Leveraging Private Sector Investment in Digital Communications Infrastructure in Eastern Africa World Bank Group July 2024 © 2024 The World Bank Group 1818 H Street NW, Washington, DC 20433 Telephone: +1 202-473-1000; Internet: www.worldbankgroup.org Some rights reserved 1 2 3 4 21 20 19 18 The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the World Bank Group, its Board of Executive Directors, or the governments they represent. The World Bank Group 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. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of the World Bank Group, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: World Bank and CEPA. 2024. Leveraging Private Sector Investment in Digital Communications Infrastructure in Eastern Africa. Washington, DC: World Bank. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by the World Bank Group and should not be considered an official World Bank Group translation. The World Bank Group shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by the World Bank Group. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by the World Bank Group. Third-party content—The World Bank Group does not necessarily own each component of the content contained within the work. The World Bank Group therefore does not warrant that the use of any third party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org 2 Leveraging Private Sector Investment in Digital Communications Infrastructure in Eastern Africa World Bank July 2024 3 Acknowledgments T his report has been prepared by Chris Doyle, Jack Kemp and Chris Sutcliffe of CEPA Ltd and associate Professor Thomas Senaji, Information Communication Technology We wish to thank Xavier Decoster (Senior Digital Development Specialist, IDD02), Thomas Chalumeau (Senior Digital Development Specialist, IDD09), Erik Consultant, United Nations Economic Commission, Africa Whitlock (Partner, Salience), and Ai Yamakami (Associate (all collectively referred to as ‘CEPA’). Investment Officer, IFC) for their insightful comments which have been incorporated into this revised version. We wish to CEPA was commissioned as consultant for the World Bank’s thank also Martha Oringo for her work in page composition Digital Development Global Practice, under the supervision of the final report. of Tim Kelly (Lead Digital Development Specialist), Aneliya Muller (Senior Digital Development Specialist) and The information contained in this document has been Evalyn Oloo (Digital Development Consultant). Funding compiled by CEPA and may include material from third for this study was generously provided by Public Private parties which is believed to be reliable but has not been Infrastructure Advisory Facility (PPIAF) under a Trust Fund independently verified or audited. No representation or to support the Kenya Digital Economy Acceleration Project warranty, express or implied, is given and no responsibility (KDEAP; P170941), the Eastern Africa Regional Digital or liability is or will be accepted by the World Bank, or by Integration Series of Projects (EARDIP, P176181; P180931), or on behalf of CEPA or by any of its directors, members, and the Digital and Energy Connectivity for Inclusion in employees, agents or any other person as to the accuracy, Madagascar project (DECIM, P178701). completeness or correctness of the material from third parties contained in this document and any such liability is The report was prepared for the purpose of undertaking expressly excluded. a study on leveraging private sector investment in Digital Communications Infrastructure (DCI) in Eastern Africa. The findings enclosed in this document may contain Specifically, the study focuses on leveraging private sector predictions based on current data and historical trends. investment in the Eastern African countries of: Djibouti, Any such predictions are subject to inherent risks and Ethiopia, Kenya, Madagascar, Somalia and South Sudan. uncertainties. An advanced draft of the report was discussed during consultations with Government officials in Kenya on The opinions expressed in this document are valid only for February 29, 2024 and with the other countries on March the purpose stated herein and as of the date stated. No 21, 2024. The final report was subject to a World Bank obligation is assumed to revise this document to reflect review, conducted virtually in June 2024, in a decision changes, events or conditions, which occur subsequent to meeting chaired by Isabel Neto, Practice Manager, Digital the date hereof. Development, Eastern and Southern Africa. 4 Contents Acknowledgments 4 Figures, tables and boxes 6 Abbreviations and acronyms 7 1. Introduction 10 1.1. Digital communications infrastructure and digital gaps 11 1.2. The roles of the private sector, government and multilateral development banks 14 1.3. Private capital mobilisation 14 1.4. Public-private partnerships, procurement and digital communications infrastructure 15 1.5. Risks, anchor tenants and price discovery 16 1.6. Structure of the report 17 2. Funding Digital Communications Infrastructure In Eastern Africa 18 2.1. Economic characteristics of digital communications infrastructure 18 2.2. Government options and funding gaps 19 2.3. Financing connectivity infrastructure 23 2.4. Conclusions 24 3. Facilitating private capital mobilisation in digital communications infrastructure 25 3.1. Current constraints and risks 25 3.2. Role of national governments 27 3.3. Role of development institutions 27 3.4. Conclusions 28 4. Approaches to incentivize private sector investment in digital 29 communications infrastructure 4.1. Subsidy methods 30 4.2. Government agencies as anchor tenants 32 4.3. Key considerations 37 5. Achieving value for money 38 5.1. Procurement and competition 38 5.2. Reverse auctions 39 5.3. Conclusion 48 6. Policy lessons and recommendations 49 6.1. Key recommendations 49 Appendix A Eastern Africa and country backgrounds 52 Appendix B Project backgrounds 57 Appendix C What is Digital Communications Infrastructure? 60 Appendix D Procurement rules 62 Appendix E World Bank project Task Team Leader (TTL) Interviews 64 Appendix F Proposed commercial transaction roadmap 72 5 Figures Figure 1‑1 The Gaps Model 12 Figure 1‑2 Digital connectivity in Africa 13 Figure 2‑1 Intermediate funder carries foreign exchange risk 21 Figure 2‑2 Sources of finance for infrastructure projects with private participation in 24 emerging and developing economies Figure 4‑1 Public sector anchor tenant model 33 Figure 4‑2 Demand and cost for communication services 34 Figure 4‑3 Anchor tenant effect 34 Figure 4‑4 Benefits of an anchor tenant 34 Figure 5‑1 Evaluation of Reverse Auction Formats 50 Figure 6‑1 The decision process for implementing a MRRA for the use of 62 dispersing subsidy funds for investment Tables Table 1‑1: Percentage of population using the internet 11 Table 1‑2: Financing digital infrastructure and access in KDEAP 15 Table 2‑1: Financial instruments for DCI 23 Table 3‑1: Perceived investment risks in the six countries in this study, CEPA estimates 26 Table 4‑1: Approaches to incentivise private sector investment in DCI 29 Table 5‑1: Bidder estimates of funding required to earn a normal return 48 Table 6‑1: Recommendation summary table for each of the six Eastern African countries 49 Boxes Box 1: Niger Smart Villages Project (P167543) 30 Box 2: Smartphone subsidies in Rwanda 31 Box 3: Demand-side Subsidy - Gigabit Broadband Voucher Scheme (UK) 31 Box 4: Anchor tenant and direct subsidy diagrams 34 Box 5: Digital Malawi Program Phase 1: Digital Foundations Project 35 Box 6: Kosovo Digital Economy (KODE) project 41 Box 7: Static Reverse Auction Format: Illustrative Example 42 Box 8: Price Discovery and Multi-Round Auctions: Illustrative Example 46 Box 9: FCC Auction 904 – MRRA clock format 47 6 Abbreviations and Acronyms 2G, 3G, 4G, 5G Second, Third, Fourth, Fifth Generation mobile telephony service AfDB African Development Bank ARRA American Recovery and Reinvestment Act of 2009 BAFO Best and Final Offer BCG Boston Consulting Group BOT Build-Operate-Transfer BTO Build-Transfer-Operate CA Communication Authority (Kenya) CAI Community Anchor Institutions CAPEX Capital Expenditure, and OPEX – Operational Expenditure. CTM Commercial Transaction Manual DARE Djibouti Africa Regional Express DCI Digital Communications Infrastructure DE4A Digital Economy for Africa Initiative DECIM Digital and Energy Connectivity for Inclusion in Madagascar Project DeFi Decentralized Finance DPI Digital Public Infrastructure EAP Electronic Auction Platform EARDIP Eastern Africa Regional Digital Integration Project ECA United Nations Economic Commission for Africa E-Commerce Electronic Commerce EPC Engineering, Procurement, Construction FCC Federal Communications Commission (United States) FDI Foreign Direct Investment FGS Federal Government of Somalia GDP Gross Domestic Product GSMA Global System for Mobile Communications Association ICA The Infrastructure Consortium of Africa ICT Information and Communication Technology IDA International Development Association IFIs International Financial Institutions IFC International Finance Corporation IGAD Intergovernmental Authority on Development IP Internet Protocol IRU Indefeasible Right of Use ITU International Telecommunication Union KDEAP Kenya Digital Economy Acceleration Project 7 LICs Low Income Countries MAS Maximum Allowable Subsidy Mbit/s Megabits per second MDBs Multilateral Development Banks MIGA Multilateral Investment Guarantee Agency MNO Mobile Network Operator MoU Memorandum of Understanding MPPRA Madagascar Public Procurement Regulatory Authority MRRA Multi-Round Reverse Auction MVNO Mobile Virtual Network Operator NCA National Communications Authority of Somalia NICTBB National ICT Broadband Backbone NPCI National Payments Corporation of India PCE Private Capital Enabling PCM Private Capital Mobilisation PPA Public Procurement Act, 2011 PPIAF Public-Private Infrastructure Advisory Facility PPP Public-Private Partnership PPPC Public-Private Partnership Commission (Malawi) PPRA Public Procurement Regulatory Authority (Kenya) PRI Political Risk Insurance RAG Red Amber Green RCIP Regional Communications Infrastructure Programme RFQ Request For Quotation SDGs Sustainable Development Goals SIM Subscriber Identity Module SOE State Owned Enterprise SSA Sub-Saharan Africa TAPAS Transparency and Accountability in Public Administration and Services Project (UK) TTL Team Task Leader UAT User Acceptance Testing UCF Unguaranteed Commercial Financing UCSAF Universal Communications Service Access Fund (Tanzania) UN United Nations UPI Unified Payments Interface USAFs Universal Service and Access Funds USD United States Dollar USF Universal Service Fund WB World Bank 8 CHAPTERS 9 Introduction 1 This study provides guidance on World Bank involvement A summary of selected World Bank funded DCI proj- in Digital Communications Infrastructure (DCI)1 projects in ects in Eastern Africa. which finance has been extended to support build-out.2 In An overview of the role of the private sector, Multilat- particular, the study provides guidance to three World Bank eral Development Banks (MDBs) and governments in projects: facilitating private sector investment in DCI. A discussion of various approaches that can be used to The Kenya Digital Economy Acceleration Project incentivise and facilitate Private Capital Mobilisation (KDEAP; P170941), which aims to expand access to (PCM) in DCI. In particular, we look at the potential for high-speed internet, improve the quality and delivery of governments and their agencies to aggregate demand education and selected government services, and build for DCI services and functions and act as anchor ten- skills for the regional digital economy..3 ants. The potential for using novel procurement methods, The Eastern Africa Regional Digital Integration Project such as Multiple Round Reverse Auctions (MRRA), (EARDIP Series of Project 1; P176181, and Series of to promote value for money and an assessment of Projects 2; P180931), which has the objective of pro- the “readiness” of the six Eastern African countries in moting the expansion of an integrated digital market adopting such an approach. across Eastern Africa by increasing cross-border broad- band connectivity, data flows and digital trade in the re- The report also provides a roadmap on what should be gion.4 The first phase covers Somalia and South Sudan included in a Commercial Transaction Manual (CTM) in the and grants to the East African Community (EAC) and the case where we recommend that a borrower government Intergovernmental Authority on Development (IGAD); pilots a MRRA. The preparation of the CTM is a disbursement while the second phase covers Djibouti and Ethiopia.5 condition covering infrastructure investments in each of the three projects and this report is intended to reform the The Digital and Energy Connectivity for Inclusion in preparation of the CTMs in each of the six countries. Madagascar Project (DECIM; P178701), which aims to expand access to renewable energy and digital services The rest of this chapter is as follows. Section 1.1 introduces in Madagascar.6 the concept of DCI and discusses policy within the context of digital gaps. Section 1.2 discusses the roles of the private The study has been conducted under the supervision of the sector, government and multilateral banks in Eastern Africa. World Bank (WB) co-Task-Team Leaders for the projects Section 1.3 looks at the role expected to be played by PCM. listed in the acknowledgements. The report provides a list of Section 1.4 discusses in brief the role of Public-Private policy recommendations with regard to leveraging private Partnerships (PPPs), procurement and DCI. Section 1.5 sector investment in DCI in the six Eastern African countries discusses risks in DCI projects and their mitigation, focusing analysed: Djibouti, Ethiopia, Kenya, Madagascar, Somalia in particular on the role of government agencies as anchor and South Sudan. The content of the report covers, among tenants and in overseeing MRRAs. Our conclusions and other things: Section 1.6 highlights the remainder of this report. 1 The phrase Digital Communications Infrastructure is used here to denote digital technologies used for connectivity, such as fiber optic cables, cellular mobile services, microwave systems, satellites, etc. They are used by the public to communicate by voice, video, data, etc. Digital Communications Infrastructure may be regarded as a subset of the term “Digital Public Infrastructure” which encompasses also data exchange platforms, digital ID systems, digital financial payment systems etc, as well as connectivity. A more detailed explanation is provided in Appendix C 2 Throughout this study we use the convention that the financing of infrastructure is how governments, private companies and others that own infrastructure find the money to meet the upfront costs of building it. In contrast, the funding of infrastructure is how consumers, taxpayers and others pay for the infrastructure, including pay- ing back the finance from whichever source government or private owners choose. In some instances, private firms may face a funding gap. In this case government, and in the case of Eastern African countries multilateral development banks, can intervene and provide finance to cover the funding gap. Our report is concerned with achieving value for money in funding gap situations. 3 See https://projects.worldbank.org/en/projects-operations/project-detail/P170941 4 See https://projects.worldbank.org/en/projects-operations/project-detail/P176181 5 See https://projects.worldbank.org/en/projects-operations/project-detail/P180931 6 This project is interesting as it combines initiatives in both electricity and power supply markets and broadband communications. The two are interrelated, as new mobile services and mobile broadband require reliable power supplies. In the project, it is envisaged that an Engineering, Procurement and Construction (EPC) model may be deployed. This may involve system integrators designing, building, operating and handing over assets such as solar panels and battery packs to schools and health centres. See https://projects.worldbank.org/en/projects-operations/project-detail/P178701 10 1.1 Digital communications The figures in Table 1.1 reflect gaps formally described by Navas-Sabater, Dymond and Juntunen (2002) in their study infrastructure and digital gaps on universal access.10 The authors proposed a ‘Gaps Model’, shown in Figure 1‑1, where the y-axis is the demand side As more citizens across the globe use the internet, for a service (households ordered by income from high to governments and their agencies can deliver better public low) and the x-axis is the supply side (geography). The top services, improve accountability and promote inclusivity.7 border represents 100% of communities (i.e., geographic DCI forms a crucial part of this and includes the solutions population centres) within the region or country and the and systems that facilitate essential society-wide functions right border represents 100% of the population, typically and services such as: identity verification, payments and expressed as households. data exchange (we explore DCI in further detail in Appendix C). To achieve these benefits, many governments in low- and The model proposes two main gaps in service provision: middle-income countries, including in Eastern Africa, are (i) a market efficiency gap and (ii) an access gap, explained currently promoting investments in building out the physical further below. In Figure 1-1 current penetration is shown infrastructure and virtual architectures that underpin DCI as relatively low, as for most of the countries in this study, functions and services. and typically reflects that demand is mainly coming from higher income households (vertical axis) and supply is Most citizens in Eastern Africa gain access to the internet focused mainly in lower cost service areas (horizontal axis). through mobile devices. However, the proportion of people Notwithstanding, since the Gaps Model was originally connected to mobile broadband in Sub-Saharan Africa proposed, industry has proved adept at extending networks (SSA), at around 25 per cent in 2022, was less than half and services to low-income households in high-cost service the global average of 57 per cent.8 Table 1‑1 shows that as areas, often working with development agencies and recent as 2021 only Djibouti exceeded the global average government, as the penetration rates for mobile service in of the percentage of the population using the internet, all the countries of this study underscore.11 for the other five counties in this study the majority of the population did not use the internet. Table 1‑1 Percentage of population using the internet Year 2021 2021 2021 2021 2021 2021 est 2020 Country Global Djibouti Ethiopia Kenya Madagascar Somalia South Sudan 63% 69% 17% 29% 20% 6% 7% Percentage of population using the internet Source: ITU9 7 Milakovich, Michael. (2021). Digital Governance: Applying Advanced Technologies to Improve Public Service. 10.4324/9781003215875 8 GSMA (2023) “The State of Mobile Internet Connectivity: Key Findings 2023”. 9 Internet users are individuals who have used the internet (from any location) in the last 3 months. The internet can be used via a computer, mobile phone, personal digital assistant, games machine, digital TV etc. Source: International Telecommunication Union ( ITU ) World Telecommunication/ICT Indicators Database at https://data.worldbank.org/indicator/IT.NET.USER.ZS 10 Juan Navas-Sabater, Andrew Dymond and Niina Juntunen (2002) “Telecommunications and Information Services for the Poor: Toward a Strategy for Universal Access” World Bank. 11 According to the GSMA, the penetration rate of mobile services in Sub-Saharan Africa for 2023 was 43%, see GSMA (2023) “The Mobile Economy Sub-Saharan Africa 2023”. 11 The market efficiency gap is shown as representing the Although there are substantial DCI gaps, most of the large geographic reach and coverage that is commercially feasible cities in SSA have good communications connectivity in a fully-functioning market setting. The gap reflects various and access to the internet has improved as a result of obstacles in the market that have not enabled this gap to increased submarine cable connectivity to the African be fully closed. The ‘smart subsidy zone’ shows how both continent, deployment of new fibre backbone networks coverage and reach can be extended beyond the market and competition.14 However, there remain significant gaps, frontier after the application of one-time subsidies to assist particularly outside the larger cities where private operators start-up and cover initial network investment.12 Finally, the are less incentivised to rollout communication networks figure shows what is termed the ‘true access gap’ which due to the higher cost of network build and lower demand. refers to areas and users that require ongoing support. It is these areas where there are access funding gaps and considerable disparities in coverage, as evidenced in GSMA Figure 1‑1 The Gaps Model (2023) and by a World Bank (2023) report (see Figure 1‑2 below).15.16 100% Household (Universal Service) True Key findings from the World Bank (2023) report include: access Internet access remains low. On average, only 28.5 Low income households gap percent of individuals live in households with access Smart to internet. Among the countries surveyed, Gabon has subsidy zone the highest rate of internet access, with an estimated access rate of 64.9  percent of the population, while Zambia has the lowest rate of internet access, with only Market an estimated 6.9 percent share of the population having efficiency access. gap Access to computers and tablets is very low. On average, only around 11 percent of individuals live in households with access to a computer. Mauritius and Cabo Verde have the highest rates of computer or tablet access, with High income households 58 percent and 45 percent respectively. Meanwhile, the Central African Republic and the Democratic Republic 100% Geographical Coverage Current of Congo have the lowest rates of computer or tablet network reach access, at 0.6 percent and 1.4 percent respectively. On & access average, only 11.2 percent of the population lives in a With household with access to a computer or tablet. one time Requires Commercially subsidy, ongoing Access to mobile phones is relatively high compared to feasible reach feasible support other internet-capable devices. On average 76 percent of individuals have access to a phone. Senegal and the Arab Republic of Egypt have the highest rates of phone access, with 99.1 percent and 97.9 percent respectively. Geographical reach Source: Reproduced from World Bank and ITU. 2011. Telecommunications Electricity access varies significantly. On average, Regulation Handbook, Tenth Anniversary Edition p.15713 58.8 percent of individuals surveyed have access to electricity. However, this hides large differences between countries. The Central African Republic has the lowest electricity access rate (9.1 percent); the Seychelles has the highest electricity access rate (100 percent). 13 World Bank and ITU. 2011. Telecommunications Regulation Handbook, Tenth Anniversary Edition (2011) edited by Colin Blackman and Lara Srivastava,. 14 Simione and Li. 2021. “The Macroeconomic Impacts of Digitization in Sub-Saharan Africa: Evidence from Submarine Cables” IMF Working Paper, WP/21/110. Available at: https://www.imf.org/-/media/Files/Publications/WP/2021/English/wpiea2021110-print-pdf.ashX. Figures 1 & 2 in their paper show a dramatic increase in cable connectivity, especially since 2010, and it is estimated that this has resulted in a statistically significant effect on productivity and growth. 15 See, https://www.gsma.com/r/wp-content/uploads/2023/10/The-State-of-Mobile-Internet-Connectivity-Report-2023.pdf 16 World Bank. 2023. “The Size and Distribution of Digital Connectivity Gaps in Sub-Saharan Africa.” Available at: https://documents1.worldbank.org/curated/ en/099241003142325200/pdf/IDU0cb2e42f3050260484d0b8370b84eee303ecf.pdf 12 Figure 1‑2 Digital connectivity in Africa Source: World Bank (2023) To facilitate wider adoption of DCI in Eastern Africa, For example, DCI investments can promote inclusive policies focus on both the demand and supply sides of the economic growth (for example Kouladoum (2023)) and market. On the demand-side, government interventions strengthen the economic resilience of households (for can include digital literacy training, favourable taxation example Chetty (2023)).17 Studies also show that increasing and subsidies to end-users. Government interventions on internet availability increases jobs and reduces poverty.18,19 the supply-side include market liberalisation, favourable Finally, additional DCI investments can also help bridge the corporate tax policies and import duties, and subsidies. digital gender divide.20 More generally, DCI is recognised Further, as DCI enables new digital government services, as having substantial potential to promote many positive this provides higher demand for digital communication externalities such as improvements in health care (e.g. services and facilitates scale economies, which lowers unit through disseminating knowledge), education (e.g. access to costs. Aggregating government agency demand for DCI resources), etc..21 services can help close the usage gap by making services more affordable. While the many benefits justify public expenditures on DCI, there is a need to ensure that policies deliver value DCI investments may involve, for instance, the rolling out for money. In this study we emphasise the importance of of more datacentres, land-based fibre-optic broadband designing policy interventions that ensure value for money cables and telecom towers, which help to close digital is achieved. gaps and confer benefits on consumers and producers. 17 Jean‑Claude Kouladoum (2023) “Digital infrastructural development and inclusive growth in Sub‑Saharan Africa” Journal of Social and Economic Development 25: 403–427. Krish Chetty (2023) “Strengthening Africa’s Digital Infrastructure for Greater Economic Resilience” South African Institute of International Affairs, Policy Briefing number 277, August. See also https://www.undp.org/digital/blog/human-and-economic-impact-digital-public-infrastructure#:~:text=Using%20a%20DCI%20 approach%20can,the%20economic%20resilience%20of%20households 18 Begazo, Blimpo and Dutz (2023) “Digital Africa: Technological Transformation for Jobs” World Bank Group. 19 Ndubuisi, G., Otioma, C. and Tetteh, G. K. (2021). “Digital infrastructure and employment in services: Evidence from Sub-Saharan African countries.” Telecommunications Policy, 45(8): 102153. 20 A study conducted in Nigeria shows that after a year or more of mobile broadband coverage the total consumption in households increased by more 6 percent and that two or three years of 3G/4G coverage increased labour force participation for women (See Bahia, K., Castells, P., Cruz, G., Masaki, T., Pedros, X., Pfutze, T., Winkler, H. (2020)). The Welfare Effects of Mobile Broadband Internet. Evidence from Nigeria. Washington DC: World Bank Group. See also OECD (2018) “Bridging the Digital Gender Divide: Include, Upskill, Innovate”. 21 DCI is an important enabler of accelerating sustainable development goals, see UNDP (2023) “Accelerating the SDGs through Digital Public Infrastructure: A Compendi- um of the potential of digital public infrastructure”. 13 1.2 The roles of the private sector, The World Bank, in particular, has played a leading role in this regard, having invested a total of US$582 million in government and multilateral ICT projects in Eastern and Southern Africa in 2022, which development banks amounted to 3 per cent of its total lending in the region.26 In 2024, the World Bank has committed a further US$2.5 The bridging of digital gaps and the promotion of DCI in billion in lending in the region up to 2032 under the Eastern Africa involve all major stakeholders: the private Inclusive Digitalization in Eastern and Southern Africa sector (especially communication service and network Multiphase Programmatic Approach Project (IDEA MPA). companies and equipment manufacturers), government Such investments include: mobile infrastructure in unserved (national, regional and local) and MDBs. To date, the and underserved areas, building out and extending capacity majority of investment in DCI in SSA has been undertaken in fibre-optic backbone networks and programmes aimed by the private sector, with governments and MDBs typically at enhancing digital literacy. In promoting broadband extending some gap financing and in many cases overseeing investments, programmes have included loans to facilitate structural reforms and capacity building that have facilitated government subsidy of rural rollout and supply of internet much needed new competition.22 connectivity to schools, hospitals and other public resources.27 According to Lee and Gonzalez (2022), between 2007 and 2020 investment in telecoms SSA by the private sector Increasingly the intervention by MDBs is seeking to leverage was US$25.3 billion versus US$4.6 billion by the public more effectively funding or matching investments made by sector.23 The private sector accounted for 85 per cent of the private sector through PCM. Additionally, reforms and all investment in SSA over 2007-20, whereas in the energy public investments part-funded by the Bank perform an sector private sector investment was lower at 43 per cent important role in Private Capital Enabling (PCE).28 and in the transport sector lower still at 32 per cent. Much of the private investment in telecoms occurred in the earlier part of the period when new costly mobile networks were 1.3 Private capital mobilisation being built. According to ICA Africa “The most notable distinguishing feature of investment in ICT infrastructure in Projects that seek to improve the digital economy and DCI Africa is that…it requires little support, financial or otherwise, in Eastern Africa involve a considerable amount of private from national governments or IFIs [International Financial finance and in some instances has involved public funding. Institutions]. Almost all ICT infrastructure is financed by the Although to date much investment has been led by the private sector. Within the ICT sector there is government private sector, it seems that further progress in extending of support for offshore cables (notably in the form of guarantees) DCI coverage into high-cost service areas and low-income as well as fiber-optic data backbones. The amounts involved are demand areas will require more public funding. A key focus overshadowed by the mobile telecommunications expenditure of this report is to examine how best to achieve value for by the private sector.”24 money new public funding investments and how such interventions can best leverage PCM. Whilst private sector investment has led to marked improvements in telecommunications and DCI coverage The relative immaturity of capital markets, low saving and usage in Eastern Africa, there remain substantial gaps. It rates in SSA, and constrained fiscal positions of many is estimated by the World Bank that countries in SSA would governments mean that, to achieve marked reductions in need to invest between 2 and 8 per cent of GDP annually DCI gaps, a considerable fraction of new investment will until 2030 to help close these gaps.25 be required from external sources, including from MDBs. Njenga et al (2022) remark “As most SSA countries do not To stimulate further investment and reap the benefits of new have well-established and liquid capital markets, external debt DCI in SSA, there is a need for intervention by government, has become the only source of capital. External debt is driven MDBs and international financial institutions (IFIs) to by low revenues and high expenditure needs, particularly in address both institutional and market failures. infrastructure development.”29 22 Emmanauelle Auriol (2005) “Telecommunication Reforms in Developing Countries” Communications & Strategies, November. 23 Nancy Lee and Mauricio Cardenas Gonzalez. 2022. “Stuck Near Ten Billion: Public Private Infrastructure Finance in Sub-Saharan Africa.” CGD Policy Paper 251. Washing- ton, DC: Center for Global Development. https://www.cgdev.org/publication/stuck-near-ten-billion-public-private-infrastructure-finance-sub-saharan-africa 24 ICA. 2018 The Infrastructure Consortium for Africa “Infrastructure Financing Trends in Africa – 2018” page 66. 25 Rozenberg, Julie and Marianne Fay. 2019. “Beyond the Gap: How Countries Can Afford the Infrastructure They Need while Protecting the Planet” World Bank. 26 Begazo, Blimpo and Dutz. 2023. “Digital Africa: Technological Transformation for Jobs” World Bank Group. 27 For example, on 5 April 2023 the World Bank announced it was lending $390m to the Government of Kenya for the first phase of a program that aims to expand access to high-speed internet, improve the quality and delivery of education and selected government services, and build skills for the regional digital economy. This forms part of the KDEAP. See https://www.worldbank.org/en/news/press-release/2023/04/05/kenya-afe-and-the-world-bank-group-provide-a-390-million-boost-the-digital-econo- my#:~:text=NAIROBI%2C%20April%205%2C%202023%20%E2%80%94,for%20the%20regional%20digital%20economy. 28 See World Bank. 2022. “Evolving the World Bank Group’s Mission, Operations, and Resources: A Roadmap” 18 December especially para. 19. 29 Njenga, Githinji, Josphat Machagua, and Samwel Gachanja. 2022. “Capital markets in sub-Saharan Africa” WIDER Working Paper 2022/112. 