## wpiea2025139-print-pdf

## Source details

**Canonical URL:** [wpiea2025139-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025139-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025139-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025139-print-pdf.pdf.json)

---

### III. Introduction: scope and objectives
- Purpose: explore the evolving landscape of funding flows to sub-Saharan Africa (SSA), with focus on Eurobonds, syndicated bank loans, and non-resident flows into local bond markets.
- Key comparative questions:
  - Is there an “African risk premium” in the Eurobond market and are credit rating agencies biased?
  - What is the relative importance of syndicated loans versus Eurobond issuances for African issuers and what drives access to international investors?
  - How have domestic bond markets evolved over time and what is the role of foreign investors?
- Data sources: transaction-level data from the international bond primary market, sovereign spreads from the secondary bond market, sovereign ratings, cross-border syndicated loans, local bond yields, and macroeconomic/financial indicators.

### III. Major empirical approaches and models
- Primary market Eurobond premium estimation (panel regression with fixed effects):
  - Specification: Y_{c,i,t} = α + β(X_{c,i,t}) + SSA_c + Γ_i + Λ_t + ε_{c,i,t}
  - Y: yield to maturity at issuance; X: issuance characteristics (maturity, tranche size); SSA: sub-Saharan Africa dummy; Γ_i: rating fixed effects; Λ_t: quarterly time fixed effects.
- Eurobond spread determinants (panel regression with fixed effects):
  - Specification: Y_{c,t} = α + β(X_{c,t}) + γ Global_t + SSA_c + Γ_c + Λ_t + ε_{c,t}
  - Global push factors: US dollar index, short-term interest rates, VIX; Pull factors: economic, financial, political risk ratings (ICRG).
- Likelihood of issuing Eurobonds vs. bank loans (probit model):
  - Baseline: Issuance_{c,t} = α + β1 RiskRating_{c,t} + γ Global_t + ε_{c,t}
  - Global factors: VIX_t, 3M rate_t, USD index_t; RiskRating from ICRG (overall and sub-indexes).
  - Extended specification includes SSA interactions to test differential effects for SSA countries.
- Local-currency yields and non-resident ownership (pooled OLS with time fixed effects):
  - Specification: LC_Yield_{c,t} = α + β NonResident_{c,t} + γ X_{c,t} + Λ_t + ε_{c,t}
  - Dependent variable: yield to maturity on five-year local-currency bonds; Key regressor: share of non-resident ownership of domestic-currency debt stock; controls include trade-weighted dollar index, policy rates, economic risk indicators.
- Variable importance: Dominance Analysis (Budescu, 1993) implemented via the domin command in Stata to decompose explanatory power into contributions of domestic versus global drivers.

### III. Key empirical findings (stylized facts)
- African risk premium:
  - Estimated average African risk premium in the primary Eurobond market: about 50 basis points.
  - Premium increases substantially during periods of stress, indicating non-linearities and liquidity constraints.
  - No statistically significant African premium detected in the secondary market once controls (including risk ratings and governance) are included.
  - Limited evidence of systematic bias by credit rating agencies against SSA countries.
- International bond market size and evolution:
  - Market-access SSA countries issued on average about $9 billion per year, or 1.5 percent of the issuers’ GDP, over the past decade.
  - SSA countries borrowed nearly $40 billion between 2020 and 2024, with a similar average yearly issuance (approximately $8 billion per year) as in the pre-pandemic decade.
  - Average annual issuance during 2007-2016: $4 billion per year.
  - In 2024, Eurobonds issued by SSA countries represented 5 percent of EMDEs total; excluding G20 countries, region’s 2024 issuances represent about 7 percent of total EMDE issuances.
  - Number of distinct SSA Eurobond issuers: 5 countries prior to the GFC, 17 countries in the decade prior to the pandemic, 11 different countries post-pandemic.
  - Median bond maturity lengthened from 10 to 11 years between pre- and post-pandemic periods.
  - Surge in external borrowing has led to significantly higher external debt service for public and private sectors.
- Issuance costs and examples:
  - Côte d’Ivoire’s $2.6 billion Eurobond issuance (January 2024) had a weighted-average spread of about 400 basis points and was its most expensive to date; recent cost surge mainly reflects higher US bond yields.
  - Yield at issuance in 2024 for SSA countries was on average 300 basis points higher than in previous issuances, mirroring increases in benchmark yields.
  - Côte d’Ivoire’s spread at issuance was about 50 basis points above pricing for a similar bond from a developing country outside SSA with the same credit rating.
- Syndicated loans versus Eurobonds:
  - Syndicated bank loans are relatively more important for African countries compared to other regions; historically SSA relies more on bank loans than Eurobonds for government financing.
  - Likelihood of issuing Eurobonds or borrowing from syndicated loan market is highly sensitive to global financing conditions, particularly the US dollar.
  - US dollar strength is associated with: higher bond spreads, higher likelihood of issuing Eurobonds (as opposed to syndicated loans), and lower ratings for EMDEs; effects are stronger for SSA.
  - Likelihood of borrowing from syndicated loan market increases when the US dollar weakens; local-currency yields comove with the dollar.
  - Higher uncertainty and tighter global financial conditions are associated with an increased probability of relying on loans.
- Non-resident ownership and local yields/volatility:
  - Higher non-resident ownership of domestic-currency debt is generally associated with lower yields in EMDEs; effect is stronger for SSA.
  - Quantified example: SSA countries may experience a reduction in yields by 0.4 percentage points for a 10 percent increase in non-resident holdings.
  - In high-debt countries, an increase in non-resident ownership can be associated with higher yields.
  - Across EMDEs, higher non-resident share is associated with higher yield volatility; however, SSA countries exhibit a tendency where higher non-resident ownership correlates with reduced volatility.

