## 2. Capital Flows to Sub-Saharan Africa: Causes and Consequences (Annex 2.1) ## Source details **Canonical URL:** [2. Capital Flows to Sub-Saharan Africa: Causes and Consequences (Annex 2.1)](https://www.imf.org/-/media/files/publications/reo/afr/2018/october/pdf/sreo1018background.pdf) ## Other formats - [Markdown version](/-/media/files/publications/reo/afr/2018/october/pdf/sreo1018background.pdf.md) - [Structured JSON version](/-/media/files/publications/reo/afr/2018/october/pdf/sreo1018background.pdf.json) --- ### Data sources, sample, and country coverage - Primary data sources: - IMF’s World Economic Outlook (WEO) and International Financial Statistics (IFS) databases. - Monthly analysis: Emerging Portfolio Fund Research (EPFR), Haver Analytics, Bloomberg. - Sample: - The sample for sub-Saharan Africa comprises 45 countries listed in Annex Table 2.1.2. - Country composition varies across analyses due to data limitations. - Country groups used include: middle-income, low-income, oil exporters, resource-intensive, non-resource-intensive, and comparator emerging market economies (EMEs). - Note: - Countries marked with an asterisk (*) in Annex Table 2.1.2 are SSA countries included in the analysis using monthly data. ### Variables and data description - Capital flows and related variables (units and sources preserved): - Asset flows — In USD billion. Excludes reserve assets, and official other investment asset flows — IMF, WEO database - Bank flows — In USD billion — BIS Locational Statistics - Bond fund flows — In USD billion — EPFR Global - Equity fund flows — In USD billion — EPFR Global - Errors and omissions — In USD billion — IMF, WEO database - FDI liability flows — In USD billion — IMF, WEO database - Inward direct investment — In USD billion — IMF, Coordinated Direct Investment survey database - Inward portfolio investment — In USD billion — IMF, Coordinated Portfolio Investment survey database - Liability flows — In USD billion. Excludes official other investment liability flows — IMF, WEO database - Net capital flows — In USD billion. Excludes reserve assets, and official other investment asset and liability flows — IMF, WEO database - Other investment liability flows — In USD billion. Excludes official flows — IMF, WEO database - Portfolio liability flows — In USD billion — IMF, WEO database - Broad money — In billions of national currency — IMF, IFS database - Capital account openness index — Index (high values: more open) — Chinn and Ito (2006) - Commodity prices — Index — IMF, WEO database - Compensation of employees — In percent of GDP — IMF, WEO database - Consumer price index (CPI) — Index — IMF, WEO and INS databases - Current account balance — In percent of GDP — IMF, WEO database - Domestic credit to private sector — In billions of national currency — IMF, IFS database - Exchange rate regime (de facto) — Index (1=hard or conventional peg; 2=basket peg/band/crawl/managed float; 3=free float) — Ghosh, Ostry, and Qureshi (2015) - External debt — In percent of GDP — IMF, WEO database - General government expense — In percent of GDP — IMF, WEO database - Inflation — In percent. — IMF, WEO database - Institutional quality index — Average of 12 political risk components normalized between 0 (low) and 1 (high) — International Country Risk Guide - Nominal GDP — In billions (USD and national currency) — IMF, WEO database - Official development assistance — In USD billion — World Bank, International Development Statistics database - Oil prices — Index — IMF, WEO database - Output gap — Difference between actual real GDP and trend (obtained from HP filter), in percent of trend real GDP — Staff calculations - Overvaluation — Difference between actual REER and trend (obtained from HP filter), in percent of trend REER — Staff calculations - Policy rate — Central bank policy rate or discount rate, in percent — IMF, IFS database - Private investment — In percent of GDP — IMF, WEO database - Public consumption — In percent of GDP — IMF, WEO database - Public debt — In percent of GDP — IMF, WEO database - Public investment — In percent of GDP — IMF, WEO database - Purchases/use of goods & services — In percent of GDP — IMF, WEO database - Real effective exchange rate (REER) — Index — IMF, WEO database - Real GDP — In billions of national currency — IMF, WEO database - Real GDP growth — In percent — IMF, WEO database, Penn World Tables 9.0 database - Real GDP growth in trading partners — In percent — IMF, WEO database - Real GDP per capita — In PPP terms — IMF, WEO database, Penn World Tables 9.0 database - Real GDP per capita growth — In percent — IMF, WEO database, Penn World Tables 9.0 database - Real U.S. govt. bond yield — (100+nominal govt. bond yield)/(100-future inflation) — Staff calculations - S&P 500 index returns volatility — Standard deviation of monthly returns of the S&P500 index, in percent — Bloomberg and staff calculations - Social benefits — In percent of GDP — IMF, WEO database - Sovereign bond issuances — In USD billion — Bloomberg LLP - Stock of external liabilities and assets — In USD billion — External Wealth of Nations Mark II database - Stock of reserves — In USD billion — IMF, WEO database - Terms of trade of goods — US Dollars — IMF, WEO database - Total investment — In percent of GDP — IMF, WEO database - Trade openness — Sum of exports and imports, in percent of GDP — IMF, WEO database - U.S. 10-year govt. bond yield — In percent. — IMF, IFS database (Bloomberg for monthly data) - VIX/VXO — In logs — Chicago Board Options Exchange ### Figures and notes (selected) - Annex Figure 2.1.2: Sub-Saharan Africa (excl. South Africa): Financial Flows, 1980-2017 (in USD billion and percent of GDP). Note: Statistics for 2017 are provisional. Negative values indicate outflows. Flows exclude reserve asset and official other investment flows. Net financial flows in percent of GDP is the sum of financial flows to the region in percent of regional GDP. - Annex Figure 2.1.1 and 2.1.3: Net Financial Flows, 1980-2017 (Percent of M2 and Percent of GDP). Note: No statistics for EMEs including China are reported for 1980-89 because of lack of data availability on Chinese reserve asset flows for that period. For 1990-99, China's data is available for three years 1997-99. - Annex Figure 2.1.6: Composition of Net Financial Flows, 1980-2017 (in USD billion). Note: Statistics for 2017 are provisional. Negative values indicate outflows. Flows exclude official other investment flows. The components do not necessarily add up to total net financial flows because of lack of data availability. - Annex Figure 2.1.5: Sub-Saharan Africa Frontier Markets: Sovereign Bond Issuance, 2007–18 (in USD billion). Note: Other countries are Cameroon, Ethiopia, Gabon, Namibia, Rwanda, Tanzania, and Zambia. Data is as of June 2018. - Annex Figures 2.1.9–2.1.10: Errors and Omissions, 1980-2017 (in percent of GDP). Note: SSA (EME) errors and omissions are in percent of SSA (EME) GDP. Statistics for 2017 are provisional. Mauritius is excluded from the sample. Negative values indicate outflows. - Annex Figure 2.1.11–2.1.12: Stock of External Liabilities and Assets, Short-term External Debt, 1980-2017 (In USD billion). ### Distribution and transition statistics (liability flows, 1980–2017) - Transition probabilities (values preserved exactly): - FDI - Positive → Positive: 96.4 - Positive → Negative: 3.6 - Negative → Positive: 98.7 - Negative → Negative: 1.4 - Portfolio - Positive → Positive: 84.7 - Positive → Negative: 15.3 - Negative → Positive: 81.5 - Negative → Negative: 18.6 - Other investment - Positive → Positive: 76.3 - Positive → Negative: 23.8 - Negative → Positive: 75.8 - Negative → Negative: 24.2 - Portfolio debt flows - Positive → Positive: 80.0 - Positive → Negative: 20.0 - Negative → Positive: 78.8 - Negative → Negative: 21.2 - Portfolio equity flows - Positive → Positive: 82.5 - Positive → Negative: 17.5 - Negative → Positive: 75.2 - Negative → Negative: 24.8 --- ### Drivers of financial flows: framework, global push and domestic pull factors ### Estimation framework - Period: 1980-2017. - Dependent variable K_jt: nonofficial capital flows (net financial flows, liability flows, asset flows) in percent of GDP. - Estimation method: ordinary least squares (OLS) with standard errors clustered at the country level. - All domestic variables are lagged one period to address potential endogeneity. - Model (equation A2.1.1): - K_jt = Σ_i β_i x_i,t + Σ_k γ_k z_k,jt + μ_j + ε_jt - x: global push factors; z: domestic pull factors; μ_j: time-invariant country-specific fixed effects; ε_jt: error term. ### Global push factors considered - US interest rate: 10-year nominal US government bond yield (results robust to real yield or other US yield proxies). - Global market uncertainty: volatility of S&P 500 index returns (used because VIX is available only from 1990; correlation with VIX is 0.53). - International commodity price index (log). ### Domestic pull factors considered - Lagged current account balance (percent of GDP) as a proxy for external financing need. - De jure financial account openness (Chinn and Ito, 2008). - Real GDP growth rate. - Trade openness. - Institutional quality. - Real GDP per capita (log). - External debt ratio (external debt/GDP). - De facto exchange rate regime (lower values = less flexible regimes). - Note: domestic policy rate / discount rate, expected real exchange rate depreciation, and stock of foreign exchange reserves were considered but found statistically insignificant in these specifications. --- ### Key estimation findings and quantitative magnitudes ### Net, liability, and asset flows (annual regressions) - Net financial flows: - A 100 basis point decline in the nominal US government bond yield, on average, implies an increase in net flows by about 0.2 to 0.4 of a percent of GDP. - Commodity prices: statistically significant when only global factors and country fixed effects are included; coefficient weakens when additional domestic factors are added. - Among domestic factors, higher real GDP growth, higher real GDP per capita, greater trade openness, and a larger external financing need are associated with more net inflows. - Liability flows: - Much of the effect of lower US interest rates on net flows operates through liability flows. - A 100 basis point decline in the US government bond yield, on average, increases liability flows by about 0.3 to 0.5 of a percent of GDP. - A 10 percent increase in the commodity price index implies an increase in liability flows by about 0.2 to 0.3 of a percent of GDP. - Asset flows: - Evidence that higher US interest rates and greater global market volatility lead to greater asset inflows, consistent with domestic residents retrenching from foreign assets as global financial conditions tighten. - By flow type (qualitative magnitudes): - US interest rates and commodity prices have a stronger effect on inward direct investment (FDI) than on other types of flows. - Global market volatility has a statistically stronger effect on foreign portfolio flows than on FDI. ### Quantitative examples by flow type - FDI: - A 100-basis point reduction in the US government bond yield implies an increase of about 0.2 of a percent of GDP in FDI flows to sub-Saharan Africa. - Portfolio