## backgroundpapers

## Source details

**Canonical URL:** [backgroundpapers](https://www.imf.org/-/media/files/publications/reo/afr/2019/october/english/backgroundpapers.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/reo/afr/2019/october/english/backgroundpapers.pdf.md)
- [Structured JSON version](/-/media/files/publications/reo/afr/2019/october/english/backgroundpapers.pdf.json)

---

### Executive Summary — 1. NAVIGATING UNCERTAINTY
- Growth projections:
  - Growth in sub-Saharan Africa is projected to remain at 3.2 percent in 2019 and rise to 3.6 percent in 2020.
  - Growth is forecast to be slower than previously envisaged for about two-thirds of the countries in the region.
  - Growth prospects vary considerably across countries: non-resource-intensive countries average about 6 percent; resource-intensive countries about 2½ percent.
  - 24 countries, home to about 500 million people, will see their per capita income rise faster than the rest of the world; 21 countries are projected to have per capita growth lower than the world average.
- Inflation and debt:
  - Inflation is expected to ease going forward.
  - The average sub-Saharan African-wide debt burden is stabilizing, but elevated public debt vulnerabilities and low external buffers will continue to limit policy space in several countries.
- Downside risks:
  - External: rising protectionism; sharp increase in risk premiums or reversal in capital inflows owing to tightening global financial conditions; faster-than-anticipated slowdown in China and in the euro area.
  - Regional: climate shocks; intensification of security challenges; potential spread of the Ebola outbreak beyond the Democratic Republic of the Congo.
  - Domestic: fiscal slippages (including those ahead of elections) and lack of reform in key countries.
  - Medium-term upside: successful implementation of reforms, including in the context of the African Continental Free Trade Area (AfCFTA).
- Three-pronged strategy (summary of policy approach):
  - Carefully calibrate the near-term policy mix:
    - Room for supporting growth is mainly on the monetary policy side and is restricted to countries where inflation pressures are muted and growth is below potential.
    - If downside risks materialize, fiscal and monetary policy could be carefully recalibrated to support growth, consistent with debt sustainability and available financing, and as part of a credible medium-term adjustment plan.
    - In slowly growing countries, the pace of adjustment could be more gradual, provided financing is available, or its composition fine-tuned to minimize impact on growth.
    - In fast-growing countries with elevated debt vulnerabilities, priority remains rebuilding buffers.
  - Build resilience:
    - Mobilize domestic revenue, streamline inefficient subsidies, and improve public financial management to strengthen sovereign balance sheets and create fiscal space for development needs.
    - Promote economic diversification, improve macroeconomic policy frameworks, and reduce nonperforming loans (NPLs).
  - Raise medium-term growth:
    - Tackle tariff and nontariff barriers comprehensively in the context of the AfCFTA, develop regional value chains, and implement reforms to boost investment and competitiveness.

### Competition, Competitiveness, and Growth in Sub‑Saharan Africa
- State of competition:
  - More than 70 percent of the countries in the region are in the bottom half of countries globally in terms of competition indicators.
  - Firm markups are about 11 percent higher in sub-Saharan African countries relative to other emerging market economies and developing countries and are more persistent.
  - State-owned firms are more prevalent.
- Estimated economic effects:
  - Empirical analysis suggests increased competition can boost real per capita GDP growth rate by about 1 percentage point.
  - Channels: improved export competitiveness, productivity growth, and investment.
  - Competition lowers prices, especially for food and essential items, and helps increase labor’s share in output.
- Key regression and IV coefficients (selected, preserved exactly):
  - World sample (OLS) local competition coefficient = 1.500*** (0.529).
  - EMEDEV sample (OLS) local competition coefficient = 1.510** (0.628).
  - SSA sample (OLS) local competition coefficient = 1.035 (0.809).
  - IV-2SLS examples: world = 1.978* (1.063); EMEDEV = 2.646** (1.311); SSA = 3.337** (1.652).
  - Improvement in domestic competition in SSA from the median value to the top quartile of the global distribution implies a boost in real GDP per capita growth of about one percentage point.
  - Non-oil exports channel: improvement implies an increase in non-oil exports by about 2 percent of GDP for SSA.
  - Welfare/prices: sub-Saharan Africa shows a higher price level (about 20 percent) for the individual consumption basket relative to comparable countries; competition improvement from median to top quartile implies decline in that price level by about 10 percent.
- Firm-level findings (WBES and Orbis):
  - Markup defined as log ratio of sales to cost (turnover to costs) or Lerner Index proxies.
  - WBES-derived markup and Orbis measures both show:
    - Higher markups associated with lower investment (markup = -0.724*** (EMEDEV), -0.539*** (SSA) in WBES examples).
    - Higher markups associated with lower exports and lower labor share.
    - Orbis manufacturing sample: markup coefficients for investment example = -2.353*** (0.19), -2.420*** (0.21).
  - Annex Table 2.23 (selected exact coefficients):
    - Markup (cols. (1)-(3)): -0.844*** (0.19), -1.077*** (0.27), -1.037*** (0.27).
    - Markup (manufacturing, investment cols. (1)-(2)): -2.353*** (0.19), -2.420*** (0.21).
- Policy recommendations to improve product market competition:
  - Reduce structural and regulatory barriers.
  - Implement an effective competition policy framework comprising a strong competition law backed by an independent and adequately resourced competition authority.
  - Adopt trade and investment policies that encourage foreign competition.
  - Align fiscal and procurement policies to be supportive of competition.
- Data sources and coverage notes:
  - GCI (World Economic Forum) Pillar 6: Goods Market Efficiency; competition intensity indicator values between 1 and 7; GCI data used over 2006–17 for 44 SSA countries; regression analysis covers 25 SSA countries and 95 other countries (sample ends in 2014 for some variables).
  - WBES: 39 SSA countries over 2006–18; after cleaning about 10,000 SSA firm-level observations retained.
  - Orbis: 28 SSA countries over 2000-17; about 73 percent of SSA Orbis firms pertain to South Africa, Nigeria and Mauritius.
  - Product Market Liberalization dataset covers 14 SSA countries (accounting for about 80 percent of the region’s output).

### Domestic Arrears in Sub‑Saharan Africa: Causes, Symptoms, and Cures
- Magnitude and trends:
  - Domestic arrears have been pervasive and increased to about 3.3 percent of GDP in 2018 following the 2014 commodity price shock.
- Macro and micro effects (empirical findings):
  - Real GDP per capita growth: A 1 percentage point increase in arrears accumulation is associated with a -0.3 percentage point fall in real GDP per capita growth (Table A.3.4.1, Col. 1: Arrears accumulation, % GDP = -0.003**).
  - Banks’ asset quality (NPLs-to-gross loans): A 1 percentage point increase in arrears accumulation is associated with a 0.3 percentage point rise in the NPL ratio (Table A.3.4.2, Col. 1: Arrears accumulation, % GDP = 0.323*).
  - Tax revenue mobilization: A 1 percentage point increase in arrears accumulation is associated with a -0.213 percentage point change in tax revenue as a share of GDP (Table A.3.4.3, Col. 1: Arrears accumulation, % GDP = -0.213*).
  - Access to electricity: A 1 percentage point increase in arrears accumulation is associated with a -0.589 percentage point change in the percentage of the population with access to electricity (Table A.3.4.4, Col. 1: Arrears accumulation, % GDP = -0.589**).
  - Private sector (firm-level) outcomes:
    - Sales growth: arrears accumulation coefficient = -0.047*** (Table A.3.4.5, Col. 1).
    - Productivity growth: arrears accumulation coefficient = -0.054*** (Table A.3.4.5, Col. 2).
    - Employment growth: arrears accumulation coefficient = -0.025** (Table A.3.4.5, Col. 3).
- Interaction effects and heterogeneous impacts (total marginal effects, Annex Table A.3.4.6):
  - Government contract dummy (sales growth):
    - Yes: -0.069*** (0.021).
    - No: -0.041* (0.017).
  - Bank credit dummy (employment growth):
    - Yes: -0.030*** (0.011).
    - No: -0.022* (0.011).
  - Tax visit requirement dummy (employment growth):
    - Yes: -0.026*** (0.010).
    - No: -0.021** (0.010).
  - Prepaid sales, % total sales (employment growth):
    - 0%: -0.019*** (0.004).
    - 50%: -0.012*** (0.004).
    - 100%: -0.004 (0.006).
- Approaches to clearing and preventing arrears:
  - Clearing arrears: verification, prioritization, liquidation.
  - Preventing accumulation: public financial management reforms, building buffers, and timely external supports.
- Modeling and additional evidence:
  - DSGE model with domestic arrears financing captures transmission: arrears reduce entrepreneurs’ collateral value, tighten borrowing constraints, and lower fiscal multipliers.
  - Calibrations used in SSA application:
    - Government spending: 25 percent of GDP.
    - Domestic arrears: 20 percent of government spending.
  - Simulation experiment: unanticipated 1 percent of GDP increase in government expenditure.
  - Main simulation result: accumulation of domestic arrears lowers the fiscal multiplier; prolonged arrears financing can make the fiscal multiplier substantially lower and even negative (down to -0.2 Percent of GDP shown in scenarios).
- Synthetic Control Method (SCM) case results (selected SSA examples after 2015 oil price shock):
  - Selected cases with sharp buildups:
    - Chad: +2.3 ppts of GDP.
    - Angola: +5.7 ppts of GDP.
    - Congo Rep.: +3.5 ppts of GDP.
  - One year after the buildup (post-2015), real GDP per capita relative to synthetic counterfactual:
    - Chad: 11 percent lower than counterfactual.
    - Congo Rep.: 7 percent lower than counterfactual.
    - Angola: 7 percent lower than counterfactual.
  - Results remain below counterfactual through 2018 and are robust to donor-pool broadening and placebo checks.

