## 1. Large uncertainties characterize the major trends determining the future of work in sub-Saharan Africa (SSA) and connectivity is a key policy area.

## Source details

**Canonical URL:** [1. Large uncertainties characterize the major trends determining the future of work in sub-Saharan Africa (SSA) and connectivity is a key policy area.](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019210-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019210-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019210-print-pdf.pdf.json)

---

### Demographic and policy context
- Population and labor supply projections and requirements:
  - Population projected to reach about 1.7 billion by 2040 from 1.0 billion currently.
  - United Nations projects a net increase in the working-age population (15–64 years) in SSA of about 20 million people per year.
  - The need to generate 20 million jobs per year during the next two decades is identified as the key challenge for SSA policy makers.
- Policy framing:
  - The October 2018 Regional Economic Outlook for sub-Saharan Africa (IMF, 2018) identifies connectivity as a key policy area to promote job creation and yield dramatic improvements in living conditions.
  - Connectivity extends beyond traditional physical infrastructure (roads, railways, ports) to digital connectivity, which is critical for taking advantage of technical change and growth opportunities.

### Digital connectivity: prerequisites, constraints, and recent investments
- Complementary policies and conditions required to realize digital dividends:
  - Higher digital connectivity coupled with an improved business climate, strong investment in people’s education and health, and good governance would deliver digital dividends (World Bank, 2016).
  - Adequate digital infrastructure and a supportive business environment enable new forms of business that create jobs for both the educated and less educated.
- Quality and access considerations:
  - Important to consider infrastructure quality and costs to users.
  - Reforms to policy and regulatory frameworks to make broadband access more affordable, accessible and universal need to be accompanied by skills development.
  - Population capacity to access the Internet depends on cultural acceptance, supporting policy, and availability of smartphones and computers at the household level.
- Recent investments:
  - The region has been investing heavily in Information, Communications and Technology (ICT) infrastructure, including internet and mobile-cellular signal coverage.

### Data, scope and methodological approach
- Data scope and improvements over prior work:
  - Assesses a significantly higher number of ICT indicators.
  - Uses the most recent (2016–17) available data for a comprehensive set of countries based on data availability (193 economies).
- Unsupervised machine learning techniques:
  - Implemented k-means clustering to assess existence of global digital divide; algorithm run 100 times for different random initial configurations and best solution minimizing the objective is selected.
  - Cluster selection uses elbow technique, silhouette method, and gap statistics to derive K; K-means broadly groups countries into three general classes; optimal K taken as three for the world.
  - Implemented principal components analysis (PCA); PCA indicates heterogeneity in digital connectivity can be largely explained by the first principal component, which explains nearly half of the sample variation.
- Composite index and regressions:
  - Constructs an Enhanced Digital Access Index (EDAI) using the Mazziotta-Pareto methodology.
  - Estimates fractional logit regressions with EDAI as dependent variable using over 100 independent variables and step-wise regressions minimizing the quasi Akaike Information Criterion.
  - Models estimated for full sample and subgroups: Advanced Economies (AEs), Emerging Market and Middle-Income Countries (MICs), Low Income Developing Countries (LIDCs), and SSA; SSA-specific checks include CPIA components while controlling for per capita income.
- Data sources and software:
  - Primary data source: World Telecommunication/ICT Indicators Database, augmented by the UN E-Government Survey and UNESCO Institute for Statistics (UIS) database.
  - Country list and ISO codes: 193 countries based on data availability (Appendix II).
  - All estimations done using R software.
  - Averages for analytical groups calculated as weighted averages using PPP GDP shares from the World Economic Outlook database (IMF, 2019a).

### Composite index: EDAI (Enhanced Digital Access Index)
- Five fundamental categories (sub-indices) used:
  - (i) infrastructure;
  - (ii) knowledge;
  - (iii) affordability;
  - (iv) quality; and
  - (v) actual internet usage.
- Indicator set and rescaling:
  - Augmented DAI with indicators from the ICT Development Index of ITU and the Digitization Index (Katz and others, 2014).
  - All indicators rescaled to a [0, 100] interval following the transformation used by the Inclusive Internet Index (Facebook and Economist).
- Aggregation: Mazziotta-Pareto Index (MPI)
  - MP퐼퐼푖 = 푀푀푖 ⋅ (1 − 퐶퐶푉푉푖2) = 푀푀푖 − 푆푆푖 ⋅ 퐶퐶푉푉푖, where 푀푀푖, 퐶퐶푉푉푖, 푆푆푖 denote the i-th country mean, coefficient of variation, and its standard deviation.
  - MPI advantages: avoids imposing equal weights; introduces a penalty for units with unbalanced sub-index values; normalizes indicators independent of an “ideal unit”; simplifies computations.
- Caveat: country specific values for EDAI and sub-indices available in Appendix VI (in source).