flows: - A 100-basis point reduction in the US government bond yield implies an increase of about 0.03 of a percent of GDP in foreign portfolio flows. - A one standard deviation shock to the global market volatility index reduces portfolio flows by about 0.1 of a percent of GDP; no statistically significant effect on FDI from the same shock. ### Model fit and sample statistics (selected) - Observations (examples across specifications): 1,531; 1,456; 1,343; 811. - Countries (examples across specifications): 44; 43; 29; 44. - Adjusted R-squared (examples across specifications): 0.230; 0.276; 0.292; 0.281. - Significance notation: ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. --- ### Surge and reversal episodes: definitions and probit results ### Surge definition - An observation is a large inflow episode ("surge") if it is above the top 30th percentile of the country’s own distribution of liability flows (percent of GDP) and in the top 30th percentile of the entire sample’s distribution of liability flows (percent of GDP). - Formally, S_jt = 1 if K_jt ∈ {top 30th percentile of country j’s distribution} ∩ {top 30th percentile of sample distribution}; S_jt = 0 otherwise. - Rationale: dual criterion ensures surges are large relative both to the country’s historical experience and to cross-country standards. ### Probit results — likelihood of surge and reversal (1980–2017) - Surges: - Global drivers: US interest rates and commodity prices are statistically significant determinants of surge occurrence. - Magnitude example: "Against an unconditional surge probability of about 23 percent in the estimated sample, a 100 basis point decline in the US government bond yield is associated with a 2 percentage point higher likelihood of a surge (evaluated at mean values of other regressors)." - Pseudo-R2 (global factors only): 0.17. - With pull factors added: pseudo-R2 increases to a range of 0.21–0.26. - Probit sensitivity (proportion of surges correctly called): about 27 percent with global factors only; increases to 33–55 percent with pull factors. - Overall classification: "the probit calls 90 percent of the observations correctly" when pull factors included. - Reversals: - Global drivers: US interest rates show somewhat symmetric effects for reversals; commodity prices do not have symmetric effects (commodity price increases lower likelihood of a reversal in resource-intensive countries only). - Domestic factors that lower reversal likelihood: higher real GDP growth, greater exchange rate flexibility, better institutional quality. - Probit performance: - Pseudo-R2 with pull factors: around 0.16–0.34. - Sensitivity (proportion of reversals correctly called): about 21–28 percent. - Specificity (proportion of non-occurrences correctly classified) examples: 92.72, 93.58, 93.84, 93.93 (surge models) and 95.52, 95.81, 94.88, 95.11, 96.22 (reversal models). --- ### Monthly analysis: fund flows, stock returns, and bond yields (2000–2017) ### Data and approach - Monthly bond and equity fund flows, sovereign bond yields, and real stock returns estimated controlling for global and domestic variables, with country fixed effects and country-year or country-quarter dummies in variants. ### Selected magnitudes and representative coefficients - Fund flows: - A 100 basis point decline in the US interest rate is, on average, associated with about 0.3 of a percent of GDP higher bond fund flows, and about 0.1 of a percent of GDP higher equity fund flows. - A one standard deviation decrease in (log) VIX increases bond and equity fund flows by 0.2 and 0.1 of a percent of GDP, respectively. - Sovereign bond yields and stock returns: - Bond yields: "A 100 basis points increase in the US interest rate and a one standard deviation increase in the VIX (in log terms) leads to an increase in sovereign bond yields by about 111 and 18 basis points, respectively, while a one standard deviation increase in commodity prices lowers the yields by about 21 basis points." - Stock returns: "the effect of the VIX index is statistically the strongest, with a one standard deviation increase in the VIX index on average implying a drop in stock returns by about 1.7 percentage points." - Pre- versus post-global financial crisis: - Effects of VIX and US interest rate on fund flows are larger postcrisis. - Example: "a one standard deviation decrease in the VIX (in log terms) implies an increase in fund flows to sub-Saharan Africa by about 0.2 of a percent of GDP before 2009, but by about 0.5 of a percent of GDP after the crisis." ### Representative monthly coefficient examples (preserved) - US interest rate on bond flows (full sample examples): -0.048***, -0.327***, -0.467***. - Commodity prices on bond+equity flows (full sample): 0.814***. - S&P500 index volatility examples: -0.998***, -1.401***, -2.028*** in panel specifications; large negative effects on stock returns and bond yields in several columns (e.g., -9.942***, -15.194*** in some bond yield-related regressions). - Postcrisis monthly sample (2009M1–17M12) examples: US interest rate on bond flows -0.030, -0.386***, -0.634***; Commodity prices showing 0.220***, 1.362***; S&P500 volatility coefficients -0.413***, -0.773***, -0.746***. --- ### Macroeconomic consequences: vulnerabilities ### Vulnerabilities analyzed and model - Vulnerabilities examined: currency overvaluation (deviation of the real effective exchange rate from its trend), output gap (deviation of real GDP from its trend), and rapid credit growth (change in the domestic credit-to-GDP ratio). - Estimated model (A2.1.6): - V_jt = ρ f_jt + ∑_k γ_k w_k,jt + δ_j + φ_t + ε_jt - V denotes vulnerability in country j at time t; f is financial flow (in percent of GDP); w are control variables (per capita income, real GDP growth, credit-to-GDP ratio, trade openness, etc.). Regressions include country and year-fixed effects. ### Key regression coefficients and findings (Annex Table 2.1.11, preserved) - Net financial flows/GDP coefficients: -0.005; -0.025; -0.000. Standard errors: (0.042); (0.019); (0.011). - Liability flows/GDP coefficients: 0.011; -0.018; -0.000. Standard errors: (0.043); (0.022); (0.011). - Asset flows/GDP coefficients: -0.016; 0.020; -0.010; -0.016; -0.021; -0.013. Standard errors: (0.048); (0.054); (0.026); (0.037); (0.018); (0.028). - FDI liability flows/GDP coefficients: 0.004; -0.016; -0.009. Standard errors: (0.112); (0.056); (0.022). - Portfolio liability flows/GDP coefficients: 0.276**; 0.157***; 0.099**. Standard errors: (0.124); (0.031); (0.045). - Other investment liability flows/GDP coefficients: 0.027; -0.029; 0.006. Standard errors: (0.047); (0.032); (0.015). - Trade openness coefficients (examples): -0.018; -0.020; -0.017; 0.053***; 0.052***; 0.055***; -0.001; -0.002; 0.001. Standard errors provided in table. - Real GDP per capita (log) coefficients (examples): 0.913; 0.903; 0.734; 3.681***; 3.662***; 3.766***; 0.216; 0.152; 0.096. - Initial credit to GDP coefficients: -0.044**; -0.047**; -0.056***. ### Text summary of vulnerabilities - Aggregate result: Neither net, nor aggregate liability or asset flows are on average statistically associated with currency overvaluation, output gap, and credit growth for sub-Saharan African countries. - Disaggregated result: Portfolio flows are associated with generating vulnerabilities. - Quantified effect: "a 10 percent of GDP increase in portfolio inflows is, on average, associated with about a 3 percentage point deviation of the real exchange rate and output from trend and a 1 percent of GDP increase in domestic credit to the private sector." - Sample-split note: Splitting the sample into middle and low-income countries suggests much of the effects of portfolio flows are in the middle-income country sub-sample. --- ### Domestic investment: impact of capital flows ### Investment model and estimation - Investment_it (total, private, public investment in percent of GDP) regressed on various flow measures (total net capital flows, asset and liability flows, FDI, portfolio, other investment flows, all in percent of GDP), country-fixed effects, year effects, and controls (log real GDP per capita, real GDP growth, trade openness, exchange rate flexibility measure, external and total public debt). - All regressors lagged; standard errors clustered at country level. - Robustness: results robust to 5-year averages and two-stage least squares instrumental variable strategy. ### Key econometric findings (Annex Tables 2.1.12–2.1.14, preserved) - Aggregate finding: - Net and liability flows have a statistically strong positive association with domestic investment. - Interpretation: "a 1 percent of GDP increase in these flows is associated with about 0.2 of a percent of GDP increase in total investment." - FDI: - FDI liability flows/GDP coefficients: 0.553***; 0.457***; 0.474***. Standard errors: (0.088); (0.100); (0.111). - Text interpretation: "a one percent of GDP increase in FDI being associated with a 0.5 of a percent of GDP increase in domestic investment." - Private investment (Annex Table 2.1.13): - Net flows/GDP coefficients: 0.218***; 0.212***; 0.222***; 0.234***. - FDI liability flows/GDP coefficients: 0.485***; 0.509***; 0.548***. - Public investment (Annex Table 2.1.14): - No financial flows of any type are found to be statistically significantly associated with public investment. - Net flows/GDP coefficients: 0.017; -0.002; 0.002; 0.003. - Portfolio liability flows/GDP coefficients: -0.048; -0.024; -0.028. - Portfolio flows: - Little evidence that portfolio flows, on average, affect domestic investment; some association with private investment is weaker and less consistent. --- ### Public expenditure: impact of portfolio flows (Annex Table 2.1.16) ### Main regression findings (public consumption to GDP) - Net flows/GDP examples: - Column (1): 0.051 (0.049) - Column (2): 0.041 (0.036) - Column (3): 0.037 (0.036) - Column (4): 0.036 (0.036) - Portfolio liability flows/GDP: - Column (1): 0.211** (0.103) - Column (2): 0.241** (0.101) - Column (3): 0.222** (0.094) - Asset flows/GDP and liability flows/GDP reported across columns with standard errors preserved. - Trade openness/GDP often positive and significant (examples: 0.070** (0.026); 0.072*** (0.026); 0.074** (0.028); 0.081*** (0.029); 0.091*** (0.030)). - Estimation details: - Country-fixed effects: Yes (all specifications). - Year effects: Yes (all specifications). - Observations by column: 1,424; 1,365; 1,324; 1,287; 1,404; 1,346; 1,304; 1,267; 1,263; 1,216; 1,142. - R-squared by column: 0.052; 0.102; 0.112; 0.119; 0.051; 0.095; 0.106; 0.111; 0.068; 0.125; 0.145. - Statistical notes: - Standard errors in parentheses clustered by country. - Extreme observations (top and bottom percentile) excluded. - All regressors are lagged to mitigate endogeneity. - ***,**, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. ### Disaggregated government expenditure regressions (selected findings) - Portfolio liability flows/GDP positively and often significantly associated with current government expenditures components: - Examples with significance: - 0.249** (0.103) - 0.275** (0.114) - 0.219** (0.095) - 0.079*** (0.025) - 0.082*** (0.026) - 0.085*** (0.029) - 0.110*** (0.036) - 0.117** (0.047) - 0.127** (0.051) - 0.049** (0.018) - 