec- tions and public standposts. This results in significantly lower levels of consumption per capita. Residential water consumption tends to be very low; in many utilities, less than 50 lcd (the minimum volume of water set by World Health Organization [WHO] to guarantee good health and hygiene). Hence, even though tariffs are not necessarily low, the very low consumption levels provide utilities with a relatively weak revenue basis (especially when the utility is small in size). This low consumption also makes it less compelling for water consumers to use or connect to the service compared with other water sources at least in terms of service delivered (that is, consumption levels, convenience of service which is limited if households share connections and/or are dependent on public taps, and the often ­ high number of supply interruptions). It was found that higher levels of service as measured by customer performance are positively connected to water coverage. Hence, when alterna- tive water sources are available, the low levels of service provided can deter households from connecting to the piped water network. Lesson 3: The Major Drivers of Water Utility Performance Are Linked to the Cost of Service, While an Enabling Environment as Reflected in Good Economic Management also Matters It was observed that one of the major drivers of utility performance is economies of scale: size (as measured by the size of the population served with water) matters, but when a utility becomes too large the benefits become disadvantages. Economies of scope also matter. Utilities with sewerage coverage tend to show better performance, although this result has to be interpreted against a context in which sewerage coverage is very low especially in low-income countries in Africa; and this may be linked with higher gross domestic product (GDP) in environments where utilities that provide wastewater services are located. Another finding, linked to the economies of scale and scope, is the impact of costs (either measured 94 Performance of Water Utilities in Africa by the operation and maintenance [O&M] costs per cubic meter of water produced or the share of labor in the total costs): higher O&M costs per cubic meter of water produced are linked with lower utility performance. A large part of the O&M costs are the result of investment decisions, and these decisions lock in costs for decades to come. It is, therefore, important to assess investment decisions properly. Some of the other drivers of utility ­ performance are beyond the control of the utilities. The quality of economic management in the countries in which utilities operate is affecting utility performance: the higher the qual- ity of economic management, the higher the utility performance. Lesson 4: Improving Water Coverage in Africa Will Require Large Investments that Will Have to Be Mostly Paid for by Government Funds Customer performance (that is, service quality) has a direct impact on improvements in water coverage. Better economic management and higher gross national income (GNI) growth have a positive impact on coverage as it is likely linked to the availability of invest- ment funding for the sector. As per capita income rises, a larger amount of money can be spent on water, ensuring a more sustainable footing for utilities. Experience shows that most utilities are able to provide more or less universal access to water supply only when they reach a certain level of economic development; and even higher levels of economic develop- ment are needed to ensure proper management of wastewater services. In the sample of utilities, it was found that the link between utilities’ financial performance and water coverage was only in place for certain groups of utilities. For other utilities, the contribution of good financial performance to water coverage is very weak. This suggests that most utilities, especially smaller ones in low-income countries are not able to improve access through improved financial performance but depend on external funds to do so. The case studies confirmed these findings: many of the best performing utilities still depend on some form of government funding to finance their investments. Hence, increases in water coverage will have to be matched by government priorities to fund the sector, especially as the current investment in the sector is mostly limited to water supply (that is, most utilities in Africa, especially in low-income countries, do not provide wastewater services). Utilities with few exceptions are also not able to provide services effi- ciently, as was reflected in the results of the data envelopment analysis (DEA)—most utili- ­ vailable—showed ties were very far from the DEA frontier. The case studies—where data were a that more efficient levels of investment spending allow more financial space for utilities. As many utilities are underperforming, government funding for the sector should be subject to much more stringent criteria for investment selection and priority setting as these ­ decisions will determine the O&M costs for decades after. The investment selection criteria must be linked to, among others: (a) More comprehensive processes and methods for investment project selection (with proper technical, economic, financial, and social due diligence); Performance of Water Utilities in Africa 95 (b) Sufficient attention to the performance of the utility to operate and maintain its existing and new infrastructure so as to ensure the long-term sustainability of water (and waste- water services) by ensuring that these investments will generate the most value for money and through the inclusion of all life-cycle costs in investment decision making; (c) Ensuring that investment expansion and institutional capacity to manage the new infra- structure assets are in synch; and (d) Putting in place adequate and transparent sector financing policies (that is, tariff and subsidy policies that ensure that the assets can be properly operated and maintained). Lesson 5: Specific Measures Are Needed to Ensure that Progress in Financial Cost Recovery Does Not Translate into Less Affordable Services The better performing utilities tend to charge higher water rates to their customers. Yet, affordability remains a significant challenge for many utilities, even though most utilities in  Africa provide only water services.1 In some cases, top-performing utilities have used cross-subsidies to much effect to reduce the burden on the residential consumer, but this  has  at times resulted in a very skewed customer base, with nonresidential water users opting out. In the five case study utilities, most of these utilities were able—with cross-­ subsidization—to make the water supply services more affordable. Yet, only two of the utilities (Nairobi’s NCWSC and Senegal’s SDE) were able to provide piped five  case study ­ water services at levels that are above the thresholds set by WHO as necessary for good health and  hygiene (i.e., 50–100 lcd for residential water consumers). Cross-subsidies can help make the service more affordable, but care should be taken to ensure that the high tariffs charged to nonresidential users are not set higher than the costs of alternative water supplies (such as groundwater) so that the nonresidential users opt out of the piped network system. Lesson 6: Availability of Data is Critical to Assess Performance and Guide Sector Planning To enable anyone to investigate the performance of utilities in Africa and elsewhere, avail- ability of reliable and complete data is critical. For this analysis, the team could benefit from a tested instrument like the IBNET Toolkit. Even so, the quality and, especially, the completeness of the collected data remained a major challenge. Organizational data were ­ ­ collected but overall response rate for this type of information was low; utilities do not always collect even basic data on their performance. There is a huge demand from many professionals for more data beyond the basic financial and operational data collected for this study. Basic performance monitoring is most common in countries where a regulator is active. In countries where utilities have no specific regulator in place, performance monitor- ing is generally underdeveloped. Yet, it is hard to improve performance without reliable data and basic reporting mechanisms in place that are available to the public (externally audited financial statements are often missing). 96 Performance of Water Utilities in Africa It should be noted that focusing on a very small set of indicators may result in a distorted picture. For instance, cost recovery—at least O&M cost recovery—has become a major area of focus to measure utility performance. Although, in theory, many utilities in Africa cover their O&M costs, the actual cash flow performance of utilities is a major challenge because collection efficiencies are typically significantly below 100 percent. This means that the cash flowing into the utilities is insufficient to cover the basic expenditures, resulting in a decline in service levels and slow progress in increasing access to piped water supply. Data on collection efficiency (let alone working capital) are often either left unreported or are only ­ partially reported (with certain groups of consumers missing) and so, on paper, a utility may be able to generate sufficient billed revenue to pay for its O&M. Finally, most utilities have little insight into their customers’ behavior. There are large dis- crepancies between the data provided by utilities on the access to water supply services and those registered in household surveys. Utilities should at frequent intervals calibrate their insights into their customers’ behavior to better predict the demand for their services and the investment plans that are based on this demand. Note In general, the provision of wastewater services is significantly more expensive than the provision of water supply 1.  services. Performance of Water Utilities in Africa 97 Appendix A Customer Performance and Water Coverage Performance of Water Utilities in Africa 99 100 Table A .1. Customer Performance and Water Coverage, National Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Benin SONEB S.A.U. 68.44 24.0 37.94 1.5 0.752 0.814 0.628 0.752 Burkina Faso ONEA 78.63 23.0 47.44 0.83 0.793 0.783 0.690 0.711 Congo, Dem. REGIDESO 25.69 69.59 0.63 0.303 0.370 0.454 0.493 Rep. Cote dIvoire SODECI 66.74 16.6 39.33 1.04 0.645 0.721 0.621 0.731 Guinea-Bussau EAGB Guinea- 21.23 8.0 166.12 0.483 0.612 0.558 0.705 Bissau Lesotho Lethoto WASCO 56.00 18.0 135.42 0.98 0.770 0.799 0.780 0.816 Malawi LWB 65.79 20.0 92.05 0.86 0.830 0.819 0.829 0.815 Mali SOMAGEP-Mali 67.68 72.71 1.15 0.779 0.834 0.668 0.779 Niger SPEN 86.71 22.0 63.85 0.851 0.888 0.818 0.878 Rwanda Rwanda WASAC 77.72 12.0 31.01 0.612 0.709 0.667 0.778 Senegal SDE 97.94 22.8 58.85 1.39 0.874 0.906 0.811 0.874 Togo SODECA 40.01 40.38 0.65 0.326 0.392 0.489 0.522 Uganda NWSC 77.82 20.0 51.97 1.36 0.739 0.804 0.691 0.794 Table A .2. Customer Performance and Water Coverage, Regional Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Kenya Tavevo 72.13 103.99 0.1 0.907 0.702 0.861 0.603 Kenya Thika 95.09 15.0 94.93 1.12 0.875 0.906 1.000 1.000 Kenya Runda 100.00 15.0 145.21 1.09 0.875 0.903 1.000 0.996 Kenya Nyeri 85.07 14.07 95.65 1.08 0.862 0.893 1.000 0.995 Kenya Garissa 86.00 40.14 1.19 0.859 0.894 0.789 0.859 Kenya Nairobi 75.23 18.0 110.38 1.18 0.834 0.876 0.876 0.917 Kenya Kericho 77.33 21.02 0.95 0.770 0.793 0.656 0.724 Kenya Kakamega Busia 72.51 37.56 1.88 0.766 0.824 0.648 0.766 Kenya Mandera 25.84 102.44 1.09 0.753 0.812 0.629 0.749 Kenya Malindi 85.02 15.0 59.76 0.98 0.750 0.785 0.812 0.838 Kenya Moyale 66.67 8.22 1.0 0.727 0.773 0.590 0.698 Kenya Naivasha 64.73 7.39 1.39 0.720 0.790 0.580 0.720 Kenya Eldama Ravine 56.25 44.14 0.87 0.716 0.734 0.574 0.646 Kenya Mombasa 56.79 41.6 1.05 0.716 0.776 0.574 0.701 Kenya Nyanas 61.45 1.9 0.86 0.706 0.725 0.559 0.634 Kenya Tililbei 54.89 24.96 0.89 0.698 0.725 0.547 0.633 Kenya Kiamumbi 77.78 15.0 62.62 1.83 0.678 0.759 0.705 0.803 Kenya Isiolo 40.00 61.64 1.02 0.676 0.739 0.515 0.652 Kenya Eldoret 71.73 15.0 72.72 0.83 0.667 0.690 0.689 0.711 Kenya Nanyuki 91.23 8.75 64.28 1.3 0.667 0.750 0.818 0.879 Kenya Rumuruti 45.45 16.44 0.95 0.661 0.712 0.492 0.616 Kenya Muranga 78.13 14.16 53.15 0.98 0.660 0.717 0.694 0.759 Kenya Gusii 45.20 12.3 0.92 0.658 0.703 0.487 0.604 Kenya Machakos 41.21 14.03 1.19 0.646 0.734 0.468 0.646 Kenya Nakuru Rural 24.22 68.71 0.88 0.630 0.672 0.445 0.563 Kenya Kisumu 66.80 15.0 37.51 0.99 0.621 0.692 0.620 0.714 Kenya Kapenguria 29.31 37.07 0.83 0.621 0.654 0.432 0.539 table continues next page 101 102 Table A .2. continued Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Kenya Embu 62.25 14.17 67.91 1.19 0.620 0.715 0.634 0.756 Kenya Kiambu 34.74 11.7 85.51 1.01 0.612 0.687 0.674 0.754 Kenya Nakuru 93.47 5.53 51.81 1.12 0.612 0.709 0.802 0.868 Kenya Iten Tambach 19.23 57.53 2.45 0.604 0.703 0.405 0.604 Kenya Amatsi 24.06 35.96 1.59 0.603 0.702 0.404 0.603 Kenya Mavoko 79.62 11.63 14.9 1.96 0.602 0.702 0.661 0.774 Kenya Lamu 68.18 11.7 60.27 0.99 0.598 0.673 0.654 0.735 Kenya Meru 58.72 13.47 66.78 1.05 0.597 0.686 0.615 0.727 Kenya Ruiru Juja 60.78 15.0 21.43 1.29 0.590 0.693 0.573 0.715 Kenya Gulf 20.43 28.84 0.82 0.586 0.626 0.379 0.501 Kenya Kilifi Mariakani 55.76 15.0 23.14 0.93 0.575 0.643 0.550 0.650 Kenya Tarda Kiambere 69.14 11.43 19.08 0.89 0.567 0.627 0.613 0.678 Kenya South Nyanza 12.04 15.53 1.42 0.549 0.662 0.324 0.549 Kenya Kitui 46.20 15.0 12.54 1.08 0.536 0.647 0.492 0.654 Kenya Limuru 40.00 14.99 32.72 1.04 0.529 0.634 0.481 0.637 Kenya Nzoia 62.77 8.82 39.85 1.02 0.524 0.625 0.602 0.710 Kenya Nyahururu 48.18 11.26 51.69 1.12 0.518 0.639 0.543 0.695 Kenya Kibwezi Makindu 38.38 14.98 21.07 0.84 0.515 0.576 0.461 0.560 Kenya Karuri 59.21 9.56 17.96 0.83 0.507 0.570 0.562 0.627 Kenya Olkalou 36.36 15.0 7.83 2.8 0.501 0.626 0.439 0.626 Kenya Narok 36.96 10.27 70.91 1.81 0.484 0.613 0.512 0.675 Kenya Mathira 31.54 2.89 82.19 0.97 0.479 0.579 0.658 0.731 Kenya Mwala 26.92 15.0 11.74 0.99 0.472 0.578 0.395 0.562 Kenya Lodwar 47.86 7.62 32.78 1.6 0.452 0.589 0.520 0.680 Kenya Wote 20.31 15.0 14.75 1.03 0.451 0.572 0.364 0.554 Kenya Kapsabet Nandi 45.45 5.74 60.27 0.94 0.440 0.543 0.540 0.645 Kenya Kikuyu 26.42 9.57 54.14 0.94 0.424 0.531 0.437 0.575 Kenya Namanga 57.89 3.01 27.4 1.75 0.418 0.564 0.565 0.710 table continues next page Table A .2. continued Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Kenya Oloolaiser 27.07 9.12 40.33 0.97 0.409 0.528 0.424 0.577 Kenya Olkejuado 32.50 7.58 33.72 0.25 0.401 0.358 0.444 0.372 Kenya Kirinyaga 29.42 7.57 37.52 1.01 0.394 0.525 0.433 0.595 Kenya Matungulu 26.09 8.06 27.4 1.15 0.382 0.537 0.406 0.604 Kangundo Kenya Yatta 29.03 6.72 18.26 0.91 0.368 0.482 0.411 0.550 Kenya Sibo 23.17 7.4 26.32 1.0 0.363 0.498 0.390 0.562 Kenya NolTuresh Loitoktok 18.84 7.2 36.35 0.75 0.352 0.434 0.379 0.479 Kenya Mikutra 19.89 2.33 7.61 1.29 0.270 0.452 0.356 0.571 Kenya Kwale 17.01 1.41 28.8 0.99 0.261 0.420 0.362 0.540 Malawi CRWB 73.53 22.0 81.34 1.12 0.884 0.913 0.868 0.912 Malawi SRWB 77.82 22.0 73.97 1.36 0.786 0.840 0.721 0.814 Malawi NRWB 79.00 20.0 71.8 0.89 0.760 0.773 0.724 0.752 Malawi BWB 73.80 20.0 62.98 0.92 0.735 0.761 0.685 0.736 Mozambique Inhambane 95.43 24.0 82.85 0.88 1.000 0.951 1.000 0.935 Mozambique Tete 75.99 22.0 86.13 1.15 0.892 0.919 0.880 0.920 Mozambique Xai-Xai 107.54 24.0 67.13 0.83 0.881 0.851 0.822 0.801 Mozambique Chókwé 99.18 22.0 56.37 0.83 0.844 0.821 0.808 0.789 Mozambique Beira 71.04 24.0 65.61 1.2 0.783 0.838 0.675 0.783 Mozambique AdeM 63.13 17.0 81.6 0.93 0.780 0.797 0.816 0.827 Mozambique Maxixe 69.40 24.0 50.37 0.77 0.765 0.749 0.647 0.666 Mozambique Chimoio 53.27 24.0 65.56 1.1 0.724 0.793 0.586 0.724 Mozambique Pemba 72.78 10.0 111.97 1.22 0.715 0.786 0.864 0.909 Mozambique Quelimane 54.35 20.0 50.85 0.78 0.660 0.495 0.573 0.382 Mozambique Angoche 24.86 22.0 36.05 1.0 0.578 0.660 0.408 0.574 Mozambique Nacala 40.57 15.0 42.45 0.93 0.538 0.614 0.494 0.611 Mozambique Lichinga 21.79 17.0 63.54 0.86 0.520 0.587 0.426 0.546 Mozambique Nampula 50.11 9.0 43.16 1.06 0.487 0.605 0.543 