14 This situation is likely to prevail in the near term and the The KDEAP PAD states “Among the many possible models for International Development Association (IDA) credit and combining public and private funds for network roll-out, one grant30 facilities will continue to feature in many new DCI of the models, for both the extension of the backbone and last investment projects.31 mile connectivity, would involve matching investments from network operators in return for project contracts awarded An example of a project which is anticipated to involve a through a reverse auction model.”33 This approach is also considerable amount of PCM is KDEAP in Kenya. This project advocated for PCM in the DECIM project in Madagascar, involves the IDA extending credit of US$390 million to the which “may potentially include, for instance, the use of an Government of Kenya over the first phase of the project, interactive electronic auction platform to optimize the best 2023-28, with a planned additional US$100 million to be value in competitive, multi-round bidding processes.”34 leveraged in PCM, or unguaranteed commercial financing (UCF), for projects targeting improvements in digital infrastructure and access through expansion of broadband 1.4 Public-private partnerships, coverage.32 The financing under KDEAP is highlighted in Table 1‑2. procurement and digital communications infrastructure Table 1‑2: Financing digital infrastructure and access in KDEAP The mobilisation of private capital in projects in receipt of IDA credit Potential gap financing for DCI brings together the public and private KDEAP Investment, sectors. Public sector gap financing incentivises participation (US$, PCM (US$, Component 1 million) million) by the private sector in projects that may involve small scale local initiatives (e.g. installing connectivity at a village school 1.1 Extending the reach of the 60 60 in Kenya in the KDEAP project, and installing mobile base- backbone network (middle mile) stations in rural areas in Madagascar in the DECIM project) 1.2 Increasing last mile through to large scale national and regional initiatives 90 30 connectivity for education (e.g. investing in additional national middle-mile backbone 1.3 Enhancing government fibre capacity in Ethiopia, Somalia and South Sudan and 35 0 connections to bordering countries in the EARDIP project). connectivity 1.4 Strengthening the digital 5 0 In some cases, arrangements between the public and private enabling environment sectors in infrastructure projects fall within the scope of R1.5 Enhancing regional digital 30 10 PPPs which are broadly defined as a “long-term contract infrastructure between a private party and a government entity, for Source: Para. 28, Table 1 Project Information Document, KDEAP op cit providing a public asset or service, in which the private party Note  : This table covers KDEAP Component 1 only. Adding funding for bears significant risk and management responsibility, and Components 2, 3 and 4 brings the total to US$390m plus US$100m in remuneration is linked to performance”.35 Typically a PPP Private Capital Mobilization (PCM). involves the delivery of project finance through a project- specific company established to deliver infrastructure. It is The WB anticipates that potential PCM will be leveraged the company that borrows the money and contracts transfer through credit facilities supplied by IDA. Ideally the PCM responsibility for matters such as design, build, operation induced by the WB gap funding (sometimes referred to and maintenance to the company in which the investors as matching investments) is expected to yield a ratio of have managerial responsibilities. There are a number of roughly 2:3 (i.e., US$2 of commercial investment for every PPPs in telecoms in Africa, an example is the high capacity US$3 of public funds). Achieving this ratio of private and backbone network part funded by the WB in the Democratic public funding is anticipated to be achieved by leveraging Republic of Congo.36 government agencies demand for digital services and by promoting greater competition for subsidies and public sector contracts through the use of novel procurement models, including MRRAs. 30 The IDA extend assistance to low-income countries via grants, which offer funds in return for results, or through loans (credit) that result in debt and require servicing and eventual redemption. Subsidy funding to support DCI can be provided through either channel, and for some projects it may involve a mix of grants and loans. 31 See the Digital Moonshot Report (2019) for how the World Bank Group will mobilize finance through the IDA under a Mobilizing Finance for Development approach to distribute $25 billion earmarked for the DE4A Initiative. 32 See Project Information Document – Kenya Digital Economy Acceleration Project – P170941 26 February 2023, at https://documents.worldbank.org/en/publication/ documents-reports/documentdetail/099094702272313420/p17094109fabfa0508c6204a93160f0598 33 Para. 31 in Project Information Document KDEAP. 34 Para. 48 in Project Information Document DECIM. Under DECIM, a similar model of “matching grants” is also proposed for power sector infrastructure in rural areas, such as mini-grids. 35 World Bank. 2014. “Public-Private Partnership Reference Guide Version 2.0”. There are many variants of PPPs, but the typical PPP arrangements used by the WB include BOO: Build, Own, Operate; BOT: Build, Operate, Transfer; BOOT: Build, Own, Operate, Transfer. PPPs can be structured in many different ways and one recent interest- ing example promoting DCI in Indonesia and the Philippines involves in-kind payments by private sector telcos expanding into negative net present value areas in return for access and use of national fiber backbone facilities – the Palapa ring model, see https://baktikominfo.id/layanan/palapa-ring. 36 See World Bank. 2012. “Public-Private Partnership in Telecommunications Infrastructure Projects: Case of the Republic of Congo.” Available at: https://documents1. worldbank.org/curated/en/677871468244179240/pdf/687020ESW0P1220cover0PO1223950Congo.pdf 15 PPPs have been researched extensively and continue to One way to mitigate some moral hazard risk in DCI projects be created but are not the primary focus of this study. Our is through the aggregation of demand for digital services by concern involves PCM in projects that have many sub- government agencies in the form of anchor tenants.40 For projects in which private sector parties may compete to example, government agencies could commit to purchasing supply matching investments. It is widely recognised (for substantial broadband capacity over the lifetime of a example, Bhattacharya et al. (2022)) that for Eastern Africa broadband connectivity project. In such a scenario, demand to boost economic growth and achieve the United Nations side risk faced by a private party would be lowered and (UN) Sustainable Development Goals (SDGs) there is a need private firms would be more willing to share more of the for greater leverage to be placed on the private sector.37 upfront costs of infrastructure investment as the funding gap would be lower. In projects involving last mile connectivity Where PCM is anticipated in projects in receipt of financing for education, as forms part of KDEAP, the aggregation from the IDA, the public and private sectors will be subject of demand for broadband capacity by public educational to any applicable national PPP and procurement legislation. institutions would help in this regard, particularly for For example, the projects specified in KDEAP are subject to regional projects. both national procurement legislation in Kenya.38 To ensure PCM is leveraged by public financing to greatest effect, it is When seeking PCM through public sector gap financing, necessary to understand the constraints arising from the public agencies are exposed to the risk of adverse selection. applicable legislative frameworks. Prospective private partners seeking public finance may overstate deliverables and offer terms in responses to tender that are unrealistic, in efforts to win the initial public finance. This risk can be problematic if private bidders know 1.5 Risks, anchor tenants and price they have the possibility to hold-up the public party and discovery renegotiate down the line.41 To help mitigate this risk, a project needs to be well-specified to ensure that sufficient bids are made by private entities and bids received are Infrastructure projects are risky because of their often meaningfully comparable (i.e., different bids are not based long term, large scale characteristics. Demand side revenue on varying underlying assumptions). streams and supply side infrastructure construction costs are not known for certain. Market risks can be managed by Some of the risks public finance faces when seeking PCM in contracts that allocate risks to incentivise participants to DCI projects arise because of asymmetric information about choose actions that result in desirable outcomes. Many PPPs the funding gap. For example, government is not usually in a are structured in ways that allocate risks to help mitigate the good position to know how much gap funding is required. An costs associated with uncertainty. important challenge for the public sector entities providing finance is how best to mitigate paying too much (i.e. the However, while contracts can target risks presented adverse selection risk) and achieve value for money. in market situations, behavioural risks can arise due to asymmetric information within the contractual setting, such as moral hazard.39 For example, where the private sector co-finances infrastructure in conjunction with a public subsidy, a well-designed contract will likely encourage the private party to deliver services efficiently at lowest cost. By contrast, where the private sector operates infrastructure services financed entirely by the public sector, it might be less incentivised to minimise costs if it perceives losses will be made up by the public sector. 37 The authors state that “there is significant potential to leverage official finance with double the amount of private finance. Use of such instruments could lead to a dou- bling of private finance compared with 2019 – an additional $395 billion”, see p. 46 in : Bhattacharya A et al. 2022. Financing a big investment push in emerging markets and developing economies for sustainable, resilient and inclusive recovery and growth. London: Grantham Research Institute on Climate Change and the Environment, London School of Economics and Political Science, and Washington, DC: Brookings Institution. 38 In Kenya the relevant PPP legislation is The Public Private Partnerships Act, 2021. Part V of the Act specifies the types of procurement that are feasible. The Government of Kenya issued a policy on procurement “National Public Procurement and Asset Disposal Policy”, in November 2020 which is in accordance with The Public Procure- ment and Asset Disposal Act, 2015 (revised 2022). 39 Moral hazard is a situation where an economic entity has an incentive to increase its exposure to risk because it does not bear the full costs of that risk. In practice PPPs are prone to moral hazard risks, often exemplified by post-bid negotiations. For example, see Gifford et al. (2014) who discuss PPPs in the United States where contract renegotiation may dampen the market and adversely affect national infrastructure investment efforts. See https://ppp.worldbank.org/public-private-partnership/library/ renegotiation-transportation-public-private-partnerships. In the United Kingdom a government review found that negotiations at the preferred bidder stage led to price increases in many projects, see HM Treasury (2014). “Investing in UK Infrastructure.” Available at: https://assets.publishing.service.gov.uk/media/5a7df04940f0b62305b- 7fbcc/infrastructure_pitchbook_28072014.pdf. A WB study has also found that the nature of the long-term contracts between participants that constitute PPPs are at risk of re-negotiation and cited four causes behind this: (i) unexpected exogenous changes, (ii) complexity of the contractual relationship, (iii) winner’s curse and (iv) rent seeking behaviour. See World Bank (2014) “Renegotiation of Transportation Public-Private Partnerships”, 40 The aggregation of demand is also known as demand pooling. This has been studied recently in the context of pooling demand by schools in Africa for Internet related services, see Deloitte (2024) “Market assessment: Connectivity solutions for schools in Eastern, Western and Southern Africa”, Project Giga, ITU and UNICEF, April. 41 This risk of opportunism can be reversed in telecoms, as private investors seeking to partner with government agencies to roll-out next generation networks (NGNs) might limit their investment fearing opportunism by a future different government, see : Howell, Bronwyn and Sadowski, Bert. 2014. “Anatomy of a Public-Private Partnership: Hold-up and regulatory risk in an NGN PPP”, 20th Biennial Conference of the International Telecommunications Society (ITS): "The Net and the Internet – Emerging Markets and Policies" , Rio de Janeiro, Brazil, 30th-03rd December, 2014, International Telecommunications Society (ITS), Calgary. 16 One way to mitigate the adverse selection risk is via the 1.6 Structure of the report application of procurement methods that facilitate ‘price discovery’. Price discovery is desirable because it leads to resources being allocated more efficiently – at the margin The remainder of our study is structured as follows. In $1 of public funds should yield a marginal benefit of $1 – and chapter 2 we look at the specific challenge of funding DCI the public sector should obtain value for money. Competitive in Eastern Africa. This looks at the economic characteristics markets operate via the “invisible hand” to discover prices that affect DCI funding and the options available to – rivalry among suppliers drives down prices to reflect government in funding gaps and financing connectivity underlying costs. Procurement methods that promote infrastructure. This is followed in chapter 3 with a look competition among private suppliers seeking government facilitating private capital mobilisation in DCI, assessing subsidies for investments in DCI can bring about enhanced risks and probing into the role that can be played by national value for money by enabling price discovery. governments and development institutions. Over the years, a limited form of price discovery has been In chapter 4 of the report we take a deep dive into the ways facilitated through many MDB loans used to contribute to private sector investment can be incentivised by public funding gaps. These have relied on a procurement method funding. We describe direct subsidy methods and look known as ‘Best and Final Offer’ or BAFO. According to World closely at the opportunity for government agencies to act as Bank (2018) “BAFO is appropriate when the procurement anchor tenants and enable scale economies. This is followed process may benefit from Bidders/Proposers having a final in chapter 5 by looking at how the provision of funding opportunity to improve their Bid/Proposal, including by can be best structured to achieve value for money. In this reducing prices, clarifying or modifying their Bid/Proposal, or chapter our main recommendation of piloting multi-round providing additional information. It is normally particularly reverse auctions is described. effective when markets are known to be highly competitive and there is strong competitive tension between Bidders/ The final chapter 6 sets out our policy lessons and Proposers.”42 recommendations. We also set out a road map for commercial transaction manuals in regard of deploying While BAFO provides an opportunity for Bidders/Proposers multi-round reverse auctions. This is followed by a number to reduce prices and improve terms, its application often of appendices that look at a range of related and relevant takes place against a backdrop of negotiation that lacks issues, including procurement rules. We also provide some market discipline. Further, competing bidders and proposers interview summaries conducted with a number of World may have little visibility about other offers. An improvement, Bank Task Team Leaders. and in many ways an extension of the principles enshrined in BAFO, are methods that allocate subsidy finance by allowing bidders to change prices (and terms) more than once and in response to sight of some summary of bid information made by competitors. This can be achieved using a MRRA framework. Where a MRRA framework has been used to allocate public subsidies, as in India43 and the United States,44 savings have been identified and consequently the public enjoys better value for money. In World Bank and CEPA (2023) it was argued that application of MRRAs in Tanzania, where millions of dollars of subsidy have been awarded by way of single round auctions, the authorities could have achieved better value for money had MRRAs been used.45 In this report we analyse how a MRRA is likely to be beneficial within the context of DCI investments seeking PCM.46 42 World Bank. 2018. Procurement Guidance “Negotiations and Best and Final Offer (BAFO). Use of Negotiations and BAFO in procurement of Goods, Works, and Noncon- sulting Services”. 43 The application of multi-round reverse auctions in India is discussed in Wu, Irene S. 2010. “Maximum Impact for Minimum Subsidy: Reverse Auctions for Universal Access in Chile and India” FCC Staff Working Paper 2. More recently, Shivam Bajpai and A.K Malviya. 2023. “Efficacy of Electronic Reverse Auction (eRA) in the Public Procurement System in India” International Journal of Creative Research Thoughts, Volume 11, Issue 7 July, illustrates the benefits enjoyed in practice by MRRA schemes in public procurement in India. 44 World Bank and CEPA. 2023. “Allocating universal service subsidies using electronic multi-round reverse auctions: Telecommunications in Tanzania” discusses multi- round auctions that have taken place in the United States. The benefit of MRRA public procurement schemes in the United States are quantified in David C. Wyld. 2011. “Reverse Auctioning: Saving Money and Increasing Transparency” IBM Center for the Business of Government. 45 World Bank and CEPA. 2023. op cit 46 Examples of the use of MRRAs in the energy sector include: a) Brazil’s electricity auction in descending clock format (Box 5.1 IRENA Report); b) The UK’s Capacity Market Auction in descending clock format to award capacity market agreements. 17 Funding Digital Communications Infrastructure in Eastern Africa 2 2.1 Economic characteristics of digital Uncertainty. DCI projects are typically long-term in outlook and in the near term they involve considerable communications infrastructure outlays with revenues appearing further down the line. Returns for investors are therefore influenced by future To build the foundations needed to help countries in Eastern growth and future regulatory and governance environ- Africa connect to, and benefit from, the digital economy, ments, presenting risks that affect the cost of capital as discussed in chapter 1, significant investment finance and the amount of finance needed. There is a role for is required. 47 DCI projects are often large-scale, capital- government and MDBs to help manage these risks, in- intensive, long-term and involve considerable planning cluding aggregating demand for services through gov- and construction. Within the scope of this study, the ernment agencies and offering guarantees, as well as extension to the backbone national fibre network and last- establishing clear and credible regulatory rules. mile connectivity in the KDEAP project in Kenya is a prime example. Network effects. DCI services are delivered on plat- forms that may feature network effects in the form of The economic characteristics of these DCI investments externalities across different participants – different include: types of users avail of and provide services over the Large scale assets, irreversible investments and platforms on different sides of the market (some as buy- long-lasting. The KDEAP project includes extending ers, others as suppliers).48 For example, services may the national fiber backbone network (the long distance be accessed by end-users as consumers and supplied and high capacity routes between interconnected net- over DCI platforms by firms and government agencies. works and core routers in the internet) into the middle Network externalities within and across user groups on mile bringing services closer to users in the last mile. these platforms may lead to market failure in the form Investments involve substantial capital outlays and of under provision of DCI. This is likely to arise when construction costs which are estimated to lie between investors are unable to monetise externalities.49 As US$15,000 to US$30,000 per kilometer, depending many DCI projects have wider social benefits that are on the topography. In the KDEAP, as shown in Table not monetised through usage fees, this can provide a ra- 1‑2, the World Bank has allocated US$60 million to tionale for public support to correct this form of market extending the backbone to the middle mile and a fur- failure. ther US$90 million to last mile connectivity, along with US$35 million on connecting government agencies and The level of DCI in a country is affected by both demand and services (e.g. hospitals, educational institutions, etc.) supply-side forces, as discussed in the context of the gaps and US$30m for investments in cross-border connec- model above. On the demand side, income is the key driver tivity. This is estimated to be sufficient to leverage a fur- and as income increases, demand for more data and better- ther US$100m in unguaranteed commercial financing quality services increases which in turn increases the use of through “matching investment” by the private sector. digital infrastructure. For example, faster broadband speeds will become more affordable as income increases. Higher demand for high-speed broadband makes it more profitable to invest in expanding fibre and wireless networks. 47 See For example, Bandwidth & Cloud Services Group (BCS) expansion of fibre infrastructure in DRC, Malawi, Zambia, Mozambique and Zimbabwe involves an estimated total project cost of US$185m where 70% of the capex is concentrated in DRC. The IFC recently committed a debt package structured as an unsecured covenant-lite (“UCL”) loan of up to US$200m (US$[165]m A-loan and up to US$[35]m in mobilization from the Managed Co-Lending Portfolio Program (MCPP)) for Airtel Africa. The proceeds will support capex financing which will contribute to expanding and upgrading Airtel Africa network and also refinance part of its debt. e https://empowerafrica. com/internet-infrastructure-in-africa/ 48 A platform is an intermediary connecting two or more different sides of a market. For example, a two-sided market has (1) two sets of agents interact through an inter- mediary or platform [in this study the intermediary or platform is viewed as the physical and virtual networks enabling DCI services], and (2) the decisions of each set of agents affects the outcomes of the other set of agents, typically through an externality. Externalities can be positive (e.g. a consumer values services more when there are more suppliers on the other side to choose from) and negative (e.g. too many consumers on a platform may lead to congestion such as too many adverts, etc.). See Rysman, Marc. 2009. "The Economics of Two-Sided Markets." Journal of Economic Perspectives, 23 (3): 125-43. DOI: 10.1257/jep.23.3.125 49 Jacobides, M. G., Cennamo, C., & Gawer, A. .2024. Externalities and Complementarities in Platforms and Ecosystems: From Structural Solutions to Endogenous Failures. Research Policy, 53(1), Article 104906. https://doi.org/10.1016/j.respol.2023.104906 notes platform operators may have the scope to internalise the problem of mone- tising externalities by setting appropriate discriminatory tariffs. 18 Investments in DCI by government can stimulate future fiscal is a greater need for public support of DCI projects to help benefits and improve government efficiency which together promote growth and achieve the UN’s SDGs. The source of enable funds to support additional DCI related projects. In public funding will often rely to some extent on loans from the Project Appraisal Document for the KDEAP it is noted MDBs and IFIs. “The public sector would enjoy cost savings stemming from digitization and automation of core administrative functions A further risk arises in Eastern Africa in respect of currency and services to relieve strain on public finances…”.50 risk, where some operators manage exposure to currency risk by charging customers in US dollars. For example, BCS, a Businesses will demand more and better quality DCI, so wholesale fiber provider, charges all its wholesale customers, that they can further improve productivity and access wider except Airtel Uganda, in USD. markets. Some countries in Eastern Africa have strong growth potential, so demand for more and better DCI is However, with DCI projects there is the potential for likely to be high over the foreseeable future. Government considerable private sector funding and operational policy often seeks to facilitate demand for DCI, for example involvement and the key issue is how best to leverage this. Ethiopia’s Digital 2025 Policy set the objective of becoming a lower-middle income country by 2025 and involves various programmes.51 2.2 Government options and funding gaps Public sector participation in Digital Communications Infrastructure For many DCI projects in Eastern Africa there is a funding gap and the private sector alone is unlikely to deliver the If a DCI project is commercially viable from a risk-return socially optimal levels of investment. There is scope for perspective, then the private sector can be left alone to public finance interventions and public funding to help make finance investments. In middle- and high-income countries, up gaps in investment. A key problem, however, is imperfect the majority of investments in DCI are privately financed information about the magnitude of the funding gaps. This and public and private interests are fully aligned. Further, in asymmetry in information raises a question about the best middle- and high income countries access gaps are relatively approach to the funding of gaps.54 small. Traditionally telecoms funding gaps have been managed by However, even in middle- and high-income countries there one or more the following approaches:55 is widespread public funding of DCI programmes.52 These interventions arise for both efficiency and fairness reasons. Regulation: setting licence obligations that require uni- For example, the social benefits arising from investments versal service and other objectives to be met and/or the in DCI are not fully accounted for by private investors and application of controls on prices and the nature of com- public funding becomes a means to unleashing greater gains. petition. There is also pressure on many governments to deliver equality of access to DCI for inclusivity reasons.53 Universal service and access funds (USAFs)56: using a mix of industry levies and general taxation to disburse subsidies.57 For example, a mobile operator in Tanzania In low- and low-to-middle income countries, and especially is subject to a universal service levy of up to 1.5% ap- in Eastern Africa, immature private capital markets plied to gross operating revenue.58 combined with governments having weak fiscal positions and inadequate or under-resourced capacities mean there 50 Para. 70 in Project Appraisal Document for KDEAP, Report No. PAD4616, March 14, 2023. 51 On Digital Ethiopia 2025 see https://dfp.gov.et/. In a recent paper by Adam L, E Alemneh, Omar N and Partridge A (2024) African After Access 2023 Ethiopia, Post the Pandemic Policy Paper Series, Policy Paper 3, Research ICT Africa, Cape Town, the authors argue that Ethiopia’s ambitions will fall short of the targets set. 52 For example, the EU’s Connecting Europe Broadband Fund invests in underserved areas where there are strong opportunities for profitability. By 2021 it had raised €555 million for broadband investment and is expected to unlock total investments of €1-1.7 billion. See https://digital-strategy.ec.europa.eu/en/library/connecting-eu- rope-broadband-fund. 53 For example, see BEAD in US, Project Gigabit in UK, Broadband Europe in EU. 54 For related reports identifying digital infrastructure gaps and various financing options and partnership opportunities to address those gaps in Kenya and the Horn of Africa respectively see (a) World Bank and Salience Consulting. 2022. “Kenya Public Digital Infrastructure Options Assessment and Pre-Feasibility Study” and (b) World Bank and TMG. 2022. “Identification of Missing Broadband Links in the Horn of Africa Region”. 55 Many of the issues in this chapter were addressed in detail by Arturo Muente-Kunigami and Juan Navas-Sabater. 2010. “Options to Increase Access to Telecommunica- tions Services in Rural and Low-Income Areas”, World Bank: Washington DC. 56 GSMA. 2014. “Sub-Saharan Africa – Universal Service Fund study” September is a good, though slightly dated, overview of universal service policies in most SSA countries. In Table 3.2.3 of the report it can be seen that most countries apply a levy of 1% or higher on some measure of revenues. In a more recent report GSMA. 2023. “Mobile Tax Policy and Digital Development: A study of markets in Sub-Saharan Africa” October “Mobile consumers and operators in SSA are subject to a substantial tax burden, increasingly driven by sector-specific taxes and fees, which exacerbates affordability and coverage barriers. The multitude of taxes and frequent tax changes negatively impacts the business environment and hinders operators’ capacity to invest in network expansion and coverage.” 57 For example, see : Thakur, D. and Potter, L. 2018. Universal Service and Access Funds: An Untapped Resource to Close the Gender Digital Divide. Washington DC: Web Foundation. 58 See https://www.tcra.go.tz/download/sw-1619084421-The%20Universal%20Communications%20Service%20Access%20Act%20of%202006.pdf 19 PPPs: co-financing arrangements between the private Enforcement of coverage licence obligations is usually and public parties in an infrastructure project. In tele- one of the duties of the communications regulator, and in communications projects often involve foreign capital many instances this process is supported by the courts. and this presents currency risks. Obligations may define outputs and quality of service, and may also determine the type of service supplied. For Guarantees: underwriting commercial investments by example, it is often a requirement that licensees achieve offering insurance against losses. at least pre-defined coverage levels within a period of time after the grant of licence. For example, Safaricom Ethiopia, Anchor tenants: facilitating and channelling public sec- surpassed a 25 percent nationwide population coverage it tor demand for DCI enabled services to promote scale was required to hit as part of its licence, acquiring over 5 economies and lower unit costs of service delivery. million customers since launching commercial services in October 2022.59 In this study we are interested in how funds are disbursed and/or provided by the public sector. The approach taken Inserting obligations within licences can help promote towards funding gap investments can have a material investment but carries a risk that licensees will seek bearing on how much the private sector contributes towards concessions in other areas – such as higher prices. Further, investment. For example, a USAF manager could choose to the insertion of obligations may be misjudged and lead to disburse subsidies using a single-round, sealed-bid auction. inefficient outcomes. Obligations may also lead to less market This approach has to date been taken by the UCSAF in interest and diminished competition – something which has Tanzania. Once World Bank funds had been disbursed, been a feature at times in various Africa countries.60 UCSAF itself stepped in to offer a higher level of subsidy to operators for those sites that had missed out on bids through Universal service access funds, subsidies and the initial rounds of bids. guarantees In addition to directly decreeing objectives through licence Arguably if the UCSAF had instead used MRRAs it would obligations, government can also provide subsidies to help likely result in better value for money by promoting more fund and steer investments in DCI. Subsidy finance can investment and high private sector funding (see World Bank derive from levies on the telecoms industry, general taxation and CEPA, 2023). and borrowing. Where the telecoms industry faces levies, these raise costs61 and result in cross-subsidy: users with a This section gives a brief introduction to the various options high willingness to pay help to subsidise services provided available to mobilise private funding. to consumers with a lower willingness to pay. The impact on efficiency depends on the type of levies applied.62 According Regulation to Matheson and Petit (2017) “Universal service fund charges Regulation of private undertakings that supply networks also reduce profitability by reducing output and/or operator and services is commonplace and is a way for governments margins”.63 to achieve desirable policy objectives that would otherwise not arise absent regulation. One method used to regulate On the demand side, government can subsidise products entities is through obligations in regard of geographic and services to help bridge the gap between the returns and population coverage attached to licences assigned to an investor would require and the current willingness operators and service providers. Such obligations can also to pay of households and businesses. For example, the act as a means of channelling private investments in lower UK government operates a voucher scheme to connect profitability areas. unserved addresses to gigabit broadband services.64 In Africa the government of Rwanda applied a subsidy scheme This could be done in respect of DCI. However, it requires to purchasers in rural areas of mobile handsets. Analysis government to know what services should be delivered as of this scheme by Björkegren and Karaca (2022) showed part of DCI, who the recipients are and the implications for that “welfare impacts are positive, with a social rate of return end user prices. of at least 44%. A substantial portion of impact arises from spillovers”.65 However, schemes of this kind appear to be the exception rather than the norm in Sub-Saharan Africa, possibly because of concerns about corruption and a lack of capacity in public administration.66 59 See Capacity. 2023. “A brief guide to Ethiopia’s attempts to licence a third operator” 60 Ethiopia suspended the award of a third national telecoms licence in December 2021 (partly as a result of conflict in the north of the country) following the failure of bids to materialise after expressions of interest were received. According to the regulatory authority, the process is currently “parked”. 61 See Gregoire Rota-Graziosi and Fayçal Sawadogo. 2022. “The tax burden on mobile network operators in Africa”, Telecommunications Policy, Volume 46, Issue 5 who find that telecommunications faces a very high tax burden in many African countries. 62 Sector specific taxes that target excess profit can be efficient. See Thornton Matheson and Petit. 2017. “Taxing Telecommunications in Developing Countries” IMF Work- ing Paper, Fiscal Affairs Department, WP/17/247. 63 Matheson and Petit. 2017. page 16, op cit 64 For example, see “Gigabit Broadband Voucher Scheme information”, February 2024. 65 Daniel Björkegren and Burak Ceyhun Karaca. 2022. “Network adoption subsidies: A digital evaluation of a rural mobile phone program in Rwanda”, Journal of Develop- ment Economics, Volume 154, 102762, ISSN 0304-3878. 