### Methodological background and estimation notes
- Panel regression frameworks and firm-level augmented specifications:
  - Macro regressions: arrears accumulation expressed in percent of GDP (lagged by one year); RHS variables lagged by one year; country and year fixed effects; standard errors clustered at the country level.
  - Firm regressions: include conditioning variables and interactions; country-sector fixed effects; standard errors clustered at the country-sector level.
- Preferred macro magnitudes (parsimonious specifications):
  - Real GDP per capita growth: arrears accumulation, % GDP = -0.003** (implying -0.3 percentage point per 1 percentage point of GDP arrears).
  - Banks’ asset quality: arrears accumulation, % GDP = 0.323*.
  - Tax revenue: arrears accumulation, % GDP = -0.213*.
  - Access to electricity: arrears accumulation, % GDP = -0.589**.
- Robustness and inference:
  - Multiple robustness checks with additional macro controls reported; lagging RHS variables and fixed effects used to mitigate endogeneity concerns.
  - SCM placebo tests and sensitivity checks support identified effects for treated SSA cases.

### Statistical appendix, data vintage, and aggregation notes
- Data and projections: IMF staff estimates as of September 30, 2019, consistent with October 2019 WEO.
- Coverage: 45 sub-Saharan African countries in IMF African Department datasets.
- Country group thresholds:
  - Oil exporters: net oil exports ≥ 30 percent of total exports.
  - Other resource-intensive countries: nonrenewable natural resources ≥ 25 percent of total exports.
  - Middle-income threshold: US$1,025.00 (World Bank, Atlas method) for 2016–18 average per capita GNI.
  - Fragility CPIA cutoff: 3.2 (average over 2016–18).
- Aggregation methods: group composites weighted by GDP at PPP or market exchange rates as specified per table groupings (SA1–SA28 enumerated).

_Italic: Executive Summary and Annexes, Regional Economic Outlook: Sub‑Saharan Africa (October 2019), BACKGROUND PAPER: ONLINE ANNEXES (backgroundpapers.pdf), IMF staff estimates as of September 30, 2019._

### Executive Summary ......................................................................................................

### Executive Summary

### 1. NAVIGATING UNCERTAINTY

- Growth projections:
  - Growth in sub-Saharan Africa is projected to remain at 3.2 percent in 2019 and rise to 3.6 percent in 2020.
  - Growth is forecast to be slower than previously envisaged for about two-thirds of the countries in the region.
  - Growth prospects vary considerably across countries: non-resource-intensive countries average about 6 percent; resource-intensive countries about 2½ percent.
  - 24 countries, home to about 500 million people, will see their per capita income rise faster than the rest of the world; 21 countries are projected to have per capita growth lower than the world average.

- Inflation and debt:
  - Inflation is expected to ease going forward.
  - The average sub-Saharan African-wide debt burden is stabilizing, but elevated public debt vulnerabilities and low external buffers will continue to limit policy space in several countries.

- Downside risks:
  - External: rising protectionism; sharp increase in risk premiums or reversal in capital inflows owing to tightening global financial conditions; faster-than-anticipated slowdown in China and in the euro area.
  - Regional: climate shocks; intensification of security challenges; potential spread of the Ebola outbreak beyond the Democratic Republic of the Congo.
  - Domestic: fiscal slippages (including those ahead of elections) and lack of reform in key countries.
  - Medium-term upside: successful implementation of reforms, including in the context of the African Continental Free Trade Area (AfCFTA).

- Three-pronged strategy (summary of policy approach):
  - Carefully calibrate the near-term policy mix:
    - Room for supporting growth is mainly on the monetary policy side and is restricted to countries where inflation pressures are muted and growth is below potential.
    - If downside risks materialize, fiscal and monetary policy could be carefully recalibrated to support growth, consistent with debt sustainability and available financing, and as part of a credible medium-term adjustment plan.
    - In slowly growing countries, the pace of adjustment could be more gradual, provided financing is available, or its composition fine-tuned to minimize impact on growth.
    - In fast-growing countries with elevated debt vulnerabilities, priority remains rebuilding buffers.
  - Build resilience:
    - Mobilize domestic revenue, streamline inefficient subsidies, and improve public financial management to strengthen sovereign balance sheets and create fiscal space for development needs.
    - Promote economic diversification, improve macroeconomic policy frameworks, and reduce nonperforming loans (NPLs).
  - Raise medium-term growth:
    - Tackle tariff and nontariff barriers comprehensively in the context of the AfCFTA, develop regional value chains, and implement reforms to boost investment and competitiveness.

### 2. COMPETITION, COMPETITIVENESS, AND GROWTH IN SUB-SAHARAN AFRICA

- State of competition:
  - More than 70 percent of the countries in the region are in the bottom half of countries globally in terms of competition indicators.
  - Firm markups are about 11 percent higher in sub-Saharan African countries relative to other emerging market economies and developing countries and are more persistent.
  - State-owned firms are more prevalent.

- Economic effects of increased competition:
  - Empirical analysis suggests increased competition can boost real per capita GDP growth rate by about 1 percentage point.
  - Channels: improved export competitiveness, productivity growth, and investment.
  - Competition can substantially improve consumers' purchasing power by lowering prices of goods and services, especially food and other essential items.
  - Competition helps increase labor’s share in output, with potential important distributional consequences.

- Reform recommendations to improve product market competition:
  - Reduce structural and regulatory barriers.
  - Implement an effective competition policy framework comprising a strong competition law backed by an independent and adequately resourced competition authority.
  - Adopt trade and investment policies that encourage foreign competition.
  - Align fiscal and procurement policies to be supportive of competition.

- Empirical and data notes (Annex highlights):
  - Macro-level competition indicators are obtained from the World Economic Forum’s Global Competitiveness Index (GCI), available annually over 2006–17 for 44 sub-Saharan African countries; regression analysis covers 25 sub-Saharan African countries and 95 other countries (sample ends in 2014 due to availability of some variables).
  - Firm-level evidence: Annex tables and figures report sectoral profitability and markups based on World Bank Enterprise Survey (WBES) and Orbis data; markup defined as log ratio of sales to cost (or turnover to costs).

### 3. DOMESTIC ARREARS IN SUB-SAHARAN AFRICA: CAUSES, SYMPTOMS, AND CURES

- Magnitude and trends:
  - Domestic arrears have been pervasive in many countries, reflecting weak public financial management.
  - Arrears have increased in recent years to about 3.3 percent of GDP in 2018, following the 2014 commodity price shock.

- Effects of domestic arrears:
  - Domestic arrears negatively impact private sector activity and the delivery of social services.
  - Arrears increase banking sector vulnerabilities and undermine citizens’ trust in the government.
  - Arrears weaken the ability of fiscal policy to support growth, casting doubt on the merit of relying on arrears financing to avoid spending cuts.

- Causes of accumulation (high level):
  - Weak public financial management is a core factor behind pervasive arrears accumulation.

- Approaches to clearing and preventing arrears:
  - Clearing arrears: verification, prioritization, liquidation.
  - Preventing accumulation: public financial management reforms, building buffers, and timely external supports.

- Modeling and analysis:
  - Chapter includes a DSGE model with domestic arrears financing and analysis of macroeconomic effects of domestic arrears accumulation (detailed in the chapter and annex).

*Source: Executive Summary, Regional Economic Outlook: Sub‑Saharan Africa (October 2019).*

### Annex Table 2.4. Sub-Saharan Africa Country Coverage of the WEF Competition Index

### Annex Table 2.4. Sub-Saharan Africa Country Coverage of the WEF Competition Index

### WEF Competition Index: scope and components
- The subcomponents of the GCI used to capture the state of competition at the macro-level belong to Pillar 6: Goods Market Efficiency.
- For the regression analysis, the indicator capturing the intensity of local competition is used to capture the market structure and extent of domestic competition facing firms.
- Annex Table 2.5 — Components of the World Economic Forum’s Competition Index:
  - Competition → Domestic competition / Foreign competition
  - Domestic competition subcomponents:
    - Intensity of local competition
    - Effectiveness of anti-monopoly policy
    - Extent of market dominance
    - Effect of taxation on incentives to invest
    - Total tax rate
    - Number of procedures required to start a business
    - Time required to start a business
    - Agricultural policy costs
  - Foreign competition subcomponents:
    - Prevalence of trade barriers
    - Trade tariffs
    - Prevalence of foreign ownership
    - Business impact of rules on FDI
    - Burden of customs procedures
    - Imports as a percentage of GD
- Source for WEF methodology: World Economic Forum, 2018, Methodology and Computation of the Global Competitiveness Index 2017–2018.
- Note: The GCI has been revised since 2018. Data for the new GCI 4.0 version is available for 2018 only and is not included in the analysis as it is not comparable to earlier data. GCI 4.0 is a weighted average of 11 pillars; components of interest belong to Pillar 7 (Product Market). The values for the components of the GCI 4.0 range between 0 to 100 (best).

### Bertelsmann Stiftung Transformation Index (BTI): description and coverage
- The BTI assesses development and socioeconomic aspects; index is the average of two main components: democracy status and market economy status.
- The market economy status contains seven subcomponents; the subcomponents considered here are: market-based competition and anti-monopoly policy, which range between 0 and 10 (best).
- Data is based on opinion surveys among country experts and is available every two years between 2006 and 2018 for 36 sub-Saharan African countries (Annex Table 2.6).
- Additional note: As the index is not available annually, it is not used in the regression analysis.

