## wpiea2023027-print-pdf

## Source details

**Canonical URL:** [wpiea2023027-print-pdf](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023027-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023027-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023027-print-pdf.pdf.json)

---

### Major findings and headline estimates
- Additional investment needs to provide universal broadband connectivity:
  - $418 billion globally.
  - Approximately 0.45 percent of global GDP.
- Investment as a share of GDP by region:
  - Sub-Saharan Africa: 4.49 percent of GDP.
  - Advanced economies: 0.02 percent of GDP.
- Global connectivity context and digital divide:
  - Only approximately 60 percent of the global population is online.
  - Unconnected population estimates cited: 2 billion in emerging market economies; 1 billion in low-income developing countries.
  - Regional magnitudes noted: parts of Emerging and Developing Asia (1.7 billion) and Sub-Saharan Africa (0.75 billion).

### Role and benefits of digital infrastructure
- Digital infrastructure links to United Nations Sustainable Development Goals via Target 9.1.
- Availability of Internet connectivity can:
  - Provide new economic opportunities for unconnected communities.
  - Foster structural labor market shifts toward more productive, digitally-enabled activities.
  - Improve educational outcomes, healthcare access, economic productivity, and job creation.
- COVID-19 observed effects:
  - The digital divide impeded effective social distancing where citizens could not work remotely.
  - More digitalized industries experienced lower revenue losses.
  - Areas with superior broadband showed better labor market resilience, improved educational outcomes, and easier access to government support.
- Wider public-benefit outcomes from broader broadband access:
  - Expand the tax base and strengthen revenue collection.
  - Transform public financial management through modernized systems, improved public service delivery, digital payments, and increased transparency.
  - Contribute to lower carbon emissions via 'smart' management across energy, utilities, manufacturing, agriculture, buildings, services, transportation, and traffic management.

### Limitations and complementary requirements for adoption
- Investing in digital infrastructure alone does not guarantee digital adoption.
- Barriers to adoption include:
  - Lack of necessary skills to access relevant content (e.g., language, digital literacy).
  - Mistrust or lack of knowledge about technologies in poorer, less-educated communities.
  - Lack of proof of identity to access online secured services.
  - Legal and regulatory weaknesses (user protection, data privacy, cybersecurity).

### Research questions, method, and DICE model overview
- Key research questions:
  - How much investment is required in digital infrastructure to achieve affordable universal connectivity?
  - How do data consumption and quality of service affect the necessary investment?
- DICE (Digital Infrastructure Costing Estimator) purpose and scope:
  - Estimates comparative country-specific investment to achieve universal broadband connectivity, assuming predominantly terrestrial 4G deployment with satellite fallback for very hard-to-serve locations.
  - Estimates investments for first mile, middle mile, and last mile infrastructure.
  - Expected smartphone adoption rate for the end of the assessment period (2030) set at 90 percent.
- Core modeling features and equations (structure preserved):
  - Demand: U_i,t = (P_i * (1 + G/100)^y) * (A_i/100) * (MMS_i/100) * (AU_i/100)
  - Per-user rate conversion: RRR_i = MRR_i * 1000 * 8 * (1 / n_d) * (f_dh / 100) * (1 / 3600)
  - Spatial rate: R_i = (U_i,t * RRR_i) / Area_i
  - Area capacity C_i,t (Mbps/km^2) and spectral efficiency via Monte Carlo simulation.
  - Site additions and upgrades modelled with NAN_Mi,t,A,E_i = TTT_tATMMi,t,A,E_i − EEE_iEE t in EMMit tA E_i (Equation (6) structure preserved).
  - New site capex components include RAN capex, backhaul capex, civil engineering, power system cost, and labor (Equation (7) structure preserved).
  - Mean fiber backhaul cost computed using mean distance, number of new sites, backhaul-core splitting factor α, and cost per fiber meter (Equation (8) structure preserved).
  - Opex approximated as 15 percent of the initial asset value annually plus labor (Equation (9) structure preserved).
  - Policy/regulation allocation roughly $2/user; skills and content allocation estimated at $12/user.
  - Satellite fallback: $200 monthly subscription split between 12, 8, or 4 users for low-income developing countries, emerging market economies, or advanced economies, respectively.
  - Total Cost of Ownership Per User (TTPU_c) and Total Cost per country (TTC_c) computed by summing components and weighting by population and adoption (Equations (10) and (11) structures preserved).
- Model data sources:
  - WorldPop 1 km^2 population mosaic for 2020; GADM levels 1 and 2.
  - Population growth forecasts from the United Nations (IMF 2021) for 2030 population.
  - Tower counts from TowerXchange; cell coordinates from OpenCelliD.
  - Cellular population coverage from ITU (2022).
  - Backhaul technology shares from GSMA 2019: 81 percent North East Asia, 73 percent North America, 33 percent Europe, 21 percent Latin America and the Caribbean, 20 percent Middle East and North Africa, 17 percent South and South East Asia, 15 percent Sub-Saharan Africa.
  - Labor costs from IMF 2022, with 16 hours allocated per labor component.
  - Internet adoption rates from World Bank 2022.

