## Appendix I: Main Data Sources and Description

## Source details

**Canonical URL:** [Appendix I: Main Data Sources and Description](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0678.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0678.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0678.pdf.json)

---

### I. Study objective and context
- Objective: examine impact of the IMF’s data standards initiatives—the Special Data Dissemination Standard (SDDS) and the General Data Dissemination System (GDDS)—on sovereign borrowing costs for emerging market and developing countries (EMCs) issuing sovereign bonds.
- Context: SDDS guides countries with or seeking access to international capital markets; GDDS guides statistical development for countries not yet aspiring to SDDS.
- Key empirical claim:
  - GDDS discount: about 8 percent, equivalent to about 20 basis points.
  - SDDS discount: about 20 percent, equivalent to about 50 basis points.
- SDDS approved March 1996; GDDS approved December 1997.
- Participation (end-2005): SDDS subscription = 62 countries; GDDS participation = 88 countries; 5 GDDS graduates to SDDS.

### II. Data sources and sample
- Panel: some 320 sovereign bonds issued by 26 EMCs and developing countries over 1989–2004; unbalanced panel.
  - Maturities: ranged from 1 to 30 years; median = 7 years.
  - Currency composition: 55 yen-denominated bonds (17 percent), 97 euro-denominated bonds (30 percent); remainder U.S. dollar.
  - Sample coverage: 24 of 26 countries accounted for an average of 68 percent of value of all new bond issues by EMCs and developing countries during 2000–2004.
- Data sources:
  - Bond characteristics and issuance data: IMF’s Bonds, Equities, Loans (BEL) database (sourced from Dealogic).
  - Macroeconomic variables: IMF International Financial Statistics and World Economic Outlook.
  - External debt stocks: World Bank Global Development Finance.
  - Sovereign credit ratings: Standard and Poor’s, Moody’s, Fitch—alphanumeric ratings transformed to numerical ratings (mapping provided in source).
  - Institutional quality: International Country Risk Guide (PRS Group Inc.).
  - SDDS/GDDS/IMF program status: IMF records; IMF DSBB website for SDDS/GDDS dates.
- Totals (sample counts):
  - 26 Countries
  - Macro variable sample 317
  - Credit rating sample 322

### III. Dependent variables and empirical specification
- Dependent variables:
  - Launch spread (SPi,t) = annual yield to maturity at launch minus “risk-free” benchmark yield (industrial country government bond of same currency and maturity).
  - Launch yield (YLDi,t).
- Core log-linear specification: ln(Ci,t) = f(Xi,t) + ui,t where Ci,t is SP or YLD and Xi,t includes bond characteristics, macro indicators or credit rating, institutional quality, SDDS and GDDS dummies, IMF program dummy, time trend.
- SDDS and GDDS dummies: zero prior to subscription/participation, one in quarter of subscription/participation and thereafter.
- Macroeconomic regressors:
  - Real GDP growth (YDOT)
  - Inflation differential vis-à-vis the United States (DPDOT)
  - Primary fiscal balance change (∆GPBAL)
  - Debt-export ratio (DXR)
  - Alternative specification replaces macro variables with credit rating (CR)
- Estimation methods:
  - Seemingly Unrelated Regression (SUR) with serial correlation correction.
  - Pooled least squares with fixed effects as robustness check.
  - Panel unit root tests: stationarity rejected for credit ratings only.
  - Granger causality tests: exogeneity of maturity accepted for all but four countries.

### IV. Core model equations (as in source)
- Spread equation (macro indicators):
  - ln(SPi,t) = β0 + β1 YDOTi,t + β2 DPDOTi,t + β3 (∆GPBALi,t) + β4 ln(DXRi,t) + β5 ln(MATi,t) + β6 ln(INSTi,t) + β7 YENi,t + β8 EUROi,t + β9 IMFi,t + β10 SDDSi,t + β11 GDDSi,t + β12 TIME + ui,t
- Spread equation (credit ratings):
  - ln(SPi,t) = β0 + β1 ln(CRi,t) + β2 ln(MATi,t) + β3 ln(INSTi,t) + β4 YENi,t + β5 EUROi,t + β6 IMFi,t + β7 SDDSi,t + β8 GDDSi,t + β9 TIME + ui,t
- Yield equation (macro indicators):
  - ln(YLDi,t) = β0 + β1 YDOTi,t + β2 DPDOTi,t + β3 (∆GPBALi,t) + β4 ln(DXRi,t) + β5 ln(INTRi,t) + β6 ln(MATi,t) + β7 ln(INSTi,t) + β8 YENi,t + β9 EUROi,t + β10 IMFi,t + β11 SDDSi,t + β12 GDDSi,t + β13 TIME + ui,t

