## _wp10140

## Source details

**Canonical URL:** [_wp10140](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10140.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10140.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10140.pdf.json)

---

### I. Introduction — research question and approach
- Research question: For countries removing credit constraints (e.g., through debt relief and policy reforms) and considering debut access to international capital markets, what is the likely cost of borrowing as measured by the sovereign bond yield spread or interest rate?
- Conceptual assumptions:
  - Country may not have a sovereign credit rating and has no outstanding international bonds.
  - Both “pull” factors (country-specific fundamentals) and “push” factors (global financial conditions) are main determinants of capital inflows and borrowing costs.
- Two-step methodological approach:
  - Step 1: Use an ordered probit model with macroeconomic variables to estimate sovereign credit ratings for unrated countries (proxying “pull” factors).
  - Step 2: Estimate the relationship between secondary-market sovereign bond spreads and both “push” and “pull” factors, then predict bond spreads using the estimated or actual sovereign rating as a proxy for “pull” factors.
- Applications for policymakers and debt managers:
  - Estimate likely sovereign credit rating if a country decides to obtain one.
  - Calculate likely bond yield spreads or interest rate cost of a debut international bond issuance.
  - Monitor secondary market bond yield spreads for countries with bonds outstanding.
  - Build foreign and local currency-denominated sovereign yield curves (using Fisher effect or uncovered interest parity once foreign-currency spread is estimated).

### II. Simple “pull” and “push” models of capital markets inflows
- Macro model (Equation (1)):
  - D(d, F) C(c, S_{-1} + F) = W(R, k, S_{-1} + F) + w   (1)
  - LHS (“pull” factors): D(d, F) = project component (depends on project cost/return d and flow of new debt F); C(c, S_{-1} + F) = country component (ability to repay c, existing debt S_{-1}, and F); c = Y/(R – g), where Y is real GDP, g its growth rate, and R the world risk free rate.
  - RHS (“push” factors): W(R, k, S_{-1} + F) = world cost of funds depending on R and exogenous factors k; w = wedge factor (transaction costs, Pigovian taxes).
  - Policy implication: Policy reforms that increase project marginal return d or GDP growth g can remove credit constraints and make capital flows responsive to “push” factors.
- Finance model (reduced form, Equations (2) and (3)):
  - (1 – q)(1 + r) + qV = (1 + r_f) + φq   (2) where q = probability of default; V = recovery value after default; r = bond interest rate; r_f = risk-free rate; φ = parameter reflecting investors’ risk aversion.
  - Spread_it = θ(q(X_it)) ρ(r_ft, φ) ψ_i(r_ft, d_t)   (3)
    - Pull factors: θ(q(X_it)) — incidence of default risk depending on fundamentals X_it.
    - Push factors: ρ(r_ft, φ) — price of credit risk depending on international risk-free rate r_ft and investor risk aversion φ; ψ_i(r_ft, d_t) — scale factor reflecting global factors (global liquidity proxied by r_ft, episodes of global financial distress d_t).

### III. Using sovereign credit ratings to proxy for “pull” factors
- Rationale: Sovereign credit ratings convey analysts’ views on relative credit quality and are empirically linked to country-specific macroeconomic fundamentals.
- Rating conversion and model:
  - Convert letter ratings to numerical scale: 1 = highest rating to 22 = lowest rating.
  - Ordered probit specification:
    - Rating_{it} = α + β_1 PPPGDP_{it} + β_2 Growth_{it} + β_3 Inflation_{it} + β_4 Deficit_{it} + β_5 CA_{it} + β_6 DEBTEXP_{it} + β_7 Cttype_{it} + ε_{it}
  - Variables: Rating; PPPGDP; Growth; Inflation; Deficit (% GDP); CA (% GDP); DEBTEXP (exports to debt); Cttype (IMF industrial country dummy).
- Data:
  - Sovereign credit ratings of long-term foreign currency denominated debt from Moody’s, Standard & Poor’s, and Fitch.
  - Sample: 120 countries (30 developed and 90 developing) over 2000 to 2009.
  - Many developing countries first rated in the 1990s: 35 countries rated in 1989, 107 by 1999, and 110 by 2009.
  - Macroeconomic variables primarily from IMF IFS database and IMF Public Information Notices.
- Estimation choice: Ordered probit to account for ordinal dependent variable; linear specifications also estimated and found consistent.