### Firm-level profitability and markups: definitions and measurement
- Common indicators:
  - Lerner Index defined as the price-marginal cost difference relative to the price (P − MC)/P; in practice approximated by the ratio of Earnings Before Interest and Taxes (EBIT) to operating revenue (IMF, 2019).
  - Markups proxied following De Loecker and Warzynski (DW, 2012), where markup can be expressed using output elasticities (βk) and expenditure shares (αk) of inputs (see equation A.2.1 and A.2.2).
- Practical approximations with cross-sectional/limited panel data:
  - From WBES: EBIT not provided; firm profitability constructed as the difference between annual sales (proxy for revenue) and the cost of variable inputs.
  - Markup proxy from WBES: use log ratio of annual sales to variable costs (sum of labor cost and the cost of raw materials and intermediate goods) as a firm-specific proxy for α/(1−α) in equation (A.2.2).
  - From Orbis: two approaches to compute markups:
    - (a) OR/COGS (Operating Revenue to Cost of Goods Sold) as a proxy comparable to WBES-derived measure.
    - (b) Estimate output elasticities (the β’s in equation A.2.1) by sector using the Olley and Pakes (1996) control function approach and GMM, following DW and Diez and others (2019).

### Data sources, country coverage, and key sample sizes
- World Bank Enterprise Survey (WBES):
  - Focus: formal firms in manufacturing and services with 5 or more employees; mostly cross-sectional.
  - WBES data used in the chapter comprises 39 sub-Saharan African countries over 2006–18 (Annex Table 2.7).
  - Data cleaning: variables trimmed at top and bottom percentiles; no sector with less than 100 observations included in regression analysis.
  - After cleaning, about 10,000 firm-level observations are retained for sub-Saharan Africa and about 41,000 observations for other emerging market and developing economies (Annex Table 2.8).
  - Example country-year-observation entries (from Annex Table 2.7): Angola 2006, 2010 — 238; Nigeria 2007, 2014 — 1,355; South Africa 2007 — 677; Zambia 2007, 2013 — 540.
- Orbis (Bureau van Dijk):
  - Harmonized cross-country financial information for private and listed firms; raw data undergoes extensive cleaning (drop duplicates, outliers, scale for comparability).
  - Data for 28 sub-Saharan African countries over 2000-17 (with gaps for some countries) are obtained (Annex Table 2.9).
  - About 73 percent of the firms for sub-Saharan Africa in Orbis pertain to three countries: South Africa, Nigeria and Mauritius.
  - Sectoral composition: about 30 percent of observations in the emerging markets sample and 40 percent in the sub-Saharan Africa sample belong to the manufacturing sector; the rest dominated by services (Annex Table 2.10).
  - Example Orbis country-year-observation entries (Annex Table 2.9): Mauritius 2001 - 2017 — 984; South Africa 2000 - 2017 — 3,432; Nigeria 2001 - 2017 — 1,334; Zambia 2002 - 2017 — 147.
- Industry coverage notes (WBES and Orbis):
  - WBES industry counts (selected entries from Annex Table 2.8): Food products and beverages — 7,777 (Emerging Market Economies and Developing Countries) and 2,693 (Sub-Saharan Africa); Manufacture of basic metals — 125 (EM) and 168 (SSA); Total listed as 412019979 (as presented in Annex Table 2.8).
  - Orbis industry counts (selected entries from Annex Table 2.10): Manufacturing — 573,862 (Emerging Market Economies and Developing Countries) and 33,85 (Sub-Saharan Africa entry as in table); Wholesale/retail trade; repair of — 668,317 (EM) and 1,326 (SSA).
- Notes on estimation:
  - To estimate β’s (output elasticities) a two-step approach per NACE Rev. 2 industry is used: first estimate expected output from variable inputs and fixed capital; residuals interpreted as productivity; second step uses Markov assumption on productivity and GMM to obtain β (see equation A.2.3; methodology in Diez and others 2019).

### Product market liberalization and competition policy framework data
- Product Market Liberalization dataset:
  - Source: Alesina and others (forthcoming).
  - Sample includes 14 sub-Saharan African countries: Burkina Faso, Cameroon, Côte D’Ivoire, Ethiopia, Ghana, Kenya, Madagascar, Mozambique, Nigeria, Senegal, South Africa, Tanzania, Uganda and Zimbabwe.
  - These 14 countries account for about 80 percent of the region’s output.
  - Database covers reforms in trade, current and capital accounts, telecommunications and electricity, agriculture, and domestic financial liberalization; data available generally from 1973 to 2014.
- Data on Competition Policy Frameworks:
  - World Bank – African Competition Forum (World Bank, 2016): surveys over 2010-15, covers 36 countries; provides cross-sectional information on existence and enforcement of competition laws, price controls, independence, budgets and staff size of competition authorities.
  - IMF Survey of Competition Authorities: in-house survey conducted May-June 2019; elicited responses from 37 jurisdictions (29 separate jurisdictions and one regional body).
  - Key survey findings:
    - Most countries have a competition law or unified legal framework to deal with competition issues (Annex Figure 2.7).
    - Gap between legal setup and effective implementation: on average, a lag of about 6 years is reported.
    - Only in 32 percent of the cases, implementation of anti-trust frameworks occurred in tandem with legal setup.
    - In at least three cases, it took more than a decade for the legal framework to be implemented.

### Variable definitions and data sources (selected entries from Annex Table 2.11)
- Antitrust frameworks data — Data on anti-trust frameworks — World Bank (2016)
- Competition frameworks in SSA — Data on competition frameworks — IMF, AFR survey
- Enterprise Survey data — Diverse firm level data — World Bank
- General gov. gross debt to GDP — In percent — IMF, WEO database
- Global Competitiveness Index — Scores — World Economic Forum
- Market liberalization data — Scores — Alesina, et al (2019)
- Orbis data — Diverse firm level data — ORBIS Bureau van Dijk
- Population — Millions — World Bank, WDI
- Private investment — In percent of GDP — IMF, WEO database
- Real GDP growth in trading partners — In percent — IMF, WEO database
- Real price of investment goods — Index — PWT 9.0
- Years of schooling — Years — World Bank, WDI

*Source: IMF — backgroundpapers - Annex Table 2.4. Sub-Saharan Africa Country Coverage of the WEF Competition Index (backgroundpapers.pdf).*

### Annex Figure 2 .7. Anti-Trust   Framework S urvey R esults

### Annex Figure 2 .7. Anti-Trust   Framework S urvey R esults

### Survey note
- The scores denote the percentage of respondent authorities who answered yes to the question. Total respondents equal to 37.

### Estimation methodologies
- Growth regressions estimated using equation (A.2.4): dependent variable y_{i,t} represents alternative macroeconomic indicators (real GDP per capita growth rate, share of private investment in GDP, non-oil exports in percent of GDP, measures of innovation and technological readiness). Com_{i,t} is the World Economic Forum Intensity of Local Competition index (values between 1 and 7). X_{i,t} is a vector of controls including lagged (log) real GDP per capita, share of investment in output, trade openness, change in terms of trade, years of schooling, trading partner growth, institutional quality, and public debt in percent of GDP. α_i and α_t are country and year fixed effects. Model estimated using ordinary least squares (OLS) with standard errors clustered at the country level.
- Welfare regressions estimated using equation (A.2.5): dependent variable P_i is an internationally comparable price (ICP, year 2011) for products/services or aggregate individual consumption basket. Com_i is the competition intensity indicator (1 to 7). X includes (log) real GDP per capita, logistical efficiency, trade openness, and FDI in percent of GDP. EME, DEV and SSA are dummy variables for emerging market economy, developing country, and sub-Saharan African country, respectively. Equation estimated by OLS with robust standard errors.
- Endogeneity addressed using IV-2SLS with two instruments: (i) the first two lagged values of the competition index; and (ii) a regional competition index equal to the average of competition indices for nearby countries.

### Estimation results: Growth (key findings)
- Competition is positively associated with higher economic growth in both the world and EMEDEV samples:
  - World sample: Local competition coefficient = 1.500*** (0.529)
  - EMEDEV sample: Local competition coefficient = 1.510** (0.628)
  - SSA sample (OLS): Local competition coefficient = 1.035 (0.809)
  - Alternative specifications (cols. 4-6): World = 1.353*** (0.502); EMEDEV = 1.268** (0.593); SSA = 0.233 (0.589)
- Based on these results: an increase in domestic competition in SSA from the median value to the top quartile of the global distribution implies a boost in real GDP per capita growth of about one percentage point.
- Restricting the sample to sub-Saharan African countries, the coefficient remains positive but loses statistical significance (p-value: 0.2).
- IV-2SLS results (Annex Table 2.14) using instruments (lagged competition; regional average competition) generally support positive association for world and EMEDEV samples; columns reported use the same dependent variable (real GDP growth per capita in PPP terms).