### Global findings, digital divide, and distance-to-frontier (DTF)
- Evidence of a global digital divide with clustering of countries into three main groups; gap narrowed between 2003 and 2017.
- SSA distance-to-frontier (DTF):
  - DTF in 2003 = 45.1
  - DTF in 2017 = 17.3
- Comparative normalized indices (as reported):
  - DAI_NwwO = 69.8
  - DAI_SSSA = 24.7
  - EDAI_WwWO = 85.1
  - EDAI_SSSA = 67.8
- Geographic and income heterogeneity:
  - High-ranking regions: North America, Europe, Western Asia, Australia and Oceania.
  - Low-ranking regions: Sub-Saharan Africa (SSA) and LIDCs.
  - EDAI decile mapping reveals geography remains important to the digital divide.
  - EDAI distribution across income groups: AEs perform best, then MICs, then LIDCs; distribution more dispersed for LIDCs.

### Sub-index summary (PPP-weighted by GDP)
- Infrastructure:
  - World = 80.23
  - AEs = 87.83
  - MICs = 78.11
  - LIDCs = 54.64
- Quality:
  - World = 25.21
  - AEs = 31.26
  - MICs = 22.06
  - LIDCs = 15.62
- Affordability:
  - World = 20.47
  - AEs = 22.55
  - MICs = 19.18
  - LIDCs = 18.65
- Knowledge:
  - World = 78.99
  - AEs = 92.06
  - MICs = 73.54
  - LIDCs = 48.28
- Internet Usage:
  - World = 60.94
  - AEs = 84.80
  - MICs = 48.28
  - LIDCs = 24.67

### EDAI for SSA: patterns and heterogeneity
- SSA shows substantial heterogeneity; top-ranking SSA countries include Botswana, Cabo Verde, Gabon, Ghana, Lesotho, Mauritius, Rwanda, Seychelles, South Africa.
- Median for the world = 78; four SSA countries above median: Cabo Verde, Mauritius, Seychelles, South Africa.
- SSA vs. LIDC and world on sub-indices:
  - SSA similar to LIDC average.
  - SSA and LIDCs close to world in quality (maximum download speeds) and affordability (similar SMS and internet prices in US$ across countries).
  - SSA lags in infrastructure, internet usage, and knowledge (human capital).
- Affordability caveat:
  - Affordability comparable in US$ terms, but relative to per capita income affordability remains an issue for SSA; fixed broadband cost noted as highest in sub-Saharan Africa (as cited in source).

### Drivers of digital connectivity: fractional logit regressions and variable sets
- Methodology:
  - Fractional logit regressions to handle dependent variable bounded in [0,1] and extremes 0 and 1.
  - Step-wise regressions used to select explanatory variables minimizing quasi-Akaike Information Criterion.
  - Total candidate explanatory variables = 127, classified into 18 thematic groups.
- Table 2: counts by variable group (as presented)
  - Balance of Payments = 9
  - Climate = 3
  - Corruption, transparency and country risk = 8
  - CPIA = 16
  - Debt statistics = 6
  - Demographics = 2
  - Ease of doing business = 19
  - Education = 5
  - Employment = 7
  - Financial access = 5
  - Fiscal = 6
  - Geography = 4
  - Health = 11
  - Logistics = 5
  - Macro Indicators = 5
  - National accounts and real sector = 9
  - Social Development = 1
  - Urbanization = 6
  - Grand Total = 127
- Preprocessing:
  - Variables with absolute pairwise correlation > 0.9 addressed by removing the variable with the highest mean absolute correlation.

### Full sample regression highlights
- Variables positively associated with higher digital connectivity (selected):
  - Better business enabling and regulatory environment.
  - Higher tax revenue yield.
  - Higher share of renewable energy in total energy production.
  - Higher urban access to electricity and urbanization.
  - Higher private consumption (captures affordability and ownership of devices).
- Negative association highlighted:
  - Higher dependency on remittances associated with lower digital connectivity (remittances as % of GDP).
- Interpretation example (as presented):
  - A decrease of 1 percentage point in share of rural population leads to the e^(−0.656)−1 i.e., 0.48 odds increase in digital connectivity.
  - World average share of rural population ≈ 40 percent; AEs average = 21 percent.
  - If world average halves to 20 percent (a 20 percentage points decline), odds of higher digital connectivity would rise by 0.09 (as presented).

