## _wp0427 - References

## Source details

**Canonical URL:** [_wp0427 - References](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0427.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0427.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0427.pdf.json)

---

### I. Introduction — market growth and role of credit derivatives
- Market expansion:
  - From US$1.8 billion in 1993 to US$2.0 trillion in 2002 (British Bankers’ Association survey).
  - Estimated total outstanding value will reach US$4.8 trillion by 2004 (BBA survey).
- Purpose and structure:
  - Credit derivatives permit trading of credit risk without transferring ownership of underlying assets; isolate credit risk from market risk.
  - Created by combining financing strategies such as swap contracts or pooled securities; provide insurance-like protection from credit/default risk.
- Market interaction implications:
  - Changes in a firm’s credit risk affect CDS prices, bond prices, and equity prices.
  - Distress increases default risk → bond prices and equity prices fall → CDS spreads rise.
  - Sellers of credit protection may hedge by shorting bonds or equity, exerting further downward pressure on prices.
- Theoretical and empirical context:
  - Duffie (1999) and Hull and White (2000): absent frictions, arbitrage forces CDS spreads to be approximately equal to underlying bond spreads; CDS and bond spreads positively correlated.
  - Merton (1974): firm value, debt, and equity link; default when asset value falls below debt face value; bond and equity prices positively correlated when default risk is high.
  - Empirical evidence: CDS (and equity) markets often lead bond markets in price discovery during deteriorations (Hull, Predescu, and White (2003); Blanco, Brennan, and Marsh (2003); Longstaff, Mithal, and Neiss (2003)).
- Study objective:
  - Examine equilibrium price relationships and price discovery among CDS, bond, and equity markets for eight emerging-market countries: Brazil, Bulgaria, Colombia, Mexico, the Philippines, Russia, Turkey, and Venezuela.
  - Attempt to extend Merton’s theory to sovereign issuers (appendix justification).

### II. Background — credit derivatives and CDS specifics
- Market overview and participants:
  - Credit derivatives combine derivatives and securitization to isolate and transfer credit risk at relatively low cost.
  - Standardization (ISDA Credit Derivatives Definitions) and broker trading (GFINet, CreditTrade) increased market activity.
  - Main participants: banks (largest buyers and sellers), insurance companies (second largest sellers), securities houses, hedge funds.
  - In September 2003, banks sold US$1.3 trillion of gross credit protection though they were net buyers of protection in the amount of US$229 billion; insurance companies sold protection in a net amount of US$303 billion (FitchRatings, 2003).
- Market size and emerging markets:
  - Global credit derivatives: about $2.0 trillion in 2002; in 1997 total global value was $180 billion; estimated growth to $4.8 trillion by end-2004.
  - Emerging market notional amounts outstanding in 2002: US$300 billion, roughly equal to 15 percent of the market (Deutsche Bank/Xu, 2003).
  - CDS liquidity depends on liquidity of sovereign bonds — countries with liquid bond markets (Mexico, Brazil, Colombia, Venezuela) have easier price quotes.
- Principal product types:
  - Four broad types: total return swaps, credit default swaps (CDSs), credit linked notes, synthetic collateralized debt obligations (CDO).
  - CDS mechanics:
    - Protection buyer pays periodic fee (default swap premium or default swap spread) expressed in basis points per notional; fee paid quarterly or semiannually.
    - Typical credit events: bankruptcy, insolvency, credit downgrade, failure to make a scheduled payment (ISDA definitions).
    - Settlement: physical settlement (delivery of reference security for par) or cash settlement (par minus recovery determined by dealer poll or price quote services); most contracts subject to physical settlement.
    - Typical contract sizes and maturities: notional amounts of US$5 million and US$10 million with a five-year maturity; trading activity greatest for five-year contracts for corporate/financial institutions.
  - Analogy: CDS akin to swapping payments from a risky security for a risk-free security plus a contingent default payment (see Duffie (1999)).

### III. Price relationships and theoretical framework
- CDS–bond parity:
  - Arbitrage: buying a risky bond and buying a CDS to insure default should eliminate spread differential; theory predicts one-to-one changes between CDS and bond spreads (Duffie (1999); Hull and White (2000)).
  - Required assumptions: par-floating rate securities, no transaction costs, negligible tax effects, CDS spread payments stop if credit event occurs, protection buyer paid on next coupon date after event.
  - Practical divergences:
    - Cheapest-to-deliver option: protection buyer can deliver the cheapest bond on credit event, causing sellers to charge a higher CDS spread (cheapest-to-deliver premium).
    - Relative liquidity differences: liquidity differentials can cause CDS and bond spreads to diverge.
- Bond–equity relationship (Merton 1974):
  - Equity analogous to a call option on firm assets with strike equal to debt face value; if assets < face value → default; bond and equity prices positively correlated, stronger with higher debt-to-asset ratios.
  - Implication: CDS spreads and equity prices should move in opposite directions if CDS↔bond and bond↔equity relationships hold.
- Extending Merton to sovereigns:
  - Sovereign-specific considerations: introduce “willingness-to-pay factor”; two critical asset values: “default value” and “no-payment value.”
  - For sovereigns: “default value” > face value of debt and “no-payment value” > zero due to continued asset value after cessation of debt payments.
  - Implication: for every asset value, default risk is higher for sovereigns than for comparable corporates; relationship between debt and equity prices remains qualitatively similar for sovereigns.