66 For example, see Lamin Cessay (2019) “Corruption in Sub-Saharan Africa - An Impediment to Economic Growth” European Scientific Journal, April, vol.15, no.10. 20 A demand side measure that can stimulate PCM is through In some instances local guarantee companies could packaging public sector demand for digital services and be established, with advocates claiming such schemes providing guaranteed payments for services. This is can unlock lending from local pension funds, insurance considered below in section 2.2.4 within the context of companies, and the like, for projects that commercial banks anchor tenants . On the supply side, government can give are reluctant to finance. An illustration of directing financing subsidies to the private sector to build infrastructure, as in this way is shown in Figure 2‑1 below: occurs in Tanzania. By lowering the funding gap, private sector interests are incentivised to build out more infrastructure. As guarantees are defined in local currency, exchange rate As an example, see Appendix F.1 for a discussion of supply- risk is decreased and transferred from the telecom operator side subsidies to allocate funds for investment in mobile or supplier of DCI services, which is a significant source of broadband digital infrastructure in Tanzania. risk in many African countries. Local currency solutions are appealing, especially as in recent years currencies in many On the supply side, government and MDBs could offer Africa countries have lost value or been highly volatile. a capital market ‘subsidy’ by offering guarantees in collaboration with commercial banks.67 This would provide An example of a guarantee company in Kenya is InfraCo a guarantee that money loaned to projects will be repaid, Africa, a finance vehicle backed by the UK, the Netherlands, which can be an important stimulant of PCM. The guarantors and Switzerland, which has made some local currency charge borrowers a fee for taking on this risk. investments supporting guarantees, mainly in Kenya.69 However, while this option is a positive step forward in PCM, the investment amounts involved to date have been relatively small. Figure 2‑1 Intermediate funder carries foreign exchange risk Standard Credit Facility Loan Agreement Funder International Telecom Refinancing Credit Agreement Bank Operator Principal & Interest Guarantee Guarantor Guarantee On-Lending Credit Facility Loan Agreement Loan Agreement Telecom Funder Refinancing Credit Agreement International Intermediate Operator Bank Funder Principal & Interest Principal & Interest Source: European Investment Bank (2021) 68 67 Guarantees enable financing partners to transfer certain risks that they cannot easily absorb or manage on their own to other parties such as the WB or government. For example, partial risk guarantees cover lenders against nonpayment by the borrower caused by political risk events. See World Bank Group Guarantee Products (2016), Guidance Note. 68 European Investment Bank. 2021. “Unlocking digital connectivity in Africa” page 40 69 FSD Africa. 2022. Guarantee companies unlock African infrastructure finance 21 Public-private partnerships In any case, for many telecom DCI projects PPPs may not be As we have already set out, PPPs may act as vehicles for appropriate because relative to other infrastructure sectors government to provide funding in collaboration with private there is often considerable private sector involvement and sector interests to deliver public services. Over the term of a investor appetite. Instead, subsidies and PCM are more contract revenues are drawn from taxpayers and/or users for effective mechanisms to bridge the funding gap. However, profit. Thus, a PPP is one way to fund DCI gaps by blending caution is needed to ensure that investment appetite does users’ revenues, taxation and government borrowing. PPPs, not translate into entrenchment of market power. which are championed by development institutions such as the African Development Bank (AfDB), are becoming Anchor tenants more and more common in Africa.70 As alluded to in section We discussed benefits that could derive from anchor tenants 1.4, there are many potential benefits to PPPs and, when in section 1.5. In the commercial world of real estate, an designed carefully, they can foster efficiency gains. However, anchor tenant is a store whose presence increases footfall because PPPs often involve very complex contracts and, for nearby stores in a shopping mall. The shopping mall as a over time, these are often imperfectly implemented and platform typically offers generous and possibly below cost subject to renegotiation. rents to an anchor tenant and recovers costs by charging higher rents to smaller stores willing to pay more because PPPs can appeal as they relieve government funding of the higher demand (footfall) due to the anchor tenant.72 problems with projects relying more on private sector funding. However, government fiscal commitments to In a DCI setting, anchor tenants are potential large buyers of PPPs can be unclear, and treating PPPs as off-balance data services such as universities, hospitals, schools, etc. The sheet can lead to governments accepting higher fiscal presence of government agencies as anchor tenants can be commitments and risk than prudent financial management structured through funding to provide guaranteed income would suggest.71 Furthermore, many African countries for infrastructure investors and thereby lower project risks. lack the capacity to implement and monitor the delivery For example, a government agency may agree to a long-term of PPPs, and private sector participation may be limited, contract to pay a fixed amount upfront for the rights to using especially as they involve considerable bureaucracy. This say 50% of fibre capacity at no usage charge over ten years.73 can result in a lack of competition for PPP contracts and the potential for contract renegotiation following award of Scale economies afforded by anchor tenants lead to lower contracts. Strong governance is key to ensuring PPPs are incremental costs for access by other users of DCI. Thus, implemented effectively. Governments can also decide to government funding via long-term contracts for services reverse PPP arrangements, as happened in the case of the supplied to anchor tenants can result in higher coverage Burundi Backbone System where the Government decided and welfare. Unlike in the commercial world of real estate, to nationalise a PPP that had originally been set up under a where smaller stores pay more, a government anchor tenant World Bank lending program, chilling the environment for acts to transfer benefits to smaller users and helps increase private sector investment. overall welfare. We revisit this idea of anchor tenants in chapter 4.2. 70 See the African Development Bank “PPP STRATEGIC FRAMEWORK 2021-2031”. 71 See World Bank. 2022. “Fiscal Accounting and Reporting for PPPs.” Available at: Fiscal Accounting and Reporting for PPPs Public Private Partnership (worldbank.org). Since PPPs transfer risks (such as construction or demand risk) to a private party, under accounting rules most PPPs are not recognised in debt statistics. Most accounting and reporting standards do not require governments to recognise contingent liabilities, including those arising from accepting risk under PPP contracts. 72 An anchor tenant would usually bargain with the mall owner over the division of rents arising from its presence. Konishi, H., and M. T. Sandfort. 2003. “Anchor Stores.” Journal of Urban Economics 53 (3): 413–435. https://doi.org/10.1016/S0094-1190(03)00002-0. 73 For example, BCS's fibre construction in DRC has involved laying 1,200km of backbone (no metro yet), and a plan to build an additional 6,000km of backbone and 3,000km of metro. BCS has built backbone and metro fibre with 96 cores plus a submarine cable of 24 cores. Out of the 96 cores, 24 are owned by the Government (form- ing an anchor tenant) with remaining owned by 48 BCS plus 24 Vodacom (operator). The total capacity of the 96 cores is 1.6Tbps. 22 2.3 Financing connectivity infrastructure Securing financing is critical for DCI projects to make it past the planning stage. There are a range of financial instruments that can be used, outlined in Table 2‑1. Table 2‑1: Financial instruments for DCI Financial Instrument Merits Challenges Public budgets are sums of money Backed by state policy Influenced by changes in national allocated for public expenditure by a Steady flow and stable value priorities state entity Subject to public scrutiny Dependent on political buy-in Grants are sums of money that are Aligns DCI with welfare goals Unreliable for long-term financing as commitment-based or philanthropic in Ideal for funding the R&D aspects most grants are provided for a fixed nature of DCI period of time with uncertain terms Often large amounts of money, of renewal with limited control over usage Limited public accountability Private capital is a sum of money Possibility to raise more capital Typically focused on short-term goals invested in an entity or infrastructure Forges strategic partnerships with Could create the micro-economic in exchange for ownership rights investors issue of principal-agent conflict Ideal for funding the R&D aspects of DCI Debt is a sum of money borrowed Lowest risk for funder Interest payments can cause fiscal for a fixed period of time, with Ownership rights remain with the burden accompanying interest payments entity/country Restrictive terms of operation Source: UNDP 2023 74 There are numerous examples around the world of how It has also seen varying contributions from UNDP, these financial instruments have been used to develop DCI. Sweden, Swiss-funded E-Governance for Accountability and Participation Program (EGAP),76 EU4Digital, and Public budgets – for example, India’s National Health others over the years. The app provides online access to Authority was allocated US$41 million from the more government services to all Ukrainians with digital national budget in 2023-24 for the Ayushman Bharat documents such as ID-cards, student ID, driver’s licenses, Digital Mission which aims to support the integrated vehicle registration certificates, vehicle insurance policy, tax digital health infrastructure of the country. It will bridge numbers, birth certificates and resettlement certificates and the existing gap amongst different stakeholders of foreign biometric passports. healthcare ecosystem through digital highways. Within ten months of its nationwide launch, 20 government Private capital - As of June 2023, 65 banks were platforms and 32 private health applications were shareholders of the National Payments Corporation of integrated and contributing to scaling the network.75 India (NPCI), an umbrella organization that operates the Unified Payments Interface in India. There are currently Grants in response to a shift in geopolitical conditions 22 third-party applications developed by private sector in 2022, USAID contributed US$25 million to further firms for end-users to avail UPI for digital payments.77 develop Ukraine’s e-governance app Diia. With support from the UK Transparency and Accountability in Public Debt – MDBs such as the WB provide debt finance Administration and Services (TAPAS) project, Diia has for DCI projects. For example, a US$250 million loan been in development since 2019. The first version was to strengthen population and civil registration while developed by EPAM Systems, a private software firm. increasing the usage of digital identification for service delivery in Indonesia.78 74 See https://www.undp.org/sites/g/files/zskgke326/files/2023-08/undp-the-DCI-approach-a-playbook.pdf 75 See https://abdm.gov.in/ 76 See https://eef.org.ua/en/program/programa-egap/ 77 See https://www.npci.org.in/ 78 See https://projects.worldbank.org/en/projects-operations/project-detail/P175218 23 Figure 2‑2 Sources of finance for infrastructure projects with private participation in emerging and developing economies Total Investment (100%) Sources of Financing Equity Debt (24%) (67%) Private Equity Subsidy (24%) International Debt (9%) (35%) Local Debt Public Equity (32%) (0.3%) Decentralized DeFi Debt Non DeFi Debt Finance (15%) (20%) Multilateral Bilateral Commercial Public Institutional (10%) (5%) (25%) (26%) (0.9%) Source: PPI Database (2022)79 *All figures as a percentage of total investment 2.4 Conclusions Funding the build out of DCI in Eastern Africa will face the challenge of funding gaps over the coming years. In this chapter we have set out how funding can be structured given the economic characteristics of DCI. Measures such as regulation can be applied to achieve desirable outcomes, but other approaches that may utilise competitive tendering, such as the allocation of subsidies, might yield better value for money. These themes will be examined further in the next few chapters . 79 Source https://ppi.worldbank.org/en/ppidata 24 Facilitating Private Capital Mobilisation in Digital Communications Infrastructure 3 3.1 Current constraints and risks The development of DCI in Eastern African needs to ensure 2. Currency risk. For international investors a key concern that funds invested by the public sector deliver the best is currency volatility and depreciation, as revenues are value for money and PCM is maximised. This is all the more defined in domestic currency. This can be caused by important as macroeconomic indicators for the region have a lack of prudential macroeconomic policy and poor generally declined in recent years, partly as a result of the foreign exchange reserve management in addition COVID-19 pandemic and other overlapping crises, placing to other economic factors. Djibouti is an example a greater strain on government accounts. According to of a country with relatively low currency risk since the World Bank Africa’s Pulse October 2023 report, rising the Djibouti Franc is pegged to the dollar and is fully instability, weak growth in the region’s largest economies, covered by foreign exchange reserves. Ethiopia has a and lingering uncertainty in the global economy are dragging high currency risk, as the Birr is not freely convertible down growth prospects in the region.80 and the country lacks sufficient foreign reserves.84 Further, debt distress remains widespread, with 21 3. Political risk. Political instability is a major concern for countries at high risk of external debt distress or in debt private sector investors, particularly in parts of Eastern distress as of June 2023. In 2022, the median debt-to-GDP Africa. Unstable governments may suddenly cut funding ratio for low-income-countries (LICs) in Sub-Saharan Africa and short political cycles may challenge commitments reached 55 percent, while it reached 58 percent for middle- to long-term infrastructure projects. The possibility income-countries in the region—in both cases, there was an of regional violence or conflict can completely halt increase in the debt ratio of around 27 percentage points local infrastructure investment. Limited competition since 2012.81 Rising interest rates have increased the cost of in the telecommunications sector makes foreign borrowing to fund infrastructure investments and budgets investors fearful of unfair treatment due to government are constrained by rising interest payments on existing debt. preference for state owned firms, particularly if it There are four key risks which act to constrain private is a market structure in which the fibre backbone is investment in DCI in Africa: dominated by a state-owned incumbent. Ethiopia, Somalia, and South Sudan are classified by the World 1. Project risk. Although there is a vast array of business Bank as Fragile and Conflict-Afflicted Situations (FCS). opportunities, the pipeline of projects that are ‘investment ready’ is limited. Investors are cautious of 4. Exit risk. Underdeveloped financial markets can prevent investing in unfamiliar markets or early-stage concepts investors from exiting by issuing shares and capital which have not been implemented before. Particularly controls can slow down or increase the cost of exiting. for last-mile digital infrastructure, these projects are Legal frameworks are often weak and investors have unprofitable and costly to deliver with unpredictable limited protections available. If investors are unable to revenues. Governance and regulatory ambiguity can recoup their gains from investment, they will be unlikely also deter investors. For example, the US Department to invest in the first place. Apart from Kenya, the project of State finds that foreign investment in Madagascar countries all have relatively weak and under-developed is limited by the government’s reported lack of fair and capital markets. consistent implementation of existing laws.82 Other rules and restrictions, such as foreign ownership in the Malagasy telecommunications sector being restricted to 66 percent, means that foreign investors might look elsewhere.83 80 Africa’s Pulse: Delivering Growth to People through Better Jobs, October 2023, World Bank 81 Africa’s Pulse, October 2023 pp. 44-45 op cit. 82 See https://www.state.gov/reports/2023-investment-climate-statements/madagascar/ 83 The recent licensing of Starlink in Madagascar suggests the government’s willingness to waive this foreign ownership limitation. 84 The credit rating of Ethiopia was downgraded on several occasions toward the end of 2023. The rising cost of external finance is resulting in government seeking more domestic finance to fund government deficits. This environment is putting increasing pressure on the funding of large infrastructure projects. See Addis Standard “Ethio- pia’s credit rating downgraded by Fitch, amplifying high default risk”, 3 November 2023. 25 Table 3‑1: Perceived investment risks in the six countries in this study, CEPA estimates Country Project Currency Political Exit Description Risk Risk Risk Risk Seen as a relative safe haven in the region and currency pegged to the dollar. Some political risk Djibouti from Chinese investment and spillovers from conflict in Middle East and the Red Sea. Weak         financial sector and opaque business practices. Ethiopian birr not freely convertible and lack of foreign reserves. Areas of conflict and violence. Ethiopia Lack of developed capital market but reform minded government. No functioning stock market.         Positive investment climate with relatively sophisticated capital markets. Recent currency Kenya appreciation relative to US$. Some political risks.         Lack of comprehensive legal and regulatory framework and presence of al-Shabaab terrorist Somalia group. Government welcomes FDI and has maintained macroeconomic stability. Weak         financial system. Reported unfair practices such as expropriation of assets and unpredictable tax policies, years of South conflict and internal displacement. No functioning Sudan stock market.         Government publicly welcomes FDI but in practice has challenging business environment. No stock Madagascar market, but developing one, and foreigners cannot participate in the bond market.         Legend: is high risk; is medium risk; and is benign or normal risk. Mitigating risks is complex and various institutions have a vital role to play to facilitate private sector investment in DCI in Africa. Sections 3.2 and 3.3 below focus on specific actions that local governments and development institutions can enact. While local governments and development institutions are key to creating a conducive environment for DCI, collaboration between the public and private sectors will help catalyse implementation and effectiveness. 26 3.2 Role of national governments 3. Promoting effective governance and security Governments play a key role in facilitating private sector a Good governance: promoting transparency, checks investment in DCI. We have identified three broad areas for on corruption, improving regulatory quality, and reform. political stability all help to incentivise private sector investment. Good governance improves other areas which affect the investment climate, such as policy 1. Improving the business climate uncertainty, macro instability, cost of access to finance, insecurity and crime.87 In particular, promoting voice a Macroeconomic stability: prudential macroeconomic and accountability and improving regulatory quality are policy and sound foreign exchange reserve management especially important, see Saha et al. (2022).88 can limit currency risk which will attract foreign investors and can make projects more profitable. b Transparent regulations and protections: Transparent regulations, adequate legal protection for investors – 3.3 Role of development institutions especially foreign investors, and easy permitting and land acquisition can reduce legal and operational risks The involvement of development institutions such as the and enhance the commercial viability of projects. A World Bank is important to incentivise greater private good investment climate provides opportunities and sector involvement in DCI projects in Africa, effectively incentives for firms to compete and innovate. PCE. The role of development banks is leveraged through a wide range of mechanisms. 2. Providing financial incentives Laying the foundations. The presence of past projects a Subsidies: particularly for projects that are not seen (e.g., in digital infrastructure) makes it easier to as profitable but still provide a wider social benefit, implement similar projects in the future. Lessons can such as developing last-mile digital infrastructure, the be learnt from failures and successful strategies can be government can provide subsidies to make projects replicated. It also acts as a signal that the government viable. This is already done extensively in other parts is committed to investment in infrastructure projects so of the world. In East Asia, 90 percent of infrastructure there might be less perceived political risk surrounding projects with private participation receive government questions over ongoing support. However, direct support.85 The private sector will only get involved in involvement by the World Bank ends once a project projects with adequate risk returns, so government formally closes. backing can provide both higher potential returns and lower risk. Multiplier effect. Investment by MDBs such as the World Bank, often working in collaboration with b Reallocating funding: given budget constraints and regional economic commissions such as IGAD and ECA, limited availability of public funds, governments can have a multiplier effect, where the total investment may consider reallocating resources used for public created is a multiple of the initial investment provided. investment towards financing public incentives for Private investors will have increased confidence in the private projects. The private sector is typically more country and projects and will be more likely to invest if effective in allocating resources and suffers less from MDBs have previously invested. external influences, such as trying to appease voters or the status quo bias, which may affect project feasibility. Improved local governance. Development institutions The McKinsey Center for Government analysis finds can help local governments improve governance by that governments across most countries and at all making funding contingent on meeting specific criteria levels are less effective in allocating resources than or by providing support and expertise. An example of the private sector, being slower to reallocate resources success is the World Bank’s ICT Sector Support Program where needed and delivering little change in spending for Somalia,89 which ran from 2014 to 2020, and had a allocations from year to year.86 To avoid crowding out focus on promoting good governance and ICT sector the private sector, public finance can be reallocated regulation. The government was helped to develop from the most commercially viable asset classes to those telecommunications legislation and encouraged that provide lower returns which are more suitable for private sector investment in ICT infrastructure, such government investment. as the development of mobile money and improving connectivity. 85 See https://www.adb.org/news/features/how-much-should-asia-spend-infrastructure-0 86 See https://www.mckinsey.com/industries/public-sector/our-insights/how-governments-can-be-more-effective-by-reallocating-their-resources 87 World Development Report (2005), World Bank. 88 Saha, S., Sadekin, M. and Saha, S. K. 2022. “Effects of institutional quality on foreign direct investment inflow in lower-middle income countries.” Heliyon, 8(10): e10828. 89 See https://www.worldbank.org/en/news/feature/2021/09/27/supporting-information-and-communications-technology-advances-in-somalia 27 Data from the International Telecommunication Union Risk Sharing. MDBs can leverage the private sector shows that the cost of a low-user package of mobile by making investments less risky. For example, the services fell by 76 percent between 2015 and 2019, Multilateral Investment Guarantee Agency (MIGA) when the project was in operation.90 provided an initial US$1 billion guarantee for political risk insurance (PRI) cover against the equity investment Funding and expertise. Many governments and in the first privately owned telecommunications network developers lack the capabilities and capacity, as well in Ethiopia, a consortium led by Safaricom of Kenya, and as the budgets, to design and implement infrastructure IFC has taken an equity share in the consortium.93 projects with commercial potential. Very few projects reach financial close, with 80 percent of African infrastructure projects failing at the feasibility and business-plan stage91. MDBs can provide the funding 3.4 Conclusions and expertise necessary to develop profitable projects. In this chapter we have set out the risks that surround Trust. MDBs can help build trust between the public DCI investments in Eastern Africa and considered ways and private sectors and facilitate cooperation. Trust governments and MDBs can mitigate them. Policies to is defined by the expectation that the other party will manage and mitigate risks play an important role in affecting take the trusting party’s interests into account and the environment for private sector interests and the extent will not behave opportunistically.92 Trust is necessary to which PCM occurs. In the next chapter we look in more for the private sector to enter investments with the depth at how approaches to incentivising private sector public sector due to the perceived risks involved. For investment in DCI. example, there is the risk that an actor will abuse power in the project or abandon the cooperation, which forces the other actor to bear the costs. Involvement of a trusted third party without vested interests, such as the World Bank, can encourage the private sector to enter contracts as they trust that they will be treated fairly and not left with any unexpected costs. 90 See https://www.itu.int/en/ITU-D/Statistics/Dashboards/Pages/IPB.aspx 91 See https://www.mckinsey.com/capabilities/operations/our-insights/solving-africas-infrastructure-paradox 92 Warsen, R., Nederhand, J., Klijn, E., Grotenbreg, S. and Koppenjan, J. 2018. “What makes public-private partnerships work? Survey research into the outcomes and the quality of cooperation in PPPs.” Public Management Review, 20(8): 1165-1185. 93 See https://www.worldbank.org/en/results/2023/12/01/mobilizing-the-private-sector-to-drive-development-in-africa 28 Approaches to Incentivize Private Sector Investment in Digital communications 4 Infrastructure As already stated, the private sector has played a crucial In the case of last mile connectivity infrastructure role in developing telecoms infrastructure in SSA.94 investment, methods such as tax allowances are likely to be Notwithstanding, the private sector is not in a position to ineffective due to the political risk involved and the fact that deliver within a timescale preferred the scale of investment the burden is still placed on the private sector to raise capital. required to achieve ambitious government policy goals Contracts with the private sector, such as the provision of associated with DCI. Policy ambitions, along with market direct subsidies, will be more effective due in this case due to failures and institutional shortcomings, necessitate the sharing of risk between public and private sector and the intervention by governments and MDBs. As discussed in ability for the government to state what investment needs Chapter 10, substantial funding gaps in both connectivity to occur as part of the contractual conditions. and access justify public finance intervention. The focus of this chapter is to look more closely at how governments can The key policy question is how public and private funds can leverage private sector finance and investments in DCI. be best used in combination to maximise DCI investments. The remainder of this chapter outlines different approaches There are a range of approaches governments could use to leveraging the private sector. First, we consider subsidies to incentivise private sector investment in connectivity – outlining both the traditional methods used to allocate infrastructure and DCI. These range from traditional fiscal subsidies in telecoms and in particular direct subsidies, approaches, such as tax incentives, to long term contracts which we believe are most likely to be appropriate to with the private sector through PPPs. Table 4‑1 summarises maximise the benefit derived from the funding. We also these two direct approaches. Other methods outside of the develop the anchor tenant model further, which can be used scope of this report, such as policies to improve governance in conjunction with some form of subsidy rather than simply and the investment environment, can be effective alongside as a direct alternative. these more direct approaches. Table 4‑1 Approaches to incentivise private sector investment in DCI Approach Advantages Disadvantages Altering taxes e.g., lower Reducing the tax burden Political risk – tax often altered so no guarantee the favourable corporation tax for companies who invest in tax policy would last. connectivity infrastructure Burden on private sector to raise capital, which would be can make projects more expensive due to risk of last mile infrastructure. profitable. Can be targeted Often not a strong enough incentive for investment in risky and at high-cost rural areas. unknown projects. Fiscal pressures, though these can be mitigated to some extent by MDB lending. Contracts involving Share risks between public Political risk – government can break commitment e.g., private sector e.g., PPPs, and private sector. nationalise PPP, but more difficult to do than altering tax policy. subsidy Contract can specify quality Potential to distort market competition if PPP project given targets and give government protections. more control over output. Source: CEPA analysis 94 See the discussion paper from the International Telecommunication Union, available at: https://www.itu.int/ITU-D/ict/papers/bmi/bmi98.pdf 29 4.1 Subsidy methods Such an agreement relies on a strong legal framework in the particular country where the investment is aimed for; We considered subsidy approaches through USAFs in however, such arrangements are now standard practice chapter 2.2.2 and noted subsidies can be provided to either in most countries including those in Africa. Supply-side service providers (supply-side subsidy) or to customers of subsides are well suited for DCI projects in Eastern Africa, communications services (demand-side subsidy). Both aim such as rollout of last- and middle-mile fiber, because it helps to improve the commercial viability of a project, but on to provide gap funding for projects that might otherwise different sides of the market. be uncommercial. Box 1 describes a supply-side subsidy implemented by the WB in Niger to speed up construction A supply-side subsidy is where the government or other of cell towers in 2,000 plus underserved villages.95 public body such as an MDB provides funds to a private company in return for delivery of certain obligations, An alternative to subsidising infrastructure cost and build is such as building and operating DCI in areas that might to stimulate demand by providing users with a subsidy – this not otherwise be served. The subsidy would typically can be focused on both consumer related infrastructure build be funded through fiscal measures (general taxation or and/or devices needed by consumers to avail of services. government borrowing) and enable the private party to Demand-side subsidies have occurred in Eastern Africa, an obtain privately funded capital to fund the development. example is shown from Rwanda in Box 2. In 2023, Airtel in The subsidy payment being made to the private party upon Rwanda partnered with Reed Hastings (founder of Netflix) competition of the contractually agreed obligations is a form to provide subsidized 4G smartphones for around USD$20 of commitment device: it ensures the private sector delivers coupled with 4G daily data and unlimited calls and texts at on its agreements otherwise it will not receive the subsidy, USD$3/month.96 Subsidisation of last-mile fibre rollout in see Box 1 below for a discussion in the context of Niger. This rural areas occurs in many high-income countries and an protects the government or MDB from providing funds for a example is provided in Box 3 from the United Kingdom.97 project that is subsequently not delivered as promised. Box 1: Niger Smart Villages Project (P167543) – subsidies for CAPEX Background: Niger is a large, landlocked, country in the Sahel region of West Africa and is one of the least developed countries in the world. Niger’s digital infrastructures are significantly under-developed, rated last out of 176 countries in the 2017 ICT Development Index of the ITU. At least 10 percent of the population has no mobile phone telephone coverage and around half the population is not covered by mobile broadband. The Niger Smart Villages Project was launched in 2020 with the aim of increasing access to cell phone and broadband services in rural areas and bringing digital financial services to selected underserved areas. Component 2 of the project aims to develop connectivity for approximately 2,111 “smart villages” out of a total of 6,092 rural villages lacking connectivity. Subsidies: Supply-side subsidies for capex of around US$60,000 per village are provided for underserved villages with a population between 250 to 2,500. This population range was chosen because larger villages would likely be served even without a subsidy if enabling polices and an appropriate regulatory environment were in place. Smaller villages would likely be uneconomical to serve using today’s technology without also subsidising operating expenditures. Subsidies were most commonly used for construction of additional cell towers in areas with limited or no cellular coverage. Capital subsides were provided to the private sector through a reverse auction subsidy tender process where the winning bidders request the lowest levels of subsidy. Additional funding to promote connectivity is expected to be supplied by winning bidders, among others. The payment of subsidies, however, is made contingent on winning bidders in the process undertaking the investments promised. This illustrates the importance of post-auction audit in ensuring commitment. 