## Network costs for 4G/5G deployment in SSA (content unit) ## Source details **Canonical URL:** [Network costs for 4G/5G deployment in SSA (content unit)](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019249-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2019/wpiea2019249-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019249-print-pdf.pdf.json) --- ### I. Introduction, context, and objectives - Purpose: - Estimate cost of introducing a full-scale 4G network by 2025 in SSA and an operable 5G network by 2040. - Methodological approach: - Follows Lombardo (2019) (European Broadcasting Union) assuming a non-standalone network architecture. - Uses a synthetic country (average of France, Germany, Italy, Spain, and United Kingdom) adjusted for SSA-specific characteristics. - Accounts for demographic transitions, rapid urbanization, population density projections for highly decentralized administrative areas, heterogeneity in electricity costs, current 4G coverage, number of licensed mobile operators, and potential lack of fiber backbones. - Scope limitation: - Focuses on infrastructure costs to bridge the mobile Digital Divide; affordability and user capacity issues are acknowledged but excluded from the costing scope. ### Key demographic and urbanization projections used - SSA population: projected to rise to 1.7 billion in 2040 from 1 billion today. - Labor force: projected to increase by 20 million per year over the next two decades. - Urbanization: - Urban population has on average risen by approximately 4.4 percent annually over the last 50 years; average urbanization rate now exceeding 40 percent. - Urban population expected to double over the next 25 years. - Population distribution shifts: - Share of population in rural areas projected to fall to less 60 percent by 2040 from 78 percent in 2010. - Share of population in densely urban areas projected to increase to roughly a third of total population by 2040 from below 15 percent in 2010. ### Technical framing: 4G and 5G network model - 5G architecture and approach: - Focus on heterogeneous 5G networks (non-standalone architecture) requiring concurrent operation of macro-, micro-, pico-cells. - Non-standalone approach: use existing sites/infrastructure and 4G spectrum as anchor connection with a densified network of small cells and higher-band spectrum. - Network parameters and cell geometry: - Macro-cells operate on 700 MHz; Micro-cells operate on 1800 MHz; Pico-cells operate on 2600 MHz. - Spectral efficiencies: 4G: 15 bit/s/Hz; 5G: 23 bit/s/Hz. - Cell radius by geotype: - Rural (0 – 500 pop/km2): Macro 10 km; Micro 2 km; Pico Not applicable. - Urban (500 – 1500 pop/km2): Macro 2.5 km; Micro 1.25 km; Pico 0.2 km. - Dense Urban (>1500 pop/km2): Macro 1 km; Micro 0.5 km; Pico 0.1 km. - Hexagonal cell approximation used for coverage; final calculated number of cells augmented by 30 percent (following Lombardo, 2019). ### Traffic, capacity, and area-throughput transformation - Inputs and conversions: - Current mobile traffic demand input: ITU ICT Indicators database average monthly mobile data per capita in GB. - Busy hour traffic per capita: - Convert GB to megabytes by multiplying by 1000. - Convert months to days by dividing by 30. - Assume hourly busy data consumption accounts for 20 percent of total daily traffic. - Busy hour bitrate: - Convert megabytes per hour to megabits per second by multiplying by 8 and dividing by 3600. - Network capacity requirement: - Network should provide at least 80 percent of busy hour bitrate all the time (Network capacity = Busy hour bitrate / 0.8). - Area throughput: - Multiply projected population densities by network capacities to derive area throughput. - Population density classification for planning: - Rural: less than 500 people per square kilometer. - Urban: greater than 500 but less than 1,500 people per square kilometer. - Dense Urban: exceeds 1,500 people per square kilometer. ### Scenarios and traffic benchmarks used for planning - 2020 baseline: current mobile consumption traffic assumed uniformly distributed across administrative areas. - Two 2025 4G scenarios: - Scenario 1: assume 23 percent of connections in 2025 are 4G (versus 7 percent in 2018) and impose a minimum of 0.5 GB monthly mobile traffic by 2025. - Scenario 2: world mean monthly consumption benchmark scaled to equal 7.3 GB monthly traffic in 2025 (calculation based on 2.03/5.6⋅20.1 = 7.3 GB). - 2040 5G target benchmark: - European Broadcasting Union objective: guarantee a minimum of 44 GB of monthly mobile traffic by 2040. - Smartphone adoption assumption: - Current smartphone adoption rate in SSA is 39 percent, predicted to rise to 66 percent by 2025. ### Cost components and core assumptions - Two main cost components: Capex and Opex (Pereira, 2007). - Capex: equipment costs, tower