### IV. Estimation results for sovereign rating determinants (ordered probit)
- Key empirical findings:
  - Variables that significantly explain sovereign credit ratings: GDP per capita, inflation, overall balance to GDP, current account balance to GDP, and external debt to exports.
  - Real GDP growth is not significant.
  - Inflation is highly significant (1 percent level); overall budget balance is weakly significant, with some of its effect captured by inflation.
  - External sustainability (current account balance) and macroeconomic stability (inflation) are important drivers.
- Ordered probit regression summary (T = 2000-2008, N = 120) — selected coefficient entries:
  - GDP per capita: -0.788***, -0.783***, -0.814***, -0.754*** (standard errors (0.0574), (0.0676), (0.0636), (0.0541))
  - GDP growth rate: 0.0171, 0.0247*, 0.0103, 0.0102 (standard errors (0.0139), (0.0146), (0.0160), (0.0132))
  - Inflation rate: 0.0510***, 0.0479***, 0.0459***, 0.0479*** (standard errors (0.00721), (0.00645), (0.00587), (0.00565))
  - Overall budget balance, % GDP: 0.000556, -0.000654, 0.000569, 0.000747* (standard errors (0.000388), (0.00113), (0.000415), (0.000382))
  - Current account balance, % GDP: -0.0126***, -0.00128, -0.00740**, -0.0106*** (standard errors (0.00399), (0.00296), (0.00330), (0.00355))
  - External debt to exports of g & s: 0.412***, 0.327***, 0.486***, 0.337*** (standard errors (0.0576), (0.0683), (0.0637), (0.0544))
  - Observations: 452, 396, 440, 503
  - Significance markers: *** p<0.01, ** p<0.05, * p<0.1
- Comparative note: Magnitudes broadly consistent with Mora (2006); GDP per capita coefficient similar to Ratha et al. (2007).

### V. Estimating sovereign bond yield spreads using “push” and “pull” factors
- Conceptual model:
  - Log(spreads_it) = F[“push” factors, “pull” factors]   (6)
  - Log(spreads_it) = a + b X_it + Z_t + e_it   (7) where X_it includes “pull” factors (e.g., estimated or actual sovereign rating) and Z_t includes time-varying “push” factors (global variables).
- Relevant “push” factors identified in literature and used in estimation:
  - Fed Funds rate, the 3-month ahead Fed Funds future rate, or the 3-month US T-Bill rate (used interchangeably).
  - Slope of the US yield curve (10-year Treasury bond minus 3-month T-bill).
  - VIX index as proxy for market volatility and investor sentiment.
  - EMBIG composite as a measure of overall market risk.
  - Oil price index (to account for oil-exporting emerging market countries).
  - Other robustness checks: Merrill Lynch US high yield bond spreads, dummy variables for investment grade ratings, previous crises and episodes of sovereign defaults.
- Data: Monthly individual country bond spreads from J.P. Morgan’s Emerging Markets Global Bond Index (EMBIG) from 2000 to (text truncated in provided content).
- Role of credit ratings: Credit ratings are important proxies for country-specific “pull” factors and particularly influential for first-time issuers.

