## Sub-Saharan Africa’s risk perception premium: in the search of missing factors

## Source details

**Canonical URL:** [Sub-Saharan Africa’s risk perception premium: in the search of missing factors](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023130-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023130-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023130-print-pdf.pdf.json)

---

### I. Introduction — context and research question
- Borrowing from international financial markets by Sub-Saharan African (SSA) governments has grown exponentially over the past fifteen years; as of 2021, fifteen countries from the region have tapped international markets through the issuance of Eurobonds.
- Drivers of increased SSA access:
  - For global investors: high potential return, improved macroeconomic policies and development prospects, low global interest rates, increased global liquidity, portfolio diversification.
  - For SSA countries: Eurobonds free up cash, offer additional development financing without conditionalities amid drying up of concessional loans; used for infrastructure financing, domestic bond market benchmark, private sector access, and debt management/restructuring.
- Research questions:
  - Document heterogeneity in borrowing costs across regions at issuance and assess whether SSA countries are charged higher costs at issuance beyond credit ratings and bond characteristics.
  - Assess whether there is any bias against SSA countries in the secondary market that results in higher refinancing cost—not explained by country-specific characteristics and global risk factors.

### II. Data sources and stylized facts
- Primary market:
  - Bond Radar: total of 1592 international primary sovereign fixed coupon bonds issued in USD and Euro between 2003-2021 from 89 countries.
  - Dataset includes coupon rates, maturity date, issuance date, country risk ratings at issuance, and other bond characteristics.
  - Country effective risk ratings based on Fitch or S&P converted to numerical scale from 1 (AAA) to 23 (D).
  - Only bonds issued outside issuing country’s jurisdiction retained; financial centers and other advanced economies excluded.
- Secondary market:
  - Bloomberg: unbalanced panel of 107 countries from 1990 – 2022 at quarterly frequency.
  - Bond spread measured by JP Morgan’s Emerging Market Bond Index (EMBI) spread or CDS spread when EMBI not available.
- Controls and indices:
  - Quarterly International Country Risk Guide (ICRG) risk indices (political, financial, economic) composed of 22 components; higher values reflect lower risks.
  - Global variables: VIX index and US federal funds rate.
  - Other country-level annual data: fiscal crises, diversification index, size of informal sector, transparency, financial development.
- Stylized facts:
  - Between 2006-2021, 15 SSA countries issued fixed coupon international sovereign bonds; most SSA countries had not issued such bonds prior to 2010 except Gabon, Ghana, Seychelles, South Africa.
  - Yearly SSA issuance: below USD 5 billion before 2013; picked up from 2014 with record issuances in 2018 (USD 19.2 billion) and 2019 (USD 15.5 billion); 2020 fell to USD 5 billion and bounced back in 2021 to USD 13.5 billion.
  - Maturities concentrated around 10-year; some recent long maturities include 30-year and 40-year issues.
  - Ratings patterns:
    - Except South Africa and Namibia, none of SSA issuers had an investment grade (IG) rating at issuance; since 2019 all SSA sovereign foreign currency bonds in sample are rated HY by all rating agencies.
    - Between 2014-2021, around 85 percent of SSA issuances are rated HY compared with 38 percent for EMDEs from other regions.
  - Sovereign coupon and spread patterns:
    - On average between 2004 and 2021, SSA countries paid around 2.1 percentage points higher coupon rate than countries from other regions.
    - Across all maturities, average coupon rate for SSA issuances is persistently higher by 1.3 percentage points per year than the average coupon rate of EMDE countries from other regions between 2014 and 2021.
    - Between 2014–2021, SSA countries borrowed in the Eurobond market at higher cost than peers from other regions by 155 basis points on average (simple average, 4-quarter moving average shown in Figure 1.D).

### III. Empirical strategy
- Primary market (bond-level) regression:
  - lnCoupon_it = c + α Z_it + γ SSA + π_t + ν_it
  - lnCoupon_it: natural log of coupon rate at issuance; Z_it includes sovereign rating, maturity, volume; π_t: time fixed effects; SSA: dummy = 1 if country is SSA. γ captures potential market bias toward SSA.
- Secondary market (country-level) regressions:
  - lnSpread_it = β X_it + π_t + μ_i + ε_it
  - Augmented: lnSpread_it = ω X_it + ϑ SSA + π_t + ε_it
  - lnSpread_it: natural log of sovereign bond spreads; X_it: country controls (economic, financial, political risk ratings, history of defaults); global controls: VIX and US federal funds rate; μ_i: country fixed effects; π_t: time fixed effects.
  - Time-invariant SSA dummy identified via Pesaran and Zhou (2018) two-step approach (FE without SSA, then regress residuals on SSA dummy) when needed.

### IV. Results — Primary market (bond-level)
- Core findings:
  - SSA countries incurred higher cost to issue Eurobonds over 2004-2021.
  - Unconditional: average coupon rate for SSA countries was around 66 percent higher than the average coupon rate of countries from other regions (Column (1), Table 1).
  - After controlling for rating, maturity, volume and other bond characteristics, SSA countries still pay significantly higher coupon (Columns (3)-(5), Table 1).
- Interaction effects and selected coefficients (dependent variable ln(coupon), selected estimates from Table 1):
  - SSA: 0.506*** (Column (1)), 1.155*** (Column (2)), 0.234*** (Column (3)), 0.286*** (Column (4)), 0.215*** (Column (5))
  - APD: 0.556*** (Column (2))
  - EUR, developing: 0.666*** (Column (2))
  - LAC: 0.865*** (Column (2))
  - MCD: 0.797*** (Column (2))
  - High Yield: 0.714*** (Column (1)), 0.664*** (Column (2)), 0.422*** (Column (3)), 0.419*** (Column (4))
  - Not Rated: 0.509*** (Column (1)), 0.189*** (Column (2))
  - Rating*SSA: 0.084*** (Column (6))
  - Rating*High Yield: 0.179*** (Column (7))
  - Rating, Log: 0.456*** (Column (1)), 0.457*** (Column (2)), 0.444*** (Column (3))
  - Maturity, Log: 0.211*** (Columns shown)
  - Volume, Log: 0.064*** (Columns shown)
  - Sinkable: 0.203** (Column (1)), 0.196** (Column (2)), 0.303*** (Column (3))
  - Callable: 0.133*** (Column (1)), 0.134*** (Column (2)), 0.115*** (Column (3))
  - Observations: 1592 (various columns show 1592, 1448)
  - Adjusted R-squared ranges: 0.035, 0.184, 0.275, 0.448, 0.541, 0.541, 0.540
- Interpretation:
  - A downgrade of sovereign rating tends to increase coupon rates; Rating*SSA significant implies for the same magnitude of downgrade, SSA coupon rates increase more relative to non-SSA peers.
  - Downgrades for HY bonds lead to higher increases in borrowing cost relative to downgrades of IG bonds.
  - Higher borrowing costs for SSA might partly reflect lower bond liquidity; liquidity measures at issuance unavailable in the bond-level data.