0.046** (0.018) - 0.039** (0.019) - Other investment liability flows/GDP often negative or insignificant for certain current expenditure components (examples include -0.083* -0.069 -0.046). - Text interpretation: "Disaggregating government expenditure into compensation of public employees (as well as the sum of compensation of public employees and purchases of goods/services) and social benefits shows that these current expenditures are statistically strongly (and positively) associated with portfolio inflows (Annex Table 2.1.16)." - Robustness: - Results robust to using alternate variables for government current and capital expenditure (fiscal accounts) instead of national accounts public consumption and investment. - Splitting the sample pre/post global financial crisis: unconditional correlations show portfolio flows positively associated with public expenditure but not with public investment in both periods. - Resource intensity: correlation between public investment and portfolio flows is positive for nonresource intensive countries. --- ### The Future of Work in Sub-Saharan Africa (Section 3: modeling and scenarios) ### Modeling approach - Framework extended from Berg, Buffie and Zanna (2018) to include two regions: sub-Saharan Africa (SSA) and the rest of the world (ROW), with trade/financial linkages. - Time: discrete time, no uncertainty. Each region populated by a single infinitely-lived household that owns labor (L), capital (K) and robots (Z), and firms; households can borrow/save vis-à-vis the other region. - Three assets with returns: rKt (capital), rZt (robots), rBt (financial). No-arbitrage implies equalized net returns across countries in equilibrium. - Accumulation rules: - Kt+1 = IKt + (1–δK)Kt - Zt+1 = IZt + (1–δZ)Zt - δK and δZ are depreciation rates of capital and robot stock. - Production: - Labor and robots combined with a CES technology to produce “tasks”; tasks combined with capital via Cobb-Douglas to produce final output. - Elasticity of substitution between labor and robots denoted σ; CES allows modeling complements or substitutes. - Technological change modeled as an increase in robot productivity bt. Total factor productivity Ai is the only parameter varying across regions. - Calibration: - For each σ, calibrate SSA total factor productivity and the labor share in task production (e) to match relative GDP per capita between regions and the robot share in income in ROW. ### Key model findings and role of σ - General equilibrium: a doubling of robot productivity raises GDP per capita steady states in both regions; relative gains depend on σ. - If robots and labor are complements: - SSA can grow faster than ROW; increase in GDP per capita will be higher for SSA. - Labor share can increase, particularly in SSA. - Lower initial wages in SSA make it well-positioned to take advantage of higher robot productivity when robots complement workers. - If robots and labor are substitutes: - ROW benefits more; SSA may be hurt and can experience a fall in GDP per capita in the short run. - Labor share declines in both regions; decline is larger in ROW because workers are relatively more expensive and replaced by robots. - ROW has greater incentive to invest in robots when robots substitute labor because wages are high. - Three σ parameter cases used for dynamics: - σ=2 corresponds to substitutes. - σ=1 corresponds to a Cobb-Douglas production function. - σ=0.8 corresponds to complements. ### Financial flows and transition dynamics - When robots and labor are substitutes: - Higher robot demand in ROW can channel resources out of SSA. - SSA can run a current account surplus in transition, accumulating foreign assets. - Short-run fall in SSA GDP per capita can occur in addition to declines in wages and labor share. - In the final steady state, SSA consumption can exceed output financed by interest on accumulated assets. - When robots and labor are complements: - Resources may flow into SSA, further boosting SSA GDP per capita beyond the direct productivity effect. ### Three scenario narratives to 2040 - Africa Arisen: - Technical change improves productivity of low-skilled workers and creates new professional services jobs. - Global integration deepens after initial frictions. - Climate impacts offset by productivity and technological innovations; self-sufficiency in food. - Economic landscape: thriving urban innovation hubs; project-based gig economy; informality receding. - Skills: top-end creativity/STEM/entrepreneurship; low-end basic digital literacy and emotional intelligence. - Social policies: expanded social safety nets; portable health insurance and retirement schemes. - Winners: expanding middle class; large multinational tech companies. - Losers: rentiers (weakened commodity monopolies); people outside social safety net. - Policy challenges: facilitating high job mobility and continuous learning; allocating work and meaning beyond work; making megacities efficient; reducing informality while strengthening social safety nets. - Africa for Africa: - Technical change: replacement of workers across skill spectrum globally; technological adoption slower in SSA due to lower labor costs. - Global integration: regional integration spurred growth; large African market. - Climate: reduces growth potential. - Politics & state: strong push for regional integration; states drive growth with limited tax bases. - Economic landscape: regional transport infrastructure creates 1½ billion consumers regional market; informality remains a defining feature. - Skills: entrepreneurship, vocational skills, digital literacy for micro manufacturing and 3D printing. - Social policies: bare-bones safety nets; family/community networks important. - Winners: entrepreneurs serving regional demand; returning diaspora. - Losers: workers stuck in rural/old sectors. - Policy challenges: promoting vocational and entrepreneurial skills; balancing regional integration with limited fiscal resources. - Africa Adrift: - Technical change: localization and reshoring of manufacturing render earlier infrastructure investments obsolete. - Global integration: inward-looking policies, declining trade flows and development assistance. - Climate: hits harder and faster; regular food shortages and mass migration. - Politics & state: frequent economic crises, elites strengthen control, weak states, high debt. - Economic landscape: stagnant economies dependent on primary commodities; subsistence activities dominate; large-scale economic migration. - Skills: low-quality government education; elites use private schools. - Social policies: governments focused on control/security initially; later modest improvements from a low base. - Winners: migrants who moved to advanced economies; commodity-owning elites. - Losers: non-elite common workers; those outside elite networks. - Policy challenges: immediate priority on security and crime prevention; later, adjustment to technological upheaval and catching up. ### Policy implications (model and scenario-driven) - Enabling complementary investments when robots complement workers can help SSA capture larger gains. - Mitigating adverse distributional and transition dynamics when robots substitute for labor: - Need for policies to manage short-run capital outflows and labor-share declines, particularly where ROW demand for robots draws resources away from SSA. - Broader policy priorities across scenarios: - Facilitate high job mobility, continuous learning, and reskilling. - Strengthen digital and regional infrastructure. - Expand and adapt social safety nets (portable health insurance, retirement saving) suited to a more project-based/gig economy. - Invest in education focused on creativity, STEM, vocational skills and basic digital literacy. - Address urban megacity productivity and informality reduction. ### Caveats and limitations - Modeling technological change remains an open question; the nature and magnitude of new technologies are still nascent. - The model omits inequality within countries, unemployment, human capital dynamics, role of government, capital market frictions, endogenous innovation, and political economy factors. - Given uncertainty about technological change, results are qualitative and scenario-driven rather than precise quantitative forecasts. *Source: IMF, Regional Economic Outlook: Sub‑Saharan Africa — Annex 2.1 (Capital Flows to Sub‑Saharan Africa: Causes and Consequences).* ### 2. Capital Flows to Sub-Saharan Africa: Causes and Consequences Annex ...............................1 ### 2. Capital Flows to Sub-Saharan Africa: Causes and Consequences (Annex 2.1) ### Data sources and country coverage - Primary data sources: - IMF’s World Economic Outlook (WEO) and International Financial Statistics (IFS) databases. - Monthly analysis: Emerging Portfolio Fund Research (EPFR), Haver Analytics, Bloomberg. - Full set of data sources listed in Annex Table 2.1.1. - Sample: - The sample for sub-Saharan Africa comprises 45 countries listed in Annex Table 2.1.2. - Country composition varies across analyses due to data limitations. - Country groups used include: middle-income, low-income, oil exporters, resource-intensive, non-resource-intensive, and comparator emerging market economies (EMEs). - Note: Countries marked with an asterisk (*) in Annex Table 2.1.2 are SSA countries included in the analysis using monthly data. ### Data description and variables (as listed in Annex Table 2.1.1) - Capital flows variables (units and sources preserved): - Asset flows — In USD billion. Excludes reserve assets, and official other investment asset flows — IMF, WEO database - Bank flows — In USD billion — BIS Locational Statistics - Bond fund flows — In USD billion — EPFR Global - Equity fund flows — In USD billion — EPFR Global - Errors and omissions — In USD billion — IMF, WEO database - FDI liability flows — In USD billion — IMF, WEO database - Inward direct investment — In USD billion — IMF, Coordinated Direct Investment survey database - Inward portfolio investment — In USD billion — IMF, Coordinated Portfolio Investment survey database - Liability flows — In USD billion. Excludes official other investment liability flows — IMF, WEO database - Net capital flows — In USD billion. Excludes reserve assets, and official other investment asset and liability flows — IMF, WEO database - Other investment liability flows — In USD billion. Excludes official flows — IMF, WEO database - Portfolio liability flows — In USD billion — IMF, WEO database - Broad money — In billions of national currency — IMF, IFS database - Capital account openness index — Index (high values: more open) — Chinn and Ito (2006) - Commodity prices — Index — IMF, WEO database - Compensation of employees — In percent of GDP — IMF, WEO database - Consumer price index (CPI) — Index — IMF, WEO and INS databases - Current account balance — In percent of GDP — IMF, WEO database - Domestic credit to private sector — In billions of national currency — IMF, IFS database - Exchange rate regime (de facto) — Index (1=hard or conventional peg; 2=basket peg/band/crawl/managed float; 3=free float) — Ghosh, Ostry, and Qureshi (2015) - External debt — In percent of GDP — IMF, WEO database - General government expense — In percent of GDP — IMF, WEO database - Inflation — In percent. — IMF, WEO database - Institutional quality index — Average of 12 political risk components normalized between 0 (low) and 1 (high) — International Country Risk Guide - Nominal GDP — In billions (USD and national currency) — IMF, WEO database - Official development assistance — In USD billion — World Bank, International Development Statistics database - Oil prices — Index — IMF, WEO database - Output gap — Difference between actual real GDP and trend (obtained from HP filter), in percent of trend real GDP — Staff calculations - Overvaluation — Difference between actual REER and trend (obtained from HP filter), in percent of trend REER — Staff calculations - Policy rate — Central bank policy rate or discount rate, in percent — IMF, IFS database - Private investment — In percent of GDP — IMF, WEO database - Public consumption — In percent of GDP — IMF, WEO database - Public debt — In percent of GDP — IMF, WEO database - Public investment — In percent of GDP — IMF, WEO database - Purchases/use of goods & services — In percent of GDP — IMF, WEO database - Real effective exchange rate (REER) — Index — IMF, WEO database - Real GDP — In billions of national currency — IMF, WEO database - Real GDP growth — In percent — IMF, WEO database, Penn World Tables 9.0 database - Real GDP growth in trading partners — In percent — IMF, WEO database - Real GDP per capita — In PPP terms — IMF, WEO database, Penn World Tables 9.0 database - Real GDP per capita growth — In percent — IMF, WEO database, Penn World Tables 9.0 database - Real U.S. govt. bond yield — (100+nominal govt. bond yield)/(100-future inflation) — Staff calculations - S&P 500 index returns volatility — Standard deviation of monthly returns of the S&P500 index, in percent — Bloomberg and staff calculations - Social benefits — In percent of GDP — IMF, WEO database - Sovereign bond issuances — In USD billion — Bloomberg LLP - Stock of external liabilities and assets — In USD billion — External Wealth of Nations Mark II database - Stock of reserves — In USD billion — IMF, WEO database - Terms of trade of goods — US Dollars — IMF, WEO database - Total investment — In percent of GDP — IMF, WEO database - Trade openness — Sum of exports and imports, in percent of GDP — IMF, WEO database - U.S. 10-year govt. bond yield — In percent. — IMF, IFS database (Bloomberg for monthly data) - VIX/VXO — In logs — Chicago Board Options Exchange ### Country coverage groups (Annex Table 2.1.2 — country lists preserved) - Sub-Saharan Africa (SSA) — Angola*, Benin, Botswana, Burkina Faso, Burundi, Cabo Verde, Cameroon*, Central African Republic, Chad, Comoros, Democratic Republic of the Congo, Republic of the Congo, Côte d'Ivoire*, Equatorial Guinea, Eritrea, Ethiopia*, Gabon*, The Gambia, Ghana*, Guinea, Guinea-Bissau, Kenya*, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritius*, Mozambique*, Namibia*, Niger, Nigeria*, Rwanda*, Senegal*, Seychelles, Sierra Leone, South Africa*, South Sudan, Swaziland, São Tomé and Príncipe, Tanzania*, Togo, Uganda, Zambia*, Zimbabwe - Low-income — Benin, Burkina Faso, Burundi, Central African Republic, Chad, Comoros, Democratic Republic of the Congo, Eritrea, Ethiopia, The Gambia, Guinea, Guinea-Bissau, Liberia, Madagascar, Malawi, Mali, Mozambique, Niger, Rwanda, Sierra Leone, South Sudan, Tanzania, Togo, Uganda, Zimbabwe - Middle-income — Angola, Botswana, Cabo Verde, Cameroon, Republic of the Congo, Côte d'Ivoire, Equatorial Guinea, Gabon, Ghana, Kenya, Lesotho, Mauritius, Namibia, Nigeria, Seychelles, São Tomé and Príncipe, Senegal, South Africa, Swaziland, Zambia - Oil exporters — Angola, Cameroon, Chad, Republic of the Congo, Equatorial Guinea, Gabon, Nigeria, South Sudan - Other resource intensive — Botswana, Burkina Faso, Central African Republic, Democratic Republic of the Congo, Ghana, Guinea, Liberia, Mali, Namibia, Niger, Sierra Leone, South Africa, Tanzania, Zambia, Zimbabwe - Non-resource intensive — Benin, Burundi, Cabo Verde, Comoros, Côte d'Ivoire, Eritrea, Ethiopia, The Gambia, Guinea-Bissau, Kenya, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Rwanda, Senegal, Seychelles, Swaziland, São Tomé and Príncipe, Togo, Uganda - Emerging market economies (EMEs) comparator group — Argentina, Brazil, Bulgaria, Chile, China, Colombia, Hungary, India, Indonesia, Malaysia, Mexico, Peru, Philippines, Poland, Russia, Thailand, Turkey, Venezuela ### Additional stylized facts and figures (figure captions and notes preserved) - Annex Figure 2.1.2: Sub-Saharan Africa (excl. South Africa): Financial Flows, 1980-2017 (in USD billion and percent of GDP). Source: IMF, World Economic Outlook database. Note: Statistics for 2017 are provisional. Negative values indicate outflows. Flows exclude reserve asset and official other investment flows. Net financial flows in percent of GDP is the sum of financial flows to the region in percent of regional GDP. - Annex Figure 2.1.1: Sub-Saharan Africa and Emerging Markets: Net Financial Flows, 1980-2017 (Percent of M2). Source: IMF, WEO database, and IMF staff calculations. Note: Statistics represent sum of flows to the region in percent of total M2 for the region. No statistics for EMEs including China are reported for 1980-89 because of lack of data availability on Chinese reserve asset flows for that period. For 1990-99, China's data is available for three years 1997-99. - Annex Figure 2.1.3: Sub-Saharan Africa and Emerging Markets: Net Financial Flows, 1980-2017 (Percent of GDP). Source: IMF, WEO database, and IMF staff calculations. Note: Statistics represent sum of flows to the region (excluding South Africa) in percent of total GDP for the region (excluding South Africa). No statistics for EMEs including China are reported for 1980-89 because of lack of data availability on Chinese reserve asset flows for that period. For 1990-99, China's data is available for three years 1997-99. - Annex Figure 2.1.4: Sub-Saharan Africa: Net Official Development Assistance, 1980-2016 (in USD billion). Source: World Bank, World Development Indicators database. - Annex Figure 2.1.6: Sub-Saharan Africa: Composition of Net Financial Flows, 1980-2017 (in USD billion). Source: IMF, WEO database. Notes: Statistics for 2017 are provisional. Negative values indicate outflows. Flows exclude official other investment flows. The components do not necessarily add up to total net financial flows because of lack of data availability. - Annex Figure 2.1.5: Sub-Saharan Africa Frontier Markets: Sovereign Bond Issuance, 2007–18 (in USD billion). Source: Bloomberg LLP. Notes: Other countries are Cameroon, Ethiopia, Gabon, Namibia, Rwanda, Tanzania, and Zambia. Data is as of June 2018. - Annex Figure 2.1.7: SSA: Portfolio Debt and Equity Flows, 1980-2017 (in USD billion). Source: IMF, WEO database. Notes: Portfolio debt and equity flows do not add up to total portfolio liability flows for the region because of lack of data availability. Disaggregated data is also unavailable for South Africa. - Annex Figure 2.1.8: Sub-Saharan Africa (excl. Mauritius): Composition of Financial Flows, 1980-2017 (in USD billion). Source: IMF WEO database. Notes: Statistics for 2017 are provisional. Negative values indicate outflows. Flows exclude official other investment flows. The components do not necessarily add up to total liability and asset flows because of lack of data availability. - Annex Figure 2.1.9 and 2.1.10: Errors and Omissions, 1980-2017 (in percent of GDP). Source: IMF, WEO database. Notes: SSA (EME) errors and omissions are in percent of SSA (EME) GDP. Statistics for 2017 are provisional. Mauritius is excluded from the sample. Negative values indicate outflows. - Annex Figure 2.1.11: Sub-Saharan Africa: Stock of External Liabilities and Assets, 1980-2015 (In USD billion). Source: External Wealth of Nations Mark II database. - Annex Figure 2.1.12: Sub-Saharan Africa: Short-term External Debt, 1980-2017 (In USD billion). Source: IMF, WEO database. - Annex Figure 2.1.13: Top Three Source Countries for Inward Direct Investment, 2009-16 (In USD billion). Source: Coordinated Direct Investment Survey database. Note: Statistics are period averages, and represent the outstanding stock of inward direct investment in Sub-Saharan Africa. Right-hand panel excludes South Africa and Mauritius. - Annex Figure 2.1.14: Top Three Source Countries for Inward Portfolio Investment, 2009-16 (In USD billion). Source: IMF Coordinated Portfolio Investment Survey database. Note: Statistics are period averages, and represent the outstanding stock of portfolio investment in Sub-Saharan Africa based on the reported outstanding stock of portfolio assets by reporting countries. - Annex Figure 2.1.15: Top Three Source Countries for Cross-Border Bank Investment, 2001-16 (In USD billion). Source: Bank for International Settlements database and Cerutti and Zhou (2018). Note: Statistics are period averages, and represent the outstanding stock of cross-border banking claims in Sub-Saharan Africa based on the stock of banking claims by reporting countries. South Africa's outstanding stock of banking claims in South Africa are excluded. While the BIS does not release China’s bilateral exposure to individual counterparties, data limitations apply. ### Distribution and transition statistics (preserved numeric values) - Annex Figure 2.1.16: Sub-Saharan Africa: Distribution of Financial Flows, 1980-2017 — transition probabilities between inflows and outflows for different liability flow types (source: IMF staff estimates). Note: Outliers (defined as those in the top and bottom percentile of the full distribution) are excluded. - Transition probabilities table (values preserved exactly as presented): - FDI - Positive → Positive: 96.4 - Positive → Negative: 3.6 - Negative → Positive: 98.7 - Negative → Negative: 1.4 - Portfolio - Positive → Positive: 84.7 - Positive → Negative: 15.3 - Negative → Positive: 81.5 - Negative → Negative: 18.6 - Other investment - Positive → Positive: 76.3 - Positive → Negative: 23.8 - Negative → Positive: 75.8 - Negative → Negative: 24.2 - Portfolio debt flows - Positive → Positive: 80.0 - Positive → Negative: 20.0 - Negative → Positive: 78.8 - Negative → Negative: 21.2 - Portfolio equity flows - Positive → Positive: 82.5 - Positive → Negative: 17.5 - Negative → Positive: 75.2 - Negative → Negative: 24.8 - Comparative panels: Sub-Saharan Africa versus Emerging Markets (labels preserved: "Sub-Saharan Africa" and "Emerging Markets"). *Source: IMF, Regional Economic Outlook: Sub‑Saharan Africa — Annex 2.1 (Capital Flows to Sub‑Saharan Africa: Causes and Consequences).