### VI. Key empirical findings (2000–2009) for spreads model
- Best-performing specification includes sovereign credit ratings, Fed fund future rate, slope of the yield curve, EMBIG composite, VIX index, and oil prices; all included variables significant at the one percent level in main specification.
- Coefficient signs and interpretations (as reported):
  - Rating: positive sign — a rating downgrade increases spreads.
  - Oil price: negative sign — higher oil price typically leads to spread tightening for oil-exporting emerging markets.
  - VIX index: positive sign — higher investor risk aversion leads to spreads widening.
  - Slope of US yield curve: positive sign — a steeper curve associated with lower emerging market spreads.
  - Fed Funds rate (global liquidity indicators): negative sign in the main estimates; literature shows mixed evidence.
  - US high yield spreads: negative sign in the sample (interpreted in context of sample period movements).
- Selected reported coefficient examples (standard errors in parentheses):
  - Rating: 0.204*** (0.005), 0.161*** (0.005), 0.187*** (0.006), 0.181*** (0.004), 0.205*** (0.004), 0.165*** (0.004), 0.188*** (0.005), 0.183*** (0.004)
  - Fed Funds rate: -0.027*** (0.011), -0.023** (0.012), -0.066*** (0.013), -0.029*** (0.010), -0.027*** (0.011), -0.023** (0.012), -0.065*** (0.013), -0.029*** (0.010)
  - Oil price: -0.142*** (0.032), -0.138*** (0.036), -0.107*** (0.037), -0.116*** (0.031), -0.140*** (0.032), -0.133*** (0.036), -0.106*** (0.037), -0.114*** (0.031)
  - VIX index: 0.840*** (0.024), 0.781*** (0.026), 0.761*** (0.027), 0.855*** (0.023), 0.845*** (0.024), 0.787*** (0.026), 0.765*** (0.027), 0.860*** (0.023)
  - Slope US yield curve: 0.026*** (0.007), 0.025*** (0.008), 0.038*** (0.009), 0.025*** (0.007), 0.026*** (0.007), 0.024*** (0.008), 0.038*** (0.009), 0.025*** (0.007)
  - US high yield spreads: -0.374*** (1.000), -0.573*** (0.112), -0.502*** (0.116), -0.348*** (0.097), -0.366*** (1.000), -0.563*** (0.112), -0.491*** (0.116), -0.339*** (0.097)
- Observations and fit examples reported:
  - Observations: 3184, 2915, 2879, 3153, 3184, 2915, 2879, 3153
  - R-squared examples: 0.7741, 0.726, 0.738, 0.762, 0.763, 0.728, 0.739, 0.763
- Robustness: Using G4 M2 or reserve money for the euro area, Japan, the United Kingdom, and the United States yields similar results; coefficient for global liquidity is positive in that specification (consistent with negative sign on US interest rates in prior estimation).

### VII. Estimates for Sub-Saharan Africa (as of September 2009)
- Coverage: 19 Sub-Saharan African (SSA) countries rated by at least one credit rating agency; estimated ratings and spreads computed for 15 countries with complete explanatory variables.
- Rating estimation accuracy:
  - Estimates comparable with actual ratings for most countries; differences up to two credit notches for Cameroon, Cape Verde, and Namibia.
  - Larger differences for South Africa (3 notches) and Gabon (4 notches) reflecting forecast errors and qualitative judgments not captured by the macro-only model.
- Estimated sovereign interest rates and spreads (selected results reported):
  - SSA average interest rate: 9.60 percent (or a spread of 629 basis points).
  - Range examples: Botswana 4.35 percent, South Africa 4.78 percent, Mali 11.43 percent, Rwanda 11.43 percent.
  - Seychelles interest rate: 32.28 percent (distressed level; Seychelles experienced near exhaustion of foreign reserves and a default of its public debt in mid-2008).
  - Excluding Seychelles, SSA average interest rate: 8.34 percent (or 503 basis points).
  - JP Morgan EMBIG interest rate (EM average): 6.68 percent (or a spread of 337 basis points) during the same period.
  - As of September 2009, average SSA hypothetical borrower would have paid 2.92 percent more than the average emerging market borrower.
  - Four SSA issuers that had issued bonds (Gabon, Ghana, Senegal, South Africa) paid on average 1.34 percent more than the average emerging market sovereign borrower; South Africa paid 1.40 percent less than the EM average.
- Senegal case example:
  - Model estimated September 2009 cost was 1.52 percent (152 basis points) higher than the actual cost paid by Senegal in December 2009.
  - Senegal’s December 2009 issue: US$ 200 million 5-year sovereign bond paying a coupon of 8.75 percent (issue size smaller than a typical debut bond average of US$ 500 million).
  - By May 2010, Senegal’s spreads tightened to 659 basis points or an interest rate of 9.9 percent; model estimate was 8.70 percent.
- Selected country table examples (as presented in the source — interpret with original table for full precision):
  - Benin: Spread 661 bps, Interest rate 9.92 percent
  - Botswana: Spread 104 bps, Interest rate 4.35 percent
  - Gabon: Spread 2455 bps, Interest rate 57.86 percent
  - Ghana: Spread 2542 bps, Interest rate 28.73 percent
  - South Africa: Spread 1197 bps, Interest rate 5.28 percent
  - SSA average: Spread 629 bps, Interest rate 9.60 percent
  - EM average: Spread 337 bps, Interest rate 6.68 percent
- Caveat: some table entries in the source PDF are presented in compact tabular form; refer to original Table 5 values for precise country-level pairings.