### 2. Eurobond Yields to Maturity at Issuance — detailed findings
- African Risk Premium (primary market):
  - Premium for sovereign Eurobond issuances (primary market) is estimated at 46 basis points with a standard deviation of 16 basis points (Table 1, column 1).
  - This gap widens during global shocks to more than 120 basis points.
- SSA SOEs and corporations:
  - SSA premium is not statistically significant for SOEs and corporations issuing Eurobonds; borrowing costs for these issuers are not found different from peers outside the region.
- Sovereign spreads (secondary market, J.P. Morgan EMBI+ index):
  - Sovereign spreads are on average higher in SSA than in other emerging and frontier market economies, particularly after 2020 and early 2022.
- Determinants and governance:
  - Domestic factors—financial, political, and economic risks—significantly explain spread variations.
  - Dominance analysis: government stability, law and order, historical economic growth, and accumulated foreign debt explain more than 80 percent of spread variation (Figure A1 reference).
  - Including an SSA dummy is statistically non-significant when economic and sociopolitical determinants are included—no evidence of an SSA premium once these determinants are controlled for.
  - Governance quality significantly reduces spreads; governance explains about one third of the variance in sovereign spreads (Figure A1).
  - A notch improvement in sovereign ratings is associated, on average, with about a 130-140 basis points reduction in sovereign spread.
  - Testing for bias in sovereign ratings:
    - Model (3): SSA dummy negative and statistically significant at the 5 percent level—SSA countries associated with lower sovereign ratings by 1.2 notch on average.
    - Model (4): when governance index is introduced, SSA dummy decreases to 0.6 notch and loses statistical significance.
    - Conclusion: once economic, sociopolitical risks and governance are considered, differences in sovereign ratings and spreads between SSA and other EMDEs largely diminish.

### IV. Syndicated Loans — trends, volumes, and likelihood of issuance
- Definitions:
  - Eurobonds: tradable debt securities issued in a foreign currency to a global investor base.
  - Syndicated loan: large loan provided by a syndicate of lenders to a single borrower; private credit agreement with extensive covenants.
- Frequency of issuances (2000-2023, sample of 102 EMDEs):
  - Average monthly probability of Eurobond issuance:
    - Full sample: 4.7 percent
    - EMDEs excluding SSA: 6.2 percent
    - SSA countries: 0.9 percent
  - Average monthly probability of loan issuance:
    - Overall sample: 7.2 percent
    - Non-SSA: 7.8 percent
    - SSA: 5.8 percent
- Financing volumes (2000-2023):
  - Eurobonds issued:
    - EMDEs excluding SSA: 1,896 billion USD
    - SSA: 107 billion USD
  - Loans received:
    - EMDEs excluding SSA: 1,015 billion USD
    - SSA: 171 billion USD
  - For SSA countries, loans were a much more important source of financing (171 billion USD) than Eurobonds (107 billion USD) over 2000-2023.
- Post Covid-19 (2020-2023):
  - SSA loans: 52 billion USD
  - SSA Eurobonds: 24 billion USD
  - Average annual issuance by SSA governments:
    - Loans: increased from 6 billion USD (2000-2019) to 13 billion USD (2020-2023)
    - Eurobonds: increased from 4 billion USD (2000-2019) to 6 billion USD (2020-2023)
- Maturities and costs:
  - Maturity of Eurobonds and loans remain similar across SSA and other EMDEs; loans tend to have shorter maturities than Eurobonds.
  - Cost of borrowing remained higher for SSA countries, particularly for Eurobonds.
- Likelihood of issuance (probit avg. marginal effects; sample: 29,038 observations, 102 countries):
  - Eurobond issuance more likely for countries with sound economic, financial, and political environment (positive and statistically significant coefficients).
  - Eurobond issuance less likely when VIX is higher (VIX coefficients negative and significant).
  - Eurobond issuance less likely when US 3M rate is higher (US 3M rate negative and significant).
  - Stronger USD (USD index) is associated with higher likelihood of Eurobond issuance (USD index positive and significant).
  - Loan issuance more likely during periods of high uncertainty (VIX coefficients positive and significant) and when USD index is weaker (USD index negative and significant).
  - Dominance analysis:
    - Overall risk rating explains 73 percent of variation in Eurobond issuance.
    - Overall risk rating explains 50 percent of variation in loan issuance.
    - USD index explains around 37 percent of variation for loans, but only 6 percent for Eurobonds.

### Predictive performance (ROC / AUC)
- Threshold interpretation: a threshold of 0.75 and false positive rate 0.25 "represents a threshold which predicts issuance when there is issuance in the data around 75 percent of the time and predicts an issuance when there is no issuance around 25 percent of the time."
- AUC results (probit regressions):
  - Eurobonds:
    - Global-factors-only model: AUC = 0.55; standard error = 0.008.
    - Full model (adds overall risk rating): AUC = 0.63; standard error = 0.007.
  - Bank loans:
    - Global-factors-only model: AUC = 0.59; standard error = 0.007.
    - Full model (adds overall risk rating): AUC = 0.62; standard error = 0.006.
- Interpretation: combining global and domestic factors improves predictive power; full-model AUCs are informative but not close to perfect prediction.

### Recent trends in local-currency debt markets in SSA
- Issuance volumes and maturities:
  - Total domestic-currency issuances across SSA (excluding maturities < one year) averaged $10.9bn per month in 2024 — more than double the amount issued in 2019.
  - Average issuance duration fell to around three and a half years in 2024, compared with an average of more than 6 years in 2019 (after excluding securities with durations < one year).
  - Shorter maturities increase rollover frequency and potential financing risks.
- Yield dynamics:
  - Average yields on domestic-currency debt in SSA were on a downward trend between 2017 and 2021; yields have risen globally since 2022 and SSA yields increased as well.
  - Since 2012, domestic-currency yields have been higher than Eurobond yields around 80 percent of the time.
- Cross-country yield dispersion in 2024:
  - Seychelles: lowest debt interest costs at 3.2 percent.
  - Sierra Leone: highest at 39.8 percent.
  - Between 2023 and 2024, nominal yields increased in 24 of 35 countries with available data (just under 70 percent).
  - Real interest rates increased in 27 countries in 2024 (77 percent of countries with available data).