0.682 103 table continues next page 104 Table A .2. continued Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Mozambique Cuamba 14.61 10.0 47.81 0.57 0.386 0.419 0.370 0.420 Nigeria KSWB, Kaduna State 91.62 15.0 68.89 0.6 0.758 0.705 0.825 0.732 Nigeria JSWB, Jigawa State 100.00 16.0 14.05 0.11 0.730 0.573 0.762 0.542 Nigeria FCT WB, Abuja 32.78 24.0 202.28 5.52 0.690 0.767 0.535 0.690 Nigeria KGWB, Kogi State 53.58 18.0 58.17 0.635 0.727 0.578 0.719 Nigeria SSWB, Sokoto State 65.00 6.0 85.97 0.66 0.633 0.624 0.825 0.749 Nigeria KBSWB, Kebbi State 32.76 12.0 147.07 1.25 0.609 0.707 0.664 0.776 Nigeria KNSWB, Kano State 41.23 18.0 40.41 0.71 0.580 0.596 0.495 0.545 Nigeria BYSWB, Bayelsa 99.36 5.0 9.86 0.0 0.575 0.431 0.758 0.506 State Nigeria LWC, Lagos 39.83 17.7 17.89 0.38 0.556 0.503 0.465 0.425 Nigeria ASWB, Adamawa 33.33 8.0 143.21 0.52 0.556 0.534 0.667 0.601 State Nigeria ZSWB, Zamfara State 75.00 6.0 74.18 0.0 0.555 0.416 0.707 0.471 Nigeria ENSWC, Enugu State 69.37 9.0 14.22 1.34 0.531 0.648 0.609 0.740 Nigeria WCOS , Oyo State 53.90 12.0 4.98 1.3 0.516 0.637 0.524 0.682 Nigeria CRSWBL, Cross River 55.37 24.0 27.18 2.48 0.701 0.776 0.552 0.701 State Nigeria YSWC, Yobe State 67.00 6.0 58.97 0.07 0.514 0.401 0.646 0.451 Nigeria TSWSA, Taraba State 36.09 12.0 49.4 0.04 0.486 0.373 0.480 0.331 Nigeria PSWB, Plateau State 15.82 7.0 142.84 1.47 0.483 0.612 0.579 0.719 Nigeria ESUWB, Edo State 64.29 7.0 0.78 0.11 0.479 0.383 0.572 0.414 Nigeria ANSWC, Anambra 49.25 9.0 25.51 0.46 0.471 0.459 0.519 0.486 State Nigeria BSWB, Bauchi State 55.06 5.0 63.39 0.63 0.464 0.492 0.592 0.586 Nigeria EBSWC, Ebonyi State 12.42 18.0 7.4 0.26 0.462 0.405 0.318 0.290 Nigeria GSWB, Gombe State 41.48 9.0 44.41 0.09 0.459 0.366 0.501 0.363 Nigeria ABSWB, Abia State 60.00 5.0 16.54 0.07 0.446 0.351 0.565 0.398 table continues next page Table A .2. continued Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Nigeria KTSWB, Katsina 40.00 9.0 30.57 1.09 0.444 0.582 0.478 0.650 State Nigeria OSWC, Osun State 37.00 9.82 22.04 0.23 0.440 0.382 0.455 0.373 Nigeria OGSWC, Ogun State 44.66 8.0 6.51 0.5 0.430 0.438 0.479 0.472 Nigeria NSWB, Nasarawa 30.55 6.0 72.65 0.15 0.405 0.339 0.483 0.368 Sate Nigeria AKWCL, Akwa Ibom 40.75 3.0 4.51 1.38 0.347 0.510 0.458 0.638 Nigeria DSUWB, Delta State 26.60 6.0 11.32 0.08 0.345 0.277 0.393 0.286 Nigeria RSWB, Rivers State 4.87 8.0 55.0 0.65 0.331 0.397 0.330 0.418 Nigeria BSWB, Benue State , 5.50 10.0 8.81 0.26 0.329 0.307 0.285 0.270 Makurdi. Nigeria ODWC, Ondo State 11.45 8.0 15.27 0.1 0.325 0.266 0.321 0.244 Nigeria NSWB, Niger State 12.64 5.0 30.0 0.0 0.297 0.223 0.341 0.227 Nigeria EKSWC, Ekiti State 20.00 4.0 7.25 0.04 0.293 0.228 0.356 0.249 Nigeria ISWC, Imo State 2.60 6.0 18.78 0.84 0.270 0.392 0.280 0.440 Zambia Southern WSC 89.24 21.0 82.25 1.26 0.958 0.969 1.000 1.000 Zambia Lusaka WSC 86.48 20.0 85.63 1.26 0.944 0.958 1.000 1.000 Zambia Kafubu WSC 86.71 17.0 91.64 1.45 0.903 0.927 1.000 1.000 Zambia Nkana WSC 94.87 15.0 78.49 1.44 0.875 0.906 1.000 1.000 Zambia North Western WSC 82.78 23.0 46.01 1.39 0.864 0.898 0.795 0.864 Zambia Lukanga WSC 69.23 20.0 122.93 0.99 0.842 0.856 0.846 0.863 Zambia Eastern WSC 114.25 20.0 66.79 0.83 0.826 0.807 0.822 0.798 Zambia Mulonga WSC 94.69 18.0 218.47 1.39 0.812 0.859 0.843 0.896 Zambia Chambeshi WSC 70.85 10.0 49.74 0.83 0.575 0.619 0.654 0.686 Zambia Luapula WSC 19.86 9.0 101.3 0.46 0.525 0.498 0.599 0.539 Zambia Western WSC 35.94 13.0 54.88 0.83 0.504 0.566 0.486 0.574 105 106 Table A .3. Customer Performance and Water Coverage, Municipal Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance South Africa Emfuleni 100.00 24.0 147.52 1.18 1.000 1.000 1.000 1.000 South Africa Mogale 100.00 24.0 142.42 0.85 1.000 0.943 1.000 0.924 South Africa Buffalo City 100.00 24.0 131.84 0.64 1.000 0.896 1.000 0.861 South Africa George 100.00 24.0 128.72 0.68 1.000 0.905 1.000 0.874 South Africa Nelson Mandela 100.00 24.0 140.05 1.35 1.000 1.000 1.000 1.000 Bay South Africa Rustenburg 100.00 24.0 145.19 0.98 1.000 0.974 1.000 0.965 South Africa Silulumanzi 100.00 24.0 153.33 0.15 0.993 0.778 0.989 0.704 South Africa EWS 100.00 24.0 157.68 0.66 0.984 0.888 0.976 0.850 South Africa The Msunduzi 100.00 24.0 159.39 0.84 0.980 0.925 0.971 0.900 South Africa Ekurhuleni 100.00 24.0 164.68 1.33 0.970 0.978 0.955 0.970 South Africa Drakenstein 100.00 24.0 182.04 1.39 0.941 0.956 0.912 0.941 South Africa Cape Town 100.00 24.0 184.4 0.7 0.938 0.862 0.907 0.816 South Africa Mangaung 100.00 24.0 196.13 0.92 0.922 0.900 0.882 0.867 South Africa Stellenbosch 100.00 24.0 196.46 1.16 0.921 0.941 0.882 0.921 South Africa Joburg Water 100.00 24.0 207.72 0.96 0.907 0.900 0.861 0.866 South Africa Sol Plaatje 100.00 24.0 208.34 1.41 0.907 0.930 0.860 0.907 South Africa Tshwane 100.00 24.0 246.11 1.2 0.870 0.902 0.805 0.870 South Africa Newcastle 100.00 24.0 0.0 0.18 0.833 0.666 0.750 0.555 South Africa Polokwane 100.00 24.0 0.0 0.61 0.833 0.763 0.750 0.685 table continues next page Table A .3. continued Full Definition Limited Definition Hours of Operation, Total Consumption Operating Cost (including hours of operation) (excluding hours of operation) Country Utility Coverage, % hours a day (lpcd) Coverage Sustainability Absolute Sustainability Absolute Performance Performance Performance Performance Tanzania Tanga 94.50 23.5 81.74 1.12 1.000 1.000 1.000 1.000 Tanzania Moshi 89.40 20.0 110.15 1.05 0.944 0.947 1.000 0.984 Tanzania Iringa 95.80 24.0 69.64 1.01 0.884 0.893 0.826 0.857 Tanzania Mbeya 96.90 21.0 64.14 1.01 0.837 0.856 0.818 0.850 Tanzania Dodoma 70.20 19.0 83.15 1.05 0.831 0.863 0.851 0.887 Tanzania Musoma 66.50 18.9 96.36 0.88 0.818 0.814 0.833 0.822 Tanzania Morogoro 72.00 17.4 88.44 0.47 0.815 0.717 0.860 0.715 Tanzania Mwanza 72.50 22.0 70.12 1.04 0.765 0.811 0.689 0.775 Tanzania Arusha 70.40 12.0 78.43 1.17 0.735 0.801 0.852 0.901 Tanzania Babati 84.20 14.0 26.19 1.0 0.710 0.759 0.774 0.818 Tanzania Bukoba 56.90 21.3 65.61 0.91 0.699 0.732 0.604 0.680 Tanzania Mtwara 51.60 13.6 78.1 0.79 0.694 0.700 0.758 0.744 Tanzania Tabora 71.00 18.2 49.77 0.92 0.689 0.727 0.655 0.716 Tanzania Songea 76.50 17.1 36.18 0.74 0.682 0.679 0.667 0.668 Tanzania Shinyanga 22.00 21.3 187.18 0.97 0.636 0.699 0.511 0.636 Tanzania Singida 81.20 5.7 38.71 0.84 0.604 0.644 0.787 0.780 Tanzania Kigoma 42.80 9.1 122.88 0.81 0.602 0.635 0.714 0.721 Tanzania Sumbawanga 61.60 10.6 20.84 0.73 0.532 0.564 0.577 0.605 Tanzania Bariati 33.61 16.0 15.87 0.33 0.510 0.457 0.432 0.387 Tanzania DAWASCO 62.28 8.0 38.28 0.81 0.510 0.566 0.598 0.643 Tanzania Lindi 70.00 6.0 14.11 0.32 0.492 0.442 0.612 0.506 Tanzania Mpanda 71.24 3.0 44.0 1.17 0.474 0.606 0.649 0.766 Tanzania Njombe 53.00 6.0 61.74 0.91 0.470 0.560 0.580 0.663 Tanzania Geita 6.59 4.0 19.68 0.53 0.256 0.312 0.300 0.361 107 Appendix B Case Studies Burkina Faso, ONEA (L’Office National de l’Eau et de l’Assainissement) Background ONEA is a national utility that provides water supply, wastewater, and excreta treatment services to 54 towns in Burkina Faso (Sawadogo 2015). Ouagadougou, the capital, accounts for more than 60 percent of customers. ONEA was established as a public water utility in 1985. Access was low and reliability was poor throughout the early 1990s. In 1994, ONEA was corporatized. In the same year, the first contract plan (performance contract with operational targets) was established between the utility and the government of Burkina Faso. A performance-based service contract with Veolia was in place from 2001 to 2006. Today, ONEA is publicly owned and operated. Its contract plan with the government is supervised by a multi-stakeholder committee comprising representatives of customers, nongovernmental organizations, and donors. The committee monitors performance of both the utility and the government under the contract, on the basis of independently audited financial and technical reports. The committee’s monitoring role is centered on an annual meeting. Before the meeting, committee members receive not just a report from ONEA on its performance against the contract, but also the report of a financial auditor and a technical auditor whose job it is to assure the quality of the information. The auditors’ reports indicate the degree of confidence they have in the information presented. The auditors appear in person before the committee and explain their reports. Performance Customer Performance Access to Water Services. In 2000, ONEA served about half of its service area population—2.3 million people. By 2014, nearly 4 million people were served out of 4.6 million people (86 percent coverage). This progress is extraordinary considering that from 2000 to 2015 Ouagadougou was the fastest growing Sub-Saharan African city (out of the set of cities with a 2015 population above 1 million). Ouagadougou grew at an average annual rate of 7.5 per- cent during this period (United Nations 2014). Growth in the service area population and population served are shown in figure B.1. Figure B.2 shows the growth in water coverage, defined as the proportion of the service area population served by ONEA. A breakdown of those served by direct or shared taps and those served by public water points is included for years 2006–14. Performance of Water Utilities in Africa 109 Figure B.1. Population Served Compared with Service Area The decrease in population served from 2008 to 2009 Population, 2000–14 is due to revised estimation techniques. As shown in figure B.2, ONEA reports that the percentage of people 7 served by public water points has remained relatively 6 constant since 2006, at 20 percent. Meanwhile, the share of those served by direct or shared taps has 5 increased from 41 percent to 65 percent. People, millions 4 However, household survey data for Ouagadougou present a different picture. In 2010, only about 48 per- 3 cent of people reported accessing water from a tap piped to their premises.1 Forty-six percent reported accessing 2 water from a public tap. In other words, the DHS data (for 1 Ouagadougou only) show lower access to piped water to their premises, higher access to public taps, and higher 0 total access in  comparison with the utility-reported 10 00 02 03 04 08 09 12 13 14 01 05 06 11 07 20 20 20 20 ­ eople data. The DHS data do confirm that the share of p 20 20 20 20 20 20 20 20 20 20 20 Population in service area Population served with access to a household connection has increased—in Source: IBNET, www.ib-net.org. 2003, just 37 percent reported accessing piped water sup- plied to their premises. DHS data for 1993, 1995,  2003, Figure B.2. Water Coverage (Population Served Divided by and 2010 are shown in ­ figure B.3. The number of people Service Area Population), 2001–14 served per connection is shown in figure B.4. ONEA’s strategy is to first focus on achieving good access 100 for the poor through public taps and then increase individ- ual connections. The above data show that the utility is 80 executing this strategy successfully. ONEA plans to serve 80 percent of its service area population with piped Coverage (%) 60 water  to the premises by 2030. For the time being, how- ever, the average number of ­ connection people  per  ­ 40 remains high. There were 13 people per connection in 2014, down from 19 people per connection in 2000. 20 0 Network Expansion and Growth in Connections. A Bank- funded expansion project, implemented in 2004, led to 10 02 03 08 09 12 13 04 14 11 01 06 05 07 20 20 20 20 20 20 20 20 20 20 20 20 20 20 rapid growth in connections and the distribution network Water coverage, direct and shared taps Water coverage, public water points (figures B.5 and B.6). From 2004 to 2014, the annual aver- Water coverage age rate of connections was 15 percent, up from an average Source: IBNET, www.ib-net.org. increase of 7 percent in the previous decade. Reliability. In the 1990s, Burkina Faso faced a severe water shortage. Data from this period are not available, but it is known that ONEA’s water supply was intermittent. After construction 110 Performance of Water Utilities in Africa Figure B.3. Access to Water Service, Ouagadougou Only of the Ziga Dam near Ouagadougou, ONEA has achieved figure B.7). near-perfect reliability at 23 hours per day (see ­ 100 This is significant, especially considering that the country 90 has just 732 m3 of renewable internal freshwater resources Access, Ouagadougou only (%) 80 per capita. The Ziga II Project aims to attain 24  hours of 70 service by 2017 (Sawadogo 2015). 60 50 residential water consump- Water Consumption. Average ­ 40 tion per capita is relatively low at 39 lcd. Residents con- 30 sume about 87 percent of the volume of water sold. 20 10 0 Affordability. One measure of affordability is the percentage 1993 1995 2003 2010 of income spent on water consumption. A proxy for the Public tap Piped to premise average annual per capita expenditure on water is total rev- Source: IBNET, www.ib-net.org. enue from sales divided by the number of people served. A proxy for average per capita income is Burkina Faso’s GNI Figure B.4. People Served per Connection, 2001–14 per capita. Dividing the proxy for water expenditure into the proxy for income gives an average expenditure on water 25 at 2.7 percent of income.2 By contrast, the regional bench- mark for this indicator is 1.22 percent. The benchmark is the 20 first quartile of all African water utilities included in the People per conneciton sample; in other words, it is the middle number between 15 the most affordable water (lowest ratio) and the median. Affordability is reduced by the low consumption levels of 10 residential water users (at about 39 lcd; see figure B.8). 5 Safety. ONEA tests water quality and reports that the sam- 0 ples consistently pass a defined standard for drinking water. However, drinking water quality data are neither published 09 10 03 08 02 04 12 13 14 01 11 06 05 07 20 20 20 20 20 20 20 20 20 20 20 20 20 20 nor independently verified. Source: IBNET, www.ib-net.org. Operational Performance NRW. ONEA has maintained low levels of NRW since 1995, the first year with data available. Measured in percentage terms, NRW has been about 20 percent for the last 20 years. NRW per connection per day was more than halved during this period, dropping to just 135 liters per connection per day. This is close to the global benchmark of 121 liters per connection per day and better than the African benchmark (at 205 liters per connection per day). NRW trends over time, measured in liters per connection per day and percentage of production, are shown in figure B.9 and  figure B.10, respectively. NRW spikes in 2004 correspond to Performance of Water Utilities in Africa 111 Figure B.5. Number of Water Connections when the Ziga scheme began ­supplying water. 350 NRW in percentage terms is relatively flat because produc- ­ 300 tion and consumption increase at about the same average 250 annual rate (6 percent). NRW in liters per  connection per day Water connections 200 reduces because of a significant increase in connections, even as 150 the total amount of NRW rises. Metering across the service area 100 has been nearly universal (about 97 percent) since 2000. 50 Staff Productivity. Staff productiv- 0 ity has improved significantly since 2000, with the number of 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 20 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 staff per 1,000 water connections Source: IBNET, www.ib-net.org. falling from 10 to 3 by 2014 (see figure B.11). This progress was Figure B.6. Network Expansion achieved as both staff numbers 9,000 and productivity grew. This was possible because network expan- 8,000 sion significantly outpaced hiring 7,000 of new labor. ONEA’s staff efficiency—­ revenue 6,000 per employee divided by labor 5,000 costs per employee, or simply rev- KM pipe enue divided by labor costs—has 4,000 fallen from about 5 in 2003 and 3,000 2004 to 4 from 2009 to 2013. At 3.94 in 2013, ONEA is below the 2,000 regional benchmark of 4.21.3 1,000 0 Financial Performance Collection Efficiency. ONEA had a 95 96 7 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 9 20 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 performance-based service con- Source: IBNET, www.ib-net.org. tract for commercial management 112 Performance of Water Utilities in Africa Figure B.7. Hours of Supply per Day, 2006–14 with Veolia from 2001 to 2006 (Marin, Fall, and Ouibiga 2010). Veolia provided two deputy managers, in addition to other 24 short-term advisers, for ONEA’s commercial and finance departments. They set up new accounting and customer 20 management systems and helped ONEA identify illegal cus- 16 tomers, improve meter reading and meter repairs, and Hours per day improve customer service. After an initial decline from 85 12 percent to 78 percent, collection efficiency rose to 95 percent 8 by the end of the contract (see figure B.12). ONEA has maintained high cash collection rates above 4 95 percent since the end of the contract. This includes col- lections from the government, which is obliged to settle its 0 water bills in the terms of the performance contract. ONEA’s 09 10 08 12 13 14 06 11 07 20 20 20 collection ratio from 2001 to 2013 is shown in figure B.12. 