costs, installation costs; paid at the beginning of each year. - Opex: energy consumption, heat dissipation, site rental, site O&M; paid at the end of the year. - Electricity costs per kW/h: from World Bank DB16-methodology; assumed constant over next two decades. - Heat dissipation: assumed to double electricity costs ("This implies a doubling of energy consumption costs."). - Operational labor: 6–7 days annually for an expert for macro cells; 2–4 days for micro- and pico-cells. - Sharing and mark-ups: - Operators share parts of Capex and Opex (see Box 2). - Final costs augmented by 30–50 percent to account for sparse fiber backhaul and secondary costs: 50 percent mark-up for Capex and 30 percent for Opex in SSA. - Exchange rate assumption: 1.1 US$/EUR (assumed constant). - Equipment lifetimes and renewals: - Tower lifetime: around 20-30 years — does not need renewal until 2040. - Active equipment lifetime: around 5-10 years — renew active equipment in 2020, 2032, and 2040. - Renewal of 5G additional cells at least once. - Installation costs incurred for each equipment renewal cycle. - Fixed equipment price assumed from 2020–40 (no price decline modeled). ### Mobile network unit cost inputs (preserved values) - Capex per cell type: - Macro: Equipment €32k; Tower €123k; Installation €30k - Micro: Equipment €5k; Tower €20k; Installation €10k - Pico: Equipment €2k; Tower €3k; Installation €2k - Opex per cell type: - Macro: Power 100 W; Efficiency 50%; O&M €3.3k; Rental €5k - Micro: Power 50 W; Efficiency 50%; O&M €1k; Rental €2.5k - Pico: Power 10 W; Efficiency 50%; O&M €0.9k; Rental €0.01k ### Aggregate cost estimates and annualized equivalents (headline findings) - Total aggregate costs: - Cost to implement full 4G in SSA by 2025: US$14 billion. - 4G Capex: US$12.5 billion by 2025 for SSA. - 4G Opex: US$1.5 billion by 2025 for SSA. - Conditional additional cost to make 5G operable by 2040: US$57 billion. - 5G Capex: US$49 billion by 2040 for SSA. - 5G Opex: US$8 billion by 2040 for SSA. - Annualized capital and operational cost equivalents: - Full 4G (annual): US$2.5 billion (0.04 percent of 2025 SSA GDP). - Operable 5G (annual): US$3.9 billion (0.03 percent of 2040 SSA GDP). - Decomposition of annual costs: - To achieve full 4G connectivity by 2025: - Capital expenditure: US$2.1 billion. - Operational expenditure: US$0.4 billion. - Annual costs for 5G connectivity in 2040: - Capital expenditure: US$3.3 billion. - Operational expenditure: US$0.6 billion. - Financing cadence assumption: - 4G Capex compressed into 5 years prior to 2025 (higher annual Capex). - 5G Capex spread uniformly over 15 years (2026-2040). ### Per-subscriber cost metrics (median/aggregate SSA) - Per-subscriber income equivalents (median basis): - 4G connectivity in 2025: 8.4 percent of income per subscriber. - 5G connectivity in 2040: 4.9 percent of income per subscriber. - Method for per-subscriber calculation: - Annual total cost / (Population ⋅ Penetration rate) = Annual total cost / (Penetration rate ⋅ GDP) using projected GDP, population, and linear extrapolation of GSMA Intelligence 2017 penetration rates to 2025 and 2040. - Appendix V (selected country values): - Examples (Total Cost per Subscriber — Percent of income per capita): - AGO: 4G 8.66; 5G 4.81 - BDI: 4G 451.08; 5G 43.58 - BEN: 4G 4.06; 5G 2.89 - CIV: 4G 1.84; 5G 1.77 - KEN: 4G 5.85; 5G 2.35 - NGA: 4G 2.80; 5G 2.02 - TZA: 4G 19.75; 5G 5.25 - ZAF: 4G 0.42; 5G 1.16 - ZWE: 4G 4.54; 5G 7.60 ### Heterogeneity across SSA countries and fiscal burdens - Significant variation driven by current infrastructure, population growth, urbanization rates, and real GDP growth projections. - Examples of annual Capex as percent of GDP: - 4G (2025): - Mauritius and South Africa: less than 0.004 percent of GDP annually. - Burundi: total annual cost reaches 2 percent of GDP. - 5G (2040): - Mauritius, Seychelles and Ghana: less than 0.01 percent of projected GDP annually. - Burundi: 0.36 percent of projected GDP annually. - Central African Republic: 0.85 percent of projected GDP annually. - Appendix IV-A (selected Total Capital Expenditures — Millions of USD): - AGO: 2025 Ericsson 551.89; 2025 GSMA 545.34; 2040 Ericsson 1000.27; 2040 GSMA 997.00; 2040 Ericsson (decreased radius) 2541.72; 2040 GSMA (decreased radius) 2538.45 - NGA: 2025 Ericsson 1083.07; 2025 GSMA 1072.62; 2040 Ericsson 3850.62; 2040 GSMA 3845.40; 2040 Ericsson (decreased radius) 4986.14; 2040 GSMA (decreased radius) 4980.91 - TZA: 2025 Ericsson 1677.96; 2025 GSMA 1674.69; 2040 Ericsson 2266.60; 2040 GSMA 2264.96; 2040 Ericsson (decreased radius) 3256.28; 2040 GSMA (decreased radius) 3254.64 ### Operational expenditures (selected country figures) - Appendix IV-C (Operational Expenditures — Millions of USD): - AGO: 2020 0.78; 2025 Ericsson 9.83; 2025 GSMA 9.80; 2040 9.25; 2040 (decreased radius) 14.25 - KEN: 2020 7.11; 2025 Ericsson 29.74; 2025 GSMA 29.58; 2040 30.27; 2040 (decreased radius) 39.86 - NGA: 2020 11.26; 2025 Ericsson 28.38; 2025 GSMA 28.26; 2040 24.67; 2040 (decreased radius) 44.71 - TZA: 2020 7.99; 2025 Ericsson 63.83; 2025 GSMA 63.79; 2040 64.19; 2040 (decreased radius) 73.71 ### Policy conclusions and recommendations (preserved formulation) - Infrastructure alone is insufficient; complementary policies needed to address affordability and knowledge gaps. - Key policy actions to complement infrastructure spending: - Public-private collaboration and updated Technological Infrastructure Policy to ensure widespread implementation beyond urban upgrades. - Endow the poor with smartphones in rural and undeveloped areas of SSA to expand benefits of 5G adoption. - Reconsider taxation of mobile phone-based transactions that currently slows the mobile revolution in SSA. - Encourage competition. - Implement appropriate licensing frameworks to achieve inclusive coverage. - Address tax issues that hamper the provision of mobile services. - Regulators and policymakers should ensure affordable connectivity for everyone, including addressing the gender divide. - Use directed subsidies in rural areas and by gender (e.g., free SIM cards, costless subscriptions, incentives to increase affordability of smartphones). - Trade-offs and policy questions highlighted: - Who should bear the cost — public sector or consumers/firms? - Should countries with significant mobile connectivity costs consider delaying 4G connectivity implementation to 2030 to reduce annual Capex needs at the cost of delaying catch-up? - Consider competing spending needs (education, health, electricity access, water and sanitation) and the sizable additional spending requirement for SSA to meet the SDGs by 2030 (estimated at US$400 billion). ### Model limitations and upside risks - Estimates do not include costs from signal loss caused by topology (mountains, deserts) or interference; constant throughput across service areas is assumed. - Calculated costs should be viewed as a lower bound. - More sophisticated models (SEAMCAT, nonlinear constrained discrete optimization approaches) could refine estimates but require data unavailable for all SSA countries. - Final costs may underestimate expenses in low-income areas due to assumed expert availability and fixed equipment prices. *Source: IMF Working Paper WP/19/249.* ### 8.4 percent of annual subscriber income, on a median basis, by 2025 for 4G and 4.9 percent of ### I. INTRODUCTION ### Context and objectives - Improving digital connectivity is crucial in SSA given demographic expansion and rapid urbanization to create jobs and exploit opportunities from the Fourth Industrial Revolution. - Paper aims to estimate the cost of introducing a full-scale 4G network by 2025 in SSA and an operable 5G network by 2040. - Methodological approach: - Follows the methodology developed by Lombardo (2019) of the European Broadcasting Union assuming a non-standalone network architecture. - Uses a synthetic country (average of France, Germany, Italy, Spain, and United Kingdom) adjusted for SSA-specific characteristics. - Accounts for demographic transitions, rapid urbanization, population density projections for highly decentralized administrative areas, heterogeneity in electricity costs, current 4G coverage, number of licensed mobile operators, and potential lack of fiber backbones. ### Key demographic and urbanization projections used in the analysis - SSA population projected to rise to 1.7 billion in 2040 from 1 billion today. - Labor force projected to increase by 20 million per year over the next two decades. - Urban population has on average risen by approximately 4.4 percent annually over the last 50 years; average urbanization rate now exceeding 40 percent. - Urban population expected to double over the next 25 years. ### 4G and 5G role and technical framing - 4G and 5G are seen as transformative for economic and social opportunities, enabling IoT and advanced applications. - The paper focuses on heterogeneous 5G networks (non-standalone architecture) requiring concurrent operation of macro-, micro-, pico-cells. - Non-standalone approach: mobile operators use existing sites/infrastructure and 4G spectrum as an anchor connection with a densified network of small cells and higher-band spectrum. ### Policy actions highlighted - Infrastructure alone is insufficient; policies are needed to address affordability and knowledge gaps. - Recommended policy actions (identified in the paper): - Public-private collaboration and updated Technological Infrastructure Policy to ensure widespread implementation beyond urban upgrades. - Endow the poor with smartphones in rural and undeveloped areas of SSA to expand benefits of 5G adoption. - Reconsider