95 See https://projects.worldbank.org/en/projects-operations/project-detail/P167543 and https://documents1.worldbank.org/curated/en/315991594519247904/pdf/ Niger-Smart-Villages-for-Rural-Growth-and-Financial-Inclusion-Project.pdf 96 See https://taarifa.rw/airtel-unveils-rwandas-most-affordable-lte-smartphone-with-an-exclusive-offer/ 97 See https://www.topafricanews.com/2024/01/22/rwanda-pushes-africa-towards-universal-smartphone-access/ and the #connectrwanda programme at https://www. minict.gov.rw/ 30 Box 2: Smartphone subsidies in Rwanda Rwanda recently-announced a three-year target for smartphone ownership and is one of the most ambitious in Africa. As part of the innovative funding scheme, citizens can obtain handsets at a subsidised price of US$16, complemented by a data plan costing just US$1 for 1GB, valid for 30 days and inclusive of unlimited text messaging. “We are optimistic that in the next three years, at least every citizen will have this level of access that allows them to benefit directly,” Communications Minister Paula Ingabire told reporters in Davos in early 2024. Box 3: Demand-side subsidies - Gigabit Broadband Voucher Scheme (UK) The United Kingdom government has introduced a policy called Project Gigabit to promote faster broadband in areas of the country that lie in the access gap. As part of the £5 billion (US$6.4 bn) Project Gigabit, the UK Government is providing up to £210 million (US$268 m) worth of voucher funding for people experiencing slow broadband speeds in rural areas. Vouchers help to cover the costs of installing gigabit broadband. In effect, the voucher bridges the gap between the required returns for a fiber network investor and the current willingness to pay of households and businesses to pay for these services. Only suppliers can request vouchers on behalf of their customers. The value of the voucher contributes to the build cost of installing a gigabit-capable connection at the customer’s premises. Testimonies state how broadband upgrades resulting from the scheme have greatly increased the utility of those who live in rural areas. Users can download large documents quickly and engage in video-conferencing calls that support business development. Residents of supported areas are now able to work from home when they would have had to drive for hours to reach a quality signal. This case study highlights the success of demand-side subsidies in a UK context. However, this does not mean that it would be successful in Eastern Africa since construction costs might be so high that it is impossible for the company to finance and build the infrastructure without a supply-side subsidy. In addition, the sufficient level of demand-side subsidy required per individual in the East Africa context may be significantly higher than the level required to achieve the desired outcomes in the UK. Below we consider the advantages and limitations of using subsidies against other interventions such as direct regulation, public ownership models or government or public supply. Advantages Supply-side subsidies compatible with competitive allocation mechanisms to achieve value for money: If service providers are competing solely on the level of subsidy required, government can establish a procurement process that is compatible with an auction format to promote better value for money. These are known as reverse auctions and are discussed further in chapter 5. Broad experience: Subsidies are used around the world to support the rollout of DCI, such as the construction of last mile infrastructure to bring broadband to rural areas. Subsidies have also been widely used in WB programs such as Niger Smart Villages (P167543), Digital Malawi Foundations Project (P160533), and the Digital Tanzania Project (P160766). Efficiency: Following the award of a subsidy, recipient private companies have an incentive to build the required infrastructure for as low a cost as possible so as to receive all the subsidy funds. 31 Greater transparency and accountability when This is particularly important when funds are scarce. combined with auction allocations: A direct subsidy Many factors, including the current areas of coverage, when allocated through use of a reverse auction, can be need to be determined to assess which areas require associated with a process that has greater transparency subsidy and which do not. This is a challenging and and accountability. It is clear who is allocated what lengthy process. It is also potentially susceptible to subsidy, and whether there are any unallocated funds, lobbying by vested interests and risk the result of an following competition in an auction process. This is inefficient allocation. in contrast to where subsidy funds are allocated by negotiation or application, as features in in many Red tape: Subsidies and grants are conditional on USAFs, where outcomes often lack transparency and numerous requirements which may be quite stringent accountability. on the private party in terms of security, open access, social and environment safeguards requirements and Limitations other obligations. This may lead to concerns of either Effective procurement mechanism requires undue government control or an unjustified burden competition: Subsidies may not provide value for money placed on the private sector which would be inefficient. if the method of allocation does not involve competition. However, in most subsidy allocation areas there is likely Buyer determined award rule: Rather than award the to be some competition, though policy makers need to subsidy funds to the lowest bidder via an auction, it is take care when it is very limited. possible for governments to allocate funds using an arbitrary scoring system. Bureaucrats usually score Challenging to determine appropriate level of subsidy: applications using a degree of discretion, which can For many DCI projects in Eastern Africa, such as last mile lead to corrupt influences, especially in countries where fiber, expected costs vary significantly to the areas in oversight and court resources are limited. This is a which networks have already been developed. It makes commonly used approach in public sector procurement it more difficult for private companies to bid for a subsidy and also widely used in the private sector. if they are very unsure about costs and demand. There is a risk of winning bidders receiving a subsidy award that is less than the amount required to complete the project 4.2 Government agencies as anchor without making a loss. This phenomenon is analogous to the concept of the winner’s curse – the idea that, if you tenants are the winning bidder, it is because you have bid too The idea of using government agencies as anchor tenants low (or too high, in the case of an ascending auction).98 as part of communications policy has developed traction Given bidders are aware of the winner’s curse, they will over the last fifteen years. Above, we mentioned that BCS tend to include a relatively large risk premium within hosts a number of anchor tenants on its fibre network in their bid prices to avoid winning an insufficient subsidy. Eastern Africa.99 Stenberg (2010) described the merits of With imperfect information it will be rational for all anchor tenants in helping to improve broadband coverage bidders to do so, risking an inefficient outcome and the in the United States.100 He showed that over 32,000 procurement not delivering value for money. A multiple anchor institutions received subsidy funding under the round auction for subsidy finance can help mitigate the American Recovery and Reinvestment Act of 2009 (ARRA) winner’s curse. to help improve broadband connectivity. Also in the United States, the Benton Foundation (2016) published an action Difficulty in deciding where to allocate the subsidy: The plan on behalf of schools, health providers and libraries government or public body in charge of allocating the to promote broadband connectivity using Community subsidy funds needs to decide carefully what geographic Anchor Institutions (CAI).101 Part 7 of the ten-part plan regions the funding should be assigned to. recommended:102 98 The winner’s curse that may arise in an auction setting requires the value of items to have a common value component. Common value is where the intrinsic value of an item is the same for everyone. The bidders in an auction form their own expectations about the common value of the item, informed by research and information. This typically results in a distribution of value estimates, with each bidders’ estimate known only by that bidder. In theory the average of these estimates is likely to be close to or equal to the true ‘hidden’ common value. Therefore, bidders’ possessing estimates above the average are likely to overbid (or in a reverse auction, ask for less) and win – as a result there is a higher chance of winning and paying too much (or receiving too little) – the winner’s curse. Rational calculating bidders are aware of this and in a single round setting would adjust bids upwards in a reverse auction, to reduce the possibility of receiving too little. 99 BCS Group owns one of the biggest terrestrial fiber networks in Africa. Its network spans from the East Coast of Africa connecting to multiple submarine cables through Mombasa in Kenya and extends to the West Coast of Africa through Muanda in the Democratic Republic of Congo and Luanda in Angola, providing capacity to land locked countries in the continent as well as ensuring network redundancy to our customers in the event of unplanned network outage. BCS is connecting up to 5,660 institutions inclusive of 3,125 schools, 1,533 hospitals and 1,002 government office as anchor tenants. See https://www.bcs-ea.com/about.html 100 Peter L. Stenberg. 2010. “American Policy and the Evolving broadband Internet Network”, Choices, 25(4). 101 CAIs are Schools, Health & Libraries Broadband (SHLB) Coalition. “Connecting Anchor Institutions: A Broadband Action Plan.” Evanston, IL: Benton Foundation, July 2016. 102 Page vi op cit. 32 “Studies show that CAIs often cannot afford to purchase the In 2018 the ITU, UNICEF and the private sector established broadband capacity they need to serve their communities. an initiative ‘Giga’105—having the goal of connecting every Policymakers can address CAIs’ financial constraints with school to the internet. Analysis for Giga shows that schools direct subsidies to CAIs, encouraging them to work together can be anchor tenants and facilitate access and digital in planning joint procurement of broadband services, and skills training across the neighbouring communities.106 In expediting review of consortium applications for funding that a wide-ranging report for Giga on connecting schools by can yield cost savings.” the Boston Consulting Group (BCG) (2021), three primary forms of funding connectivity are identified: (1) Private The World Bank has also recommended policies around the sector; (2) Government and (3) Community from which anchor tenant concept. Kim, Kelly and Raja (2010) stated numerous funding models can be derived. BCG state that a that the “Government’s main pump-priming function on the combination of funding models is needed in most countries, demand side is to serve as an anchor tenant for broadband as single solutions cannot bridge the funding gap in isolation services.”103 In the influential Moonshot for Africa plan of and note that “anchor clients stand as a good option to provide 2019, anchor tenants were recognised to form an important stable revenues and thus decrease risk”.107 part of connectivity policy:104 An illustration of the anchor tenant model is provided in “Procurements of telecommunications network infrastructure Figure 4‑1 below. and service, establishing broadband connectivity at government offices, including wide area networks and extensions to town and village centers, which can also serve as “anchor tenants” that can support costs of commercial operators’ overall network infrastructure deployment.” Figure 4‑1 Public sector anchor tenant model Local Govenment, Schools,Hospitals, Surgeries, Police,etc. Local agencies as anchor tenants: LT contract for capacity (de-risk demand) Co-invest in frastructure in return for services (lower setup cost) MNO installs new or upgrade coverage to 4G or 5G Local business Metro Fibre Network and residents Money Flow Telecom service flow Local businesses and residents: Receive higher quality services sooner and at lower cost - without anchor possibly no services Source: CEPA analysis 103 Yongsoo Kim, Tim Kelly, and Siddhartha Raja. 2010. Building broadband: strategies and policies for the developing world, World Bank. 104 Broadband Commission Working Group on Broadband for All: A “Digital Infrastructure Moonshot” for Africa, October (2019) “Connecting Africa Through Broadband: A strategy for doubling connectivity by 2021 and reaching universal access by 2030” Broadband Commission for Sustainable Development. ITU, UNESCO. 105 See https://giga.global/ 106 The State of Broadband 2022: Accelerating broadband for new realities, ITU/UNESCO Broadband Commission for Sustainable Development, September 2022, page 48. 107 BCG for Giga. 2021. “Meaningful school connectivity: An assessment of sustainable business models” Giga in collaboration with Boston Consulting Group (BCG). 33 The possibility for government agencies to act as anchor tenants within the development of communications infrastructure and services continues to grow as DCI policies on the digitalisation of taxation, education, health and identity services expand.108 Box 4: Anchor tenants and direct subsidy diagrams $ A B Subsidy Figure 4‑2 Demand and cost for communication services K DCI communications infrastructure exhibits high fixed and sunk costs (high capex) and relatively low marginal costs (low opex), giving declining average cost. In areas where demand (willingness to pay) D C is low, service may require substantial subsidy. Average Cost In Figure 4-2 the subsidy required to incentivise Marginal Cost service quantity Q* is the area ABCD which exceeds Marginal Demand consumer surplus shown as triangle KDC. Revenue Q* Quantity Figure 4‑3 Anchor tenant effect $ K DCI service demand by local government agencies Subsidy increases demand. In Figure 4-3, local government F agencies demand QT (anchor tenant effect) and have E a higher willingness to pay, reflecting social gains H G (externalities). With higher demand, scale economies result in lower unit costs and a lower subsidy EFGH is required. The subsidy is less than the consumer Average Cost surplus measured by the triangle KHG. Marginal Cost Marginal Demand Revenue QT Q* Quantity $ Subsidy Figure 4‑4 Benefits of an anchor tenant QT of DCI services are purchased by the anchor Community consumer surplus tenant. By increasing demand for connectivity, average costs are lowered and this enables the Subsidy absent anchor tenant community to enjoy a consumer surplus directly. The subsidy EFGH is shown in orange in Figure 4-4. The subsidy without an anchor tenant exceeds the Average Cost subsidy applied with an anchor tenant. Marginal Cost Marginal Revenue Demand QT Q* Quantity 108 Digital identities issued by government are also a person’s legal identity – such as passports, birth certificates, and national identity cards. These identities are usually required to access services offered by government and private providers; for example, travel, bank accounts, and educational opportunities. The main driver for digital ID globally is the aspiration to account for every person in line with the sustainable development goals (SDGs). An indicator for SDG 16.9 is to issue a legal identity to every person by 2030. The World Bank is helping to fund a national digital ID programme being piloted in Madagascar. The system is expected to facilitate access to public ser- vices and the technology procured for the national ID project will form part of the upgrade the country’s digital infrastructure – enabling government agencies to become anchor tenants. See https://www.biometricupdate.com/202211/madagascar-signs-with-mosip-for-national-digital-id-pilot-plans-1k-enrollments 34 Box 5: Digital Malawi Program Phase 1 “Digital Foundations Project” use of anchor tenants Background: Malawi ranks poorly compared to its peers in the development of its telecommunications market, being ranked 168 out of 175 countries in the 2016 International Telecommunication Union’s Global ICT Development Index. It has extremely low mobile and internet penetration, high cost and low-quality ICT services, and a large digital divide between rural and urban areas. Lack of knowledge of how to use the internet and affordability issues act as the main barriers to internet access. Approach: The project (2017 to 2024) aimed to build the digital foundations needed to help Malawi connect to the global digital economy. Access to the internet will be expanded by making internet connections more affordable, reliable and higher quality in all parts of the country. Importantly, the project aimed to leverage significant private sector infrastructure investment and support regulatory and policy measures aimed at increasing competition in the ICT sector. Government procurement was moved online which improved transparency and efficiency. The project was implemented by the Public-Private Partnership Commission (PPPC), with other key players including the Ministry of Information and Digitalization (MID), the Malawi Research and Education Network (MAREN) and the Malawi Communications Regulatory Authority. Funding came from the World Bank with a US$72.4 million International Development Association (IDA) credit. The tender process was usually a single stage process. Bidders were evaluated (e.g., see contract for design, installation, and commissioning of a data centre) to identify a preferred supplier. The main model used was a pre-purchase of international connectivity for government, under a 10-year Indefeasible Rights of Use (IRU) contract, where the government acts as an anchor tenant to encourage investment from the private sector. Through a competitive tender process, SimbaNet of Kenya was selected to invest in a fibre backbone network under the earlier RCIP project (2009-2016) and three companies were selected for different lots to provide international bandwidth, BengolNet umentto provide last mile-connectivity in the South and North of the country, and Datacon/Luna to serve the Central region. Lessons: The anchor tenant model was effective because it engaged and benefitted both the public and private sectors. Public funds are used to benefit the public sector, which itself can have positive spillover effects for society (e.g., improved educational outcomes), but also later benefits the private sector by enabling their use of the fibre backbone. This can promote local economic growth and productivity and drive innovation. A key feature of the project design is open access, which reduced the price of access to the network for the government and also to private operators. The wholesale price fell from over US$10,000 per Mbit/s to US$481 Mbit/s under RCIP, and has subsequently fallen to below US$10 per Mbit/s under Digital Malawi. The contracts were won by private sector firms which helped with efficient network planning and operation, and achieved ‘matching investments’ where firms invested their own funds alongside the government contracts. The competitive tender for contracts could be improved. For contracts we examined, the bid price range was very large (often millions of USD for bigger contracts). A multi-round process, by enabling price discovery, would make it more likely winning bids reflected the true value of the contract, ensuring better value for money. This is proposed for use in the new Digital Malawi Acceleration Program (P505095), approved in June 2024. This model has been adapted to DCI to leverage private sector investment and can be used as part of direct subsidy arrangement. Anchor tenancy involves a government agency being tied into a long-term agreement, with guaranteed revenue for an ICT provider acting as a commercial incentive for investment. For example, a number of local councils in the UK have used the anchor tenant model to improve broadband access. West Sussex County Council signed a 30-year Indefeasible Right of Use (IRU) with CityFibre agreeing to build fibre network access to 152 public buildings.109 The argument for the use of the anchor tenant model in conjunction with a direct subsidy is demonstrated diagrammatically in Box 4 below. Malawi and South Africa provide two examples of how the government can act as an anchor tenant for the private sector for DCI projects in Africa. 109 See https://static1.squarespace.com/static/5ef3391483c1fe1e25c1e871/t/5fa2df58b3960047e82eead4/1604509539068/C0094+-+TF+Connectivity+Invest- ment+and+Use+Cases.pdf 35 Digital Malawi Foundations Program (World Bank): Positive externalities for the public: Public funds are provides a good example of the anchor tenant model and used to benefit the public sector, which can have positive how it can be successful in attracting private investment spillover effects on society, such as better internet in DCI (see Box 5 and Appendix F.2). Part of the project access directly improving educational and health involved developing high-speed connectivity for priority outcomes. For example, Project Giga, the ITU-UNICEF public institutions, which acted as anchor tenants to initiative for schools connectivity, has achieved a 57 incentivise private network investment. These included percent reduction in the average price per Mbit/s for schools, community centers, post offices, health centers, schools. More generally, where the anchor tenant model airports, etc. For the private operator, expanding to results in deployment of DCI that would otherwise not nearby households becomes commercially viable, as the have occurred absent the anchor client, a consumer incremental cost of expanding from the nearby public surplus is generated. institution is far lower than the incremental cost of expanding the network from zero. The government as a Benefits to the economy: The prices for communications client provides some financial stability for networks and services also benefit businesses as well as household makes revenues from improving their networks more and public institutions, as they too benefit from cheaper certain. and better-quality services. Given the need to be digitally connected for most business activities in the South African Government’s “Connected Government” 21st Century, this is an important benefit. plan: provincial treasuries would act as anchor tenants to provide a degree of financial stability for start-up Limitations networks to incentivise private investment in DCI.110 In addition, the government wanted to aggregate Requires substantial oversight and governance: local council demand to become an anchor tenant Successful anchor tenant projects have also required and stimulate other government agencies to join the a focus on technical assistance, procurement services, broadband services rollout. capital deployment and contract management. Monitoring is essential to ensure connectivity is Below we consider the advantages and limitations of deployed efficiently and maintained over time. promoting investments in DCI using anchor tenants, where the latter can be used with subsidies and other Clashes between public and private sectors: Public interventions such as direct regulation, public ownership bodies may have different needs and ideas for how the models or government or public supply. infrastructure should be used compared to the private sector. There may be instances where the immediate Advantages most efficient network expansion might not be near a school or other public institution, but the anchor tenant May reduce level of subsidy: Where the anchor tenant model often requires such institutions to generate model is used in conjunction with direct subsidies, demand and lower unit costs. the level of subsidy required to be provided by the government may be lower. This is because the anchor Expansion limited to areas near public institution: tenant provides an assured higher level of demand and although relatively rare, in instances where there are through exploiting scale economies, unit costs are lower rural areas far away from public buildings such as a helping to stimulate take up of services in the wider school or hospital, the anchor tenant model is not community. The subsidy may be lower because the feasible and direct supply-side subsidies must be used funding behind the public sector anchor tenant takes either in isolation or in conjunction with some other account of productivity gains in the future. This assumes form of support, such as demand-side subsidies. that the operator that wins the bid continues to own and operate the network infrastructure and not transfer it Less common model: Using an anchor tenant to attract to the Government or local client (such as a school). private sector investment in DCI has been used relatively infrequently. There could be unknown issues with Improves commercial viability of projects: The deploying direct subsidies alongside anchor tenants, so government reduces project risk by providing a it would be advisable to initially test this approach using more assured revenue for the private operator. This a small-scale pilot to assess the effectiveness of the guarantee provides the security needed to receive to model. obtain private capital, either at a lower rate or financing that the service provider would be unable to obtain entirely. 110 https://www.westerncape.gov.za/sites/www.westerncape.gov.za/files/connected_government_-western_cape_broadband_implementation_plan.pdf 36 Operational costs. Although operations and Government preferences – if the government wants to maintenance costs usually costs only a few percentage use its own funds for DCI projects that directly benefit points of the up-front capacity costs, they can become the public sector, such as digitising government services proportionally more significant towards the end of a or providing broadband connection to rural schools, long IRU period, especially if donor funding has expired. then an anchor tenant could be used to leverage private An emerging practice here is to require operators to sector investment on top of it e.g., nearby households factor in future O&M costs in their bid price. using the school’s fibre network. Such an arrangement can bring significant efficiencies and benefits for society. If the government wants to be 4.3 Key considerations heavily involved in the project, in terms of management (e.g., joint venture) or ownership (e.g., BTO) then a PPP Whether it is appropriate to use direct subsidies, either with could meet their requirements whilst still providing or without the use of anchor tenants, to leverage private efficiency by involving the private sector. sector involvement in the rollout of DCI depends on the specific characteristics of the particular country and the Development of the telecoms market – where a market nature of the project(s) involved. When making this decision, is less developed and there is limited competition, there are three key issues which the government and MDBs especially in the case of a single monopoly provider, should consider: it will be difficult or impossible to implement a competitive subsidy allocation process. In this instance, Involvement of foreign firms – in the case that the awarding subsidies to promote DCI projects will domestic private sector does not have the relevant necessitate regulatory measures to protect against the knowledge or expertise, technology, or management abuse of market power and inefficiency. This could, for best-practices, for delivering a DCI project, then foreign example, lead to infrastructure sharing and open access firms should be sought to provide what is lacking regulations, as well as cost-based pricing regulations.111 domestically and allow for some capacity building and transfer of know-how. However, some African countries still retain minimum local equity investment requirements for telecom operators. 111 For further information on regulatory options, see chapter 5 in https://broadbandcommission.org/wp-content/uploads/2021/09/WGDigitalMoonshotforAfrica_Re- port2020-1.pdf 37 Achieving value for money 5 5.1 Procurement and competition E-procurement systems are also advantageous because they can facilitate transparent complaint procedures and It is well understood that, when procuring goods or services, produce a digital trail which helps with accountability. competition among suppliers helps to deliver better value for money. For example, Istache and Iimi (2008) demonstrated The models described in chapter 4 for facilitating private that the competition effect was underutilized in many sector investment in DCI require a procurement mechanism infrastructure projects funded by development partners, to allocate funds to the private sector. For a subsidy, the arguing “auction design, especially lot division, is crucial for procurer needs to decide how much to allocate and who reducing unit costs of infrastructure” and “competition [in] the should receive funds. Strong public procurement systems, developing world might be able to save at most 8.2 percent of with rules that encourage competitive bidding, are essential total infrastructure development costs”.112 for efficient procurement. This includes: More recently the National Audit Office (2023) in the United Transparent tender documents so that all eligible Kingdom has published a report on the efficiency of public firms can be involved with no discrimination, and with procurement models in which it states “competition can help no charge, or only minimal administrative charges, to support efficiency, innovation and quality in public services, by acquire bidding documents. allowing buyers to select the bid that can supply the optimal balance of benefits and cost. When competition is lacking or An independent procurement agency that is held to ineffective, other safeguards are required, or value for money account including in its role of setting standards and can be reduced through higher prices, inefficiencies and poorer monitoring their enforcement.116 outcomes”.113 E-procurement systems to reduce transaction costs and Competition is recognised to be an effective way of selecting thus increase efficiency and value for money. the lowest cost suppliers. Absent competition, well-designed regulatory oversight can help mitigate problems, but this Today, the typical procurement method in Eastern African requires well-resourced and experienced government telecommunications markets is largely an administrative administration. In the countries that form the focus of process. This is evident for the majority of procurements our study, competition is not always apparent and well- managed through USAFs in SSA.117 Administrative resourced and experienced government administration is an processes typically involve detailed application forms exception rather than the norm. and scoring schemes that serve as a proxy for social objectives. In practice these administrative processes often E-procurement is often advocated as one way to improve lack transparency, and many countries in Africa lack the and make more accountable procurement, especially in bureaucratic capacity or legal frameworks to oversee such developing economy contexts (see Komakech, 2016).114 The a process and ensure sufficient accountability.118 This can World Bank has published a number of benchmarking studies mean that, in the projects of interest in Eastern Africa, a on procurement.115 Their reports suggest that e-portals largely administrative process for allocating subsidy funds for procurement can reduce information asymmetries may not provide the best value for money and other more between parties in the procurement process which can curb competitive market-based processes, such as reverse opportunistic behaviour and unfair advantages. auctions, might better deliver value. 112 Antonio Estache and Atsushi Limi. 2008 “Procurement Efficiency for Infrastructure Development and Financial Needs Reassessed”, World Bank Policy Research Working Paper 4662. 113 Cabinet Office & National Audit Office. 2023. “Lessons learned: competition in public procurement” Session 2022-23, 19 July 2023, HC 1664. 114 Robert Agwot Komakech. 2016. “Public Procurement in Developing Countries: Objectives, Principles and Required Professional Skills”, Public Policy and Administration Research, vol, 6, no. 8, pp. 20-29. 115 See https://documents1.worldbank.org/curated/en/121001523554026106/pdf/Benchmarking-Public-Procurement-2017-Assessing-Public-Procurement-Regulato- ry-Systems-in-180-Economies.pdf 116 See https://blogs.worldbank.org/voices/value-money-public-procurement-beyond-rules-measurement 117 See GSMA (2023) “Universal service funds in Africa: Policy reforms to enhance effectiveness” October, by Kenechi Okeleke, Kalvin Bahia and Sayali Borole. 118 Jones M. 1990. “Efficiency and Effectiveness in an African Public Administration Context.” International Journal of Public Sector Management, 3(1). More recently AfDB (2014) “Summary of Literature on Fraud & Corruption in Public Procurement”. 