### IV. Empirical context, data, and methodology
- Countries analyzed: Brazil, Bulgaria, Colombia, Mexico, the Philippines, Russia, Turkey, Venezuela.
- Credit-rating context: these countries have foreign currency–denominated external debt that are below investment grade except Mexico, which has been rated investment–grade since March 2002.
- Data frequency and sample period:
  - Daily data for the period March 19, 2001 through May 29, 2003.
  - For cross-country comparisons, country bond and equity indexes used.
- Data sources and measurement:
  - CDS spreads: mid-price quotes on five-year contracts obtained from CreditTrade and Deutsche Bank. CreditTrade used for plots; Deutsche Bank used for empirical analysis.
  - Bond spreads: JPMorgan Chase Emerging Market Bond Index Plus (EMBI+).
  - Equity prices: U.S. dollar–denominated Morgan Stanley Composite Index (MSCI).
- Descriptive relationships (Figure 2 summary):
  - Bond spreads and equity prices: most countries exhibit an inverse relationship.
  - CDS spreads and equity prices: CDS spreads and equity prices are negatively correlated for most countries.
  - CDS spreads and bond spreads: most countries have positive correlation between these two markets.
- Empirical methodology:
  - Two-step cointegration approach:
    - Step 1: stationarity/unit root tests using Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests (null: unit root).
    - Step 2: Johansen (1991) cointegration rank test on a VAR of order p; cointegration if Π has reduced rank equal to 1 (Π = α βᵀ).
  - Granger causality tests via first-difference regressions; horizons tested: 1, 5, 10, and 20 business day lags.
  - Price discovery measures:
    - Hasbrouck (1995) lower and upper bounds from VMA long-run impact matrix Ψ(1).
    - Gonzalo and Granger (1995) GG statistic.
    - VECM specification: ∆Yt = α β′ Yt−1 + Σ Aj ∆Yt−j + e t; lag selection by Schwarz information criterion (SIC).
  - Interpretation: insignificance of an error-correction coefficient for one market suggests price discovery happens in the other market; significance in both implies shared price discovery.

### V. Unit root and cointegration test results
- Unit root test (ADF and PP) summary:
  - Adjusted t-statistics for all variables characterized by unit roots at the 90 percent and above confidence level, implying need for cointegration rank tests.
  - Reported adjusted t-statistics (ADF / PP) by country and variable as presented in source (exact figures preserved in original table).
- Johansen cointegration rank test (trace statistics) — key outcomes:
  - CDS and bond spreads are cointegrated in Brazil, Bulgaria, Colombia, Russia, and Venezuela (trace statistics reported: Brazil 37.33; Bulgaria 41.235; Colombia 64.4; Russia 21.637; Venezuela 20.081 — significant rejections of null of no cointegration).
  - No CDS–bond cointegration in Mexico, the Philippines, and Turkey.
  - Bond spreads and equity prices: cointegration found for Russia (17.874) and Venezuela (11).
  - CDS spreads and equity prices: cointegration found for Russia (18.646).
- Interpretations:
  - Existence of CDS–bond cointegration supports arbitrage forces converging CDS and bond spreads despite frictions.
  - Absence of cointegration in Mexico, Philippines, Turkey may reflect exploitable-arbitrage impediments or EMBI+ not being an appropriate proxy for 5-year bond spreads.
  - For most countries, equity prices are not cointegrated with bond spreads or CDS spreads; possible reasons: low debt-to-asset ratios, widely fluctuating leverage, nonlinear relationships, or MSCI indices not precise equity proxies.