### SSA-specific empirical findings and magnitudes
- General SSA drivers:
  - Better business enabling and regulatory environment, financial access, urbanization, and availability of postal services are associated with higher digital connectivity.
  - Leveling the playing field for female entrepreneurs and reducing property registration costs are positively related to higher digital connectivity.
- Rural population effects:
  - A decrease of 1 percentage point in share of rural population leads 0.77 odds increase in digital connectivity.
  - Average share of rural population in SSA is about 57 percent compared to 20 percent average in AEs.
  - If the SSA average reduces to 21 percent (a 36 percentage points decline), the odds of higher digital connectivity would rise by 0.17.
- Financial access effects:
  - An increase of 1 percentage point in financial access as measured by account ownership leads to 3.1 odds increase in digital connectivity.
  - Average share of account ownership is 41 percent in SSA compared to 95 percent in AEs.
  - If SSA account ownership were to improve to AE levels, odds of higher digital connectivity would rise by 1.67.

### Key fractional logit estimates for SSA (selected coefficients from Table 3)
- Observations: 45
- Coefficients report changes in the odds ratio; value greater than 0 indicates increase in the odds ratio relative to the unconditional odds. Standard errors in parentheses. All regressions control for per capita GDP in PPP terms.
- Selected SSA column coefficients (coefficient, standard error):
  - Renewables share (% of energy consumption): 0.710 (0.521)
  - Registering Property - Procedures (number): -0.044 (0.034)
  - Registering Property - Cost (% of property value): -3.522 ** (1.657)
  - Starting a Business - Time - Women (days): -0.015 *** (0.003)
  - Account ownership (% of population ages 15+): 1.412 *** (0.461)
  - Tax revenue (% of GDP): 1.215 (1.745)
  - Private consumption expenditure (% of GDP): 0.226 (0.574)
  - Gross fixed capital formation (% of GDP): -0.517 (1.044)
  - Services, value added (% of GDP): 1.276 (0.834)
  - International Health Regulations capacity: 0.477 ** (0.221)
  - Rural population (% of total): -1.488 *** (0.503)
  - Population without postal services (% of total): -0.485 ** (0.197)

### CPIA-based regressions for SSA (selected)
- Fractional logit regressions using the 16 CPIA indicators (controlled for per capita income) identify only the CPIA environmental sustainability rating as surviving stepwise regression.
- Environmental sustainability rating coefficients by subgroup (coefficient, standard error; Observations):
  - SSA: 0.550 *** (0.203); Observations: 45
  - Oil Exporters: 2.305 (1.697); Observations: 7
  - Other Resource Intensive Exporters: 0.278 ** (0.106); Observations: 16
  - Non-resource Intensive Countries: 0.640 *** (0.164); Observations: 22
  - Countries in Fragile Situations: 0.573 ** (0.231); Observations: 17
- Interpretation: With the exception of oil exporters, the environmental sustainability rating variable is robustly related to digital connectivity, suggesting the role of responsive governance.

### Robustness checks and methodological notes
- Clustering robustness:
  - K-means clustering results robust to alternative K specifications (K=2 and K=4 yield similar groupings).
  - Hierarchical clustering (Divisive) considered and yielded similar results; K-means preferred due to computational and methodological considerations.
- Index robustness:
  - EDAI construction checked against alternative weighting schemes: equal, gradual, and Wroclaw weightings.
  - Mazziotta-Pareto aggregation allows summarizing indicators assumed not fully substitutable.
- PCA justification:
  - PCA’s first component explains nearly half of sample variation; first two components plotted to visualize groupings and justify quasi-linear composite index approach.
- Caveats:
  - PPP GDP per capita used as control in all regressions; 16 regional dummies used in world regression.
  - Potential endogeneity not addressed due to lack of instruments and cross-sectional data; results emphasize correlations, not causation.
  - CPIA variables used only in SSA regression leading to 110 variables in global regression.

### Policy implications and areas for action
- Broad policy levers associated with higher digital connectivity:
  - Improve business enabling and regulatory environment.
  - Strengthen fiscal capacity (tax revenue yield).
  - Expand renewable energy share and urban electricity access.
  - Promote urbanization and increase urban access to electricity.
  - Promote private consumption to enhance affordability and device ownership.
  - Expand financial access (account ownership, mobile/digital finance).
  - Improve logistics infrastructure (noted for MICs).
- SSA-specific policy actions:
  - Target infrastructure, internet usage, and knowledge (education) deficits to raise connectivity.
  - Complement investments in ICT infrastructure with skills development, affordability measures, and measures that enhance household access to devices and cultural acceptance of Internet use.
  - Reduce barriers for female entrepreneurs and lower property registration costs.
  - Expand postal services where lacking.
- Research and monitoring:
  - Channels through which variables affect connectivity and endogeneity concerns require further research.
  - EDAI can support multi-year analysis as data availability improves.