38 However, while competitive multi-round reverse auctions To achieve value for money, governments in Eastern Africa have the potential to deliver value for money, we are mindful should focus on using methods that facilitate competition that in Eastern Africa there is limited experience of applying and price discovery. them and governments face budgetary constraints. These present challenges and raise the costs of applying multi- round reverse auctions, so policy makers need to consider 5.2 Reverse auctions carefully trade-offs involved. Notwithstanding, where a competitive multi-round reverse auction is considered Auctions are a recognised way to elicit valuations (discover appropriate, we feel it would deliver better value for money. prices) from buyers (in forward auctions) and sellers (in This is likely because administrative processes for allocating reverse auctions) in circumstances when such valuations are funds are impacted by the following. hidden and market prices not readily available as guides.119 In a reverse auction, there are many sellers who bid for the Misappropriation of funds or rent extraction Public prices at which they are willing to sell their goods or services officials might divert funds towards areas that suit to a single buyer. The winner is the seller who offers the them best or to their own pockets, and applicants lowest price. For example, in the case of allocating a subsidy for funding might extend gifts to sway decisions in for building last-mile fiber infrastructure, the winner of the their favour. Administrative processes require robust reverse auction would be the firm who asks for the lowest mechanisms in place to stop agents from abusing the subsidy amount. This contrasts with a regular forward process. Well-designed reverse auctions can limit auction in which multiple buyers bid as high as they can for opportunistic behaviour and result in more transparent an item from a single seller. and accountable procurement processes. Reverse auctions, if designed well, are superior to Complexity Allocating a small number of subsidy administrative processes in terms of achieving value for payments administratively may make sense, but in money.120 This is due to the price discovery characteristic situations where there are many micro-projects of reverse auctions, where bidders have the incentive to involved (e.g. thousands of new base stations have compete against each other to offer the lowest bid possible. featured in the Tanzania subsidisation programme for This means that the winning bid price should reflect the rural connectivity) it becomes extremely complex and true value of the object. To leverage the benefits of reverse challenging to process who should be awarded what auctions, it is important to follow best practice. In general, subsidy. Carefully designed e-procurement auctions successful auction design relies on two factors: can help overcome these problems resulting in a faster and more efficient allocation of funds. However, a Attracting, or allowing, entry – an auction will not be multi-round reverse auction with many lots can also be competitive if there are limited bidders in the auction, challenging to implement and should, at this time, be making it less likely for price discovery to occur. executed by countries in eastern Africa only if they have the support of experienced delivery partners. Preventing collusion – explicit or tacit collusion removes the competitive element of an auction which results in Relies on administrative judgement Administrative limited price discovery and an inefficient allocation. judgement is susceptible to unconscious or even conscious bias which can result in distorted outcomes. To highlight the importance of auction design, consider The difficulty of making administrative procedures the European 3G mobile-phone licence auctions that were credibly transparent undermines trust and confidence, held in the early 2000s. Auction design in some countries leading to inefficient outcomes. A reverse auction may have facilitated tacit collusion and high prices in early procedure is less exposed to such biases, as participants auctions acted to deter entrants in later auctions, apart focus on monetary implications. from existing 2G operators. Furthermore, the sequential nature of the 3G auctions in Europe (occurring in different To achieve DCI policy goals and the SDGs, and the possible countries at different times) enabled some firms (European allocation of many subsidies across numerous localities, wide incumbents) to benefit from gaming. This and other suggests that flexible and more effective decentralised factors, see below, resulted in large differences by country procurement models than those currently deployed in the revenues per pop per MHz from the auctions. These are needed. In situations where thousands of potential ranged from 20 Euros per capita in Switzerland to 650 Euros subsidy payments may be involved, procurement should be per capita in the UK, though the values of the licences sold streamlined and agile rather than bureaucratic. were similar in terms of spectrum availability. 121 119 For example, radio spectrum is often sold by auction, as the seller usually cannot appeal to the market and determine the value of the spectrum. Companies seeking ser- vice providers often do not know what the market price is as the bespoke characteristics of their requirements are not easy to value. In both scenarios auctions provide a way to elicit the hidden values of the buyer (in the case of spectrum) or seller (in the case of the service provider). Paul Milgrom (2017) “Discovering Prices: Auction Design in Markets with Complex Constraints (Kenneth J. Arrow Lecture Series)” Columbia University Press provides an excellent account of the issues involved. 120 See documents1.worldbank.org/curated/en/099651211282373597/pdf/IDU0dbf327200338f04d3e0baf708920f13a346c.pdf 121 Klemperer, P. 2002. “How (not) to run auctions: The European 3G telecom auctions.” European Economic Review, 46(4-5): 829-845. 39 Despite similar auction designs, country specific factors and There are a few important trade-offs in auction design that timing can result in significant differences in auction results, policymakers should consider. Firstly, auctions with greater for example between the very high values paid in the UK 3G levels of price discovery, such as MRRA, are typically more spectrum auction which was held at the peak of the dotcom complex to execute. Even if the format may theoretically bubble in March 2000 and lower competition in the Italian be optimal, in practice capacity and experience may not 3G spectrum auction held later in 2000 after a downward support its implementation. For example, a complex format market correction in tech stock valuations. It is important to will be more expensive to implement, and agents involved be mindful of such variations and effects when considering in the auction might not fully understand the rules or best the practical challenges for implementing reverse auctions strategy which could result in irrational strategies. in the Eastern African countries considered in this study. The key characteristics in any reverse auction are: In the context of Eastern Africa, where experience with multi-round reverse auctions is limited and budgets are Lot category – A key component of designing an auction tight, this trade-off should be carefully considered. is the areas in which the subsidy is to be offered. For example, one can choose areas that align with political Secondly, there is a trade-off between promoting administration areas to ease the administrative process, competition in the market and minimising subsidy allocation though a trade-off with service operational areas should in the auction. Minimising subsidy allocation could result also be considered. in larger incumbents receiving the bulk of the funding due to economies of scale. However, from a dynamic efficiency Lot size - How lots are dimensioned can influence perspective, it might be better to allow smaller entrants to the way bidders approach bid strategy and impact win so as to promote competition in the future. The procurer on auction outcomes. The level of granularity needs must make clear what their main intentions are for the to be selected to best suit the needs of the country in auction to be able to properly tailor the design to achieve question. The simplest case would be having one subsidy potentially conflicting aims. that covers all the desired areas which eases auction logistics. Disaggregating lots allows bidders to assemble Single round reverse auction their preferred packages.122 A single round reverse auction is the simplest reverse auction Maximum Allowable Subsidy (MAS) – The MAS should to organise. In Tanzania, the Universal Communications be sufficiently high to attract credible interest and not Service Access Fund (UCSAF) has used this format using deter participation, but not too high to avoid poor value a first-price rule for allocating subsidies under the Digital for money. A higher MAS should attract more interest Tanzania Project to extend rural mobile and broadband but may result in successful applicants receiving funds coverage over the last decade. In eleven funding rounds, on far above what is needed. average 61 per cent of areas received subsidy funding, with the rate of success declining to roughly 45 per cent in the Pricing rule – Determines what the winning bidders last four rounds. It is estimated that competitive bidding receive in terms of the subsidy. Rules can be first-price, saved the Government US$3.3 million (see Appendix F.1 for where bidders receive as bid, or second-price, where further discussion on these auctions) and there was further the winner receives the amount equal to the nearest scope for saving where government selection was used, in a unsuccessful bidder.123 second round to the auction. 122 Increasing the number of lots in an auction offers more discretion for bidders but comes at a risk of additional computational complexity, especially in regard of bid strate- gy, winner determination (depending on how bids are expressed) and valuation. In some circumstances this can result in inefficiency. In an econometric analysis of the US ‘C Block’ spectrum auctions of 1995-96, in which the FCC offered 480 distinct licence areas, Fox and Bajari (2013) argued that had the FCC chosen fewer licensed areas the outcome would have been more efficient. 123 Where the winning subsidy is calculated with reference to the second lowest subsidy requested, this is known as a Vickrey auction or sealed-bid second-price auction. In a Vickrey sealed-bid reverse auction bidders’ submit written bids without knowing the bid of the other participants in the auction. The lowest bidder wins, but the subsidy received by the winner is the second-lowest requested subsidy. In a Vickrey auction, bidders’ are incentivised to bid at their willingness to accept (i.e., the lowest subsidy each bidder needs to cover its cost of capital) and are not influenced by views on other bidder’s bid amounts. In other words, bidders’ request subsidies according to their true requirements and this leads to an efficient outcome – the bidder requesting the lowest subsidy wins. 40 Box 6 discusses the Kosovo Digital Economy (KODE) project and its use of single-round reverse auctions for procuring contracts.124 Also see Appendix F.3 for the TTL interview for the project. Box 6: Kosovo Digital Economy (KODE) project Background: Kosovo is a mountainous European country with a low population density which means that telecom network operators were unable to provide good connectivity on a profitable basis. Rural populations had particularly poor connections with around 200 villages totally unconnected. This means that there was a large geographical divide in the access to social and economic opportunities. Remote areas were losing young families to cities due to the inability to work or study online and school and health centres struggled with poor or no internet connection. The population also had low trust in institutions which limited collaboration between the public and private sector to close the digital divide. Approach: The KODE project began in 2018 and aimed to improve access to high-speed broadband services in project areas. The Ministry of Economic Development provided strategic direction and technical oversight to the project and partnered with the Kosovo Telecommunication Regulatory Authority, University of Prizren, and the pan-European network of universities GÉANT. For component 1, a “matching grants” model was used, where private sector operators compete for a grant of up to 80% of capital expenditure for a project and the rest is provided by the operator. The rural broadband program effectively leveraged private capital and the World Bank was able to use its convening power to build trust between public and private parties. However, the ratio of private sector investment to public sector investment fell over time, from nearly a 1:1 ratio in 2019 to a ratio of 1:4 in 2021, in part due to the fact that more commercially viable projects were chosen first and the more challenging higher cost projects were left. Overall, the success of the project, in terms of building digital infrastructure, has made it a blueprint for similar World Bank projects in other parts of the world. Contracts were procured with a variety of methods. For example, request for quotations (RFQ) was used to procure the purchasing and installation of internal network equipment for schools in Prizren municipality (September 2023). 10 private firms bid in the single round sealed-bid auction competed to provide the lowest quote. All but one of the companies were based in Kosovo. The range of bid prices was large, at more than EUR  100,000 (US$108k) between highest and lowest. Lessons: The strengths of the single round process are simplicity and the fact that it is commonly used, meaning it is less costly to implement both in terms of monetary and other non-pecuniary factors. This is important, particularly in the Eastern African context, where there is limited budget and institutions have less capacity. The project was able to attract sufficient competition in bidding for contracts, which is a prerequisite for an effective auction. However, the range of bid prices was large which suggests that a multi-round process, such as BAFO or MRRA, could have been used to facilitate price discovery and ensure more effective competition. The large bid price range that was observed implies winning bids were unlikely to reflect the true cost of projects. If the bidders (firms) all bid conservatively together, then the procurer will get poor value for money. In a well-designed multi-round process, any bidder with a losing bid will have the incentive to bid higher (or reduce their quote) in the next round, up until their valuation. Alternatively, a firm might be suffering from the winner’s curse, where they overbid by overestimating the true value of the contract. A multi-round process allows bidders to learn about all the other bidders’ valuations, limiting the welfare loss from the winner’s curse. 124 See https://public.tableau.com/app/profile/basani.club/viz/KODEProject-BroadbandGrants-Oct2022/6-Averagesandtotalsiinfo for data on the KODE project. 41 The single round format involves the auctioneer announcing a MAS for each lot and requires eligible parties to submit one set of bids for funding.125 Each bidder can express no more than one bid value for each lot in which they would like to invest and is not permitted to request an amount of subsidy above the MAS. Box 7 provides an explanation of the mechanics of the single round “static” reverse auction. Box 7: Static Reverse Auction Format: Illustrative Example Ten areas (lots A, B, … ,J) are potential recipients of funding to support new communications investment. A total of $120 is available in the form of subsidy. The maximum allowable subsidy is calculated by supposing an efficient provider in receipt of this amount would achieve a normal rate of return in a lot. On this basis, assume the maximum allowable subsidy in lots A and B is $20 and for lots C through J it is $10, reflecting lower costs and/or higher demand. Assume three bidders qualify to bid for subsidy support. The ten lots fall into three categories. Each bidder forms an estimate of how much subsidy is required to make a normal return for each lot in each category. The estimate of each bidder is known only by that bidder and are shown in the table below. Bidder Lots A,B Lots C,D,E,F Lots G,H,I,J 1 22 6 9 2 24 4 13 3 17 7 8 The bidders express the lots they are interested in bidding for and make one set of bids. Bidder 1 bids $10 for lots C through J. Bidder 2 bids $8 for lots C, D, E, F and does not bid for any other lots. Bidder 3 bids $20 the lots A and B, and bids $11 for lots C thorough J. Winning bids: Lots A and B are won by bidder 3 for $20 each. Lots C, D, E, F are won by Bidder 2 for $8 each. Lots G, H, I, J are won by Bidder 1 for $10 each. Of the total $120 funding available, $116 is disbursed. The bid success rate is defined as the proportion of lots receiving subsidy over total lots in the auction which in this case is 100%. Achieving a 100% bid success rate involved using 96.7% of available total funds. If a country currently uses a non-competitive process to allocate funds or if they want to introduce more competition into the market, a single round sealed-bid reverse auction could be advantageous. In certain cases, using a single round sealed- bid reverse auction is more practical and effective than using more complicated versions of reverse auctions. Single round sealed-bid auctions have some key theoretical advantages, though they depend critically on how the winning subsidy amount is calculated in relation to bids received. For example, there is a substantial difference between a single round sealed-bid auction where the winning bidder receives a subsidy as bid, versus a winner receiving the second lowest subsidy bid (in a second-price auction). 125 The MAS is the equivalent in concept to the role played by the reserve price in an auction. 42 Simplicity – as discussed above in Box 6 in the case of Winner’s curse –the cost of infrastructure build for the KODE project, a single round auction is the simplest DCI projects may in some circumstances have similar to implement. Bidders and sellers are less likely to make cost components across bidders, it is reminiscent of mistakes when competing because the format is easy to a common value auction. In this setting, bidders may understand. Where the auction is second-price, in theory ask for subsidies that are too high, lying above their this encourages all participants to bid truthfully and this individual values, compromising price discovery and is desirable for efficiency. However, wide gaps between potentially undermining efficiency. winning requested subsidies and the next lowest subsidies may present public perception problems, Exposure risk – With multiple objects, it may be difficult which can be avoided by deploying multi-round with a single round of bidding to organise combinatorial descending clock reverse auctions.126 Furthermore, packages and as a result bidders are exposed to the winning bidders know that they will receive either zero risk of paying too much for an inferior combination or an amount bid in a first-price auction, or receive of objects if some objects that are desired are won by either zero or some amount no less than what they bid other bidders. in a second-price auction. Experience from the RCIP Tanzania WB Project (see Experience – as already discussed, single round reverse Appendix F.1) shows that a lack of experience and limited auctions are commonly used to allocate funds, so expertise can make implementing even a single round Eastern African countries can draw on wide experience reverse auction difficult. In the project, the first use of reverse to guide implementation of the format. This makes the auctions in 2012 failed to attract any bidders due to a lack of process less costly and less prone to mistakes. understanding on the part of the operators. In subsequent rounds, certain operators submitted bids higher than the No opportunity for gaming – in multiple round formats, MAS which highlights a continued lack of understanding. bidders may game and choose bids that seek to influence Further problems arose due to the participation of a state- other bidders or encourage collusion, or possibly act as a owned enterprise because none of the privately owned punishment on other bidders who choose bids perceived operators were willing to bid against it because of the fear as unfavourable. In a sealed-bid single round auction of consequences. bids are blind and such behaviour cannot occur. On the other hand, common value information is impossible to Problems surrounding complexity and lack of understanding share and a lack of price discovery means the winner’s will be worse for more involved auction formats such as curse may lead to excessive subsidy requests. Further, MRRA. Effective communication between stakeholders and if the auctioneer makes public auction result details, training is therefore vital for the success of reverse auctions. this can lead bidders to anticipate regret if they ask for too little in a first-price auction and potentially result in Best and final offer (BAFO) higher subsidy requests.127 Best and final offer (BAFO) is a simple extension of the However, single round auctions may not always be the most single round process by breaking it down into two stages. effective at achieving value for money and can result in After receiving bids in the first round, the auctioneer asks inefficient auction allocations. This is because: for the best and final offer from each bidder, allowing them to improve their offers by reducing the amount of funds they No opportunity for price discovery – as bidders submit are asking for. The information policy can vary, but typically one bid per lot or package, the single round format the bidders will be told if they are winning or not. BAFO is does not give the bidders an opportunity to learn from the middle path between a single round auction and MRRA. information about common value revealed in other bids By extending the number of rounds, a limited form of price submitted. Revealing information about bids values discovery can occur because bidders are allowed to adjust and allowing bidders to discover more about the other their bid to better reflect their true valuation, although bidders’ bids encourages competition and facilitates there are not enough rounds for proper price discovery to price discovery. occur. When there is large information asymmetry, revealing information about other bids allows bidders to update their valuation and condition future bids on this information. 126 A wide gap between the lowest subsidy request and the second lowest subsidy request may be large, which can lead to public disquiet if this is known. The latter oc- curred in a sealed-bid forward auction in New Zealand in the 1990s involving the sale of UHF TV frequencies, when one bidder won a lot by bidding NZ$7m and paying NZ$5,000. See https://blogs.cornell.edu/info2040/2022/10/31/when-an-auction-fails-new-zealand-tv-frequencies-case/ and Mueller, Milton (1993) ‘New Zealand’s Revolution in Spectrum Management’, Information Economics and Policy, vol. 5, no. 2, pp.159–77. https://doi.org/10.1016/0167-6245(93)90020-H 127 In a first-price auction, ex post regret may affect bids in a direction similar to the winner’s curse. A bidder experiences winner regret after winning a subsidy the bidder discovers they could have received more subsidy by asking for more – there is a gap between their request and the second lowest request. If the auctioneer makes auction results information available to bidders after the auction (what is called winner feedback), it could influence how bidders approach an auction. If regret matters, a bidder would ask for more subsidy to lower expected winner regret, and trade this off against lowering the probability of winning the subsidy. Filiz-Ozbay and Ozbay (2007) “Auction with Anticipated Regret: Theory and Experiments” American Economic Review, April, 54 (4), 1407–1418, proposed this possibility in both a theoretical and experimental setting. However, the notion that regret is of relevance to corporate entities can be questioned. Evidence supporting the notion of regret derives largely from experimental settings with individuals. Further, Peter Katuščák, Fabio Michelucci and Miroslav Zajíček (2015) “Does feedback really matter in one-shot first-price auctions?” Journal of Economic Behavior & Organization, vol. 119, pp. 139-152, in a series of experiments have called the effect of regret into question. 43 The complexity of the format is slightly increased by These cover three dimensions: (i) the minimum quantity of including an additional round, but not noticeably. The trade- lots a bidder is required to bid for; (ii) the maximum quantity off between complexity and price discovery for the single of lots a bidder can bid for, and (iii) an overall bid activity round vs. BAFO should be worth it, because asking for constraint. Their purpose is to promote straightforward everyone’s best and final offer is not much more complex. bidding, competition, and progression. With no activity rules, a bidder may wait till later in the auction process to Multi-Round Reverse Auction (MRRA) place a bid so that the price is not pushed too low, resulting in poor value for money for the auctioneer. Multiple round reverse auctions (MRRA) promote price discovery and have the potential to mitigate the winner’s MRRA auctions are best executed using an Electronic curse and, where well-designed, the exposure problem. The Auction Platform (EAP), in which an auction is hosted online modern default format for a MRRA is a clock auction, where allowing the auctioneer and bidders to participate remotely. the auctioneer announces prices (subsidy amounts) which This lowers transaction costs and enables a complete digital decline over successive rounds. audit trail to be maintained which increases transparency. An EAP can either be developed and maintained by the A simple clock auction format is as follows. Assume a country or procured under a licence. It is typically more cost policymaker has $100 for an area in which one operator is effective to procure an EAP under licence and outsource invited to supply service. Suppose three bidders pre-qualify its management and maintenance, since it involves highly to bid for subsidy. The auctioneer starts by announcing specialised skills that may be limited domestically. Funds support at $100 and all three bidders’ express interest. The from development institutions, such as the World Bank, can three bidders are tied at $100. The auctioneer lowers the be used to help with the procurement of an EAP. The EAP price in a subsequent round and announces $95 support. may also be hired at a regional level, such as by the regional Assume one bidder drops out, leaving two contesting the financing facilities which is proposed under the World Bank’s subsidy. After a further round at which the auctioneer IDEA program, to share costs. However, the application of announces $90, only one bidder is left in the auction. See a MRRA using an EAP will present institutional challenges, Appendix D of and the World Bank and CEPA (2023) report and require capacity building and education, both for the “Allocating universal service subsidies using multi-round auctioneer and the bidders. The end-to-end process for a reverse auctions: Telecommunications in Tanzania” for a full-scale EAP can be lengthy and the software produced detailed illustrative example of a clock auction involving needs to undergo User Acceptance Training (UAT) and multiple areas.128 penetration testing.129 The information made available to bidders is critical to If the right conditions are in place, a MRRA format offers the success of auctions. As per convention, the identity of the best opportunity for price discovery. Box 8 outlines an bidders in the MRRA is kept anonymous until the auction illustrative example of price discovery in a MRRA, building process has ended. Eligible bidders are assigned anonymous on from the example in Box 7. labels. Permitting the possibility of multiple rounds allows for price The bidders will be told when there is more demand for and possibly package discovery. This approach is superior to subsidies than funds available. Limiting information about any of the above approaches when demand exceeds supply participants is intended to minimise gaming. The auction at the reserve prices, as the outcome of the process has a should incentivise truthful bidding based on the bidder’s greater chance of attaining efficiency. The advantages of required support amounts, rather than strategically bid using a MRRA format are outlined below. based on what they perceive other bidders doing. In Eastern African countries with limited domestic competition, it is often possible for bidders to work out who is who (e.g., in the duopoly case). Therefore, it is better to attract a large number of bidders (including from abroad), if possible. Activity rules should be in place to ensure the pace of the auction moves along. These rules govern the number and types of lots a bidder can bid for in an auction. 128 See Appendix D of the report, found at: https://documents1.worldbank.org/curated/en/099651211282373597/pdf/IDU0dbf327200338f04d3e0baf708920f13a346c.pdf 129 UAT is where the auction software is tested by the client or a third-party on behalf of the client, to determine whether it can be accepted or not by the users of the soft- ware. In effect UAT is an approval process, validating the software against the auction requirements and is carried out by entities familiar with the auction requirements. Penetration testing is a security exercise where a cybersecurity expert attempts to find and exploit vulnerabilities in a computer system. 44 Price discovery – Open bidding process reveals EAPs are well suited to handling multiple lots. For information about valuations; this information instance under the Digital Malawi Acceleration Program promotes the efficient assignment of funds since bidders (P505095), approved in June 2024, at least 2,500 can condition their bids on more information.130,131 lots are envisaged – 500 government sites and 2,000 Bidders will bid more aggressively since they have schools. better information about the item’s value and the level of competition they face.132 The generally accepted However, the price discovery benefits only materialise if figures are that reverse auctions can produce savings of bidders know what to do. In the Eastern African context, between 10 – 40%.133 there are limitations and practical issues of the MRRA format, outlined below, which should be considered. They all Time savings – One of the principal advantages of arise out of the complexity of the format. MRRA are that they save time due to the compressed nature of negotiations. Negotiations that may have Complexity – The multi-round clock format is more taken weeks or months to produce price concessions difficult to understand than a single round reverse become immediate pricing decisions in the auction auction. It is also more complex to implement and ensure environment.134 it is well-designed with no perverse incentives. Bidders may struggle to know the best strategy, particularly Mitigates winner’s curse - The format enables bidders if there are many lots involved. A lack of experience to update estimates of common value based on with the format makes this more difficult to implement. information disclosed about other bidders’ estimates However, training and good preparation through mock during the bidding rounds. The winner will be less likely auctions helps. to overpay, meaning they can bid their true value. Cost – Implementing a MRRA using an EAP could be Package discovery –Bidders discover information about costly. For small value projects, the cost savings from the hidden values of combinations of lots available price discovery may not be enough to offset the higher based on information revealed during the course of the costs for paying for a license to use an EAP. This may be auction. This promotes a more efficient allocation of a factor in considering regional bidding processes, such lots. Depending on the specific format applied, exposure as under IDEA. risks and substitution risks can be mitigated. Large player advantage – Firms with large budgets Clock format and information policy reduce gaming to spend will be able to pay for tools to help them – The clock format means that bidders are not able to make the optimal decision in the complicated auction. signal using their bids. An information policy based For example, in the Indian spectrum auction in 2015, on limiting the information about who bids what Vodafone paid for a tool which enabled them to build a (anonymity) means collusion is less likely to occur as clear picture of bidder activity throughout the auction ‘cheating’ cannot easily be detected. and strategically hide their bidding behaviour.136 Small firms or new entrants may not have access to EAP provides transparency – Using an electronic this technology, reducing the competitiveness of the platform to run the auction, which the MRRA format auction. lends itself to, leaves a complete electronic record for auditing and compliance.135 130 See Milgrom, P. 2017)\. “Discovering Prices: Auction Design in Markets with Complex Constraints.” Columbia University Press. 131 Milgrom, P. R. and Weber, R. 1982. “A Theory of Auctions and Competitive Bidding.” Econometrica, 50(5): 1089–1122. 132 Cramton, P. 2002. “Spectrum Auctions,” in Martin Cave, Sumit Majumdar, and Ingo Vogelsang, eds., Handbook of Telecommunications Economics, Amsterdam: Elsevier Science B.V., Chapter 14, 605‐639. 133 Tassabehji, R. 2010. “Understanding e-auction use by procurement professionals: motivation, attitudes and perceptions.” Supply Chain Management: An International Journal, 15(6), 425–437. 134 Carter, C. R., Kaufmann, L., Beall, S., Carter, P. L., Hendrick, T. E., & Petersen, K. J. 2004.”Reverse auctions: Grounded theory from the buyer and supplier perspective.” Transportation Research, 40(3), 229–254. 135 Shalev, M.E. & Asbjornsen, S. 2010. “Electronic reverse auctions and the public sector—Factors of success.” Journal of Public Procurement, 10(3), 428–452. 