### VI. Granger causality and price discovery (VECM, Hasbrouck, Gonzalo-Granger)
- Granger causality (qualitative summary):
  - Very low p-values overall indicate causality between markets with no single market universally dominating price discovery.
  - One-day horizon: CDS dominates price discovery in most countries.
  - Longer horizons: price discovery typically occurs in both CDS and bond markets; in some countries bond market leads CDS for longer horizons.
  - Equity markets mostly play a secondary role in price discovery.
- VECM price discovery measures — selected reported findings (error-correction coefficients, Hasbrouck bounds, Gonzalo-Granger statistics preserved as presented):
  - CDS vs. bond spreads (examples from table):
    - Brazil: CDS equation error-correction coefficient -0.0421 (t-statistic -2.041); bond equation 0.004 (t-statistic 0.352); Hasbrouck lower bound: 10.0; Hasbrouck upper bound: 0.677; Gonzalo-Granger: 7.96, 8.68.
    - Bulgaria: CDS -0.1015 (t 3.653); bond 0.1875 (t 3.301); Hasbrouck lower bound: 31.52; Hasbrouck upper bound: 31.52; Gonzalo-Granger: 61.48, 64.89.
    - Colombia: CDS -0.0439 (t 1.484); bond 0.0416 (t 2.387); Hasbrouck lower bound: 29.64; Hasbrouck upper bound: 29.64; Gonzalo-Granger: 88.54, 48.59.
    - Russia: CDS -0.0026 (t -0.342); bond 0.0138 (t 2.513); Hasbrouck lower bound: 62.56; Hasbrouck upper bound: 62.56; Gonzalo-Granger: 98.85, 84.16.
    - Venezuela: CDS -0.0317 (t -1.079); bond 0.0197 (t 1.37); Hasbrouck lower bound: 13.48; Hasbrouck upper bound: 13.48; Gonzalo-Granger: 91.64, 38.33.
  - CDS vs. equity prices (selected):
    - Russia: CDS error-correction -0.0249 (t -2.273); equity -0.0061 (t -2.202); Hasbrouck lower bound: 35.72; Gonzalo-Granger: 61.92 (equity column n.a. in source).
    - Venezuela: CDS -0.03 (t -2.254); equity -0.0018 (t 3.064); Hasbrouck lower bound: 4.69; Gonzalo-Granger: 10.23 (equity column n.a.).
  - Bond vs. equity prices (selected):
    - Russia: bond -0.0067 (t -0.9747); equity -0.0007 (t -2.623); Hasbrouck lower bound: 63.09; Gonzalo-Granger: 1.29, 18.59.
- Interpretative notes:
  - Wide gaps between Hasbrouck lower and upper bounds complicate precise attribution of price discovery contribution (example from source: Brazil CDS contribution could be as little as 1 percent or as much as 78 percent).
  - Overall, Hasbrouck bounds and Gonzalo-Granger statistics suggest CDS may have a slight edge over bond market for price discovery in most countries; Russia and Venezuela: equity market plays important role.

### VII. Results synthesis and interpretation
- Existence of equilibrium price relationships:
  - CDS–bond cointegration present in Brazil, Bulgaria, Colombia, Russia, Venezuela.
  - No CDS–bond cointegration in Mexico, Philippines, Turkey (possible frictions or proxy issues).
  - For most countries, equity prices are not cointegrated with bond spreads or CDS spreads; consistent with Merton’s theory implications (low/volatile debt-to-asset ratios) or proxy limitations.
- Price discovery heterogeneity:
  - No single market universally dominates price discovery across countries and horizons.
  - Short horizon (one-day): CDS often leads.
  - Longer horizons: price discovery shared or bond leads in some countries.
  - Country-specific patterns:
    - Bulgaria, Colombia, Venezuela: CDS and bond markets equally important for price discovery.
    - Brazil: bond market more important.
    - Colombia and Russia: CDS often most important source of price discovery.
    - Russia and Venezuela: equity market plays a notable role.
- Liquidity dynamics explanation:
  - During distress, liquidity migrates toward CDS; CDS better reflects default risk and leads price discovery.
  - In emerging markets, when CDS investor base (e.g., banks) is buy-and-hold, bond market can have greater trading volume and liquidity and lead price discovery.

### VIII. Conclusions and implications
- Main conclusions (period March 2001–May 2003):
  - Strong correlation and cointegration between CDS and bond spreads in Brazil, Bulgaria, Colombia, Russia, Venezuela — consistent with arbitrage convergence.
  - No stable equilibrium between equity and bond markets in most countries — consistent with Merton’s theory or proxy/nonlinearity issues.
  - Mixed price discovery results: no universal dominant market; CDS often leads at short horizons, bonds often important or dominant at longer horizons in emerging markets.
  - Contrast with mature-market corporate studies where CDS dominates price discovery; in emerging markets bond liquidity often exceeds CDS liquidity except during high-default-risk episodes.
- Recommendations for further study:
  - Investigate alternating high/low default risk episodes and their impact on liquidity migration and price discovery.
  - Examine proxy appropriateness (MSCI for equity, EMBI+ for 5-year bond spreads) and potential nonlinearities.

### IX. Appendix — Merton model and sovereign adaptation (summary)
- Merton (1974) corporate debt model recap:
  - Equity is a call option on assets; leverage d = Fe^(−rT)/V; bond and equity prices positively correlated; correlation increases with leverage; if B/E ≈ 0, correlation negligible.
- Sovereign adaptation:
  - Incorporate “willingness-to-pay” so country may default while technically solvent.
  - Feasible debt/equity value functions satisfy balance sheet identity, incentive compatibility, and limit condition.
  - Constructed value functions show:
    - Country can default while technically solvent.
    - Debt and equity values increase with country asset value → positive correlation.
    - Debt less sensitive to asset changes when assets large relative to debt; equity more sensitive → correlation stronger near default point.