### V. Results — Secondary market (country-level) baseline
- Baseline coefficients (Table 2, column (1), All; standard errors in parentheses):
  - Economic risk, Log: -0.7149*** (0.231)
  - Financial risk, Log: -1.9522*** (0.313)
  - Political risk, Log: -2.6734*** (0.388)
  - US Federal Fund rate: 0.1859*** (0.047)
  - VIX, Log: 0.3490*** (0.109)
  - Debt crisis, Lagged: 0.2446*** (0.046)
  - SSA (binary): 0.5109*** (0.130)
  - Constant: 24.6688*** (1.999)
- Sample coverage and fit:
  - Observations: 4,508 (All), 3,969 (Non-SSA countries), 3,082 (Non-SSA emerging countries), 539 (SSA countries).
  - Number of countries: 87 (All), 74 (Non-SSA), 53 (Non-SSA emerging), 13 (SSA).
  - R-squared: 0.6289 (All), 0.6265 (Non-SSA), 0.6096 (Non-SSA emerging), 0.5883 (SSA).
- Sub-sample contrasts (SSA countries, column (4)):
  - Economic risk, Log: 0.7874 (0.851) — not statistically significant
  - Financial risk, Log: -2.9287*** (0.967)
  - Political risk, Log: -1.5812** (0.762)
  - US Federal Fund rate: -0.1446 (0.097) — not statistically significant
  - VIX, Log: -0.3244 (0.249) — not statistically significant
  - Debt crisis, Lagged: 0.3964** (0.172)
  - Constant: 21.8418*** (4.951)
- Interpretation:
  - Investors are more sensitive to political and financial developments than to economic fundamentals: financial and political risk ratings have larger effects than economic risk rating.
  - An increase of the financial risk rating by 1 percent associated with a reduction of spreads by around 2 percent; political risk by 1 percent associated with reduction of spreads by around 2.7 percent; economic risk by 1 percent associated with reduction of spreads by around 0.7 percent.
  - Global factors: VIX and US Federal Funds rate coefficients positive and highly significant; an increase of the VIX by 1 percent correlated with an increase of spreads by around 0.3 percent; an increase of the US Federal Funds rate by 1 percent correlated with an increase of spreads by around 0.3 percent.
  - Being an SSA country is associated with an increase of 0.5 percent in sovereign bond spreads (SSA coefficient 0.5109***), but sensitivity to specific factors differs for SSA sub-sample.

### VI. Decomposition of SSA premium: structural “missing factors”
- Strategy: include structural variables (financial development, Open Budget Index, informal sector size, regulatory quality) and test SSA interactions.
- Financial development (Table 3, column (2)):
  - Financial development: -1.4030*** (0.469)
  - Financial development * SSA: -1.2080 (2.644) — not significant
  - SSA: 0.3921 (0.468) — not significant
  - Interpretation: controlling for financial development removes significance of SSA dummy; stronger financial development associated with lower sovereign spreads.
- Budget transparency (Open Budget Index, column (3)):
  - Open budget index: -0.0096* (0.005)
  - Open budget index * SSA: 0.0086 (0.009) — not significant
  - SSA: 0.0426 (0.435) — not significant
  - Interpretation: greater fiscal transparency associated with lower sovereign spreads (weakly significant); SSA premium vanishes when controlled.
- Informal sector (column (4)):
  - Informal sector: 0.0239*** (0.006)
  - Informal sector * SSA: 0.0186 (0.019) — not significant
  - SSA: -0.6955 (0.727) — not significant
  - Interpretation: larger informal sector positively and strongly associated with higher sovereign spreads; controlling for informality removes SSA premium.
- Regulatory quality (column (5)):
  - Regulatory Quality: -0.6475*** (0.129)
  - Regulatory Quality * SSA: -0.5647 (0.352) — not significant
  - SSA: -0.1253 (0.245) — not significant
  - Interpretation: better regulatory quality associated with lower sovereign spreads; SSA premium not significant when institutions included.
- Full model (column (6), all structural controls):
  - SSA: -1.8181 (1.325) — not statistically significant
  - Conclusion: the previously estimated SSA premium vanishes when financial development, budget transparency, informal sector size, and regulatory quality are jointly accounted for.

### VII. Robustness checks and alternative specifications — selected evidence
- Excluding outliers (5% and 10%), excluding South Africa and Nigeria, excluding advanced countries — SSA coefficient remains positive and in several specifications significant:
  - SSA: 0.4662*** (0.117) — Excluding outliers (5%)
  - SSA: 0.3768*** (0.111) — Excluding outliers (10%)
  - SSA: 0.5966*** (0.122) — Excluding South Africa and Nigeria
  - SSA: 0.2325** (0.118) — Excluding advanced countries
- Alternative macro-fundamentals (Table 5):
  - GDP growth: coefficients range from -0.0097** (0.004) to -0.0215** (0.009)
  - Inflation: 0.0302** (0.015) in column (1); 0.0363** (0.017) in column (7)
  - Reserves, % of GDP: -0.0137** (0.006) in columns (1)-(2); -0.0187** (0.008) in column (7)
  - Debt, % of GDP, Lagged: 0.0230*** (0.004) to 0.0318*** (0.006)
  - US Federal Fund rate: 0.1921*** (0.040) in column (1); 1.5031*** (0.280) in column (6) (sample-specific large estimate)
  - SSA: 1.5033*** (0.384) in column (1) of Table 5; SSA becomes not statistically significant in columns 3-6 of Table 5.
  - Financial development index: -3.8887*** (1.022) where included.
  - Regulatory Quality: -1.0572*** (0.175) where included.
- Alternative risk indices (EIU-based, Table 6):
  - Currency risk: -2.6512*** (0.676)
  - Economic structure risk: -1.9325*** (0.614)
  - VIX, Log: 1.9447*** (0.515)
  - SSA: 0.3138** (0.154) in column (1) — significant; SSA becomes insignificant when structural factors are added.
- Two-step empirical strategy (Table 7):
  - First-step coefficients consistent with baseline.
  - Second-step (residuals on time-invariant vars): SSA: 0.5023*** (0.128) when structural controls absent; SSA coefficients not statistically significant once financial development, budget transparency, informality, and regulatory quality enter.