136 See https://www.smithinst.co.uk/case-studies/adaptive-bidding-strategies/ 45 Box 8: Price Discovery and Multi-Round Auctions: Illustrative Example Assume the same conditions as in Box 7. The auctioneer starts in round 1 by asking whether a bidder would accept a subsidy at an amount announced by the auctioneer. If demand for subsidies exceeds funds available, the auction proceeds to round 2 in which the auctioneer announces lower subsidies than in round 1 and asks the same question again. The auction continues until demand for subsidy equals or is less than the supply of subsidy. Bidding in each category ends when the number of bidders is 1 or less. Process Round 1: the auctioneer starts by asking whether bidders would accept 30 for A and B, and 15 for the other lots. Each bidder responds yes. As demand for subsidy exceeds the funds available, the auction proceeds to round 2. The auctioneer discloses information to the bidders before round 2 starts that demand in round 1 for each lot was 3 bidders per lot. Each bidder therefore discovers before the start of round 2 that the other two bidders were prepared to accept the announced subsidy in round 1. This discovery may affect the bidders’ estimates of required subsidy to make a normal return. Round 2: the auctioneer lowers the subsidy to 25 and 13 respectively. All three bidders respond yes. Again, demand for subsidies exceeds total funds available and the auction proceeds into round 3. Bidders discover demand remains at 3 in each lot. Round 3: the auctioneer reduces funding to 20 and 10 respectively. Bidder 1 lower demand and bids for lots C to J. Bidder 2 lowers demand and bids only on lots C, D, E, F. Bidder 3 bids for every lot. Thus, only Bidder 3 says yes to lots A and B so bidding on lots A and B ends. Round 4: the auctioneer reduces the subsidy for lots C to J to 9 respectively. All bidders submit bids on C, D, E, F. Bidders 1 and 3 bid on G, H, I, J. Round 5: the auctioneer announces 8 for the other lots C through J. Only Bidder 3 bids for lots G, H, I, J. All bidders bid for lots C, D, E, F. Round 6: the auctioneer changes the subsidy available for C, D, E, F only and it is reduced to 6. Bidders 1 and 2 bid for the subsidy and each recognizes that there is at least one other bidder wanting to receive subsidy. Round 7: and lowers the amount of subsidy to 5. In this instance, only Bidder 2 submits a bid. The auction ends. Results Bidder 3 receives $32 for G, H, I, J and $40 for A and B. Bidder 2 receives $20 for C, D, E, F. The total amount of subsidy requested is $92 and the bid success rate is 100%. Achieving a 100% bid success rate involved using 76.7% of available total funds. The example shows that the total subsidy used is below that for the static reverse auction in Box 1 which also had the same bid success rate 100%. This stylized example suggests price discovery, only possible with multiple rounds, appropriate disclosure of information to bidders and well-designed bidder activity rules, has the potential to improve auction outcomes in terms of a high bid success rate and attaining this at a lower cost. 46 Box 9: FCC Auction 904 – MRRA clock format The USD 20.4 billion Rural Digital Opportunity Fund was established in 2020 to bring high speed fixed broadband service to rural homes and small businesses not currently having high speed broadband services. The Phase I auction, which began on October 29, 2020, and ended on November 25, 2020, awarded support to bring broadband to over five million homes and businesses in census blocks that were entirely unserved by voice and broadband with download speeds of at least 25 Mbits/s. Auction 904 involved the disbursement of funds of up to USD 16 billion to subsidise broadband rollout. Auction 904, and its smaller precursor Auction 903, was a multiple round reverse clock auction featuring a descending price in which bidders submitted demands for financial support over multiple rounds. Auction 904 invited participants on a technological neutral basis and involved important financial and quality of service dimensions, where the latter had assigned weights reflecting a judgment about performance “Tier” (speed of service) and “Latency” (T+L weights). The motivation for assigning weights was to incentivise higher-speed (T) and lower latency (L) services. As we shall discuss, the weights serve to favour bidders offering higher speed and lower latency. The scale of the auction was immense, with up to USD 16 billion of universal service funding available to support service in over 5 million areas. The FCC used a descending clock auction format. In each area (or group of areas) a maximum allowable subsidy (reserve price) was set, and the auction started with a ‘percentage price’ multiplied into the reserve prices. The percentage price was common to all areas, and it was the percentage that descended round-to-round whenever the total demand for funds exceeded the available funds. The auction cleared (ended) when demand for support in aggregate was equal to USD 16 billion or less. The auction closed after: the budget cleared, and there were no areas remaining with competing bids. The outcome of the auction resulted in winning bidders receiving funding to deploy high-speed broadband to over 5.2 million unserved homes and businesses. This was a bid success rate of almost 99 percent, higher than that achieved in any of the UCSAF reverse auctions, for instance. A total of 180 bidders won auction support from the 386 qualified bidders that participated. The auction unleashed robust price competition that resulted in more locations being awarded at less cost to Americans who pay into the Universal Service Fund. Some 5,220,833 locations were assigned support in the auction had an initial reserve price of over $26 billion. Competition among bidders over multiple rounds meant that the final price tag to cover the locations was just over $9 billion. Only 56 percent of the universal service funds USD 16 billion was needed to achieve a near 100 percent bid success rate In Eastern African countries without past experience of Lessons can be drawn from international experience and implementing a MRRA but where conditions indicate it in Box 9 one of the largest reverse auctions (by value) for would be beneficial, it makes sense to pilot the format. This universal service fund allocations in telecommunications, would enable stakeholders and government institutions FCC Auction 904 in America, is described. The format to learn about the format and invest in capacity. It would implemented represents international best practice. also enable the MRRA to be designed in a way that best accommodates local conditions. For KDEAP, for instance, an initial pilot of last-mile connectivity for 213 educational institutions is planned, though eventually the goal is to cover more than 2,000. 47 5.3 Conclusion In this chapter we have identified three classes of reverse auctions for disbursing subsidy funds to support investment in DCI. They increase in complexity from a single round, one bid blind auction through to a multiple round reverse auction that may feature many rounds. In regard of value for money, we view the MRRA as the best format. However, as this format has not yet been applied in the countries which are the focus of this study, we recommend that should it be applied initially in the form of a pilot. However, the costs of implementing a MRRA may be high relative to the gains in the form of value for money because of the need to develop an EAP. In this regard, there may be merit in the WB investing in the development of a generic EAP that it could lease to governments. A possible roadmap towards a MRRA is discussed further in the next chapter. Figure 5‑1 Evaluation of Reverse Auction Formats Auction Pros Cons Appropriate scenario for deployment format Single Simple to implement Unlikely to achieve value for Two or more bidders Round No opportunity to signal money – no price discovery Limited number of lot categories with bids Hard to bid for multiple lots – Information about investment costs exposure risk well understood Lack of transparency Very limited capacity to implement auctions BAFO Extra round gives Unlikely achieve value for Two or more bidders some element of price money Limited number of lot categories discovery, albeit limited Hard to bid for multiple lots – Uncertainty about investment Simple to implement exposure risk costs Lack of transparency Limited capacity to implement auctions MRRA Price discovery ensures Relatively complex to Three or more bidders value for money implement and participate in Multiple lot categories Easier to bid for multiple Requires EAP which Uncertainty about investment costs lots increases dependency on Start with small scale pilot Provides transparency outside agencies Sufficient capacity or experience Aligns with international within government administration best practice Source: CEPA analysis 48 Policy Lessons and Recommendations 6 6.1 Key recommendations We should emphasise that when designing a competitive tender and a MRRA, although competition may be effective Wherever possible we propose that public funding of DCI is in retail markets it may not always be the case that specific allocated as a subsidy to private sector firms by way of MRRA. geographic wholesale markets will be competitive. The In this chapter we highlight the conditions under which this latter will likely depend on a whole host of factors related to recommendation holds. In particular, running a successful the specifics of what one auctions and how. MRRA is dependent on there being a sufficient level of competition in the defined market and adequate capacity in Table 6‑1 Recommendation summary table for each of the government administration. That is, in the sense that there six Eastern African countries are several firms who could realistically be in a position to Country Level of Legal Recommend use deliver the required connectivity infrastructure. In such Competition barriers of MRRA pilot cases, utilising a MRRA would help to drive an efficient allocation of subsidy funds and enable the government to achieve value for money. Partially Djibouti Yes No sufficient At a high level, the decision process facing each country Partially looking to disburse public subsidies by way of a MRRA is Ethiopia No137 Yes sufficient shown in the flow chart in Figure 6-1. Kenya Sufficient No Yes Whilst a MRRA may be perceived to be a relatively complex Madagascar Sufficient No Yes solution, training can mitigate the risk that it would be too complex for bidders and government administration to Somalia Sufficient No138 Yes participate in and for stakeholders to understand. It is best practice to first ensure the agency overseeing an MRRA South Sufficient No139 Yes has the required capacity and training to conduct a MRRA. Sudan Under the three World Bank projects considered here, the preparation of a commercial transaction manual is used as In addition to the method of disbursing funds government a disbursement condition to ensure that the rules of the needs to consider whether there is scope to aggregate procurement process are set out beforehand. demand and create anchor tenants in areas considered for gap funding. This can have the benefit of lowering the A further consideration when looking to utilise a MRRA demand for gap funding if scale economies are leveraged in is the legislative background of the relevant country and favour of the private sector. Further, if credible long-term whether its procurement rules enable the mechanism to be contracts can be written and agreed between government utilised (see Appendix A.7). For each of the six countries in and the private sector, it may ease fiscal pressures in the short this study, we provide in Table 6‑1 a summary on the level of term. For example, government could structure long-term competition in the telecoms market, whether legal barriers capacity purchase agreements with payments structured present obstacles and our recommendation whether to over years. The downside of this, however, is heightened risk allocate subsidy funds via a MRRA in a pilot. if the private sector is concerned about future policy change. 137 No mention of e-reverse auctions but no apparent legal restrictions. 138 No mention of e-reverse auctions but no apparent legal restrictions. 139 No mention of e-reverse auctions but no apparent legal restrictions. 49 Figure 6‑1 The decision process for implementing a MRRA for the use of dispersing subsidy funds for investment Identify areas where DCI gap funding is needed Is there Consider methods sufficient NO to increase competition competion and/or for gap funding apply different Identify areas auction format government Invest in agencies can be administrative anchor tenants capacity and YES training NO Has Consider YES Are there government reforming legal barriers NO capacity for, legislation to to deploying a or experience permit MRRAs MRRA? with MRRAs? YES Procure EAP Write auction Mock auction from WB or rules and & Bidder Pilot MRRA tender regulations, Training consult Source: CEPA analysis Given the use of a MRRA would constitute a step change in In countries where there may not be sufficient competition procurement procedures in each country, we propose that to run a MRRA (i.e. Djibouti), we still recommend that any MRRA is first conducted via a pilot. This is shown as the subsidy funds are allocated via auction. Instead, the BAFO end process in Figure 6‑1. In addition, as setting up an EAP model could be used to allocate funds, as long as there is would involve relatively large-fixed costs, we believe there is price disclosure after the first round. This approach has the merit in the WB investing in a generic EAP that it could lease benefit of reducing the potential for collusion by soliciting to a government looking to allocate subsidy funding via a bids in a sealed-bid manner, important given the increased MRRA, probably via a regional facility. This would require risk for collusion where there is a lack of competition, whilst the regional facility to develop the capability to develop and also achieving better value for money than a single round maintain an EAP, with World Bank support. Such capabilities auction through the additional final offer round. BAFO could be built-up internally or outsourced to specialists. has the added benefit of being simple to implement, as it is The EAP should be generic so that it can be adapted for easy for stakeholders to understand and straightforward to the particular circumstances in each country and for each implement, given a less sophisticated EAP is required. award process. This includes being able to set the number of geographic regions in each country, the number of lots to be In Appendix G we propose out how a commercial transaction auctioned and the number of bidders who will participate. manual might be structured as part of a programme to implement a MRRA in a pilot setting, and also provides guidance on the selection of an electronic auction platform provider. 50 APPENDICES 51 APPENDIX A Eastern Africa and Country Backgrounds The Eastern Africa region of SSA spans from Sudan to Uganda 3.55 Tanzania and the island of Madagascar, is home to over 400 million people with a wide range of countries in terms of Zimbabwe 4.76 levels of socio-economic development. Population growth in Average of selected Eastern African countries 7.94 some of the countries is expected to increase by 30 per cent to 2030, and up to two-fold by 2050.140 While the population China 0.42 has been rapidly growing across the region, economic growth has been uneven. Djibouti, Ethiopia, and Kenya recorded the India 0.99 highest per capita growth rates, while in Somalia economic South Africa 1.65 growth was broadly in line with estimated population growth; GDP per capita ranges from US$238 in Burundi United Kingdom 0.32 to US$2,099 in Kenya. Countries in the region are also United States 0.67 divergent in terms of their human and social development. For example, secondary school enrolment ranges from 25 Source: ITU Statistics143 per cent in Uganda to 45 per cent in Burundi.141 Figure A-1 shows that connectivity to the Internet and In 2022 around 36 per cent of Africa’s overall population had mobile device usage is significantly lower in SSA versus global access to broadband internet, chiefly via mobile devices, data. In SSA, 15 per cent of the population is not covered and the continent had one of the widest digital gender gaps by mobile broadband, whereas globally only 5 per cent are worldwide.142 A major stumbling block to adoption and uncovered. Compared to the global average of 49 per cent, usage of the Internet in Eastern Africa is affordability. While only 17 per cent of the population of SSA is connected to the competition and scale economies have helped to lower retail Internet with a smartphone. prices, Table A-1 shows that in Eastern Africa broadband data is relatively expensive and unaffordable for many. The Figure A-1: Regional connectivity and usage gap by device following sections look at the state of connectivity in the type, 2022 countries in this study. Table A-1: Mobile Data Pricing as Percentage of Monthly 100% 5% Income 15% 90% 2023 Mobile Data Pricing, 2GB as Percentage of monthly per capita gross national income 80% 26% Burundi 12.59 70% 8% Djibouti 6.58 40% 60% 5% Ethiopia 2.42 50% 8% Kenya 1.97 40% Madagascar 8.95 16% 30% 49% Malawi 8.78 4% 20% 8% Mozambique 9.04 Rwanda 2.46 10% 17% Somalia 5.11 0% Global Sub-Saharan Africa South Sudan 32.59 Connected with smartphone Connected without smartphone Smartphone but not connected Other phone but not connected Tanzania 4.44 Source: GSMA (2023) 144 140 UN World Population Prospects: Available at: https://population.un.org/wpp/DataQuery/ 141 Para. 1 in the Project Information Document (PID) for EARDIP. 142 See World Bank (2023) Results Brief “From Connectivity to Services: Digital Transformation in Africa” 26 June. 143 Available at: https://www.itu.int/en/ITU-D/Statistics/Dashboards/Pages/IPB.aspx 144 Available at: https://www.gsma.com/r/somic/ 52 A.1 Djibouti A.2 Ethiopia Djibouti is a small low-to-middle income country with a Ethiopia, a low-income country, is the second most populous population of just over 1.1 million people and a GDP of US$3.5 country in Africa with an estimated population of 123 million billion as of 2022.145 Mobile and fixed telecommunications people and a GDP of US$126.8 billion in 2022.151 Historically, services are provided by state-owned (monopoly) provider telecoms and internet services in the country were provided Djibouti Telecom. There were plans to partially privatise exclusively by a state-owned firm, Ethio Telecom, but in 2019 Djibouti Telecom by selling a minority stake in 2022, following the passage of a new law which partially liberalised with ownership of its assets transferred to the Djibouti the telecommunications sector, and the establishment of the Sovereign Fund in 2020 to facilitate this privatisation,146 Ethiopian Communications Authority (ECA), new licences but the proposition of a significant minority stake generated for private telecoms operators have been issued.152 insufficient interest.147 There is also a second privately owned operator Afrifiber that has been licensed since 2019 In 2021 the ECA awarded Safaricom Ethiopia (owned by offering a range of internet services. Safaricom plc, Vodafone Group, Vodacom Group, Sumitomo Corporation, British International Investment and the Djibouti Telecom offers a wide range of communications IFC which holds around 16% of the equity) Ethiopia’s first infrastructure services including network connectivity, data private sector operating licence for US$850m.153 In 2-023, centres and teleport facilities. This includes a significant it paid a further US$150 million for a mobile money license. joint venture with BringCom, a US-based communications The Ethiopian government stated that the move will create technology provider, called Djibouti Teleport. This private jobs for 1.5 million people and stimulate over US$8 billion company provides IP connections and international in domestic investment. Conditions of the licence included backhaul services geared at aeronautical and maritime commitments to invest on new transmission lines but satellite services. Safaricom is currently no allowed to bring in a third-party to build infrastructure, such as a specialist tower company.154 Djibouti is a strategically important country for submarine fibre cables; as of 2023 hosting nine landing points with Attitudes towards private investment have changed in more expected.148 In 2022, South Sudan, which is landlocked, Ethiopia in recent years. The government introduced the signed a memorandum of understanding (MoU) with Foreign Investment Law in 2020 to attract more Foreign Djibouti related to fibre-optic interconnection to benefit Direct Investment (FDI), a move which has enabled Ethio from Djibouti’s subsea cable connections. Djibouti Telecom Telecom to offer a 45 per cent stake to foreign investors; is part of a consortium that connected the Djibouti Africa though no takers have yet been found. Ethio Telecom also Regional Express (DARE) submarine fibre optic cable which plans to issue 10 percent of its shares on the new Ethiopian spans 5,500km.149 stock exchange. This strategy for growth appears to have arisen out of the increased competition the firm now China plays a key role in financing communications faces following the establishment of Safaricom Ethiopia. infrastructure in Djibouti, with Djibouti Telecom partnering The government of Ethiopia has also begun implementing with Chinese firms such as Huawei. Although Djibouti is the PPP Proclamation to allow for private investment in relatively well connected, with the status as the largest power generation and road construction sectors. This was fibre-optic hub in the region, services are less affordable motivated by government recognising that the private than in Ethiopia and Kenya (see Table A-1) reflecting a lack sector is essential to support the country’s economic growth of competition in the domestic sector. Nevertheless, the and improve the quality of public services, particularly in proportion of the population using the internet is slightly infrastructure.155 Joint ventures are also encouraged by the above the global average, suggesting a population with a Government of Ethiopia.156 The changing investment climate high appetite for internet access.150 has resulted in new investments in fibre transmission lines. For example, Liquid Intelligent Technologies announced in 2023 new projects linking Kenya and Ethiopia which involve participation by the Ethiopian Electric Power company.157 145 https://data.worldbank.org/country/DJ 146 See https://oxfordbusinessgroup.com/reports/djibouti/2023-report/ict/widening-the-net-already-a-prominent-regional-data-centre-player-the-country-looks-to-infra structure-projects-and-privatisation-to-build-capacity-overview/ 147 See https://www.imf.org/en/News/Articles/2022/12/16/pr22437-imf-staff-completes-2022-article-iv-mission-to-djibouti 148 The most recent landing point was Meta’s 45,000km 2Africa connecting Europe, Africa and Asia, see https://www.datacenterdynamics.com/en/news/worlds-longest- subsea-cable-lands-in-djibouti-east-africa/ 149 The members of the consortium are: Africa Marine Express, Djibouti Telecom, Golis Telecommunications, Hormuud Telecom Somalia, Somtel International, Telesom and TeleYemen. See https://itweb.africa/content/lLn14MmjadPqJ6Aa 150 The ITU notes that over half the homes in Djibouti are connected to the Internet, see https://datahub.itu.int/dashboards/umc/?e=DJI&i= 151 https://data.worldbank.org/country/ET 152 https://eca.et/ As of April 2024, more than 3,000 licenses have been issued. A further liberalisation of the ISP sector is expected in July 2024. 153 https://www.aa.com.tr/en/africa/global-consortium-wins-ethiopias-telecom-license-for-850m/2251286 154 Ethiopia raises $850m from historic telecoms auction (ft.com) 155 https://www.mofed.gov.et/programmes-projects/ppp/ 156 Ethiopia - Joint Ventures/Licensing | Privacy Shield 157 See https://liquid.tech/about-us/news/liquid-intelligent-technologies-announces-two-new-cross-border-fibre-routes/ 53 Investment in network expansion and mobile penetration Figure A-2: Fibre backbone networks in Kenya164 has increased internet coverage from 1.1 per cent in 2011 to 22 per cent in 2022, but coverage is still low.158 The history of telecom monopoly, a lack of reliable power supply (in 2020 only 40 per cent of households were electrified159) and unaffordable devices represent key challenges to Ethiopia’s further DCI expansion, and therefore World Bank funding is essential. A.3 Kenya Kenya is a lower-middle income country with a population of 54 million people and a GDP of US$113.42 billion as of 2022.160 Compared to other countries in SSA, the telecoms market is very competitive with five mobile network operators (MNOs). However, Safaricom PLC holds a substantial share of the market at 66.1 percent of mobile SIM subscriptions and 62.8 percent market share of mobile broadband subscriptions.161 This is reinforced by the market dominance of its mPesa mobile money platform. The Communications Authority (CA) of Kenya was established in 1999 by the Kenya Information and Communications Act (1998) and is the regulatory authority Kenya has a Universal Service Fund (USF),165 which is responsible for facilitating the development of the ICT funded from levies on licensees, appropriations from the sectors. Importantly, in 2020 Kenya’s National Information government, as well as grants and donations. The fund aims and Communications Technology Policy Guidelines adjusted to support widespread access to ICT services, promote the requirement for Kenyan ownership in foreign ICT capacity building and innovation in ICT services. The fund companies from 20 to 30 percent. This was reversed in April finances national projects that have a significant impact on 2023 when the President announced that they would lift this the availability and accessibility of ICT services in poor and 30 percent domestic equity requirement for ICT-licensed rural areas. For example, the fund has supported operators companies. This allowed, for instance, a license to be issued to increase last-mile connectivity in rural and underserved to Starlink, a US-owned low-earth orbit satellite operator. areas and improved broadband connectivity to public In general, Kenya has a positive investment climate, with secondary schools. Projects have so far ensured enhanced the government focusing on attracting more FDI to the mobile connectivity coverage for over 700,000 Kenyans.166 country.162 The government has allocated KES 111 billion (US$853m) for A.4 Madagascar implementing the National Broadband Strategy 2018-2023 (equivalent to 0.3 percent of GDP spent on broadband each Madagascar is a low-income country with a GDP of US$15 year) to increase access to broadband coverage and improve billion and a population of almost 30 million people, as of digital literacy. To achieve these goals, the government 2022.167 As a whole, Madagascar is a well-connected country understands that the private sector will play an important with 88 per cent mobile network population coverage role and is focusing on creating a conducive environment and four submarine cable links.168 However, rural areas for attracting private investment.163 This includes a focus on are still largely unconnected and only around a quarter of promoting further investment in fibre backbone technology, households are supplied with electricity.169 as shown in Figure A-2. 158 See ITU at https://www.itu.int/en/ITU-D/Statistics/Documents/DDD/ddd_ETH.pdf 159 Source Energy Catalyst, Country Guide: Ethiopia June 2020 160 https://data.worldbank.org/country/kenya 161 Communications Authority of Kenya (2023). See https://www.ca.go.ke/sites/default/files/2023-09/Sector%20Statistics%20Report%20Q4%202022-2023.pdf 162 https://www.state.gov/reports/2023-investment-climate-statements/kenya/ 163 See https://www.ict.go.ke/wp-content/uploads/2019/05/National-Broadband-Strategy-2023-FINAL.pdf 164 Source: Figure 1, page 15 in the Project Appraisal Document, KDEAP at https://documents1.worldbank.org/curated/en/099195503152337871/pdf/BOSIB155727ac80b51b9f718ac424208264.pdf 165 https://www.ca.go.ke/universal-service-fund 166 https://www.ca.go.ke/government-starts-search-universal-service-advisory-council-board-members 167 https://data.worldbank.org/country/MG 168 See ITU https://datahub.itu.int/data/?i=100095&e=MDG and https://www.submarinecablemap.com/ 169 See Medium (2023) “Connecting Rural Madagascar to Clean and Reliable Electricity” 54 The telecoms market is competitive and comprises of four The Federal Government of Somalia (FGS) encourages network-based companies: Airtel, Telma, Orange and FDI. Support of FDI is underpinned by the Foreign Blueline. Telma, a state-owned incumbent founded in 1896 Investment Law of 2015 which guarantees foreign investors following the first telephone line service in Madagascar,170 treatment equal to that given to domestic investors. This was privatised in 2001 following liberalisation of the sector includes provisions around compensation in the event of in 1996. This liberalisation paved the way for foreign-owned expropriation, international arbitration of disputes between subsidiaries of Airtel and Orange to enter the market. the investors and the Government, the right to remit profits However, foreign ownership in the telecommunications and access to foreign exchange. Investment incentives and sector is notionally restricted to 66 percent ownership, special tax incentives to encourage FDI are also in place.179 though, as in Kenya, Starlink has recently gained a license.171 All telco operators are privately owned, as are all businesses There is a legal framework for PPPs in Madagascar, as in the country.180 established by the Public Private Partnerships Law which enables partnerships under the BOT model and variants of it.172 This was aimed to help alleviate problems foreign invested faced when looking to invest in the country. A.6 South Sudan These problems stemmed from the perceived lack of fair and consistent implementation surrounding the laws. The South Sudan, a low-income country, is the world’s youngest government of Madagascar recommend foreign investors nation, has an estimated population of 10.9 million people look to enter joint ventures with local partners; however and was estimated to have a GDP of US$12 billion in 2015m such cooperation is not mandatory.173 The government has though this is highly affected by oli prices.181 The telecoms also announced a number of recent decrees in relation to market comprises three MNOs: MNT, Zain Telecom and spectrum, open access, and reform of the universal service Digitel Telecommunications. MNT and Zain are both fund (FDTC), which have been supported by World Bank foreign-owned subsidiaries, with the launch of Digitel in technical assistance and DECIM funding. 2021 significant given it is a South Sudanese-owned telecom company. A.5 Somalia The government stated that it played a role in ensuring Digitel operates in the country and signalled intentions to explore tax exemptions to encourage the import of Somalia is a low-income country with an estimated population network equipment and telecommunication tools.182 The of 17.6 million people and a GDP of US$8.1 billion as of government also has a partnership with Liquid Intelligent 2022.174 The telecoms market is competitive, comprising Technologies to build critical fibre infrastructure in South seven MNOs and one MVNO, several of which also offer Sudan. The connection runs between the capital, Juba, and fixed-line and internet services.175 The operators are all Uganda, with Liquid Telecom managing the infrastructure. licensed by the regulator, the National Communications 183 The first phase of the project was completed in January Authority (NCA), which was founded in 2017 with World 2020 and includes a 200km fibre backbone.184 More Bank support, and is tasked with encouraging competition generally, legislation states that foreign individuals are in the market. The competitive environment means that allowed to invest in medium and large size companies in the tariffs offered to consumers are among the lowest in the country provided that a South Sudanese individual has Africa.176 The quality of network coverage is also improving, at least a 31 percent share.185 Government officials will not in part driven by the interconnection of six telecom issue a business licence to a foreign investor without such a operators in 2023, supported by the World Bank.177 Despite local shareholder. However, foreign companies that want to high mobile penetration, internet coverage remains among establish a subsidiary in South Sudan are not subject to local the lowest in the world.178 The current focus is on developing shareholder requirements.186 a national fibre-optic backbone network. 170 https://www.telma.mg/data/corporate/2011_en.pdf 171 https://www.export.gov/apex/article2?id=Madagascar-openness-to-foreign-investment 172 Op cit. 173 Madagascar - Joint Ventures/Licensing | Privacy Shield 174 https://data.worldbank.org/country/SO 175 https://developingtelecoms.com/telecom-business/telecom-trends-forecasts/8945-competition-strong-in-somalia-despite-political-turbulence-r-m.html 176 Op cit. 177 The NCA state that “The interconnection will enable seamless connectivity between the operators and provide millions of customers with more options and competitive pricing.”https://nca.gov.so/interconnection-among-telecom-operators-goes-successful-enhancing-connectivity-and-competition/ 178 See World Bank. 2017. “Strategy and PPP Options for Supporting the ICT Sector and Broadband Connectivity in Somalia.” As recently as 2017, Somalia’s mobile telephony market had a penetration rate of about 60 pecent, but less than 3 percent of the population had broadband access, one of the lowest rates in the world. 179 https://sominvest.gov.so/why-invest/ 180 https://www.state.gov/reports/2021-investment-climate-statements/somalia/ 181 https://data.worldbank.org/country/SS 182 https://developingtelecoms.com/telecom-business/operator-news/11508-south-sudanese-owned-operator-launches-in-juba.html 183 https://www.afdb.org/sites/default/files/documents/publications/south_sudan_cfr_2023_.pdf 184 https://liquid.tech/the_first_to_bring_fibre_to_south_sudan/ 185 See the Companies Act (2012). 186 https://www.state.gov/reports/2022-investment-climate-statements/south-sudan/ 55 South Sudan has one of the lowest rates of mobile access In the financial year 2021-22 alone, the WB has developed and connectivity in the world, with a penetration rate of over 40 digitalization projects for a total of $4.5 billion in 22 percent in 2022. Poor transport conditions and high 29 SSA countries to support the building of digital economy fuel prices are major barriers for MNOs to implement and foundations, including digital connectivity, digital public maintain DCI, and violence and conflict has resulted in infrastructure, digital businesses, and digital skills. More MNOs suffering major infrastructure losses and caused recently, the World Bank has accelerated scaling up its high population displacement which means that many support to governments to enable digital transformation and have moved from areas of connectivity to areas with no aims to commit around $7 billion to Africa over the period connectivity.187 2023-25. The three projects outlined below represent the World Bank’s increasing ambition to facilitate private investment in DCI to support digitalisation in Africa. A.7 Country Summary Mobile Foreign Telecoms Independent Country USF broadband Electrification investment Key points competition regulator development climate Quasi-monopoly provider. Djibouti Relatively well connected. Recently introduced second Ethiopia MNO. Teledensity remains low. Five MNOs Kenya but Safaricom dominant. Four MNOs, high coverage but large digital Madagascar divide due to low income and poor electrification. Seven MNOs. Internet coverage Somalia among lowest in the world. Three MNOs, two foreign South Sudan owned. Very low connectivity. Legend: is high risk, is medium risk and is benign or normal risk. Risk assessment based on CEPA analysis. 