*Source: wpiea2019210-print-pdf - 1; 20; 36; APPENDIX II (IMF).*

### 1.      Large  uncertainties  characterize  the  major  trends  determining  the  future

### 1.      Large uncertainties characterize the major trends determining the future of work in sub-Saharan Africa (SSA) and connectivity is a key policy area.

### Demographic and policy context
- Population and labor supply projections and requirements:
  - Population projected to reach about 1.7 billion by 2040 from 1.0 billion currently.
  - United Nations projects a net increase in the working-age population (15–64 years) in SSA of about 20 million people per year.
  - The need to generate 20 million jobs per year during the next two decades is identified as the key challenge for SSA policy makers.
- Policy framing:
  - The October 2018 Regional Economic Outlook for sub-Saharan Africa (IMF, 2018) identifies connectivity as a key policy area to promote job creation and yield dramatic improvements in living conditions.
  - Connectivity extends beyond traditional physical infrastructure (roads, railways, ports) to digital connectivity, which is critical for taking advantage of technical change and growth opportunities.

### Digital connectivity, prerequisites and constraints
- Complementary policies and conditions required to realize digital dividends:
  - Higher digital connectivity coupled with an improved business climate, strong investment in people’s education and health, and good governance would deliver digital dividends (World Bank, 2016).
  - Adequate digital infrastructure and a supportive business environment enable new forms of business that create jobs for both the educated and less educated.
- Quality and access factors beyond infrastructure quantity:
  - Important to consider infrastructure quality and costs to users.
  - Reforms to policy and regulatory frameworks to make broadband access more affordable, accessible and universal need to be accompanied by skills development.
  - Population capacity to access the Internet depends on cultural acceptance, supporting policy, and availability of smartphones and computers at the household level.
- Recent investments:
  - The region has been investing heavily in Information, Communications and Technology (ICT) infrastructure, including internet and mobile-cellular signal coverage.

### Data, scope and methodological approach
- Data scope and improvements over prior work:
  - The paper assesses a significantly higher number of ICT indicators.
  - Uses the most recent (2016–17) available data for a comprehensive set of countries based on data availability (193 economies).
  - Applies several methodologies, including machine learning techniques, to investigate existence of a global digital divide and to formulate a composite index.
- Unsupervised machine learning techniques:
  - Implemented k-means clustering to assess existence of global digital divide.
  - Implemented principal components analysis (PCA) for dimensionality reduction to investigate variation in digital connectivity.
  - For k-means: the algorithm is run 100 times for different random initial configurations and the best solution minimizing the objective is selected.
  - Cluster selection uses elbow technique, silhouette method, and gap statistics to derive K.
  - K-means broadly groups countries into three general classes; optimal K taken as three for the world.
  - PCA indicates heterogeneity in digital connectivity can be largely explained by the first principal component, which explains nearly half of the sample variation, motivating a quasi-linear composite index.
- Composite index construction and categories:
  - Constructs an Enhanced Digital Access Index (EDAI) using the Mazziotta-Pareto methodology (De Muro and others, 2011).
  - Five fundamental categories (sub-indices) used:
    - (i) infrastructure;
    - (ii) knowledge;
    - (iii) affordability;
    - (iv) quality; and
    - (v) actual internet usage.
  - EDAI improves upon ITU’s Digital Access Index by expanding variables, using improved aggregation, and covering a larger number of countries with recent data.
- Regression analysis and explanatory variables:
  - Estimates fractional logit regressions with EDAI as dependent variable.
  - Uses over 100 independent variables and step-wise regressions to reduce explanatory variables by minimizing the quasi Akaike Information Criterion.
  - Explanatory variables include SDG indicators, ease of doing business, regulatory environment, transparency, country risk, employment, climate, corruption perceptions, and usual macroeconomic indicators.
  - Models estimated for full sample and subgroups: Advanced Economies (AEs), Emerging Market and Middle-Income Countries (MICs), Low Income Developing Countries (LIDCs), and SSA; SSA-specific checks include CPIA components while controlling for per capita income.
- Data sources and software:
  - Primary data source: World Telecommunication/ICT Indicators Database, augmented by the UN E-Government Survey and UNESCO Institute for Statistics (UIS) database.
  - Country list and ISO codes: 193 countries based on data availability (Appendix II).
  - All estimations done using R software.
  - Averages for analytical groups calculated as weighted averages using PPP GDP shares from the World Economic Outlook database (IMF, 2019a).