### VIII. Synthesis of findings
- Baseline models measure a positive and significant SSA perception premium (e.g., SSA = 0.5109*** in Table 2, column (1)).
- The SSA premium is larger than for other emerging markets in many specifications, though other EMDEs also show risk premia relative to advanced economies.
- The SSA premium vanishes when structural factors are jointly controlled:
  - Financial development: negative association with spreads (e.g., -1.4030***)
  - Regulatory quality: negative association (e.g., -0.6475***)
  - Open Budget Index (fiscal transparency): negative (e.g., -0.0096*)
  - Informal sector: positive association (e.g., 0.0239***)
- Results robust across samples, macro controls, alternative risk indices, and the two-step approach.

### IX. Policy implications and recommendations
- Strengthen sovereign risk management to ensure financing needs are met at the lowest possible cost given a certain level of risk.
- Targeted reforms indicated by empirical results:
  - Improve transparency of the budget process to increase quality of publicly available fiscal data and investor confidence.
  - Strengthen the quality of public institutions to enhance ability to formulate and implement sound macroeconomic policies.
  - Create a business environment conducive to private sector development and reduce informality (smaller informal sector increases tax base and de-facto fiscal space).
  - Deepen domestic financial markets to reduce bond illiquidity risk and lower premia charged by investors.
- These reforms address the “missing factors” that empirically explain the apparent SSA risk premium and can reduce borrowing costs both at issuance and in the secondary market.

### X. Key statistics and magnitudes (selected)
- Bond-level sample: 1592 international primary sovereign fixed coupon bonds (2003–2021) from 89 countries.
- Average coupon differentials:
  - SSA paid around 2.1 percentage points higher coupon rate on average between 2004 and 2021 compared with other regions.
  - Average coupon rate for SSA issuances persistently higher by 1.3 percentage points per year than EMDEs from other regions between 2014 and 2021.
- Secondary-market baseline (Table 2, All):
  - SSA coefficient: 0.5109*** (0.130)
  - Economic risk, Log: -0.7149*** (0.231)
  - Financial risk, Log: -1.9522*** (0.313)
  - Political risk, Log: -2.6734*** (0.388)
  - US Federal Fund rate: 0.1859*** (0.047)
  - VIX, Log: 0.3490*** (0.109)
  - Debt crisis, Lagged: 0.2446*** (0.046)
- Structural factor estimates (Table 3, selected):
  - Financial development: -1.4030*** (0.469)
  - Open budget index: -0.0096* (0.005)
  - Informal sector: 0.0239*** (0.006)
  - Regulatory Quality: -0.6475*** (0.129)
- Robustness examples:
  - Excluding outliers (5%): SSA = 0.4662*** (0.117)
  - Excluding South Africa and Nigeria: SSA = 0.5966*** (0.122)
  - Two-step residual second-stage: SSA = 0.5023*** (0.128) before structural controls; SSA not significant after including structural controls.

*International Monetary Fund — IMF Working Papers: “Sub-Saharan Africa’s risk perception premium: in the search of missing factors” (excerpts and results summarized from the provided content).*

### References .................................................................................................. 31

### Sub-Saharan Africa’s risk perception premium: in the search of missing factors

### I. Introduction — context and research question
- Borrowing from international financial markets by Sub-Saharan African (SSA) governments has grown exponentially over the past fifteen years; as of 2021, fifteen countries from the region have tapped international markets through the issuance of Eurobonds.
- Drivers of increased SSA access to international markets:
  - For global investors: high potential return, improved macroeconomic policies and development prospects, low global interest rates, increased global liquidity, portfolio diversification.
  - For SSA countries: Eurobonds free up cash, offer additional development financing without conditionalities amid drying up of concessional loans; used for infrastructure financing, domestic bond market benchmark, private sector access, and debt management/restructuring.
- Perceived problem: policymakers in SSA flag a mispricing / “risk perception premium” that leads to “unjustifiably” high borrowing costs compared with peers.
  - Example comparisons highlighted in the text:
    - Argentina (defaulted nine times) issued a 100-year bond in 2017 with a coupon of 7 percent and was oversubscribed; Angola (no default since 2002) paid over 9 percent for a 30-year bond issued in 2018 and its 10-year Eurobond issued in 2015 carried a yield of 9.5 percent.
- Two-fold empirical approach of the paper:
  1. Document heterogeneity in borrowing costs across regions at issuance and assess whether SSA countries are charged higher costs at issuance beyond credit ratings and bond characteristics.
  2. Assess whether there is any bias against SSA countries in the secondary market that results in higher refinancing cost—not explained by country-specific characteristics and global risk factors.

### II. Literature review — gaps and hypothesized “missing factors”
- Sovereign bond yields and CDS are widely used measures of sovereign risk premia; determinants studied include domestic macro fundamentals, liquidity and solvency indicators, institutions and policies, and global factors (risk appetite, global liquidity, contagion).
- Existing work on SSA is limited and inconclusive:
  - Gueye and Sy (2015): SSA country bonds overpriced; actual bond spread is 338 basis points below what is implied by fundamentals (reported in text).
  - Olabisi and Stein (2015): governments in SSA pay about 2.9% points more to borrow compared to other countries after controls.
  - Morsy and Moustafa (2020): find clustering/herding and mispricing mainly due to discriminatory investor behavior.
- Potentially overlooked factors that could explain SSA premium:
  - Data transparency and fiscal transparency (Choi and Hashimoto (2018); Kemoe and Zhan (2018)).
  - Degree of financial development and domestic bond market depth/liquidity.
  - Size of the informal sector (Aizenman and Jinjarak, 2012).
  - Quality of public institutions (Chen and Chen (2018); Huang et al. (2019)).
- Paper’s contribution: bond-level analysis using sovereign fixed coupon Eurobond characteristics and ratings at issuance (Bond Radar); first systematic exploration of structural drivers (budget transparency, informality, financial development, institutions) of an apparent SSA premium.