### Non-resident holdings: effects on yields and volatility
- Sample and data:
  - Monthly data from January 2011 to September 2024; final sample: 24 emerging market economies with complete information.
  - Yield series: 5-year local-currency bond yields; foreign holdings from Haver, Bloomberg, national exchanges, and country authorities.
- Effects on yields:
  - All emerging markets (Model 1): a ten-percentage point increase in non-resident ownership is associated with a 0.13 percentage point decrease in yields. Statistically significant at 1 percent.
  - SSA-specific effect (Model 2): for SSA, a ten-percentage point increase in non-resident holdings is associated with a 0.4 percentage point decrease in yields.
  - High-debt interaction (Model 3): for low-debt countries, a ten-percentage point increase in non-resident holdings is associated with a 0.09 percentage point decline in yields; for high-debt countries (public debt to GDP ratio above 60 percent), the same increase is associated with a 0.3 percentage point increase in yields.
  - Selected coefficient examples:
    - Model 1 (pooled): -0.1079*** (0.005) effect on 5-year T-bond yield.
    - Table A16 pooled models show nonresident_holdings coefficients ranging from -0.1119*** to -0.0096** depending on specification; later specifications controlling for more factors show coefficients like -0.0129*** (Model 10).
- Effects on yield volatility:
  - Volatility measure: standard deviation of bond yields over a rolling window (nineteen months for monthly series).
  - Pooled OLS result (Model 4): a ten-percentage point increase in non-resident ownership is associated with a 0.02 percentage point increase in yield standard deviation (all emerging markets). Statistically significant at 1 percent.
  - SSA-specific volatility:
    - On average, SSA countries have lower yield volatility than other EMDEs: standard deviation reduced by 0.3 percentage points.
    - For non-SSA countries: a ten-percentage point increase in non-resident holdings increases yield volatility by 0.05 percentage points.
    - For SSA countries: a ten-percentage point increase in non-resident holdings reduces volatility by 0.06 percentage points.
  - High-debt cases: higher non-resident holdings typically associated with higher yield volatility for high-debt countries.
  - Robustness: monthly-volatility results considered most robust; daily 90-day rolling-window results are mixed for SSA.

### Additional empirical robustness and sample facts
- Domestic bond market summary (Table A3 selected stats):
  - 5-year local currency government bond yield (Overall sample): Obs. 3,337; Mean 7.93; Std. Dev. 5.05; Min 0.70; Max 20.07.
  - Yield volatility: Obs. 3,291; Mean 0.75; Std. Dev. 0.59; Min 0.00; Max 3.52.
  - Non-resident holdings (percent): Obs. 3,415; Mean 18.03; Std. Dev. 12.72; Min 0.00; Max 58.33.
  - Policy rate (percent): Obs. 3,585; Mean 7.21; Std. Dev. 5.42; Min 0.25; Max 50.00.
- Eurobond and loan facts (selected):
  - Eurobond spreads and maturities (Table A12):
    - SSA: Maturity median 10; Mean 13.6; Spread median 441.5; Mean 433.6.
    - Non-SSA: Maturity median 10; Mean 12.7; Spread median 242.7; Mean 284.6.
  - Loan facts (Table A13):
    - SSA loan maturity median 8; mean 9.6; spread median 210; mean 247.8.
    - Non-SSA loan maturity median 8; mean 9.4; spread median 150; mean 189.7.

### Key policy-relevant findings and recommendations
- Drivers of issuance and costs:
  - Both global factors (VIX, US 3M rate, USD index) and domestic factors (overall risk rating) materially influence issuance likelihood for Eurobonds and loans; combining both improves predictive power (AUC rises from ~0.55–0.59 to ~0.62–0.63).
  - SSA faces an African risk premium in Eurobond markets that increases during crises; governance and institutional improvements can help reduce this premium and borrowing costs.
- Debt composition and market choice:
  - Policymakers should weigh trade-offs between foreign-currency-denominated external debt (subject to exchange rate risk and US dollar cycles) and domestic-currency debt (which can reduce FX exposure but may limit foreign investor participation where local markets are underdeveloped).
- Managing external vulnerabilities:
  - Sensitivity of issuance costs and access to global factors (notably the US dollar and global risk aversion) suggests importance of macroeconomic buffers and debt management strategies that account for global financing cycles.
- Domestic market deepening:
  - Increasing non-resident participation in domestic-currency markets can lower yields and, for SSA in the sample, may also reduce yield volatility — implying potential gains from policies that develop local-currency debt markets and attract stable foreign investor bases.
- Crisis-period readiness:
  - The increase in African premium during stress periods highlights the need for contingency plans, liquidity backstops, and access to alternative financing (e.g., syndicated loans) when global conditions tighten.
- Trade-offs and cautions:
  - Increasing non-resident participation generally reduces yields but can increase yield volatility in many emerging markets; for high-debt countries, non-resident participation may raise yields and volatility (risk of capital flight).
  - Policymakers should weigh potential financing-cost reductions against volatility and rollover risks, especially in high-debt contexts.
- Recommended policy directions (implied by findings):
  - Strengthen governance, transparency, and institutional frameworks to improve risk ratings and attract stable foreign investors.
  - Develop domestic debt markets to diversify financing sources and reduce foreign-currency exposure while managing rollover and financial-stability risks where fiscal dominance or tight sovereign-bank nexus exists.
  - Implement debt-management strategies that consider interactions between non-resident participation, public debt levels, and potential volatility.

### Conclusion and research priorities
- The financing landscape in SSA is shifting: increased reliance on Eurobonds and domestic-currency issuance amid rising borrowing costs and elevated debt levels.
- Empirical evidence supports a nuanced role for non-resident investors: potential to lower yields and, in SSA, reduce volatility — but with adverse outcomes for high-debt countries.
- Future research priorities identified:
  - Impact of governance reforms on credit ratings and non-resident investment flows in SSA.
  - Mechanisms driving lower international bond issuances by SSA countries and implications for regional financing strategies.

*International Monetary Fund — Navigating the Evolving Landscape of External Financing in Sub-Saharan Africa (excerpted content).*

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

### wpiea2025139-print-pdf - References .............................................................................................................

### III. Introduction: scope and objectives
- Purpose: explore the evolving landscape of funding flows to sub-Saharan Africa (SSA), with focus on Eurobonds, syndicated bank loans, and non-resident flows into local bond markets.
- Key comparative questions:
  - Is there an “African risk premium” in the Eurobond market and are credit rating agencies biased?
  - What is the relative importance of syndicated loans versus Eurobond issuances for African issuers and what drives access to international investors?
  - How have domestic bond markets evolved over time and what is the role of foreign investors?
- Data sources: transaction-level data from the international bond primary market, sovereign spreads from the secondary bond market, sovereign ratings, cross-border syndicated loans, local bond yields, and macroeconomic/financial indicators.