### Key empirical findings and patterns
- Existence and structure of digital divide:
  - There is a global digital divide with a clustering of countries into three main groups.
  - Variation in digital connectivity across countries can be broadly approximated by the first principal component, motivating a quasi-linear index.
  - Significant heterogeneity in digital connectivity across country groupings based on income and geography.
- SSA-specific performance and outliers:
  - The majority of SSA countries lag behind in digital connectivity, with exceptions: Botswana, Cabo Verde, Gabon, Lesotho, Mauritius, Seychelles, and South Africa, and LIDCs such as Ghana and Rwanda.
  - Among the five EDAI dimensions, SSA countries on average perform well in terms of affordability and quality, but do less well on infrastructure, internet usage, and knowledge.
- Regression results — drivers of digital connectivity:
  - Across samples, business and regulatory environment is significantly correlated with digital connectivity.
  - Fractional logit regressions underscore importance of:
    - regulatory and business enabling environment;
    - higher urbanization and urban access to electricity.
  - SSA-specific regression findings:
    - Better business enabling and regulatory environment, financial access, urbanization, and availability of postal services are associated with higher digital connectivity.
    - Leveling the playing field for female entrepreneurs and reducing property registration costs are positively related to higher digital connectivity.

### Methodology and robustness steps (high level)
- Clustering robustness:
  - K-means clustering results robust to alternative K specifications (K=2 and K=4 yield similar groupings).
  - Hierarchical clustering (Divisive) considered and yielded similar results, but K-means preferred due to computational and methodological considerations.
- Index robustness:
  - EDAI construction checked against alternative weighting schemes: equal, gradual, and Wroclaw weightings.
  - Mazziotta-Pareto aggregation allows summarizing indicators assumed not fully substitutable.
- PCA justification:
  - PCA’s first component explains nearly half of sample variation; first two components plotted to visualize groupings and justify quasi-linear composite index approach.

### Implications for policy and areas for action
- Digital connectivity policy levers highlighted by results:
  - Improve business enabling and regulatory environment to enhance digital connectivity.
  - Expand financial access and postal services to support digital uptake.
  - Promote urbanization policies and increase urban access to electricity to enable connectivity.
  - Reduce barriers for female entrepreneurs and lower property registration costs to positively affect digital connectivity.
  - Complement investments in ICT infrastructure with skills development, affordability measures, and measures that enhance household access to devices and cultural acceptance of Internet use.

*Source: wpiea2019210-print-pdf - 1.      Large  uncertainties  characterize  the  major  trends  determining  the  future (IMF).*

### 20.      We propose a new composite index to measure digital connectivity: EDAI.

### 20.      We propose a new composite index to measure digital connectivity: EDAI.

### Methodology: indicators and rescaling
- Five sub-categories of digital connectivity:
  - availability of infrastructure
  - affordability of access
  - educational level of the population
  - quality of information and communication technology services
  - internet usage
- Augmented DAI with:
  - indicators from the ICT Development Index of ITU
  - indicators from the Digitization Index (Katz and others, 2014)
- All indicators rescaled to a [0, 100] interval following the transformation used by the Inclusive Internet Index (Facebook and Economist), via:
  - normalization formula (transformation to [0, 100] as shown in source)

### Construction and aggregation: Mazziotta-Pareto Index (MPI)
- EDAI values constructed using the Mazziotta-Pareto Index (MPI) methodology.
- MPI formula (as presented):
  - MP퐼퐼푖 = 푀푀푖 ⋅ (1 − 퐶퐶푉푉푖2) = 푀푀푖 − 푆푆푖 ⋅ 퐶퐶푉푉푖
  - where 푀푀푖, 퐶퐶푉푉푖, 푆푆푖 denote respectively the i-th country mean, coefficient of variation, and its standard deviation.
- Advantages of MPI:
  - avoids imposing equal weights
  - introduces a penalty for units with unbalanced sub-index values (each lagging sub-index acts as a bottleneck)
  - normalizes indicators independent of an “ideal unit”
  - simplifies computations
- Country specific values for EDAI and sub-indices available in Appendix VI (in source).

### Global findings, digital divide, and distance-to-frontier (DTF)
- EDAI provides evidence of a global digital divide, but gap narrowed between 2003 and 2017.
- SSA distance-to-frontier (DTF):
  - DTF in 2003 = 45.1
  - DTF in 2017 = 17.3
- Comparative normalized indices cited (values as presented in source):
  - DAI_NwwO = 69.8
  - DAI_SSSA = 24.7
  - EDAI_WwWO = 85.1
  - EDAI_SSSA = 67.8
- Note: caveats due to differences in aggregation method and indicator set; reliance on DTF helps mitigate such caveats.