### III. Data sources and stylized facts
- Primary market data:
  - Bond Radar: total of 1592 international primary sovereign fixed coupon bonds issued in USD and Euro between 2003-2021 from 89 countries.
  - Dataset includes coupon rates, maturity date, issuance date, country risk ratings at issuance, and other bond characteristics.
  - Country effective risk ratings based on Fitch or S&P converted to numerical scale from 1 (AAA) to 23 (D).
  - Only bonds issued outside issuing country’s jurisdiction retained; financial centers and other advanced economies excluded.
- Secondary market data:
  - Bloomberg: unbalanced panel of 107 countries from 1990 – 2022 at quarterly frequency.
  - Bond spread measured by JP Morgan’s Emerging Market Bond Index (EMBI) spread or CDS spread when EMBI not available.
- Control variables and indices:
  - Baseline uses quarterly International Country Risk Guide (ICRG) risk indices (political, financial, economic) composed of 22 components; higher values reflect lower risks.
  - Global variables: VIX index and US federal funds rate.
  - Other country-specific annual data (interpolated) from IMF and World Bank: fiscal crises (Medas et al. (2018)), diversification index, size of informal sector, transparency, financial development.
- Stylized facts:
  - Between 2006-2021, 15 SSA countries issued fixed coupon international sovereign bonds (Euro and USD); most SSA countries had not issued such bonds prior to 2010 except Gabon, Ghana, Seychelles, South Africa.
  - Yearly SSA issuance amounts were below USD 5 billion before 2013, picked up from 2014 with record issuances in 2018 (USD 19.2 billion) and 2019 (USD 15.5 billion); 2020 fell to USD 5 billion and bounced back in 2021 to USD 13.5 billion.
  - Maturities concentrated around 10-year; some recent long maturities include 30-year and 40-year issues by certain SSA countries.
  - Sovereign rating patterns:
    - Except South Africa and Namibia, none of SSA issuers had an investment grade (IG) rating at issuance; since 2019 all SSA sovereign foreign currency bonds in sample are rated HY by all rating agencies.
    - Between 2014-2021, around 85 percent of SSA issuances are rated HY compared with 38 percent for EMDEs from other regions.
    - Among HY SSA issuers, best rating at issuance observed: BB- (Cote d’Ivoire 2017-2021, Gabon 2013, Ghana 2015, Nigeria 2011, 2013, Senegal 2017-2021); worst rating B- (Gabon in 2021).
    - Between 2010-2021, rating changes: 3 countries improved (Cote d’Ivoire, Senegal, Rwanda); 5 countries downgraded (Angola, Gabon, Ghana, Nigeria, Zambia).
  - Sovereign rating and coupon rates:
    - On average between 2004 and 2021, SSA countries paid around 2.1 percentage points higher coupon rate than countries from other regions.
    - Across all maturities, average coupon rate for SSA issuances is persistently higher by 1.3 percentage points per year than the average coupon rate of EMDE countries from other regions between 2014 and 2021.
  - Sovereign spread:
    - Between 2014–2021, SSA countries borrowed in the Eurobond market at higher cost than peers from other regions by 155 basis points on average (simple average, 4-quarter moving average shown in Figure 1.D).

### IV. Empirical strategy — models estimated
- Primary market (bond-level) regression (equation (1)):
  - lnCoupon_it = c + α Z_it + γ SSA + π_t + ν_it
  - lnCoupon_it: natural log of coupon rate of country i at issuance t.
  - Z_it: bond characteristics including sovereign rating, maturity and volume.
  - π_t: time fixed effects; SSA: dummy = 1 if country is SSA.
  - Coefficient γ captures any potential market bias toward SSA.
- Secondary market (country-level) regression (equation (2) and (3)):
  - lnSpread_it = β X_it + π_t + μ_i + ε_it
  - lnSpread_it: natural log of sovereign bond spreads for country i at time t.
  - X_it: country controls — economic, financial, political risk ratings, history of defaults; global controls: VIX and US federal funds rate.
  - μ_i: country fixed effects; π_t: time fixed effects.
  - Augmented specification includes SSA dummy:
    - lnSpread_it = ω X_it + ϑ SSA + π_t + ε_it
  - Estimation considerations:
    - Country fixed effects identify time-varying coefficients in X but render time-invariant SSA dummy unidentified; robustness uses Pesaran and Zhou (2018) two-step approach (FE without SSA, then regress residuals on SSA dummy).

### V. Results
- A. Bond-level (Primary market) analysis — key findings
  - SSA countries incurred higher cost to issue Eurobonds over 2004-2021.
  - Unconditional result: Column (1) of Table 1 indicates the average coupon rate for SSA countries was around 66 percent higher than the average coupon rate of countries from other regions.
  - Even after controlling for rating, maturity, volume and other bond characteristics, SSA countries still pay significantly higher coupon (Columns (3)-(5) in Table 1).
  - Interaction effects:
    - A downgrade of sovereign rating tends to increase coupon rates (Column (5)).
    - The coefficient of Rating*SSA is significant (Column (6)), implying for the same magnitude of downgrade, SSA country coupon rates increase more relative to non-SSA peers.
    - Column (7) suggests downgrades for HY bonds lead to higher increases in borrowing cost relative to downgrades of IG bonds.
  - Key numerical estimates from Table 1 (selected coefficients and notes — dependent variable ln(coupon)):
    - SSA: 0.506*** (Column (1)), 1.155*** (Column (2)), 0.234*** (Column (3)), 0.286*** (Column (4)), 0.215*** (Column (5))
    - APD: 0.556*** (Column (2))
    - EUR, developing: 0.666*** (Column (2))
    - LAC: 0.865*** (Column (2))
    - MCD: 0.797*** (Column (2))
    - High Yield: 0.714*** (Column (1)), 0.664*** (Column (2)), 0.422*** (Column (3)), 0.419*** (Column (4))
    - Not Rated: 0.509*** (Column (1)), 0.189*** (Column (2))
    - Rating*SSA: 0.084*** (Column (6))
    - Rating*High Yield: 0.179*** (Column (7))
    - Rating, Log: 0.456*** (Column (1)), 0.457*** (Column (2)), 0.444*** (Column (3))
    - Maturity, Log: 0.211*** (Columns shown)
    - Volume, Log: 0.064*** (Columns shown)
    - Sinkable: 0.203** (Column (1)), 0.196** (Column (2)), 0.303*** (Column (3))
    - Callable: 0.133*** (Column (1)), 0.134*** (Column (2)), 0.115*** (Column (3))
    - Observations: 1592 (various columns show 1592, 1448)
    - Adjusted R-squared ranges displayed: 0.035, 0.184, 0.275, 0.448, 0.541, 0.541, 0.540
  - Note on interpretation: higher borrowing costs for SSA might partly reflect lower bond liquidity; liquidity measures at issuance unavailable. Secondary market analysis controls for financial development correlated with bond liquidity.
- B. Country-level (Secondary market) analysis — baseline findings
  - Table 2 (summary in text):
    - Economic, financial and political ICRG risk ratings: coefficients negative and strongly significant — stronger fundamentals associated with lower sovereign spreads.
    - Financial and political risk ratings have larger effects than economic risk ratings:
      - An increase of the financial risk rating by 1 percent associated with a reduction of spreads by around 2 percent.
      - An increase of the political risk rating by 1 percent associated with a reduction of spreads by around 2.7 percent.
      - An increase of the economic risk rating by 1 percent associated with a reduction of spreads by around 0.7 percent.
    - Debt crisis history: positive and strongly significant — experiencing debt crisis in the past associated with an increase of the spreads by 0.2 percent.
    - Global factors:
      - VIX index and US Federal Funds rate coefficients positive and highly significant at the 1 percent level.
      - An increase of the VIX index by 1 percent is correlated with an increase of the spreads by around 0.3 percent.
      - An increase of the US Federal Funds rate by 1 percent is correlated with an increase of the spreads by around 0.3 percent (text states both increase spreads by around 0.3 percent; sentence cut in source).
  - Overall interpretation: investors are more sensitive to political and financial developments than to economic fundamentals; global risk aversion and tighter US monetary policy increase sovereign spreads.