taxation of mobile phone-based transactions that currently slows the mobile revolution in SSA. - Address affordability (ability to purchase devices and pay monthly data fees) and user knowledge/skills gaps to increase demand and adoption. ### Scope limitation - The paper focuses on the cost to provide necessary infrastructure to bridge the mobile Digital Divide; issues of affordability and user capacity are acknowledged as important but are outside the paper’s costing scope. ### Findings — headline cost estimates and per-subscriber metrics - Overall capital and operational cost estimates: - Cost to implement full 4G in SSA by 2025: US$14 billion. - Conditional additional cost to make 5G operable by 2040: US$57 billion. - Annualized capital and operational cost equivalents: - Full 4G (annual): US$2.5 billion (0.04 percent of 2025 SSA GDP). - Operable 5G (annual): US$3.9 billion (0.03 percent of 2040 SSA GDP). - Per-subscriber income equivalents: - 8.4 percent of income per subscriber for 4G connectivity in 2025. - 4.9 percent of income per subscriber for 5G connectivity in 2040. - Decomposition of annual costs into capital and operational components: - To achieve full 4G connectivity by 2025: - Capital expenditure: US$2.1 billion. - Operational expenditure: US$0.4 billion. - Annual costs for 5G connectivity in 2040: - Capital expenditure: US$3.3 billion. - Operational expenditure: US$0.6 billion. - Heterogeneity across SSA: - Significant variation in costs per subscriber income across SSA countries driven by differences in current mobile infrastructure, population growth, urbanization rates, and real GDP growth projections. ### Analytical contributions and novelty - Central contribution: incorporation of SSA-specific characteristics (demographic transitions, rapid urbanization, projected population densities for decentralized administrative areas) into the costing framework. - Uses observed trends in population densities (2000–2020) and United Nations medium-variant population projections to infer 2025 and 2040 densities for each SSA country to determine infrastructure and operational expenditure requirements. *Source: IMF Working Paper WP/19/249.* ### 21. Despite the proliferation of mobile phones, there remains a global mobile Digital ### 21. Despite the proliferation of mobile phones, there remains a global mobile Digital Divide ### Mobile Digital Divide: current patterns and heterogeneity - SSA is the fastest expanding region in number of mobile subscribers with a composite average growth rate of 4.6 percent per year and an additional 167 million subscribers expected to materialize over the period to 2025 (GSMA, 2019a). - The total subscriber base in SSA is projected to increase to slightly over 600 million, representing around half the population. - Large heterogeneity in average monthly mobile data consumption (GB per capita): - Countries with average monthly traffic exceeding 10 GB: Kuwait, 31 GB; Finland, 23 GB; Saudi Arabia, 13 GB. - Countries with average monthly traffic less than 0.08 GB: Afghanistan, Ukraine, and Bolivia. - In SSA, only three countries consume more than 2.03 GB per month (global average): Seychelles, Equatorial Guinea, and Gabon. - Alper and Miktus (2019) note a narrowing of the global Digital Divide from 2005 to 2017, but absent active policies addressing infrastructure, affordability, knowledge, quality, and internet access, the Digital Divide may persist or widen. ### Economic contribution of mobile technologies and 5G potential - Digitalization is estimated to contribute US$12.3 trillion of global economic output by 2035 (HIS, 2017). - In SSA, mobile technologies and services generated US$144.1 billion, about 8.6 percent of SSA GDP in 2018 (GSMA, 2019a). - Projected contribution of 5G in Africa: USD 10.5 billion by 2026 (from virtually zero currently), with highest shares in manufacturing and energy sectors. - 5G potential impacts: - Creation of new high-tech employment opportunities. - Revolutionizing rural jobs, particularly in agriculture, via quicker transfer of weather and other data. - Improved network connectivity in rural schools and positive effects on local government and health care service provision. ### Policy implications and risks for 5G rollout - Early-stage 5G deployment investment will be commercially viable solely for densely populated urban areas, risking initial widening of the Digital Divide for rural and suburban areas. - Non-infrastructure factors (affordability, knowledge, gender dimension) may continue to prevent poorer and rural households from benefiting from 5G; leveling the playing field for female entrepreneurs is positively related to higher digital connectivity (Alper and Miktus, 2019). - Active policy intervention is required to implement 5G to bridge the mobile Digital Divide in SSA. - The