20 20 20 20 20 20 Source: IBNET, www.ib-net.org. Operating Cost Coverage. ONEA’s OCCR has fluctuated over Figure B.8. Sufficiency of Residential Consumption, 2001–14 time, reaching a high of 1.63 in 2007. This high is attributed 60 to an increase in revenue after completion of the Ziga dam project and  related network expansion efforts. In 2012 50 ­owest and 2013, the operating cost ratio dropped to its l levels since 2002: 1.19 and 1.13, respectively. Operating Litres per person per day 40 expenses nearly doubled from 2008 to 2013, with the share of non-labor expenses also rising. The trend in the 30 operating cost recovery ratio is shown in figure B.13. Real average tariffs and real average costs per m3, in the local 20 currency, are shown in figure B.14. The U.S. dollar equiv- alents for 2013 are also noted in the figure. 10 In real terms, ONEA’s average tariff has declined at an average annual rate of 1.7 percent since 2002. In 2013, it 0 was equivalent to US$1.12. Real average costs have fluctu- 01 02 03 04 05 06 07 08 09 10 11 12 13 14 20 20 20 20 20 20 20 ated significantly from 2002 to 2013. They were US$0.99 20 20 20 20 20 20 20 Source: IBNET, www.ib-net.org. in 2013. ONEA subsidizes consumption for basic needs by charging much higher tariffs for what it regards to be excessive consumption. ONEA’s tariff for the first consumption block (up to 8 m3 per month) is just 18 percent of the tariff for consumption in excess of 30 m3. This subsidized tariff is US$0.39 per m3, compared with US$2.16 per m3 for consumption above 30 m3. The latter is designed to be a disincentive for excessive water use.4 There are two other resi- dential tariff blocks—above 8 m3 and up to 15 m3 (US$0.89) and above 15 m3 and up to 30 m3 (US$1.06). The  standpipe tariff is equal to the basic needs tariff (US$0.39 per m3).5 subsidize are set at US$2.16 per m3 regardless of Nonresidential tariffs to help cross-​ Performance of Water Utilities in Africa 113 Figure B.9. Nonrevenue Water, by Connection, 1995–2014 amount consumed. Industrial customers account for 5 percent 350 of consumption by volume.6 300 Financial Transparency. ONEA does not publish an annual report or 250 financial statements. The 2015 Liters per connection per day contract (with tar- performance ­ gets) is published on its website, 200 but actuals are not included. The tariff ­ schedule is posted online. 150 Investment. For ONEA to increase 100 access and improve service, invest- ment totaling about US$600 mil- lion was required from 2002 to 50 2013.7 This amounts to about US$23  per person served per 0 year  over this 12-year period. 95 96 98 99 00 01 02 03 97 04 05 06 07 08 13 14 09 10 11 12 Figure B.15 shows the sources and 20 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 amounts of investment financing Source: IBNET, www.ib-net.org. during this period. While about Figure B.10. Nonrevenue Water as a Percentage of Production 52  percent was grant  financed, 19  percent was financed by own 60 percent from  loans. cash and 29  ­ 50 ONEA can service its  debt with operating cash flows. 40 A major program was the NRW (%) 30 Ouagadougou Water Supply 20 Project (US$269 million, 2001–07). The Bank was a major financier 10 along with 10 other donors. The 0 Bank lent US$70 million to the government of Burkina Faso. The 5 96 7 8 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 9 9 9 20 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 amount on-lent to ONEA was Source: IBNET, www.ib-net.org. ­ illion and the remaining US$28 m US$42 million was given as an equity contribution. The interest rate was 5.4 percent and the tenor was 20 years (including a 10-year grace period). The program included the extension of the distribution network and construction of the Ziga dam, Boudtenga res- ervoir (5,400 m3), a water treatment plant, and pumping station. Increased water 114 Performance of Water Utilities in Africa Figure B.11. Staff Numbers and Staff per 1,000 Water and perfect water supply reliability production led to near-​ Sewer Connections, 2000–14 (23  hours per day), whereas before the project service was intermittent. Another component of the project was 1,200 12 the hiring of Veolia under a performance-based service Sta per 1,000 connections 10 962 1,000 10 contract to help manage the commercial function. 800 8 Increasing operating cash flow was key to achieving Number of sta 575 600 6 expansion in service. As the cash flows allowed ONEA 400 3 4 to service the debt and invest directly, 48 percent of 200 2 investment was thus supported by the free cash flow ONEA created through its operations. Important driv- 0 0 ers of increasing cash flow were increasing collection 20 0 20 0 20 8 20 9 02 20 03 20 2 20 3 14 20 1 20 11 20 4 20 6 20 05 20 7 0 1 1 0 0 0 1 0 0 0 20 20 efficiency (78 percent in 2002 to 97 percent in 2013), Total staff Staff per 1,000 connections maintaining low levels of NRW, and increasing labor Source: IBNET, www.ib-net.org. productivity by limiting the growth of staff numbers as connections increased. Figure B.12. Collection Ratio, 2001–13 95% 97% Organization 100 85% 90 78% Human Resources 80 Collection ratio (%) ONEA has instituted a performance management system 70 60 with clearly documented processes, job descriptions, and Veolia contract 50 (2001–06) targets for each position. Annual performance reviews let 40 30 team members know where they have to improve. Targets 20 from the utility’s contract with the government are 10 0 reflected in the board’s contract with the managing direc- tor. The managing director then cascades these targets 01 02 03 04 05 06 07 08 09 10 11 12 13 20 20 20 20 20 20 20 20 20 20 20 20 20 down to lower-level managers in the organization. This Source: IBNET, www.ib-net.org. method of performance-based management helped to build strong managers throughout the company. By cre- Figure B.13. Operating Cost Recovery Ratio, 2002–14 ating a deep management bench, ONEA has reduced the 1.8 risk that losing a leader will undo its success. 1.6 Operating cost recovery 1.4 Strategic Planning 1.2 ONEA uses strategic planning to develop the contract plans, 1.0 0.8 which are at the heart of its accountability arrangements. 0.6 The plans set out what the utility needs to do to meet 0.4 service targets and what this will cost. The plans are inde- 0.2 0 pendently scrutinized by external stakeholders. When these stakeholders are satisfied that the plans are reason- 10 09 03 08 12 13 14 02 04 11 05 06 07 20 20 20 20 20 20 20 20 20 20 20 20 20 able, the service targets in the plan and the tariffs required Source: IBNET, www.ib-net.org. to cover the costs of services are incorporated in a multiyear Performance of Water Utilities in Africa 115 Figure B.14. Average Tariffs, Average Costs per m3, 2002–14 contract plan. ONEA started strategic planning in 2004. It is now on its third plan. 800 ONEA is ISO 9001 certified. ONEA reports that certifica- CFAF (2013) per m3 of water sold 700 US$1.12 tion provides internal discipline in  the company and a 600 credible external validation of reliability. In addition, 500 ONEA’s annual reports against the key performance indica- 400 US$0.99 tors in its contract with the government are audited by 300 financial and technical auditors, who submit their reports 200 to a multi-stakeholder supervisory committee. 100 0 Commercial Practices ONEA’s mandate prevents it from providing services 10 03 08 09 13 12 02 04 14 11 06 05 07 20 20 20 20 20 20 20 20 20 20 20 20 20 Real average tari Real average costs directly to informal settlements because they are unplanned, and so the residents lack formal title. The utility Source: IBNET, www.ib-net.org. has delegated service provision to small entrepreneurial providers. These providers typi- Figure B.15. Sources and Amounts of Investment Financing, 2002–13 cally start business by operating a water kiosk. However, they can 40 35 then expand their delivery 30 options, with some offering flexi- CFAF, billions 25 ble piped connections to the 20 home, run from the water point 15 controlled by the small entrepre- 10 5 neurial provider. ONEA controls 0 the prices that the small provid- ers charge, helping reduce the 08 09 10 02 03 13 04 12 06 11 05 07 20 20 20 20 20 20 20 20 20 20 20 20 From grants From loans From own cash risk of small providers using local monopoly power to on-sell water Source: IBNET, www.ib-net.org. at excessive prices. Even so, ONEA does not regard this as a long-term solution. It would like to progressively extend formally piped water connections to more and more households in the informal settlements. Summary As a result of ONEA’s turnaround, about 78 percent of people in the utility service area in 2013 have access to piped water services, up from just 50 percent in 2001. On average, water is available for 23 hours per day. Cross-subsidies help ensure that water is affordable to resi- dential customers. Revenue covers all operating costs and some capital costs, a result of operational efficiency and a cost recovery tariff. In the 1990s, Burkina Faso faced a severe water shortage. A new water source was des- perately needed. The success of Senegal’s affermage contract made that an obvious choice 116 Performance of Water Utilities in Africa for Burkina Faso. However, the government was committed to public sector control of the utility. An alternative model was developed involving strong multi-stakeholder account- ability arrangements and a performance-based contract with a specialist firm to boost ONEA’s commercial performance. Financing partners found this proposal credible, and the bulk water scheme and institutional reforms to ONEA proceeded in parallel. Veolia’s contract from 2001 to 2006 increased collection efficiency from 85 percent to 95 ­ percent, success ONEA has maintained in the decade after the contract’s end. At the same time, the Ouagadougou Water Supply Project ensured greater supply reliability for ONEA’s customers, and included network expansion and the installation of  new con- nections and standposts. Low levels of NRW were maintained. Staff productivity rose while new staff were hired. Today, ONEA’s innovative Supervisory Committee, strong performance culture, and operational efficiency are sustaining the benefits achieved by these reforms. ­ Côte d’Ivoire, SODECI (la Société de Distribution d’Eau de la Côte d’Ivoire) Background SODECI, founded in 1960, is the oldest PPP water utility in the developing world, serving over 11 million customers. The partnership between the national government and the private operator has allowed effective expansion of access to water services, while increasing the efficiency and profitability of the national utility over time. This PPP has endured even as utilities in neighboring countries were nationalized. It has remained private since its founding more than 50 years ago. More notably, SODECI has shown remarkable resilience and continuity of service during and after the First Ivorian Civil War from 2002 until 2007 and the Second Ivorian Civil War from late 2010 until early 2011. The history of SODECI is summarized in five time periods below (Marin et al. 2009). 1960–74. Côte d’Ivoire gained independence in 1960. The year before that, the French water operator SAUR had been awarded a concession contract to provide water supply services in Abidjan. SODECI, the new Ivorian company established in 1960, gradually signed operating contracts with municipalities in 10 other cities. 1974–88. In 1974, Côte d’Ivoire’s water sector was reformed. Municipal water utilities were consolidated into one national utility, which signed an operations contract with SODECI. From 1974 to 1988, US$400 million of public investment were used to install piped water in 200 additional towns. Before these reforms, only 34 percent of households had access to piped water. By 1983, household coverage almost doubled to 63 percent. The ­company’s management was gradually transferred to Ivorian nationals. In 1978, the French group SAUR sold 48 percent of the shares in an initial public offering on the Abidjan stock exchange. Performance of Water Utilities in Africa 117 1988–2001. From 1988 to 2001, SODECI financed rapid expansion of the water network through a tariff surcharge. SODECI more than doubled the number of households served by piped water. During this period, major gains in efficiency were achieved (losses decreased from 180 liters per connection per day to 130 liters per connection per day). At  the end of this period, SODECI’s operational performance was at the level of ­ present-day global benchmarks. The expansion of the network added many more cus- tomers. Increased revenues as a result of expansion, as well as efficiency gains, gradually improved the overall financial performance of the utility. Average tariffs decreased in real terms over the same time period. 2002–11. During the First Ivorian Civil War (2002–07) and Second Ivorian Civil War (2010–11), SODECI continued to provide water services to existing and new customers. ­ Nearly  4  million additional people were served during this period.8 As would be expected, however, service and operational performance declined during this time. service was nearly halved. NRW rose. Revenue fell slightly. Reliability of ­ 2012–14. After 2011, when the conflict stabilized, performance began improving. While pre-conflict performance has not been attained since then, SODECI is still a successful utility in the context of Sub-Saharan Africa. Most indicator values for 2014 are on par with or better than the regional benchmark, a significant accomplishment given nearly a decade of politi- cal instability and factual division of the country. The Eranove Group, formerly known as Finagestion, is a main shareholder of SODECI. Performance Customer Performance Access to Water Services. Between 2000 and 2014, SODECI extended water service to million additional people. This was achieved despite the civil wars from 2002 to 2007 5.5 ­ and 2010 to 2011. Owing to rapid migration to major cities and towns during the crises, water coverage declined in the early 2000s. Since 2011, coverage has been steadily rising as SODECI regains its footing. Growth in SODECI’s service area population and population served over this period is shown in figure B.16. Figure B.17 shows the growth in water coverage, defined as the propor- tion of the service area population served by SODECI. As of 2006, about 25 percent of people in the served area population were served by commu- nity tap stands (Fall et al. 2009). In 2014, the average number of people served per connection connection during the 2000–04 period (see figure B.18). was 15, up from 12 people per ­ Sufficiency. Average residential water consumption per capita has  declined over time, from about 40 lcd in the early 2000s to 32 lcd in 2014 (see figure B.19). Historically, per capita con- sumption has been higher in Abidjan than in secondary centers. 