### Channels: investment, exports, productivity, innovation
- Investment:
  - Positive but statistically weak association between competition and private investment rate in world and EMEDEV samples.
  - Significantly positive relationship between competition and investment in SSA (Annex Table 2.15; example coefficient shown in IV-2SLS panels: SSA column coefficients include 3.337** (1.652) in col. 3 and 1.161 (1.751) in col. 6 for local competition in various IV specifications).
- Non-oil exports:
  - Strong positive association between competition and non-oil exports.
  - Improvement in domestic competition in SSA from the median value to the top quartile of the global distribution implies an increase in non-oil exports by about 2 percent of GDP.
- Labor productivity:
  - Annex Table 2.15 (cols. 7-9) shows labor productivity improves when competition strengthens (example IV coefficients reported: 11.624** (5.303) in col. 7; 11.532 (7.933) in col. 8; 6.814 (4.254) in col. 9).
- Innovation and technological readiness:
  - Competition is strongly positively associated with greater innovation and technological readiness across countries (Annex Table 2.16).
  - Example coefficients (OLS):
    - Innovation index: World = 0.168*** (0.040); EMEDEV = 0.181*** (0.048); SSA = 0.108 (0.086)
    - Technological readiness index: World = 0.166*** (0.051); EMEDEV = 0.200*** (0.054); SSA = 0.134** (0.054)
  - Investment is also strongly associated with innovation (Investment coefficient in innovation regressions: 11.835*** (3.852) for world sample).

### Estimation results: Welfare (prices)
- Without competition index: prices of most products and services (food, utilities, furniture, health services, communication, education, machinery and equipment) are on average significantly higher in sub-Saharan Africa than in other EMEDEV countries (Annex Table 2.17).
  - Higher price levels across product categories translate into a higher price level (about 20 percent) for the individual consumption basket in sub-Saharan Africa relative to other countries at a similar level of development.
  - The sub-Saharan African dummy is statistically insignificant for several product categories, implying parity with advanced countries for those goods/services.
- Including local competition intensity and foreign competition indicators (trade openness, FDI):
  - Competition is strongly negatively associated with the prices of most goods and services (Annex Table 2.18 summary).
  - Moving from the median value of the competition index for sub-Saharan Africa to the top quartile of the global distribution implies, on average, a decline in the price level of the individual consumption basket by about 10 percent.
  - Greater trade openness and FDI are also strongly negatively associated with price levels across countries.
  - Including these variables lowers the price difference between sub-Saharan Africa and other EMEDEVs, but the differential is not eliminated in most cases.

### Key reported coefficients and statistics (selected)
- Total respondents to anti-trust framework survey = 37.
- Competition index range = 1 to 7 (best).
- Selected coefficients (Annex Table 2.13, OLS, real GDP per capita growth, competition variable):
  - World (col. 1): 1.500*** (0.529)
  - EMEDEV (col. 2): 1.510** (0.628)
  - SSA (col. 3): 1.035 (0.809)
  - World (col. 4): 1.353*** (0.502)
  - EMEDEV (col. 5): 1.268** (0.593)
  - SSA (col. 6): 0.233 (0.589)
- Selected control coefficients (Annex Table 2.13):
  - Real GDP per capita (lag): -14.122*** (1.947) in col. 1; -17.050*** (2.084) in col. 4.
  - Investment: 14.997*** (3.495) in col. 1; 12.025*** (3.138) in col. 4.
  - Public debt: -0.053*** (0.018) reported in col. 1 (and significant in other samples).
- IV-2SLS selected local competition coefficients (Annex Table 2.14 IV panels using lagged competition and regional average instruments):
  - Examples: 1.978* (1.063) in world col. 1; 2.646** (1.311) in EMEDEV col. 2; 3.337** (1.652) in SSA col. 3.
- Non-oil exports channel (Annex Table 2.15, IV examples):
  - Local competition coefficients (cols. 4-6, IV): 1.598* (0.832) world; 1.186 (0.884) EMEDEV; 2.704* (1.544) SSA.
- Labor productivity IV coefficients (Annex Table 2.15, cols. 7-9 examples): 0.916** (0.452) world; 1.090* (0.556) EMEDEV; 1.200 (0.934) SSA.
- Innovation and technological readiness OLS coefficients (Annex Table 2.16):
  - Innovation (cols. 1-3): 0.168*** (0.040) world; 0.181*** (0.048) EMEDEV; 0.108 (0.086) SSA.
  - Technological readiness (cols. 4-6): 0.166*** (0.051) world; 0.200*** (0.054) EMEDEV; 0.134** (0.054) SSA.
- Welfare: estimated higher price level (about 20 percent) for the individual consumption basket in SSA relative to comparable countries; competition improvement from median to top quartile implies decline in individual consumption basket price by about 10 percent.

*Source: IMF Staff Calculations, BACKGROUND PAPER: ONLINE ANNEX 2—COMPETITION, COMPETITIVENESS, AND GROWTH IN SUB-SAHARAN AFRICA*

### Annex Table 2.18. Internationally Comparable Price Levels Across Regions

### Annex Table 2.18. Internationally Comparable Price Levels Across Regions

### Regional price-level findings
- On average, east Africa has the lowest prices for goods, followed by west Africa.
- Prices of most services (such as health, communications, restaurants and hotels, as well as utilities) are not statistically different among the regions (Annex Table 2.19).
- Higher price levels in the services sector are consistent with the higher firm markups across regions in the services sector reported in Annex Figure 2.3.
- Differentiating countries by exchange rate regime (CFA franc zone vs. non-CFA franc zone) shows no statistically significant difference in price levels in most cases, except for some nontradable sectors such as health services, communication, and recreation (Annex Table 2.20).
- It is noted that proxies for competition considered here may not fully account for actual competition.

### Selected regression coefficients and significance (internationally comparable prices by item)
- Local competition coefficients (examples shown across items and specifications):
  - Local competition = -0.082**, -0.086**, 0.008, 0.009, -0.294***, -0.306***, -0.129***, -0.137***, 0.042, 0.034, -0.074, -0.082*, -0.058**, -0.062**
  - Local competition = -0.145*, -0.156**, -0.107*, -0.111*, -0.155**, -0.170**, -0.138***, -0.148***, -0.144***, -0.152***, -0.106***, -0.115***, -0.120***, -0.127***
- Real GDP per capita (log) coefficients (examples):
  - Real GDP per capita (log) = 0.165***, 0.163***, 0.176***, 0.169***, 0.240***, 0.237***, 0.199***, 0.197***, 0.295***, 0.291***, 0.218***, 0.217***, 0.008, 0.007
  - Real GDP per capita (log) = 0.326***, 0.324***, 0.069*, 0.066, 0.211***, 0.213***, 0.224***, 0.222***, 0.283***, 0.284***, 0.208***, 0.207***, 0.171***, 0.169***
- Emerging market and developing country dummies (examples indicate lower internationally comparable prices relative to advanced countries):
  - Emerging market = -0.320***, -0.306***, -0.209***, -0.203**, -0.279**, -0.242**, -0.230***, -0.204***, -0.588***, -0.559***, -0.359***, -0.338***, -0.072*, -0.061
  - Developing country = -0.344***, -0.330***, -0.399***, -0.403***, -0.392***, -0.351***, -0.346***, -0.318***, -0.665***, -0.635***, -0.459***, -0.435***, -0.114**, -0.103**
- Sub-Saharan Africa dummy (examples):
  - Sub-Saharan Africa = -0.107, -0.086, -0.090, -0.088, -0.581***, -0.525***, -0.203**, -0.164**, -0.024, 0.019, -0.418***, -0.385***, -0.003, 0.013
  - Sub-Saharan Africa = -0.057, -0.004, -0.612***, -0.592***, -0.215, -0.152, -0.252***, -0.204**, -0.044, -0.009, -0.260***, -0.222***, -0.336***, -0.302***
- Logistics index coefficients (examples indicating positive association with price levels):
  - Logistics index = 0.046, 0.077, 0.159*, 0.174*, 0.109, 0.186*, 0.064, 0.119**, 0.275***, 0.337***, 0.183***, 0.227***, 0.003, 0.026
  - Logistics index = 0.486***, 0.560***, 0.062, 0.093, 0.026, 0.104, 0.091, 0.156**, 0.083, 0.129, 0.163***, 0.216***, 0.132**, 0.180***
- Trade openness coefficients (examples showing negative association with prices):
  - Trade openness = -0.001***, -0.001***, -0.000, -0.001, -0.001**, -0.002***, -0.001**, -0.001***, -0.001***, -0.002***, -0.001***, -0.001***, -0.000**, -0.001**
- FDI to GDP (examples negative and sometimes significant):
  - FDI to GDP = -0.002**, -0.001, -0.005***, -0.003***, -0.004, -0.003**, -0.001**

### Regional dummies and CFA vs non-CFA (selected coefficients)
- East Africa coefficients (examples):
  - East Africa = -0.236***, -0.217**, -0.047, -0.050, -0.669***, -0.620***, -0.311***, -0.277**, -0.151, -0.113, -0.542***, -0.512***, -0.024, -0.010
  - East Africa = -0.198, -0.151, -0.600***, -0.586***, -0.261, -0.199, -0.368***, -0.324***, -0.284*, -0.249, -0.367***, -0.334***, -0.454***, -0.424***
- West Africa coefficients (examples):
  - West Africa = -0.048, -0.026, -0.199*, -0.200*, -0.600***, -0.541***, -0.280***, -0.239**, -0.103, -0.057, -0.478***, -0.442***, 0.036, 0.053
  - West Africa = -0.146, -0.090, -0.730***, -0.713***, -0.173, -0.101, -0.229**, -0.177, -0.006, 0.035, -0.268***, -0.228***, -0.346***, -0.310***
- Central Africa and Southern Africa coefficients (examples show heterogeneous patterns across items; some positive coefficients for Central Africa in select specifications):
  - Central Africa examples: -0.023, 0.009, -0.086, -0.078, -0.481***, -0.404**, -0.016, 0.039, 0.176, 0.238, -0.234, -0.188, -0.036, -0.014
  - Southern Africa examples: -0.068, -0.046, -0.011, -0.004, -0.505***, -0.451***, -0.077, -0.038, 0.119, 0.163, -0.294***, -0.262***, -0.006, 0.010
- CFA franc zone dummy examples (CFA franc zone = 1 if country is part of CFA franc zone):
  - CFA franc zone = 0.077, 0.077, -0.007, -0.009, -0.007, -0.009, 0.134*, 0.135*, 0.136, 0.138, 0.154, 0.156, -0.006, -0.005
  - CA franc zone = 0.208*, 0.211*, 0.007, 0.008, 0.346***, 0.350***, 0.175**, 0.177**, 0.154, 0.156, 0.088, 0.090, 0.078, 0.079

### Model fit and sample sizes (examples)
- Observations in price-level regressions repeatedly reported as: 122, 121, 122, 121, 122, 121, 122, 121, 122, 121, 122, 121, 122, 121 (across different columns/items).
- R-squared examples across specifications: 0.658, 0.669, 0.656, 0.667, 0.599, 0.615, 0.690, 0.711, 0.753, 0.762, 0.823, 0.830, 0.202, 0.224; other reported R-squared values include 0.711, 0.720, 0.625, 0.634, 0.402, 0.416, 0.692, 0.710, 0.641, 0.647, 0.799, 0.812, 0.756, 0.770.