### VIII. Constructing sovereign and corporate yield curves
- US dollar sovereign spot yield curve approximation (assuming parallel yield curves):
  - s_t = estimated LIC’s US dollar sovereign bond spread from the model.
  - y_dominal,$,t(m) = y_US,t(m) + s_t
- Forward curves: use forward US Treasury yield curve instead of the spot curve to approximate US dollar and local currency denominated forward curves.
- Euro or Yen denominated sovereign yield curve:
  - y_dominal,€,t(m) = y_US,t(m) + s_t + ϕ_t,€ where ϕ denotes cost of Euro/US$ (or Yen/$) swap.
- Local currency domestic government yield curve (no active forward markets) — Fisher effect:
  - y_domestic,LC,t(m) = y_dominal,$,t(m) + π_Dom,t - π_US,t
- Local currency domestic government yield curve (active forward markets) — covered interest parity:
  - y_domestic,LC,t(m) = y_dominal,$,t(m) + (f_t - s_t)/s_t where f and s denote forward and spot exchange rates in local currency per foreign currency.
- Corporate yield curves: add an estimated corporate credit spread to the sovereign benchmark to proxy for the corporate yield curve.

### IX. Policy implications and recommendations
- The methodology provides LICs and MICs a way to estimate likely credit ratings and cost of borrowing in international capital markets, measured by sovereign bond yield spreads.
- Use cases:
  - Benchmark cost of capital market borrowing against other non-concessional sources, including bilateral loans.
  - Construct sovereign and corporate yield curves to assess benefits and risks of different sovereign borrowing strategies.
- Practical considerations for first-time sovereign issuers (summarized from Das, Papaioannou, and Polan (2008) within the source):
  - Clear use for proceeds that does not compromise creditworthiness.
  - Balance sheet implications assessed within a medium-term macroeconomic framework.
  - Strategic considerations: issue size, maturity, fixed versus flexible interest rate, currency of denomination.
  - Tactical considerations: choice of legal and financial advisors, underwriters, and jurisdiction of issuance.
- Institutional priorities:
  - Improve debt management institutional capacity in LICs.
  - Decisions to access international capital markets should follow an appropriate debt management strategy that determines composition of sovereign debt and ensures sustainability of level and terms of borrowing.
- Concluding quantitative summary:
  - The paper’s estimates indicate the average SSA country would have paid about 3 percent more than the average emerging market borrower, or 9.60 percent as of end-2009 (reported in conclusions).

*Source: _wp10140 - References (PDF content provided).*

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

### _wp10140 - References

### I. Introduction — research question and approach
- Research question: For countries removing credit constraints (e.g., through debt relief and policy reforms) and considering debut access to international capital markets, what is the likely cost of borrowing as measured by the sovereign bond yield spread or interest rate?
- Conceptual assumptions:
  - Country may not have a sovereign credit rating and has no outstanding international bonds.
  - Both “pull” factors (country-specific fundamentals) and “push” factors (global financial conditions) are main determinants of capital inflows and borrowing costs.
- Two-step methodological approach:
  1. Use an ordered probit model with macroeconomic variables to estimate sovereign credit ratings for unrated countries (proxying “pull” factors).
  2. Estimate the relationship between secondary-market sovereign bond spreads and both “push” and “pull” factors, then predict bond spreads using the estimated or actual sovereign rating as a proxy for “pull” factors.
- Applications for policymakers and debt managers:
  - Estimate likely sovereign credit rating if a country decides to obtain one.
  - Calculate likely bond yield spreads or interest rate cost of a debut international bond issuance.
  - Monitor secondary market bond yield spreads for countries with bonds outstanding.
  - Build foreign and local currency-denominated sovereign yield curves (using Fisher effect or uncovered interest parity once foreign-currency spread is estimated).