118 Performance of Water Utilities in Africa Figure B.16. Population Served Compared with Service Area Population, 2000–14 Reliability. Reliability of water supply was perfect (24 hours per 20 day) from 2000 to 2004. By 2009, 18 between the two conflicts, it had 16 dropped to just 13 hours per day. Since then, SODECI has improved 14 Number of people, millions substantially on this indicator, 12 with water available 20 hours per 10 day on average, as of 2014 (see 8 figure B. 20). ­ 6 Affordability. One measure of 4 affordability is the percentage of 2 income spent on water consump- 0 tion. A proxy for the average annual per capita expenditure on water is 00 01 02 03 04 05 06 07 09 08 10 11 12 13 14 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 total revenue from sales divided Population in service area Population served by  the number of people served. Source: IBNET, www.ib-net.org. A  proxy for average per capita income is Côte d’Ivoire’s GNI per Figure B.17. Water Coverage (Population Served Divided by Service Area Population) capita. Dividing the proxy for water 100 expenditure into the proxy for income gives an average expendi- 90 ture on water at  0.96 percent of 80 ­ gure B.21). This is income (see fi better than the regional benchmark 70 for this indicator, 1.22 percent, Water coverage (%) 60 indicating that SODECI’s service is ­ more affordable than over 75 per- 50 cent of African water utilities. 40 Quality. Results of water quality 30 tests are published in SODECI’s 20 Annual Reports. About 38,000 tests were conducted in 2012. 10 The passing rate for residual 0 chlorine tests fell to 91 percent after the First Ivorian Civil War, 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 then recovered to pre-conflict Source: IBNET, www.ib-net.org. levels of 98  percent in 2012 Performance of Water Utilities in Africa 119 Figure B.18. People Served per Connection, 2000–14 (SODECI 2012). From 2007 to 2012, 99 percent of microbio- logical tests passed in Abidjan (the capital), on average. 18 Passing rates outside of  Abidjan were lower, at times as 16 low as 88 percent (SODECI 2012). A description of passing 14 requirements is not included in these reports. People per connection 12 Operational Performance 10 NRW. SODECI was very successful in reducing NRW in the 8 late 1990s, reaching a low of 131 liters per connection per day (or 17 percent) in 2000. However, by 2011, NRW levels 6 had risen to 204 liters per connection per day or 28 per- 4 cent. By regional standards, these levels are still quite 2 good, slightly better than the African benchmark of 205 0 liters per connection per day. In percentage terms, since 2011, SODECI has made rapid progress in reducing NRW, 00 10 02 03 04 09 12 13 14 11 01 20 20 20 20 20 20 20 20 20 20 20 although it still  has not been brought down to pre-­ Source: IBNET, www.ib-net.org. (figure B.22). As of 2014, it was 174 liters conflict levels ­ per connection per day or 24 percent (see figure B.23). Figure B.19. Sufficiency of Consumption, 2000–14 Metering has been virtually universal since 2000. 45 Staff Productivity. Staff productivity is excellent and has 40 remained stable since 2000, at less than three staff per 35 1,000  water connections. This was achieved even as 900 Litres per person per day 30 staff were employed. Good performance on this indicator is 25 attributed to SODECI’s long history of being privately oper- ated since 1960. 20 Figure B.24 shows staff per 1,000 water connections 15 only. The productivity ratio for water and sewer connec- 10 tions is even lower—at just 2.2 in 2000. Data on the num- 5 ber of sewer connections were not available after 2004. 0 This indicator was not graphed. SODECI’s staff efficiency (revenue per employee divided 14 01 02 03 04 09 10 11 12 13 00 20 20 20 20 20 20 20 20 20 20 20 by labor costs per employee, or simply revenue divided by Source: IBNET, www.ib-net.org. labor costs) has ranged from 4.9 to 5.55 from 2000 to 2004 and averaged 5.4 between 2009 and 2014. At 5.84 in 2014, SODECI was above the Africa bench- mark of 4.27. Financial Performance Operating Cost Coverage. SODECI’s OCCR has remained relatively stable since 2000, just slightly over 1.0 in most years and below 1.0 in two years (figure B.25). It  seems that SODECI earns just enough revenue to cover operating expenses, but nothing more. This 120 Performance of Water Utilities in Africa Figure B.20. Hours of Water Supply per Day, 2000–14 is surprising considering SODECI’s excellent perfor- mance in the early 2000s and postconflict improvements 24 since 2011. A relatively low OCCR close to 1.0 implies that SODECI would have difficulty financing network expan- 20 sion, servicing debt for major capital projects, and pay- Hours of service per day 16 ing dividends to its shareholders. Real average tariffs rose at an average annual rate of 10.3 12 percent from 2011 to 2014, while operating expenses per m3 8 rose at an average annual rate of 7.5 percent. At the 2014 U.S. dollar value, the average tariff and average cost per m3 4 were US$0.99 and US$0.94, respectively (see figure B.26). 0 Collection Efficiency. From 2002 to 2004, SODECI’s collection 00 01 02 03 04 09 10 11 12 13 14 20 20 20 20 20 20 20 20 20 20 20 ratio was 95 percent. This dropped to below 90 percent Source: IBNET, www.ib-net.org. from 2009 to 2011—and the latter values represent the col- Figure B.21. Affordability, 2000–14 ­ rivate customers and large consumers lections rate from p only (see figure B.27).9 3 It is likely that the conflict prevented SODECI from collect- Percentage of income spent on water ing bills from all customers, and that customers were less willing to pay  their bills due to declining service levels. 2 SODECI’s collection period has been over one year since 2010, reaching a high of  533 days in 2011.10 As of 2014, SODECI’s receivables have decreased, indicating that cus- 1 tomers are paying their outstanding bills. For now, however, the collection period remains high at 457 days. Financial Transparency. SODECI publishes annual reports 0 with audited financial statements on its website. As of 00 01 02 03 04 09 10 11 12 13 14 20 20 20 20 March 2016, reports for the years 2010–14 are available. 20 20 20 20 20 20 20 Source: IBNET, www.ib-net.org. Organization Human Resources. SODECI began as the subsidiary of its French mother company, SAUR, in gradually transferred to Ivorian nationals. Today, the com- 1960. Early on, management was ­ pany’s management is almost universally Ivorian. SODECI strives for corporate social responsibility. The utility has set up several funds for its employees—such as an AIDS fund offering free treatments, an employee shareholder scheme, a supplementary pension scheme, and a mutual financing fund offering financial services. In partnership with its mother company, the Eranove Group, SODECI operates a training cen- ter, the Centre des Métiers de l’Eau (Water Training Center) which covers production, transport, distribution, and commercialization.11 Performance of Water Utilities in Africa 121 Figure B.22. Nonrevenue Water as a Percentage of Commercial Techniques. In addition to traditional payment Production, 2000–14 methods—cash, direct debit, and check—SODECI has intro- ­ duced new electronic payment mechanisms. SODECI part- 30 several mobile money providers and offers bill ners with ­ 25 payment at select automated teller machines. 20 Summary SODECI remains a relatively well-performing African water NRW (%) 15 utility today, despite two civil wars that split the country 10 from 2002 to 2007 and 2010 to 2011. Founded as a PPP in 1960, the utility achieved remarkable efficiency in the 5 1990s. In 2000—with just three staff per 1,000 water con- nections, NRW of 17 percent, and service 24 hours a day— 0 SODECI’s performance was virtually unmatched in the 00 01 02 03 04 09 10 11 12 13 14 region. 20 20 20 20 20 20 20 20 20 20 20 National conflict took a toll on this success, but the com- Source: IBNET, www.ib-net.org. pany was resilient. NRW worsened, reaching a high of 28 Figure B.23. Nonrevenue Water by Connection, 2000–14 percent in 2011. Collection efficiency fell to 86  percent. Although performance was worse than in the earlier decade, 250 SODECI still compared favorably to its peers. Access, in absolute terms, actually rose from 2002 to 2011. 200 Since 2011, when the conflict ended, service has improved Litres per connection per day again. Reliability is up from a low of 13 hours to 20 hours per 150 day, on average. Water coverage, as a percentage, is now increasing, from 65 percent in 2011 to 69 percent in 2014. 100 Kenya, NCWSC (Nairobi City Water and 50 Sewerage Company) Background 0 The NCWSC is Kenya’s largest water service provider (WSP) 0 01 02 03 04 09 10 11 12 13 14 by  service area population, with responsibility for the 0 20 20 20 20 20 20 20 20 20 20 20 3.8 million residents of Nairobi County (NCWSC 2014). The Source: IBNET, www.ib-net.org. NCWSC was incorporated in 2003 as a wholly owned subsid- iary of the Nairobi City County. The service area is divided into six principal administrative regions (Northern, Eastern, North Eastern, Central, Southern, and Western), which are further subdivided into 25 zones. The county borders largely align with the borders of Nairobi City. The NCWSC is a licensee of the Athi Water Services Board (WSB). The WSBs are licensed by the government of Kenya, under the Water Act 2002, to be responsible for water resources management in a particular area.12 The WSPs (such as the NCWSC) are licensed in turn by the WSBs to be responsible for water and sewerage services provision in a particular area. Athi 122 Performance of Water Utilities in Africa Figure B.24. Staff Numbers and Staff per 1,000 Water and Sewer Connections WSB’s jurisdiction covers a popu- lation of 5.5 million people and 12 2,500 4 2,317 WSPs, including the NCWSC.13 The constitution of 2010 devolved Sta per 1,000 connections 2,000 3 3 water and sewerage services pro- 3 vision to the counties. A national Number of sta 1,500 1,406 2 Water Bill is currently being con- 1,000 sidered in parliament to opera- tionalize this devolution of 1 500 responsibilities. Once the Water Bill is enacted, it will be the county 0 0 10 of Nairobi that delegates water 00 09 03 08 14 13 12 02 04 01 11 06 05 07 service responsibilities to the 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Total sta Sta per 1,000 connections NCWSC. The Water Act 2002 also estab- Source: IBNET, www.ib-net.org. lishes the Water Services Figure B.25. Operating Cost Recovery Regulatory Board (WASREB), which is the national regulatory 1.2 body for Kenya’s water services sector. WASREB oversees the 1.0 implementation of policies and Operating cost coverage ratio strategies relating to the provi- 0.8 sion of water and sanitation ser- vices and monitors and regularly 0.6 reports on the performance of the WSBs and WSPs (WASREB 2015). 0.4 Performance 0.2 Customer Performance Access to Water Services. Coverage 0 in the NCWSC service area has 10 00 09 08 02 04 12 13 03 14 11 01 06 05 07 increased from 66 percent (2008) 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 to 80 percent (2014), while the Source: IBNET, www.ib-net.org. service area population grew from 3.1 million to 3.7 million. The NCWSC forecasts that the population in its service area million by 2019 (NCWSC 2014). Figure B.28 shows growth in the NCWSC’s will grow to 4.5 ­ service area population and the population served over this period. Figure B.29 shows the NCWSC’s growth in water coverage; that is, the proportion of the NCWSC’s service area pop- ulation being served by the NCWSC. The NCWSC is also providing sewerage services. Sewerage coverage in 2013 was 28 percent. Performance of Water Utilities in Africa 123 Figure B.26. Average Tariffs, Average Costs per m3 According to the NCWSC, about 50 percent of Nairobi’s residents 600 have access to water piped to their US$0.99 premises. The rest obtain water 500 from kiosks, vendors, and illegal CFAF (2014) per m3 of water sold connections. These estimates US$0.94 400 match household survey data for Nairobi City. In the 2012–13 Kenya 300 Baseline “State of the City” sur- vey, about 55 percent of house- 200 holds reported access to piped water to their premises, while 23 100 percent reported access to a water kiosk. 0 A major challenge in increasing water coverage has been the lim- 00 02 03 04 08 09 10 12 13 14 06 11 01 05 07 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 ited water distribution network Real average tari Real average costs, per m3 within Nairobi’s densely popu- Source: IBNET, www.ib-net.org. lated informal settlements. To meet this challenge, in 2008, the NCWSC formed an admin- Figure B.27. Collection Efficiency, 2000–11 istrative unit solely responsible for improving access to 100 services in these settlements. The unit was later upgraded 90 to a region to give it greater status with respect to invest- 80 ment and operations. This pro-poor initiative has contrib- Collection cost ratio (%) 70 uted to the marked increase in the NCWSC’s water coverage 60 between 2008 and 2014 (see figures B.28 and B.29).14 The 50 NCWSC also provides sewerage services. In 2014, 46 per- 40 cent of the service area population was covered, up from 28 30 percent sewerage coverage in 2010. 20 10 0 Reliability. Reliability of water supply increased from 16 hours per day in 2011 to 18 hours per day in 2014. This brings the 00 09 10 03 08 02 04 01 11 20 20 20 20 20 20 20 20 20 NCWSC’s reliability toward the higher end of WASREB’s Source: IBNET, www.ib-net.org. “acceptable” rating, and in line with the national average for Kenya (WASREB 2015). It falls short of the benchmark for Africa, however, which is 21 hours. According to the NCWSC, about 40 percent of its customers  currently receive water for 24 hours per day.15 The utility aims to attain reliability of 24 hours per day for all customers by 2018 (NCWSC 2014). Figure B.30 shows the average daily duration of the NCWSC’s water supply in its service area between 2011 and 2014. 124 Performance of Water Utilities in Africa Figure B.28. Population Served Compared with Service Sufficiency. Average residential consumption is good, at Area Population, 2008–14 about 70 liters per person served per day (see figure B.31). This is more than double the level reported at well-­ 7 average performing utilities in Sub-Saharan Africa. SODECI’s ­ 6 residential consumption is 32 liters per person per day whereas the NWSC’s is just 23 liters per person per day. If the 5 NCWSC were able to reduce NRW from current levels of 700 liters per connection per day (see figure B.32), this high level People, millions 4 of sufficiency could be maintained without significant 3 increases in water supply. The cause of the large jump in reported volume sold to residential customers between 2010 2 million m3 2011—is and 2011—46.6 million m3 in 2010 to 73.2 ­ unknown. 1 Affordability. One measure of affordability is the percentage 0 of income spent on water consumption. A proxy for the 09 10 13 08 12 14 11 20 20 20 20 20 20 20 average annual per capita expenditure on water within the Population in service area Population served NCWSC’s service area is total revenue from water sales Source: IBNET, www.ib-net.org. divided by the number of people served. A proxy for aver- age per capita income is Kenya’s Figure B.29. Water Coverage (Population Served Divided by Service Area Population) GNI per capita (US$1,280 in 2014). Dividing the proxy for water 100 expenditure into the proxy for 90 income gives average expenditure 80 on water at 2.1 percent of income 70 in 2014. The regional benchmark 60 for this indicator is 1.22 percent. Coverage (%) 50 Safety. The NCWSC’s water qual- 40 ity performance compares favor- 30 ably to other WSPs in Kenya, with 20 an overall water quality perfor- 10 mance of 95 percent in the 2012/13 0 to 2013/14 reporting period (mea- sured by adherence to  WASREB 98 99 00 03 04 08 09 10 01 02 12 13 14 96 06 11 05 5 97 07 9 20 20 20 20 20 20 20 19 19 20 19 20 20 19 20 20 19 20 20 20 water quality standards) (WASREB Source: IBNET, www.ib-net.org. 2015). Kenya’s overall water qual- ity performance against these standards was lower, at 91 percent. WASREB has stated that the Athi WSB, along with most other WSBs, needs to do more to ensure compliance with its Performance of Water Utilities in Africa 125 Figure B.30. Hours of Water Supply per Day, 2011–14 water quality and effluent monitoring standards, including further investment in laboratory facilities and ensuring 24 adequate provision for water quality analysis in tariff pro- posals (WASREB 2015). 