- C. Decomposition of SSA premium and role of structural factors
  - The paper explores structural dimensions where SSA faces acute challenges: budget transparency, size of informal sector, level of financial development, quality of public institutions.
  - Key empirical finding (text summary):
    - The excess premium estimated for SSA countries vanishes when these structural factors are accounted for in the regression.
    - Results are robust to several robustness checks.
  - Policy-relevant implication drawn in the paper:
    - Reforms to improve the transparency of the budget process, strengthen the quality of public institutions, create a business environment conducive to private sector development and reduce informality, and measures to deepen domestic financial markets would help ensure that governments’ financing needs are met at the lowest possible cost.

### VI. Policy implications and recommendations (as presented in the paper)
- Strengthen sovereign risk management to ensure financing needs are met at the lowest possible cost given a certain level of risk.
- Targeted reforms indicated by empirical results:
  - Improve transparency of the budget process to increase quality of publicly available fiscal data and investor confidence.
  - Strengthen the quality of public institutions to enhance ability to formulate and implement sound macroeconomic policies.
  - Create a business environment conducive to private sector development and reduce informality (smaller informal sector increases tax base and de-facto fiscal space).
  - Deepen domestic financial markets to reduce bond illiquidity risk and lower premia charged by investors.
- These reforms address the “missing factors” that empirically explain the apparent SSA risk premium and can reduce borrowing costs both at issuance and in the secondary market.

*International Monetary Fund — IMF Working Papers: “Sub-Saharan Africa’s risk perception premium: in the search of missing factors” (excerpts and results summarized from the provided content).*

### 0.2 percent, respectively.

### Sub-Saharan Africa’s risk perception premium: in the search of missing factors

### Baseline results (Table 2) — core coefficients and sample splits
- Sample coverage and splits:
  - Observations: 4,508 (All), 3,969 (Non-SSA countries), 3,082 (Non-SSA emerging countries), 539 (SSA countries).
  - Number of countries: 87 (All), 74 (Non-SSA), 53 (Non-SSA emerging), 13 (SSA).
  - R-squared: 0.6289 (All), 0.6265 (Non-SSA), 0.6096 (Non-SSA emerging), 0.5883 (SSA).
- Key coefficient estimates (column (1), All):
  - Economic risk, Log: -0.7149*** (0.231)
  - Financial risk, Log: -1.9522*** (0.313)
  - Political risk, Log: -2.6734*** (0.388)
  - US Federal Fund rate: 0.1859*** (0.047)
  - VIX, Log: 0.3490*** (0.109)
  - Debt crisis, Lagged: 0.2446*** (0.046)
  - SSA (binary): 0.5109*** (0.130)
  - Constant: 24.6688*** (1.999)
- Sub-sample contrasts (column (4), SSA countries):
  - Economic risk, Log: 0.7874 (0.851) — not statistically significant
  - Financial risk, Log: -2.9287*** (0.967) — larger magnitude than in non-SSA samples
  - Political risk, Log: -1.5812** (0.762) — lower magnitude than in rest of world
  - US Federal Fund rate: -0.1446 (0.097) — not statistically significant
  - VIX, Log: -0.3244 (0.249) — not statistically significant
  - Debt crisis, Lagged: 0.3964** (0.172)
  - Constant: 21.8418*** (4.951)
- Interpretation:
  - Investors in SSA appear less sensitive to economic risk and global factors (economic risk, US Federal Fund rate, VIX) but more sensitive to financial risk and sovereign debt crisis risk.
  - Being an SSA country is associated with an increase of 0.5 percent in sovereign bond spreads (SSA coefficient 0.5109***).

### Evidence of a region-specific premium and time dynamics (Figure 2)
- Main patterns from the difference between actual and estimated bond spreads (actual − estimated, Log):
  - Both SSA and non-SSA emerging countries exhibit a positive risk premium relative to estimated spreads; advanced countries show actual spreads lower than estimated.
  - The risk premium is higher for SSA countries than for other emerging countries except during the global financial crisis period 2007-11, when SSA’s premium dropped significantly.

### Structural “missing factors” (Table 3) — testing candidate explanations
- Approach: include each structural factor and its interaction with SSA to test whether the SSA premium persists.
- Financial development (column (2)):
  - Financial development: -1.4030*** (0.469)
  - Financial development index * SSA: -1.2080 (2.644) — not significant
  - SSA: 0.3921 (0.468) — not significant
  - Interpretation: once financial development is controlled for, SSA dummy is no longer significant; stronger financial development is associated with lower sovereign spreads.
- Budget transparency (Open Budget Index, column (3)):
  - Open budget index: -0.0096* (0.005)
  - Open budget index * SSA: 0.0086 (0.009) — not significant
  - SSA: 0.0426 (0.435) — not significant
  - Interpretation: greater fiscal transparency is associated with lower sovereign spreads (weakly significant at 10%), and SSA premium vanishes when controlled.
- Informal sector (column (4)):
  - Informal sector: 0.0239*** (0.006)
  - Informal sector * SSA: 0.0186 (0.019) — not significant
  - SSA: -0.6955 (0.727) — not significant
  - Interpretation: larger informal sector is positively and strongly associated with higher sovereign spreads; controlling for informality removes the SSA premium.
- Regulatory quality (column (5)):
  - Regulatory Quality: -0.6475*** (0.129)
  - Regulatory Quality * SSA: -0.5647 (0.352) — not significant
  - SSA: -0.1253 (0.245) — not significant
  - Interpretation: better regulatory quality is associated with lower sovereign spreads; SSA premium not significant when institutions included.
- Full model (column (6), all structural controls included):
  - SSA: -1.8181 (1.325) — not statistically significant
  - Conclusion: the previously estimated SSA premium vanishes when financial development, budget transparency, informal sector size, and regulatory quality are jointly accounted for.