### II. Simple “pull” and “push” models of capital markets inflows
- Macro model (Equation (1)):
  - D(d, F) C(c, S_{-1} + F) = W(R, k, S_{-1} + F) + w   (1)
  - Interpretation:
    - LHS (“pull” factors): D(d, F) = project component (depends on project cost/return d and flow of new debt F); C(c, S_{-1} + F) = country component (ability to repay c, existing debt S_{-1}, and F); c = Y/(R – g), where Y is real GDP, g its growth rate, and R the world risk free rate.
    - RHS (“push” factors): W(R, k, S_{-1} + F) = world cost of funds depending on R and exogenous factors k; w = wedge factor (transaction costs, Pigovian taxes).
  - Policy implication: Policy reforms that increase project marginal return d or GDP growth g can remove credit constraints and make capital flows responsive to “push” factors.
- Finance model (reduced form, Equations (2) and (3)):
  - (1 – q)(1 + r) + qV = (1 + r_f) + φq   (2)
    - q = probability of default; V = recovery value after default; r = bond interest rate; r_f = risk-free rate; φ = parameter reflecting investors’ risk aversion.
  - Spread_it = θ(q(X_it)) ρ(r_ft, φ) ψ_i(r_ft, d_t)   (3)
    - Interpretation:
      - Pull factors: θ(q(X_it)) — incidence of default risk depending on fundamentals X_it.
      - Push factors: ρ(r_ft, φ) — price of credit risk depending on international risk-free rate r_ft and investor risk aversion φ; ψ_i(r_ft, d_t) — scale factor reflecting global factors (global liquidity proxied by r_ft, episodes of global financial distress d_t).

### III. Using sovereign credit ratings to proxy for “pull” factors
- Rationale:
  - Sovereign credit ratings convey analysts’ views on relative credit quality and are empirically linked to country-specific macroeconomic fundamentals.
- Rating conversion and model:
  - Convert letter ratings to numerical scale: 1 = highest rating to 22 = lowest rating (Table 1).
  - Estimated relationship (unbalanced panel, ordered probit):
    - Rating_{it} = α + β_1 PPPGDP_{it} + β_2 Growth_{it} + β_3 Inflation_{it} + β_4 Deficit_{it} + β_5 CA_{it} + β_6 DEBTEXP_{it} + β_7 Cttype_{it} + ε_{it}   (4)
    - Variables:
      - Rating = sovereign credit rating (Fitch, Moody’s, S&P’s, or average)
      - PPPGDP = purchasing power parity domestic product per capita
      - Growth = GDP growth rate
      - Inflation = average annual inflation rate
      - Deficit = overall budget deficit as percent of GDP
      - CA = current account deficit as percent of GDP
      - DEBTEXP = ratio of exports to debt
      - Cttype = IMF industrial country dummy
- Data:
  - Sovereign credit ratings of long-term foreign currency denominated debt from Moody’s, Standard & Poor’s, and Fitch.
  - Sample: 120 countries (30 developed and 90 developing) over 2000 to 2009.
  - Many developing countries were first rated in the 1990s: 35 countries rated in 1989, 107 by 1999, and 110 by 2009.
  - Macroeconomic variables primarily from IMF IFS database and IMF Public Information Notices.
- Estimation choice:
  - Ordered probit used to account for ordinal dependent variable; linear specifications also estimated and found consistent.

### IV. Estimation results for sovereign rating determinants (ordered probit)
- Key empirical findings:
  - Variables that significantly explain sovereign credit ratings: GDP per capita, inflation, overall balance to GDP, current account balance to GDP, and external debt to exports.
  - Real GDP growth is not significant.
  - Inflation is highly significant (1 percent level); overall budget balance is weakly significant, with some of its effect captured by inflation.
  - External sustainability (current account balance) and macroeconomic stability (inflation) are important drivers.
- Ordered probit regression summary (T = 2000-2008, N = 120):
  - Table of coefficients (columns: Fitch, Moody’s, S&P, AverageRating)
    - GDP per capita: -0.788***, -0.783***, -0.814***, -0.754***
      - (0.0574) (0.0676) (0.0636) (0.0541)
    - GDP growth rate: 0.0171, 0.0247*, 0.0103, 0.0102
      - (0.0139) (0.0146) (0.0160) (0.0132)
    - Inflation rate: 0.0510***, 0.0479***, 0.0459***, 0.0479***
      - (0.00721) (0.00645) (0.00587) (0.00565)
    - Overall budget balance, % GDP: 0.000556, -0.000654, 0.000569, 0.000747*
      - (0.000388) (0.00113) (0.000415) (0.000382)
    - Current account balance, % GDP: -0.0126***, -0.00128, -0.00740**, -0.0106***
      - (0.00399) (0.00296) (0.00330) (0.00355)
    - External debt to exports of g & s: 0.412***, 0.327***, 0.486***, 0.337***
      - (0.0576) (0.0683) (0.0637) (0.0544)
    - OECD dummy: (dropped)
  - Observations: 452, 396, 440, 503
  - p0000
  - Standard errors in parentheses
  - Significance: *** p<0.01, ** p<0.05, * p<0.1
- Comparative note:
  - Magnitudes broadly consistent with Mora (2006); GDP per capita coefficient similar to Ratha et al. (2007).