* ### Annex Table 2.1.3. Transition Probabilities for Liability Flows, 1980–2017 ### Annex Table 2.1.3. Transition Probabilities for Liability Flows, 1980–2017 ### Drivers of financial flows: framework and data - Sample and estimation: - Period: 1980-2017. - Dependent variable K_jt: nonofficial capital flows (net financial flows, liability flows, asset flows) in percent of GDP. - Estimation method: ordinary least squares (OLS) with standard errors clustered at the country level. - All domestic variables are lagged one period to address potential endogeneity. - Model (equation A2.1.1): - K_jt = Σ_i β_i x_i,t + Σ_k γ_k z_k,jt + μ_j + ε_jt - x: global push factors; z: domestic pull factors; μ_j: time-invariant country-specific fixed effects; ε_jt: error term. ### Global push factors considered - US interest rate: 10-year nominal US government bond yield (results robust to real yield or other US yield proxies). - Global market uncertainty: volatility of S&P 500 index returns (used because VIX is available only from 1990; correlation with VIX is 0.53). - International commodity price index (log). ### Domestic pull factors considered - Lagged current account balance (percent of GDP) as a proxy for external financing need. - De jure financial account openness (Chinn and Ito, 2008). - Real GDP growth rate. - Trade openness. - Institutional quality. - Real GDP per capita (log). - External debt ratio (external debt/GDP). - De facto exchange rate regime (lower values = less flexible regimes). - Note: domestic policy rate / discount rate, expected real exchange rate depreciation, and stock of foreign exchange reserves were considered but found statistically insignificant in these specifications. ### Key estimation findings (net, liability, and asset flows) - Net financial flows: - A 100 basis point decline in the nominal US government bond yield, on average, implies an increase in net flows by about 0.2 to 0.4 of a percent of GDP. - Commodity prices: statistically significant when only global factors and country fixed effects are included; coefficient weakens when additional domestic factors are added. - Among domestic factors, higher real GDP growth, higher real GDP per capita, greater trade openness, and a larger external financing need are associated with more net inflows. - Liability flows: - Much of the effect of lower US interest rates on net flows operates through liability flows. - A 100 basis point decline in the US government bond yield, on average, increases liability flows by about 0.3 to 0.5 of a percent of GDP. - A 10 percent increase in the commodity price index implies an increase in liability flows by about 0.2 to 0.3 of a percent of GDP. - Asset flows: - Evidence that higher US interest rates and greater global market volatility lead to greater asset inflows, consistent with domestic residents retrenching from foreign assets as global financial conditions tighten. - By flow type (qualitative magnitudes): - US interest rates and commodity prices have a stronger effect on inward direct investment (FDI) than on other types of flows. - Global market volatility has a statistically stronger effect on foreign portfolio flows than on FDI. ### Quantitative examples by flow type - FDI: - A 100-basis point reduction in the US government bond yield implies an increase of about 0.2 of a percent of GDP in FDI flows to sub-Saharan Africa. - Portfolio flows: - A 100-basis point reduction in the US government bond yield implies an increase of about 0.03 of a percent of GDP in foreign portfolio flows. - A one standard deviation shock to the global market volatility index reduces portfolio flows by about 0.1 of a percent of GDP; no statistically significant effect on FDI from the same shock. ### Model fit and sample statistics (selected) - Observations (examples across specifications): 1,531; 1,456; 1,343; 811. - Countries (examples across specifications): 44; 43; 29; 44. - Adjusted R-squared (examples across specifications): 0.230; 0.276; 0.292; 0.281. - Notes: Sample size varies across estimations because of data availability. Clustered standard errors (by country) reported in parentheses in the original table. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. ### Definition of surge episodes (surge and reversal) - Surge definition (equation A2.1.2): - An observation is a large inflow episode ("surge") if it is above the top 30th percentile of the country’s own distribution of liability flows (percent of GDP) and in the top 30th percentile of the entire sample’s distribution of liability flows (percent of GDP). - Formally, S_jt = 1 if K_jt ∈ {top 30th percentile of country j’s distribution} ∩ {top 30th percentile of sample distribution}; S_jt = 0 otherwise. - Rationale: - The dual criterion ensures surges are large relative both to the country’s historical experience and to cross-country standards, preventing misclassification for countries that typically experience very small inflows or frequent outflows. - Classification of surge type: - To classify each surge observation as an asset-flow or liability-flow surge, the analysis compares whether the change in asset flows is greater or smaller than the change in liability flows (see Ghosh and others, 2014 for details referenced in the original). *Source: IMF staff estimates.* ### Annex Table 2.1.5 Sub-Saharan Africa: Drivers of Different Types of Liability Flows, 1980–2017 ### Annex Table 2.1.5 Sub-Saharan Africa: Drivers of Different Types of Liability Flows, 1980–2017 ### Key regression findings (liability flows by type) - US interest rate - Negative and statistically significant effects on several liability flow measures (examples from table coefficients): -0.296***, -0.194***, -0.179***, -0.381**, -0.082**, -0.090**, -0.094***, -0.091***, -0.104***, -0.230***. - Commodity prices (log) - Positive and often statistically significant association with liability inflows across specifications: 2.303***, 1.475**, 1.561**, 1.080, 0.147, 0.164*, 0.607, 0.672***, 0.489***, 0.485***, 0.373*, 0.757**. - S&P500 index volatility (and VIX in monthly analysis) - Mixed effects: some negative coefficients (e.g., -0.311***, -0.806***, -0.998*** in monthly analyses) indicating higher global volatility reduces flows; some small positive or insignificant coefficients in certain cross-sectional specifications (example coefficient 0.013 or 0.021). - Domestic pull factors - Current account balance/GDP: negative coefficients where significant (examples: -0.030***, -0.032***, -0.033***, -0.049***), implying larger external financing needs (more negative current account) are associated with higher inflows. - Trade openness: positive and significant in many specifications (examples: 0.061***, 0.054***, 0.053, 0.014***, 0.013***, 0.014***). - Real GDP growth: positive and significant for surge likelihood and some flow measures (examples: 0.037***, 0.035***, 0.032*), mixed/insignificant in some other liability flow regressions. - Real GDP per capita (log): large and positive in some specifications for “other investment” (examples: 3.612***, 4.233***), mixed or insignificant elsewhere (examples: 0.720, 0.724, -1.652). - Exchange rate regime: coefficients indicating more flexible regimes associated with higher surge likelihood in some specifications (examples: 1.127**, 0.921**, 0.276*, 0.249*), but mixed elsewhere. - Institutional quality: positive and sometimes significant for surges and flows (examples: 3.807, 6.682*, 4.734*, -4.666** in interacted specifications). ### Probit results — likelihood of surge and reversal (1980–2017) - Surges - Global drivers: US interest rates and commodity prices are statistically significant determinants of surge occurrence. - Magnitude example: "Against an unconditional surge probability of about 23 percent in the estimated sample, a 100 basis point decline in the US government bond yield is associated with a 2 percentage point higher likelihood of a surge (evaluated at mean values of other regressors)." - Pseudo-R2 and classification performance: - Pseudo-R2 (global factors only): 0.17 (reported as 17 percent for the full model comparison). - With pull factors added: pseudo-R2 increases to a range of 0.21–0.26. - Probit sensitivity (proportion of surges correctly called): about 27 percent with global factors only; increases to 33–55 percent with pull factors. - Overall classification: "the probit calls 90 percent of the observations correctly" when pull factors included. - Reversals - Global drivers: US interest rates show somewhat symmetric effects for reversals; commodity prices do not have symmetric effects (commodity price increases lower likelihood of a reversal in resource-intensive countries only). - Domestic factors that lower reversal likelihood: higher real GDP growth, greater exchange rate flexibility, better institutional quality. - Probit performance: - Pseudo-R2 with pull factors: around 0.16–0.34. - Sensitivity (proportion of reversals correctly called): about 21–28 percent. - Specificity (proportion of non-occurrences correctly classified) reported in tables with values such as 92.72, 93.58, 93.84, 93.93 (for surge models) and 95.52, 95.81, 94.88, 95.11, 96.22 (for reversal models) — exact values vary by specification. ### Monthly analysis — fund flows, stock returns, and bond yields (selected magnitudes) - Data and approach - Monthly bond and equity fund flows, sovereign bond yields, and real stock returns estimated controlling for global and domestic variables, with country fixed effects and country-year or country-quarter dummies in variants. - Fund flows (bond and equity) - US interest rate: "a 100 basis point increase decline in the US interest rate is, on average, associated with about 0.3 of a percent of GDP higher bond fund flows, and about 0.1 of a percent of GDP higher equity fund flows." - VIX (global risk appetite): "a one standard deviation decrease in (log) VIX increases bond and equity fund flows by 0.2 and 0.1 of a percent of GDP, respectively." - Sovereign bond yields and stock returns - Bond yields: "A 100 basis points increase in the US interest rate and a one standard deviation increase in the VIX (in log terms) leads to an increase in sovereign bond yields by about 111 and 18 basis points, respectively, while a one standard deviation increase in commodity prices lowers the yields by about 21 basis points." - Stock returns: "the effect of the VIX index is statistically the strongest, with a one standard deviation increase in the VIX index on average implying a drop in stock returns by about 1.7 percentage points." - Pre- versus post-global financial crisis - Strengthening after the crisis: effects of VIX and US interest rate on fund flows are larger postcrisis. - Example: "a one standard deviation decrease in the VIX (in log terms) implies an increase in fund flows to sub-Saharan Africa by about 0.2 of a percent of GDP before 2009, but by about 0.5 of a percent of GDP after the crisis." - Representative monthly coefficient examples (from Annex Tables) - US interest rate on bond flows (full sample): -0.048***, -0.327***, -0.467*** (various specifications). - Commodity prices on bond+equity flows (full sample): 0.814***. - S&P500 index volatility on bond+equity flows and returns: coefficients such as -0.998***, -1.401***, -2.028*** in panel specifications; large negative effects on stock returns and bond yields in several columns (e.g., -9.942***, -15.194*** in some specifications for bond yield-related regressions). - Postcrisis monthly sample (2009M1–17M12) coefficient examples: US interest rate on bond flows -0.030, -0.386***, -0.634*** across columns; Commodity prices showing 0.220***, 1.362*** in specific flow specifications; S&P500 volatility coefficients -0.413***, -0.773***, -0.746*** in flow specifications and large negative effects on bond yields and stock returns in related columns. ### Model performance and data notes - Cross-sectional regressions (1980–2017) report Adjusted R2 values ranging across specifications (examples: 0.397, 0.445, 0.437, 0.338, 0.069, 0.072, 0.060, 0.190, 0.124, 0.170, 0.191, 0.198). - Probit models report Pseudo-R2 values and sensitivity/specificity measures; illustrative values include Pseudo-R2: 0.170, 0.214, 0.226, 0.231, 0.262 (surge columns) and 0.143, 0.164, 0.174, 0.198, 0.340 (reversal columns). - Monthly regressions report Adjusted R-squared values that are materially larger in many specifications (examples: 0.0803, 0.321, 0.486; and in some country*period fixed-effect specifications 0.987, 0.990, 0.994). - Sample coverage and specification notes: - Flows are in percent of GDP; stock returns in percentage points; bond yields in percent. - All domestic variables in annual regressions are one-period lagged to mitigate endogeneity. - Standard errors are clustered by country. - Significance notation: ***,**, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. - Sample sizes and country counts vary across specifications because of data availability. *Source: IMF staff estimates.