### Geographic and income-based heterogeneity
- High-ranking regions in digital connectivity:
  - North America, Europe, Western Asia, Australia and Oceania
- Low-ranking regions:
  - Sub-Saharan Africa (SSA) and LIDCs
- EDAI decile mapping reveals geography remains important to the digital divide.
- EDAI distribution across income groups:
  - AEs perform best, then MICs, then LIDCs
  - Distribution more dispersed for LIDCs

### Sub-index values (Table 1: Sub-indices values; PPP-weighted by GDP)
- Infrastructure:
  - World = 80.23
  - AEs = 87.83
  - MICs = 78.11
  - LIDCs = 54.64
- Quality:
  - World = 25.21
  - AEs = 31.26
  - MICs = 22.06
  - LIDCs = 15.62
- Affordability:
  - World = 20.47
  - AEs = 22.55
  - MICs = 19.18
  - LIDCs = 18.65
- Knowledge:
  - World = 78.99
  - AEs = 92.06
  - MICs = 73.54
  - LIDCs = 48.28
- Internet Usage:
  - World = 60.94
  - AEs = 84.80
  - MICs = 48.28
  - LIDCs = 24.67

### EDAI for SSA: patterns and within-region heterogeneity
- SSA shows substantial heterogeneity; SSA deciles highlight top-ranking SSA countries:
  - Botswana, Cabo Verde, Gabon, Ghana, Lesotho, Mauritius, Rwanda, Seychelles, South Africa
- Median for the world = 78; four SSA countries above median: Cabo Verde, Mauritius, Seychelles, South Africa.
- SSA vs. LIDC and world on sub-indices:
  - SSA similar to LIDC average
  - SSA and LIDCs close to world in:
    - quality (maximum download speeds)
    - affordability (similar SMS and internet prices in US$ across countries)
  - SSA lags in:
    - infrastructure
    - internet usage
    - knowledge (human capital)
- Note on affordability: comparable in US$ terms, but relative to per capita income affordability remains an issue for SSA; fixed broadband cost noted as highest in sub-Saharan Africa (as cited in source).

### Drivers of digital connectivity: fractional logit regressions
- Methodological approach:
  - fractional logit regressions to handle dependent variable bounded in [0,1] and extremes 0 and 1
  - step-wise regressions used to select explanatory variables minimizing quasi-Akaike Information Criterion
  - total candidate explanatory variables = 127, classified into 18 thematic groups (see Table 2)
- Table 2: counts by variable group (as presented)
  - Balance of Payments = 9
  - Climate = 3
  - Corruption, transparency and country risk = 8
  - CPIA = 16
  - Debt statistics = 6
  - Demographics = 2
  - Ease of doing business = 19
  - Education = 5
  - Employment = 7
  - Financial access = 5
  - Fiscal = 6
  - Geography = 4
  - Health = 11
  - Logistics = 5
  - Macro Indicators = 5
  - National accounts and real sector = 9
  - Social Development = 1
  - Urbanization = 6
  - Grand Total = 127
- Preprocessing: variables with absolute pairwise correlation > 0.9 addressed by removing the variable with the highest mean absolute correlation.
- Full sample regression results (variables reduced from 110 to 14 by stepwise selection):
  - Positive associations with higher digital connectivity:
    - better business enabling and regulatory environment
    - higher tax revenue yield
    - higher share of renewable energy in total energy production
    - higher urban access to electricity and urbanization
    - higher private consumption (captures affordability and ownership of devices)
  - Negative association highlighted: higher dependency on remittances associated with lower digital connectivity (remittances as % of GDP)
  - Interpretation example (odds change):
    - a decrease of 1 percentage point in share of rural population leads to the e^(−0.656)−1 i.e., 0.48 odds increase in digital connectivity (as presented)
    - world average share of rural population ≈ 40 percent; AEs average = 21 percent
    - if world average halves to 20 percent (a 20 percentage points decline), odds of higher digital connectivity would rise by 0.09 (as presented)
- Heterogeneity by income group regressions:
  - AEs:
    - only account ownership significantly related to digital connectivity (suggests promoting financial access, FinTech, mobile banking, mobile money, e-wallets)
  - MICs:
    - positive association with better regulatory and business enabling environment
    - better logistics
    - higher tax revenue capacity
  - LIDCs:
    - importance of higher electricity access in cities
    - improved financial access
    - improved business facilitating environment
- Caveats and limitations noted in source:
  - PPP GDP per capita used as control in all regressions; 16 regional dummies used in world regression
  - potential endogeneity not addressed due to lack of instruments and cross-sectional data; results emphasize correlations, not causation
  - CPIA variables used only in SSA regression leading to 110 variables in global regression