### Cost drivers for digital infrastructure investment
- Population density is the primary determinant of per-user investment economics.
- Decile methodology:
  - Local statistical areas grouped into population density deciles; Decile 1 = top 10 percent densest areas; Decile 10 = bottom 10 percent least dense areas.
- Observations across income groups (selected exact figures):
  - Advanced economies: first decile contains 420 million people (39 percent of total population); final decile 22 million (2 percent).
  - Low-income developing countries: first decile 317 million (21 percent); final decile 70 million (5 percent).
  - Area share: first decile 2 percent of overall area.
  - Bottom decile area shares: 51 percent (advanced), 45 percent (emerging market), 31 percent (low-income developing).
  - Population density examples: Decile 1 density — emerging market economies 5,901 persons/km^2; advanced 659 persons/km^2; low-income developing 4,640 persons/km^2.
  - Decile 10 density: advanced economies falls to 1 person/km^2.

### Digital infrastructure investment needs (detailed estimates)
- Aggregate global digital infrastructure investment need: $418 billion.
  - Based on assumption of providing universal 4G cellular broadband to users with approximately 40-50 GB of monthly data at 95 percent reliability.
- By income group:
  - Emerging market economies: $305 billion (73 percent).
  - Advanced economies: $11 billion (3 percent).
  - Low-income developing countries: $102 billion (24 percent).
  - Low-income developing countries composition: digital infrastructure capex 27 percent, metro and backbone fiber 24 percent, infrastructure opex 33 percent.
- By region:
  - Emerging and Developing Asia: $176 billion (42 percent).
  - Sub-Saharan Africa: $91 billion (22 percent).
  - Middle East and North Africa, Afghanistan, and Pakistan (MENAP): $69 billion (17 percent).
  - Emerging and Developing Europe: $14 billion (3 percent).
  - Caucasus and Central Asia: $7 billion (2 percent).
- Decile and regional distribution highlights:
  - In low-income developing countries: decile investments example — $3.6 billion (first decile), $13.8 billion (fifth decile), $6.7 billion (bottom decile).
  - Advanced economies: deciles five to ten each below $2.5 billion.
  - Per-decile regional maxima: Emerging and Developing Asia up to $28.3 billion per decile; Sub-Saharan Africa up to $11.9 billion per decile; Latin America and the Caribbean up to $9.3 billion per decile; MENAP up to $9.2 billion per decile; Emerging and Developing Europe up to $3.4 billion; Caucasus and Central Asia up to $1.7 billion.
- Investment relative to GDP:
  - Low-income developing countries: about 3.5 percent of GDP.
  - Emerging market economies: 0.7 percent of GDP.
  - By region: Sub-Saharan Africa 4.5 percent of GDP; MENAP 1.7 percent; Caucasus and Central Asia 1.6 percent; Latin America and the Caribbean 1.0 percent.
- Spatial granularity:
  - Subnational investments range from below $10 million up to more than $100 million per statistical area.
  - Highest needs: high-latitude regions, the Sahara, the Amazon, central/western Australia, and the Tibetan Plateau.
  - Areas requiring no investment include the United States’ coastal areas, Latin America, Western and Southern Europe, Eastern China, and large parts of India.
- Comparison with other estimates:
  - ITU Connecting Humanity estimated $428 billion — 1.4 percent higher than the DICE estimate of $418 billion.
  - Noted differences concentrated in mobile infrastructure operational expenditure and East Asia and the Pacific estimates.
  - An alternative IMF staff estimate of $14 billion for full 4G in Sub-Saharan Africa is much lower due to differing methodological assumptions (conservative adoption and data consumption, 2025 target, and omission of topological radio effects).