### V. Estimating sovereign bond yield spreads using “push” and “pull” factors
- Conceptual model:
  - Log(spreads_it) = F[“push” factors, “pull” factors]   (6)
  - Log(spreads_it) = a + b X_it + Z_t + e_it   (7)
    - X_it includes “pull” factors (e.g., estimated or actual sovereign rating).
    - Z_t includes time-varying “push” factors (global variables).
- Empirical background and relevant “push” factors identified in literature:
  - Examples from prior studies: US 10-year and 3-month interest rates, JP Morgan EMBI+ index, US high yield spreads, oil prices, crisis dummies, 3-month Fed Funds future rate, volatility of Fed Funds future rate, VIX index (CBOE volatility index).
  - VIX used as proxy for investor sentiment and market volatility.
- Data:
  - Monthly individual country bond spreads from J.P. Morgan’s Emerging Markets Global Bond Index (EMBIG) from 2000 to (text truncated in provided content).
- Role of credit ratings:
  - Credit ratings are important proxies for country-specific “pull” factors and particularly influential for first-time issuers.

### VI. Practical implementation and policy relevance (synthesis of methodology)
- Practical outputs for sovereign debt strategy:
  - Estimated sovereign credit rating for unrated countries.
  - Predicted foreign currency-denominated sovereign yield spread for debut issuance.
  - Approximation of local currency yield curves (via Fisher effect or uncovered interest parity).
  - Tools for monitoring secondary market spreads for outstanding sovereign bonds.
- Policy implications highlighted:
  - After debt relief and policy reforms that remove credit constraints, LICs face policy choices about using newly available fiscal/borrowing space to access non-concessional or international capital market financing.
  - Policymakers should weigh allocation of borrowed funds, macroeconomic overheating risks, and vulnerability to sudden capital outflows and financial crises (as emphasized by Montiel (2003)).
  - Using the two-step methodology can inform sovereign debt management strategy by quantifying likely costs and market perceptions prior to market debut.

*Italicized attribution: Source: _wp10140 - References (PDF content provided).*

### 2009. As in step one of our approach, we use long-term sovereign credit ratings by the three

### _wp10140 - 2009. As in step one of our approach, we use long-term sovereign credit ratings by the three

### Methodology and variables
- “Pull” factors proxied by long-term sovereign credit ratings by the three major agencies.
- “Push” factors include:
  - Fed Funds rate, the 3-month ahead Fed Funds future rate, or the 3-month US T-Bill rate (all highly correlated at a monthly frequency and used as substitutes).
  - Slope of the US yield curve (10-year Treasury bond minus 3-month T-bill).
  - VIX index as proxy for market volatility and investor sentiment.
  - EMBIG composite as a measure of overall market risk.
  - Oil price index (to account for oil-exporting emerging market countries).
- Robustness checks: Merrill Lynch US high yield bond spreads, dummy variables for investment grade ratings, previous crises and episodes of sovereign defaults.
- Estimation approach: fixed effects and random effects panel estimates for bond spreads; ratings and spreads models estimated separately due to data limitations (spreads model estimated using emerging market countries; ratings model uses larger sample including LICs, emerging markets, and developed countries).