### Notes on dependent variable and controls
- Dependent variable is internationally comparable price of the respective items.
- Emerging market, developing country, sub-Saharan Africa (and regional) dummies are (mutually exclusive) with advanced country as the base category.
- CFA franc zone is a dummy variable equal to one if the country is part of the CFA franc zone and zero otherwise.
- All specifications include a constant. Robust standard errors are computed.
- ***, **, * indicate statistical significance at the 1, 5 and 10 percent levels, respectively.

### Firm dynamics and competition (World Bank Enterprise Survey and Orbis results)
- Regression setup (WBES): Dependent variables normalized by value added and log transformed (except exports). Explanatory variables include markups (competition intensity), firm-specific controls (size in logs; Direct Exports; Foreign ownership dummy; Private ownership dummy; Age; competitor count dummy for <5 competitors), and macro controls (log real GDP per capita PPP, industry, country, and year fixed effects). Benchmark regressions restricted to industries with at least 100 observations.
- Markup and investment:
  - Higher firm markups are associated with lower investment with an elasticity of about 0.5-0.7, highly statistically significant across both EMEDEV and SSA samples.
  - Result robust to alternative specifications (scaling by sales instead of value added and including other controls).
  - Majority private-owned firms invest more relative to public firms.
  - Larger and older firms tend to invest less.
  - No strong evidence that foreign ownership or exporting status matter for investment.
- Markup and exports (WBES benchmark specification):
  - Higher markups are associated with significantly lower exports in both the EMEDEV and SSA samples.
  - Firms facing fewer competitors and older firms export less; larger and foreign firms tend to export more.
- Selected WBES coefficient examples (columns labeled Investment, Exports, Labor Share; EMEDEV and SSA):
  - Markup = -0.724*** (EMEDEV), -0.539*** (SSA) in an investment/export/labor-share set of regressions; other reported markup coefficients include -0.168***, -0.053***, -1.236***, -1.311*** with standard errors shown in parentheses.
  - Log real GDP per capita (PPP) examples: 0.335, 0.107, -0.114*, -0.043, 0.125**, 0.223**
  - No. of competitors examples: -0.015, 0.026, -0.124***, -0.027***, -0.011, 0.003
  - Size (in logs) examples: -0.193***, -0.116***, 0.142***, 0.078***, -0.000, -0.000
  - Observations: 17,933 (EMEDEV investment), 3,598 (SSA investment), 41,956 (EMEDEV exports), 8,110 (SSA exports), 41,956 (EMEDEV labor share), 8,110 (SSA labor share).
  - R-squared examples: 0.117, 0.068, 0.158, 0.164, 0.408, 0.407.
- Markup and labor share:
  - Effect of markups on labor share is negative and significant across samples, with an elasticity of about one.
  - Older, exporting, and foreign-owned firms tend to have lower labor shares.
- Orbis database regressions (examples across investment, labor share, labor productivity growth, total factor productivity growth):
  - Markup coefficients (examples): -1.275***, -1.313***, -1.421***, -0.580***, -0.587***, -0.687***, -0.829***, -0.841***, -0.828***, -0.832***, -0.827***, -0.728*** (standard errors reported in parentheses).
  - Firm size, Age, Log Real GDP per capita, Real GDP growth, Trade openness, Institutional quality, Labor productivity (lag), Total factor productivity (lag) reported with coefficients and significance in various columns; see table for exact values.
  - Observations and No. of firms reported for different specifications: Observations examples = 19,422; 19,018; 9,913; 33,563; 32,951; 13,206; 12,736; 12,477; 5,433; 105,760; 101,353; 73,095. No. of firms examples = 7,769; 7,637; 3,051; 11,296; 11,124; 3,529; 5,281; 5,200; 1,932; 68,368; 64,491; 52,542.
  - R-squared examples: 0.048, 0.048, 0.073, 0.107, 0.108, 0.110, 0.122, 0.124, 0.115, 0.358, 0.356, 0.356.
- Notes on firm-behavior regressions:
  - Dependent variables: (log) investment to value added (cols. (1)-(3)), (log) exports to value added (cols. (3)-(4) for WBES), (log) share of labor in value added (cols. (4)-(6)), labor productivity growth (cols. (7)-(9)), total factor productivity growth (cols. (10)-(12)).
  - Markup defined as log of sales to cost of inputs.
  - No. of competitors is binary = 1 if reported number of competitors < 5, else 0.
  - Foreign and private ownership binary = 1 if >50% share, else 0.
  - All specifications include log real GDP per capita, a constant, and industry, year, and country fixed effects (WBES). Orbis regressions include firm/country/year fixed effects and industry x year effects as noted.
  - Clustered or robust standard errors reported in parentheses. ***, **, * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.

*Source: IMF staff estimates.*

### Annex Table 2.23. Sub-Saharan Africa: Competition and Firm Behavior

### Annex Table 2.23. Sub-Saharan Africa: Competition and Firm Behavior

### Key regression findings — all samples (investment, labor share, labor productivity growth, TFP growth)
- Markup
  - Columns (1)-(3): Markup -0.844*** (0.19), -1.077*** (0.27), -1.037*** (0.27)
  - Columns (4)-(6): Markup -0.699*** (0.14), -0.848*** (0.20), -0.865*** (0.21)
  - Columns (7)-(9): Markup -0.304 (0.21), -0.317 (0.20), -0.333 (0.20)
  - Columns (10)-(12): Markup -0.659*** (0.05), -0.562*** (0.07), -0.562*** (0.08)
- Age
  - Columns (1)-(3): Age -0.0235 (0.02), -0.0282 (0.03), -0.0255 (0.03)
  - Columns (4)-(6): Age 0.106*** (0.02), 0.0799** (0.02), 0.0769*** (0.02)
  - Columns (7)-(9): Age 0.0557*** (0.01), 0.0442*** (0.01), 0.0403** (0.01)
  - Columns (10)-(12): Age -0.0383*** (0.00), -0.0345*** (0.01), -0.0430*** (0.01)
- Log Real GDP per capita
  - Columns (1)-(3): (blank) 0.554 (0.48), 0.762 (0.60), 0.778 (0.80)
  - Columns (4)-(6): 0.307 (0.30), 1.034* (0.40), 0.579 (0.42)
  - Columns (7)-(9): -0.517 (0.61), -0.246 (0.55), 0.0258 (0.54)
  - Columns (10)-(12): 0.0436 (0.07), 0.00669 (0.12), 0.22 (0.22)
- Firm size
  - Columns (1)-(3): 0.275*** (0.07), 0.265*** (0.07), 0.0909 (0.06)
  - Columns (4)-(6): 0.09990 (0.06), 0.261* (0.11), 0.263* (0.10)
  - Columns (7)-(9): 0.0165 (0.01), 0.0186 (0.01)  (note: formatting in source shows fewer cells)
- Real GDP growth / Trade openness / Institutional quality
  - Real GDP growth coefficients reported in some columns: 0.0126 (0.01), -0.0175* (0.01), -0.0111 (0.01), -0.00996* (0.00)
  - Trade openness coefficients reported in some columns: -0.0056 (0.01), -0.004 (0.01), -0.0054 (0.01), 0.00112 (0.00)
  - Institutional quality coefficients reported in some columns: 0.0245 (0.19), 0.113 (0.12), -0.0938 (0.16), -0.0817 (0.05)
- Productivity lags
  - Labor productivity (lag): -0.186** (0.07), -0.172* (0.07), -0.165* (0.07)
  - Total factor productivity (lag): -0.0116 (0.04), -0.0509 (0.08), -0.0604 (0.08)
- Sample size, fit, and controls
  - Observations across columns: 3,192; 1,923; 1,879; 4,105; 2,376; 2,320; 1,123; 1,123; 1,104; 4,496; 2,209; 2,159
  - R-squared reported: 0.037; 0.072; 0.074; 0.076; 0.069; 0.077; 0.147; 0.158; 0.157; 0.288; 0.270; 0.282
  - Firm/country/year fixed effects: Yes in all columns
  - Industry x Year effects: Yes in all columns
  - Other macro control variables: No in some columns, Yes in others (as indicated in source)
- Notes on specification and inference
  - Dependent variables: (log) investment to value added in cols. (1)-(3); (log) share of labor in value added in cols. (4)-(6); labor productivity growth in cols. (7)-(9); total factor productivity growth in cols. (10)-(12).
  - Lagged values of markup and firm-level characteristics used in cols. (7)-(12) to mitigate potential reverse causality concerns.
  - Clustered standard errors at the firm level reported in parentheses.
  - ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.