20 Operational Performance 16 Hours per day NRW. NRW levels remained at a consistently high level 12 between 2010 and 2014, ranging from 600 liters per connec- 8 tion per day to 700 liters per connection per day. In percentage terms, there was a slight downward trend over that period 4 (from 42 percent in 2010 to 39 percent in 2014). This level of 0 inefficiency is in line with average performance in Kenya 2011 2012 2013 2014 (42 percent), but far short of the “acceptable” standard set by WASREB (less than 25 percent) and the NCWSC’s own target of Source: IBNET, www.ib-net.org. 30 percent. According to the NCWSC, this underperformance Figure B.31. Sufficiency of Consumption, 2010–14 is mainly due to low investment in NRW reduction projects. The utility included NRW reduction efforts in its strategic plan 90 for 2019; it estimates the cost to be K Sh 3.2 million (US$37 mil- 80 lion) to reduce NRW to 16 percent by 2019 (NCWSC 2014). 70 Figure B.32 shows the NCWSC’s NRW trend over time mea- Litres per person per day sured in liters per connection per day. Figure B.33 shows NRW 60 as a percentage of the NCWSC’s total water production. 50 40 Staff Productivity. The ratio of staff employed per 1,000 water 30 and sewer connections has been fairly constant since 2009, at 5, while staff numbers rose by about 700. These trends 20 are shown in figure B.34. 10 The NCWSC’s staff efficiency (revenue per employee 0 divided by labor costs per employee, or simply revenue 10 12 13 14 11 divided by labor costs) has stayed relatively constant from 20 20 20 20 20 2009 to 2013, slightly above 2. At 2.13 in 2013, the NCWSC is Source: IBNET, www.ib-net.org. far below the regional benchmark of 4.21, indicating that staff expenses are a large portion of their operating expenses. Financial Performance Collection Efficiency. Collection efficiency increased substantially in 2014, reaching 91 per- cent. This is a significant improvement from previous years, in which this ratio ranged figure  B.35). However, it remains lower than many from 75 percent to 85 percent (see ­ well-performing utilities in Africa—such as ONEA and the NWSC—who almost consistently achieve collection efficiency rates of 95 percent and above. This improvement can be attributed to a number of factors. In 2014, the NCWSC introduced Jisomee Mita, which 126 Performance of Water Utilities in Africa Figure B.32. Nonrevenue Water, by Connection, 2010–14 enables customers to use a mobile phone to receive their ­eliability of water bill and pay for water use. Increased r 800 water supply (see figure B.30) could be another factor. 700 According to WASREB, improvement in hours of supply Liters per connection per day 600 increases customer satisfaction, which translates to will- ingness to pay, which has a direct correlation with collec- 500 tion efficiency (WASREB 2015). 400 300 Operating Cost Coverage. The trend in the operating cost 200 recovery ratio is shown in figure B.36. Real average tariffs and real average costs per m3, in Kenya shillings, are shown 100 in figure B.37. The U.S. dollar equivalents for 2014 are also 0 noted in the figure. 10 11 12 13 14 20 The NCWSC’s OCCR has been slightly above 1 since 2009, 20 20 20 20 Source: IBNET, www.ib-net.org. except in 2012 when the ratio fell to 0.96. Staff expenses comprise a large proportion of operating expenses—on Figure B.33. Nonrevenue Water as a Percentage of average, about 40 percent during this period. In real terms, Production, 2010–14 average labor costs per employee rose by 6.7 percent per 100 year during this period. The OCCR may be overstated, how- 90 ever, because it seems that the NCWSC is underproviding 80 for bad debts. From 2009 to 2014, provisions accounted for 70 11  percent of revenue. However, the collection ratio (see 60 figure B.35) mostly ranged from about 75 percent to 85 per- NRW (%) 50 cent, rising to 91 percent in 2014 only. This suggests that 40 provisions for bad debts should have been in between 15 30 percent and 25 percent of revenue for much of this period. 20 Since 2010, the real average tariff (measured in 2014 Kenya 10 shillings) has declined at an average annual rate of 8.9 per- 0 cent. Tariffs were not changed during this period. At the same time, real operating expenses per m3 sold decreased at 10 12 13 14 11 20 20 20 20 20 a similar rate.16 Source: IBNET, www.ib-net.org. In U.S. dollars, the average tariff (water and sewerage) was US$0.66 in 2014 (see figure B.37). This is low compared with some other large utili- tariff is US$1.12, the NWSC’s average tariff is US$1.16, ties in the region—ONEA’s average ­ and SODECI’s average tariff is US$0.99. In November 2015, a new  tariff structure came into effect. Consumption up to 6  m3 per month is charged at a flat rate of K Sh 204 (US$2.00)—about K Sh 34 per m3 (US$0.33 per m3) if 6 m3 were consumed. This rate applies to both domestic and commercial customers. The highest block, for consumption greater than 60 m3 per month, is K Sh 64 (US$0.63). Sewerage is charged at 75 percent of water billed for all customers with a connection. Performance of Water Utilities in Africa 127 Figure B.34. Staff Numbers and Staff per 1,000 Water and Sewer Connections, Financial Transparency. The 2009–14 NCWSC does not publish an annual report or financial state- 3,000 8 2,612 ments on its website. Up-to-date 7 Sta per 1,000 connections 2,500 tariffs are not posted either. The 1,918 6 2,000 Strategic Plan for the  2014/2015 Number of sta 5 1,500 5 4 to 2018/2019 planning period 5 3 (dated March 2014) is available 1,000 2 on the NCWSC website. WASREB 500 1 publishes the NCWSC’s key per- 0 0 formance indicators in its annual 10 12 13 09 14 11 impact report. 20 20 20 20 20 20 Total sta Sta per 1,000 connections Investment. No data were avail- Source: IBNET, www.ib-net.org. able on past investment Figure B.35. Collection Ratio, 2009–14 ­ illion expenditure. According to the NCWSC, about K Sh 42 b (US$478 million, 2014) is required to effectively implement 100 its strategic plan over the 2014/15 and 2018/19 financial Collection ratio (%) 80 years (NCWSC 2014). This would be around US$24 per per- 60 son served per year. The NCWSC expects to finance a large 40 portion of this plan with internally generated funds. This 20 will require a combination of cost savings and an increase 0 in revenues. As of 2013, the utility was barely able to pay its 09 10 11 12 13 14 O&M through its billed revenues. As collection efficiencies 20 20 20 20 20 20 were below 1, there was insufficient cash available in 2013 Source: IBNET, www.ib-net.org. for the utility to pay for its investments through internal Figure B.36. Operating Cost Recovery Ratio, 2009–14 funds. Development partners will help finance the larger projects, such as new dams, water intake works, water 1.2 treatment works, water trunk mains, and the distribution 1.0 network. The utility expects a limited number of projects to Operating cost recovery be financed through the PPPs. 0.8 0.6 Organization Human Resources. Upon formation, the NCWSC inherited the 0.4 majority of its staff from the Nairobi City Council, which 0.2 was previously responsible for Nairobi’s water services. 0 With the support of the Bank, the NCWSC reduced its work- force. This included the introduction of human resource 14 09 10 11 12 13 20 20 20 20 20 20 management and payroll software (Mugo 2006). The Source: IBNET, www.ib-net.org. NCWSC recruits its senior management team competitively, 128 Performance of Water Utilities in Africa Figure B.37. Average Tariffs, Average Costs per m3, advertising for the positions. Senior management are on 2010–14 performance contracts with specific targets. All directors and senior management are bound by a code of ethics and 90 all staff are bound by a code of conduct which aims to K Sh (2014) per m3 of water sold 80 US$0.66 70 enhance integrity and improve service delivery. 60 Strategic Planning and Accountability. The NCWSC has used 50 40 strategic planning since its inception in 2003 (NCWSC 2014). US$0.65 30 The NCWSC pursued two, three-year rolling strategic plans up 20 to June 2010 and a five-year strategic plan between 2011 and 10 2015. The current strategic plan covers 2014–19. The NCWSC’s 0 strategic planning includes a monitoring and evaluation 11 12 10 13 14 framework, which provides for assessment of performance 20 20 20 20 20 Real average tari Real average costs against a number of key performance indicators over time. Source: IBNET, www.ib-net.org. Commercial Techniques. A key element of the NCWSC’s stra- tegic planning is to leverage technology to improve performance. A particular area of focus has been on initiatives which take advantage of the high rates of mobile phone use in Kenya, including Jisomee Mita, which enables water consumers to use a mobile phone to query and receive current water bills and pay for water use; MajiVoice, a mobile phone cus- tomer complaint resolution tool; and Mobile Field Assistant, a mobile meter reader which allows staff to collect information on geo-references, meter readings, and location of households through a smartphone (Ndaw 2015). Summary The NCWSC is expanding coverage, which rose from 66 percent in 2009 to 80 percent in 2014. On average, water is available for 18 hours per day. Average residential consumption is at a good level by regional standards, at 70 liters per person served per day. However, opera- tional performance as measured by NRW, metering, and staff efficiency (as discussed in chapter 2) is below what is seen in the typical utility in Africa. However, NRW is high at 700 liters per connection per day (39 percent). Poor performance on this indicator could be one of the contributing factors to the NCWSC’s mediocre operating cost recovery ratio—which has hovered around 1.0 since 2009. The NCWSC supplies a lot of water to its network (more than 200 million m3 in 2014), yet just about 60 percent of what is supplied is actually billed. Relatively low tariffs could be another contributing factor. On other efficiency indicators, the NCWSC is a fair performer—better than the “typical” African water utility, but not on par with the best performers in the region. The staff produc- tivity ratio has been maintained at five staff per 1,000 water and sewer connections. The collection ratio is rising—at 91 percent in 2014, compared with an average of about 80 per- cent in the previous five years. Mobile bill payment and other innovative commercial tech- niques are contributing factors. Performance of Water Utilities in Africa 129 Nairobi is a rapidly growing city, expected to reach 4.5 million people by 2019. The NCWSC will need to finance significant capital investment to keep pace with population growth and reach the 20 percent still unserved today, while improving supply reliability. The required investment (2014–19) is estimated to be US$478 million. Senegal, SDE and SONES Background In 1995, the GoS split SONEES, the existing national utility, into three entities: an asset-holding company for water service assets (SONES), a private operator for water ser- vices engaged through an affermage contract (SDE), and a public, combined asset-owner and operator for wastewater services (ONAS). SONES and ONAS have entered into perfor- mance contracts with the GoS, represented by the Ministry of Water and Sanitation (Ministère de l’Hydraulique et de l’Assainissement) and the Ministry of Finance (Ministère de l’Économie, des Finances et du Plan). The SDE has an affermage contract with SONES and the government. The SDE, the private operator, is responsible for water supply services in 66 towns. Dakar, the capital, accounts for more than 50 percent of connections. The provision of water supply and sanitation services in Senegal is governed by the Water and Sanitation Law (Loi portant organisation du service public d’eau potable et d’assainissement des eaux usées domestiques) of September 24, 2008, which defines the responsibilities for managing urban and rural water and sanitation services and their delegation (including to private entities), the principles for delivering services, the monitoring and controlling of the delivery of services, and the cost recovery of these services. Performance Customer Performance Access to Water Services. In 1995, the SDE served about 69 percent of its service area ­ population—3.6 million people. By 2013, nearly 5.8 million people were served, out of 5.9 million people (98 percent coverage).17 Growth in the service area population and the population served are shown in figure B.38. Figure B.39 shows the growth in water coverage, defined as the proportion of the service area population served by the SDE. A breakdown of those served by connections and those served by standpipes is also included. Today, the SDE serves most people (89 percent) with a direct connection. Just 9 percent are served by public taps. These utility-reported estimates closely match household survey data for Dakar. In 2014, 86 percent of people reported access to piped water to their prem- ises and 9 percent of people reported access to standpipes (see figure B.40).18 In 2005, how- ever, the DHS data differed from the utility-reported data. Eighty-eight percent of those living in Dakar reported accessing water through a connection piped to their premises. For the same year, the SDE reported direct access for just 64 percent of its service area. This seems to indicate that direct connections were first promoted in Dakar and then in other 130 Performance of Water Utilities in Africa Figure B.38. Population Served Compared with Service Area Population, 1996–2014 centers. Alternatively, the SDE may have historically underesti- 7 mated the number served by 6 direct connections. There is a jump in the coverage data from 5 2008 to 2009, which could indi- People, millions 4 cate that the estimation method- ology was revised. The household 3 survey data for Dakar is shown in 2 figure B.40.19 The differences in coverage estimates between the 1 household survey and the SDE 0 estimates persisted in the 2013 household survey—with the SDE 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 20 20 20 20 20 20 19 20 20 20 20 19 19 19 20 20 20 20 20 estimates significantly higher Population in service area Population served than the household survey’s esti- Source: IBNET, www.ib-net.org. mates. It is important that the Figure B.39. Water Coverage (Population Served Divided by Service Area Population), utility undertakes some research 1995–2014 on the number of people it is actually serving as lower service 100 coverage would mean signifi- cantly higher water consumption 80 levels than are currently reported. Despite the improvement in the Coverage (%) 60 type of service provided to cus- tomers, the average ­ number of 40 people served per connection has declined only slowly. In 1995, an 20 average of 12 people were served per connection. By 2013, that 0 ratio was 11 people per connec- 00 01 02 03 04 05 06 20 7 08 09 10 11 12 13 14 95 96 97 98 99 tion (see figure B.41). 0 20 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 Water coverage, standpipes Water coverage Water coverage, connections Network Expansion and Growth in Connections. Between 1995 and Source: IBNET, www.ib-net.org. 2013, the distribution network in the SDE/SONES service area more than doubled, with 4,800 km of pipe added (see figure B.42). Active connections rose by 332,000 during this period (see figure B.43). Reliability. In the mid-1990s, Dakar faced a severe water shortage. SONEES (the utility at the time) could only supply about 60 percent of demand in the city. Overexploitation of Performance of Water Utilities in Africa 131 Figure B.40. Access to Water Service, Dakar Only groundwater resources led to the risk of saline intrusion. In some areas, groundwater was declining at 1.5 m per year. 100 This crisis forced the utility to ration water. Service was 90 available for about 16 hours per day on average. A new 80 Access, total population (%) production source was needed to provide 24-hour ser- ­ 70 vice.20 To improve service in the short term, new boreholes 60 were constructed and the existing treatment works were 50 expanded. This program led to increased supply reliability 40 figure B.44, in the late 1990s and early 2000s. As shown in ­ 30 average hours of service per day increased to 18 hours 20 (1999–2000) and then to 20 hours (2001–04). 