### Sensitivity analysis (data consumption and QoS)
- Data consumption sensitivity:
  - Baseline data consumption: 40-50 GB/Month per user.
  - Lower-data scenario: Emerging market and advanced economies at 20 GB/Month per user; low-income developing countries at 10 GB/Month per user.
    - Aggregate cost decreases from $418 billion to $201 billion (52 percent decrease).
    - Emerging market economies decrease from $305 billion to $136 billion (55 percent reduction).
  - Higher-data scenario: Emerging market and advanced economies at 100 GB/Month per user; low-income developing countries at 80 GB/Month per user.
    - Aggregate cost increases from $418 billion to $783 billion (more than 90 percent rise).
    - Emerging market economies increase from $305 billion to $597 billion (49 percent growth).
    - Low-income developing countries increase from $102 billion to $172 billion.
  - Implication: broadband policy assessment must specify per-user data allowances.
- Quality of Service (QoS) sensitivity:
  - Baseline reliability: 95 percent.
  - Lowering service reliability to 5 percent reduces overall cost from $418 billion to $176 billion.
    - Emerging and Developing Asia: cost decreases by 77 percent relative to baseline.
    - Sub-Saharan Africa: cost decreases by 51 percent relative to baseline (greenfield infrastructure remains significant even for low QoS).

### Policy implications and recommendations
- Place digital infrastructure at the center of economic policy agendas to:
  - Unlock economic opportunities, create jobs, and generate growth.
  - Improve quality of life through better access to healthcare, education, social safety nets, and financial services.
  - Support reskilling for workers affected by technology-driven economic transformation.
  - Help reduce depopulation in rural and remote areas.
- Complement infrastructure investment with policies to increase adoption:
  - Invest in digital skills and relevant content in users' languages.
  - Strengthen legal and regulatory frameworks for user protection, data privacy, and cybersecurity.
- Policy and regulatory measures and business models can reduce costs and increase ARPU; one cited assessment suggests implementing all cost-reducing measures could reduce necessary investment for all low- and middle-income countries by half.

### Limitations, implementation risks, and avenues for future research
- Key limitations and risks:
  - Country capacity to deliver strong public investment management and capital spending implementation, particularly in Sub-Saharan Africa.
  - Global analysis uses harmonized datasets and sacrifices country-specific detail; not a substitute for detailed country modeling.
- Future research directions:
  - Develop more detailed demand-side models reflecting affordability and digital literacy constraints, country-by-country.
  - Explore global cost implications of deploying Fiber-To-The-Premises (FTTP) as a longer-term goal.
  - Assess the economic benefits of broadband infrastructure investment beyond construction costs.

*Source: wpiea2023027-print-pdf - References*

### References .............................................................................................................

### wpiea2023027-print-pdf - References

### Major findings and headline estimates
- The method generally assumes the use of 4G cellular technology to provide access to Internet.
- Additional investment needs to provide universal broadband connectivity:
  - $418 billion globally.
  - Approximately 0.45 percent of global GDP.
- Investment as a share of GDP by region:
  - Sub-Saharan Africa: 4.49 percent of GDP.
  - Advanced economies: 0.02 percent of GDP.
- Global connectivity context and digital divide:
  - Only approximately 60 percent of the global population is online.
  - Unconnected population estimates cited: 2 billion in emerging market economies; 1 billion in low-income developing countries.
  - Regional magnitudes noted: parts of Emerging and Developing Asia (1.7 billion) and Sub-Saharan Africa (0.75 billion).