### Manufacturing-only sample (Orbis database) — investment, labor share, labor productivity growth, TFP growth
- Markup
  - Investment cols. (1)-(2): -2.353*** (0.19), -2.420*** (0.21)
  - Labor share cols. (3)-(4): -1.017*** (0.13), -1.196*** (0.18)
  - Labor productivity growth cols. (5)-(6): -1.381*** (0.19), -1.556*** (0.23)
  - TFP growth cols. (7)-(8): -0.648*** (0.03), -0.614*** (0.03)
- Firm size
  - Investment cols. (1)-(2): 0.170* (0.07), 0.174* (0.07)
  - Labor share cols. (3)-(4): 0.0755 (0.05), 0.0338 (0.06)
  - Labor productivity growth cols. (5)-(6): 0.371*** (0.07), 0.349*** (0.07)
  - TFP growth cols. (7)-(8): 0.00938* (0.00), 0.00764 (0.00)
- Age
  - Investment cols. (1)-(2): -0.868 (0.57), -1.148 (1.55)
  - Labor share cols. (3)-(4): -0.417 (0.44), -0.726 (0.72)
  - Labor productivity growth cols. (5)-(6): -0.0276 (0.11), -0.390 (1.02)
  - TFP growth cols. (7)-(8): 0.038 (0.03), -0.0737 (0.17)
- Log Real GDP per capita (selected reported)
  - Investment cols. (1)-(2): 58.93 (40.12), 94.06 (116.00)
  - Labor share cols. (3)-(4): -25.21 (32.17), 12.43 (9.28)
  - Labor productivity growth cols. (5)-(6): 1.362 (7.23), 0.68 (66.40)
  - TFP growth cols. (7)-(8): -3.298 (1.68), 3.855 (10.59)
- Domestic ownership
  - Investment cols. (1)-(2): -0.203 (0.75), -0.199 (0.74)
  - Labor share cols. (3)-(4): 0.584 (0.81), 0.569 (0.82)
  - Labor productivity growth cols. (5)-(6): 0.536*** (0.06), 0.530*** (0.06)
  - TFP growth cols. (7)-(8): 0.0927** (0.03), 0.0893** (0.03)
- Other controls (selected)
  - Real GDP growth: -0.468 (0.89), 0.738 (0.75), -0.006 (0.17), 0.0207 (0.01)
  - Trade openness: -0.688 (0.95), 0.0516 (0.10), -0.261 (0.76), -0.0425 (0.05)
  - Institutional quality: 19.742 (16.29), 9.52 (30.92), -1.340 (3.70), 0.0203 (0.45)
  - Lagged labor productivity: -0.044 (0.04), -0.0253 (0.05); additional reported -0.0594** (0.05)
  - Lagged TFP: -0.0229, -0.0249 (standard errors not fully aligned in source)
- Sample size, fit, and controls
  - Observations reported across columns: 6253; 5389; 8460; 6821; 3582; 2970; 24719; 23036 (note: formatting in source mixes spacing)
  - R-squared reported: 0.111; 0.118; 0.17; 0.174; 0.122; 0.122; 0.336; 0.359
  - Firm/country/year fixed effects: Yes in all columns
  - Industry x Year effects: Yes in all columns
- Notes on specification and inference
  - Dependent variables: (log) investment to value added in cols. (1)-(2); (log) share of labor in value added in cols. (3)-(4); labor productivity growth in cols. (5)-(6); total factor productivity growth in cols. (7)-(8).
  - Sample restricted to manufacturing firms in the Orbis database.
  - Lagged values of markup and firm-level characteristics used in cols. (5)-(8) to mitigate potential reverse causality concerns.
  - Clustered standard errors at the firm level reported in parentheses.
  - ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.

### Interpretation and substantive implications (as reported)
- Higher markups are consistently associated with:
  - Lower investment (large negative and statistically significant coefficients, especially in the manufacturing sample: e.g., -2.353*** (0.19) and -2.420*** (0.21)).
  - Lower labor share (negative and statistically significant coefficients across samples).
  - Lower labor productivity growth and lower total factor productivity growth (negative and statistically significant in several specifications).
- Firm size and domestic ownership have heterogeneous associations:
  - Firm size tends to be positively associated with investment and productivity in several specifications (e.g., firm size 0.275*** (0.07) in broader sample; 0.371*** (0.07) for labor productivity growth in manufacturing).
  - Domestic ownership is positively associated with labor productivity growth and TFP growth in the manufacturing sample (e.g., 0.536*** (0.06), 0.530*** (0.06)).
- Robustness and controls
  - Results control for firm, country and year fixed effects, and industry x year effects.
  - Lagged firm-level variables used to mitigate reverse causality.

*Source: IMF staff estimates.*

### ANNEX 3.4. METHODOLOGICAL BACKGROUND FOR THE SECTION MACROECONOMIC

### ANNEX 3.4. METHODOLOGICAL BACKGROUND FOR THE SECTION MACROECONOMIC EFFECTS OF DOMESTIC ARREARS ACCUMULATION

### Methodological overview
- The Annex describes two empirical approaches used to assess the macroeconomic effects of domestic arrears accumulation: (i) panel regressions and (ii) synthetic control method estimator.
- Panel regressions are estimated in the form of (Eq. 1) where the outcome ܻ௜,௧ is measured for country ݅ in year ݐ; arrears accumulation ݏݎܽ݁ݎݎܣ
௜,௧ିଵ is expressed in percent of GDP (lagged by one year); ܺ௜,௧ିଵ is a vector of covariates; ܦ
௜ and ܦ
௧ are country and year fixed effects. Standard errors are clustered at the country level.  
- For firm-level regressions an augmented specification (Eq. 2) is used that includes a conditioning variable ܼ௜௞,௧ and its interaction with arrears accumulation; country-sector fixed effects are included and standard errors are clustered at the country-sector level.
- RHS variables are lagged by one year in macro regressions; for firms’ performance regressions RHS variables are generally not lagged (except the independent variable of interest) to protect sample size.

### Panel regression specifications and controls
- Real GDP per capita growth (Table A.3.4.1)
  - Baseline controls: level of development, investment % GDP, gross secondary school enrollment ratio, trade openness (log), terms of trade shocks, WGI institutional quality.
  - Robustness checks: credit to private sector % GDP, CPI inflation %, fiscal balance % GDP, public debt % GDP.
- Nonperforming loans-to-gross loans ratio (Table A.3.4.2)
  - Baseline controls: real GDP growth %, credit growth, REER index (log), terms of trade shocks.
  - Robustness checks: CPI inflation %, regulatory capital to risk-weighted assets, return on assets, foreign-currency-denominated liabilities % total liabilities.
- Tax revenue-to-GDP ratio (Table A.3.4.3)
  - Baseline controls: level of development, agricultural value-added % GDP, trade openness (log), terms of trade shocks, CPI inflation %, net ODA % GDP, external debt % GDP.
  - Robustness checks: WGI control of corruption, WGI political stability & no violence, natural resource rents % GDP, current account balance % GDP.
- Access to electricity (Table A.3.4.4)
  - Baseline controls: level of development, urban population % total population, population density per km2, population growth %, WGI control of corruption, fiscal balance % GDP.
  - Robustness checks: age dependency ratio % working-age population, unemployment %, tax revenue % GDP, grants % GDP.
- Corporate sector performance (Table A.3.4.5)
  - Outcomes: sales growth, productivity growth, employment growth.
  - Firm-level controls: log firm size, log firm age, foreign ownership dummy, exporter dummy, ISO certified dummy, log GDP per capita.
  - Interaction terms: participation in public procurement markets (Gov. contract dummy), access to a bank loan or line of credit (Bank loan dummy), requirement to meet with tax officials (Tax visit dummy), prepaid sales share (Prepaid sales % total sales).