### Robustness checks and alternative specifications
- Alternative samples (Table 4):
  - Excluding outliers (top/bottom 5% and 10%), excluding South Africa and Nigeria, excluding advanced countries — SSA coefficient remains positive and in several specifications significant:
    - SSA: 0.4662*** (0.117) — Excluding outliers (5%)
    - SSA: 0.3768*** (0.111) — Excluding outliers (10%)
    - SSA: 0.5966*** (0.122) — Excluding South Africa and Nigeria
    - SSA: 0.2325** (0.118) — Excluding advanced countries
- Alternative macro-fundamentals (Table 5):
  - Key macro coefficients (selected):
    - GDP growth: coefficients range from -0.0097** (0.004) to -0.0215** (0.009) across columns
    - Inflation: e.g., 0.0302** (0.015) in column (1); 0.0363** (0.017) in column (7)
    - Reserves, in % of GDP: -0.0137** (0.006) in columns (1)-(2); -0.0187** (0.008) in column (7)
    - Debt, in % of GDP, Lagged: 0.0230*** (0.004) to 0.0318*** (0.006)
    - US Federal Fund rate: 0.1921*** (0.040) in column (1); 1.5031*** (0.280) in column (6) (sample-specific large estimate)
  - SSA binary in these specifications:
    - SSA: 1.5033*** (0.384) in column (1) of Table 5 (positive and strongly significant)
    - SSA becomes not statistically significant in columns 3-6 of Table 5.
  - Financial development and regulatory quality remain strongly negative (lower spreads) where included:
    - Financial development index: -3.8887*** (1.022) in column (1) where included
    - Regulatory Quality: -1.0572*** (0.175) in column (1) where included
- Alternative risk indices (EIU-based, Table 6):
  - Currency risk: -2.6512*** (0.676) in column (1)
  - Economic structure risk: -1.9325*** (0.614) in column (1)
  - VIX, Log: 1.9447*** (0.515) in column (1)
  - SSA: 0.3138** (0.154) in column (1) — significant and positive using EIU risk variables; SSA becomes insignificant when interacting structural factors are added.
- Alternative empirical strategy (two-step approach, Table 7):
  - First-step coefficients remain consistent with baseline (economic, financial, political risks, US rate, VIX, debt crisis significant with expected signs).
  - Second-step (residuals regressed on time-invariant variables):
    - SSA: 0.5023*** (0.128) in column (1) — confirms baseline SSA premium when structural controls absent.
    - SSA coefficients are not statistically significant in columns (2)–(6) once financial development, budget transparency, informality, and regulatory quality enter the analysis.

### Synthesis of findings
- There is an empirically measured perception premium for SSA countries when using traditional specifications: SSA dummy is positive and significant in baseline models (e.g., 0.5109***).
- This SSA premium is not unique to SSA in the sense that other emerging markets also display risk premia relative to advanced economies, but the SSA premium is larger in many specifications and periods.
- The SSA premium vanishes once structural factors are controlled for jointly:
  - Financial development: negative association with spreads (e.g., -1.4030***)
  - Regulatory quality: negative association with spreads (e.g., -0.6475***)
  - Open Budget Index (fiscal transparency): negative (e.g., -0.0096*)
  - Informal sector: positive association with spreads (e.g., 0.0239***)
- Robustness: results hold across multiple robustness checks (alternative samples, alternative macro controls, alternative risk indices, two-step empirical strategy), with the consistent message that structural factors explain the SSA premium when included.

### Policy implications and recommendations (drawn from empirical conclusions)
- Structural reforms that reduce the empirically observed drivers of higher spreads could lower SSA borrowing costs:
  - Develop and deepen domestic financial markets and liquidity (financial development).
  - Improve transparency of the budget process (Open Budget Index improvements).
  - Strengthen the quality of public institutions and regulatory frameworks (Regulatory Quality).
  - Reduce the size and economic relevance of the informal sector.
- Meeting these structural challenges could reduce the perceived risk premium and unlock more sustainable and lower-cost financing for SSA countries’ development needs.

*Source: IMF Working Paper — Sub-Saharan Africa’s risk perception premium: in the search of missing factors (excerpts and tables from the supplied PDF content).*

### Annex 1. Robustness check: excluding outliers (5 percent of observations)

### Annex 1. Robustness check: excluding outliers (5 percent of observations)

### Key regression results (columns (1)–(6))
- Economic risk, Log
  - (1) -0.8968*** (0.245)
  - (2) -0.5930*** (0.186)
  - (3) -1.0671*** (0.285)
  - (4) -0.8630*** (0.253)
  - (5) -0.9471*** (0.250)
  - (6) -0.7986*** (0.274)
- Financial risk, Log
  - (1) -1.3462*** (0.279)
  - (2) -1.5393*** (0.294)
  - (3) -1.7415*** (0.332)
  - (4) -1.4369*** (0.298)
  - (5) -1.3888*** (0.247)
  - (6) -1.9079*** (0.333)
- Political risk, Log
  - (1) -2.0974*** (0.299)
  - (2) -1.9075*** (0.336)
  - (3) -2.0758*** (0.344)
  - (4) -2.0422*** (0.312)
  - (5) -1.4131*** (0.386)
  - (6) -1.3801*** (0.452)
- US Federal Fund rate
  - (1) 0.1847** (0.094)
  - (2) 0.1688* (0.099)
  - (3) 0.0808** (0.033)
  - (4) 0.1515** (0.060)
  - (5) 0.1696*** (0.055)
  - (6) 0.0508 (0.047)
- VIX, Log
  - (1) 0.1344 (0.149)
  - (2) 0.2504 (0.173)
  - (3) 0.5926*** (0.111)
  - (4) 0.2069** (0.093)
  - (5) 0.1475 (0.147)
  - (6) 0.6208*** (0.123)
- Debt crisis, Lagged
  - (1) 0.2239*** (0.048)
  - (2) 0.2687*** (0.054)
  - (3) 0.2089*** (0.055)
  - (4) 0.2106*** (0.050)
  - (5) 0.1620*** (0.045)
  - (6) 0.2346*** (0.059)
- SSA (Sub-Saharan Africa indicator)
  - (1) 0.4662*** (0.117)
  - (2) 0.1979 (0.429)
  - (3) -0.6735 (0.788)
  - (4) 0.2718 (0.400)
  - (5) 0.0280 (0.162)
  - (6) -2.6496 (2.006)
- Open budget index
  - (5) -0.0058 (0.004)
  - (6) 0.0026 (0.003)
- Open budget index * SSA
  - (5) 0.0043 (0.008)
  - (6) 0.0482 (0.034)
- Informal sector
  - (5) 0.0229*** (0.006)
  - (6) 0.0107 (0.007)
- Informal sector * SSA
  - (5) 0.0173 (0.015)
  - (6) 0.0146 (0.021)
- Financial development index * SSA
  - (5) -0.0124 (2.009)
  - (6) -2.1322 (2.281)
- Financial development
  - (5) -0.8006 (0.490)
  - (6) -0.7851 (0.527)
- Regulatory Quality
  - (5) -0.5167*** (0.087)
  - (6) -0.3616*** (0.121)
- Regulatory Quality * SSA
  - (5) -0.2647 (0.177)
  - (6) -0.5478** (0.242)
- Constant
  - (1) 21.3662*** (1.629)
  - (2) 20.2930*** (1.718)
  - (3) 21.7027*** (1.989)
  - (4) 21.4586*** (1.728)
  - (5) 18.9440*** (1.846)
  - (6) 19.1036*** (2.376)