*Source: IMF working paper content unit _wp0427 (PDF chapter/section content provided).*

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

### _wp0427 - References

### I. Introduction — market growth and role of credit derivatives
- Credit derivatives market expansion:
  - From US$1.8 billion in 1993 to US$2.0 trillion in 2002 (British Bankers’ Association survey).
  - Estimated total outstanding value will reach US$4.8 trillion by 2004 (BBA survey).
- Credit derivatives purpose and structure:
  - Allow credit risk to be traded without transferring ownership of underlying assets.
  - Strip out and isolate credit risk from other factors such as market risk.
  - Created by combining financing strategies such as swap contracts or pooled securities.
  - Provide insurance-like protection from credit/default risk.
- Market interaction and implications:
  - Changes in a firm’s credit risk affect CDS prices, bond prices, and equity prices.
  - Distress increases default risk → bond prices and equity prices fall → CDS spreads rise.
  - Sellers of credit protection may hedge by shorting bonds or equity, exerting further downward pressure on prices.
- Relevant theoretical and empirical literature:
  - Duffie (1999) and Hull and White (2000): absent market frictions, arbitrage forces CDS spreads to be approximately equal to underlying bond spreads; CDS and bond spreads are positively correlated.
  - Merton (1974): links firm value, debt, and equity; when asset value falls below debt face value, default occurs; bond and equity prices positively correlated when default risk is high.
  - Empirical findings for corporate issuers: Hull, Predescu, and White (2003); Blanco, Brennan, and Marsh (2003); Longstaff, Mithal, and Neiss (2003) — evidence that CDS (and equity) markets often lead bond markets in price discovery during deteriorations.
- Study objective:
  - Examine equilibrium price relationships and price discovery among CDS, bond, and equity markets for eight emerging market countries: Brazil, Bulgaria, Colombia, Mexico, the Philippines, Russia, Turkey, and Venezuela.
  - Attempt to extend Merton’s theory to sovereign issuers (justification provided in appendix).

### II. Background — credit derivatives and CDS specifics
- Market overview:
  - Credit derivatives combine derivatives and securitization to isolate and transfer credit risk at relatively low cost.
  - Increased standardization (ISDA Credit Derivatives Definitions) reduced legal obstacles and increased broker trading (GFINet, CreditTrade).
  - Contracts offered by market makers have become almost identical for CDS, allowing price shopping.
- Market size and players:
  - Credit derivatives market about $2.0 trillion in 2002; in 1997 total global value was $180 billion; estimated growth to $4.8 trillion by end-2004.
  - Main participants: banks (largest buyers and sellers of protection), insurance companies (second largest sellers), securities houses, hedge funds.
  - In September 2003, banks sold US$1.3 trillion of gross credit protection though they were net buyers of protection in the amount of US$229 billion; insurance companies sold protection in a net amount of US$303 billion (FitchRatings, 2003).
- Principal product types:
  - Four broad types: total return swaps, credit default swaps (CDSs), credit linked notes, synthetic collateralized debt obligations (CDO).
  - Total return swap: exchange total return on a bond/reference asset for LIBOR plus a spread.
  - CDS: payments by protection buyer to protection seller in exchange for contingent payment if specified credit event occurs.
  - Credit linked notes: securitized form of credit derivatives.
  - Synthetic CDO: asset-backed security whose collateral is typically a portfolio of bonds or bank loans.
- Credit default swap mechanics:
  - CDS: protection buyer pays periodic fee (default swap premium or default swap spread) expressed in basis points per notional; fee paid quarterly or semiannually.
  - Typical credit events: bankruptcy, insolvency, credit downgrade, failure to make a scheduled payment (as defined by ISDA).
  - Settlement types: physical settlement (buyer delivers reference security and receives par) or cash settlement (payment equal to par minus recovery value determined by dealer poll or price quote services). Most contracts subject to physical settlement.
  - Typical contract sizes and maturities: notional amounts of US$5 million and US$10 million with a five-year maturity; many contracts written on 5-year maturity but banks also write on client-specified notional amounts and maturities.
  - CDS analogous to swapping payments from a risky security for a risk-free security plus a contingent payment in case of default (see Duffie (1999)).
  - For corporate and financial institutions, trading activity greatest for five-year contracts.
- CDS liquidity and market drivers:
  - Most activity references mature market issuers; emerging market issuers comprise a smaller share but have evolved rapidly.
  - Emerging market notional amounts outstanding in 2002: US$300 billion, roughly equal to 15 percent of the market (Deutsche Bank/Xu, 2003).
  - CDS liquidity depends on liquidity of sovereign bonds — countries with liquid bond markets (Mexico, Brazil, Colombia, Venezuela) have easier price quotes.
  - Major emerging-market CDS players: hedge funds, emerging market dedicated mutual funds, pension funds, and banks.