### Key empirical findings (aggregate and sectoral effects)
- Summary causal association estimates (preferred parsimonious specifications, column (1) in tables):
  - Real GDP per capita growth: A 1 percentage point increase in arrears accumulation is associated with a -0.3 percentage point fall in real GDP per capita growth. (Table A.3.4.1, Col. 1: Arrears accumulation, % GDP = -0.003**)
  - Banks’ asset quality (NPLs-to-total gross loans ratio): A 1 percentage point increase in arrears accumulation is associated with a 0.3 percentage point rise in the NPL ratio. (Table A.3.4.2, Col. 1: Arrears accumulation, % GDP = 0.323*)
  - Tax revenue mobilization: A 1 percentage point increase in arrears accumulation is associated with a -0.213 percentage point change in tax revenue as a share of GDP. (Table A.3.4.3, Col. 1: Arrears accumulation, % GDP = -0.213*)
  - Access to electricity: A 1 percentage point increase in arrears accumulation is associated with a -0.589 percentage point change in the percentage of the population with access to electricity. (Table A.3.4.4, Col. 1: Arrears accumulation, % GDP = -0.589**)
  - Private sector performance (firm-level outcomes):
    - Sales growth: arrears accumulation coefficient = -0.047*** (Table A.3.4.5, Col. 1)
    - Productivity growth: arrears accumulation coefficient = -0.054*** (Table A.3.4.5, Col. 2)
    - Employment growth: arrears accumulation coefficient = -0.025** (Table A.3.4.5, Col. 3)
- Additional regression coefficients and model fit indicators (selected):
  - Real GDP per capita growth regressions (Annex Table A.3.4.1): Log GDP per capita coefficient ranges: -0.256***, -0.270***, -0.297***, -0.276***, -0.255***; Observations across columns: 152, 145, 144, 143, 152; R-squared: 0.440, 0.433, 0.446, 0.438, 0.440.
  - Banks’ asset quality regressions (Annex Table A.3.4.2): Arrears accumulation coefficients: 0.323*, 0.360*, 0.357**, 0.341*, 0.286*; Observations: 216, 206, 204, 154, 136; R-squared: 0.627, 0.620, 0.640, 0.743, 0.740.
  - Tax revenue regressions (Annex Table A.3.4.3): Arrears accumulation coefficients: -0.213*, -0.204*, -0.206*, -0.216*, -0.213*; Observations: 238 in each column; R-squared: 0.896, 0.897, 0.896, 0.900, 0.897.
  - Access to electricity regressions (Annex Table A.3.4.4): Arrears accumulation coefficients: -0.589**, -0.543**, -0.583**, -0.587**, -0.584**; Observations: 237, 237, 237, 234, 237; R-squared: 0.960, 0.960, 0.960, 0.959, 0.960.
  - Private sector regressions (Annex Table A.3.4.5): Arrears accumulation coefficients across columns: -0.047*** (sales), -0.054*** (productivity), -0.025** (employment), -0.041** (sales with interaction), -0.022** (employment with interaction), -0.021** (employment with interaction), -0.019*** (employment in col. 7). Observations and R-squared vary by specification: Observations 9,739; 9,466; 11,717; 9,739; 11,518; 11,649; 4,280. R-squared: 0.197; 0.157; 0.104; 0.198; 0.107; 0.105; 0.103.

### Interaction effects and heterogeneous firm-level impacts
- Total marginal effects reported in Table A.3.4.6 (summary of interaction results):
  - Exposure to public procurement:
    - A 1 percentage-point increase in arrears accumulation is associated with a -6.9 percentage-point fall in annual sales growth for firms that participate in public procurement markets.
    - For firms that did not attempt to secure a government contract, the associated fall in annual sales growth is -4.1 percentage points. (Table A.3.4.5, Col. 4 and Table A.3.4.6)
  - Exposure to the financial sector:
    - Firms with a bank loan or line of credit witness a -3 percentage point drop in employment growth versus -2.2 percentage points for firms without such exposure. (Table A.3.4.5, Col. 5 and Table A.3.4.6)
  - Role of prepaid sales:
    - Arrears buildup reduces job creation by -1.9 percentage point for firms with zero percent of prepaid sales in total sales; the negative effect shrinks as the share of prepaid sales increases and ultimately turns non-significant. (Table A.3.4.5, Col. 7 and Table A.3.4.6)
  - Regulatory burden:
    - Firms facing burdensome regulatory procedures (tax visits, red tape) experience a larger deceleration in employment growth when arrears build up. (Table A.3.4.5, Col. 6 and Table A.3.4.6)

### Robustness, identification, and estimation notes
- Multiple robustness checks augment the baseline specifications with additional macroeconomic indicators to assess sensitivity of the arrears coefficient.
- For firm-level regressions, country-sector fixed effects are included and standard errors clustered at the country-sector level.
- Use of lagged RHS variables and fixed effects aims at mitigating endogeneity concerns; the overfitting problem precluded the use of GMM estimators (Roodman, 2009) in this context.
- Dependent variables and estimation details in tables:
  - Real GDP per capita growth: dependent variable is the growth of real GDP per capita. Robust standard errors in parentheses, clustered at the country level. RHS variables lagged by one year. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent level respectively.
  - NPL share: dependent variable is percent share of nonperforming loans to total gross loans, winsorized to set 1st and 99th percentiles at the cutoff levels. Robust standard errors in parentheses, clustered at the country level. RHS variables lagged by one year. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent level respectively.
  - Tax revenue: dependent variable is tax revenue as a share of GDP. Robust standard errors in parentheses, clustered at the country level. RHS variables lagged by one year. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent level respectively.
  - Access to electricity: dependent variable is the percentage share of the population with access to electricity. Robust standard errors in parentheses, clustered at the country level. RHS variables lagged by one year. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent level respectively.
  - Firm outcomes: dependent variables are sales/productivity/employment growth. Robust standard errors in parentheses, clustered at the country-sector level. Arrears accumulation is lagged by one year. ***, **, *, and + indicate statistical significance at the 1, 5, 10, and 15 percent level respectively.

_Italic: ANNEX 3.4. METHODOLOGICAL BACKGROUND FOR THE SECTION MACROECONOMIC EFFECTS OF DOMESTIC ARREARS ACCUMULATION (backgroundpapers - ANNEX 3.4. METHODOLOGICAL BACKGROUND FOR THE SECTION MACROECONOMIC)_

### Annex Table A.3.4.6. Sub-Saharan Africa: Total Marginal Effect of Arrears Accumulation

### Annex Table A.3.4.6. Sub-Saharan Africa: Total Marginal Effect of Arrears Accumulation on Private Sector Sales and Employment Growth

### Total marginal effects (table results and notes)
- Dependent variable is sales or employment growth. Coefficients are total marginal effects computed using parameter estimates from columns (4)-(7) of Table A.3.4.5. Robust standard errors in parentheses, clustered at the country-sector level. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent level respectively.
- Government contract dummy (sales growth)
  - Yes: -0.069*** (0.021)
  - No: -0.041* (0.017)
- Bank credit dummy (employment growth)
  - Yes: -0.030*** (0.011)
  - No: -0.022* (0.011)
- Tax visit requirement dummy (employment growth)
  - Yes: -0.026*** (0.010)
  - No: -0.021** (0.010)
- Prepaid sales, % total sales (employment growth)
  - 0%: -0.019*** (0.004)
  - 50%: -0.012*** (0.004)
  - 100%: -0.004 (0.006)

### Key interpretation points from table
- Presence of government contract attempts, access to bank credit, and tax visit requirements amplify the negative total marginal effect of arrears accumulation on firm-level outcomes (sales and employment), with coefficients that are statistically significant at conventional levels.
- Higher shares of prepaid sales mitigate the negative marginal effect of arrears accumulation on employment growth (coefficients move from -0.019*** at 0% prepaid sales to -0.004 at 100% prepaid sales).

*Italic: Note from source — Dependent variable and estimation details as stated above.*

### Synthetic Control Method (SCM): Purpose and implementation
- Research question: Do countries with sharp buildups of domestic arrears record lower levels of economic growth in subsequent years?
- Selected SSA cases with significant arrears increases after 2015 oil price shock:
  - Chad: +2.3 ppts of GDP
  - Angola: +5.7 ppts of GDP
  - Congo Rep.: +3.5 ppts of GDP
- Sharp accumulation defined as increase in stock of arrears above the 95th percentile of the sample distribution.
- SCM calibration and controls:
  - Synthetic control calibrated over 2011-15; results shown through 2018.
  - Pre-2015 controls: investment as share of GDP, human capital (gross secondary school enrollment), trade openness, terms of trade shocks, oil exporter status, institutional quality; ensure comparable pre-treatment log real GDP per capita.
  - Donor pool restricted to SSA, excluding countries with sharp arrears buildup (above 90th percentile) in 2014/15 or missing arrears data for these dates.
  - Congo Rep.: human capital excluded from pre-treatment characteristics due to lack of data.

### SCM baseline results (outcome magnitudes)
- One year after the buildup (post-2015), real GDP per capita relative to synthetic counterfactual:
  - Chad: 11 percent lower than counterfactual
  - Congo Rep.: 7 percent lower than counterfactual
  - Angola: 7 percent lower than counterfactual
- Actual outcomes remain below counterfactual after 2015 for all three countries; the gap tends to grow larger over time.

### Robustness checks and inference
- Broadening donor pool to the World:
  - Results that significant domestic arrears buildup can trigger a shift in real GDP per capita still hold.
- Placebo tests (iterative SCM applied to each control country):
  - Annex Figure A.3.4.10 plots treated-control gaps (solid red) and control placebo gaps (dotted grey).
  - Representation restricted to gaps satisfying:
    - (i) ratio of placebo RMSPE to the “true” RMSPE is lower than one, and
    - (ii) pre-2015 actual vs. counterfactual log real GDP per capita difference is lower than 5 percent in absolute value.
  - Findings: The negative effect of the 2015 arrears accumulation on log real GDP per capita is by far the largest for the three SSA countries, suggesting robust baseline effects.
- Caveat: SCM aims to maximize pre-treatment fit but does not guarantee perfect similarity across all pre-2015 characteristics (e.g., oil exporter dummy).

### DSGE model with domestic arrears financing: Model structure and transmission
- Model features four agents: government, households, entrepreneurs, banks.
- Government finances spending q_t through lump-sum taxes T_t and arrears accumulation AR_t; government clears arrears after n periods without compensating for inflation. Arrears defined and follow AR(1).
- Entrepreneurs:
  - Production uses capital and labor.
  - Borrowing constraint: cannot borrow more than discounted value of expected capital net of government domestic arrears (constraint expressed with capital, price of capital, discount factor, arrears coefficient).
  - Accumulated government arrears reduce entrepreneurs’ available resources, tighten collateral valuation, and raise haircuts on collateral, thereby tightening access to bank credit.
- Banks:
  - Combine bank capital and household deposits to issue loans.
  - Face quadratic cost when capital-to-loans ratio deviates from regulatory capital requirement.
- Standard new Keynesian features: capital producers, monopolistic retailers, Central Bank for price stability.