10 In a second phase of improvement, 10 years later, a new 0 130 million liter per day treatment plant and associated 2000 2005 2011 2014 Public tap Piped to premise transmission works were constructed. Upon completion of this project, the SDE could meet water demand in Dakar. Sources: DHS 2005, 2011, 2014; MICS 2000. Average reliability across the net- work reached 24 hours per day Figure B.41. People Served per Connection, 1995–2013 from 2006 to 2008. Recently, however, supply constraints have 14 reemerged as an issue and new 12 production facilities are needed Peole per connection 10 to return to 24-hour service. 8 6 Sufficiency. The SDE reports that 4 average residential consumption has remained relatively constant 2 from 2006 to 2013, at 55 liters per 0 person per day, up from 44 liters 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 20 per person per day in 2004 (see 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 figure B.45). This increase corre- Sources: DHS 2005, 2011, 2014; MICS 2000. sponds to the completion of the new treatment plant to serve Dakar. Before this project, demand was not met. Over this period, the share of residential consumption has risen—from 67 percent in 2004 to 93 percent in 2013. This could be due to the high nonresidential tariffs depressing nonresi- dential consumption. Affordability. One measure of affordability is the percentage of income spent on water con- sumption. A proxy for the average annual per capita expenditure on water is total revenue from sales divided by the number of people served. A proxy for average per capita income is Senegal’s GNI per capita. Dividing the proxy for water expenditure into the proxy for income gives average expenditure on water at 2.2 percent of income. By contrast, the regional 132 Performance of Water Utilities in Africa Figure B.42. Network Expansion, 1995–2013 benchmark for this indicator is 1.22 percent. 10,000 9,000 Safety. The SDE has conducted 8,000 water quality tests since 1996. 7,000 6,000 At least 99 percent of samples Km pipe 5,000 passed water quality tests from 4,000 2010 to 2014. However, the data 3,000 2,000 shows a huge break in 2008/09, 1,000 with the absolute number of 0 water quality tests dropping to 95 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 about 8,000 per year, which 20 20 20 20 20 19 19 20 20 20 20 19 19 20 20 20 20 20 translates to only one  test Sources: DHS 2005, 2011, 2014; MICS 2000. every three days in every town served by the SDE. The data on Figure B.43. Number of Water Connections, 1995–2013 water quality testing are not 600 published. 500 Operational Performance Water connectionse 400 NRW. The SDE steadily reduced 300 NRW from 1996 to 2003 200 percent (29  ­ to 20 percent) 100 because the affermage contract 0 included strong incentives to do so. Since the early 2000s, how- 99 00 09 10 02 03 08 12 13 04 98 11 96 01 06 05 95 97 07 20 20 20 20 20 19 20 20 19 19 20 19 20 19 20 20 20 20 20 ever, NRW has plateaued at about Sources: DHS 2005, 2011, 2014; MICS 2000. 20 percent. In 2013, NRW was 159 liters per connection per day. Figure B.4 4. Hours of Supply per Day, 1995–2013 This is close to the global bench- mark of 121 liters per ­ connection 24 per day and better than the 20 African benchmark (at  205 liters 16 per connection per day). NRW Hours per day trends over time, measured  in 12 liters per connection per day 8 and  percentage of production, 4 are shown in figure B.46 and 0 figure B.47, respectively. ­ Metering across the service area 10 98 9 00 08 09 02 12 13 03 04 01 06 11 96 5 05 7 07 9 9 9 20 20 20 20 20 20 19 20 19 19 19 20 20 20 19 20 20 20 20 has been nearly universal (about Sources: DHS 2005, 2011, 2014; MICS 2000. 97 percent) since 2000. Performance of Water Utilities in Africa 133 Figure B.45. Sufficiency of Consumption, 2004–13 Staff Productivity. Staff productivity has improved signifi- cantly since 1996, with the number of staff per 1,000 water 60 connections falling from seven to two by 2013 (see figure 50 B.48). Staff numbers declined only slightly during this period. Liters per person per day 40 Staff efficiency (revenue per employee divided by labor costs per employee, or simply revenue divided by labor 30 costs) has risen from about four in 1998 to slightly more 20 than five (since 2007). This puts the SDE significantly above 10 the regional benchmark of 4.21. 0 Financial Performance 09 10 08 4 2 3 1 6 5 07 1 1 1 0 0 0 20 20 20 20 20 20 20 20 20 20 Collection Efficiency. The SDE has consistently maintained a Sources: DHS 2005, 2011, 2014; MICS 2000. figure B.49. A target for high collection ratio, as shown in ­ collections is included in the Figure B.46. Nonrevenue Water, by Connection, 1995–2013 affermage contract, and the oper- 450 ator has financial incentives to 400 ensure collections. In 2013, the Liters per connection per day 350 collection ratio fell to 94 percent. 300 It is not clear from the data 250 whether this includes collections 200 150 from all customers or only from 100 residential consumers, as anec- 50 dotal evidence seems to suggest 0 that government users have trou- 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 ble paying their bills on time. 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 Sources: DHS 2005, 2011, 2014; MICS 2000. Operating Cost Coverage. The Figure B.47. Nonrevenue Water as a Percentage of Production, 1995–2014 SDE’s OCCR has fluctuated over time, reaching a high of 1.55 in 50 2007. This could be attributed to 45 40 an increase in revenue after com- 35 pletion of the new production 30 NRW (%) facility near Dakar. Since then, 25 20 the OCCR has fluctuated, but 15 remained well above 1. The trend 10 in the operating cost recovery 5 0 ratio is shown in figure B.50. Real average tariffs and real average 11 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 13 14 12 20 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 costs per m3, in the local currency, Sources: DHS 2005, 2011, 2014; MICS 2000. are shown in figure B.51. The U.S. 134 Performance of Water Utilities in Africa Figure B.48. Staff Numbers and Staff per 1,000 Water and Sewer Connections, dollar equivalents for 2013 are 1996–2013 also noted in the figure. 1,600 8 1,400 Sta per 1,000 connections 1,400 1,271 7 Real Average Tariff. In real terms, 1,200 7 6 the average tariff has declined at Number of sta 1,000 5 an average annual rate of 2.8 per- 800 4 ­ gure B.51). cent since 2009 (see fi 600 3 400 2 This could be attributed to the 2 200 1 increased share of residential 0 0 consumption because domestic customers have a lower tariff. In 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 20 20 20 20 20 19 19 20 20 20 20 19 19 19 20 20 20 20 20 Total sta Sta per 1,000 connections 2013, the average tariff was equiv- alent to US$1.08. Sources: DHS 2005, 2011, 2014; MICS 2000. Senegal’s increasing block tariff structure has a subsidized social Figure B.49. Collection Ratio, 1996–2013 tariff for levels of consumption 100 below 20 m3 (CFAF 202; US$0.40) per two months. There is also a Collection ration (%) 80 regular tariff for consumption 60 from 21 m3 to 40 m3 (CFAF 697.97; 40 US$1.39) and a “dissuasive” tariff 20 for consumption above 40 m3 0 (CFAF 878.35; US$1.75). The dis- suasive tariff is designed to be a 96 98 10 11 12 13 97 00 01 02 03 04 05 06 07 08 09 99 20 20 20 20 20 19 20 20 20 20 19 19 19 20 20 20 20 20 disincentive for excessive water Sources: DHS 2005, 2011, 2014; MICS 2000. use. It can be seen that the tariff for household consumption of Figure B.50. Operating Cost Recovery Ratio, 1998–2013 less than 20 m3 per 60 days is less than a third of the regular tariff, 1.8 1.6 and less than a quarter of the tar- iff for consumption in the top Operating cost recovery 1.4 1.2 block. Bills are sent every two 1.0 months based on meter readings, 0.8 and the SDE can cut off water 0.6 supply for nonpayment. 0.4 Nonresidential, nongovernmen- 0.2 0 tal customers must pay the dissuasive ­ tariff regardless of 8 10 11 12 13 00 01 02 03 04 05 06 07 08 09 99 9 20 20 20 20 20 20 20 20 20 19 19 20 20 20 20 20 amount consumed. As of 2013, just Sources: DHS 2005, 2011, 2014; MICS 2000. 7 percent of SDE customers were Performance of Water Utilities in Africa 135 Figure B.51. Average Tariffs, Average Costs per m3, 1997–2013 classified as nonresidential, down from 33 percent in 2004. High tar- 700 US$1.08 iffs could be a contributing factor CFAF (2013) per m3 of water sold 600 to this trend. Government custom- 500 ers pay more than twice the dissua- 400 sive tariff—their tariff is CFAF 300 1,868.88 per m3 (US$3.72 per m3). US$0.81 The structure has been in place 200 since 2007. In  that year, the gov- 100 ernment agreed to raise tariffs for 0 government customers by 70 per- 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 20 cent, while keeping tariffs for other 20 20 20 20 19 20 20 20 20 19 19 20 20 20 20 20 Real average tari Real average costs customers constant. This was Sources: DHS 2005, 2011, 2014; MICS 2000. introduced as a way to keep domes- tic tariffs from rising while still Figure B.52. Sources and Amounts of Investment Financing, 1996–2013 ensuring cost recovery for SONES/ SDE. In 2015, domestic tariffs were 70 also raised—the lowest tariff block 59 60 was raised by 4 percent and other 50 rates were increased by 9 percent. Billions FCFA 40 33 30 Real Average Cost per m3 Sold. From 25 23 30 19 18 17 14 13 1998 to 2005, real average costs 12 12 12 11 20 10 11 9 9 declined by an average of 4.5 per- 10 cent per year (see figure B.51). In 0 recent years, real average costs have fluctuated. Average costs 98 00 03 99 02 04 01 06 6 05 10 13 08 09 12 97 11 07 9 20 20 20 20 20 20 20 19 19 19 20 20 19 20 20 20 20 20 per m3 were equivalent to US$0.81 From grants From donor loans From own cash per m3 in 2013. Sources: DHS 2005, 2011, 2014; MICS 2000. Financial Transparency. The SDE and SONES do not publish annual reports or financial statements. Tariffs are posted on the SDE website, but they are out of date (from 2003). Performance results are also out-of-date— the most recent data published are from 2012. Investment. For the SDE and SONES to increase access and improve service, investment total- ing about US$770 million was required from 1996 to 2013.21 This amounts to about US$10 per person served per year. Figure B.52 shows the sources and amounts of investment financing during this period. While about 29 percent was grant financed, 23 percent was financed by own cash and 47 percent from loans from donors. The loans from donors are repaid from tariff revenue, which is allocated to SONES for this purpose. 136 Performance of Water Utilities in Africa Two major programs implemented during this period were the Senegal Water Project (US$223 million, 1996–2004) and the Long Term Water Project (US$255 million, 2002–09).22 The Bank was a major financier for both projects, providing US$85 million and US$146 million, respectively. similar to International Bank for Reconstruction and Overall borrowing terms were ­ Development terms, with interest rates around 6 percent and tenors of 20 years (including five-year grace periods).23 The cost of finance was kept down through the financing structure used. Because the investments were the responsibility of the publicly owned SONES, the funds were lent to the government on concessional International Development Association terms. The first project focused on urgent investments needed to increase water supply in Dakar. This included additional boreholes, expansion of a treatment plant, and leakage reduction works. A private operator was brought in through this project as well. Signing a PPP contract satisfactory to the Bank was a condition of the Bank loan. The second phase of reforms involved the construction of the much larger Keur Momar Sarr water treatment plant (in 2005, 65,000 m3 per day; upgraded in 2008 to 130,000 m3 per day) along with a continued expansion in the distribution network. Organization Human Resources. Management capability came from both the private and public sectors. The new state-run asset holder, SONES, maintained continuity by retaining the managing director of the forerunner institution, SONEES. All permanent employees of SONEES had guaranteed employment in the new structure. Meanwhile, the affermage contract also brought in private ­ ompany, SAUR. Senegal’s sector expertise from an experienced and specialized international c long experience with private sector management—it had an affermage contract before the 1972 nationalization and a consulting contract with SAUR between 1972 and 1995—facilitated trust, familiarity, and cooperation. That operator in turn brought in modern management sys- tems and techniques and skilled managers. The operator trained local staff, so the company is now run almost entirely with national staff at the same high levels of efficiency. Profit incen- tives for owners in turn led to performance-oriented management for the staff. Strategic Planning. The various contracts establish a process for investment planning. SONES  (the public asset-holding company) has to prepare a 10-year investment plan and a 3-year rolling investment plan, based on the SDE’s inputs on demand and service needs. The parties then have to agree on a three-year investment convention containing the detailed planning of the works in the coming three years. No investment work can be carried out if it has not been previously included in that investment convention. Summary In Senegal, the SDE estimates that about 98 percent of people in its service area now have access to piped water services. Eighty-nine percent have access to a connection on premises while 9 percent have access to a standpipe. On average, water is available for 23 hours per day. Residential customers consume 92 percent of all water sold, at an average of 55 lcd. Performance of Water Utilities in Africa 137 This situation is a significant improvement over that of the mid-1990s, when Dakar faced a water shortage. With groundwater resources depleting rapidly, supply was rationed to 16  hours per day. The utility needed to finance a new production source and turned to development partners for support. The Water Sector Project, supported by the Bank, developed a plan to finance short-term production sources. A condition of this loan was the introduction of a private operator to improve efficiency and management. In 1995, the GoS split SONEES, the existing national utility, into three entities: an asset-holding com- pany for water service assets (SONES), a private operator for water services engaged through an affermage contract (SDE), and a public, combined asset-owner and operator for wastewater services (ONAS). Private operation did improve efficiency, with NRW falling from 29 percent to 20 percent in eight years (1995–2003) and the collection ratio rising to 98 percent by 1997. Staff produc- tivity improved from seven staff per 1,000 connections (1996) to two staff per 1,000 connec- tions (2013). Total capital investment in the sector from 1996 to 2013 totaled about US$770 ­ m illion. About a quarter of this was financed by own funds and half was financed by loans from donors. The SDE is limited by its contract to charge a fixed operator tariff, so the average retail tariff paid by customers has to cover the operator tariff, in addi- tion to an amount to service the debt taken on by SONES to finance the infrastructure. Tariffs are kept at affordable levels by cross-subsidizing households that consume less than 20 m 3 every two months. The nonresidential tariff is four times this social tariff (US$1.75 per m 3), while the tariff for government customers is nine times the social tariff (US$3.72 per m 3). A consistent increase in the residential share of con- sumption in recent years (92 percent as of 2013 with only 8 percent being consumed by nonresidential water customers) could threaten the sustainability of this cross-sub- sidy mechanism. Uganda, NWSC (National Water and Sewerage Corporation) Background In 2013 the NWSC was providing water supply and (sometimes) sewerage services to 23 urban centers and towns in Uganda. 