### Sample and fit statistics
- Observations: (1) 4,075; (2) 3,328; (3) 2,445; (4) 3,741; (5) 3,828; (6) 2,050
- Number of countries: (1) 87; (2) 68; (3) 80; (4) 85; (5) 76; (6) 61
- R-squared: (1) 0.5529; (2) 0.4868; (3) 0.5146; (4) 0.6301; (5) 0.6760; (6) 0.6692

### Notes
- Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

---

### Annex 2. Robustness check: excluding outliers (10 percent of observations)

### Key regression results (columns (1)–(6))
- Economic risk, Log
  - (1) -0.8310*** (0.263)
  - (2) -0.5868*** (0.200)
  - (3) -0.9859*** (0.308)
  - (4) -0.7951*** (0.276)
  - (5) -0.8696*** (0.260)
  - (6) -0.6754** (0.283)
- Financial risk, Log
  - (1) -1.3465*** (0.285)
  - (2) -1.4986*** (0.281)
  - (3) -1.7095*** (0.331)
  - (4) -1.4206*** (0.304)
  - (5) -1.4236*** (0.259)
  - (6) -1.9259*** (0.319)
- Political risk, Log
  - (1) -1.9647*** (0.324)
  - (2) -1.9582*** (0.331)
  - (3) -1.9360*** (0.328)
  - (4) -1.8855*** (0.345)
  - (5) -1.3228*** (0.416)
  - (6) -1.1121** (0.4733)
- US Federal Fund rate
  - (1) 0.0894 (0.064)
  - (2) 0.0893 (0.063)
  - (3) 0.0748 (0.047)
  - (4) 0.0938 (0.088)
  - (5) 0.1190** (0.058)
  - (6) 0.0352 (0.057)
- VIX, Log
  - (1) 0.0988 (0.210)
  - (2) 0.2564 (0.198)
  - (3) 0.5976*** (0.103)
  - (4) 0.1921* (0.112)
  - (5) 0.1239 (0.201)
  - (6) 0.5937*** (0.125)
- Debt crisis, Lagged
  - (1) 0.2219*** (0.048)
  - (2) 0.2930*** (0.048)
  - (3) 0.2148*** (0.055)
  - (4) 0.2074*** (0.051)
  - (5) 0.1675*** (0.046)
  - (6) 0.2466*** (0.060)
- SSA
  - (1) 0.3768*** (0.111)
  - (2) 0.2924 (0.427)
  - (3) -0.3295 (0.545)
  - (4) 0.1751 (0.351)
  - (5) 0.0120 (0.148)
  - (6) -2.4903 (2.011)
- Open budget index
  - (5) 0.0000 (0.003)
  - (6) -0.0043* (0.002)
- Open budget index * SSA
  - (5) 0.0001 (0.008)
  - (6) 0.0429 (0.033)
- Informal sector
  - (5) 0.0199*** (0.004)
  - (6) 0.0095* (0.005)
- Informal sector * SSA
  - (5) 0.0081 (0.010)
  - (6) 0.0172 (0.022)
- Financial development index * SSA
  - (5) 0.1661 (1.721)
  - (6) -1.9781 (2.161)
- Financial development
  - (5) -0.7504 (0.489)
  - (6) -0.6726 (0.479)
- Regulatory Quality
  - (5) -0.4803*** (0.083)
  - (6) -0.3961*** (0.123)
- Regulatory Quality * SSA
  - (5) -0.2281 (0.171)
  - (6) -0.4257 (0.227)
- Constant
  - (1) 20.9115*** (1.642)
  - (2) 20.1724*** (1.478)
  - (3) 20.8438*** (1.888)
  - (4) 20.7793*** (1.928)
  - (5) 18.6749*** (1.870)
  - (6) 17.6292*** (2.502)

### Sample and fit statistics
- Observations: (1) 3,616; (2) 3,051; (3) 2,241; (4) 3,352; (5) 3,412; (6) 1,900
- Number of countries: (1) 85; (2) 64; (3) 78; (4) 83; (5) 85; (6) 59
- R-squared: (1) 0.487; (2) 0.492; (3) 0.477; (4) 0.551; (5) 0.577; (6) 0.628