text notes an investment magnitude as: US$10.52 billion (context: early-stage investment implications). ### Costing methodology: scope and objectives - Objective: evaluate costs for full introduction of 4G networks in SSA by 2025 and assess 5G implementation by 2040, assuming full 4G implementation by 2025, and estimate costs for providing universal coverage. - Methodology accounts for heterogeneity across SSA in current 4G adoption, population growth rates, pace of urbanization, and electricity prices. - Spatial inputs: - Administrative areas from GADM at the finest level of granularity. - Population maps from WorldPop for 2000, 2010, and 2020 (predicted number of people per ~100 m pixel estimated using a random forest model). - Population density projections for 2025 and 2040 obtained via quadratic Lagrange polynomials using nodes 2000, 2010, 2020; projected weights truncated to [0,1] and normalized so weights sum to 1; projected densities multiplied by UN medium-variant projected population size ratio to current region area. ### Transforming traffic and capacity into area throughput (technical steps) - Current mobile traffic demand input: ITU ICT Indicators database average monthly mobile data consumed per month per capita in GB. - Busy hour traffic calculation (per capita): - Convert GB to megabytes by multiplying by 1000. - Convert months to days by dividing by 30. - Assume hourly busy data consumption accounts for 20 percent of total daily traffic. - Busy hour bitrate: - Convert megabytes per hour to megabits per second by multiplying by 8 and dividing by 3600 (60 x 60). - Network capacity: - Network should provide at least 80 percent of busy hour bitrate all the time. - Area throughput: - Multiply projected population densities by network capacities to derive area throughput. - Box 1 formulas summarized (as in methodology): - Busy hour traffic uses the 20 percent factor and the 1000 multiplier for GB-to-MB and monthly-to-daily conversion. - Busy hour bitrate converts MB/hour to Mb/s with the 8 and 3600 factors. - Network capacity = Busy hour bitrate / 0.8. - Area throughput = Network capacity × Population density. ### Population density classification and urbanization projections - Administrative area classification by population density: - Rural: less than 500 people per square kilometer. - Urban: greater than 500 but less than 1,500 people per square kilometer. - Dense Urban: exceeds 1,500 people per square kilometer. - Projected shifts in SSA: - Share of population in rural areas projected to fall to less 60 percent by 2040 from 78 percent in 2010. - Share of population in densely urban areas projected to increase to roughly a third of total population by 2040 from below 15 percent in 2010. - Rapid urbanization and emergence of mega-cities by 2040 will increase reliance on micro- and pico-cells and raise infrastructure costs. ### Network parameters, cell geometry, and capacity planning - Assumed spectrum and cell operation: - Macro-cells operate on 700 MHz. - Micro-cells operate on 1800 MHz. - Pico-cells operate on 2600 MHz. - Spectral efficiencies: - 4G: 15 bit/s/Hz. - 5G: 23 bit/s/Hz. - Cell radius by geotype (Table 1 values): - Rural (0 – 500 pop/km2): Macro cell 10 km; Micro cell 2 km; Pico cell Not applicable. - Urban (500 – 1500 pop/km2): Macro cell 2.5 km; Micro cell 1.25 km; Pico cell 0.2 km. - Dense Urban (>1500 pop/km2): Macro cell 1 km; Micro cell 0.5 km; Pico cell 0.1 km. - Hexagonal cell approximation used for coverage: - Number of base stations per cell class i for area A uses hexagon geometry with maximum radius r_i (formula given in text). - Throughput provided by n_i base stations in area A computed using maximum capacity τ_i and n_i. - Capacity planning and densification logic: - Start with macro-cells to guarantee full coverage and compute delivered throughput. - If demand not met, add antennas (up to three per cell), then add extra macro layers (if other carriers present), then micro-cell layers or pico-cells until demand satisfied. - Augment final calculated number of cells by 30 percent (following Lombardo, 2019) to reflect practical deployment considerations. ### Scenarios and target traffic benchmarks for planning - 2020 baseline: current mobile consumption traffic assumed uniformly distributed across administrative areas; for 2020 the final number of cells equals the product of percentage of population covered by 4G network and the theoretical number of cells. - Two 2025 4G scenarios: - Scenario 1: assume 23 percent of connections in 2025 are 4G (versus 7 percent in 2018) and impose a minimum of 0.5 GB monthly mobile traffic by 2025. - Scenario 2: use world mean monthly consumption benchmarks—current global average monthly traffic estimated as 2.03 GB; Ericsson (2019) projects global mean of 5.6 GB currently and 20.1 GB by 