### V. Launch spreads — principal econometric findings
- GDDS participation:
  - SUR point estimate implies GDDS participation reduces launch spreads by over 9 percent, or 23 basis points on an illustrative total spread of 250 basis points (first column, Table 3).
  - Across specifications, GDDS estimated to reduce spreads by 20 to 35 basis points.
  - Truncated balanced-sample SUR estimate for 11 countries: GDDS spread reduction over 40 basis points.
- SDDS subscription:
  - Statistically significant negative coefficient; estimated to reduce launch spreads by 12 to 20 percent, or 30 to 50 basis points on a total spread of 250 basis points.
  - Balanced-sample SUR with fixed effects: estimated spread reduction for SDDS of 58 basis points using macro indicators and 31 percent using credit ratings.
- Other determinants (spread-equation magnitudes):
  - Real GDP growth: a ½ percentage point higher growth implied lower spreads in the range of 35 to 70 basis points.
  - Primary fiscal balance: improvement of ½ percentage point of GDP reduces spreads by 60 to 100 basis points.
  - Debt-export ratio: decline from 50 to 40 percent reduces spreads by 13 to 23 basis points.
  - Inflation differential (DPDOT): statistically insignificant, negligible magnitude.
  - Institutional quality (INST): one-standard-deviation increase around mean reduces spreads by 35 to 42 basis points.
  - Credit rating (CR): one full-notch upgrade reduces spreads by 38 basis points (SUR full panel) and 25 basis points (pooled LS with fixed effects).
  - Maturity (MAT): increase from 5 to 10 years increases spread by about 5 basis points.
  - Currency effects: yen and euro bonds have significantly lower spreads than dollar-denominated bonds.
  - IMF-supported program (IMF): when effective, launch spreads decline by about 10 basis points.
- Robustness:
  - GDDS and SDDS coefficients stable over different time periods and specifications.
  - Inclusion of institutional quality controls for concurrent institutional improvements.

### VI. Launch yields — principal econometric findings
- Yield estimates consistent with spread results; magnitudes reflect scale differences between yields and spreads.
- SDDS and GDDS discounts in yield terms:
  - SDDS: about 50 basis points.
  - GDDS: about 20 basis points.
- Main yield-equation magnitudes:
  - Real GDP growth: ½ percentage point increase reduces launch yield by 40 to 65 basis points.
  - Primary fiscal balance: improvement of ½ percentage point of GDP reduces launch yield by 55 to 110 basis points.
  - Debt-export ratio: decline from 50 to 40 percent reduces launch yield by 25 to 35 basis points.
  - Credit rating: one-notch upgrade reduces launch yields by about 40 basis points.
  - Institutional quality: one-standard-deviation improvement reduces launch yields by 25 to 35 basis points.
  - IMF program: approval reduces launch yields by an estimated 10 to 20 basis points.
- Memorandum point estimates (evaluated at illustrative spreads/yields):
  - Spread-equation point estimates (evaluated at illustrative spread of 250 basis points):
    - SDDS: 48.50, 30.50, 34.75, 44.00 (across columns 1–4, Table 3).
    - GDDS: 23.25, 33.75, 19.00, 29.00 (columns 1–4, Table 3).
  - Yield-equation point estimates (evaluated at illustrative yield of 750 basis points):
    - SDDS: 66.00, 45.75, 45.75, 48.75 (columns 1–4, Table 4).
    - GDDS: 25.50, 38.25, 18.75, 28.50 (columns 1–4, Table 4).