### III. Price relationships and theoretical framework
- CDS–bond relationship:
  - Arbitrage argument: buying a risky bond and buying a CDS to insure default should eliminate spread differential; theory predicts one-to-one changes between CDS and bond spreads (Duffie (1999); Hull and White (2000)).
  - Assumptions required for parity: par-floating rate securities for risky and risk-free bonds; no transaction costs; negligible tax effects; CDS spread payments stop if credit event occurs; protection buyers paid on next coupon date after event.
  - Practical divergences:
    - Cheapest-to-deliver option in CDS: protection buyer can deliver the cheapest bond on credit event, potentially causing sellers to charge a higher CDS spread (cheapest-to-deliver premium) relative to bond spread.
    - Relative liquidity differences: if CDS markets are less liquid than bond markets (as can be the case for sovereigns), CDS may trade at higher spreads to compensate for lower liquidity; if liquidity migrates, CDS and bond spreads may diverge.
- Bond–equity relationship (Merton 1974 framework):
  - Equity analogous to a call option on firm assets with strike equal to face value of debt.
  - If assets < face value of debt → default; equity worthless in that state.
  - Bond and equity prices are positively correlated, stronger when debt-to-asset ratios are high or default risk is high.
  - For firms near default threshold, small asset value changes can cause default, affecting both equity and bond prices.
  - Implied sign: CDS spreads and equity prices should move in opposite directions if CDS ↔ bond spreads and bond ↔ equity have the described relationships.
- Extending Merton to sovereigns:
  - Sovereign-specific considerations: sovereigns may choose to default despite technical solvency; introduce “willingness-to-pay factor.”
  - Two critical asset values for sovereigns: “default value” and “no-payment value.”
  - For corporates: “no-payment value” = zero and “default value” = face value of debt.
  - For sovereigns: “default value” > face value of debt and “no-payment value” > zero due to continued asset value after cessation of debt payments.
  - Implication: for every asset value, default risk is higher for sovereigns than for comparable corporates; relationship between debt and equity prices should remain qualitatively similar for sovereigns as for corporates (further details in appendix).

### IV. Empirical context and data overview
- Countries analyzed: Brazil, Bulgaria, Colombia, Mexico, the Philippines, Russia, Turkey, and Venezuela.
- Credit-rating context: these countries have foreign currency–denominated external debt that are below investment grade except Mexico, which has been rated investment–grade since March 2002.
- Data frequency and sample period:
  - CDS, bond, and equity data are daily data for the period March 19, 2001 through May 29,

*Source: _wp0427 - References (IMF PDF).*

### 2003. For purposes of cross-country comparisons, we use country bond and equity indexes. The

### _wp0427 - 2003. For purposes of cross-country comparisons, we use country bond and equity indexes. The

### Data and measurement
- CDS spreads: mid-price quotes on five-year contracts obtained from CreditTrade and Deutsche Bank. CreditTrade used for plots; Deutsche Bank used for empirical analysis.
- Bond spreads: JPMorgan Chase Emerging Market Bond Index Plus (EMBI+).
- Equity prices: U.S. dollar-denominated Morgan Stanley Composite Index (MSCI) used as proxy for country equity value.
- Sample period described in conclusions: March 2001–May 2003.

### Descriptive relationships (Figure 2)
- Column (1): bond spreads and equity prices — most countries exhibit an inverse relationship between these two markets.
- Column (2): CDS spreads and equity prices — CDS spreads and equity prices are negatively correlated for most countries.
- Column (3): CDS spreads and bond spreads — most countries have positive correlation between these two markets.
- Sources cited for figure: Bloomberg L.P.; and CreditTrade.

### Empirical methodology
- Two-step cointegration approach:
  - Step 1: test stationarity/unit roots using Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests. Null hypothesis: series characterized by unit root (β = 0).
  - Step 2: Johansen (1991) cointegration rank test on a vector autoregression of order p; if Π has reduced rank equal to 1, there is cointegration (Π = α βᵀ).
- Unit root test equations:
  - ADF regression (equation (1)) with ∆ operator, p lags; null β = 0.
  - Phillips-Perron regression (equation (2)) with Newey-West correction; null β = 0.
- Granger causality tests:
  - Regression form (equation (5)) using first differences ∆, lags n.
  - Granger causality tested for 1, 5, 10, and 20 business day lags (up to one-month horizon).
- Price discovery measures:
  - Hasbrouck (1995) lower and upper bounds derived from VMA long-run impact matrix Ψ(1).
  - Gonzalo and Granger (1995) GG statistic: contribution ∝ ii αα⊥⊥∑ (i=1,..,n), where f = α⊥ᵀ Y.
  - VECM specification used (equation (6)): ∆Yt = α β′ Yt−1 + Σ Aj ∆Yt−j + e t.
  - Lag selection in VECM based on Schwarz information criterion (SIC).
- Interpretation of error-correction coefficients: statistical insignificance of an error-correction coefficient in one market suggests price discovery happens only in the other market; significance in both markets implies shared price discovery.

