## _wp0458

## Source details

**Canonical URL:** [_wp0458](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0458.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0458.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0458.pdf.json)

---

### Introduction
- SDDS introduced in 1996 to guide countries with or seeking access to international capital markets in the provision of economic and financial data; four dimensions: access, integrity, quality, and the data themselves.  
- SDDS prescribes monitorable minimum timeliness and frequency standards across 18 data categories covering real, fiscal, financial, and external sectors (summarized in Table 1).  
- Prior secondary-market studies report SDDS-associated reductions in EMBI spreads: examples include 200–300 basis points (IIF, 2002), about 15 percent (Christofides, Mulder, and Tiffin, 2003), and 4–12 percent (Glennerster and Shin, 2003).  
- This paper examines primary-market sovereign bond issuance and asks whether SDDS subscription lowers launch spreads for sovereign bonds denominated in U.S. dollar, yen, and euro.

### Data and estimation methodology
- Countries in panel: Argentina, Brazil, Colombia, Mexico, the Philippines, South Africa, and Turkey.  
  - Colombia and Brazil subscribed in May 1996 and March 2001, respectively; the remaining five countries subscribed in August 1996.  
  - Estimation period generally ranges from 1990 to 2002, starting approximately six years prior to, and ending six years after, mid-1996. Specific start dates: Argentina (1994:1), Brazil (1992:4), Colombia (1995:1) for parts of the sample.  
- Data sources:
  - Sovereign bond characteristics from the Bonds, Equities and Loans (BEL) database of the IMF (sourced from Capital Data).  
  - Macroeconomic data from IMF’s International Financial Statistics (IFS).  
  - External debt indicators from World Bank’s Global Development Finance (GDF).  
- Sample restrictions:
  - Only fixed interest rate bonds denominated in U.S. dollars, yen, and euros included (spread data available only for these).  
  - Bond maturities in sample range from 1 to 30 years (sample period mean is 7 ½ years).  
- Explanatory variables considered include: real GDP growth (YDOT), investment/GDP, inflation, short- and long-term global interest rates (USFED, USLONG), fiscal and external current account balances (% of GDP), public external debt stock-to-GDP and exports ratios (DXR), debt service-to-exports, short-term external debt relative to reserves, Paris Club rescheduling (PARIS), IMF program status (IMF), bond maturity (MAT), and currency dummies (EURO, YEN).  
- SDDS influence captured by a dummy variable (SDDS = 0 prior to subscription, = 1 thereafter); subscription dates treated as exogenous.  
- Time series stationarity: ADF tests indicate all series used in reported regressions are stationary at conventional significance levels (Appendix I, Table A1). Annual external debt stock-to-exports ratios converted to quarterly frequency and smoothed with Hodrick-Prescott filter prior to testing.  
- Estimation technique: pooled time series cross-section estimation with fixed effects correcting for equation-specific serial correlation and cross-section heteroscedasticity. Smoothed (four quarter moving average) real GDP growth and U.S. interest rates used.