### III. Major empirical approaches and models
- Primary market Eurobond premium estimation (panel regression with fixed effects):
  - Specification: Y_{c,i,t} = α + β(X_{c,i,t}) + SSA_c + Γ_i + Λ_t + ε_{c,i,t}
  - Y: yield to maturity at issuance; X: issuance characteristics (maturity, tranche size); SSA: sub-Saharan Africa dummy; Γ_i: rating fixed effects; Λ_t: quarterly time fixed effects.
- Eurobond spread determinants (panel regression with fixed effects):
  - Specification: Y_{c,t} = α + β(X_{c,t}) + γ Global_t + SSA_c + Γ_c + Λ_t + ε_{c,t}
  - Global push factors: US dollar index, short-term interest rates, VIX; Pull factors: economic, financial, political risk ratings (ICRG).
- Likelihood of issuing Eurobonds vs. bank loans (probit model):
  - Baseline: Issuance_{c,t} = α + β1 RiskRating_{c,t} + γ Global_t + ε_{c,t}
  - Global factors: VIX_t, 3M rate_t, USD index_t; RiskRating from ICRG (overall and sub-indexes).
  - Extended specification includes SSA interactions to test differential effects for SSA countries.
- Local-currency yields and non-resident ownership (pooled OLS with time fixed effects):
  - Specification: LC_Yield_{c,t} = α + β NonResident_{c,t} + γ X_{c,t} + Λ_t + ε_{c,t}
  - Dependent variable: yield to maturity on five-year local-currency bonds; Key regressor: share of non-resident ownership of domestic-currency debt stock; controls include trade-weighted dollar index, policy rates, economic risk indicators.
- Variable importance: Dominance Analysis (Budescu, 1993) implemented via the domin command in Stata to decompose explanatory power into contributions of domestic versus global drivers.

### III. Key empirical findings (stylized facts)
- African risk premium:
  - Estimated average African risk premium in the primary Eurobond market: about 50 basis points.
  - Premium increases substantially during periods of stress, indicating non-linearities and liquidity constraints.
  - No statistically significant African premium detected in the secondary market once controls (including risk ratings and governance) are included.
  - Limited evidence of systematic bias by credit rating agencies against SSA countries.
- International bond market size and evolution:
  - Market-access SSA countries issued on average about $9 billion per year, or 1.5 percent of the issuers’ GDP, over the past decade.
  - SSA countries borrowed nearly $40 billion between 2020 and 2024, with a similar average yearly issuance (approximately $8 billion per year) as in the pre-pandemic decade.
  - Average annual issuance during 2007-2016: $4 billion per year.
  - In 2024, Eurobonds issued by SSA countries represented 5 percent of EMDEs total; excluding G20 countries, region’s 2024 issuances represent about 7 percent of total EMDE issuances.
  - Number of distinct SSA Eurobond issuers: 5 countries prior to the GFC, 17 countries in the decade prior to the pandemic, 11 different countries post-pandemic.
  - Median bond maturity lengthened from 10 to 11 years between pre- and post-pandemic periods.
  - Surge in external borrowing has led to significantly higher external debt service for public and private sectors.
- Issuance costs and examples:
  - Côte d’Ivoire’s $2.6 billion Eurobond issuance (January 2024) had a weighted-average spread of about 400 basis points and was its most expensive to date; recent cost surge mainly reflects higher US bond yields.
  - Yield at issuance in 2024 for SSA countries was on average 300 basis points higher than in previous issuances, mirroring increases in benchmark yields.
  - Côte d’Ivoire’s spread at issuance was about 50 basis points above pricing for a similar bond from a developing country outside SSA with the same credit rating.
- Syndicated loans versus Eurobonds:
  - Syndicated bank loans are relatively more important for African countries compared to other regions; historically SSA relies more on bank loans than Eurobonds for government financing.
  - Likelihood of issuing Eurobonds or borrowing from syndicated loan market is highly sensitive to global financing conditions, particularly the US dollar.
  - US dollar strength is associated with: higher bond spreads, higher likelihood of issuing Eurobonds (as opposed to syndicated loans), and lower ratings for EMDEs; effects are stronger for SSA.
  - Likelihood of borrowing from syndicated loan market increases when the US dollar weakens; local-currency yields comove with the dollar.
  - Higher uncertainty and tighter global financial conditions are associated with an increased probability of relying on loans.
- Non-resident ownership and local yields/volatility:
  - Higher non-resident ownership of domestic-currency debt is generally associated with lower yields in EMDEs; effect is stronger for SSA.
  - Quantified example: SSA countries may experience a reduction in yields by 0.4 percentage points for a 10 percent increase in non-resident holdings.
  - In high-debt countries, an increase in non-resident ownership can be associated with higher yields.
  - Across EMDEs, higher non-resident share is associated with higher yield volatility; however, SSA countries exhibit a tendency where higher non-resident ownership correlates with reduced volatility.

### III. Relation to literature and robustness
- Findings align with prior work showing the African premium disappears when controlling for institutional and structural factors (e.g., quality of institutions, budget transparency, informal sector).
- Literature estimates of African risk premium range from zero to 300 basis points depending on sample, period, and determinants.
- Results are consistent with literature emphasizing the US dollar and global risk aversion as key drivers of EMDE capital flows and sovereign spreads.
- Robustness checks: alternative specifications using overall risk rating, credit rating, orthogonalized risk-rating components; country and time fixed effects; inclusion of G20 dummy for robustness.

### III. Policy-relevant implications (derived from empirical results)
- Debt composition and market choice:
  - Policymakers should weigh trade-offs between foreign-currency-denominated external debt (subject to exchange rate risk and US dollar cycles) and domestic-currency debt (which can reduce FX exposure but may limit foreign investor participation where local markets are underdeveloped).
- Managing external vulnerabilities:
  - Sensitivity of issuance costs and access to global factors (notably the US dollar and global risk aversion) suggests importance of macroeconomic buffers and debt management strategies that account for global financing cycles.
- Domestic market deepening:
  - Increasing non-resident participation in domestic-currency markets can lower yields for SSA issuers and may reduce yield volatility in SSA, implying potential gains from policies that develop local-currency debt markets and attract stable foreign investor bases.
- Crisis-period readiness:
  - The increase in African premium during stress periods highlights the need for contingency plans, liquidity backstops, and access to alternative financing (e.g., syndicated loans) when global conditions tighten.