### Unit root test results (Table 1: adjusted t-statistics)
- Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) adjusted t-statistics reported; result summary: adjusted t-statistics for all variables are characterized by unit roots at the 90 percent and above confidence level, implying need for cointegration rank tests.
- Table 1 entries (Adjusted t-statistics):
  - Brazil: ADF CDS spreads -1.581; Bond spreads -1.485; Equity prices -1.920. PP CDS spreads -1.336; Bond Spreads -1.554; Equity Prices -1.734.
  - Bulgaria: ADF CDS -1.838; Bond -2.228; Equity n.a. PP CDS -2.292; Bond -1.729; Equity n.a.
  - Colombia: ADF CDS -1.409; Bond -1.886; Equity -0.395. PP CDS -1.402; Bond -1.556; Equity -0.237.
  - Mexico: ADF CDS -1.885; Bond -1.469; Equity -1.924. PP CDS -1.770; Bond -2.016; Equity -1.747.
  - Philippines: ADF CDS -3.181; Bond -1.944; Equity -1.112. PP CDS -1.910; Bond -1.719; Equity -1.426.
  - Russia: ADF CDS -1.845; Bond -2.029; Equity -1.335. PP CDS -1.722; Bond -1.543; Equity -0.890.
  - Turkey: ADF CDS -2.854; Bond -3.160; Equity -2.407. PP CDS -2.532; Bond -2.289; Equity -2.497.
  - Venezuela: ADF CDS -2.199; Bond -2.456; Equity -2.198. PP CDS -2.438; Bond -2.096; Equity -1.517.
- Note: CDS spreads correspond to Deutsche Bank daily 5-year quotes; bond spreads EMBI+; equities MSCI indices.

### Johansen cointegration rank test results (Table 2: trace statistics)
- Summary statement: CDS and bond spreads are cointegrated in Brazil, Bulgaria, Colombia, Russia, and Venezuela; not cointegrated in Mexico, the Philippines, and Turkey.
- Table 2 reported significant trace statistics (only significant rejections of null of no cointegration at 10 percent or below):
  - CDS spreads and bond spreads:
    - Brazil: 37.33 (rejects no cointegration).
    - Bulgaria: 41.235.
    - Colombia: 64.4.
    - Mexico: no cointegration.
    - Philippines: no cointegration.
    - Russia: 21.637.
    - Turkey: no cointegration.
    - Venezuela: 20.081.
  - Bond spreads and equity prices:
    - Russia: 17.874 (reported as significant).
    - Venezuela: 11 (reported as significant).
  - CDS spreads and equity prices:
    - Russia: 18.646 (reported as significant).
- Interpretations noted in text:
  - Existence of CDS–bond cointegration provides evidence that arbitrage forces CDS and bond spreads to converge despite market frictions.
  - Absence of cointegration in Mexico, Philippines, Turkey may reflect exploitable-arbitrage impediments or EMBI+ not being an appropriate proxy for 5-year bond spreads.
  - For most countries, equity prices are not cointegrated with bond spreads or with CDS spreads; possible interpretations include low debt-to-asset ratios, widely fluctuating leverage, nonlinear relationships, or MSCI indices not being precise equity proxies.

### Granger causality tests (Table 3) — qualitative summary
- Granger causality tests performed for lags of 1, 5, 10, and 20 business days; reported F-statistics and p-values.
- Main findings:
  - Very low p-values overall indicate causality between markets with no single market universally dominating price discovery.
  - For a one-day horizon, CDS dominates price discovery in most countries.
  - As horizon lengthens, price discovery typically occurs in both CDS and bond markets; in some countries bond market leads CDS for longer horizons.
  - Equity markets mostly play a secondary role in price discovery.