### VII. Institutional quality, rating mappings, and treatment of outlooks/reviews
- Institutional quality index = sum of law and order and bureaucratic quality components of the ICRG’s overall political risk rating.
  - Law and order indicator ranges from 1 to 6.
  - Bureaucratic quality indicator ranges from 0 to 4.
  - In sample:
    - Institutional quality variable varies from zero to 10
    - Mean of 5.3
    - Standard deviation of 2.2
- Sovereign rating numerical mapping and treatment of outlooks/reviews:
  - The adjustment for outlooks and watches/reviews is 0.2 for positive outlook and watchs/review qualifications while negative outlook or watchs/review are increased 0.2 each.
  - Example mapping (exact values from source):
    - A+ rating from S&P and Fitch → numerical value of 5
    - A+ with a positive outlook → 4.8
    - A+ with a positive review → 4.6
    - A+ with a negative outlook → 5.2
    - A+ with a negative review → 5.4
- Empirical note: experimentation with adding other ICRG indicators produced only negligible and statistically insignificant variation in empirical results for the GDDS and SDDS variables; law and order and bureaucratic quality produced the most robust estimates for launch spreads and yields.

### VIII. Table A2 — selected SDDS and GDDS sample excerpts (exact strings)
- Selected SDDS entries (exact strings from source):
  - Argentina August 16, 1996 1994:2 to 2002:4 1992:3 to 2002:4 24 24
  - Brazil March 14, 2001 1995:3 to 2002:4 1995:3 to 2002:4 16 16
  - Mexico August 13, 1996 1991:2 to 2002:4 1991:2 to 2002:4 24 24
  - Turkey August 8, 1996 1990:2 to 2002:4 1992:3 to 2002:4 34 34
- Selected GDDS entries (exact strings from source):
  - Barbados May 22, 2000 1994:3 to 2003:4 1995:1 to 2004:4 4 2
  - China, People’s Republic of April 15, 2002 1994:1 to 2000:4 1994:1 to 2002:4 12 13
  - Lebanon January 16, 2003 1994:4 to 2003:4 1997:2 to 2004:4 22 24
  - Romania February 14, 2001 Graduated to SDDS May 2005 1996:3 to 2001:2 1996:3 to 2002:4 7 7

### IX. Conclusions and policy implications
- Participation in the IMF’s data standards initiatives provides measurable cost savings to sovereign borrowers beyond fundamentals:
  - GDDS participation associated with a small but statistically significant interest rate discount.
  - SDDS subscription associated with a larger discount than GDDS.
- Financial incentives exist for sovereign borrowers to participate in GDDS and even larger incentives to subscribe to SDDS.
- For the IMF:
  - Maintaining credibility of the SDDS as a monitored standard is critical because lower sovereign borrowing costs for subscribers depend on observance of the standard.
  - The Fund plans to issue annual reports on SDDS observance beginning in 2007 (policy action cited in source).
- Cautions:
  - GDDS participation should not be interpreted as by itself granting market access—11 GDDS participants in study had prior market access.
  - Investors may view both SDDS subscription and GDDS participation as signals of reduced uncertainty about data reliability and serviceability; the larger SDDS discount is consistent with its more stringent and monitored requirements.

*Source: _wp0678 - Appendix I: Main Data Sources and Description (excerpts and section summary as provided).*

### Appendix I: Main Data Sources and Description ....................................................................20

### Appendix I: Main Data Sources and Description

### Appendix Tables
- A1. Alphanumeric Credit Ratings and Equivalent Numerical Ratings ...................................................................20
- A2. SDDS Subscription and GDDS Participation Dates, Sample Periods, and Number of Bonds in Panel .........................................................................................................................22

### Text Tables
- 1. Granger Causality Tests: Launch Spreads and Maturity .....................................................16
- 2. Panel Unit Root Tests ..........................................................................................................17
- 3. Panel Estimation of Spread Equations: Macro Variables and Credit Ratings .....................18
- 4. Panel Estimation of Yield Equations: Macro Variables and Credit Ratings .......................19

### Figures
- 1. Number of SDDS Subscribers and GDDS Participants.......................................................14

### References
- References ................................................................................................................................23

*Source: _wp0678 - Appendix I: Main Data Sources and Description (pages and item listings as provided).*

### 2. Recursive SDDS and GDDS Coefficient Estimates ............................................................15

### 2. Recursive SDDS and GDDS Coefficient Estimates

### I. Introduction
- Objective: examine impact of the IMF’s data standards initiatives—the Special Data Dissemination Standard (SDDS) and the General Data Dissemination System (GDDS)—on sovereign borrowing costs for emerging market and developing countries (EMCs) issuing sovereign bonds.
- Context: SDDS guides countries with or seeking access to international capital markets; GDDS guides statistical development for countries not yet aspiring to SDDS.
- Key empirical claim: strong and consistent econometric evidence of discounts for sovereign issuers participating in the GDDS and for EMCs subscribing to the SDDS.
  - GDDS discount: about 8 percent, equivalent to about 20 basis points.
  - SDDS discount: about 20 percent, equivalent to about 50 basis points.
- Results are consistent across modeling approaches, stable over time, and broadly in line with prior studies (but lower than the Institute for International Finance (2002) estimate of 200–300 basis points).