* ### Annex Table 2.1.7 Sub-Saharan Africa: Drivers of Monthly Fund Flows, Stock Return and Bond Yield, 2000–2017 ### Annex Table 2.1.7 Sub-Saharan Africa: Drivers of Monthly Fund Flows, Stock Return and Bond Yield, 2000–2017 ### Drivers in Monthly Regressions (selected coefficients) - US interest rate coefficients across columns: -0.090; -0.147*; -0.112; -0.011; -0.079*; -0.035; -0.143**; -0.427; -0.120; 0.067; -2.268*; -2.896; 0.019; 0.054; 0.057. - Reported standard errors (in same order): (0.057); (0.076); (0.086); (0.008); (0.040); (0.036); (0.044); (0.240); (0.098); (0.530); (1.207); (1.744); (0.010); (0.027); (0.031). - Commodity prices (log) coefficients across columns: 0.133; 0.21; -0.409*; 0.001; 0.005; -0.111; 0.079; 0.107; -0.564; -0.259; 2.216; 4.185; 0.016; -0.07; -0.029. - Reported standard errors (in same order): (0.083); (0.142); (0.208); (0.024); (0.174); (0.187); (0.058); (0.197); (0.557); (0.861); (2.217); (7.159); (0.009); (0.055); (0.046). - S&P 500 index volatility coefficients across columns: -0.285**; -0.290**; -0.237*; -0.117**; -0.244**; -0.510; -0.425**; -0.422**; -1.474; -4.880***; -5.638***; -5.273; 0.049*; 0.117**; 0.061. - Reported standard errors (in same order): (0.114); (0.115); (0.109); (0.039); (0.080); (0.400); (0.123); (0.138); (1.194); (0.788); (1.630); (3.251); (0.021); (0.035); (0.029). ### Regression controls, sample and fit (from table) - Country-fixed effects: Yes (all specifications). - Country*Year Dummy / Country*Quarter Dummy: Yes in specified columns. - Observations by specification: 293; 293; 293; 648; 648; 648; 150; 150; 150; 733; 733; 733; 340; 340; 340. - Adjusted R-squared by specification: 0.386; 0.388; 0.499; 0.0114; 0.0264; 0.138; 0.0434; 0.146; 0.212; 0.043; 0.064; 0.115; 0.993; 0.994; 0.998. - Countries counts by specification: 8; 8; 8; 11; 11; 11; 6; 6; 6; 9; 9; 9; 5; 5; 5. - Note: Bond and equity flows are in percent of GDP. Stock return is in percentage points and bond yield is in percent. Standard errors reported in parentheses are clustered by country. ***,**, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. ### Table scope - Dependent variables summarized in table headline: Bond flows; Equity flows; Bond+Equity flows; Stock return; Bond yield. --- ### Macroeconomic Consequences: Vulnerabilities ### Conceptual framing and econometric specification - Vulnerabilities examined: currency overvaluation (deviation of the real effective exchange rate from its trend), output gap (deviation of real GDP from its trend), and rapid credit growth (change in the domestic credit-to-GDP ratio). - Estimated model (A2.1.6): V_jt = ρ f_jt + ∑_k γ_k w_k,jt + δ_j + φ_t + ε_jt. - V denotes vulnerability in country j at time t. - f is financial flow (in percent of GDP). - w are control variables (per capita income, real GDP growth, credit-to-GDP ratio, trade openness, etc.). - δ denotes country-fixed effects; φ are year effects. ### Key findings on vulnerabilities (Annex Table 2.1.11) - Net financial flows/GDP coefficients: -0.005; -0.025; -0.000. Standard errors: (0.042); (0.019); (0.011). - Liability flows/GDP coefficients: 0.011; -0.018; -0.000. Standard errors: (0.043); (0.022); (0.011). - Asset flows/GDP coefficients (reported across columns): -0.016; 0.020; -0.010; -0.016; -0.021; -0.013. Standard errors: (0.048); (0.054); (0.026); (0.037); (0.018); (0.028). - FDI liability flows/GDP coefficients: 0.004; -0.016; -0.009. Standard errors: (0.112); (0.056); (0.022). - Portfolio liability flows/GDP coefficients: 0.276**; 0.157***; 0.099**. Standard errors: (0.124); (0.031); (0.045). - Other investment liability flows/GDP coefficients: 0.027; -0.029; 0.006. Standard errors: (0.047); (0.032); (0.015). - Trade openness coefficients (by column group): -0.018; -0.020; -0.017; 0.053***; 0.052***; 0.055***; -0.001; -0.002; 0.001. Standard errors: (0.040); (0.042); (0.049); (0.014); (0.015); (0.016); (0.009); (0.009); (0.010). - Real GDP per capita (log) coefficients: 0.913; 0.903; 0.734; 3.681***; 3.662***; 3.766***; 0.216; 0.152; 0.096. Standard errors: (0.977); (1.054); (1.074); (0.868); (0.866); (0.896); (0.480); (0.454); (0.427). - Initial credit to GDP coefficients: -0.044**; -0.047**; -0.056***. Standard errors: (0.019); (0.019); (0.020). - Notes on table construction: - Dependent variables: REER overvaluation (cols. 1–3), output gap (cols. 4–6), change in private credit/GDP (cols. 7–9). - Capital flow variables in percent of GDP. All variables are lagged one period. Initial credit to GDP ratio is lagged two periods. Regressions include country and year-fixed effects. Clustered standard errors at country level reported in parentheses. ***,**, * denote statistical significance at the 1, 5, and 10 percent levels, respectively. ### Text summary of vulnerabilities - Aggregate result: Neither net, nor aggregate liability or asset flows are on average statistically associated with currency overvaluation, output gap, and credit growth for sub-Saharan African countries. - Disaggregated result: Portfolio flows are associated with generating vulnerabilities. - Quantified effect reported: "a 10 percent of GDP increase in portfolio inflows is, on average, associated with about a 3 percentage point deviation of the real exchange rate and output from trend and a 1 percent of GDP increase in domestic credit to the private sector." - Additional sample-split note: Splitting the sample into middle and low-income countries suggests much of the effects of portfolio flows are in the middle-income country sub-sample. --- ### Domestic Investment ### Investment model specification (A2.1.7) - Investment_it (total, private, public investment in percent of GDP) regressed on various flow measures (total net capital flows, asset and liability flows, FDI, portfolio, other investment flows, all in percent of GDP), country-fixed effects, year effects, and controls (log real GDP per capita, real GDP growth, trade openness, exchange rate flexibility measure, external and total public debt). - All regressors lagged; standard errors clustered at country level. ### Key econometric findings (summary of Annex Tables) - Annex Table 2.1.12 (Total Investment, 1980–2017): - Net flows/GDP coefficients: 0.190**; 0.165**; 0.179**; 0.188**. Standard errors: (0.075); (0.069); (0.070); (0.079). - Asset flows/GDP coefficients (selected): 0.083; 0.115; 0.116; 0.115; 0.296***; 0.290***; 0.286***. Standard errors: (0.224); (0.184); (0.184); (0.193); (0.075); (0.076); (0.085). - Liability flows/GDP coefficients: 0.237***; 0.181**; 0.197***; 0.204**. Standard errors: (0.075); (0.071); (0.070); (0.077). - FDI liability flows/GDP coefficients: 0.553***; 0.457***; 0.474***. Standard errors: (0.088); (0.100); (0.111). - Portfolio liability flows/GDP coefficients: 0.245; 0.224; 0.175. Standard errors: (0.158); (0.155); (0.142). - Interpretation in text: "a 1 percent of GDP increase in these flows is associated with about 0.2 of a percent of GDP increase in total investment." - FDI effect highlighted: "a one percent of GDP increase in FDI being associated with a 0.5 of a percent of GDP increase in domestic investment." - Annex Table 2.1.13 (Private Investment, 1980–2017): - Net flows/GDP coefficients: 0.218***; 0.212***; 0.222***; 0.234***. Standard errors: (0.060); (0.059); (0.062); (0.069). - Asset flows/GDP coefficients (selected): 0.213*; 0.212*; 0.196*; 0.209*; 0.298***; 0.283***; 0.271***. Standard errors: (0.125); (0.107); (0.106); (0.112); (0.071); (0.070); (0.080). - Liability flows/GDP coefficients: 0.246***; 0.227***; 0.243***; 0.254***. Standard errors: (0.069); (0.070); (0.069); (0.078). - FDI liability flows/GDP coefficients: 0.485***; 0.509***; 0.548***. Standard errors: (0.084); (0.087); (0.098). - Portfolio liability flows/GDP coefficients: 0.290*; 0.243; 0.199. Standard errors: (0.162); (0.145); (0.125). - Text interpretation: Association of net and liability flows with private investment is of about the same magnitude as with total investment; FDI is significantly positively associated with private investment. - Annex Table 2.1.14 (Public Investment, 1980–2017): - Net flows/GDP coefficients: 0.017; -0.002; 0.002; 0.003. Standard errors: (0.032); (0.030); (0.026); (0.026). - Asset flows/GDP coefficients: -0.078; -0.054; -0.042; -0.051. Standard errors: (0.085); (0.065); (0.062); (0.063). - Liability flows/GDP coefficients: 0.033; -0.003; -0.010; -0.012. Standard errors: (0.029); (0.029); (0.024); (0.023). - FDI liability flows/GDP coefficients: 0.097; 0.007; -0.012. Standard errors: (0.060); (0.063); (0.059). - Portfolio liability flows/GDP coefficients: -0.048; -0.024; -0.028. Standard errors: (0.065); (0.056); (0.052). - Text interpretation: "no financial flows of any type are found to be statistically significantly associated with public investment." - Additional note from text: - If public consumption (in percent of GDP) is considered instead of public investment, the association with portfolio flows is positive and statistically significant (Annex Table 2.1.15). - Robustness: Results are robust to 5-year averages and two-stage least squares instrumental variable strategy. --- ### Econometric Results: Text summary and magnitudes - Aggregate finding: Net and liability flows have a statistically strong positive association with domestic investment; magnitude about 0.2 percent of GDP increase in total investment per 1 percent of GDP increase in flows. - FDI is the main driver: 1 percent of GDP increase in FDI associated with ~0.5 percent of GDP increase in domestic investment. - Portfolio flows: little evidence that portfolio flows, on average, affect domestic investment; some association with private investment is weaker and less consistent. - Public investment: no significant associations with financial flows found. - Vulnerability channel: portfolio inflows are associated with currency overvaluation, positive output gaps, and higher private credit growth (quantified: 10 percent of GDP portfolio inflows → about 3 percentage point real exchange rate and output deviation from trend; 1 percent of GDP increase in domestic private credit). *BACKGROUND PAPER: ONLINE ANNEX–CAPITAL FLOWS TO SUB-SAHARAN AFRICA: CAUSES AND CONSEQUENCES (tables and text excerpts provided).