### Key implications (as derived in source)
- Policy areas associated with higher digital connectivity include:
  - improving business enabling and regulatory environments
  - strengthening fiscal capacity (tax revenue yield)
  - expanding renewable energy share and urban electricity access
  - promoting urbanization and private consumption to enhance affordability and device ownership
  - expanding financial access (account ownership, mobile/digital finance)
  - improving logistics infrastructure (noted for MICs)
- SSA-specific implications:
  - targeting infrastructure, internet usage, and knowledge (education) deficits could raise connectivity
  - affordability measured in US$ may be comparable, but affordability relative to income remains an issue
  - heterogeneity implies tailored country-level measures beyond income-based prescriptions

*Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019210-print-pdf.pdf*

### 36.      Results  for  the  SSA-specific  sample  (column  7)  further  underscore  the

### Results for the SSA-specific sample

### SSA regression findings and magnitudes
- Importance highlighted: business enabling and regulatory environment, financial access, and urbanization; leveling the playing field for female entrepreneurs; investing in government services and health; and improving regulatory environment.
- Controlling for income per capita, higher percentage of population without postal services and lack of health regulations seem to adversely affect digital connectivity.
- Rural population effects:
  - A decrease of 1 percentage point in share of rural population leads 0.77 odds increase in digital connectivity.
  - Average share of rural population in SSA is about 57 percent (average) compared to 20 percent average in AEs.
  - If the SSA average reduces to 21 percent, i.e., a 36 percentage points decline, the odds of higher digital connectivity would rise by 0.17.
- Financial access effects:
  - An increase of 1 percentage point in financial access as measured by account ownership leads to 3.1 odds increase in digital connectivity.
  - Average share of account ownership is 41 percent in SSA compared to 95 percent in AEs.
  - Hence if SSA account ownership were to improve to AE levels, odds of higher digital connectivity would rise by 1.67.

### Key fractional logit estimates for SSA (from Table 3)
- Observations: 45
- Coefficients report changes in the odds ratio; value greater than 0 indicates increase in the odds ratio relative to the unconditional odds. Standard errors in parentheses. All regressions control for per capita GDP in PPP terms.
- Selected SSA column coefficients (coefficient, significance stars, standard error):
  - Renewables share (% of energy consumption): 0.710 (0.521)
  - Registering Property - Procedures (number): -0.044 (0.034)
  - Registering Property - Cost (% of property value): -3.522 ** (1.657)
  - Starting a Business - Time - Women (days): -0.015 *** (0.003)
  - Account ownership (% of population ages 15+): 1.412 *** (0.461)
  - Tax revenue (% of GDP): 1.215 (1.745)
  - Private consumption expenditure (% of GDP): 0.226 (0.574)
  - Gross fixed capital formation (% of GDP): -0.517 (1.044)
  - Services, value added (% of GDP): 1.276 (0.834)
  - International Health Regulations capacity: 0.477 ** (0.221)
  - Rural population (% of total): -1.488 *** (0.503)
  - Population without postal services (% of total): -0.485 ** (0.197)

### Heterogeneity and comparison with other groups
- Differences between LIDCs and SSA:
  - Access to electricity and property registration are significant for LIDCs but not for SSAs.
  - Health regulation capacity, percentage of rural population, and population without postal services are significant for SSAs but not for LIDCs.
  - Discrepancy could reflect inclusion of MICs in the SSA sample or geographic differences among LIDCs not captured by dummy variables.

### CPIA-based regressions for SSA (Table 4)
- Fractional logit regressions using the 16 CPIA indicators (controlled for per capita income) identify only the CPIA environmental sustainability rating as surviving stepwise regression.
- Environmental sustainability rating coefficients by subgroup (coefficient, significance stars, standard error):
  - SSA: 0.550 *** (0.203); Observations: 45
  - Oil Exporters: 2.305 (1.697); Observations: 7
  - Other Resource Intensive Exporters: 0.278 ** (0.106); Observations: 16
  - Non-resource Intensive Countries: 0.640 *** (0.164); Observations: 22
  - Countries in Fragile Situations: 0.573 ** (0.231); Observations: 17
- Interpretation: With the exception of oil exporters, the environmental sustainability rating variable is robustly related to digital connectivity, suggesting the role of responsive governance.