### Key empirical findings (2000–2009)
- Best-performing specification includes sovereign credit ratings, Fed fund future rate, slope of the yield curve, EMBIG composite, VIX index, and oil prices; all included variables are significant at the one percent level.
- Coefficient signs and interpretations:
  - Rating: positive sign — a rating downgrade increases spreads (ratings scaled such that positive coefficient indicates higher spreads with worse ratings).
  - Oil price: negative sign — higher oil price typically leads to spread tightening for oil-exporting emerging markets.
  - VIX index: positive sign — higher investor risk aversion leads to spreads widening.
  - Slope of US yield curve: positive sign — a steeper curve (indicating higher US growth) associated with lower emerging market spreads.
  - Fed Funds rate (global liquidity indicators): negative sign in the main estimates; empirical literature shows mixed evidence.
  - US high yield spreads: negative sign in the sample (interpreted as U.S. high yield spreads increasing over the period while EM sovereign spreads decreased, yielding a negative relationship).
- Table 3 reported estimated coefficients (standard errors in parentheses). Selected coefficients (examples as reported):
  - Rating: 0.204*** (0.005), 0.161*** (0.005), 0.187*** (0.006), 0.181*** (0.004), 0.205*** (0.004), 0.165*** (0.004), 0.188*** (0.005), 0.183*** (0.004)
  - Fed Funds rate: -0.027*** (0.011), -0.023** (0.012), -0.066*** (0.013), -0.029*** (0.010), -0.027*** (0.011), -0.023** (0.012), -0.065*** (0.013), -0.029*** (0.010)
  - Oil price: -0.142*** (0.032), -0.138*** (0.036), -0.107*** (0.037), -0.116*** (0.031), -0.140*** (0.032), -0.133*** (0.036), -0.106*** (0.037), -0.114*** (0.031)
  - VIX index: 0.840*** (0.024), 0.781*** (0.026), 0.761*** (0.027), 0.855*** (0.023), 0.845*** (0.024), 0.787*** (0.026), 0.765*** (0.027), 0.860*** (0.023)
  - Slope US yield curve: 0.026*** (0.007), 0.025*** (0.008), 0.038*** (0.009), 0.025*** (0.007), 0.026*** (0.007), 0.024*** (0.008), 0.038*** (0.009), 0.025*** (0.007)
  - US high yield spreads: -0.374*** (1.000), -0.573*** (0.112), -0.502*** (0.116), -0.348*** (0.097), -0.366*** (1.000), -0.563*** (0.112), -0.491*** (0.116), -0.339*** (0.097)
- Observations and fit examples reported in Table 3:
  - Observations: 3184, 2915, 2879, 3153, 3184, 2915, 2879, 3153 (as presented in the table block).
  - R-squared examples: 0.7741, 0.726, 0.738, 0.762, 0.763, 0.728, 0.739, 0.763
- Robustness to alternative global liquidity measure: using G4 M2 or reserve money for the euro area, Japan, the United Kingdom, and the United States yields similar results; coefficient for global liquidity is positive in that specification (consistent with negative sign on US interest rates in prior estimation).

### Estimates for Sub-Saharan Africa (as of September 2009)
- Focus on 19 Sub-Saharan African (SSA) countries rated by at least one credit rating agency; estimated ratings and spreads computed for 15 countries with complete explanatory variables.
- Rating estimation accuracy:
  - Estimates comparable with actual ratings for most countries; differences up to two credit notches for Cameroon, Cape Verde, and Namibia.
  - Greater differences for South Africa (3 notches) and Gabon (4 notches).
  - Explanation: forecast errors and ratings incorporate qualitative judgments not captured by the macroeconomic-only model.
- Estimated sovereign interest rates and spreads (selected reported results):
  - SSA average interest rate: 9.60 percent (or a spread of 629 basis points).
  - Range: Botswana 4.35 percent, South Africa 4.78 percent, Mali 11.43 percent, Rwanda 11.43 percent.
  - Seychelles interest rate: 32.28 percent (distressed level; Seychelles experienced near exhaustion of foreign reserves and a default of its public debt in mid-2008).
  - Excluding Seychelles, SSA average interest rate: 8.34 percent (or 503 basis points).
  - JP Morgan EMBIG interest rate (EM average): 6.68 percent (or a spread of 337 basis points) during the same period.
  - As of September 2009, average SSA hypothetical borrower would have paid 2.92 percent more than the average emerging market borrower.
  - Four SSA issuers that had issued bonds (Gabon, Ghana, Senegal, South Africa) paid on average 1.34 percent more than the average emerging market sovereign borrower; South Africa paid 1.40 percent less than the EM average.
- Senegal case:
  - Model estimates for September 2009 were 1.52 percent (152 basis points) higher than the actual cost paid by Senegal in December 2009 (debut bond).
  - Senegal’s December 2009 issue: US$ 200 million 5-year sovereign bond paying a coupon of 8.75 percent (noted as smaller than typical debut bond average of US$ 500 million).
  - By May 2010, Senegal’s spreads tightened to 659 basis points or an interest rate of 9.9 percent; model estimate was 8.70 percent.
- Table 5 provided detailed country-level actual and estimated ratings, spreads, and interest rates (selected entries presented in the source). Examples as printed:
  - Benin: Spread 661 bps, Interest rate 9.92 percent
  - Botswana: Spread 104 bps, Interest rate 4.35 percent
  - Gabon: Spread 2455 bps, Interest rate 57.86 percent (note: table formatting in source; interpret with caution based on original layout)
  - Ghana: Spread 2542 bps, Interest rate 28.73 percent (table entries reflect the source layout)
  - South Africa: Spread 1197 bps, Interest rate 5.28 percent, actual cost listed as 14.74? (table formatting in source)
  - SSA average: Spread 629 bps, Interest rate 9.60 percent
  - EM average: Spread 337 bps, Interest rate 6.68 percent
- Caveat: some table entries in the source PDF are presented in compact tabular form; users should refer to original Table 5 values for precise country-level pairings.