### II. Background
- SDDS approved March 1996; GDDS approved December 1997.
- SDDS requires metadata, advance release calendars, National Summary Data Page (NSDP), posting on the Dissemination Standards Bulletin Board (DSBB).
- GDDS requires metadata, statistical practices and development plans, annual metadata updates.
- Participation (end-2005): SDDS subscription = 62 countries; GDDS participation = 88 countries; 5 GDDS graduates to SDDS.
- Fund activities: outreach, technical assistance, monitoring; alignment with the Data Quality Assessment Framework (DQAF).
- Rationale: improved timeliness and quality of data reduce uncertainty and can lower risk premia; market participants view SDDS and GDDS as useful (Financial Stability Forum 2002).

### III. Data and Empirical Approach
- Panel: some 320 sovereign bonds issued by 26 EMCs and developing countries over 1989–2004; unbalanced panel reflecting issuance histories and data availability.
  - Maturities: ranged from 1 to 30 years; median = 7 years.
  - Currency composition: 55 yen-denominated bonds (17 percent), 97 euro-denominated bonds (30 percent); remainder U.S. dollar.
  - Sample coverage: 24 of 26 countries accounted for an average of 68 percent of value of all new bond issues by EMCs and developing countries during 2000–2004.
- Data sources:
  - Bond characteristics and issuance data: IMF’s Bonds, Equities, Loans (BEL) database.
  - Macroeconomic variables: IMF International Financial Statistics and World Economic Outlook; external debt stocks from World Bank Global Development Finance.
  - Sovereign credit ratings: Standard and Poor’s, Moody’s, Fitch—alphanumeric ratings transformed to numerical ratings (Table 1A mapping provided).
  - Institutional quality: International Country Risk Guide (PRS Group Inc.).
  - SDDS/GDDS/IMF program status: IMF records.
- Dependent variables:
  - Launch spread (SPi,t) = annual yield to maturity at launch minus “risk-free” benchmark yield (industrial country government bond of same currency and maturity).
  - Launch yield (YLDi,t).
- Core empirical specification: log-linear model ln(Ci,t) = f(Xi,t) + ui,t where Ci,t is SP or YLD and Xi,t includes bond characteristics, macro indicators or credit rating, institutional quality, SDDS and GDDS dummies, IMF program dummy, time trend.
- SDDS and GDDS dummies: zero prior to subscription/participation, one in quarter of subscription/participation and thereafter.
- Macroeconomic variables included: real GDP growth (YDOT), inflation differential vis-à-vis the United States (DPDOT), primary fiscal balance change (∆GPBAL), debt-export ratio (DXR). Alternative specification replaces macro variables with credit rating (CR).
- Estimation methods:
  - Seemingly Unrelated Regression (SUR) with serial correlation correction; pooled least squares with fixed effects as robustness check.
  - Panel unit root tests: stationarity rejected for credit ratings only (Table 2); cointegration not an issue for panel.
  - Granger causality tests between spreads and maturities: exogeneity of maturity accepted for all but four countries (Table 1).

### III.A Sovereign Borrowing Cost Model (formulas retained from source)
- Spread equation (macro indicators):
  - ln(SPi,t) = β0 + β1 YDOTi,t + β2 DPDOTi,t + β3 (∆GPBALi,t) + β4 ln(DXRi,t) + β5 ln(MATi,t) + β6 ln(INSTi,t) + β7 YENi,t + β8 EUROi,t + β9 IMFi,t + β10 SDDSi,t + β11 GDDSi,t + β12 TIME + ui,t
- Spread equation (credit ratings):
  - ln(SPi,t) = β0 + β1 ln(CRi,t) + β2 ln(MATi,t) + β3 ln(INSTi,t) + β4 YENi,t + β5 EUROi,t + β6 IMFi,t + β7 SDDSi,t + β8 GDDSi,t + β9 TIME + ui,t
- Yield equation (macro indicators):
  - ln(YLDi,t) = β0 + β1 YDOTi,t + β2 DPDOTi,t + β3 (∆GPBALi,t) + β4 ln(DXRi,t) + β5 ln(INTRi,t) + β6 ln(MATi,t) + β7 ln(INSTi,t) + β8 YENi,t + β9 EUROi,t + β10 IMFi,t + β11 SDDSi,t + β12 GDDSi,t + β13 TIME + ui,t