### Notes
- Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

---

### Annex 3. Robustness check: excluding South Africa and Nigeria

### Key regression results (columns (1)–(6))
- Economic risk, Log
  - (1) -0.7022*** (0.231)
  - (2) -0.4572** (0.220)
  - (3) -1.0572*** (0.338)
  - (4) -0.6985*** (0.237)
  - (5) -0.6466** (0.261)
  - (6) -0.6609** (0.324)
- Financial risk, Log
  - (1) -1.9193*** (0.334)
  - (2) -1.9783*** (0.373)
  - (3) -2.2041*** (0.360)
  - (4) -1.9389*** (0.336)
  - (5) -2.0458*** (0.331)
  - (6) -2.3620*** (0.381)
- Political risk, Log
  - (1) -2.7381*** (0.384)
  - (2) -2.3020*** (0.365)
  - (3) -2.7843*** (0.461)
  - (4) -2.5378*** (0.397)
  - (5) -1.6705*** (0.462)
  - (6) -1.1728** (0.574)
- US Federal Fund rate
  - (1) 0.1955*** (0.051)
  - (2) 0.2286*** (0.062)
  - (3) 0.0715** (0.036)
  - (4) 0.1471*** (0.054)
  - (5) 0.2088*** (0.053)
  - (6) 0.0701* (0.039)
- VIX, Log
  - (1) 0.3537*** (0.112)
  - (2) 0.3600*** (0.112)
  - (3) 0.5960*** (0.114)
  - (4) 0.1431* (0.076)
  - (5) 0.3746*** (0.114)
  - (6) 0.5646*** (0.139)
- Debt crisis, Lagged
  - (1) 0.2371*** (0.046)
  - (2) 0.2818*** (0.052)
  - (3) 0.2184*** (0.049)
  - (4) 0.2201*** (0.049)
  - (5) 0.1755*** (0.043)
  - (6) 0.2218*** (0.061)
- SSA
  - (1) 0.5966*** (0.122)
  - (2) (blank) 0.1814 (0.417)
  - (3) -1.5083 (1.218)
  - (4) 0.6013 (0.765)
  - (5) -0.2891 (0.247)
  - (6) -1.2205 (1.205)
- Open budget index
  - (5) -0.0088 (0.005)
  - (6) 0.0037 (0.003)
- Open budget index * SSA
  - (5) 0.0080 (0.009)
  - (6) 0.0426** (0.016)
- Informal sector
  - (5) 0.0243*** (0.006)
  - (6) 0.0061 (0.006)
- Informal sector * SSA
  - (5) 0.0434 (0.031)
  - (6) 0.0197 (0.018)
- Financial development index * SSA
  - (5) -2.9147 (5.081)
  - (6) 1.7247 (1.641)
- Financial development
  - (5) -1.6265*** (0.455)
  - (6) -1.5199*** (0.575)
- Regulatory Quality
  - (5) -0.6389*** (0.135)
  - (6) -0.5664*** (0.182)
- Regulatory Quality * SSA
  - (5) -0.9475*** (0.354)
  - (6) -0.2808 (0.375)
- Constant
  - (1) 24.6609*** (2.041)
  - (2) 22.7793*** (1.836)
  - (3) 26.2132*** (2.494)
  - (4) 25.2518*** (2.010)
  - (5) 20.5895*** (2.159)
  - (6) 19.9269*** (2.303)

### Sample and fit statistics
- Observations: (1) 4,313; (2) 3,479; (3) 2,552; (4) 3,936; (5) 4,055; (6) 2,095
- Number of countries: (1) 85; (2) 64; (3) 78; (4) 83; (5) 85; (6) 59
- R-squared: (1) 0.6460; (2) 0.5870; (3) 0.6640; (4) 0.7340; (5) 0.7450; (6) 0.7600

### Notes
- Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

---

### Annex 4. Robustness check: excluding advanced economies

### Key regression results (columns (1)–(6))
- Economic risk, Log
  - (1) -0.5475*** (0.202)
  - (2) -0.5287** (0.221)
  - (3) -0.6589** (0.298)
  - (4) -0.5128** (0.210)
  - (5) -0.5053*** (0.196)
  - (6) -0.5155 (0.325)
- Financial risk, Log
  - (1) -1.7942*** (0.277)
  - (2) -1.7644*** (0.307)
  - (3) -1.9528*** (0.336)
  - (4) -1.8405*** (0.289)
  - (5) -1.8399*** (0.252)
  - (6) -2.0171*** (0.317)
- Political risk, Log
  - (1) -2.4779*** (0.410)
  - (2) -2.2030*** (0.380)
  - (3) -2.5228*** (0.488)
  - (4) -2.5029*** (0.417)
  - (5) -1.8152*** (0.455)
  - (6) -1.4603*** (0.420)
- US Federal Fund rate
  - (1) 0.1628*** (0.056)
  - (2) 0.2230*** (0.068)
  - (3) 0.0738** (0.031)
  - (4) 0.1738*** (0.055)
  - (5) 0.1458** (0.058)
  - (6) 0.0613 (0.041)
- VIX, Log
  - (1) 0.3623*** (0.138)
  - (2) 0.4160*** (0.149)
  - (3) 0.7476*** (0.139)
  - (4) 0.4199*** (0.095)
  - (5) 0.3614** (0.140)
  - (6) 0.6691*** (0.149)
- Debt crisis, Lagged
  - (1) 0.2483*** (0.049)
  - (2) 0.2867*** (0.053)
  - (3) 0.2307*** (0.058)
  - (4) 0.2362*** (0.050)
  - (5) 0.1826*** (0.048)
  - (6) 0.2388*** (0.062)
- SSA
  - (1) 0.2325** (0.118)
  - (2) 0.5661 (0.421)
  - (3) -1.1155 (0.641)
  - (4) 0.4460 (0.505)
  - (5) -0.2353 (0.209)
  - (6) -0.6932 (1.448)
- Open budget index
  - (5) 0.0055* (0.003)
  - (6) 0.0029 (0.002)
- Open budget index * SSA
  - (5) -0.0068 (0.008)
  - (6) 0.0311 (0.026)
- Informal sector
  - (5) 0.0147** (0.006)
  - (6) 0.0035 (0.005)
- Informal sector * SSA
  - (5) 0.0273 (0.016)
  - (6) -0.0071 (0.017)
- Financial development index * SSA
  - (5) -1.5689 (2.801)
  - (6) -3.1434 (2.413)
- Financial development
  - (5) -0.3554 (0.571)
  - (6) -0.6208 (0.544)
- Regulatory Quality
  - (5) -0.4420*** (0.103)
  - (6) -0.4518*** (0.119)
- Regulatory Quality * SSA
  - (5) -0.7188*** (0.278)
  - (6) -0.4404* (0.225)
- Constant
  - (1) 23.0666*** (1.969)
  - (2) 21.2504*** (1.770)
  - (3) 22.9053*** (2.236)
  - (4) 23.0699*** (1.987)
  - (5) 20.3341*** (2.101)
  - (6) 18.6397*** (2.026)

### Sample and fit statistics
- Observations: (1) 3,621; (2) 3,201; (3) 2,276; (4) 3,384; (5) 3,331; (6) 2,003
- Number of countries: (1) 66; (2) 55; (3) 59; (4) 64; (5) 66; (6) 50
- R-squared: (1) 0.5960; (2) 0.6030; (3) 0.5170; (4) 0.6120; (5) 0.6040; (6) 0.7240

### Notes
- Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

*IMF WORKING PAPERS Sub-Saharan Africa’s risk perception premium: in the search of missing factors — INTERNATIONAL MONETARY FUND*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023130-print-pdf.pdf_