### DSGE calibration and simulation results (SSA application)
- Calibration inputs (exact values used):
  - Government spending: 25 percent of GDP
  - Domestic arrears: 20 percent of government spending
- Policy experiment simulated:
  - Unanticipated 1 percent of GDP increase in government expenditure
- Main simulation result:
  - Accumulation of domestic arrears lowers the fiscal multiplier: the longer arrears remain outstanding, the smaller the fiscal multiplier.
  - Mechanism: arrears reduce private sector resources for production and investment and tighten borrowing constraints, diminishing the expansionary effect of government spending on aggregate demand.
  - The impact of the accumulation period on the size of the fiscal multiplier declines over time.

_Italic: Source — backgroundpapers - Annex Table A.3.4.6. Sub-Saharan Africa: Total Marginal Effect of Arrears Accumulation (BACKGROUND PAPER: ONLINE ANNEX 3—DOMESTIC ARREARS IN SUB-SAHARAN AFRICA)._

### Annex Figure A.3.5.1. Fiscal multiplier in the presence of government domestic arrears

### Annex Figure A.3.5.1. Fiscal multiplier in the presence of government domestic arrears

### Main findings on fiscal multipliers and domestic arrears
- The combination of two channels reduces significantly the fiscal multiplier as financial frictions amplify the impact of domestic arrears on the fiscal multiplier.
- If the private sector faces liquidity constraints and the banking sector is not able to accommodate the extra demand for credit due to domestic arrears accumulation, either for liquidity or solvency reasons, the private sector will have to significantly adjust downward their level of production and investment, which reduces the expansionary effect of government spending on aggregate demand.
- The fiscal multiplier could be substantially lower and even negative if government expenditures are mostly financed through arrears accumulation for a long period and the private sector is under liquidity constraints.
- In that case, the negative impact of arrears on the private sector will offset the expansionary effect of government expenditures.
- These results are consistent with the literature that stipulates that fiscal multipliers are dependent on the state of the economy, the presence of financial frictions and the efficiency of public spending.

### Quantitative scenarios and graphical annotations (as presented)
- Impact on GDP of an unanticipated 1 percent of GDP increase in government expenditure.
- Axis tick values shown in figures: 0, 0.1, 0.2, 0.3, 0.4 (Percent of GDP).
- Quarters labeling shown in figures: 161116 (Quarters).
- Scenario labels depicted across figures:
  - Absence of domestic arrears
  - Domestic arrears (20 percent of government spending) accumulated for one quarter.
  - Domestic arrears (20 percent of government spending) accumulated for four quarters.
  - Domestic arrears (50 percent of government spending) accumulated for four quarters.
  - Domestic arrears (70 percent of government spending) accumulated for four quarters.
- In the presence of large shares of government spending financed through domestic arrears accumulation and a liquidity constrained private sector, the fiscal multiplier series include negative values down to -0.2 (Percent of GDP) as illustrated.

### Interpretive implications for policy
- Accumulating domestic arrears weakens the stimulative effect of government spending, especially when:
  - The private sector is liquidity constrained.
  - The banking sector cannot supply additional credit due to liquidity or solvency limits.
- Prolonged financing of government expenditures through arrears can not only neutralize but reverse expansionary effects (fiscal multiplier can be negative).
- Policy emphasis should consider the interaction between public payment practices (arrears accumulation), private sector liquidity conditions, and banking sector capacity when assessing fiscal stimulus effectiveness.

*BACKGROUND PAPER: ONLINE ANNEX 3—DOMESTIC ARREARS IN SUB-SAHARAN AFRICA*

### References

### backgroundpapers - References

### Scope and data and projection vintage
- Data and projections presented in this Regional Economic Outlook are IMF staff estimates as of September 30, 2019, consistent with the projections underlying the October 2019 World Economic Outlook.
- The data and projections cover 45 sub-Saharan African countries in the IMF’s African Department.
- Data definitions follow established international statistical methodologies to the extent possible; in some cases, data limitations limit comparability across countries.

### Country groupings (membership rules and thresholds)
- Countries are aggregated into three (non-overlapping) groups: oil exporters, other resource-intensive countries, and non-resource-intensive countries.
  - Oil exporters: countries where net oil exports make up 30 percent or more of total exports.
  - Other resource-intensive countries: nonrenewable natural resources represent 25 percent or more of total exports.
  - Non-resource-intensive countries: those not classified as either oil exporters or other resource-intensive countries.
- Countries are also aggregated into four (overlapping) groups: oil exporters, middle-income, low-income, and countries in fragile situations.
  - Middle-income countries: per capita gross national income in the years 2016–18 of more than US$1,025.00 (World Bank, Atlas method).
  - Low-income countries: average per capita gross national income in the years 2016–18 equal to or lower than US$1,025.00 (World Bank, Atlas method).
  - Countries in fragile situations: average CPIA scores of 3.2 or less in the years 2016–18 and/or had the presence of a peace-keeping or peace-building mission within the last three years.

### Regional cooperation bodies and aggregation coverage
- Membership of sub-Saharan African countries in major regional cooperation bodies (listed on page 60) includes:
  - CFA franc zone (WAEMU and CEMAC), COMESA, EAC-5, ECOWAS, SADC, and SACU.
  - Note: EAC-5 aggregates include data for Rwanda and Burundi, which joined the group only in 2007.

### Methods of aggregation (tables and weighting)
- In Tables SA1–SA3, SA6–SA7, SA13, SA15–SA16, and SA22–SA23:
  - Country group composites are calculated as the arithmetic average of data for individual countries, weighted by GDP valued at purchasing power parity as a share of total group GDP.
  - Source of purchasing power parity weights: World Economic Outlook (WEO) database.
- In Tables SA8–SA12, SA17–SA21, and SA24–SA26:
  - Country group composites are arithmetic averages weighted by GDP in U.S. dollars at market exchange rates as a share of total group GDP.
- In Tables SA4–SA5 and SA14:
  - Country group composites are geometric averages of data for individual countries, weighted by GDP valued at purchasing power parity as a share of total group GDP. Source of PPP weights: WEO database.
- In Tables SA27–SA28:
  - Country group composites are calculated as the unweighted arithmetic average of data for individual countries.

### Statistical appendix: tables and primary series (inventory)
- List of Appendix Tables SA1—SA28 (topics only):
  - SA1. Real GDP Growth
  - SA2. Real Non-Oil GDP Growth
  - SA3. Real Per Capita GDP Growth
  - SA4. Consumer Prices, Average
  - SA5. Consumer Prices, End of Period
  - SA6. Total Investment
  - SA7. Gross National Savings
  - SA8. Overall Fiscal Balance, Including Grants
  - SA9. Overall Fiscal Balance, Excluding Grants
  - SA10. Government Revenue, Excluding Grants
  - SA11. Government Expenditure
  - SA12. Government Debt
  - SA13. Broad Money
  - SA14. Broad Money Growth
  - SA15. Claims on Nonfinancial Private Sector Growth (Percent change)
  - SA16. Claims on Nonfinancial Private Sector (Percent of GDP)
  - SA17. Exports of Goods and Services
  - SA18. Imports of Goods and Services
  - SA19. Trade Balance on Goods
  - SA20. External Current Account
  - SA21. Net Foreign Direct Investment
  - SA22. Real Effective Exchange Rates
  - SA23. Nominal Effective Exchange Rates
  - SA24. External Debt, Official Debt, Debtor Based
  - SA25. Terms of Trade on Goods
  - SA26. Reserves
  - SA27. Banking Penetration
  - SA28. Banking Sector: Loan-to-Deposit Ratio

### Notable data and footnote points (country-specific caveats)
- Zimbabwe:
  - In February 2019, Zimbabwe adopted a new local currency unit, the RTGS dollar, which has become the official unit of account.
  - Efforts are underway to revise and update all national accounts series to the new RTGS dollar.
  - Current data are based on IMF staff estimates of price and exchange rate developments in US (and RTGS) dollars. Staff estimates of US dollar values may differ from authorities’ estimates.
- Table SA26 (Reserves) sources: IMF, Common Surveillance database, and IMF, World Economic Outlook database, October 2019.
  - Notes indicating use of WAEMU or CEMAC aggregates for certain members for reserves data, and flags for fiscal-year data where applicable.
- Tables SA1–SA3, SA6–SA19, SA21, SA24–25 source: IMF, Common Surveillance database and IMF, World Economic Outlook database, October 2019.
  - Note: “...” denotes data not available.
  - Note: Fiscal year data noted with footnote where applicable.

### Representative numeric precisions and examples preserved verbatim from the appendix
- IMF staff estimates date: September 30, 2019.
- Coverage: 45 sub-Saharan African countries.
- Oil-exporter classification threshold: 30 percent or more of total exports.
- Other resource-intensive classification threshold: 25 percent or more of total exports.
- Income threshold for middle-income / low-income classification: US$1,025.00 (World Bank, Atlas method).
- Fragility CPIA cutoff: 3.2 (average over 2016–18).
- Table numbering: SA1–SA28 (themes listed above).

*Source: backgroundpapers - References (Regional Economic Outlook: Sub‑Saharan Africa; IMF staff estimates and statistical appendix, as provided).*

---


_Source: https://www.imf.org/-/media/files/publications/reo/afr/2019/october/english/backgroundpapers.pdf_