2024; adjusted 2025 target scaled by initial conditions to equal 7.3 GB monthly traffic in the second scenario (calculation based on 2.03/5.6⋅20.1 = 7.3 GB). - 2040 5G target: - European Broadcasting Union objective of guaranteeing a minimum of 44 GB of monthly mobile traffic by 2040 is used as the planning benchmark. - This does not imply full SSA uptake but gives the infrastructure cost needed to allow 44 GB monthly per user. - Note on smartphone adoption: - Current smartphone adoption rate in SSA is 39 percent, predicted to rise to 66 percent by 2025 (GSMA, 2019). ### Model limitations and upside risks - Estimates do not include costs from signal loss caused by topology (mountains, deserts) or interference; constant throughput across service areas is assumed. - Calculated costs should be viewed as a lower bound. - More sophisticated models (SEAMCAT, nonlinear constrained discrete optimization approaches) could refine estimates but require data unavailable for all SSA countries. *Source: wpiea2019249-print-pdf - 21. Despite the proliferation of mobile phones, there remains a global mobile Digital Divide (chapter content as provided).* ### 39. Finally, we calculate network costs. We distinguish between two main cost ### Network costs for 4G/5G deployment in SSA (content unit) ### Cost components and assumptions - Two main cost components distinguished: capital expenditures (Capex) and operating expenditures (Opex) (Pereira, 2007). - Capex includes equipment costs, tower costs, and installation costs, paid at the beginning of each year for simplicity. - Opex incorporates energy consumption, heat dissipation, site rental, and site operation and maintenance expenses (O&M), paid at the end of the year. - Electricity costs per kW/h are from the World Bank database following the DB16-methodology and are assumed to remain constant in the next two decades. - Heat dissipation is assumed to double electricity costs ("This implies a doubling of energy consumption costs."). - Operational labor assumptions: 6–7 days of work annually for an expert for macro cells; 2–4 days of work for micro- and pico-cells. - Operators share parts of Capex and Opex costs (see Box 2). - Expert availability assumed in every region (may underestimate costs in low-income areas). - Final costs are augmented by 30–50 percent to account for sparse fiber backhaul and secondary costs; specifically a 50 percent mark-up for Capex and 30 percent for Opex in SSA. - Exchange rate assumption: 1.1 US$/EUR (assumed constant). ### Mobile network costs (Table 2 — input numbers preserved) - Capital Expenditures (Capex) - Macro: Equipment €32k; Tower €123k; Installation €30k - Micro: Equipment €5k; Tower €20k; Installation €10k - Pico: Equipment €2k; Tower €3k; Installation €2k - Operational Expenditures (Opex) - Macro: Power 100 W; Efficiency 50%; O&M €3.3k; Rental €5k - Micro: Power 50 W; Efficiency 50%; O&M €1k; Rental €2.5k - Pico: Power 10 W; Efficiency 50%; O&M €0.9k; Rental €0.01k - Source for table inputs: Lombardo (2019). ### Box 2: calculating network costs — equipment lifetimes and renewal timing - Passive "Tower" component lifetime: around 20-30 years — does not need renewal until 2040. - Active "Equipment" component lifetime: around 5-10 years — necessitates renewal cycles: - Renew active equipment in 2020, 2032, and 2040 (i.e., two renewals between 2020 and 2040). - Renewal of 5G additional cells at least once. - Installation costs are assumed incurred for each equipment renewal cycle; no separate installation costs for Tower vs Equipment are assumed. - Fixed equipment price assumed from 2020–40 (no price decline modeled). A potential refinement could include 5 to 10 percent annual reduction in active equipment unit cost. ### Total estimated costs for SSA (aggregate and components) - Aggregate required capital and operational expenditures: - Full-scaled 4G by 2025: about US$14 billion total. - 4G Capex: US$12.5 billion by 2025 for SSA. - 4G Opex: US$1.5 billion by 2025 for SSA. - Operable 5G by 2040: about US$57 billion total. - 5G Capex: US$49 billion by 2040 for SSA. - 5G Opex: US$8 billion by 2040 for SSA. - Capex financing assumption: required Capex spending cost is financed uniformly across years. - 4G implementation by 2025 compressed into 5 years (higher annual Capex). - 5G implementation by 2040 spread uniformly over 15 years (2026-2040) (lower annual Capex). - Aggregate annual spending needs (median/aggregate SSA): - 4G implementation climbs annually to approximately US$2.5 billion (0.04 percent of SSA GDP) by 2025. - 5G implementation overall costs amount to US$3.9 billion (0.03 percent of SSA GDP) annually in 2040. ### Heterogeneity across countries, rural share, and per-country burdens - Significant heterogeneity in required annual Capex spending across SSA due to differences in current