### VECM price discovery measures (Table 4: error-correction coefficients, Hasbrouck bounds, Gonzalo-Granger statistics)
- Error-correction coefficients and t-statistics (italic t-statistics presented in original table); Hasbrouck lower and upper bounds (percent contribution range); Gonzalo-Granger statistics (percent contribution).
- CDS spreads and bond spreads (error-correction coefficients, t-statistics in italics; Hasbrouck lower bound; Hasbrouck upper bound; Gonzalo-Granger statistics):
  - Brazil:
    - CDS equation error-correction coefficient: -0.0421 (t-statistic -2.041).
    - Bond equation error-correction coefficient: 0.004 (t-statistic 0.352).
    - Hasbrouck lower bound: 10.0; Hasbrouck upper bound: 0.677. (Note: table formatting indicates bounds then percent contributions; original table juxtaposes numbers—preserve as reported.)
    - Gonzalo-Granger statistics: 7.96 (CDS?), 8.68 (Bond?). 
  - Bulgaria:
    - CDS equation: -0.1015 (t-statistic 3.653).
    - Bond equation: 0.1875 (t-statistic 3.301).
    - Hasbrouck lower bound: 31.52; Hasbrouck upper bound: 31.52 (table shows 31.52 then 61.48/64.89 contextually).
    - Gonzalo-Granger statistics: 61.48, 64.89 (as presented).
  - Colombia:
    - CDS equation: -0.0439 (t-statistic 1.484).
    - Bond equation: 0.0416 (t-statistic 2.387).
    - Hasbrouck lower bound: 29.64; Hasbrouck upper bound: 29.64 (table shows 29.64 then 88.54/48.59).
    - Gonzalo-Granger statistics: 88.54, 48.59.
  - Russia:
    - CDS equation: -0.0026 (t-statistic -0.342).
    - Bond equation: 0.0138 (t-statistic 2.513).
    - Hasbrouck lower bound: 62.56; Hasbrouck upper bound: 62.56 (table shows 62.56 then 98.85/84.16).
    - Gonzalo-Granger statistics: 98.85, 84.16.
  - Venezuela:
    - CDS equation: -0.0317 (t-statistic -1.079).
    - Bond equation: 0.0197 (t-statistic 1.37).
    - Hasbrouck lower bound: 13.48; Hasbrouck upper bound: 13.48 (table shows 13.48 then 91.64/38.33).
    - Gonzalo-Granger statistics: 91.64, 38.33.
- CDS spreads and equity prices:
  - Russia:
    - CDS equation error-correction: -0.0249 (t-statistic -2.273).
    - Equity equation error-correction: -0.0061 (t-statistic -2.202).
    - Hasbrouck lower bound: 35.72; Hasbrouck upper bound: 35.72 (table shows 35.72 then 61.92/ n.a.).
    - Gonzalo-Granger statistics: 61.92; equity column marked n.a.
  - Venezuela:
    - CDS equation: -0.03 (t-statistic -2.254).
    - Equity equation: -0.0018 (t-statistic 3.064).
    - Hasbrouck lower bound: 4.69; Hasbrouck upper bound: 4.69.
    - Gonzalo-Granger statistics: 10.23; equity column marked n.a.
- Bond spreads and equity prices:
  - Russia:
    - Bond equation error-correction: -0.0067 (t-statistic -0.9747).
    - Equity equation error-correction: -0.0007 (t-statistic -2.623).
    - Hasbrouck lower bound: 63.09; Hasbrouck upper bound: 63.09.
    - Gonzalo-Granger statistics: 1.29, 18.59.
- Interpretation notes from original text:
  - Wide gap between Hasbrouck lower and upper bounds makes precise attribution of price discovery contribution difficult (example: Brazil CDS contribution could be as little as 1 percent or as much as 78 percent as described in text).
  - Overall, Hasbrouck bounds and Gonzalo-Granger statistics suggest CDS may have a slight edge over bond market for price discovery in most countries; Russia and Venezuela: equity market plays important role.

### Results: existence of equilibrium price relationships (section V.A)
- Key findings:
  - CDS–bond cointegration present in Brazil, Bulgaria, Colombia, Russia, Venezuela.
  - No CDS–bond cointegration in Mexico, Philippines, Turkey (possible explanations: market frictions, EMBI+ not appropriate proxy for 5-year bond spreads).
  - For most countries, equity prices are not cointegrated with bond spreads or CDS spreads.
  - Interpretations: consistent with Merton’s theory — low debt-to-asset values or widely fluctuating leverage; nonlinear relationships or MSCI indices not accurate equity proxies.

### Price discovery results (section V.B and V)
- Granger causality tests:
  - Evidence of causality between markets; no single market universally dominates.
  - One-day horizon: CDS often dominates price discovery.
  - Longer horizons: price discovery shared between CDS and bond markets; in some countries bond leads CDS at longer horizons.
  - Equity markets generally secondary in price discovery.
- VECM / Hasbrouck / Gonzalo-Granger:
  - Country-level heterogeneity:
    - Bulgaria, Colombia, Venezuela: CDS and bond markets are equally important for price discovery.
    - Brazil: bond market more important for price discovery.
    - Russia, Venezuela: price discovery occurs in CDS and equity markets; in Russia bond market also leads equity market.
  - Liquidity dynamics explanation:
    - During distress, liquidity migrates toward CDS; CDS prices default risk better then and lead price discovery.
    - In emerging markets, bond market can dominate when CDS investor base is banks with buy-and-hold positions; bond market then has greater trading volume and liquidity and leads price discovery.
  - Equity markets:
    - Generally negligible role except Russia.
    - Equity prices convey useful default-risk information when default risk is high (countries out-of-the-money); when in-the-money, equity movements driven by non-default factors.

### Conclusions (section VI)
- Study analyzed equilibrium price relationships and price discovery for eight emerging market countries, period March 2001–May 2003.
- Main conclusions:
  - Strong correlation between CDS and bond spreads in Brazil, Bulgaria, Colombia, Russia, Venezuela — consistent with arbitrage forces converging CDS and bond spreads.
  - No equilibrium relationship between equity and bond markets in most countries — consistent with Merton’s theory (low or volatile debt-to-asset ratios) or nonlinear relationships.
  - Price discovery: mixed results; no single market dominates across countries and time horizons.
    - Colombia and Russia: CDS most important source of price discovery.
    - Brazil and Bulgaria: CDS and bond markets equally important.
    - Liquidity migration explains alternation in dominance between CDS and bond markets.
  - Emerging markets vs. mature markets:
    - Findings contrast studies on corporate issuers in U.S./Europe where CDS dominates price discovery in mature markets.
    - In mature markets CDS more liquid than bonds; in emerging markets bond market tends to be more liquid and often dominates price discovery except during high-default-risk episodes when CDS leads.
- Further study recommended due to alternating high/low default risk episodes and potential proxy issues with MSCI and EMBI+.