### III.B Launch Spreads — Key econometric findings
- GDDS participation:
  - SUR point estimate implies GDDS participation reduces launch spreads by over 9 percent, or 23 basis points on an illustrative total spread of 250 basis points (first column, Table 3).
  - Across specifications (SUR and pooled LS with fixed effects; macro indicators vs. credit rating), GDDS estimated to reduce spreads by 20 to 35 basis points.
  - Truncated balanced-sample SUR estimate for 11 countries: GDDS spread reduction over 40 basis points.
- SDDS subscription:
  - Statistically significant negative coefficient; estimated to reduce launch spreads by 12 to 20 percent, or 30 to 50 basis points on a total spread of 250 basis points (Table 3).
  - Balanced-sample SUR with fixed effects: estimated spread reduction for SDDS of 58 basis points using macro indicators and 31 percent using credit ratings.
- Other determinants and magnitudes (spread equation):
  - Real GDP growth: a ½ percentage point higher growth implied lower spreads in the range of 35 to 70 basis points.
  - Primary fiscal balance: improvement of ½ percentage point of GDP reduces spreads by 60 to 100 basis points.
  - Debt-export ratio: decline from 50 to 40 percent reduces spreads by 13 to 23 basis points.
  - Inflation differential (DPDOT): statistically insignificant, negligible magnitude.
  - Institutional quality (INST): one-standard-deviation increase around mean reduces spreads by 35 to 42 basis points.
  - Credit rating (CR): one full-notch upgrade (e.g., Baa1 to A3 in Moody’s mapping) reduces spreads by 38 basis points (SUR full panel) and 25 basis points (pooled LS with fixed effects).
  - Maturity (MAT): increase from 5 to 10 years increases spread by about 5 basis points.
  - Currency effects: yen and euro bonds have significantly lower spreads than dollar-denominated bonds (yen and euro dummy coefficients highly significant).
  - IMF-supported program (IMF): when effective, launch spreads decline by about 10 basis points.
- Robustness:
  - GDDS and SDDS coefficients stable over different time periods (Figure 2) and specifications.
  - Inclusion of institutional quality controls for concurrent institutional improvements.

### III.C Launch Yields — Key econometric findings
- Yield estimates consistent with spread results; magnitudes reflect scale differences between yields and spreads.
- SDDS and GDDS discounts in yield terms (basis points), closer to spread-equation estimates using credit ratings:
  - SDDS: about 50 basis points.
  - GDDS: about 20 basis points.
- Benchmark international yield (INTR) included and highly significant; specification supports view of yield = benchmark international yield + country spread.
- Main yield equation results summarized:
  - Real GDP growth: ½ percentage point increase reduces launch yield by 40 to 65 basis points.
  - Primary fiscal balance: improvement of ½ percentage point of GDP reduces launch yield by 55 to 110 basis points.
  - Debt-export ratio: decline from 50 to 40 percent reduces launch yield by 25 to 35 basis points.
  - Credit rating: one-notch upgrade from adequate to strong payment capacity reduces launch yields by about 40 basis points.
  - Institutional quality: one-standard-deviation improvement reduces launch yields by 25 to 35 basis points.
  - IMF program: approval reduces launch yields by an estimated 10 to 20 basis points.
- Memorandum point estimates (from Tables):
  - Spread-equation point estimates (evaluated at illustrative spread of 250 basis points):
    - SDDS: 48.50, 30.50, 34.75, 44.00 (across columns 1–4, Table 3).
    - GDDS: 23.25, 33.75, 19.00, 29.00 (columns 1–4, Table 3).
  - Yield-equation point estimates (evaluated at illustrative yield of 750 basis points):
    - SDDS: 66.00, 45.75, 45.75, 48.75 (columns 1–4, Table 4).
    - GDDS: 25.50, 38.25, 18.75, 28.50 (columns 1–4, Table 4).