### Estimation results — main findings
- Basic specification (Equation (1), estimation period 1990:2 to 2002:4) produces a relatively high degree of fit; coefficients generally of expected sign and statistically significant.
- Key estimated coefficient results from Table 2 (Equation 1):
  - YDOT (real growth): -6.61 (t = -2.88)
  - MAT (maturity): 3.11 (t = 4.38)
  - DXR (debt-exports ratio): 1.37 (t = 1.97)
  - USFED (federal funds rate): 0.16 (t = 2.21)
  - USLONG (10-year bond rate): -0.63 (t = -5.29)
  - EURO (euro-denominated): -94.69 (t = -8.60)
  - YEN (yen-denominated): -126.26 (t = -11.27)
  - PARIS (Paris Club): 38.56 (t = 2.19)
  - IMF (IMF arrangement): -27.04 (t = -1.82)
  - SDDS (subscription): -74.86 (t = -3.97)
- Additional model statistics (Equation 1):
  - Adjusted R2: 0.797
  - Durbin-Watson: 2.24
  - Number of observations: 303
  - Mean of the dependent variable: 340.30
- Robustness and alternative specifications:
  - Equation (2) (common constant term) yields SDDS coefficient = -82.56 (t = -4.36); Adjusted R2: 0.793; observations: 303; mean dependent variable: 304.30.
  - Equation (3) (shortened sample period 1992:3 to 2000:3) yields SDDS coefficient = -65.49 (t = -3.15); Adjusted R2: 0.740; observations: 205; mean dependent variable: 322.16.
  - A 90 percent confidence interval for the SDDS estimate (Equation 1 context) ranges from -51 to -99 basis points.
- Interpretation of SDDS effect:
  - Estimated SDDS “discount” of about 75 basis points (Equation 1), representing a reduction of about 19 percent of the average sample spread over 1996:3 to 2002:4.
  - SDDS coefficient is negative, strongly significant, and stable across recursive estimations and alternative specifications (see Figure 1: recursive SDDS coefficient estimates show stability as sample is expanded from the shortened sample to full sample).
- Other substantive findings:
  - Favorable macroeconomic performance (higher real GDP growth) narrows spreads (negative YDOT coefficient).
  - Longer maturities increase spreads (positive MAT coefficient).
  - Higher U.S. federal funds rate raises spreads (positive USFED coefficient), while higher long-term U.S. yields (USLONG) are negatively correlated with spreads—together implying a steeper U.S. yield curve is associated with lower emerging market spreads.
  - Higher external debt stock-to-exports (DXR) increases spreads.
  - Euro- and yen-denominated bonds exhibit systematically lower spreads relative to U.S. dollar-denominated bonds in the sample period (negative EURO and YEN coefficients); the EURO and YEN coefficients are statistically insignificantly different from each other.
  - Paris Club rescheduling episodes are associated with spreads widening by about 40 basis points (PARIS ≈ 38.56), while IMF program participation is associated with spreads declining by an estimated 27 basis points (IMF ≈ -27.04).

### Extensions and implications
- Evidence supports the existence of an SDDS-associated reduction in primary-market launch spreads across U.S. dollar, yen, and euro issues.  
- Recursive estimates indicate the SDDS coefficient is relatively stable over alternative sample windows.  
- Suggested extensions noted in the paper:
  - Compare SDDS discount between subscribers and non-subscribers as subscription becomes more widespread (does the discount diminish when SDDS subscription becomes the norm?).  
  - Investigate whether the IMF’s General Data Dissemination System (GDDS) also reduces international borrowing costs for participating countries.  
- Additional robustness checks performed:
  - Inclusion of a dummy for public release of IMF ROSCs yielded an insignificant ROSC coefficient and an SDDS estimate close to -75 basis points, supporting interpretation that the SDDS coefficient reflects SDDS subscription rather than other IMF transparency initiatives.

### CONCLUSIONS
- Main findings:
  - The paper presents econometric evidence of a SDDS discount in primary international capital markets for sovereign countries issuing foreign currency-denominated bonds.
  - Specific bond characteristics, global monetary conditions, and a country’s fundamental macroeconomic and debt situation remain the primary determinants of sovereign borrowing costs.
  - Subscription to the SDDS also appears to entail significant savings.
  - Based on sovereign foreign currency bond issues in primary capital markets over the period 1990 to 2002, the SDDS spread discount for a group of emerging market economies borrowing in the three principal global currencies is estimated to be about 75 basis points.
- Policy implications:
  - Macroeconomic and public debt fundamentals are of primary importance in the determination of international borrowing costs.
  - Subscription to the SDDS can entail important economies to the sovereign borrower and ultimately its taxpayers.
- Note on the GDDS:
  - The GDDS also guides member countries in the dissemination of their economic and financial data to the public.
  - Introduced in 1997, the GDDS is open to all IMF member counties, but not intended for those seeking access to international capital markets.