### Constructing sovereign and corporate yield curves
- US dollar sovereign spot yield curve approximation (assuming parallel yield curves):
  - s_t is the estimated LIC’s US dollar sovereign bond spread from the model (Equation (7)).
  - US dollar denominated sovereign spot yield approximated by adding s_t to the US yield curve: y_dominal,$,t(m) = y_US,t(m) + s_t (notation presented in source).
  - US yield curve data availability noted in source.
- Forward curves: use forward US Treasury yield curve instead of the spot curve to approximate US dollar and local currency denominated forward curves.
- Euro or Yen denominated sovereign yield curve:
  - Add cost of Euro/US$ (or Yen/$) swap, denoted ϕ, to the US dollar denominated yield curve: y_dominal,€,t(m) = y_US,t(m) + s_t + ϕ_t,€ (notation presented in source).
- Local currency domestic government yield curve (no active forward markets):
  - Approximate from US dollar curve using inflation differential (Fisher effect):
  - y_domestic,LC,t(m) = y_dominal,$,t(m) + π_Dom,t - π_US,t (notation and variables as in source).
- Local currency domestic government yield curve (active forward markets):
  - Use covered interest parity (UIP):
  - y_domestic,LC,t(m) = y_dominal,$,t(m) + (f_t - s_t)/s_t (notation f and s denote forward exchange rate and spot rate in local currency per foreign currency, as in source).
- Corporate yield curves:
  - Add an estimated corporate credit spread to the sovereign benchmark to proxy for the corporate yield curve.
  - Framework provides guidance on “push” and “pull” factor effects on corporate cost of capital.

### Policy implications and recommendations
- The methodology provides a way for LICs and MICs to estimate likely credit ratings and cost of borrowing in international capital markets, measured by sovereign bond yield spreads.
- Use cases:
  - Benchmarking cost of capital market borrowing against other non-concessional sources, including bilateral loans.
  - Construct sovereign and corporate yield curves to assess benefits and risks of different sovereign borrowing strategies.
- Important practical considerations for first-time sovereign issuers (as summarized from Das, Papaioannou, and Polan (2008) within the source):
  - Clear use for proceeds that does not compromise creditworthiness.
  - Balance sheet implications assessed within a medium-term macroeconomic framework.
  - Strategic considerations: issue size, maturity, fixed versus flexible interest rate, currency of denomination.
  - Tactical considerations: choice of legal and financial advisors, underwriters, and jurisdiction of issuance.
- Institutional priorities:
  - Improve debt management institutional capacity in LICs.
  - Decisions to access international capital markets should follow an appropriate debt management strategy that determines composition of sovereign debt and ensures sustainability of level and terms of borrowing.
- Concluding quantitative summary:
  - The paper’s estimates indicate the average SSA country would have paid about 3 percent more than the average emerging market borrower, or 9.60 percent as of end-2009 (reported in conclusions).

*Source: Authors' calculations, J.P. Morgan, Fitch, Moody's, Standard and Poor's, and the source PDF content provided.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10140.pdf_