### Role and benefits of digital infrastructure
- Digital infrastructure is essential for Internet adoption and is linked to United Nations Sustainable Development Goals via Target 9.1.
- Availability of Internet connectivity can:
  - Provide new economic opportunities for unconnected communities.
  - Foster structural labor market shifts toward more productive, digitally-enabled activities.
  - Improve educational outcomes, healthcare access, economic productivity, and job creation.
- During the COVID-19 pandemic:
  - The digital divide impeded effective social distancing policies where citizens could not work remotely.
  - More digitalized industries experienced lower revenue losses.
  - Areas with superior broadband showed better labor market resilience, improved educational outcomes, and easier access to government support.
- Wider broadband access can yield additional public-benefit outcomes:
  - Expand the tax base and strengthen revenue collection.
  - Transform public financial management through modernized systems, improved public service delivery, digital payments, and increased transparency.
  - Contribute to lower carbon emissions by enabling 'smart' management across energy, utilities, manufacturing, agriculture, buildings, services, transportation, and traffic management.

### Limitations and complementary requirements for adoption
- Investing in digital infrastructure alone does not guarantee digital adoption.
- Barriers to adoption include:
  - Lack of necessary skills to access relevant content (e.g., language, digital literacy).
  - Mistrust or lack of knowledge about technologies in poorer, less-educated communities.
  - Lack of proof of identity to access online secured services.
  - Legal and regulatory weaknesses (user protection, data privacy, cybersecurity) that need addressing to realize full benefits.

### Research questions, method, and paper structure
- The paper addresses two key questions:
  - How much investment is required in digital infrastructure to achieve affordable universal connectivity?
  - How do data consumption and quality of service affect the necessary investment?
- Methodological approach:
  - Develop a method to estimate universal broadband costs for each country and aggregate results for cross-country comparisons by income group and region.
- Organization of the paper (as described):
  - Section II: overview of evolution of digital connectivity.
  - Section III: description of the method for costing digital infrastructure investments.
  - Section IV: discussion of cost drivers of digital infrastructure investment.
  - Section V: estimates of additional investment needs.
  - Section VI: sensitivity analysis around the cost estimates.

### Policy implications and recommendations
- Place digital infrastructure at the center of economic policy agendas to:
  - Unlock economic opportunities, create jobs, and generate growth.
  - Improve quality of life through better access to healthcare, education, social safety nets, and financial services.
  - Support reskilling for workers affected by technology-driven economic transformation.
  - Help reduce depopulation in rural and remote areas.
- Complement infrastructure investment with policies to increase adoption:
  - Invest in digital skills and relevant content in users' languages.
  - Strengthen legal and regulatory frameworks for user protection, data privacy, and cybersecurity.

*Source: wpiea2023027-print-pdf - References*

### conclusions being provided in Section VII.

### wpiea2023027-print-pdf - conclusions being provided in Section VII.

### Evolution of Digital Connectivity
- Broadband technologies:
  - Fixed technologies use physical fiber optic or legacy copper/coaxial cable; deployment costs can be high and are most common in advanced high-income countries.
  - Wireless technologies include mobile cellular (e.g., 4G), Wi-Fi, and satellite architectures; they provide lower-cost broadband by eliminating the expense of laying a physical cable to each user but still require a fixed high-capacity connection at some point.
  - This assessment mainly focuses on using 4G as a low-cost technology for delivering wide-area mobile broadband.
- Generational cellular overview:
  - 1G: 1980s.
  - 2G (GSM): 1990s, enabled mobile voice and text.
  - 3G (UTMS): early 2000s, basic data rates (e.g., below a peak of 10 Mbps, often experienced below 0.5 Mbps per user).
  - 4G (LTE): deployed from approximately 2010 onwards; enabled mass-market consumer broadband, video streaming up to 100 Mbps, regular experienced speed of 2-10 Mbps.
  - 5G: introduces enhanced mobile broadband and ultra-reliable low-latency communication; enables new use cases (VR/AR across industrial sectors).
  - 6G: R&D underway targeting deployment from 2028 onwards.
- Practicality for universal broadband:
  - 4G is emphasized as the more practical solution for many developing countries due to cost and device affordability constraints for 5G.
  - 5G incremental value compared to 4G is not necessarily clear; 5G viability is poor in hardest-to-reach locations.
- Coverage and adoption:
  - As of 2021, approximately 87 percent of the global population was ‘covered’ by a basic 4G signal from at least one mobile operator.
  - Despite coverage, approximately one third of citizens remain offline due to high cost of Internet access relative to income.
- Data consumption:
  - 2022 mean global smartphone user: 15 GB per month.
  - 2022 Sub-Saharan Africa: 5 GB per month.
  - 2022 India or parts of Southwest Asia (e.g., the UAE): 25 GB per month.
  - Forecast by 2028: mean global smartphone user to reach 46 GB per month; low of 18 GB per month in Sub-Saharan Africa; high of 55 GB per month in North America and North East Asia.
- COVID-19 impacts:
  - Pandemic amplified demand for Internet-based activities across work, shopping, education, public services.
  - Household surveys (Brazil) show increases across multiple online activities between pre-pandemic (2018/2019) and COVID-19 panel.