*International Monetary Fund — Navigating the Evolving Landscape of External Financing in Sub-Saharan Africa (excerpted content).*

### 2. Eurobond Yields to Maturity at Issuance

### 2. Eurobond Yields to Maturity at Issuance

### African Risk Premium
- Controlling for issuer-specific fundamentals (proxied by the issuer’s credit rating), global factors (proxied by time fixed effects), and bond characteristics (e.g., maturity, currency):
  - Premium for sovereign Eurobond issuances (primary market) is estimated at 46 basis points with a standard deviation of 16 basis points (Table 1, column 1).
  - This gap widens during global shocks to more than 120 basis points.
- For Eurobonds issued by sub-Saharan African (SSA) SOEs and corporations:
  - The SSA premium is not statistically significant.
  - Borrowing costs for these issuers are not found different from peers outside the region.
- Possible explanation:
  - Most sub-Saharan African corporations issuing Eurobonds tend to be rated higher (investment grade), reflecting better governance and management standards, and healthier balance sheets.
- Caveat:
  - The analysis takes credit ratings as given and does not resolve the debate on the objectivity of credit ratings.

### Spreads analysis (sovereign spreads vs. EMDEs)
- Sovereign spreads (J.P. Morgan EMBI+ index) overview:
  - Sovereign spreads are on average higher in SSA than in other emerging and frontier market economies.
  - This was particularly apparent after the onset of Covid-19 in 2020 and the start of the war in Ukraine in early 2022.
- Sample and variables:
  - 67 countries; generally covering 2000m1-2023m12 depending on first issuance date.
  - Global factors: US Dollar Index, VIX (CBOE Volatility) index, and U.S. 3-month T-bill yield.
  - Country risk: International Country Risk Guide (ICRG) scores for financial, political, and economic risks (0 to 100, higher = lower risk).
  - Governance: World Governance Indicators (WGI) aggregate measures (government effectiveness, rule of law, control of corruption, voice and accountability, political violence).
  - Aggregate measures constructed by averaging relevant sub-indicators; regressions use variants of equation (2) with fixed effects.

### Determinants of sovereign spreads and governance role
- Main regression results (Table 2 and Table 3 summaries):
  - Domestic factors—financial, political, and economic risks—significantly explain spread variations.
  - Dominance analysis: government stability, law and order, historical economic growth, and accumulated foreign debt explain more than 80 percent of spread variation (Figure A1 reference).
  - Including an SSA dummy:
    - SSA dummy is statistically non-significant when economic and sociopolitical determinants are included.
    - No evidence of an SSA premium once these determinants are controlled for.
  - Governance:
    - Model (7) in Table 2 shows governance quality significantly reduces spreads.
    - Dominance analysis indicates governance explains about one third of the variance in sovereign spreads (Figure A1).
    - Interaction results: benefits of improving governance on reducing sovereign spreads are larger in SSA than in other regions (Table A6).
- Sovereign ratings and spreads (Table 3):
  - A notch improvement in sovereign ratings is associated, on average, with about a 130-140 basis points reduction in sovereign spread.
  - Sovereign rating captures part of governance effects; when sovereign rating is included, the governance index can become insignificant.
  - Including an SSA dummy with sovereign rating remains statistically insignificant, reinforcing absence of SSA premium after controls.
- Testing for bias in sovereign ratings (Table 4):
  - Model (1): lower risk (higher risk rating) is associated with higher sovereign ratings.
  - Model (2): governance index correlates with higher sovereign ratings.
  - Model (3): SSA dummy is negative and statistically significant at the 5 percent level—SSA countries associated with lower sovereign ratings by 1.2 notch on average.
  - Model (4): when governance index is introduced, SSA dummy decreases to 0.6 notch and loses statistical significance.
  - Conclusion: once economic, sociopolitical risks and governance are considered, differences in sovereign ratings and spreads between SSA and other EMDEs largely diminish.

### IV. Syndicated Loans — Trends and developments
- Definitions and distinctions:
  - Eurobonds: tradable debt securities issued in a foreign currency to a global investor base.
  - Syndicated loan: large loan provided by a syndicate of lenders to a single borrower; private credit agreement with extensive covenants.
- Frequency of issuances (2000-2023, sample of 102 EMDEs):
  - Average monthly probability of Eurobond issuance:
    - Full sample: 4.7 percent
    - EMDEs excluding SSA: 6.2 percent
    - SSA countries: 0.9 percent
  - Average monthly probability of loan issuance:
    - Overall sample: 7.2 percent
    - Non-SSA: 7.8 percent
    - SSA: 5.8 percent
- Financing volumes (2000-2023):
  - Eurobonds issued:
    - EMDEs excluding SSA: 1,896 billion USD
    - SSA: 107 billion USD
  - Loans received:
    - EMDEs excluding SSA: 1,015 billion USD
    - SSA: 171 billion USD
  - For SSA countries, loans were a much more important source of financing (171 billion USD) than Eurobonds (107 billion USD) over 2000-2023.
- Post Covid-19 (2020-2023):
  - SSA loans: 52 billion USD
  - SSA Eurobonds: 24 billion USD
  - Average annual issuance by SSA governments:
    - Loans: increased from 6 billion USD (2000-2019) to 13 billion USD (2020-2023)
    - Eurobonds: increased from 4 billion USD (2000-2019) to 6 billion USD (2020-2023)
- Maturities and costs:
  - Maturity of Eurobonds and loans remain similar across SSA and other EMDEs; loans tend to have shorter maturities than Eurobonds (Tables A12 and A13).
  - Cost of borrowing remained higher for SSA countries, particularly for Eurobonds.