### IV. Conclusions and Policy Implications
- Participation in the IMF’s data standards initiatives provides measurable cost savings to sovereign borrowers beyond fundamentals:
  - GDDS participation associated with a small but statistically significant interest rate discount.
  - SDDS subscription associated with a larger discount than GDDS.
- Financial incentives exist for sovereign borrowers to participate in GDDS and even larger incentives to subscribe to SDDS.
- For the IMF:
  - Maintaining credibility of the SDDS as a monitored standard is critical because lower sovereign borrowing costs for subscribers depend on observance of the standard.
  - The Fund plans to issue annual reports on SDDS observance beginning in 2007 (policy action cited in source).
- Cautions and interpretations:
  - GDDS participation should not be interpreted as by itself granting market access—11 GDDS participants in study had prior market access.
  - Investors may view both SDDS subscription and GDDS participation as signals of reduced uncertainty about data reliability and serviceability; the larger SDDS discount is consistent with its more stringent and monitored requirements.
- Broader implication: technical assistance and monitoring that promote high-quality data dissemination may contribute to lower sovereign borrowing costs, reinforcing a virtuous cycle between financial incentives and data quality improvements (Fischer 2002 quotation referenced).

*Source: IMF working paper text (Section 2. Recursive SDDS and GDDS Coefficient Estimates).*

### 0.2 for positive outlook and watchs/review qualifications while negative outlook or

### _wp0678 - 0.2 for positive outlook and watchs/review qualifications while negative outlook or

### Sovereign rating numerical mapping and treatment of outlooks/reviews
- The adjustment for outlooks and watches/reviews is 0.2 for positive outlook and watchs/review qualifications while negative outlook or watchs/review are increased 0.2 each.
- Example mapping (exact values from source):
  - A+ rating from S&P and Fitch → numerical value of 5
  - A+ with a positive outlook → 4.8
  - A+ with a positive review → 4.6
  - A+ with a negative outlook → 5.2
  - A+ with a negative review → 5.4

### Data sources used
- International Monetary Fund DSBB website for GDDS participation and SDDS subscription, and Fund record for the quarters in which a Fund arrangement became effective and expired.
- International Country Risk Guide, PRS Group Inc., for indicators of law and order and bureaucratic quality.
- IMF Statistics Department; and the IMF’s BEL database (sourced from Dealogic).

### Institutional quality index and variables (ICRG components)
- Institutional quality index = sum of law and order and bureaucratic quality components of the ICRG’s overall political risk rating.
- Law and order indicator ranges from 1 to 6.
- Bureaucratic quality indicator ranges from 0 to 4.
- In this study’s sample of countries and time period:
  - Institutional quality variable varies from zero to 10
  - Mean of 5.3
  - Standard deviation of 2.2
- Empirical note: experimentation with adding other ICRG indicators produced only negligible and statistically insignificant variation in the empirical results for the GDDS and SDDS variables; law and order and bureaucratic quality produced the most robust estimates for launch spreads and yields.

### Table A2 — SDDS Subscription and GDDS Participation (sample excerpts and totals)
- Selected SDDS entries (exact strings from source):
  - Argentina August 16, 1996 1994:2 to 2002:4 1992:3 to 2002:4 24 24
  - Brazil March 14, 2001 1995:3 to 2002:4 1995:3 to 2002:4 16 16
  - Mexico August 13, 1996 1991:2 to 2002:4 1991:2 to 2002:4 24 24
  - Turkey August 8, 1996 1990:2 to 2002:4 1992:3 to 2002:4 34 34
- Selected GDDS entries (exact strings from source):
  - Barbados May 22, 2000 1994:3 to 2003:4 1995:1 to 2004:4 4 2
  - China, People’s Republic of April 15, 2002 1994:1 to 2000:4 1994:1 to 2002:4 12 13
  - Lebanon January 16, 2003 1994:4 to 2003:4 1997:2 to 2004:4 22 24
  - Romania February 14, 2001 Graduated to SDDS May 2005 1996:3 to 2001:2 1996:3 to 2002:4 7 7
- Totals (exact values from source):
  - 26 Countries
  - Macro variable sample 317
  - Credit rating sample 322

### Empirical and methodological points
- The study experimented with including other components of the ICRG’s overall political risk rating but found law and order and bureaucratic quality to yield the most robust estimates for launch spreads and yields.
- Adding other ICRG indicators produced only negligible and statistically insignificant variation in empirical results for GDDS and SDDS variables.

*Source: excerpt from the IMF working paper PDF (_wp0678).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0678.pdf_