### Method for Costing Digital Infrastructure Investments (DICE)
- Purpose and scope:
  - DICE (Digital Infrastructure Costing Estimator) estimates comparative country-specific investment to achieve universal broadband connectivity, assuming predominantly terrestrial 4G deployment with satellite connectivity for very hard-to-serve locations.
  - Model estimates investments for first mile, middle mile, and last mile infrastructure.
  - Expected smartphone adoption rate for the end of the assessment period (2030) is set at 90 percent.
- Modeling approach:
  - Uses a hypothetical mobile network operator with market share parameter (MMS_i) to model per-network costs and derive cost per additional user.
  - Active users at busiest hour assumed to be between 5-20 percent of customer base; model sets percentage of active users (AU_i) accordingly.
- Demand estimation (key equations and variables preserved as described):
  - Equation (1): U_i,t = (P_i * (1 + G/100)^y) * (A_i/100) * (MMS_i/100) * (AU_i/100)
  - Equation (2): RRR_i = MRR_i * 1000 * 8 * (1 / n_d) * (f_dh / 100) * (1 / 3600)
  - Equation (3): R_i = (U_i,t * RRR_i) / Area_i
- Network dimensioning:
  - Area capacity C_i,t (Mbps/km^2) computed by summing spectral efficiency * channel bandwidth across frequencies (Equation (4)).
  - Mean network spectral efficiency estimated via Monte Carlo simulation and decomposed into average cells per site and density of cells transmitting in same frequency (Equation (5)).
  - New/upgrade sites required: NAN_Mi,t,A,E_i = TTT_tATMMi,t,A,E_i − EEE_iEE t in EMMit tA E_i (Equation (6)).
  - Site capacity lookup tables produced using simulations with free space path loss and log-normal shadow fading (μ=2, σ=10).
- Cost components and formulas:
  - New site capex components include RAN capex, backhaul capex, civil engineering cost, power system cost, and labor components for planning, transport, construction, and installation (Equation (7) structure preserved).
  - Mean fiber backhaul cost per region uses mean distance between new and existing sites, number of new sites, backhaul-core splitting factor α (e.g., 10 percent), and cost per fiber meter (Equation (8)).
  - Upgrading existing sites excludes civil engineering tower cost but includes RAN equipment and backhaul upgrades.
  - Operational expenditure (Opex) approximated as 15 percent of the initial asset value annually plus labor (Equation (9) structure preserved).
  - Policy and regulation cost allocated roughly $2/user for telecom regulator governance (PTTic y _R A E aT A t i T n_i example).
  - Skills and content allocation estimated at $12/user (ICT_MBITTE_CTnntAnt_i example).
  - Satellite fallback: $200 monthly subscription split between 12, 8, or 4 users for low-income developing countries, emerging market economies or advanced economies, respectively.
  - Total Cost of Ownership Per User (TTPU_c) computed by summing capex, backhaul, Opex, policy/regulation, and skills/content across deciles and weighting by population and adoption (Equation (10) structure preserved).
  - Total Cost per country (TTC_c) = TTPU_c * UnconnectedUsers_c (Equation (11) structure preserved).
- Model data sources:
  - WorldPop 1 km^2 population mosaic for 2020; GADM levels 1 and 2 for administrative boundaries.
  - Population growth forecasts from the United Nations (IMF 2021) to estimate 2030 population.
  - Tower counts from TowerXchange; cell coordinates from OpenCelliD.
  - Cellular population coverage from ITU (2022).
  - Backhaul technology shares from GSMA 2019 (regional percentages cited: 81 percent in North East Asia, 73 percent North America, 33 percent Europe, 21 percent Latin America and the Caribbean, 20 percent Middle East and North Africa, 17 percent South and South East Asia, 15 percent Sub-Saharan Africa).
  - Labor costs from IMF 2022, with 16 hours allocated per labor component for planning, logistics, construction, installation.
  - Internet adoption rates from World Bank 2022.