### Likelihood of issuance: Main Results
- Estimation approach:
  - Probit model (eq. (3)), average marginal effects reported (Table 5). Sample: 29,038 observations, 102 countries.
  - Domestic condition proxies: overall risk rating (average of economic, financial, political risk indexes) and individual risk ratings (columns 2-4), and residualized individual ratings (column 5).
  - Global variables included: VIX, US 3M rate, USD index.
- Eurobond issuance (columns 1-5):
  - Likelihood of Eurobond issuance is higher for countries with:
    - Sound economic, financial, and political environment (positive and statistically significant coefficients).
  - During periods of:
    - Low uncertainty (lower VIX): Eurobond issuance less likely when VIX is higher (VIX coefficients negative and significant).
    - Low interest rates (US 3M rate negative and significant).
    - Stronger USD (USD index positive and significant).
- Loan issuance (columns 6-10):
  - Likelihood of loan issuance is higher for countries with sound economic, financial, and political environment.
  - Loan issuance becomes more likely during periods of:
    - High uncertainty (VIX coefficients positive and significant).
    - Weaker US dollar (USD index negative and significant).
  - US 3M rate generally negative for loans as well, but coefficients smaller in magnitude and less consistently significant.
- Interpretations (supply/demand and instrument features):
  - Investors fund countries with sound domestic conditions for both instruments (supply-side).
  - During heightened global stress (higher VIX), investors may be less willing to subscribe Eurobonds (supply-side), pushing borrowers to switch to loans (demand-side).
  - Higher global interest rates reduce issuance of both instruments (demand-side).
  - Eurobonds are typically fixed-rate securities; borrowers may avoid locking higher rates, affecting issuance choices.

### Dominance analysis (importance of estimators)
- Key dominance results (Figure 7 summary):
  - Domestic factors explain more of Eurobond issuance variation than for loans.
    - Overall risk rating explains 73 percent of the variation in Eurobond issuance.
    - Overall risk rating explains 50 percent of the variation in loan issuance.
  - External factor differences:
    - USD index explains around 37 percent of variation for loans, but only 6 percent for Eurobonds.

### Likelihood of issuance: Predictive power
- Predictive assessment approach:
  - Use receiver operating characteristic (ROC) curve to evaluate ability of models to classify issuances (true positive rate vs. false positive rate).
- (Text ends at description of axes and example interpretation; detailed ROC statistics and figures referenced but not reproduced in the source excerpt.)

*Source: IMF staff analysis in "2. Eurobond Yields to Maturity at Issuance" from the IMF Working Paper chapter provided.*

### 0.75 and false positive rate 0.25 represents a threshold which predicts issuance when there is issuance in the

### wpiea2025139-print-pdf - 0.75 and false positive rate 0.25 represents a threshold which predicts issuance when there is issuance in the

### Predictive performance of issuance models (ROC / AUC)
- A threshold of 0.75 and false positive rate 0.25 "represents a threshold which predicts issuance when there is issuance in the data around 75 percent of the time and predicts an issuance when there is no issuance around 25 percent of the time."
- AUC interpretations:
  - AUC = 0.5: model is not informative (equivalent to tossing a coin; corresponds to the 45-degree line).
  - AUC > 0.5: informative model (lies above 45-degree line).
  - AUC = 1: perfect prediction.
- Empirical AUC results (Figure 8; probit regressions):
  - Eurobonds:
    - Model with only global factors (VIX, 3-month rate, USD index): AUC = 0.55; standard error = 0.008. Statistically different than 0.5 at the 1 percent significance level.
    - Full model (adds overall risk rating): AUC = 0.63; standard error = 0.007. Statistically significantly higher than the global-factors-only model at the 1 percent significance level.
    - Interpretation: global factors add predictive power; domestic factors (risk rating) further increase predictive power; full model AUC in an acceptable range though far from perfect.
  - Bank loans:
    - Global-factors-only model: AUC = 0.59; standard error = 0.007.
    - Full model (adds overall risk rating): AUC = 0.62; standard error = 0.006.
    - Interpretation: both domestic and global factors are important in predicting loan issuances.

### Likelihood of Eurobond and loan issuances in Sub-Saharan Africa (SSA)
- Main empirical patterns (Table 6; probit average marginal effects):
  - SSA countries are less likely to issue Eurobonds relative to other countries; this effect is mitigated when domestic conditions are sound (overall risk rating positive).
  - Global factors have a more pronounced role for SSA:
    - Higher VIX or tighter global financial conditions (higher US 3M rate) reduce the likelihood SSA countries issue Eurobonds.
  - For loans:
    - SSA countries are less likely to issue loans, but less so when domestic conditions are sound (similar to Eurobonds).
    - Role of global uncertainty (VIX) and US dollar index disappears for loans in SSA.
    - Lower US 3M rate implies a higher likelihood of loan issuance for SSA.
- Selected coefficient magnitudes from Table 6 (average marginal effects; standard errors in parentheses):
  - Overall risk rating: 0.0019*** (Model 1 Eurobond); 0.0017*** (Model 2 Eurobond); 0.0033*** (Model 3 Loan); 0.0028*** (Model 4 Loan).
  - VIX: -0.0012***, -0.0011*** (Eurobond models); 0.0009***, 0.0011*** (Loan models).
  - US 3M rate: -0.0040***, -0.0032*** (Eurobond models); -0.0022**, 0.0000 (Loan models).
  - USD index: 0.0007***, 0.0006*** (Eurobond); -0.0015***, -0.0019*** (Loan).
  - SSA dummy: -0.0719***, -0.1667*** (Eurobond); -0.0105***, -0.1993*** (Loan).
  - Interaction effects (examples): Overall risk rating x SSA = 0.0029***, 0.0023***; VIX x SSA = -0.0016**, -0.0012***; US 3M rate x SSA = -0.0092***, -0.0090***.
  - Pseudo R squared values: Eurobond models 0.059, 0.059; Loan models 0.025, 0.027.
  - Observations: 29,038; Countries: 102.
- Notes: Results based on probit model in eq. (4); robust standard errors in parentheses; significance markers: *** p<0.01, ** p<0.05, * p<0.1.

### Recent trends in local-currency debt markets in SSA
- Issuance volumes and maturities:
  - Total domestic-currency issuances across SSA (excluding maturities < one year) averaged $10.9bn per month in 2024 — more than double the amount issued in 2019.
  - Average issuance duration fell to around three and a half years in 2024, compared with an average of more than 6 years in 2019 (after excluding securities with durations < one year).
  - Shorter maturities increase rollover frequency and potential financing risks.
- Yield dynamics:
  - Average yields on domestic-currency debt in SSA were on a downward trend between 2017 and 2021; yields have risen globally since 2022 and SSA yields increased as well.
  - Shorter-maturity debt tends to have lower yields on average; the recent rise in yields occurred despite shorter maturities.
  - Since 2012, domestic-currency yields have been higher than Eurobond yields around 80 percent of the time (Figure 11).
- Cross-country yield dispersion in 2024:
  - Seychelles: lowest debt interest costs at 3.2 percent.
  - Sierra Leone: highest at 39.8 percent.
  - Between 2023 and 2024, nominal yields increased in 24 of 35 countries with available data (just under 70 percent).
  - Real interest rates (correcting for inflation) increased in 27 countries in 2024 (77 percent of countries with available data).