### Policy-relevant conclusions and implications (from Conclusion)
- Digital connectivity is key to promoting job creation and improving living conditions, especially in SSA which needs to generate 20 million jobs per year in the next two decades.
- EDAI construction: a global index of digital connectivity (EDAI) was created using recent data and methodologies to assess SSA’s stance and main drivers; it can be used by policymakers to assess preparedness for the Fourth Industrial Revolution.
- Main empirical findings:
  - Evidence in favor of a global digital divide by clustering countries into three main groups.
  - Significant heterogeneity in digital connectivity across country groupings by income and geography.
  - Majority of SSA countries lag in digital connectivity; exceptions include MICs Botswana, Cabo Verde, Gabon, Lesotho, Mauritius, Seychelles, and South Africa and LICDs such as Ghana and Rwanda.
  - Among five EDAI dimensions, SSA on average performs well in affordability and quality, but lags in infrastructure, internet usage and knowledge.
  - Fractional logit regressions emphasize the importance of the business-enabling regulatory environment for improved digital connectivity. Higher urbanization, financial access, share of investment and private consumption, and share of renewable energy are also associated with digital connectivity.
  - For SSA specifically: better business enabling and regulatory environment, financial access, urbanization, and availability of postal services are associated with higher digital connectivity. Leveling the playing field for female entrepreneurs and reducing property registration costs are positively related to higher digital connectivity.
- Research caveats and avenues:
  - Channels through which variables affect connectivity and endogeneity concerns require further research.
  - EDAI can support multi-year analysis as data availability improves.

*wpiea2019210-print-pdf - 36.*

### APPENDIX II - Country groupings

### APPENDIX II - Country groupings

### Global groupings by income (Fiscal Monitor)

- Advanced Economies  
  AUS, AUT, BEL, CAN, CHE, CYP, CZE, DNK, ESP, EST, FIN, FRA, DEU, GBR, GRC, IRL, ISL, ISR, ITA, JPN, KOR, LVA, LTU, LUX, MCO, MLT, NLD, NOR, NZL, PRT, SGP, SVK, SVN, SWE, USA

- Emerging and Middle-Income Countries  
  AGO, ARE, ARG, AZE, BGR, BLR, BRA, BWA, CHL, CHN, CIV, COG, COL, CPV, DZA, DOM, ECU, EGY, GAB, GNQ, HRV, HUN, IDN, IND, IRN, KAZ, KWT, LBY, LKA, LSO, MAR, MEX, MUS, MYS, NAM, OMN, PAK, PER, PHL, POL, QAT, ROU, RUS, SAU, SMR, SRB, STP, SWZ, SYC, THA, TUR, UKR, URY, VEN, ZAF

- Low-Income Developing Countries  
  AFG, ALB, AND, ARM, ATG, BEN, BDI, BFA, BGD, BHR, BHS, BIH, BLZ, BOL, BRB, BRN, BTN, CMR, CAF, COD, COM, CRI, CUB, DJI, DMA, ERI, ETH, FJI, FSM, GEO, GHA, GIN, GMB, GNB, GRD, GTM, GUY, HND, HTI, IRQ, JAM, JOR, KEN, KGZ, KHM, KIR, KNA, LAO, LBN, LBR, LCA, LIE, MDA, MDG, MDV, MHL, MKD, MOZ, MLI, MNE, MNG, MMR, MRT, MWI, NER, NGA, NIC, NPL, NRU, PAN, PLW, PNG, PRK, PRY, RWA, SEN, SLE, SLB, SLV, SOM, SDN, SSD, SUR, SYR, TCD, TGO, TJK, TKM, TLS, TON, TTO, TUN, TUV, TZA, UGA, UZB, VCT, VNM, VUT, WSM, YEM, ZMB, ZWE

### Sub-Saharan Africa (SSA) groupings (Sub-Saharan African Regional Economic Outlook)

- Oil-exporting countries (SSA)  
  AGO, CMR, COG, GAB, GNQ, NGA, SSD, TCD

- Other resource-intensive exporters (SSA)  
  BWA, BFA, GHA, NAM, NER, SLE, SOM, TZA, ZAF, ZMB

- Non-resource-intensive exporters (SSA)  
  BEN, BDI, CIV, COM, CPV, ERI, ETH, GMB, GNB, KEN, LSO, MDG, MOZ, MUS, MWI, RWA, SEN, STP, SWZ, SYC, TGO, UGA

- Countries in Fragile situations (SSA)  
  BDI, CAF, COM, COG, CIV, COD, ERI, GIN, GMB, GNB, LBR, MWI, MLI, SSD, STP, TCD, TGO, ZWE

*Source: APPENDIX II - Country groupings*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019210-print-pdf.pdf_