### Cost Drivers for Digital Infrastructure Investment
- Population density is the primary determinant of per-user investment economics: low density areas have poor economics due to fixed costs spread across fewer users.
- Decile methodology:
  - Local statistical areas grouped into population density deciles; Decile 1 = top 10 percent densest areas; Decile 10 = bottom 10 percent least dense areas.
- Observations across income groups:
  - Population distribution:
    - Advanced economies: first decile contains 420 million people (39 percent of total population); final decile as low as 22 million (2 percent).
    - Low-income developing countries: first decile 317 million (21 percent); final decile 70 million (5 percent).
  - Area share:
    - First decile: 2 percent of overall area.
    - Bottom decile area shares: 51 percent (advanced), 45 percent (emerging market), 31 percent (low-income developing).
  - Population density examples:
    - Decile 1 density: emerging market economies 5,901 persons/km^2; advanced 659 persons/km^2; low-income developing 4,640 persons/km^2.
    - Decile 10 density: advanced economies falls to 1 person/km^2.

### Digital Infrastructure Investment Needs (Key Findings)
- Aggregate global digital infrastructure investment need: $418 billion.
  - Based on assumption of providing universal 4G cellular broadband to users with approximately 40-50 GB of monthly data at 95 percent reliability.
- By income group:
  - Emerging market economies: $305 billion (73 percent).
  - Advanced economies: $11 billion (3 percent).
  - Low-income developing countries: $102 billion (24 percent).
  - Low-income developing countries composition: digital infrastructure capex 27 percent, metro and backbone fiber 24 percent, infrastructure opex 33 percent.
- By region:
  - Emerging and Developing Asia: $176 billion (42 percent).
  - Sub-Saharan Africa: $91 billion (22 percent).
  - Middle East and North Africa, Afghanistan, and Pakistan (MENAP): $69 billion (17 percent).
  - Emerging and Developing Europe: $14 billion (3 percent).
  - Caucasus and Central Asia: $7 billion (2 percent).
- Decile and regional distribution:
  - Emerging and developing economies: decile investment distribution follows a bell-shaped curve.
  - In low-income developing countries: example decile investments range from $3.6 billion (first decile) to $13.8 billion (fifth decile) and $6.7 billion (bottom decile).
  - Advanced economies: modest investment needs in deciles five to ten, each below $2.5 billion.
  - Per-decile regional maxima: Emerging and Developing Asia up to $28.3 billion per decile; Sub-Saharan Africa up to $11.9 billion per decile; LAC up to $9.3 billion per decile; MENAP up to $9.2 billion per decile; Emerging and Developing Europe up to $3.4 billion; Caucasus and Central Asia up to $1.7 billion.
- Investment relative to GDP:
  - Low-income developing countries: about 3.5 percent of GDP.
  - Emerging market economies: 0.7 percent of GDP.
  - By region: Sub-Saharan Africa 4.5 percent of GDP; MENAP 1.7 percent; Caucasus and Central Asia 1.6 percent; Latin America and the Caribbean 1.0 percent.
- Spatial granularity:
  - Subnational visualizations indicate investments below $10 million up to more than $100 million per statistical area, with highest needs in high-latitude regions, the Sahara, the Amazon, central/western Australia, and the Tibetan Plateau.
  - Areas requiring no investment include the United States’ coastal areas, Latin America, Western and Southern Europe, Eastern China, and large parts of India.
- Comparison with other estimates:
  - ITU Connecting Humanity estimated $428 billion — 1.4 percent higher than the DICE estimate of $418 billion.
  - Largest differences are in mobile infrastructure operational expenditure and East Asia and the Pacific estimates.
  - An alternative IMF staff estimate of $14 billion for full 4G in Sub-Saharan Africa is much lower due to differing methodological assumptions (e.g., conservative adoption and data consumption, targeting 2025 rather than universal broadband horizon, not accounting for topological radio effects).