### Non-resident holdings: effects on yields (cross-country evidence)
- Data and sample:
  - Monthly data from January 2011 to September 2024.
  - Final sample: 24 emerging market economies with complete information.
  - Yield series: 5-year local-currency bond yields from Bloomberg and DataStream; foreign holdings from Haver, Bloomberg, national exchanges, and country authorities.
- Baseline result (Table 7 / text summary):
  - All emerging markets (Model 1): a ten-percentage point increase in non-resident ownership is associated with a 0.13 percentage point decrease in yields. Statistically significant at 1 percent.
  - SSA-specific effect (Model 2): for SSA, a ten-percentage point increase in non-resident holdings is associated with a 0.4 percentage point decrease in yields.
  - High-debt interaction (Model 3): for low-debt countries, a ten-percentage point increase in non-resident holdings is associated with a 0.09 percentage point decline in yields; for high-debt countries (public debt to GDP ratio above 60 percent), the same increase is associated with a 0.3 percentage point increase in yields.
- Table A14 / A16 magnitudes (selected coefficients; robust SEs in parentheses):
  - Non-resident Holdings (Model variations):
    - Model 1 (pooled): -0.1079*** (0.005) effect on 5-year T-bond yield.
    - Table A16 pooled models show nonresident_holdings coefficients ranging from -0.1119*** to -0.0096** depending on specification; later specifications controlling for more factors show coefficients like -0.0129*** (Model 10).
  - Controls commonly included: overall risk rating (negative significant effect on yields), domestic policy rate (positive significant effect), U.S. 5-year T-bond yield, dollar index, VIX.

### Non-resident holdings: effects on yield volatility
- Volatility measure: standard deviation of bond yields over a rolling window (nineteen months for monthly series; alternative daily measure uses 90-day rolling window).
- Pooled OLS yield-volatility result (Model 4 summary):
  - All emerging markets: a ten-percentage point increase in non-resident ownership is associated with a 0.02 percentage point increase in yield standard deviation. Statistically significant at 1 percent.
- SSA-specific volatility result (Model 5 summary):
  - On average, SSA countries have lower yield volatility than other EMDEs: standard deviation reduced by 0.3 percentage points.
  - For non-SSA countries: a ten-percentage point increase in non-resident holdings increases yield volatility by 0.05 percentage points.
  - For SSA countries: a ten-percentage point increase in non-resident holdings reduces volatility by 0.06 percentage points.
- High-debt cases and volatility:
  - For low-debt countries, higher non-resident ownership is associated with lower yield volatility (small and not always statistically significant).
  - For high-debt countries, higher non-resident holdings are typically associated with higher yield volatility — compatible with higher risk of capital flight.
- Robustness checks:
  - Results broadly robust to inclusion/exclusion of controls (including dollar index), country fixed effects (not heavily weighted due to small sample), and alternative volatility measure (daily 90-day rolling window). Using daily volatility, higher non-resident ownership was not found to reduce volatility in SSA (mixed result); monthly-volatility results considered most robust given data limitations.

### Additional empirical robustness and sample facts
- Domestic bond market summary (Table A3 selected stats):
  - 5-year local currency government bond yield (Overall sample): Obs. 3,337; Mean 7.93; Std. Dev. 5.05; Min 0.70; Max 20.07.
  - Yield volatility: Obs. 3,291; Mean 0.75; Std. Dev. 0.59; Min 0.00; Max 3.52.
  - Non-resident holdings (percent): Obs. 3,415; Mean 18.03; Std. Dev. 12.72; Min 0.00; Max 58.33.
  - Policy rate (percent): Obs. 3,585; Mean 7.21; Std. Dev. 5.42; Min 0.25; Max 50.00.
- Eurobond and loan facts:
  - Eurobond spreads and maturities (Table A12):
    - SSA: Maturity median 10; Mean 13.6; Spread median 441.5; Mean 433.6.
    - Non-SSA: Maturity median 10; Mean 12.7; Spread median 242.7; Mean 284.6.
  - Loan facts (Table A13):
    - SSA loan maturity median 8; mean 9.6; spread median 210; mean 247.8.
    - Non-SSA loan maturity median 8; mean 9.4; spread median 150; mean 189.7.

### Key policy-relevant findings and implications
- Both global factors (VIX, US 3M rate, USD index) and domestic factors (overall risk rating) materially influence issuance likelihood for Eurobonds and loans; combining both improves predictive power (AUC rises from ~0.55–0.59 to ~0.62–0.63).
- SSA faces an African risk premium in Eurobond markets that increases during crises; governance and institutional improvements can help reduce this premium and borrowing costs.
- Domestic-currency market development offers benefits:
  - Larger domestic issuance volumes ($10.9bn per month in 2024 across SSA, excluding <1-year maturities).
  - Attracting non-resident investors can lower yields and, for SSA in the sample, may also reduce yield volatility — but benefits vary by debt level.
- Trade-offs:
  - Increasing non-resident participation generally reduces yields but can increase yield volatility in many emerging markets; for high-debt countries, non-resident participation may raise yields and volatility (risk of capital flight).
  - Policymakers should weigh potential financing-cost reductions against volatility and rollover risks, especially in high-debt contexts.
- Recommended policy directions (implied by findings):
  - Strengthen governance, transparency, and institutional frameworks to improve risk ratings and attract stable foreign investors.
  - Develop domestic debt markets to diversify financing sources and reduce foreign-currency exposure while managing rollover and financial-stability risks where fiscal dominance or tight sovereign-bank nexus exists.
  - Implement debt-management strategies that consider interactions between non-resident participation, public debt levels, and potential volatility.

### Conclusion and research priorities
- The financing landscape in SSA is shifting: increased reliance on Eurobonds and domestic-currency issuance amid rising borrowing costs and elevated debt levels.
- Empirical evidence supports a nuanced role for non-resident investors: potential to lower yields and, in SSA, reduce volatility — but with adverse outcomes for high-debt countries.
- Future research priorities identified:
  - Impact of governance reforms on credit ratings and non-resident investment flows in SSA.
  - Mechanisms driving lower international bond issuances by SSA countries and implications for regional financing strategies.