* ### Annex Table 2.1.16 Sub-Saharan Africa: Impact of Portfolio Flows on Public Expenditure, 1980–2017 ### Annex Table 2.1.16 Sub-Saharan Africa: Impact of Portfolio Flows on Public Expenditure, 1980–2017 ### Main regression findings (public consumption to GDP, percent) - Dependent variable: public consumption to GDP (in percent) from World Economic Outlook. - Net flows/GDP: - Column (1): 0.051 (0.049) - Column (2): 0.041 (0.036) - Column (3): 0.037 (0.036) - Column (4): 0.036 (0.036) - Asset flows/GDP: - Col (1)–(7) entries: 0.022 0.047 0.044 0.040 0.028 0.023 -0.004 - Standard errors (where shown): (0.069) (0.082) (0.081) (0.085) (0.071) (0.060) (0.057) - Liability flows/GDP: - Column (1): 0.039 (0.053) - Column (2): 0.018 (0.038) - Column (3): 0.029 (0.042) - Column (4): 0.026 (0.042) - FDI liability flows/GDP: - Column (1): 0.132 (0.088) - Column (2): 0.045 (0.055) - Column (3): 0.034 (0.052) - Portfolio liability flows/GDP: - Column (1): 0.211** (0.103) - Column (2): 0.241** (0.101) - Column (3): 0.222** (0.094) - Other investment liability flows/GDP: - Column (1): -0.046 (0.043) - Column (2): -0.045 (0.036) - Column (3): -0.055 (0.040) - Trade openness/GDP: - Multiple columns show positive and significant coefficients: - Examples: 0.070** (0.026); 0.072*** (0.026); 0.074** (0.028); 0.081*** (0.029); 0.091*** (0.030) - Controls (selected): - GDP per capita (log of): values across columns include -0.208, -0.131, 0.290, 0.124, -0.108, 0.113, 0.463, 0.267, -0.114, 0.018, 0.135 (with standard errors shown). - Real GDP Growth: values reported include -0.046, -0.036, -0.023, -0.053, -0.043, -0.030, -0.068, -0.048 (with standard errors). - Exchange rate regime: e.g., -0.259 (0.593); -0.311 (0.614); -0.127 (0.486). - External Debt/GDP: e.g., 0.008 (0.012); 0.007 0.009 0.007 0.003 (with standard errors). - Public Debt/GDP: values include -0.008, -0.006, -0.004 (with standard errors). - Estimation details and fit: - Country-fixed effects: Yes (all specifications). - Year effects: Yes (all specifications). - Observations by column: 1,424 1,365 1,324 1,287 1,404 1,346 1,304 1,267 1,263 1,216 1,142 - R-squared by column: 0.052 0.102 0.112 0.119 0.051 0.095 0.106 0.111 0.068 0.125 0.145 - Number of ifs code: 43 43 42 42 43 43 42 42 42 40 40 - Statistical notes: - Standard errors in parentheses clustered by country. - Extreme observations (top and bottom percentile) excluded. - All regressors are lagged to mitigate endogeneity. - ***,**, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. ### Disaggregated government expenditure regressions (government expense; compensation of public employees; compensation + purchases of goods & services; social benefits — all to GDP, percent) - Asset flows/GDP (columns across different dependent variables): - Sequence: 0.105 0.053 0.057 0.035 0.025 0.025 0.040 0.022 0.016 0.009 0.008 0.009 - Standard errors shown: e.g., (0.089) (0.067) (0.058) (0.030) (0.023) (0.016) etc. - FDI liability flows/GDP: - Sequence: 0.086 0.004 0.010 0.025 -0.008 -0.028 0.084 0.045 0.028 0.004 0.012 0.009 - Portfolio liability flows/GDP: - Column examples with significance: - 0.249** (0.103) - 0.275** (0.114) - 0.219** (0.095) - 0.079*** (0.025) - 0.082*** (0.026) - 0.085*** (0.029) - 0.110*** (0.036) - 0.117** (0.047) - 0.127** (0.051) - 0.049** (0.018) - 0.046** (0.018) - 0.039** (0.019) - Other investment liability flows/GDP: - Entries include: -0.083* -0.069 -0.046 -0.012 -0.012 -0.025* -0.036 -0.032 -0.037 0.001 0.004 -0.002 - Standard errors examples: (0.047) (0.046) (0.041) (0.015) (0.017) (0.014) etc. - Trade openness/GDP: - Examples of coefficients and significance: 0.063** 0.058* 0.025 0.021 0.026 0.027 - Exchange rate regime: - Examples: -0.837 -1.028** -0.149 0.073 -0.138 -0.001 (with standard errors indicated). - Other details: - Country-fixed effects: Yes (all specifications). - Year effects: Yes (all specifications). - Observations (selected): 988 945 895 934 891 841 848 809 762 559 546 513 - R-squared (selected): 0.099 0.120 0.179 0.115 0.155 0.171 0.098 0.108 0.131 0.258 0.295 0.399 - Number of ifs code by column: 41 40 40 40 39 39 39 38 38 27 27 27 - Interpretation highlighted in text: - "Disaggregating government expenditure into compensation of public employees (as well as the sum of compensation of public employees and purchases of goods/services) and social benefits shows that these current expenditures are statistically strongly (and positively) associated with portfolio inflows (Annex Table 2.1.16)." ### Notes on estimation and robustness (from table footnotes and surrounding text) - Dependent variables drawn from World Economic Outlook (public consumption, government expense, compensation of public employees, purchases of goods and services, social benefits) and Penn World Tables (real GDP growth in related tables). - Extreme observations (top and bottom percentile) excluded. - All regressors are lagged (except where noted for other tables) to mitigate endogeneity concerns. - Standard errors clustered by country. - Results are robust to using alternate variables for government current and capital expenditure (fiscal accounts) instead of national accounts public consumption and investment. - Splitting the sample pre/post global financial crisis: unconditional correlations show portfolio flows positively associated with public expenditure but not with public investment in both periods. - Resource intensity: correlation between public investment and portfolio flows is positive for nonresource intensive countries. *Source: sreo1018background - Annex Table 2.1.16 Sub-Saharan Africa: Impact of Portfolio Flows on Public Expenditure, 1980–2017* ### 3. The Future of Work in Sub-Saharan Africa ### 3. The Future of Work in Sub-Saharan Africa ### Modeling approach - Framework extended from Berg, Buffie and Zanna (2018) to include two regions: sub-Saharan Africa (SSA) and the rest of the world (ROW), with trade/financial linkages. - Time: discrete time, no uncertainty. Each region populated by a single infinitely-lived household that owns labor (L), capital (K) and robots (Z), and firms; households can borrow/save vis-à-vis the other region. - Three assets with returns: rKt (capital), rZt (robots), rBt (financial). No-arbitrage implies equalized net returns across countries in equilibrium. - Accumulation rules: - Kt+1 = IKt + (1–δK)Kt - Zt+1 = IZt + (1–δZ)Zt - δK and δZ are depreciation rates of capital and robot stock. - Production: - Labor and robots combined with a CES technology to produce “tasks”; tasks combined with capital via Cobb-Douglas to produce final output. - Elasticity of substitution between labor and robots denoted σ; CES allows modeling complements or substitutes. - Technological change modeled as an increase in robot productivity bt. Total factor productivity Ai is the only parameter varying across regions. - Calibration: - Follows Berg, Buffie and Zanna (2018). - For each σ, calibrate SSA total factor productivity and the labor share in task production (e) to match relative GDP per capita between regions and the robot share in income in ROW. - Because TFP is lower in SSA, initial steady-state wages are lower in SSA. ### Key model findings and scenarios - General equilibrium result after a doubling of robot productivity: both regions reach higher GDP per capita steady states. - Effect on relative gains depends on σ (how robots and labor interact): - If robots and labor are complements: - SSA can grow faster than ROW; increase in GDP per capita will be higher for SSA. - Labor share can increase, particularly in SSA. - Rationale: lower initial wages in SSA make it well-positioned to take advantage of higher robot productivity when robots complement workers. - If robots and labor are substitutes: - ROW benefits more; SSA may be hurt and can experience a fall in GDP per capita in the short run. - Labor share declines in both regions; decline is larger in ROW because workers are relatively more expensive and replaced by robots. - Rationale: ROW has greater incentive to invest in robots (and complementary capital) when robots substitute labor because wages are high and robots easily substitute for labor. - Numerical/parameter cases illustrated in figures: - Three levels of σ used for dynamics: - σ=2 corresponds to substitutes. - σ=1 corresponds to a Cobb-Douglas production function. - σ=0.8 corresponds to complements. - Financial flows during the transition can amplify effects: - When robots and labor are substitutes, higher robot demand in ROW can channel resources out of SSA: - SSA runs a current account surplus in transition, accumulating foreign assets. - Short-run fall in SSA GDP per capita can occur in addition to declines in wages and labor share. - In the final steady state, SSA consumption can exceed output financed by interest on accumulated assets. - When robots and labor are complements, resources may flow into SSA, further boosting SSA GDP per capita beyond the direct productivity effect. ### Detailed scenario narratives (three futures) - Common framing: three scenarios termed "Africa Arisen", "Africa for Africa", and "Africa Adrift". Major contrasts summarized below. - Africa Arisen - Technical change: improves productivity of low-skilled workers and creates new professional services jobs. - Global integration: after initial frictions, deepening integration and cooperation. - Climate: productivity and technological innovations offset climate impacts; self-sufficiency in food. - Politics & state: mixed political outcomes; lean governments expand roles with growing tax base; use of technology to improve service delivery and revenue mobilization. - Economic landscape & world of work in 2040: - Thriving urban innovation hubs; integration into global value chains. - Project-based gig economy; occupations include professional services, tech entrepreneurs, low-skilled services; technology increases productivity of low-skilled (including agriculture). - Informality receding. - Skills & education: top-end creativity/STEM/entrepreneurship; low-end basic digital literacy and emotional intelligence; government-private partnerships in curricula and delivery; use of global and local online courses. - Social policies: expanded social safety nets, portable health insurance and retirement schemes. - Winners: expanding middle class; large multinational tech companies. - Losers: rentiers (weakened commodity monopolies); people outside social safety net. - Policy challenges in 2040: facilitating high job mobility and continuous learning; allocating work and meaning beyond work; making megacities efficient; reducing informality while strengthening social safety nets. - Africa for Africa - Technical change: replacement of workers across skill spectrum globally; technological adoption slower in SSA due to lower labor costs. - Global integration: regional integration spurred growth; large African market created. - Climate: climate change and pollution reduce growth potential. - Politics & state: strong push for regional integration; states drive growth but limited tax base forces prioritization; use of distributed ledger technology to facilitate African trade. - Economic landscape & world of work in 2040: - Regional transport infrastructure creates 1½ billion consumers regional market; production for local and regional market; project-based