### Sensitivity Analysis
- Data consumption sensitivity:
  - Lower-data scenario: Emerging market and advanced economies at 20 GB/Month per user; low-income developing countries at 10 GB/Month per user.
    - Aggregate cost decrease from $418 billion to $201 billion (52 percent decrease).
    - Emerging market economies decrease from $305 billion to $136 billion (55 percent reduction).
  - Higher-data scenario: Emerging market and advanced economies at 100 GB/Month per user; low-income developing countries at 80 GB/Month per user.
    - Aggregate cost increases from $418 billion to $783 billion (more than 90 percent rise).
    - Emerging market economies increase from $305 billion to $597 billion (49 percent growth).
    - Low-income developing countries increase from $102 billion to $172 billion.
- Quality of Service (QoS) sensitivity:
  - Baseline reliability: 95 percent.
  - Lowering service reliability to 5 percent reduces overall cost from $418 billion to $176 billion.
    - Emerging and Developing Asia: cost decreases by 77 percent relative to baseline.
    - Sub-Saharan Africa: cost decreases by 51 percent relative to baseline (significant greenfield infrastructure still required even for reduced QoS).

*IMF WORKING PAPERS Estimating Digital Infrastructure Investment Needs to Achieve Universal Broadband*

### 2030. To this end, we develop a new, open-source model—the Digital Infrastructure Costing Estim ator (DICE).

### 2030. To this end, we develop a new, open-source model—the Digital Infrastructure Costing Estimator (DICE).

### Methodology
- The DICE method accounts for each country’s demographic forecast trend, underlying population density, and future economic characteristics.
- The approach provides a systematic, globally harmonized assessment, trading off country-specific nuance for comprehensive coverage.

### Investment estimates for universal broadband (baseline assumptions)
- Total investment needed: $418 billion, or approximately 0.45 percent of global GDP.
- Emerging market economies: $305 billion (about 0.73 percent of their respective GDP) to connect 1.5 billion unconnected citizens.
- Advanced economies: $11 billion (about 0.02 percent of GDP) to connect 32 million currently unconnected citizens.

### Sensitivity analysis (data consumption scenarios)
- Baseline data consumption: 40-50 GB/Month per user.
- Lower consumption scenario: 10-20 GB/Month per user
  - Total cost decreases to $201 bn (from $418 bn).
- Higher consumption scenario: 80-100 GB/Month per user
  - Total cost increases up to 86 percent, to $783 bn (from $418 bn).
- Implication: future broadband policy assessment must be explicit about the quantity of data each user can consume.

### Regional and drivers of investment needs
- Key determinants of additional investment:
  - Level of past historical investment in 4G and previous cellular generations.
  - Number of new potential users to be covered.
- Sub-Saharan Africa:
  - Many unconnected users: 0.75 billion.
  - Geographic area: 19.8 million square kilometers.
  - Low Average Revenue Per User (ARPU) compounded by lower investments in newer cellular generations.
- Emerging and Developing Asia:
  - Largest estimated investment: approximately $176 billion.
  - Unconnected population: 1.7 billion.
  - Geographic area: 19.5 million square kilometers.

### Policy options and potential cost reductions
- Alternatives to business-as-usual deployment can lower costs and increase adoption:
  - Technology, business model, regulatory, and policy changes can (i) reduce infrastructure delivery costs and (ii) increase demand-side adoption and ARPU.
- One assessment cited: if all cost-reducing measures are implemented, necessary investment for providing universal broadband to all low- and middle-income countries could be reduced by half, avoiding the need for governments to provide public subsidies (Oughton et. al. 2022).

### Limitations and implementation risks
- Two key limitations:
  - Country capacity to deliver strong public investment management and capital spending implementation is a major mediating issue (particularly in Sub-Saharan Africa).
  - Global analysis requires harmonized datasets, sacrificing country-specific detail; systematic global evaluation is not a substitute for detailed country modeling needed to identify country-specific investment strategies.

### Avenues for future research
- Develop more detailed demand-side models reflecting affordability and digital literacy constraints, on a country-by-country basis.
- Explore the cost implications of deploying Fiber-To-The-Premises (FTTP) infrastructure globally as a longer-term goal beyond predominantly wireless solutions.
- Assess the economic benefits of broadband infrastructure investment, beyond the cost of building it.

*IMF Working Paper — Estimating Digital Infrastructure Investment Needs to Achieve Universal Broadband (Working Paper No. WP/2023/027)*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023027-print-pdf.pdf_