infrastructure, population growth, and urbanization. - Examples for 4G annual spending as percentage of GDP (2025): - Mauritius and South Africa: less than 0.004 percent of GDP annually. - Burundi: total annual cost reaches 2 percent of GDP (reflecting low current infrastructure). - Examples for 5G annual spending as percentage of GDP (2040): - Mauritius, Seychelles and Ghana: less than 0.01 percent of projected GDP annually. - Burundi: 0.36 percent of projected GDP annually. - Central African Republic: 0.85 percent of projected GDP annually. - A significant part of total costs owes to the extent of rural areas and population density differences. ### Per-subscriber costs and penetration assumptions - Implementation of full 4G by 2025 costs correspond annually to 8.4 percent of average subscriber income in SSA in 2025 (median basis). - Implementation of 5G by 2040 costs correspond annually to 4.9 percent of average subscriber income in SSA in 2040 (median basis). - Method: annual total cost / (Population ⋅ Penetration rate) = Annual total cost / (Penetration rate ⋅ GDP) using projected GDP, population, and linear extrapolation of GSMA Intelligence 2017 penetration rates to 2025 and 2040. - Country-level per-subscriber values for 4G in 2025 and 5G in 2040 are included in Appendix V (values not reproduced here). ### Policy conclusions and recommendations (preserved formulation) - 5G could help bridge the mobile Digital Divide in SSA, but benefits depend on policy interventions addressing issues beyond infrastructure (affordability, knowledge, cybersecurity, data privacy). - Key policy actions to complement infrastructure spending: - Encourage competition. - Implement appropriate licensing frameworks to achieve inclusive coverage. - Address tax issues that hamper the provision of mobile services. - Regulators and policymakers should ensure affordable connectivity for everyone, including addressing the gender divide. - Possible country measures (examples preserved from text): - Sell available spectrum to telecommunication companies under the condition of furnishing full geographical coverage (as in Germany or Australia). - Establish supportive legislation with a forward-looking national strategy and time-bound targets to expand digital connectivity opportunities beyond mobile. - Use directed subsidies in rural areas and by gender (e.g., free SIM cards, costless subscriptions, incentives to increase affordability of smartphones) to complement infrastructure availability. - Trade-offs and questions for policymakers: - Who should bear the cost — public sector or consumers/firms? - Should countries with significant mobile connectivity costs consider delaying 4G connectivity implementation to 2030 to reduce annual Capex needs at the cost of delaying catch-up? - Consideration of competing spending needs (education, health, electricity access, water and sanitation) and the sizable additional spending requirement for SSA to meet the SDGs by 2030 (estimated at US$400 billion). *Source: IMF working paper content unit (wpiea2019249-print-pdf).* ### 37. Qorvo, 2017. Small Cell Networks and the Evolution of 5G ### 37. Qorvo, 2017. Small Cell Networks and the Evolution of 5G ### 4G Coverage in Sub-Saharan Africa (Appendix I) - Appendix I lists 4G coverage (in percentage) by country and year. Selected entries: - Angola AGO 8 2017 - Lesotho LSO 80 2017 - Liberia LBR 4.4 2016 - Madagascar MDG 23.08 2017 - Mauritius MUS 93 2017 - Namibia NAM 39 2017 - Nigeria NGA 50.84 2017 - Rwanda RWA 89 2017 - Seychelles SYC 90 2017 - South Africa ZAF 77.62 2017 - Tanzania TZA 13 2017 - Uganda UGA 17 2017 - Zambia ZMB 43.41 2017 - Zimbabwe ZWE 34.72 2017 - Sources: ITU ICT indicators database. ### List of 4G Networks in SSA (Appendix II) - Appendix II provides operators offering 4G by country. Selected operator-country pairs: - Angola AGO: Movicel; Net One; Unitel - Benin BEN: be.Telecoms; MTN - Botswana BWA: BTC; Mascom; Orange - Cameroon CMR: MTN; Vodafone - Ghana GHA: Blu Telecom; BusyInternet; MTN; Surfline; Telesol; Vodafone - Kenya KEN: Airtel; Faiba 4G; Safaricom; Telkom Kenya - Mauritius MUS: Emtel; MTML; my.t - Nigeria NGA: Airtel; Bitflux; Cyberspace; Etisalat; Glo Mobile; InterC Network; MTN; ntel; QCell; Smile; Spectranet; SWIFT - South Africa ZAF: Cell C; MTN; Neotel; Rain Mobile; Telkom; Vodacom - Tanzania TZA: Smile; Smart; Tigo; TTCL; Vodacom; Zantel - Uganda UGA: Airtel; MTN; Orange; Smile - Sources: Wikipedia, October 8, 2019. ### Country Case Study: Rwanda (Appendix III) — Methodology and Projections - Network costing methodology steps (illustrated using Rwanda): - Incorporates population growth and rapid urbanization within each administrative area in every SSA country. - Administrative divisions plotted at the finest possible granularity (GID Level: 3). - Population density