24 The NWSC is internationally recognized for its successful performance turnaround, which began in 1998. Two important things hap- pened in 1998: a management contract with a German engineering firm (Gauff) was signed and a new managing director (Dr. William Muhairwe) was appointed. Dr. Muhairwe launched several internal programs to improve operational efficiency, such as the 100 days’ program. In 2002, a second management contract was signed, this time with Ondeo. The IDAMCs, first piloted in 2004, were another successful initiative. Each town was estab- lished as a business unit and managers were held accountable for meeting set performance targets. Kampala, which accounts for more than 60 percent of total revenue, was further 138 Performance of Water Utilities in Africa Figure B.53. Population Served Compared with Service Area divided into branches, each responsible for operational Population, 2003–14 activities in its service area and incentivized to meet targets set in the Branch Performance Contract. 7 As a result of these reforms, the NWSC’s operational 6 and financial performance improved significantly during 5 the late 1990s and early 2000s. Its achievements were maintained up to 2013, the final year analyzed in this People, millions 4 case. 3 2 Performance Customer Performance 1 Access to Water Services. In 1998, the NWSC served about 0 people). By half of its service area population (1  million ­ 10 11 12 13 14 03 04 05 06 07 08 09 2013, nearly 3 million people were served out of 3.8 million 20 20 20 20 20 20 20 20 20 20 20 20 Population in service area Population served people (78  percent coverage). Growth in the service area population and the population served over this period is Sources: DHS 2005, 2011, 2014; MICS 2000. figure B.53. Figure B.54 shown in ­ shows the growth in water cover- Figure B.54. Water Coverage (Population Served Divided by Service Area Population), age, defined as the proportion of 1998–2013 the service area population served by the NWSC. 100 According to the NWSC data, 90 the share of the population 80 served by domestic connections 70 and that served by public taps were roughly equal in 2013, with Water coverage (%) 60 50 access to domestic connections having increased slowly over 40 time. 25 However, household 30 ­ survey data for Kampala (the 20 capital city that accounts for 10 more than 60 percent of the 0 NWSC’s revenue) differ signifi- cantly. About three-quarters of 10 00 09 12 03 08 99 02 04 11 01 8 06 13 05 07 9 20 20 20 20 20 20 20 19 20 20 19 20 20 20 20 20 survey respondents with access Sources: DHS 2005, 2011, 2014; MICS 2000. to piped water reported access- ing it from a  public tap.26 Nevertheless, the average number of people served per connection has declined over time—from 15 in 2003 to nine in 2009. From 2009 to 2013, the indicator remained rela- tively ­stable at nine people per connection. Performance of Water Utilities in Africa 139 Figure B.55. Network Length, 2002–13 Network and Town Expansion. From 2002 to 2013, the distribu- tion network length tripled from 1,846 km to 5,670 km due to 6,000 (a) incorporation of new towns into the service area and (b) extension of service to new customers in Kampala and else- 5,000 where. The growth of the network over time is shown in fig- 4,000 ure B.55; the increase in towns served is shown in figure B.56. Preliminary data from 2014 and 2015 show that the number of Km pipe 3,000 towns to be served by the NWSC will expand rapidly.27 2,000 Reliability. Reliability of water service is moderate at 20 hours per day, compared with a regional benchmark of 21.6 1,000 and a global benchmark of 24 (see figure B.57). Recent chal- 0 lenges in increasing reliability have been drought, unreli- able power supply, and growing demand for services. The 10 09 08 13 12 02 03 04 11 06 05 07 20 20 20 20 20 20 20 20 20 20 20 20 NWSC’s goal for the 2015–18 period is to achieve 24–7 reli- Sources: DHS 2005, 2011, 2014; MICS 2000. ability in all towns.28 Figure B.56. Towns Served, 2003–13 Sufficiency. Average residential water consumption per cap- ita has declined steadily since 2012, from 38 lcd to 23 lcd 25 (2013) (see figure B.58). Affordability. One measure of affordability is the percentage 20 of income spent on water consumption. A proxy for the average annual per capita expenditure on water is total rev- Towns served 15 enue from sales divided by the number of people served. A 10 proxy for average per capita income is Uganda’s GNI per capita. Dividing the proxy for water expenditure into the 5 proxy for income gives the average expenditure on water at 4.3 percent of income. By contrast, the best regional perfor- 0 mance benchmark for this indicator is 1.22 percent. 10 09 13 12 08 03 11 06 05 07 20 20 20 20 20 20 20 20 20 20 Safety. Results of water quality tests have been published in Sources: DHS 2005, 2011, 2014; MICS 2000. the publicly available annual reports since 2010. In 2013, the NWSC tested more than 9,000 water samples for Escherichia coli, which translates to about one sample per town per day. At least 97 percent of samples passed E. coli tests in all 23 towns. In the same year, 22 of 23 towns in the NWSC service area produced water meeting the national standard for turbidity (less than 5 nephelometric turbidity units [NTU]).29 Operational Performance NRW. The NWSC has made significant progress in reducing NRW, which decreased from 1,210 liters per connection per day (1998) to just 200 liters per connection per day (2015). Current per- formance is slightly better than the regional benchmark of 205 liters per connection per day. 140 Performance of Water Utilities in Africa Figure B.57. Hours of Water Supply per Day, 1996–2013 NRW trends over time, measured in  liters per connection per day 24 and  percentage of production, are  shown in figure B.59 and 20 figure B.60, respectively. ­ Measured in percentage terms, 16 NRW has fallen from 50 percent to Hours per day 32 percent during the same period 12 (see figure B.60). Most progress on this indicator was achieved during 8 the 1998–2006 period during the management contracts (Gauff, 4 1998–2001; Ondeo, 2002–04) and introduction of the IDAMCs. During 0 10 reliability also this period, service ­ 00 98 09 03 08 12 99 02 04 11 01 96 06 13 05 07 97 increased (see ­ figure  B.57). Since 20 20 20 20 20 20 20 19 19 19 20 20 19 20 20 20 20 20 then, NRW levels have plateaued. Sources: DHS 2005, 2011, 2014; MICS 2000. One challenge has been reducing Figure B.58. Sufficiency of Residential Consumption, 2005–13 NRW in Kampala, the capital city of Uganda, which accounts for about 70 percent of the NWSC’s revenue. Metering across 40 the service area has been more or less universal (above 98 per- 35 cent) since 2005. 30 Staff Productivity. Staff productivity has improved signifi- Liters per person per day 25 cantly since 1998, with the number of staff per 1,000 water and sewer connections falling from 36 to 10 in the first four 20 years (1998–2002) and from 10 to 6 in the following four years 15 figure B.61). Early progress was achieved by (2002–06) (see ­ 10 laying off staff, but from 2002–06, both staff numbers and productivity grew. This was possible because network expan- 5 sion significantly outpaced hiring of new labor. 0 Since 2006, staff productivity has stagnated, equaling 10 09 13 12 08 11 06 05 07 about five staff per 1,000 water and sewer connections in 20 20 20 20 20 20 20 20 20 2013. The NWSC’s staff efficiency (revenue per employee Sources: DHS 2005, 2011, 2014; MICS 2000. divided by labor costs per employee, or simply revenue divided by labor costs) has ranged from 3 to 4.5 from 2005 to 2013. At 3.4 in 2013, the NWSC is below the African benchmark of 4.27 and the global benchmark of 4.21. Financial Performance Collection Efficiency. Collection efficiency increased from 85 percent in 2001 to 95 percent or figure  B.62). Among other initiatives, the NWSC has built higher from 2009 to 2013 (see ­ Performance of Water Utilities in Africa 141 Figure B.59. Nonrevenue Water, by Connection, 1998–2013 automatic water dispensers (standpipes that dispense water 1,400 when a prepaid token is inserted) to ensure payment. 1,200 Liters per connection per day 1,000 Operating Cost Coverage. The NWSC has had an OCCR of 1.20 or 800 higher since 2002, indicating financial sustainability (see figure 600 B.63). This is on par with the best performers in Africa. 400 Real average tariffs have 200 remained relatively constant from 2002 to 2013, between U Sh 3,000 0 per m3 and U Sh 3,500 per m3— or  US$1.16 per m3 in 2013 U.S. 10 00 09 12 03 08 99 02 04 11 01 98 06 13 05 07 20 20 20 20 20 20 20 19 20 20 19 20 20 20 20 20 dollars, as shown in figure B.64. ­ Sources: DHS 2005, 2011, 2014; MICS 2000. Real average operating costs Figure B.60. Nonrevenue Water as a Percentage of Production, 1998–2013 have  exhibited a similar pattern, amounting to about US$0.89 per 100 m3 in 2015. Since 2010, the real 90 average tariff has declined. From 2004, the NWSC’s tariff was 80 indexed to inflation. There is no 70 regulator—the indexation mecha- 60 nism was approved by parliament. NRW (%) This provided legitimacy for the 50 tariff indexation and reduced the 40 risk of future executive action to 30 prevent annual increases. Cross-subsidies are important 20 for the affordability of the NWSC’s 10 service. The domestic tariff of 0 US$0.77 is higher than the stand- pipe tariff of US$0.47. Commercial 10 00 09 12 08 98 99 02 03 11 04 01 13 05 06 07 20 20 20 20 20 20 20 19 19 20 20 20 20 20 20 20 customers are charged even more. Sources: DHS 2005, 2011, 2014; MICS 2000. However, to keep large users on the system, the tariff rate for commercial consumption above 1,500 m3 per month is lower than for consumption below that rate (US$0.93 per m3 compared with US$1.16 per m3) (NCWSC 2014). In 2013, commercial users accounted for about one-third of total volume billed. 142 Performance of Water Utilities in Africa Figure B.61. Staff Numbers and Staff per 1,000 Water and Sewer Connections, Financial Transparency. The NWSC 1997–2013 publishes an annual report that reviews its performance against 2,000 40 36 1,793 well-defined targets and includes 1,800 35 audited financial statements. 1,600 Sta per 1,000 connections 30 1,400 These reports are made available 25 on the corporation’s website, but Number of sta 1,200 1,092 1,000 20 they are not up-to-date—the lat- 800 15 est available is  2012–13. The 600 Corporate Plan 2015–18 is avail- 5 10 400 able on the website. 200 5 0 0 Investment. For the NWSC to 10 09 98 99 08 12 00 02 03 04 11 01 97 06 13 05 07 20 20 20 20 20 19 20 20 19 19 20 20 20 20 20 20 20 increase access and improve ser- Total sta Sta per 1,000 connections vice, about US$100 million in Sources: DHS 2005, 2011, 2014; MICS 2000. capital expenditure was required from 2002 to 2011 according to cash flow statement Figure B.62. Collection Ratio, 2001–14 analysis.30 This amounts to about US$3.85 per person served 100 per year. Figure  B.65 shows the sources and amounts of capital expenditure financing during this period. While Collection ratio (%) 80 about 28 percent was grant financed, 52 percent was 60 financed by internal cash flow and 16 percent from loans. A 40 small portion was financed by other sources—this includes 20 cash from nonoperating activities, such as proceeds from 0 disposal of property, plant, and equipment. All values are in nominal Uganda shillings. The exchange 10 09 13 12 08 02 03 14 04 11 01 06 05 07 20 20 20 20 20 20 20 20 20 20 20 20 20 20 rate in 2011 was U Sh 2,340 to US$1. One project the NWSC Sources: DHS 2005, 2011, 2014; MICS 2000. undertook during this period was the Gaba III water treat- ment plant and transmission mains (U Sh 52.7 billion; US$28.8 million [2006]). This project increased water production for Kampala and the nearby areas by 80,000 m3 per day. In 2010, a commercial loan of US$2 million was obtained for financing the extension of the Ggaba intake plant, which supplies water to Kampala City and the surrounding areas. This loan is being serviced from operating cash flow. Early in this period, the NWSC did not repay its loans with operating cash. The government of Uganda agreed to a moratorium on debt service for a period, which gave the NWSC finan- cial breathing space. Then, in 2007, the government converted the outstanding balance of US$47 million into equity—effectively forgiving the debt. Since then, the NWSC has bor- rowed from commercial banks and is repaying from operating cash flow. Increasing operat- ing cash flow contributed to the expansion in service. Important drivers of increasing cash flow were increasing collection efficiency (85 percent in 2001 to 95 percent in 2011), Performance of Water Utilities in Africa 143 Figure B.63. Operating Cost Recovery Ratio, 2001–14 reductions in NRW (43 percent in 2001 to 33 percent in 2011), increasing labor productivity by limiting staff growth as 1.6 connections increased, and a modest increase in the real 1.4 tariff of 3 percent annually. 1.2 Operating cost recovery Organization 1.0 Human Resources. The former utility manager described a 0.8 culture of slackness pervading the NWSC when he took 0.6 over its management (1998). Garbage was left lying 0.4 around offices. To shock the organization into change, he instituted a 100-day turnaround program (February 1999– 0.2 May 1999), in which managers and their teams committed 0 to extraordinary goals which they would achieve within 10 09 13 08 12 03 02 14 04 11 01 06 05 07 the first 100 days. This signaled that things were changing 20 20 20 20 20 20 20 20 20 20 20 20 20 20 and helped to build a performance culture within the Sources: DHS 2005, 2011, 2014; MICS 2000. organization. Other initiatives followed, including con- Figure B.64. Average Tariffs, Average Costs per m3, 2001–14 sultative strategic planning to build a sense of common purpose (Mugisha, Berg, and Muhairwe 2007). 4,000 Since 2004, the NWSC has used IDAMCs to motivate man- US$1.16 3,500 agement teams in each area of its operation. Management U Sh (2013) per m3 of water sold teams are selected competitively through an open process. 3,000 They commit to put a part of their salary at risk, in exchange 2,500 for a bonus if they meet targets for service levels and 2,000 improvements in operating cash flow. US$0.89 1,500 To help staff improve, the NWSC has a training center in Kampala. Courses include customer care, surveying tech- 1,000 niques, and ethics and integrity. The NWSC also has a voca- 500 tional training facility for technical staff. For professional 0 staff, the NWSC finances educational scholarships and pro- vides low-interest study loans to assist well-performing 10 13 09 12 20 3 08 14 02 04 11 20 1 06 20 5 20 7 0 0 0 0 20 20 20 20 20 20 20 20 20 20 Real average tari Real average costs staff with their career development. Sources: DHS 2005, 2011, 2014; MICS 2000. Strategic Planning and Accountability. In 2003, a performance contract was agreed between the government and NWSC. The contract set out targets for the NWSC, including developing plans and funding for network expansion. The contract was essentially a corporate strategic plan with quantified targets and milestones, setting out the agreed way forward.31 The NWSC has used inclusive strategic planning successfully for some time—the utility recently published its 7th Corporate Plan (2015–18), which is available on its website. In 2013, the utility published its first ever five-year Strategic Direction (2013–18), which staff 144 Performance of Water Utilities in Africa Figure B.65. Capital Expenditure Financing, 2002–07, 2009–11 at all levels were involved in preparing. 40 38 35 35 Commercial Techniques. Innovative 30 U Sh, billions 25 commercial techniques help the 20 17 16 11 19 11 11 14 NWSC serve customers better. 