187 See https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2022/10/SOUTH-SUDAN_DW.pdf 56 APPENDIX B Project Backgrounds In the financial year 2021-22 alone, the WB has developed Eastern Africa Regional Digital Integration Project over 40 digitalization projects for a total of $4.5 billion in SOP-II (P176181) 29 SSA countries to support the building of digital economy foundations, including digital connectivity, digital public The Eastern Africa Regional Digital Integration Project infrastructure, digital businesses, and digital skills. More seeks to advance regional economic integration through recently, the World Bank has accelerated scaling up its the development of cross-border digital infrastructure support to governments to enable digital transformation and and services in Eastern Africa (Burundi, Djibouti, Ethiopia, aims to commit around $7 billion to Africa over the period Eritrea, Kenya, Rwanda, Somalia, South Sudan, Sudan, 2023-25. The three projects outlined below represent Tanzania and Uganda). the World Bank’s increasing ambition to facilitate private investment in DCI to support digitalisation in Africa. Component 1: Connectivity Market Development and Integration (US$95m) - support reforms to reduce Kenya Digital Economy Acceleration barriers to the provision of cross-border telecoms services through open markets as well as broadband Project (P170941) connectivity infrastructure deployment, to increase access and affordability of connectivity for end-users. The Digital Economy Acceleration Project will be conducted in two phases to help Kenya improve its digital infrastructure Component 2: Data Market Development and and access and develop their digital economy. Digital literacy Integration (US$55m) - enable the secure exchange, remains a significant barrier to access. storage and processing of data across borders to support regional deployment and access to data-driven Component 1: Digital Infrastructure and Access services, innovation and infrastructure. (US$220m): a) Extending the reach of the backbone network; b) increasing last mile connectivity for Component 3: Online Market Development and education; c) enhancing government connectivity; d) Integration (US$25m) - support the development and strengthening the digital enabling environment; e) integration of the online market, which would enhance enhancing regional digital infrastructure. regional economic integration through trade, specifically via cross-border delivery and access of digital goods or Component 2: Digital Government and Services services. (US$104m): - (a) Digitising selected government services; b) developing the critical enablers for digital Component 4 - provide TA and capacity support for government; c) enhancing regional data governance. project preparation and implementation. Component 3: Digital Skills and Markets (US$51m): a) Eastern Africa Regional Digital Integration Project Supporting digital literacy; b) enhancing employment- ready digital skills; c) promoting device affordability; d) SOP-II (P180931) positioning Kenya as a Regional Digital Hub. The Series of Projects (SOP) development objective is to Component 4: Project Management (US$15m)– promote the expansion of an integrated digital market supporting project implementation and management of across Eastern Africa by increasing cross-border broadband regional activities. connectivity, data flows and digital trade in the region. Phase II development objective of the SOP is to advance digital • Component 5: CERC – responding to future crises. market integration in the Eastern Africa region by increasing affordable access to regional broadband connectivity and strengthening the enabling environment for cross-border digital services. The project is seeking to address gaps in regional fibre infrastructure, as shown in Figure A-3. 57 Figure A-3: Optimising the regional networks of the Horn of Africa188 The first phase of the Eastern Africa Regional Development Integration Project series of project (EARDIP SOP-1; P176181; see above) provides financing for Somalia and South Sudan and grants to two RECs, the, East Africa Community (EAC) and Intergovernmental Authority on Development (IGAD). In addition, the Kenya Digital Economy Acceleration Program (KDEAP) provides US$70m in additional regional funds for supporting Kenya’s participation in these activities in parallel with EARDIP. The plans for the second in the EARDIP series of projects (SOP-II) adds two more countries – the Republic of Djibouti and the Federal Democratic Republic of Ethiopia. The programme of projects and funds available for SOP-II sees additional funds being added to the four components described above. 188 Source: Map 1, page 7 in the Project Information Document, at https://documents1.worldbank.org/curated/en/099050423081036143/pdf/P18093108125ee- 090869a0089828ff2b8d.pdf 58 Digital and Energy Connectivity for Inclusion in Component 2: Enhancing Energy and Digital Inclusion Madagascar Project (P178701) (US$230 million) The Digital and Energy Connectivity for Inclusion in Subcomponent 2.1: Affordable off-grid solar and digital Madagascar Project aims to increase access to reliable and devices for underserved communities and marginalized affordable energy and digital services, with a focus placed on groups including underserved communities. This project differs from the other two we consider because it considers the energy Subcomponent 2.2: Digital literacy and renewable sector in addition to DCI, because energy is an enabler of energy awareness digital development and vice versa. The models considered in this report for leveraging private sector investment in Subcomponent 2.3: Support off-grid solar and DCI also apply to energy. For example, PPPs have been broadband connectivity of public institutions to used extensively around the world to involve the private increase service delivery and access for underserved sector in building energy infrastructure. In Pakistan, the communities government used PPPs and a unique credit enhancement- wind risk guarantee to attract private investment for wind Component 3: Supporting the Enabling Environment projects.189 The World Bank report “Linking Up: Public- for Green Energy and Digital Infrastructure (US$25 Private Partnerships in Power Transmission in Africa” draws million) from examples of successful PPPs in Brazil, Chile, India, Peru, and the Philippines to suggest that the private sector Subcomponent 3.1: Support for digital sector reforms can play a vital role in the financing, building, and operating power transmission lines in Africa using PPPs. 190 Subcomponent 3.2: Support for energy sector reform The project is split into 5 components: Subcomponent 3.3. Enabling environment for enhanced Component 1: Expanding Energy and Digital climate change adaptation and mitigation Infrastructure (US$160 million) Component 4: Project Management and Subcomponent 1.1: Hybridization and digitization of Implementation Support (US$20 million) isolated grids Component 5: Contingent Emergency Response Subcomponent 1.2: Deployment of digital infrastructure Component (budget to be determined) in rural areas Subcomponent 1.3: Private sector renewable energy mini grids 189 See https://blogs.worldbank.org/ppps/infrastructure-africa-s-development-ppp-imperative 190 See https://openknowledge.worldbank.org/entities/publication/201283a6-1c04-529e-a61c-5ce8df995636 59 APPENDIX C What is Digital Communications Infrastructure? As set out in chapter 1.1, Digital Communications In this study we are largely interested in funding incentives Infrastructure (DCI) may be regarded as a subset of Digital for DCI in relation to connectivity and transportation, which Public Infrastructure (DPI), which includes the solutions and comprises the physical infrastructure that carries digital systems that facilitate essential society-wide functions and data between devices, data infrastructure and services. services such as: identity verification, payments and data Therefore, we focus on that part of DCI which provides exchange platforms. However, it is acknowledged by many the foundation for a country’s digital services. Further, for that both DCI and DPI are part of an evolving concept.191 the purposes of this study we focus on leveraging private To have a clearer understanding of DCI, especially in the investment in DCI within the telecoms space, in particular context of this study, it is helpful to recognise that it forms a fibre rollout to provide last-mile or middle-mile connectivity subset of all digital infrastructure:192 and a national backbone transportation, and the deployment of mobile broadband connectivity in unserved, largely rural, Digital Infrastructure: According to the UK government, locations. at a general level digital infrastructure “underpins the digital, cultural and social infrastructures to develop places Both DCI and DPI are key pillars of the UN’s SDGs, which where people want to live, work and visit.”193 The digital are seventeen targets adopted by all UN member states infrastructure ecosystem has both physical (hardware) that aim to provide a blueprint for peace and prosperity for and non-physical elements (software) which work the whole planet.194 In 2023 the UN published an important together. Services and applications used by end-users compendium to the SDGs emphasising the important role to (real or virtual) on terminals and devices interact with be played by DCI and how lessons can be learned to ensure connectivity, transportation, storage and processing the SDGs are achieved.195 systems. Digital infrastructure can be private, closed and proprietary; or alternatively publicly accessible and The World Bank also highlights how DCI and DPI can be interoperable with open standards. leveraged to help achieve many ambitious targets across different sectors and help achieve humanitarian goals. Digital Public Infrastructure: Is that part of Digital In particular, the WB emphasises that DPI is unique and Infrastructure built for public interest, whether that important for four main, inter-related reasons:196 involves infrastructure used whole or in part by the public sector (e.g., government institutions) or services 1. It is foundational and cross-cutting. Physical and applications that provide externality benefits to infrastructure such as fibre-optic cables represent society (e.g. electronic national ID systems, government the foundation of the internet and every digital asset. e-procurement systems). DPI also involves government Virtual infrastructure to allow for identity verification orchestration and public governance and accountability. and/or making and receiving a payment are at the core of most transactions, and thus DPI prevents the need to Digital Communications Infrastructure: The focus re-invent the wheel with every new system. of this study, DCI concerns that portion of digital infrastructure that is used for voice, data and video 2. It complements and works together at policy, process, communications between individuals (one to one), and technology levels. For example, a person can use using digital technologies, such as fiber optic cables, their digital ID to exercise consent over sharing their cellular mobile communications, microwave or satellite personal data from official sources, or a small business communications. In the definition used here, it does could use payment transaction data to access cheaper not include one-to-many communications, such as credit. Physical infrastructure aimed at universal access broadcast TV or radio, and does not include equipment enables all of the population to use these virtual assets. used for data storage or analysis, such as data centers or computer servers. 191 DPI is defined by the UN as a combination of (i) networked open technology standards built for public interest, (ii) enabling governance, and (iii) a community of innovative and competitive market players working to drive innovation, especially across public programmes. The G20’s consensus on DCI defines it as a set of shared digital systems which are secure and interoperable, built on open standards and specifications to deliver and provide equitable access to public and/or private services at societal scale and are governed by enabling rules to drive development, inclusion, innovation, trust, and competition and respect human rights and fundamental freedoms. See UN (2023) https://www.undp.org/digital/digital-public-infrastructure and G20 (2023) https://www.undp.org/publications/DCI-approach-playbook. 192 See slide 4 from the G20’s DPI Playbook. Available at: https://www.undp.org/sites/g/files/zskgke326/files/2023-08/undp-the-DCI-approach-a-playbook.pdf 193 Department for Digital, Culture, Media & Sport. 2019. “Connected Growth: A manual for places working to boost their digital, cultural and social connectivity”, page 37. 194 The SDGs have a target date of 2030 and include numerous ambitious goals including ending all poverty, achieving zero hunger and taking urgent action to address climate change. See https://sdgs.un.org/goals 195 UNDP. 2023. “Accelerating the SDGs through Digital Public Infrastructure: A compendium of the potential of digital public infrastructure”. 196 See https://blogs.worldbank.org/digital-development/how-digital-public-infrastructure-supports-empowerment-inclusion-and-resilience 60 3. It enables sectoral applications to be easily built ‘on 4. Public benefit – not necessarily public ownership. top’. Government agencies and the private sector can Governments have a primary role in deciding whether focus on their core business, with the ability to innovate and how DPI is provided in the interests of the broader further boosted by common standards and open society and economy, through regulation, operation, application programming interfaces (APIs). The private and/or partnering with the private sector. sector offers innovative products that can be used in a variety of sectors. For example, the mPesa from Kenya’s An example of the many World Bank projects allocating Safaricom has spread from mobile payments to enabling public funds to promote DCI is the Log-in project operated access to other services such as energy and consumer in Georgia. This is summarised in Box C1 below. products. In India, despite the country’s public investments in DPI, Meta-owned WhatsApp became the default platform for remote schooling throughout the COVID pandemic. Box C1: Log-in Georgia Project About: In 2020, the World Bank approved a EUR 35.7 million (US$38.8m) loan for the Log-in Georgia project which aimed to improve last-mile connectivity for people, enterprises and institutions across rural Georgia and increase access to affordable high-quality broadband. The project will help connect up to 1,000 villages. Nearly 500,000 people living in regions unserved by high-quality broadband services stand to benefit from the deployment of the broadband infrastructure. Through this project, the World Bank will support Georgia’s National Broadband Development Strategy for 2020-2025. The project has three components: (i) increasing access to affordable broadband internet; (ii) promoting the use of broadband-enabled digital services; and (iii) project implementation support. So far, funds have been given to public institutions, but the hope is that the private sector will get involved in the future. How DCI relates: 1. Foundational: Fiber provides the foundation for digital services in the modern world. A good quality, fast connection is necessary for uploading and downloading large amounts of data, which is required for almost all digital services e.g., video calling, accessing online information and resources, mobile payments, etc. 2. Complementary: Broadband can be used by every part of the economy, from government services to private businesses and households. Improving connection coverage and quality can help the government focus on policy objectives such as economic growth and reducing regional inequality, for example by increasing the amount of ICT jobs in rural areas of Georgia. 3. Sectoral applications working on top: Specific sectors of the Georgian economy can build and expand their services due to the improved last mile connectivity and improved access to affordable broadband. For example, the project will promote a vast range of areas including digital financial services and e-commerce, online e-government services, remote e-learning and telemedicine. 4. Public benefit: The project will help Georgia overcome economic dualism and ensure that people in rural areas have the same access to opportunity as their urban counterparts. It will help the country harness digital technologies to increase its economic competitiveness and provide better jobs and opportunities for all its people. 61 APPENDIX D Procurement Rules Summary of key World Bank procurement rules The Procurement Documents include sufficient provisions, as agreed with the Bank, to adequately The World Bank’s Procurement Regulations197 outlines key mitigate against environmental and social (including procurement rules for entities wishing to borrow from the SEA/SH), risks and impacts; World Bank. These rules are designed to support Borrowers Contracts with an appropriate allocation of to achieve value for money and to ensure a transparent responsibilities, risks, and liabilities; procurement process. Publication of contract award information; Rights for the Bank to review procurement Section III Governance states that operations should documentation and activities; be managed through clear and transparent lines of An effective complaints mechanism; accountability and have clear definition of the roles and Maintenance of records of the Procurement Process. responsibilities of each party. According to the World Bank’s E-reverse Auction The Borrower is responsible for carrying out Guidelines,198 E-reverse auction methods should only be procurement financed by the Bank in accordance with used when: the procurement regulations. There exist a significant number of potential bidders. The Bank carries out its procurement functions, Price is an important determinant. including implementation support, monitoring and procurement oversight. The E-reverse auction199 is a particular application of a Request for Quotations (RFQ). For the auction, the WB The Bank requires the Borrower to develop a Procurement states that firms that meet the minimum qualification Plan (Section IV), which should include: criteria should receive information on: A brief description of the activities/contracts. The automated evaluation method that will be used to The selection methods to be applied. rank Bidders during the E-reverse auction; and Cost estimates. Time schedules. Any other relevant information on how the E-reverse The Bank’s review requirements. auction is to be conducted, including clear instructions The applicable Procurement Documents. on how to access and participate in the auction. Any other relevant procurement information. The WB’s preferred approach for high-value and complex An open competitive procurement approach is preferred. contracts is international competitive procurement as it can Approved methods include BAFO and E-reverse auctions. increase competition and may achieve the best value for For open competitive procurement the following money. Approaching the national market may be appropriate requirements apply: when the procurement is unlikely to attract sufficient foreign competition, or the financial and administrative Open advertising of the procurement opportunity; burdens involved are too high. In this case, the country’s own procurement procedures may be used. The procurement is open to eligible firms from any country; In guidance that particularly affects Djibouti and Ethiopia, state owned enterprises (SOE) of the borrower’s country The request for bids/request for proposals document are not normally eligible to participate in procurement. shall require that Bidders/Proposers submitting Bids/ However, when the goods, works, or services provided are Proposals present a signed acceptance at the time of of a unique and exceptional nature or the SOEs participation bidding, to be incorporated in any resulting contracts, is critical to the project, the Bank can agree to contract them confirming application of, and compliance with, the on a case-by-case basis.200 Bank’s Anti-Corruption Guidelines, including without limitation the Bank’s right to sanction and the Bank’s inspection and audit rights; 197 See World Bank (2020). “Procurement Regulations for IPF Borrowers.” Available at: https://thedocs.worldbank.org/en/doc/178331533065871195-0290022020/origi- nal/ProcurementRegulations.pdf 198 See World Bank (2005). “e-Reverse Auction Guidelines for MDB Financed Procurement.” Available at: https://documents1.worldbank.org/curated/ en/591461468162871166/pdf/883170WP0eReve00Box385191B00PUBLIC0.pdf 199 See Annex XII of World Bank (2020). “Procurement Regulations for IPF Borrowers” for details on E-reverse auctions. 200 See World Bank (2018). “A beginner’s guide for Borrowers: Procurement under World Bank Investment Project Financing.” Available at: https://thedocs.worldbank.org/ en/doc/684421525277630551-0290022018/original/BeginnersGuidetoIPFProcurementforborrowers.pdf 62 Summary of National Procurement Rules Country Procurement Institution(s) Procurement Law Restrictions on MRRA Djibouti National Commission of Public Law No. 53/AN/09/6th L on Bids submitted by envelope. No Procurement201 New Public Procurement E-reverse auctions mentioned. Margin Code202 of preference given to Djiboutian/State owned entities (up to 7.5 percent). Ethiopia Public Procurement and Federal Public Procurement Some level of detail on procurement Property Administration Directive (2010)204 options; no mention of reverse Agency203 (Federal level) auctions but no apparent restrictions. Kenya Public Procurement Regulatory Public Procurement and More detail on options in Part IX of Authority (PPRA)205 Asset Disposal Act (2015)206; PPADA including electronic reverse Public Procurement and auctions. Asset Disposal Regulations (2020) Somalia Department of Public Public Procurement Act Limited details on specific auction Procurement207 (2015)208 methods but no apparent restrictions. South Sudan Public Procurement and Disposal Public Procurement and Chapter V discusses methods including of Assets Authority Disposal of Assets Act open competitive tendering; no (2018)209 mention of reverse auctions but no apparent restrictions. Madagascar Madagascar Public Procurement Public Procurement Code A lot of government procurement Regulatory Authority (MPPRA) (2016) - LOI N° 2016-055 takes place outside the tendering process210 Article 33 (IV) states e-reverse auctions can be used under certain conditions e.g., sufficient competition.211 201 https://marchespublics.gouv.dj/commision 202 https://marchespublics.gouv.dj/procedure/1 203 http://www.ppa.gov.et/ 204 https://ethiopianlaw.weebly.com/uploads/5/5/7/6/5576668/procurement_directive_english.pdf 205 http://www.ppoa.go.ke/ 206 https://ppra.go.ke/download/the-public-procurement-and-asset-disposal-act-revised-edition-2022/ 207 https://mof.gov.so/department/public-procurement 208 https://mof.gov.so/sites/default/files/2018-09/Public%20Procurement%2C%20Concession%20and%20Disposal%20Act%202015.pdf 209 https://mofp.gov.ss/laws/PublicProcurementandDissposalofAssetsAct2018.pdf 210 https://www.trade.gov/country-commercial-guides/madagascar-selling-public-sector 211 https://www.fao.org/faolex/results/details/en/c/LEX-FAOC203722/ 63 APPENDIX E World Bank Project Task Team Leader (TTL) Interviews E1.Tanzania Task Team Leaders: For Digital Tanzania, Paul Seaden, Sara Ballan, Tim Kelly (to 2023). For RCIP Tanzania, Peter Silarsky and Rajendra Singh. Describe the project, the approach taken to DCI investment and your involvement World Bank engagements in Tanzania have followed two phases: Under RCIP Tanzania (which ran from 2009-2017), the main focus was on establishing the National ICT Backbone network (NICTBB), extending last mile connectivity (with UCSAF) and rolling out different eGovernment projects (with eGA), including for registration of births (with RITA), registration of businesses (BRELA) and digitisation of archives. The project development objective (PDO) covered reduction in the price of broadband internet and extending geographical reach. The project was rated Moderately Satisfactory at conclusion. Under Digital Tanzania, which is running from 2021 to October 2026, the same elements of Government connectivity and last-mile connectivity were retained and expanded but with some new elements introduced, notably on the creation a Digital Technology Institute, a national addressing and postcode system, expansion of eCommerce, a data center and an expansion of one-stop-shop service centers. The PDO covered increasing access to high-quality affordable broadband services and extending government delivery of eServices. The project is currently rated Moderately Satisfactory but disbursement continues to be slow. What models were used to involve the private sector? For RCIP Tanzania, three models were used for advancing Digital Public Infrastructure (DCI).: For the roll-out and expansion of the NICTBB, a conventional PPP model was used, with WB providing technical assistance and project supervision. Network roll-out was carried out by TTCL, which operates the network, and was funded by a loan from China EXIM Bank. At the time, TTCL was partially privately-owned, through an investment by Airtel, but it was fully renationalised in 2017/18. Although the first phase of the NICTBB was completed, it never really succeeded in attracting private investment, apart from in a few metropolitan area networks, and the regulatory framework actively discouraged private investors from investing in areas where NICTBB already had fiber links. The one exception to this was Halotel, the Vietnamese-owned mobile operator which was allowed to invest in fiber in rural areas after market entry, around 2015. It quickly rolled out around 18,000km of fiber, more than twice the size of the TTCL-operated fiber, but it was not allowed to resell capacity. For international bandwidth for Government, a “pre-purchase of capacity model” was used, and the tender was won by TTCL. The contract for 2 Gbit/s of capacity for Government use was concluded in 2012 and was due to expire in 2022 but had to be extended several times, under exceptional conditions, as the new bandwidth contract had not yet been awarded (see below). For last mile connectivity in rural areas, the reverse auction model was introduced, with some lots funded by UCSAF, the Universal Service Agency, and others funded from project funds. After a slow start, in which initially no bids were received from operators, the model worked well and around 2.6 million people received 2G cellular service for the first time. After project funding ended, UCSAF continued with the model with its own funds, using the reverse auction model. For Digital Tanzania, the Government had wanted to continue with a similar approach. However, further support to NICTBB was limited because, by this time, TTCL was fully-state-owned. But the other two modalities went ahead, with additional funding: b) For international bandwidth, a new approach was tried using a fixed price (US$7.5m over a period of either five or ten years), and rated criteria used to evaluate bids, based on capacity (speed), duration, service level guarantees etc. Bids that were submitted covered a wide spectrum of offers, suggesting that communication with bidders was not effective. Nevertheless, good offers were received and a consortium involving Liquid Telecom was selected. However, the tender process failed in the later stages, because of the high prices levied by TTCL for cross-connect, following the award. It was subsequently readvertised with a tighter set of criteria to try to improve competitiveness between supplier bids. c) For last-mile connectivity in rural areas, again conducted in parallel with UCSAF, a more successful outcome was achieved. Two tenders were issued, one for upgrade of brownfield sites (488 sites, from 2G to 4G) and a second tender for new greenfield sites (763 sites). Each site was treated as one lot and a reverse auction process was used again with some modifications, for instance, to require open access and national roaming after an initial exclusivity period. TTCL was initially allowed to enter the bidding process and was unchallenged in bidding to upgrade its own 2G sites to 4G. But later this approach had to be revised because TTCL should have been excluded from the process as an SOE and because of the conflict of interest with its parent organisation running the tender. Overall, the bidding process was competitive, with a good percentage of bids being competitive and below the maximum allowable subsidy. It is estimated that competitive bidding saved the Government US$3.3m (see summary table below). 64 Mobile Network 2G upgrade (Brownfield) New sites (Greenfield) Total Operator Airtel 32 111 143 Tigo 148 185 333 Vodacom 69 137 206 Total 249 433 682 Following the completion of the Bank-funded process, UCSAF went ahead and negotiated directly with operators for the allocation of the remaining sites that did not initially attract bids. Thus, this was effectively a two-round process, with UCSAF using its own funds to allocate sites in the second round as shown below: Mobile network operator Cell Sites Airtel 50 Tigo 38 Vodacom 53 Halotel 34 TTCL 100 In the WB-funded bidding rounds, the maximum allowable subsidy(MAS) was set at US$61,500 per new site and around US$22,000 for brownfield sites. The subsidies awarded by UCSAF amounted to between US$77k and US$94k. These subsidies are a lot higher than those awarded under the competitive process where, through competitive bidding, quite a few of the lots (135) were awarded below the MAS. The distribution of awards from UCSAF’s own funds look like a “evening out” processes in which all shall have prizes. TTCL received the most sites (100) even though it would only have won 4 under the competitive bidding process. On the other hand, Tigo which was the most aggressive bidder with 49% of the competitively awarded bids was awarded only 14% in the Government process. Halotel also received some sites even though, in the competitive bidding process, it consistently bid above the maximum allowable subsidy. One can conclude from that that the Government’s process of direct selection was probably not economically efficient and wasted some of the UCSAF funds compared with if a multi-round auction had been used. What were the main challenges faced to attract private sector involvement? Both phases would have benefitted from better consultation and information flow with the private sector. The first use of reverse auctions, around 2012, failed to attract any bidders because of the lack of understanding on the part of the operators. In the second round, in 2022/23, Halotel pitched all its bids at higher than the MAS, again presumably because of a lack of understanding of how the process worked. Further issues arose because none of the privately-owned operators were willing to bid against TTCL in the competitive process, presumably because of fear of consequences. Also, the cell tower operators were unwilling to bid independently of the mobile operators. What three factors contributed most to the success of each project you have been involved with? 1. The simpler procurement process – using reverse auction subsidies and pre-purchase of bandwidth, with best bid winning – worked better than attempts to set up PPP structures. The use of reverse auctions for the upgrade of 2G to 4G cell sites was innovative and is thought to have been the first time this was used in the WB. It is now being considered for wider application in WB funded lending. 2. By the second phase, the fact that at least four of the operators (Airtel, Vodacom, Tigo and TTCL) were familiar with the reverse auction process helped the competitive bidding process, though the second round of direct negotiation with Government was opaque. A fully transparent multi-round reverse auction, with price discovery, would have worked better, as documented in the WB/CEPA book on the use of multi-round auctions in Tanzania, published in 2023. 3. In the second round of bandwidth bidding, the decision to go with a fixed price revealed in advance (US$7.5m), with competitive bidding over quality criteria, seems to have worked well in generating much lower bid prices and longer IRUs. However, the wide range of offers suggests that not all bidders understood the process. 65 What three main challenges did you face which may have contributed to not fully achieving expected benefits as quickly as expected? 1. Pretty much all the competitive tendering processes were negatively affected by the participation of TTCL. Although it was partially privately-owned in the first phase, it still inherited the bad practices of a former monopoly. It was also inefficient – it was very slow in implementing the few cell tower sites it was allocated in the first round, and these are likely to have been unprofitable as the Government tends to treat it as an operator of last resort. 2. Lack of price transparency in the reverse auction process, which was initially run as a one stage process, meant the Government gained less value for money in the application of subsidies than might otherwise have been the case. Also, the use of uniform and unchanging MAS proved to be a blunt instrument, with little “learning” embedded in the process. 3. The PIU generally performed poorly, for instance by not providing sufficient opportunities for consultation with the private sector and by generating poor bid evaluation documents, sometimes with mathematical errors in calculations. Drawing on your experiences, what three factors would you change or do differently? In retrospect, it would have been better if that World Bank had confronted the TTCL problem head-on, and not appeared to endorse the business concept behind the NICTBB, as a state-owned and operated infrastructure for Government users. At the time of the first project, there were strong signs that it would be an open access network, with participation of TTCL on equal terms with other operators, but this did not turn out to be the case. The Bank’s position, that it could not support a state-owned and operated NICTBB, was weakened during project negotiations for Digital Tanzania. The use of a multi-round auction process for the cellular extension would have worked better than the modified single- round process that was ultimately used. This is well documented in the World Bank/CEPA report. In retrospect, it might have been better to have competitively selected a transaction advisor to run the procurement processes, rather than leaving it to the PIU, though it is unlikely that the Government would have accepted this. Do you think the method was successful in providing value for money? For the last-mile connectivity process, overall, the process has been a successful one and should bring services to more than three million Tanzanians that were previously outside broadband cellular coverage, in the second phase, in addition to the 2.6 million people that received cell service for the first time in the first phase. In total some 1,012 new and upgraded cell sites were provided in rural areas under Phase 2. It should be possible once the process is completed (awards were made in May 2023) to estimate the value of private sector investment that was generated from the process, but it is likely to be in the ratio of around 1:1. This is lower than under RCIP TZ, where project funds leveraged private sector investment in the ratio of around 1:2, but the economic viability of sites auctioned under Digital TZ was much lower than in the earlier process. However, it is hard to claim value for money for the international bandwidth contracts, as it became necessary to renew the expiring contract with TTCL several times due to delays in the procurement process. 