obtained for each administrative region by translating pixel colors from WorldPop images, summing, and adjusting totals to match medium-variant United Nations statistics. - Analysis performed using WorldPop pictures for 2000, 2010, and 2020; observations include increasing population trend and migration towards urban areas, especially Kigali. - Population densities projected for 2025 and 2040 through Lagrange polynomials fitted to 2000, 2010, and 2020 weights. - Observed accelerated urbanization around Kigali City, influenced by Kigali Innovation City (a technology cluster and special economic zone with preferential income tax rates). - Regional classification thresholds: - Rural: population density < 500 people per square kilometer. - Urban: population density > 500 and < 1,500 people per square kilometer. - Dense Urban: population density > 1,500 people per square kilometer. - Figures provided: administrative area maps and population density maps for years 2000, 2010, 2020, and projections for 2025 and 2040. - Sources: GADM database; WorldPop database; authors’ calculations. ### Total Capital Expenditures — Millions of USD (Appendix IV-A) - Appendix IV-A reports Total Capital Expenditures (Millions of USD) for scenarios and years. Selected country entries (columns: 2025 Ericsson; 2025 GSMA; 2040 Ericsson; 2040 GSMA; 2040 Ericsson (decreased radius); 2040 GSMA (decreased radius)): - AGO 551.89 545.34 1000.27 997.00 2541.72 2538.45 - BDI 947.30 946.13 759.68 759.10 802.79 802.21 - BEN 60.89 59.99 256.09 255.64 306.77 306.33 - BFA 237.65 237.50 446.88 446.80 655.80 655.72 - BWA 57.06 57.06 315.51 315.51 315.51 986.34 - COD 828.89 827.79 2500.56 2500.00 4282.21 4281.66 - ETH 812.98 801.68 2438.88 2433.23 3148.80 3143.15 - KEN 673.40 665.74 1149.99 1146.31 1798.51 1794.83 - NGA 1083.07 1072.62 3850.62 3845.40 4986.14 4980.91 - TZA 1677.96 1674.69 2266.60 2264.96 3256.28 3254.64 - ZAF 148.24 148.24 1470.50 1470.50 2599.18 2599.18 - Sources: GADM and WorldPop database and authors’ calculations. ### Annual Capital Expenditures — Percent of GDP (Appendix IV-B) - Appendix IV-B reports Annual Capital Expenditures as percent of GDP for scenarios and years. Selected country entries (columns: 2025 Ericsson; 2025 GSMA; 2040 Ericsson; 2040 GSMA; 2040 Ericsson (decreased radius); 2040 GSMA (decreased radius)): - AGO 0.03% 0.03% 0.01% 0.01% 0.03% 0.03% - BDI 1.56% 1.56% 0.19% 0.19% 0.20% 0.20% - BEN 0.02% 0.02% 0.01% 0.01% 0.01% 0.01% - CAF 0.55% 0.55% 0.25% 0.25% 0.78% 0.78% - KEN 0.04% 0.04% 0.01% 0.01% 0.02% 0.02% - LBR 0.30% 0.30% 0.07% 0.07% 0.12% 0.12% - MWI 0.41% 0.41% 0.07% 0.07% 0.08% 0.08% - NGA 0.01% 0.01% 0.01% 0.01% 0.01% 0.01% - TZA 0.09% 0.09% 0.03% 0.03% 0.04% 0.04% - Sources: GADM and WorldPop database and authors’ calculations. ### Operational Expenditures — Millions of USD (Appendix IV-C) - Appendix IV-C reports Operational Expenditures (Millions of USD) for 2020 baseline and scenarios for 2025 and 2040. Selected country entries (columns include 2020; 2025 Ericsson; 2025 GSMA; 2040; 2040 (decreased radius)): - AGO 0.78 9.83 9.80 9.25 14.25 - BDI 4.78 43.46 43.45 44.12 44.26 - CIV 1.50 3.21 3.21 6.81 5.92 - ETH 1.21 12.84 12.80 16.14 16.79 - KEN 7.11 29.74 29.58 30.27 39.86 - NGA 11.26 28.38 28.26 24.67 44.71 - TZA 7.99 63.83 63.79 64.19 73.71 - UGA 3.03 18.06 18.01 23.59 22.81 - ZAF 8.61 11.09 11.09 11.37 23.00 - Sources: GADM and WorldPop database and authors’ calculations. ### Operational Expenditures — Percent of GDP (Appendix IV-D) - Appendix IV-D reports Operational Expenditures as percent of GDP. Selected country entries (columns include 2020; 2025 Ericsson; 2025 GSMA; 2040; 2040 (decreased radius)): - AGO 0.00% 0.00% 0.00% 0.00% 0.00% - BDI 0.06% 0.43% 0.43% 0.16% 0.16% - ETH 0.00% 0.00% 0.00% 0.00% 0.00% - KEN 0.00% 0.01% 0.01% 0.00% 0.01% - LBR 0.00% 0.08% 0.08% 0.05% 0.06% - MWI 0.06% 0.13% 0.13% 0.05% 0.05% - NGA 0.00% 0.00% 0.00% 0.00% 0.00% - TZA 0.00% 0.02% 0.02% 0.01% 0.01% - ZWE 0.00% 0.00% 0.00% 0.00% 0.01% - Sources: GADM and WorldPop database and authors’ calculations. ### Total Cost per Subscriber — Percent of income per capita (Appendix V) - Appendix V reports Total Cost per Subscriber as percent of income per capita for 4G and 5G. Selected entries: - AGO 4G 8.66 5G 4.81 - BDI 4G 451.08 5G 43.58 - BEN 4G 4.06 5G 2.89 - CAF 4G 226.62 5G 221.13 - CIV 4G 1.84 5G 1.77 - COD 4G 40.99 5G 33.96 - ETH 4G 7.95 5G 2.81 - KEN 4G 5.85 5G 2.35 - LBR 4G 101.40 5G 38.09 - MWI 4G 150.31 5G 19.03 - NGA 4G 2.80 5G 2.02 - TZA 4G 19.75 5G 5.25 - ZAF 4G 0.42 5G 1.16 - ZWE 4G 4.54 5G 7.60 - Sources: GADM and WorldPop database and authors’ calculations. *Source: wpiea2019249-print-pdf - 37. Qorvo, 2017. Small Cell Networks and the Evolution of 5G (selected appendices and figures).* --- _Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019249-print-pdf.pdf_