15 10 Most public taps are operated as 5 0 kiosks by someone who has paid for the connection and then 8 6 9 4 2 3 5 7 0 1 0 0 0 0 0 0 0 0 1 1 20 20 20 20 20 20 20 20 20 20 on-sells the water. Some kiosks Cash from operating activities Loans are municipal, some are run by used to finance capex Grants Other sources community groups, and others are private. Although the NWSC Sources: DHS 2005, 2011, 2014; MICS 2000. standpipe tariff is about 39 per- cent less than the domestic tariff, the NWSC found that some of the kiosks charge an excessive markup for the water. Once the water is dispensed in 20 liter jerry cans at the kiosk, the effec- tive tariff could be US$1 per m3 or more (Kariuki et al. 2014). In response, the NWSC has installed standpipes that automatically dispense water when a customer inserts an electronic token. By cutting out the middlemen (the kiosk operator) these prepaid water points ensure that cus- tomers can access water at the low standpipe tariff set by the NWSC, without markups. A similar effort that helps ensure affordability is the NWSC’s recognition that some indi- vidual yard taps are in fact shared water points. If consumption of water by one direct con- nection seems high, the NWSC staff will visit to check if the yard tap serves more than two or three households. If so, the tariff for public water points is applied. Summary As a result of the NWSC’s turnaround, about 76 percent of people in the utility service area now have access to piped water services, up from just 47 percent in 1998. On average, water is available for 18 hours per day. Cross-subsidies—and standpipes that dispense water automatically—help ensure water is affordable to residential customers. Revenue covers all operating costs and some capital costs, a result of operational efficiency and a cost recovery tariff. The NWSC can borrow on commercial terms and service the debt with operating cash. The reforms were phased. Early on, local and international management models were tried in parallel, with the primary goal of continuing to serve existing customers while reducing financing losses through greater efficiency. This was successful, as evidenced by NRW falling from 50 percent to 30 percent (1998–2006) and staff productivity rising from 36  staff per 1,000 connections to six staff per 1,000 connections (1998–2006). Over time, pragmatic solutions, including tariff indexation and a performance contract rather than a regulator, were developed. The IDAMCs (since 2004) incentivize staff to meet operational and financial targets such as NRW, arrears, the working ratio, and connection efficiency. Performance of Water Utilities in Africa 145 Update: 2014–15 In 2013, the NWSC launched an impressive new “water for all” campaign. Now that the NWSC is financially sustainable, the goal is to increase water supply access across the coun- try. The number of towns served rose from 23 to 149 in less than three years (June 2013 to March 2016). In these new towns, the NWSC has mostly taken over the operations and main- tenance of existing infrastructure from the Directorate of Water Development. Data on this expansion have been incomplete and hence could not be verified during this study. Prior data collection in the newly added towns was poor, so the NWSC plans to conduct a cus- tomer survey to establish better data. Notes 1. DHS 2010 at Demographic and Health Surveys for selected countries at http://dhsprogram.com/data/ 2. This is a very basic indicator for affordability as many utilities only provide services to urban residents (whose incomes tend to be higher than the national average). In addition, the revenues spent on water also include the sales to nondomestic users (as many utilities do not separately report for residential users and hence the effect of cross-subsidies cannot always be detected). 3. This means that more than 25 percent of African utilities in the sample collected have a staff efficiency indicator value higher than ONEA’s. 4. The increasing block tariff structure will discourage the sharing of taps. 5. ONEA, “Les tarifs,” http://oneabf.com/les-tarifs/. 6. IBNET. 7. Figure quoted based on 2013 dollar value. 8. IBNET at https://database.ib-net.org/utility_profile?uid=5375 9. SODECI Annual Report 2012. 10. The collection period refers to the average number of days it takes customers to pay their bills. It is calculated as follows: accounts receivables at year end/revenue × 365. 11. For more information on the training centers, see the eranove website at http://www.eranove.com/en/collaborators​ /­training/. 12. The Water Act 2002 is under review and is subject to reform by the Water Bill 2014. According to the Kenyan Senate’s 2016 “Bills Tracker” (http://www.parliament.go.ke/the-senate/house-business/bills-tracker), the Water Bill 2014 is in the advanced stages of Kenya’s legislative process. 13. “Geographical Coverage”, Athi Water, http://awsboard.go.ke/about/our-mandate/. 14. https://www.nairobiwater.co.ke/projects. 15. https:// https://www.nairobiwater.co.ke/index.php/en/ 16. The reason for the decline in O&M costs is not clear. 17. The household survey of 2013 estimated the total population in Senegal serviced with piped water at only 5.3 million (compared with the SDE’s estimate of 5.8 million), with a different split between house and yard connections (4.7 million in ­ household survey compared with 5.3 million by the SDE) and public taps (0.6 million in household survey versus 0.5 million in the SDE’s estimates). 18. http://dhsprogram.com/pubs/pdf/FR305/FR305.pdf 19. MICS 2000, “Multiple Indicator Cluster Surveys” at http://www.unicef.org/statistics/index_24302.html. DHS 2005, 2011, 2014, “Demographic and Health Surveys” at http://dhsprogram.com/data/. 146 Performance of Water Utilities in Africa 20. Staff Appraisal Report, Republic of Senegal, Water Sector Project June 12, 1995, 2–4. 21. Figure quoted in 2013 U.S. dollars. 22. Implementation Completion and Results Report, Senegal Water Project; Implementation Completion and Results Report, Long Term Water Project. These projects included sanitation components which are overseen by ONAS, and not the SDE or SONES. 23. Staff Appraisal Report, Senegal Water Project, iv; Project Appraisal Document: Long Term Water Project, 20. According to the NWSC website, 149 towns were served as of March 17, 2016. The remainder of the case presents data up 24. to June 2013. 25. https://www.nwsc.co.ug/index.php/resources/reports 26. https://dhsprogram.com/pubs/pdf/FR264/FR264.pdf 27. Data on the expansion are not complete, neither are the underlying performance data. Hence, these two years of still incomplete data have not been included in this assessment. 28. NWSC Corporate Plan, 2015–18. 29. NWSC Annual Report, 2013. 30. Figure quoted in 2011 U.S. dollars. 31. Performance contract between the government of the Republic of Uganda and the NWSC dated October 17, 2003. References Fall, Matar, Philippe Marin, Alain Locussol, and Richard Verspyck. 2009. “Reforming Urban Water Utilities in Western and Central Africa: Experiences with Public-Private Partnerships: Volume 2, Case Studies.” Water Sector Board Discussion Paper /connect/5202b8804ba99b958e16ef1be6561834/ Series 13, World Bank, Washington, DC. http://www.ifc.org/wps/wcm​ WaterPPPvol2.pdf?MOD=AJPERES. Kariuki, Mukami, Guillaume Patricot, Rosemary Rop, Sam Mutono, and Midori Makino. 2014. Do Pro-Poor Policies Increase Water Coverage? An Analysis of Service Delivery in Kampala’s Informal Settlements. Washington, DC: World Bank, Water and Sanitation Program, and Water Partnership Program. Marin, Philippe, Matar Fall, and Harouna Ouibiga. 2010. “Corporatizing a Water Utility: A Successful Case Using a Performance Based Service Contract for ONEA in Burkina Faso.” Gridlines Note No. 53. Washington, DC: PPIAF, World Bank. Marin, Philippe, Eustache Ouayoro, Matar Fall, and Richard Verspyck. 2009. Partnering for Water in Côte d’Ivoire: Lessons from 50 Years of Successful Private Operation. Gridlines Note No. 50 (August). Washington, DC: PPIAF, World Bank. https://open- knowledge.worldbank.org/handle/10986/10529. Mugisha, Silver, Sanford V. Berg, and William T. Muhairwe. 2007. “Using Internal Incentive Contracts to Improve Water Utility Performance: The Case of Uganda’s NWSC.” Water Policy 9 (3): 271–84. Mugo, F. K. 2006. “Nairobi City Water and Sewerage Company Limited: Milestones and Challenges.” http://www.un.org/esa/ sustdev/sdissues/water/workshop_africa/presentations/nwsc.pdf. NCWSC (Nairobi City Water and Sewerage Company). 2014. Strategic Plan 2014/15–2018/19. Nairobi, Kenya: NCWSC. Ndaw, Mouhamed Fadel. 2015. “Unlocking the Potential of Information Communications Technology to Improve Water and Sanitation Services: Summary of Findings and Recommendations.” World Bank, Washington, DC. Sawadogo, Diuedonné. 2015. “Delivering City-Wide WASH Services: Reaching Informal Settlements in Ouagadougou, Burkina Faso.” PowerPoint presentation delivered at World Water Week. SIWI, Stockholm, 2016. SODECI Rapport de Geston 2011, published in 2012. https://docs.google.com/viewer?url=http://www.sodeci.ci/application/ themes/exquiso/rapports/resultats-annuels-SODECI-2011.pdf United Nations, Department of Economic and Social Affairs, Population Division. 2014. World Urbanization Prospects: The 2014 Revision. New York: United Nations. WASREB (Water Services Regulatory Board). 2015. A Performance Review of Kenya’s Water Services Sector 2013–2014. Impact Issue No. 8/2015. Nairobi, Kenya: WASREB. Performance of Water Utilities in Africa 147 Appendix C Data Quality Procedures The quality of data in International Benchmarking Network for Water and Sanitation Utilities (IBNET) depends on the quality of the submissions by utilities and their associations. IBNET, however, invests substantial efforts to ensure that the data collected are of top quality and adequately reflect the status of the utility performance. IBNET data in Africa are gained from the following sources: (a) Bank operations in Francophone Africa and Malawi; (b) regulators from Tanzania, Kenya, and Zambia; and (c) individual consultants working under the supervision of the IBNET and Bank teams from Nigeria, South Africa, and Ethiopia. The quality of data sources varied from excellent quality assurance procedures (such as regulatory data) to some with less sound procedures; how- ever, all of the data have gone through a rigorous review by the IBNET team. Data Quality When Collecting Data. The IBNET data collection tool contains ranges and built-in filters that prevent the input of obviously wrong information. More than 70 filters are set in the IBNET Toolkit that prevent input of wrong as well as non-numeric data. Data Quality When Uploading the Data. The IBNET site controls data when uploading the data: • Data are within the expected ranges. • Time trends appear to be reasonable (red flag if data or indicators changed more than 30 percent within one year, yellow flag when data changed by 10 percent to 30 percent, and green flag when changes are below 10 percent). All yellow- and red-flagged data were sent to data collectors for review and explanation. • Confidence ratings assigned are as may be expected from experience (urbanization, length of pipes, consumption, and collection rates). After the dataset is uploaded, it undergoes a review for each of the data items on outliers, data sources, and consistency through the “performers” function in the IBNET database. The IBNET website allows examination of the calculated performance levels provided by all the utilities for consistency, to ensure that data are within the ranges to be expected and time trends appear to be reasonable. The calculated averages for the given set of data help understand the utilities’ outliers, and these performance outliers are reviewed jointly with the data collector to understand the reason for these outliers. We have a full data sample of more than 1,400 observations in the database, covering 306 utilities from 40 African countries. Observations span a 20-year period (from 1995 to 2014) but not all utilities have reported information each year. There were few utilities which pro- vided information at the beginning of the period (less than 15 utilities each year from 1995 to 1999). The number of utilities in the IBNET database has increased regularly over time. In 2011 and 2012, more than 250 utilities provided information each year, because of large data collection efforts in Ethiopia and Nigeria. Because participation in the IBNET data collection Performance of Water Utilities in Africa 149 is voluntary, the sample of utilities may not be representative of the entire population of African utilities operating in the water sector. A subsample of almost 120 utilities from 14 countries covering the period 2010–13 was used for the analyses in chapters 3, 4, 5, and 6. In chapter 7, the case studies use datasets from longer time periods for the five selected utilities. 150 Performance of Water Utilities in Africa Appendix D Data Envelopment Analysis Methodology The data envelopment analysis (DEA) is a nonparametric approach to measuring the relative efficiency of firms in an industry where the firms are often referred to in the DEA literature as decision-making units. The DEA creates a performance index from indicators, referred to as inputs and outputs in the DEA literature, which can be related to other factors that drive performance. Under basic DEA, a water utility is regarded as a relatively efficient utility if its observed inputs can be scaled to yield outputs that equal or exceed any combination or scaling of what the other utilities’ observed inputs yield. The origination of the approach is frequently attributed to Charnes, Cooper, and Rhodes (1978) and has been applied in many studies, including studies of water utilities. Extensions of the basic DEA approach to accom- modate alternatives to its assumptions have been presented in the decades since the appear- ance of the seminal paper. The DEA approach uses mathematical programming methods to determine the perfor- mance ranking of firms in an industry. The DEA approach measures the efficiency of firm k in an industry of K firms as the optimal value of the objective function found by solving the following problem where yik is the amount of the ith of m outputs produced by firm k, xjk  is the amount of the jth of n inputs used by firm k, and the maximization is over the m + n nonnegative, choice variables, ui, and vj: m Maximize ∑ i =1 ui yik ,  n ∑ j =1 vi x jk m subject to ∑ i =1 ui yik ≤ 1;k = 1,2,...,K, (D.1) n ∑ j =1 vi x jk where ui, vj ≥ 0; i = 1, 2, … , m; j = 1, 2, … , n. Note that the output values, ui, and input prices, vj, are evaluated to place firm k in the best light to determine its relative performance. A transformation of variables yields an equivalent model which may be more conveniently solved through the techniques of linear programming (see for example, Alhabeeb and Moffitt 2012); that is, m Maximize ∑ i =1 ui yik ,  m n subject to ∑ i =1 ui yik − ∑ j =1 vi x jk ≤ 0, (D.2) n ∑ j =1 vi x jk = 1, ui , v j ≥ 0. Performance of Water Utilities in Africa 151 The efficiency of each of the K firms is estimated by varying k in the objective function over k = 1, 2, … , K. Collecting the optimal value of the objective function for each value of k yields a perfor- mance index known in the DEA literature as an efficiency ranking of the firms over the [0, 1] interval with a “1” signifying an efficient firm. A relative efficiency frontier can be formed by piecewise linear segments associated with observations from the efficient firms. An important advantage of the DEA relative to statistical regression techniques is that it does not require specification of a functional form relating outputs and inputs as parametric statistical techniques do. Similarly, specification of distributions for stochastic model com- ponents is avoided. An important limitation of the DEA is its sensitivity to errors in sample observations, the potential of which is ignored by the basic DEA approach. The subjectivity involved in selection of inputs and outputs is also a limitation. References Alhabeeb, M. J., and L. Joe Moffitt. 2012. Managerial Economics: A Mathematical Approach. New York: John Wiley and Sons. Charnes, A., W. W. Cooper, and E. Rhodes. 1978. “Measuring the Efficiency of Decision Making Units.” European Journal of Operational Research 2: 429–44. 152 Performance of Water Utilities in Africa SKUWXXXXX SKU W16010