*International Monetary Fund — Navigating the Evolving Landscape of External Financing in Sub-Saharan Africa (working paper content provided).*

### References

### wpiea2025139-print-pdf - References

### Studies on sovereign bond spreads, emerging markets, and credit risk
- Akıncı, Ö., 2013. “Global financial conditions, country spreads and macroeconomic fluctuations in emerging countries”, Journal of International Economics, 91(2), pp.358-371.
- Akinci, O. and Queralto, A., 2022. “Credit spreads, financial crises, and macroprudential policy”, American Economic Journal: Macroeconomics, 14(2), pp.469-507.
- Bellas, D., Papaioannou, M., and I. Petrova. 2010. “Determinants of Emerging Market Sovereign Bond Spreads: Fundamentals vs Financial Stress”, IMF Working Paper WP/10/281.
- Comelli, F., 2012. “Emerging market sovereign bond spreads: Estimation and back-testing”, Emerging Markets Review, 13(4), pp.598-625.
- Ebeke, C. and Lu, Y. 2014. “Emerging Market Local Currency Bond Yields and Foreign Holdings in the Post-Lehman Period—a Fortune or Misfortune?” IMF Working Paper WP/14/29.
- Gbohoui, W., R. Ouedraogo, and M. Some. 2023. “Sub-Saharan Africa’s Risk Perception Premium: In the Search of Missing Factors”, IMF Working Paper 2023/130, International Monetary Fund, Washington, DC.
- Morsy, H. and Moustafa, E., 2020. “Mispricing of Sovereign Risk and Investor Herding in African Debt Markets,” African Development Bank, WP No. 331.
- Nose, M. and Menkulasi, J. 2025. “Fiscal Determinants of Domestic Sovereign Bond Yields in Emerging Market and Developing Economies,” IMF Working paper WP/25/59.
- Olabisi, M. and Stein, H., 2015. “Sovereign bond issues: Do African countries pay more to borrow?” Journal of African Trade, 2(1), pp.87-109.
- Özmen, E. and Yaşar, Ö.D., 2016. “Emerging market sovereign bond spreads, credit ratings and global financial crisis,” Economic Modelling, 59, pp.93-101.
- Presbitero, A., D. Ghura, O. S. Adedeji, and L. Njie. 2016. “Sovereign bonds in developing countries: Drivers of issuance and spreads,” Review of Development Finance 6, no. 1: 1–15.
- Sy, A. N.R, 2002, “Emerging market bond spreads and sovereign credit ratings: reconciling market views with economic fundamentals,” Emerging Markets Review, Volume 3, Issue 4, Pages 380-408.
- Gueye, C.A. and Sy, A.N., 2015. “Beyond aid: how much should African countries pay to borrow?” Journal of African economies, 24(3), pp.352-366.
- Guelay? [Note: preserve original spelling as provided] — (No additional entries not present in source).

### Research on global financial conditions, currency and capital flows
- Bruno, V, I Shim and H S Shin. 2022. "Dollar beta in stock markets", Oxford Open Economics, vol 1, pp 1–10.
- Gelos, G. Patelli, P. and Shim, I. 2024. “The US dollar and capital flows to EMEs”, BIS Quarterly Review.
- Eichengreen, B. and Hausmann, R. 1999. “Exchange rates and financial fragility”, New Challenges for Monetary Policy. Jackson Hole Symposium.
- Onen, M. Shin, H. S. and von Peter, G. 2023. “Overcoming Original Sin: Insights from a New Dataset,” BIS Working Papers No. 1075.

### Policy, IMF, and multilateral reports relevant to financial stability and fiscal politics
- International Monetary Fund, 2024. “Steadying the Course: Uncertainty, Artificial Intelligence, and Financial Stability,” Global Financial Stability Report, October, Washington, DC.
- International Monetary Fund, 2025. “Enhancing Resilience amid Uncertainty,” Global Financial Stability Report, April, Washington, DC.
- Gaspar, V., S. Gupta, and C. Mulas-Granados. 2017. “Fiscal politics”, International Monetary Fund, Washington, DC.
- Kogan, J., Kazandjian, R., Luo, S., Mbohou Mama, M., and H. Miao. 2024. “The role of IMF arrangements in restoring access to international capital markets”, IMF Working Paper WP/24/173.
- Chami, R., Fullenkamp, C., & Sharma, S. 2010. “A framework for financial market development”, Journal of Economic Policy Reform, 13(2), pp.107-135.
- Fofack, H. 2021. “The Ruinous Price for Africa of Pernicious ‘Perception Premiums’.” AGI, Brookings Institution Report, October.

### Methodological and technical references
- Budescu, D. V., 1993. “Dominance analysis: A new approach to the problem of relative importance of predictors in multiple regression”, Psychological Bulletin, 114(3), 542–551.
- Luchman, J. N. 2021. “Determining relative importance in Stata using dominance analysis: domin and domme,” The Stata Journal, 21(2), 510-538. https://doi.org/10.1177/1536867X211025837 (Original work published 2021).
- De Fiore, F. and Uhlig, H., 2005. “Bank finance versus bond finance: what explains the differences between US and Europe?” (No. 547). ECB working paper.
- Hosny, A. 2020. “Non-Resident Holdings of Domestic Debt in Nigeria: Internal or External Driven?” IMF Working Paper WP/20/63.

### Research on ratings, market perception, and institutional factors
- Griffith-Jones, S., and M. Kraemer. 2021. “Credit rating agencies and developing economies.” UN/DESA Working Paper No. 175.
- Sy, A. N.R, 2002, “Emerging market bond spreads and sovereign credit ratings: reconciling market views with economic fundamentals,” Emerging Markets Review, Volume 3, Issue 4, Pages 380-408.
- Gbohoui, W., R. Ouedraogo, and M. Some. 2023. “Sub-Saharan Africa’s Risk Perception Premium: In the Search of Missing Factors”, IMF Working Paper 2023/130, International Monetary Fund, Washington, DC.

*References list for Working Paper No. WP/2025/139 — Navigating the Evolving Landscape of External Financing in Sub-Saharan Africa*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025139-print-pdf.pdf_