### Appendix — Merton model and sovereign adaptation (brief)
- Merton (1974) corporate debt model:
  - Firm’s equity is call option on assets; bond and equity prices linked by formula B = E N(d1) − ... (as presented in text).
  - Leverage d = Fe^(−rT) / V.
  - Bond and equity prices positively correlated; correlation increases with leverage; if B/E ≈ 0, correlation negligible.
- Application to sovereign issuers:
  - "Willingness-to-pay" (country may default while technically solvent) incorporated heuristically.
  - Feasible debt/equity value functions must satisfy:
    - Balance sheet identity: assets = debt + equity.
    - Incentive compatibility: country worse off when defaulting; equity value lower when country defaults.
    - Limit condition: country honors obligations when assets >> debt.
  - Constructed value functions (panel B of Figure A2) show:
    - Country can default while technically solvent.
    - Debt and equity values increase with country asset value → positive correlation.
    - Debt less sensitive to asset changes when assets large relative to debt; equity more sensitive → correlation stronger near default point.

*Source: IMF working paper content unit _wp0427 (PDF chapter/section content provided).*

### References

### _wp0427 - References

### References

- Azarchs, Tanya, 2003, “Demystifying Banks’ Use of Credit Derivatives,” Standard and Poors, December 9.  
- Blanco, Roberto, Simon Brennan, and Ian W. Marsh, 2003, “An Empirical Analysis of the Dynamic Relationships Between Investment Grade Bonds and Credit Default Swaps” (unpublished; Madrid: Banco de España, and London: Bank of England).  
- Baillie, Richard T., G. Geoffrey Booth, Yiuman Tse, and Tatyana Zabotina, 2002, “Price Discovery and Common Factor Models,” Journal of Financial Markets, Vol. 5, pp. 309–21.  
- Booth, G. Geoffrey, Teppo Martikainen, and Yiuman Tse, 1997, “Price and Volatility Spillovers in Scandinavian Stock Markets,” Journal of Banking and Finance, Vol. 21, pp. 811–23.  
- Chan-Lau, Jorge A., 2003, “Anticipating Credit Events Using Credit Default Swaps, with an Application to Sovereign Debt Crises,” IMF Working Paper 03/106 (Washington: International Monetary Fund).  
- Duffie, Darrell J., 1999, “Credit Swap Valuation,” Financial Analysts Journal, Vol. 55, pp. 73–87.  
- FitchRatings, 2003, Global Credit Derivatives: A Qualified Success, September.  
- Gonzalo, Jesús, and Clive W.J. Granger, 1995, “Estimation of Common Long-Memory Components in Cointegrated Systems,” Journal of Business and Economic Statistics, Vol. 13, pp. 27–35.  
- Granger, Clive W.J., 1969, “Investigating Causal Relations by Econometric Models and Cross-Spectral Methods,” Econometrica, Vol. 37, pp. 424–38.  
- _____, and P. Newbold, 1974, “Spurious Regressions in Econometrics,” Journal of Econometrics, Vol. 2, pp. 111–120.  
- Hamao, Yasushi, Ronald W. Masulis, and Victor Ng, 1990, “Correlations in Price Changes and Volatility Across International Stock Markets,” Review of Financial Studies, Vol. 3, pp. 281–307.  
- Hasbrouck, Joel, 1995, “One Security, Many Markets: Determining the Contributions to Price Discovery,” Journal of Finance, Vol. 50, pp. 1175–99.  
- Hayashi, Fumio, 2000, Econometrics (Princeton: Princeton University Press)  
- - 30 -  
- Hull, John C., Mirela Predescu, and Alan White, 2003, “The Relationship Between Credit Default Swap Spreads, Bond Yields, and Credit Rating Announcements” (unpublished; Toronto: University of Toronto).  
- Hull, John C., and Alan White, 2000, “Valuing Credit Default Swaps I: No Counterparty Default Risk,” Journal of Derivatives, Vol. 8, pp. 29–40.  
- International Monetary Fund, 2003, Global Financial Stability Report, September (Washington).  
- Johansen, Soren, 1991, “Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models,” Econometrica, Vol. 59, pp. 1551–80.  
- Lehmann, Bruce N., 2002, “Some Desiderata for the Measurement of Price Discovery Across Markets,” Journal of Financial Markets, Vol. 5, pp. 259–76.  
- Longstaff, Francis, Sanjay Mithal, and Eric Neiss, 2003, “The Credit Default Swap Market: Is Credit Protection Priced Correctly” (unpublished; Los Angeles: University of California).  
- Merton, Robert C., 1974, “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,” Journal of Finance, Vol. 29, pp. 449–70.  
- Stock, James H., and Mark W. Watson, 1988, “Testing for Common Trends,” Journal of the American Statistical Association, Vol. 83, pp. 1097–07.  
- Tavakoli, Janet, 1998, Credit Derivatives (New York: John Wiley and Sons).  
- Xu, David, 2003, Emerging Markets Credit Derivatives (New York: Deutsche Bank).

*Source: _wp0427 - References.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp0427.pdf_