66 E 2. Malawi Task Team Leader: Luda Bujoreanu, Tim Kelly, Ida Mboob (to 2021) for Digital Malawi; Doyle Gallegos, Casey Torgusson and Rajendra Singh for RCIP Malawi. Describe the project, the approach taken to DCI investment and your involvement World Bank engagements in Malawi have followed two phases: Under RCIP Malawi (which ran from 2009-2016), the main focus was on establishing Malawi’s national fiber backbone network and virtual landing stations in Zambia and Tanzania. The project development objective PDO) covered reduction in the price of broadband internet and extending geographical reach. The project was rated Moderately Satisfactory at conclusion. Under Digital Malawi phase 1, which is running from 2018 to October 2024, the focus has shifted to last mile connectivity with a project to connect 500 plus Government institutions, as well as higher education institutions. This phase also invoWlved increasing the amount of international bandwidth available for both Government and University users. The PDO covered increasing access to high-quality affordable broadband services and extending government delivery of eServices. The project is currently rated Satisfactory at conclusion. What models were used to involve the private sector? For Phase 1, the main model used was using pre-purchase of international connectivity for Government, under a 10-year Indefeasible Rights of Use (IRU) contract with Government as an anchor tenant to encourage investment from the private sector. Through a competitive tender process, SimbaNet of Kenya was selected and proceeded to invest in a fiber backbone network that extended from the Tanzania border, through northern Malawi to a termination poins in Lilongwe, and from the Zambian border to Blantyre. The modality included establishment of virtual landing points (VLPs) which enabled a total of eight companies to access international undersea cable (up from 0 at project inception) and allowed an increase in the number of internet service providers (ISPs) active in the market from 12 to 21. A key feature of project design was open access thus that the reduced price of access offered to Government was also offered to other private operators, at wholesale rates. All project targets were met or exceeded and the wholesale price of international bandwidth, for instance, fell from over US$10,000 per Mbit/s to just US$481. The Implementation Completion Report (ICR) for the project concluded that the procurement model used, which used the Government as an anchor tenant to incentivize investment, was more efficient that a PPP structure involving government ownership of infrastructure would have been. For Phase 2, a similar procurement model was used, with IRU purchase of long-term internet supply agreements, with the main difference being that nearly 600 terminational points were defined (500 plus for Government institutions around the country; 60 plus for higher educational institutions (HEIs) and the rest for Government Local Area Network (GLAN) sites in Lilongwe), as opposed to just eight termination points under Phase 1. Consistent with the focus on last-mile connectivity, the chosen sites were more diverse from a sectoral and geographical perspective, with a particular focus on universities, TVETs, teacher training colleges and medical schools, on post offices and public WiFi hotspots, such as markets, community centers and the airports, as well as government institutions in rural locations. The purchase of internet capacity was intended to incentivize the supply of last mile fiber to these institutions, which were selected through a feasibility study. The procurement for the connectivity transaction was conducted through five lots, of which two were for international connectivity to Blantyre via Mozambique and Zambia (won by Inq Digital), and three were for regional connectivity, in the South, North and Center of the country. The lots in the South and North were won by Bengol Net while in the Center it was won by a consortium of DataCon and Luna Technologies. What were the main challenges faced to attract private sector involvement? In both phases of the project lack of access to foreign currency and funds for investment was a major challenge. In Phase 1, SIMBAnet faced delays in raising funding, and the project had to be extended with the final payment being made only after project closure, using the Government’s own funds. In Phase 2, Digital Inq made the mistake of submitting its bid in Malawian Kwacha rather than US Dollars, and the devaluation of the Kwacha, by more than 40 percent during the course of the project, affected its ability to finance the international payments due to foreign bandwidth providers. For BengolNet and DataCon, a further challenge was related to high taxes on import of good and equipment, with a real tax rate (import tax plus sales tax) close to 65 percent. All three contractors in phase 2 faced cash flow challenges, not helped by the schedule of payments, which delayed payments to workers. 67 A further distraction, if not a challenge, stemmed from the fact that the Government contracted a loan from the China EXIM Bank for digital infrastructure at the same time as it agreed the credit from IDA. This led to a delay of a year in the project becoming effective, because the Government was undecided which loan to take (it eventually took both). The China EXIM Bank loan, which was carried out by Huawei without competitive tendering, is a more conventional contract in that the Government purchased equipment on behalf of ESCOM, the government-owned power utility, including a data center, fiber optic cable, electricity and fiber distribution pylons etc. Repayments of interest on the first tranche of debts to China EXIM Bank are now due and this has put ESCOM under considerable financial pressure as it has not been able to sell capacity at a profit. Indeed, for almost a year, the Minister of Information and Digitization (MID) placed a stop order on the World Bank connectivity transaction to try to oblige the contractors to purchase capacity from the ESCOM network. This was eventually resolved, when the Minister was removed, but caused long delays and the project had to be extended by two years. What three main challenges did you face which may have contributed to not fully achieving expected benefits as quickly as expected? 1. In both phases, the innovative nature of the procurement transaction introduced some delays in receiving World Bank procurement clearances. Under Phase 1, these delays related to the intangible nature of the asset being purchased (bandwidth) while under Phase 2 there was much discussion over the way in which operations and maintenance costs (O&M) would be handled, if they became due after the official closure of the project 2. Under Phase 1, O&M costs incurred after the end of the project closure would be payable by the Government itself. Although O&M costs typically amount to only 2 percent or so or project costs, because of the dramatic falls in bandwidth prices, by the end of the contract these because relatively more significant, and a burden for Government. Under Phase 2 this was handled by requiring the contractors to bundle future O&M costs into the upfront costs of the bandwidth. It remains to be seen whether this arrangement will hold once the project closes. The former Minister had requested that IRU contracts be reduced from ten to three years, but this would have defeated many of the original project objectives for sustainability. 3. Although not a problem for the Bank, the Government appears to have found the concept of private ownership a challenge, and some parts of Government would have preferred to stick with the Chinese model of purchase of equipment to be owned and operated by a state-owned enterprise (ESCOM), despite the losses and inefficiency of those operations. For Phase 1, this sticking point appears to have been the involvement of a foreign (Kenyan) company, SIMBANET, which requires Government departments to pay their debts, unlike ESCOM which is willing to continue service even in the face of non-payment. Under Phase 2, although all three contractors are from the Malawian business community, they are not perceived to be “indigenous”. Drawing on your experiences, what three factors would you change or do differently? In retrospect, under Phase 1, it would have been better to have confronted the problem of O&M costs at an earlier opportunity as these caused a lot of resentment among Government Ministries, even though they only constitute around 2 percent of total bandwidth costs. Under Phase 2, the Bank and the Government could have made more effort to have got operators involved in the bidding as opposed to contractors that were interested in providing the last-mile fiber. This may have avoided sustainability problems that could arise later. Also, if writing bidding documents again, the Bank may have insisted on climate resistant infrastructure (ie buried fiber or LEO satellite connectivity) rather than aerial fiber, which was proposed by the contractors. The aerial fiber is easier to repair, and has lower safeguards requirements than buried fiber, but it may be subject to more frequent cuts. In the case of the bid for international connectivity via Zambia and Mozambique, an alternative procurement model might have been used whereby the price was fixed and bidders invited to compete on quality (ie speed of connection) and length of contract, as well as coverage of O&M costs, rather than competing on price of connection. This approach was used to good effect under Digital Tanzania (P160766). One outcome of competing for bandwidth on price is that the bandwidth offer (around 3 Gbit/s under a long-term MOU) is relatively low. A second consequence is that the price was settled in Kwacha, rather than US Dollars, and was affected by devaluation. Do you think the method was successful in providing value for money? Yes, both phases provided good value for money with targets being easily exceeded. In Phase 1, for instance, the target of US$5,000 per Mbit/s was exceeded with an outcome price of US$481. In Phase 2, the outcome price is as low as US$10 per Mbit/s. Also, in Phase 2, the goal of 400 government sites connected was exceeded ad the total number of sites connected was 640, including HEIs. A simple comparison can be made with the ESCOM project, where no competitive tendering was used and where the SOE (ESCOM) was expected to own and operate the network. However, the data center constructed by ESCOM at Blantyre is largely unused while the fiber optic capacity is underutilised and pricing does not follow market trends. By contrast, internet capacity purchased under the World Bank model is very heavily utilised (over 90 percent during peak hours). 68 E3. Kosovo TTLs: Natalija Gelvanovska, Charles Hurpy Describe the project, the approach taken to DCI investment and your involvement KODE PAD, page 16: Component 1: Digital Inclusion (€15.38 million) will support digital inclusion through: (a) the expansion of digital connectivity through the co‐financing of deployment of high‐speed broadband connectivity in areas that have been identified as not connected or underserved, and (b) improving of the enabling environment for wireless broadband services, through the deployment of the National Spectrum Monitoring System (NSMS). What models were used to involve the private sector? Matching Grants Model. Private sector operators are invited to compete for a matching grant (initially up to 50%; and later up to 80%) towards capital expenditure resulting from implementation of network construction project in defined lot (bundle of few villages). More details could be found in GOM. KODE GOM: Subcomponent 1.1 will finance provision of grants to facilitate the deployment of broadband infrastructure of defined quality for unconnected settlements and public institutions (especially healthcare and educational institutions). The Grant Scheme will award grants to the selected beneficiaries. Awarded grants are dedicated to cover the costs of the eligible expenditure, such as deployment of passive and active broadband infrastructure and civil works. Grant amount can cover up to 80 percent (of the eligible expenditure only) of the total project’s budget. Remaining part of the project’s budget shall be matched by the selected beneficiary. Exact maximum percentage (not exceeding 80 percent) of the total sub-project/lot budget to be covered by Grant will be determined by the bidding process for each Call for Applications. By accepting the Grant, selected beneficiary will assume defined set of obligations. Terms and conditions of the awarded grant, incl. the obligations, will be defined by the Grant Agreement. Grant applications evaluated based on requested grant amount (70%) and the quality of the project plan (30%). What were the main challenges faced to attract private sector involvement? Few challenges. However, it is important to agree with the private sector on principal elements of the model and have open channel with them throughout the implementation. What three factors contributed most to the success of each project you have been involved with? 1. Understand key features of the model that are important for the private sector 2. Ensure model is fast disbursing 3. Develop trust between administrator of the model (Ministry), World Bank and private sector What three main challenges did you face which may have contributed to not fully achieving expected benefits as quickly as expected? 4. Initial size of the matching grant of 50% became insufficient 2 years into the project as villages started to be more expensive to cover (more remote with less households). Subsequently we have increased size of the matching grant to up to 80%. 5. Pricing for electricity poles rental became an issue with more remote villages (more of poles were needed and share of rental/OPEX went up). Subsequently Ministry negotiated special rate for pole rental for remote villages with electricity distribution company. 6. Throughout the implementation project faced only one complaint from the private sector which was not justified. If administrative process of the model had not been developed in detail, widely communicated, agreed on and followed strictly, then complaints could destroy the project. In the case of this project, the private sector knew that procedures are followed transparently and there was no scope for complaints. Drawing on your experiences, what elements would you change or do differently? Nothing major. Model worked well for Kosovo. Throughout the implementation we’ve only shortened the initial phase of consultation with the market regarding the forthcoming lots (instead of one month, Ministry is now consulting for two weeks). Do you think the method was successful in providing value for money? Method was successful. 69 E4. Gabon TTL: Michel Rogy, Charles Hurpy Describe the project, the approach taken to DCI investment and your involvement This activity mainly concerns the Central Africa Backbone (CAB), phase 4, parent project and additional financing (P122776, US$81m in IBRD financing and US$51m in counterpart financing), which ran from 2012-2020. The project development objective (PDO), consistent with the overall CAB program, was to increase the geographical reach and use of broadband internet and reduce services. Two other overlapping WB projects in Gabon were e-Gabon (P132824), which focussed on entrepreneurship and e-Health, and the new Digital Gabon project (P175987), which focusses on Digital ID but is yet to start disbursement. But neither has a strong connectivity component so is not considered here. What models were used to involve the private sector? Under CAB4, two PPP-type structures were used: To cover Gabon’s membership of the ACE cable, the World Bank covered the costs of membership (US$15m), under a classic “club cable” arrangement, with a focus on open access. Similar arrangements were used for other West African countries joining ACE. To extend Gabon’s national backbone, and to ensure connectivity with neighboring countries, a PPP arrangement was established whereby a special purpose vehicle (SPIN) was set up to manage the Government’s ownership of domestic cable, funded by the project. The network was constructed by a vendor (CCSI) and operated by a private operator, Axione, both of which were selected through a competitive process. So, effectively, this was a Build-Operate-Transfer (BOT) arrangement. What were the main challenges faced to attract private sector involvement? The main challenge appears to have been to deal with poor construction in the first phase and to re-negotiate the terms of the PPP during the additional financing to reflect the interests of both the private operator and the Government. As a consequence, there was a long delay between the completion of the construction of the second cable and its commercialisation. The project had to be restructured five times in total, in part as a result of these delays, requiring a lot of expense of management time, both for the client and the WB. Details are provided, as follows in the ICR: The first 1,140 km (“Phase 1”) of the fiber-optic backbone financed by the parent project has been in commercial operation since February 2019 under a PPP with private wholesale operator Axione, and national accessibility and quality of service are generally satisfactory, with a number of outages repaired promptly and certain infrastructural defects revealed in a February 2020 technical audit addressed by the authorities through the Additional Financing (AF). The performance of the PPP established under the parent project has produced particularly positive results. The parent project also delivered on the MOU and physical interconnection with Congo, with traffic being exchanged between operators across the border as of December 2020 after long-running infrastructural issues on the Congolese side were fixed. The 620-km extension of this fiber-optic backbone (“Phase 2”) financed through the AF was fully completed in March 2020 but was not commercialized as of the project’s closing date, due to long delays in the renegotiation of the terms of the original PPP between the Gabonese authorities and Axione (including the extension of the duration of the PPP, financial arrangements, and inclusion of the new Phase 2 links in the PPP. ). Negotiations around the network extension’s PPP concluded successfully in October 2020, and the revised PPP agreement was signed in March 2021, ensuring the overall continuity of the PPP after the project’s closing date. This revised PPP agreement is expected to pre-finance the repair of infrastructural defects revealed in the February 2020 technical audit of “Phase 1”. The AF delivered partially on interconnections with Cameroon and Equatorial Guinea, which still require the following before Internet traffic can be exchanged: (i) the finalization of the physical interconnection on the Cameroonian side, knowing that an MOU between the two countries was signed in 2019, and (ii) the signature of an MOU with Equatorial Guinea prior to the installation of the physical interconnection across the border, noting that the final draft of the MOU was finalized in September 2020 but its signature delayed due to COVID-19. 70 What three factors contributed most to the success of each project you have been involved with? 1. Most indicators have been achieved, and in some cases greatly exceeded, and it is unlikely that this would have been achieved in the absence of the project. Usage of the fiber is also high. The fact that 21 ISPs use the cable can also be considered a success. 2. Good regional cooperation and national coordination. There was good coordination among the various ministries and agencies involved (e.g., MCPEN, ARCEP, ANINF, SPIN, railway operator SETRAG, local governments) through a high-level supervisory structure, and the steering committee conducted regular meetings to follow up on progress and resolve pending issues. In addition, the Government did a very good job at building citizens’ awareness and securing their buy-in for the project through a sustained communications campaign. 3. Technical assistance to the regulator generally worked well, and the country’s score on the ITU’s ICT regulatory tracker rose during the course of the project. What three main challenges did you face which may have contributed to not fully achieving expected benefits as quickly as expected? 1. Negotiations over the extension of the terms of the PPP led to long delays (see above). 2. Long-term sustainability of the PPP arrangement is not assured as Government continues to own the network, even though it is operated by a private company. 3. It is difficult to tell what might have happened in the case of a fully private venture. Would the fiber have been built in the absence of Government involvement or ownership? Drawing on your experiences, what elements would you change or do differently? The ICR judges the project to have had substantial efficacy and efficiency and the overall rating is “Satisfactory”. The project had a transformational effect on the telecom sector, as evidenced by a dramatic increase in the geographical reach, usage, and affordability of regional broadband network services. High-level data provided by the private operator Axione in April 2021 reveal that revenues generated through the PPP were 20% higher than those expected at the time of signature. The project did not experience any cost overruns, but suffered from process inefficiencies that caused time overruns: slow initial implementation due to client-side delays in the national procedure for the approval of the loan agreement (as detailed in the ICR). An independent technical audit of the “Phase 1” (parent) backbone network financed by the project and completed in February 2020 revealed technical nonconformities in the construction process, and important repairs were estimated to cost around $US 2.8 million. The Bank’s team recommended that the authorities fully enforce the contractual guarantees with the vendor (CCSI) to proceed with these repairs as soon as possible, since they were needed to ensure the resilience and service quality of commercial operations in the long run. However, the Gabonese authorities and CCSI entered a dispute (still ongoing), and the amount required for repairs might have to be borne by the Gabonese authorities. Do you think the method was successful in providing value for money? Overall, this project has had a satisfactory outcome and the country’s usage of the internet has certainly risen substantially and prices have fallen. However, there remains a nagging doubt over whether things could have turned out better with a purely private-sector approach, or using the “government as anchor tenant” approach. One concern over value for money is that the actual amount of fiber constructed (1,760 km) seems very low in relation to the size of the Connectivity component (US$59.5m actually disbursed), though at least part of this expenditure was for ACE membership. 71 APPENDIX F Proposed commercial transaction manual roadmap A commercial transaction manual will need to be prepared to support the disbursement of funds aimed at promoting 6. Pre-qualification Determination: additional investments in DCI. In this report CEPA has Criteria for successful pre-qualification. advocated that market conditions in many of the countries Notify pre-qualified bidders. could merit the deployment of a competitive MRRA Publish list if pre-qualified bidders. deployed over an EAP. Below we present an outline of how Publish indicative timetable for auction training, mock such a manual might look and this is followed by guidance on auction and auction. the considerations required for procuring an EAP and EAP delivery partner. 7. Auction Phase Auction training manual. F.1. Introduction: Security protocols. Mock auction format. Overview of the purpose and scope of the manual. Procedures for submitting bids, including required Explanation of the competitive tender disbursement documentation. process. Handling of late bids, modifications and emergency Explanation of the MRRA format. alternative bid submission procedures. Role of the EAP. Auction procedures (online forms). The EAP delivery partner. Importance of transparency, fairness, and compliance 8. Winner Determination: with regulations. Evaluation of bids, criteria and weights. Determination of prices (how winning subsidy is 2. Legal and Regulatory Framework: calculated per lot). Overview of relevant laws, regulations, and policies Criteria for selecting winning bids. governing competitive tenders. Compliance requirements for both the buyer 9. Negotiation and Clarification: (government/agency) and the bidders. Procedures for negotiating with bidders, if allowed. Clarification process for addressing bidder queries. 3. Preparation Phase: Identification of procurement needs. 10. Award Decision: Development of tender documentation, including: Announcement of the award decision. - Tender notice or invitation to tender. Notification to successful and unsuccessful bidders. - Instructions to prospective bidders. - Technical specifications. 11. Contracting Phase: - Evaluation criteria for pre-qualification. Negotiation of final contract terms and conditions. - Contract terms and conditions. Signing of contracts. - Establishment of procurement committees and Performance bonds or guarantees, if applicable. procedures. Dispute resolution mechanisms. 4. Advertising and Publicity: Methods for advertising the tender opportunity. 12. Post-Contract Phase: Ensuring transparency and equal access to information Monitoring and performance evaluation. Contract administration procedures. for all potential bidders. Handling of variations or changes to the contract. Contract closeout procedures. 5. Application Submission: Deadline for bidder pre-qualification. Description of available lots (funding). 13. Documentation and Record Keeping: Procedures for submitting application, including Requirements for maintaining accurate and complete required documentation. records throughout the tender process. Archiving of documents for audit and compliance Application form templates. purposes. 72 14. Compliance and Ethics: Step 3 Legal and governance framework: Assess the legal Guidelines for ensuring ethical conduct and avoiding and governance framework, identify and resolve any legal conflicts of interest. obstacles to implementation. Of significance are the legal Procedures for handling complaints and grievances. positions on data management and privacy. For example, Anti-corruption measures and compliance with bribery can a bidding EAP system be hosted on servers outside the laws. jurisdiction of the auction? 15. Training and Capacity Building: Step 4 Technical implementation: Assess feasibility of Training programs for staff involved in the tender deploying a web-based bidding system. process. Mock auction procedures. Step 5 Decision on MRRA: Decide whether use of MRRA Capacity building initiatives to improve efficiency and is appropriate. If yes, proceed to B. If no, recommend alternative disbursement method. effectiveness. 16. Appendices: B. EAP Structure If a decision is made to hold a MRRA web-based auction, the Sample templates, forms and guidelines (e.g., tender relevant authority should seek to tender for an EAP provider notice, bid forms, auction user manual, etc.). and an EAP. Relevant laws, regulations, and policies. Licence template. For an EAP this should cover and set out the following Bank and funds transfer arrangements. requirements: Glossary of terms. Web-based: Requirement for a web based architecture 17. Contacts: with secure access for clients (bidders and auctioneer). Contact information for the procurement team and Requirement for bidders and auctioneers to connect through relevant authorities for inquiries or assistance. the web without the need for special application software. System to be designed to operate on common browsers including Chrome, MS Edge, Firefox, and Safari. Interaction F.2. Commercial transaction manual roadmap in should be through easy-to-use interfaces supported by clear relation to an electronic auction platform user manuals and help facilities. The CTM is intended to guide the tender process in its Secure access: Requirement for multiple authentication entirety and in line with international best practice it layers to provide robust security. facilitates a transparent and predictable process.212 A focus in the CTM should be on the EAP and include the following Auctioneer control: Auctioneers should be capable of (much of the material below would populate relevant monitoring all aspects of the auction process and set appendices in the CTM): parameters to shape the auction process. Audit trail: Need for EAP to produce a complete and A. Decision to use MRRA verifiable audit trail. Step 1 Identify lots: Identify projects delivering best value in closing digital gaps. Assess to what extent Anchor Tenants Modular architecture: Modular architecture allows could be structured within the process. Finalise lots available flexibility and build of specific auction rules. in an auction. Bid tracking tool: The EAP should also be a tool to enable Step 2 Market analysis and consultation: Determine the bidders preparing for the auctions to better understand extent of competition in the market and competition for auction dynamics, and try out different bidding strategies DCI funds. Organise effective private sector stakeholder and test procedures. consultation ahead of the tender process to identify the right universe of bidders, including towercos, satellite providers, mobile network operators and others. 212 See The World Bank and Asian Development Bank (2005) “e-Reverse Auction Guidelines for MDB Financed Procurement.” Available at https://openknowledge.world- bank.org/server/api/core/bitstreams/ff8316b3-201a-579f-9379-b606787dc548/content 73 C. EAP Provider and Role of EAP Delivery Partner G. Advertising and Stakeholder consultations The notification of the auction should be published on a Build or Procure? We strongly recommend that the publicly accessible website. Authority tasked with disbursing the DCI funds should The notification should include all the specifications, procure from an established EAP provider. A decision on this terms and conditions for the proposed contract. can be justified in terms of risk management and efficiency. There should be provision for consultations with EAP Provider Leads Implementation: The selected delivery potential bidders partner should manage the entire procurement process on behalf of the Authority, supervising auction preparations, H. Operation training, user manual production, workshops, the auction and post-auction assessments. The EAP provider shall provide sufficient resilience and contingencies to cope with power outages. D. Auction Structure Variables The EAP provider shall run the auction according to information specified in the invitation to the e-reverse The authority in consultation with stakeholder and the EAP auction. delivery partner should consider the following factors when The EAP shall be subject to rigorous user acceptance implementing the EAP: testing (UAP). A separate contract with a UAP provider Number of lots is likely required. Lot areas The EAP shall be tested rigorously in regard of its security Maximum Allowable Subsidy to prevent unauthorized access. This will require a third Pricing rule party penetration testing service provider. Winner determination The EAP will collect bids electronically and automatically Obligations – whether the connectivity infrastructure rank bids, without human intervention. built will be open access, restricted access, or becomes If a bidder submits an invalid bid, it will be notified online open after a set number of years immediately with a message explaining why the bid is rejected. E. EAP Development The Authority shall close the auction either at the time and date as previously published or when a previously Consider who will maintain the EAP over the long-term advertised time period has elapsed during which no new (for instance a national eProcurement Authority or a valid bids have been received. regulator), or whether the platform may be closed at the The Authority shall monitor whether there is improper end of the project. use of the reverse auction e.g., collusion. The selected EAP provider and delivery partner to run the auction process under the supervision of the project I. Information Policy implementation unit (PIU) for each national project. This could be delegated to a regulator, but as per WB Under no circumstances shall the identities of the rules it would be better that the Implementing Agency bidders be able to be disclosed or identified by any party supervise the process. during any phase of the auction. When deciding to give out additional information, the F. Bidding Specifications contracting authority shall verify that this information does not distort competition and informs all bidders The published specifications should contain information simultaneously. on all conditions and technical information needed to It shall inform bidders instantaneously of new ranking(s) participate in the auction, including: as they occur, together with price in such a way that The event and timing of the auction bidders are able to ascertain their ranking at any Rules for participation moment. Valid bid increments How to bid J. Ex-post Evaluation Whether the auction is divided into successive phases The information which will be made available to bidders Clear criteria should be specified to evaluate the success in the course of the electronic auction and, where of the tender process ex post. These include the amount of appropriate, when it will be made available to them private capital mobilisation, a good geographical allocation (funding for routes in remote areas), all the lots are filled (x% success rate); and smoothness of implementation. 74 75