### Appendix I — Time-series properties (selected ADF test statistics and orders of integration)
- SP (yield spread) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -4.22 *** (c,t) — I(0)
  - Brazil: -3.40 * (c,t) — I(0)
  - Colombia: -3.77 ** (c,t) — I(0)
  - Mexico: -2.85 * (c) — I(0)
  - Philippines: -2.64 * (c) — I(0)
  - South Africa: -2.69 * (c) — I(0)
  - Turkey: -3.78 ** (c,t) — I(0)
- YDOT (real growth) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -3.61 ** (c) — I(0)
  - Brazil: -6.67 *** (c,t) — I(0)
  - Colombia: -4.74 *** (c) — I(0)
  - Mexico: -4.91 *** (c) — I(0)
  - Philippines: -2.98 ** (c) — I(0)
  - South Africa: -4.66 *** (c,t) — I(0)
  - Turkey: -2.65 * (c) — I(0)
- MAT (maturity) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -4.89 *** (c) — I(0)
  - Brazil: -3.16 ** (c) — I(0)
  - Colombia: -4.35 *** (c) — I(0)
  - Mexico: -3.77 *** (c) — I(0)
  - Philippines: -3.03 ** (c) — I(0)
  - South Africa: -4.31 *** (c,t) — I(0)
  - Turkey: -4.07 *** (c) — I(0)
- DXR (debt-exports ratio) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -3.63 ** (c,t) — I(0)
  - Brazil: -5.21 *** (c,t) — I(0)
  - Colombia: -4.65 *** (c,t) — I(0)
  - Mexico: -4.77 *** (c,t) — I(0)
  - Philippines: -3.89 *** (c) — I(0)
  - South Africa: -1.97 ** (n) — I(0)
  - Turkey: -3.91 *** (c) — I(0)
- USFED (federal funds rate) — raw series:
  - -3.30 ** (c) — I(0)
  - -2.68 * (c) — I(0)
- USLONG (10-year bond rate) — raw series:
  - -3.80 ** (c,t) — I(0)
  - -3.70 ** (c,t) — I(0)

*Source: Model estimates reported in the chapter "1. Recursive Estimates of the SDDS Coefficient."*

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

### _wp0458 - References

### References
- References..........................................................................................................................13

### Tables
- 1. SDDS Data Categories and Related Periodicity and Timeliness Standards ....................4
- 2. Panel Estimation Results..................................................................................................9

### Appendix Table
- A1.      Time-Series      Properties of the Macroeconomic and Debt Variables......................12

### Figure
- Figure (unnumbered/unspecified)

### 1. Recursive Estimates of the SDDS Coefficient.......................................................11

### 1. Recursive Estimates of the SDDS Coefficient

### Introduction
- SDDS introduced in 1996 to guide countries with or seeking access to international capital markets in the provision of economic and financial data; four dimensions: access, integrity, quality, and the data themselves.  
- SDDS prescribes monitorable minimum timeliness and frequency standards across 18 data categories covering real, fiscal, financial, and external sectors (summarized in Table 1).  
- Prior secondary-market studies report SDDS-associated reductions in EMBI spreads: examples include 200–300 basis points (IIF, 2002), about 15 percent (Christofides, Mulder, and Tiffin, 2003), and 4–12 percent (Glennerster and Shin, 2003).  
- This paper examines primary-market sovereign bond issuance and asks whether SDDS subscription lowers launch spreads for sovereign bonds denominated in U.S. dollar, yen, and euro.

### Data and estimation methodology
- Countries in panel: Argentina, Brazil, Colombia, Mexico, the Philippines, South Africa, and Turkey.  
  - Colombia and Brazil subscribed in May 1996 and March 2001, respectively; the remaining five countries subscribed in August 1996.  
  - Estimation period generally ranges from 1990 to 2002, starting approximately six years prior to, and ending six years after, mid-1996. Specific start dates: Argentina (1994:1), Brazil (1992:4), Colombia (1995:1) for parts of the sample.  
- Data sources:
  - Sovereign bond characteristics from the Bonds, Equities and Loans (BEL) database of the IMF (sourced from Capital Data).  
  - Macroeconomic data from IMF’s International Financial Statistics (IFS).  
  - External debt indicators from World Bank’s Global Development Finance (GDF).  
- Sample restrictions:
  - Only fixed interest rate bonds denominated in U.S. dollars, yen, and euros included (spread data available only for these).  
  - Bond maturities in sample range from 1 to 30 years (sample period mean is 7 ½ years).  
- Explanatory variables considered include: real GDP growth (YDOT), investment/GDP, inflation, short- and long-term global interest rates (USFED, USLONG), fiscal and external current account balances (% of GDP), public external debt stock-to-GDP and exports ratios (DXR), debt service-to-exports, short-term external debt relative to reserves, Paris Club rescheduling (PARIS), IMF program status (IMF), bond maturity (MAT), and currency dummies (EURO, YEN).  
- SDDS influence captured by a dummy variable (SDDS = 0 prior to subscription, = 1 thereafter); subscription dates treated as exogenous.  
- Time series stationarity: ADF tests indicate all series used in reported regressions are stationary at conventional significance levels (Appendix I, Table A1). Annual external debt stock-to-exports ratios converted to quarterly frequency and smoothed with Hodrick-Prescott filter prior to testing.  
- Estimation technique: pooled time series cross-section estimation with fixed effects correcting for equation-specific serial correlation and cross-section heteroscedasticity. Smoothed (four quarter moving average) real GDP growth and U.S. interest rates used.