work; large regional firms prominent. - Informality remains a defining feature. - Skills & education: entrepreneurship, vocational skills, digital literacy for micro manufacturing and 3D printing; governments provide quality education across digital/vocational curricula. - Social policies: bare-bones safety nets for extreme poverty; family/community networks important. - Winners: entrepreneurs serving regional demand; returning diaspora. - Losers: workers stuck in rural/old sectors. - Policy challenges in 2040: promoting vocational and entrepreneurial skills; balancing regional integration with limited fiscal resources. - Africa Adrift - Technical change: technological change allows localization and reshoring of manufacturing, rendering earlier infrastructure investments obsolete. - Global integration: inward-looking policies, declining trade flows and development assistance; volatile geopolitics and trade wars. - Climate: climate change and pollution hit harder and faster; regular food shortages and mass migration. - Politics & state: frequent economic crises, elites strengthen control, weak states, high debt, limited tax base; governments use technology for control and censorship. - Economic landscape & world of work in 2040: - Stagnant economies dependent on primary commodities; subsistence activities dominate; large-scale economic migration. - Informality increasing. - Skills & education: low-quality government education; elites use private schools; access to global online courses and accreditation seen as escape. - Social policies: governments focused on control/security initially; later modest improvements in service delivery from a low base. - Winners: migrants who moved to advanced economies; commodity-owning elites. - Losers: non-elite common workers; those outside elite networks. - Policy challenges in 2040: immediate priority on security and crime prevention; later, adjustment to technological upheaval and catching up. ### Policy implications and challenges - Enabling complementary investments when robots complement workers can help SSA capture larger gains: - Policies to promote adoption where robots complement labor could raise SSA GDP per capita and labor share. - Mitigating adverse distributional and transition dynamics when robots substitute for labor: - Need for policies to manage short-run capital outflows and labor-share declines, particularly where ROW demand for robots draws resources away from SSA. - Broader policy priorities across scenarios: - Facilitate high job mobility, continuous learning, and reskilling. - Strengthen digital and regional infrastructure. - Expand and adapt social safety nets (portable health insurance, retirement saving) suited to a more project-based/gig economy. - Invest in education focused on creativity, STEM, vocational skills and basic digital literacy. - Address urban megacity productivity and informality reduction. ### Caveats and limitations - Modeling technological change remains an open question; the nature and magnitude of new technologies are still nascent. - The model is highly simplified and omits important features: - Inequality within countries, unemployment, human capital dynamics, role of government, capital market frictions, endogenous innovation, political economy factors. - Given uncertainty about technological change, results are qualitative and scenario-driven rather than precise quantitative forecasts. *Source: IMF staff calculations.* ### References ### sreo1018background - References ### Data and Projections - Unless otherwise noted, data and projections presented in this Regional Economic Outlook are IMF staff estimates as of September 20, 2018, consistent with the projections underlying the October 2018 World Economic Outlook. - The data and projections cover 45 sub-Saharan African countries in the IMF’s African Department. - Note: “...” denotes data not available. ### Country Groupings (definitions) - Countries are aggregated into three (non-overlapping) groups: - Oil exporters: net oil exports make up 30 percent or more of total exports. - Other resource-intensive countries: non-renewable natural resources represent 25 percent or more of total exports. - Non-resource-intensive countries: 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 2015–17 of more than US$995.00 (World Bank, using the Atlas method). - Low-income countries: average per capita gross national income in 2015–17 equal to or lower than US$995.00 (World Bank, using the Atlas method). - Countries in fragile situations: average CPIA scores of 3.2 or less in 2015–17 and/or presence of a peacekeeping or peacebuilding mission within the last three years. ### Regional Cooperation Memberships (regional bodies listed) - CFA franc zone (WAEMU and CEMAC), COMESA, EAC-5 (including Rwanda and Burundi from 2007), ECOWAS, SADC, SACU. - EAC-5 aggregates include Rwanda and Burundi, which joined the group only in 2007. ### Methods of Aggregation (by table groups) - Tables SA1–SA3, SA6–SA7, SA13, SA15–SA16, SA22–SA23: country group composites = arithmetic average 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. - Tables SA8–SA12, SA17–SA21, SA24–SA26: country group composites = arithmetic average weighted by GDP in US dollars at market exchange rates as a share of total group GDP. - Tables SA4–SA5 and SA14: country group composites = geometric average weighted by GDP valued at purchasing power parity as a share of total group GDP. Source of PPP weights: WEO database. - Tables SA27–SA28: country group composites = unweighted arithmetic average of data for individual countries. ### Statistical Appendix: Sources and Footnotes (selected) - Tables SA22–SA23 Source: IMF, Information Notice System. - 1 An increase indicates appreciation. - Table SA26 Sources: IMF, Common Surveillance database, and IMF, World Economic Outlook database, October 2018. - 1 As a member of the West African Economic and Monetary Union (WAEMU), see WAEMU aggregate for reserves data. - 2 As a member of CEMAC, see CEMAC aggregate for reserves data. - 3 Fiscal year data. - 4 In constant 2009 US dollars. The Zimbabwe dollar ceased circulating in early 2009. Data are based on IMF staff estimates of price and exchange rate developments in US dollars. Staff estimates of US dollar values may differ from authorities’ estimates. - Table SA27 Source: IMF, International Financial Statistics. - 1 Includes offshore banking assets. - Table SA28 Source: IMF, International Financial Statistics. - 1 Loan-to-deposit ratio includes deposits and loans of commercial banks to the public sector. - Tables SA1–SA3, SA6–SA19, SA21, SA24–25 Source: IMF, Common Surveillance database and IMF, World Economic Outlook database, October 2018. - 1 Fiscal year data. - 2 In constant 2009 US dollars. The Zimbabwe dollar ceased circulating in early 2009. Data are based on IMF staff estimates of price and exchange rate developments in US dollars. Staff estimates of US dollar values may differ from authorities’ estimates. - Tables SA4–SA5 Sources: IMF, World Economic Outlook database, October 2018. - 1 In constant 2009 US dollars. The Zimbabwe dollar ceased circulating in early 2009. Data are based on IMF staff estimates of price and exchange rate developments in US dollars. Staff estimates of US dollar values may differ from authorities’ estimates. ### Member Countries by Selected Groupings (as listed) - WAEMU: Benin, Burkina Faso, Côte d’Ivoire, Guinea-Bissau, Mali, Niger, Senegal, Togo. - CEMAC: Cameroon, Central African Rep., Chad, Congo, Republic of, Equatorial Guinea, Gabon. - COMESA (SSA members) include: Burundi, Comoros, Congo, Dem. Rep. of, Eritrea, Eswatini, Ethiopia, Kenya, Madagascar, Malawi, Mauritius, Rwanda, Seychelles, Uganda, Zambia, Zimbabwe. - EAC-5: Burundi, Kenya, Rwanda, Tanzania, Uganda. - SADC: Angola, Botswana, Congo, Dem. Rep. of, Eswatini, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, Seychelles, South Africa, Tanzania, Zambia, Zimbabwe. - SACU: Botswana, Eswatini, Lesotho, Namibia, South Africa. - ECOWAS: Benin, Burkina Faso, Cabo Verde, Côte d’Ivoire, Gambia, The, Ghana, Guinea, Guinea-Bissau, Liberia, Mali, Niger, Nigeria, Senegal, Sierra Leone, Togo. ### Examples of Country and Regional Statistics (preserved verbatim from tables) - Table SA1. Real GDP Growth (Percent) — Selected entries: - Sub-Saharan Africa: 2004-08 = 6.4; 2009 = 3.8; 2010 = 7.1; 2017 = 2.7; 2018 = 3.1; 2019 = 3.8. - Nigeria: 2004-08 = 7.7; 2015 = 2.7; 2016 = –1.6; 2017 = 0.8; 2018 = 1.9; 2019 = 2.3. - South Africa: 2004-08 = 4.8; 2009 = –1.5; 2016 = 0.6; 2017 = 1.3; 2018 = 0.8; 2019 = 1.4. - Table SA4. Consumer Prices, Average (Percent change) — Selected entries: - Sub-Saharan Africa: 2004-08 = 8.9; 2009 = 9.7; 2017 = 11.0; 2018 = 8.6; 2019 = 8.5. - Nigeria: 2004-08 = 11.6; 2009 = 12.5; 2017 = 12.4; 2018 = 12.4; 2019 = 13.5. - Table SA6. Total Investment (Percent of GDP) — Selected entries: - Sub-Saharan Africa: 2004-08 = 20.2; 2009 = 22.7; 2017 = 20.8; 2018 = 20.5; 2019 = 21.1. - Ethiopia: 2004-08 = 19.7; 2009 = 15.4; 2017 = 32.1; 2018 = 33.2; 2019 = 33.1. - Table SA8. Overall Fiscal Balance, Including Grants (Percent of GDP) — Selected entries: - Sub-Saharan Africa: 2004-08 = 1.7; 2009 = –4.6; 2017 = –4.8; 2018 = –4.2; 2019 = –3.9. - Nigeria: 2004-08 = 4.7; 2009 = –5.4; 2017 = –5.3; 2018 = –5.1; 2019 = –4.5. - Table SA12. Government Debt (Percent of GDP) — Selected entries: - Sub-Saharan Africa: 2004-08 = 22.5; 2009 = 20.0; 2017 = 18.8; 2018 = 17.9; 2019 = 17.8. - Mozambique: 2004-08 = 22.5; 2009 = 28.9; 2017 = 33.0; 2018 = 32.8; 2019 = 33.0. - Table SA20. External Current Account (Percent of GDP) — Selected entries: - Sub-Saharan Africa: 2004-08 = 2.1; 2009 = –2.4; 2017 = –2.3; 2018 = –2.8; 2019 = –3.4. - Angola: 2004-08 = 12.3; 2009 = –10.8; 2017 = –1.0; 2018 = –2.1; 2019 = –1.9. - Table SA21. Net Foreign Direct Investment (Percent of GDP) — Selected entries: - Sub-Saharan Africa: 2004-08 = 2.0; 2009 = 2.9; 2017 = 1.8; 2018 = 2.1; 2019 = 2.7. - Ghana: 2004-08 = 2.9; 2009 = 11.3; 2017 = 6.9; 2018 = 6.0; 2019 = 5.0. - Table SA26. Reserves (Months of imports of goods and services) — Selected entries: - Sub-Saharan Africa: 2004-08 = 5.0; 2009 = 5.1; 2017 = 4.9; 2018 = 4.9; 2019 = 4.8. - Nigeria: 2004-08 = 10.7; 2009 = 7.2; 2017 = 6.5; 2018 = 7.2; 2019 = 8.0. - Table SA27. Banking Penetration (Total banking sector assets in percent of GDP) — Selected entries: - Sub-Saharan Africa: 2004-08 = 41.9; 2009 = 48.1; 2017 = 57.4. - South Africa: 2004-08 = 116.4; 2009 = 120.9; 2017 = 113.6. - Table SA28. Banking Sector: Loan-to-Deposit Ratio (Percent of deposits) — Selected entries: - Sub-Saharan Africa: 2004-08 = 67.2; 2009 = 69.0; 2017 = 75.3. - Kenya: 2004-08 = 76.6; 2009 = 72.5; 2017 = 83.5. ### Publications and Reference Lists (selected headings preserved) - Lists of Books and Monographs, Departmental Papers, Policy Papers, Staff Discussion Notes, Working Papers, Spillover Notes, and Staff Publications are included in the appendix (titles and author lists provided in the source). *Source: sreo1018background - References (Regional Economic Outlook: Sub‑Saharan Africa — Statistical Appendix, IM F staff estimates and tables as presented).* --- _Source: https://www.imf.org/-/media/files/publications/reo/afr/2018/october/pdf/sreo1018background.pdf_