### Estimation results — main findings
- Basic specification (Equation (1), estimation period 1990:2 to 2002:4) produces a relatively high degree of fit; coefficients generally of expected sign and statistically significant.
- Key estimated coefficient results from Table 2 (Equation 1):
  - YDOT (real growth): -6.61 (t = -2.88)
  - MAT (maturity): 3.11 (t = 4.38)
  - DXR (debt-exports ratio): 1.37 (t = 1.97)
  - USFED (federal funds rate): 0.16 (t = 2.21)
  - USLONG (10-year bond rate): -0.63 (t = -5.29)
  - EURO (euro-denominated): -94.69 (t = -8.60)
  - YEN (yen-denominated): -126.26 (t = -11.27)
  - PARIS (Paris Club): 38.56 (t = 2.19)
  - IMF (IMF arrangement): -27.04 (t = -1.82)
  - SDDS (subscription): -74.86 (t = -3.97)
- Additional model statistics (Equation 1):
  - Adjusted R2: 0.797
  - Durbin-Watson: 2.24
  - Number of observations: 303
  - Mean of the dependent variable: 340.30
- Robustness and alternative specifications:
  - Equation (2) (common constant term) yields SDDS coefficient = -82.56 (t = -4.36); Adjusted R2: 0.793; observations: 303; mean dependent variable: 304.30.
  - Equation (3) (shortened sample period 1992:3 to 2000:3) yields SDDS coefficient = -65.49 (t = -3.15); Adjusted R2: 0.740; observations: 205; mean dependent variable: 322.16.
  - A 90 percent confidence interval for the SDDS estimate (Equation 1 context) ranges from -51 to -99 basis points.
- Interpretation of SDDS effect:
  - Estimated SDDS “discount” of about 75 basis points (Equation 1), representing a reduction of about 19 percent of the average sample spread over 1996:3 to 2002:4.
  - SDDS coefficient is negative, strongly significant, and stable across recursive estimations and alternative specifications (see Figure 1: recursive SDDS coefficient estimates show stability as sample is expanded from the shortened sample to full sample).
- Other substantive findings:
  - Favorable macroeconomic performance (higher real GDP growth) narrows spreads (negative YDOT coefficient).
  - Longer maturities increase spreads (positive MAT coefficient).
  - Higher U.S. federal funds rate raises spreads (positive USFED coefficient), while higher long-term U.S. yields (USLONG) are negatively correlated with spreads—together implying a steeper U.S. yield curve is associated with lower emerging market spreads.
  - Higher external debt stock-to-exports (DXR) increases spreads.
  - Euro- and yen-denominated bonds exhibit systematically lower spreads relative to U.S. dollar-denominated bonds in the sample period (negative EURO and YEN coefficients); the EURO and YEN coefficients are statistically insignificantly different from each other.
  - Paris Club rescheduling episodes are associated with spreads widening by about 40 basis points (PARIS ≈ 38.56), while IMF program participation is associated with spreads declining by an estimated 27 basis points (IMF ≈ -27.04).

### Extensions and implications
- Evidence supports the existence of an SDDS-associated reduction in primary-market launch spreads across U.S. dollar, yen, and euro issues.  
- Recursive estimates indicate the SDDS coefficient is relatively stable over alternative sample windows.  
- Suggested extensions noted in the paper:
  - Compare SDDS discount between subscribers and non-subscribers as subscription becomes more widespread (does the discount diminish when SDDS subscription becomes the norm?).  
  - Investigate whether the IMF’s General Data Dissemination System (GDDS) also reduces international borrowing costs for participating countries.  
- Additional robustness checks performed:
  - Inclusion of a dummy for public release of IMF ROSCs yielded an insignificant ROSC coefficient and an SDDS estimate close to -75 basis points, supporting interpretation that the SDDS coefficient reflects SDDS subscription rather than other IMF transparency initiatives.

*Source: Model estimates reported in the chapter "1. Recursive Estimates of the SDDS Coefficient."*

### CONCLUSIONS

### CONCLUSIONS

### Main findings
- The paper presents econometric evidence of a SDDS discount in primary international capital markets for sovereign countries issuing foreign currency-denominated bonds.
- Specific bond characteristics, global monetary conditions, and a country’s fundamental macroeconomic and debt situation remain the primary determinants of sovereign borrowing costs.
- Subscription to the SDDS also appears to entail significant savings.
- Based on sovereign foreign currency bond issues in primary capital markets over the period 1990 to 2002, the SDDS spread discount for a group of emerging market economies borrowing in the three principal global currencies is estimated to be about 75 basis points.

### Policy implications
- Macroeconomic and public debt fundamentals are of primary importance in the determination of international borrowing costs.
- Subscription to the SDDS can entail important economies to the sovereign borrower and ultimately its taxpayers.

### Note on the GDDS
- The GDDS also guides member countries in the dissemination of their economic and financial data to the public.
- Introduced in 1997, the GDDS is open to all IMF member counties, but not intended for those seeking access to international capital markets.

### Appendix I — Time-series properties (selected ADF test statistics and orders of integration)
- SP (yield spread) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -4.22 *** (c,t) — I(0)
  - Brazil: -3.40 * (c,t) — I(0)
  - Colombia: -3.77 ** (c,t) — I(0)
  - Mexico: -2.85 * (c) — I(0)
  - Philippines: -2.64 * (c) — I(0)
  - South Africa: -2.69 * (c) — I(0)
  - Turkey: -3.78 ** (c,t) — I(0)
- YDOT (real growth) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -3.61 ** (c) — I(0)
  - Brazil: -6.67 *** (c,t) — I(0)
  - Colombia: -4.74 *** (c) — I(0)
  - Mexico: -4.91 *** (c) — I(0)
  - Philippines: -2.98 ** (c) — I(0)
  - South Africa: -4.66 *** (c,t) — I(0)
  - Turkey: -2.65 * (c) — I(0)
- MAT (maturity) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -4.89 *** (c) — I(0)
  - Brazil: -3.16 ** (c) — I(0)
  - Colombia: -4.35 *** (c) — I(0)
  - Mexico: -3.77 *** (c) — I(0)
  - Philippines: -3.03 ** (c) — I(0)
  - South Africa: -4.31 *** (c,t) — I(0)
  - Turkey: -4.07 *** (c) — I(0)
- DXR (debt-exports ratio) — Augmented Dickey-Fuller Test Statistic / Order of Integration:
  - Argentina: -3.63 ** (c,t) — I(0)
  - Brazil: -5.21 *** (c,t) — I(0)
  - Colombia: -4.65 *** (c,t) — I(0)
  - Mexico: -4.77 *** (c,t) — I(0)
  - Philippines: -3.89 *** (c) — I(0)
  - South Africa: -1.97 ** (n) — I(0)
  - Turkey: -3.91 *** (c) — I(0)
- USFED (federal funds rate) — raw series:
  - -3.30 ** (c) — I(0)
  - -2.68 * (c) — I(0)
- USLONG (10-year bond rate) — raw series:
  - -3.80 ** (c,t) — I(0)
  - -3.70 ** (c,t) — I(0)

Notes: Asterisks indicate rejection of the null hypothesis that the series has a unit root at the 10 percent level (*), the 5 percent level (**), and the 1 percent level (***). The critical values of the ADF test statistics are from MacKinnon (1996). The bracketed information indicates the inclusion of exogenous variables in the ADF test equation: c=constant, t=trend, n=none; lag lengths in test equations determined using the Akaike Information Criterion.

*Source: _wp0458 - CONCLUSIONS*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0458.pdf_