1. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Santiago Forte Lidija Lovreta PhD Candidate ESADE Business School December 2008

2. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? 1.Credit risk should be implicitly or explicitly reflected through market prices of different credit sensitive claims. 2.These securities are traded in structurally different markets in terms of organization, liquidity, participants, stage in the development. 1.Credit risk should be implicitly or explicitly reflected through market prices of different credit sensitive claims. 2.These securities are traded in structurally different markets in terms of organization, liquidity, participants, stage in the development. Santiago Forte, Lidija Lovreta © ESADE Probable differences in relative speed with which respective markets respond to changes in credit conditions. Introduction

3. CDS and Bond Market Zhu (2004), Norden and Weber, (2005): Granger Causality in VAR Zhu (2004), Norden and Weber, (2005), Blanco et al, (2005): VECM Zhu (2004), Norden and Weber, (2005): Granger Causality in VAR Zhu (2004), Norden and Weber, (2005), Blanco et al, (2005): VECM CDS, Stock and Bond Market Longstaff et al. (2003), Norden and Weber (2005): lead-lag relationship in VAR Forte and Peña (2008): VECM Longstaff et al. (2003), Norden and Weber (2005): lead-lag relationship in VAR Forte and Peña (2008): VECM Price discovery mainly takes place in the CDS market. Price discovery mainly takes place in the CDS market. Stock and CDS market are more informationally efficient. Stock and CDS market are more informationally efficient. Theoretical parity relationship holds on average as an equilibrium condition. Theoretical parity relationship holds on average as an equilibrium condition. Stock market leads CDS and bond market more frequently than vice versa. Stock market leads CDS and bond market more frequently than vice versa. Literature Review Santiago Forte, Lidija Lovreta © ESADE Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads?

4. This paper Analyses credit risk discovery in the stock and CDS market in the light of two questions: Applies time-varying framework: factors underlying the price discovery process do not differ only on a cross-sectional basis but also change along time. 1.Which of these markets provides more timely information regarding the credit risk of the particular reference entity? 2.What factors influence the dynamic relationship between the stock and the CDS market? Santiago Forte, Lidija Lovreta © ESADE Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads?

5. Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads?

6. CDS Market Credit risk is explicitly traded. CDS premium (or CDS spread), a fee the buyer of the protection pays as a compensation for the protection against the default risk. Credit risk is explicitly traded. CDS premium (or CDS spread), a fee the buyer of the protection pays as a compensation for the protection against the default risk. Stock Market Information regarding credit risk is implicitly reflected through the stock market. Theoretical modelling should be applied to extract this information. Information regarding credit risk is implicitly reflected through the stock market. Theoretical modelling should be applied to extract this information. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? CDS spreads Stock market implied credit spreads - methodology of Forte (2008) Stock market implied credit spreads - methodology of Forte (2008) Credit Spreads in the CDS and Stock Market Santiago Forte, Lidija Lovreta © ESADE

7. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE

8. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Data set The initial sample of CDS and ICS series is the same as in Alonso, Forte and Marqués (2008): 96 companies during 2001-2004. For comparison purposes we exclude: two companies without available credit rating; year 2001 (CDS and ICS series are available just for 8 companies). The reduced sample contains 94 non-financial companies: 40 European 32 US 22 Japanese The reduced period covers 2002-2004. Santiago Forte, Lidija Lovreta © ESADE

9. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Data set The majority of the CDS contracts in the sample refer to A and BBB rated issuers (76 out of 94). CDS spread levels (mean 72 bp) exhibit variation over different time periods, regions and ratings: The mean level of CDS spreads reaches the maximum in 2002 while subsequent years demonstrate clear downward trend; CDS spreads are on average higher for US companies, and lower rated issuers. Derived ICS series in general follow the pattern observed in the CDS market (mean 83 bp): Alonso, Forte and Marqués (2008); Forte and Peña (2008). Santiago Forte, Lidija Lovreta © ESADE

10. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE

11. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Advantages of ICS over stock returns Enable considering eventual long-run equilibrium relationship between considered series. Enable accounting for the effect of other relevant variables (e.g. risk-free rate) in addition to stock prices that are found to be important in determining credit spreads. Take into account the non-linear functional relationship between credit spreads and considered variables. Enable considering eventual long-run equilibrium relationship between considered series. Enable accounting for the effect of other relevant variables (e.g. risk-free rate) in addition to stock prices that are found to be important in determining credit spreads. Take into account the non-linear functional relationship between credit spreads and considered variables. Santiago Forte, Lidija Lovreta © ESADE

12. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Advantages of ICS over stock returns Spreads derived from the structural model better explain changes in the CDS. Santiago Forte, Lidija Lovreta © ESADE

13. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE

14. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Strength of the stock - CDS market relationship The explanatory power decreases during the period 2002-2004, completely following the pattern of the mean CDS level. The explanatory power decreases with rating group (from BBB to AAA-AA). The explanatory power decreases during the period 2002-2004, completely following the pattern of the mean CDS level. The explanatory power decreases with rating group (from BBB to AAA-AA). The strength of the stock - CDS market relationship is positively related to the level of credit risk (i.e. increases together with the decrease in the average creditworthiness). Santiago Forte, Lidija Lovreta © ESADE

15. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE

16. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Time-series properties of CDS and ICS Engle and Granger (1987): if two time series have unit roots then their linear combination might be stationary. Those series are said to be cointegrated where cointegrating equation might be understood as a long-run equilibrium relationship. Stationarity Augmented Dickey-Fuller (ADF) test: Unit roots in both series are simultaneously detected for 66 companies. Cointegration Johansen Cointegration Test: Significant cointegration relationship is found for 17 companies. Santiago Forte, Lidija Lovreta © ESADE

17. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Vector Error Correction Model (VECM) In the case of the presence of cointegration, cointegrating relationship must be explicitly taken into account. Short-term dynamics between the CDS and ICS series is examined on the basis of the valid VECM representation. The the two-dimensional (VECM) is specified as follows: Santiago Forte, Lidija Lovreta © ESADE

18. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Measures of contribution to the price discovery process It seems that both markets almost equally contribute to price discovery. Santiago Forte, Lidija Lovreta © ESADE

19. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Measures of contribution in a time-varying framework Half-yearly loadings for each company are estimated by imposing the entire sample cointegrating vector to the half-yearly VECM. The analysis reveals no leadership concentration in neither of the markets, but, reveals a downward trend in the leading role of stock market. Santiago Forte, Lidija Lovreta © ESADE

20. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Granger Causality For the entities that either do not have unit roots or for which the presence cointegration is rejected the VECM approach is not valid. Price leadership is tested by the presence of Granger causality in VAR model of the following form: Santiago Forte, Lidija Lovreta © ESADE

21. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Granger Causality The lagged changes in ICS are important in the explanation of current changes in CDS more often than vice versa - dominance of the stock market in credit risk discovery over the entire 2002-2004 sampling period. ∆ICS do Granger-cause ∆CDS for 47 companies (61.04%); ∆CDS do Granger-cause ∆ICS for 16 companies (20.78%); The one-way influence of the stock market (∆ICS do Granger-cause ∆CDS, but ∆CDS do not Granger-cause ∆ICS) is detected for 36 companies. The opposite is true just for 5. ∆ICS do Granger-cause ∆CDS for 47 companies (61.04%); ∆CDS do Granger-cause ∆ICS for 16 companies (20.78%); The one-way influence of the stock market (∆ICS do Granger-cause ∆CDS, but ∆CDS do not Granger-cause ∆ICS) is detected for 36 companies. The opposite is true just for 5. Santiago Forte, Lidija Lovreta © ESADE

22. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Granger Causality in a time-varying framework ∆ICS do Granger-cause ∆CDS more times than vice versa; The informational precedence of the stock market appears to diminish over time (measured by the proportion of the corresponding null hypothesis rejections in the total number of examined companies). ∆ICS do Granger-cause ∆CDS more times than vice versa; The informational precedence of the stock market appears to diminish over time (measured by the proportion of the corresponding null hypothesis rejections in the total number of examined companies). Santiago Forte, Lidija Lovreta © ESADE

23. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE

24. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Factors underlying credit risk discovery process The CDS % bid-ask spread. The stock turnover ratio, defined as the number of shares traded adjusted by the number of shares outstanding. The relative frequency of adverse shocks i.e. a one-day increase in CDS level of more than 50 bp (Acharya and Johnson, 2007). Credit condition expressed by either rating (numerically expressed on a scale ranging from 23 for BB to 35 for AAA rated issuers), average credit spread level or the dummy variable which takes the value 1 if CDS level overpass 100bp during the sampling period. Trend of the level of CDS premiums over the covered period, defined as the slope of the characteristic line over the observed half-yearly period. Time dummy variables. Santiago Forte, Lidija Lovreta © ESADE

25. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Factors underlying “information shares” OLS regression is estimated on the basis of 92 half-yearly observations of the HM: Hasbrouck measures are by definition always in 0-1 range; Account for the variance-covariance matrix of residuals from the VECM specification. Santiago Forte, Lidija Lovreta © ESADE

26. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Ordered probit for the sub-sample of 77 companies The clear CDS market leadership in the credit risk discovery process: (-1) Changes in CDS Granger cause changes in ICS but not the other way around. The situation in which there is no clear interpretation. (0) The clear stock market leadership in the credit risk discovery process: (1) Dummy variables Changes in ICS Granger cause changes in CDS but not the other way around Santiago Forte, Lidija Lovreta © ESADE

27. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Ordered probit for the sub-sample of 77 companies Regression is estimated on the basis of 388 half-yearly observations. Santiago Forte, Lidija Lovreta © ESADE

28. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Ordered probit for the whole sample The clear CDS market leadership in the credit risk discovery process: (-1) Loading factor 1 is significantly negative but not the other way around or Changes in CDS Granger cause changes in ICS but not the other way around. The situation in which there is no clear interpretation. (0) The clear stock market leadership in the credit risk discovery process: (1) Loading factor 2 is significantly positive but not the other way around or Changes in ICS Granger cause changes in CDS but not the other way around. Dummy variables Santiago Forte, Lidija Lovreta © ESADE

29. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Ordered probit estimation results Regression is estimated on the basis of 480 half-yearly observations. Santiago Forte, Lidija Lovreta © ESADE

30. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Probit estimation result Regression is estimated on the basis of 151 half-yearly observations: - 1 for the clear stock market leadership and, -0 for the clear CDS market leadership in the credit risk discovery process. Santiago Forte, Lidija Lovreta © ESADE

31. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Results show: The probability of the stock market leadership is positively related with the credit risk level. The probability of the CDS market leadership is positively related with the relative frequency of severe credit deterioration shocks. Results are consistent under the: -OLS framework with HM information shares; -Ordered probit framework with categorical ordered dummies; -Probit framework that contrasts two extreme cases (strict stock vs strict CDS market leadership). Results are consistent under the: -OLS framework with HM information shares; -Ordered probit framework with categorical ordered dummies; -Probit framework that contrasts two extreme cases (strict stock vs strict CDS market leadership). Santiago Forte, Lidija Lovreta © ESADE

32. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Outline Credit spreads in the CDS and stock market Data set Advantages of the ICS over stock returns Strength of the stock - CDS market relationship Credit risk discovery Factors underlying credit risk discovery Conclusions Santiago Forte, Lidija Lovreta © ESADE

33. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Conclusions 1.The relationship between ICS and CDS is stronger at the lower levels of the credit quality. 2. Both markets contribute to credit risk discovery process with slight dominance of the stock market that declines over time. 3. The probability of the stock market leading price discovery increases with the credit risk level. 4. The probability of the CDS market leading price discovery is positively related with the frequency of severe downturns. Santiago Forte, Lidija Lovreta © ESADE

34. Thank You Comments Welcome

35. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? There is a negative relationship between credit rating and stock illiquidity. Information flows from the CDS to the stock market, especially for the entities that experience and are more likely to experience credit deterioration. The CDS market (as opposed to the bond market) is considerably more sensitive to the stock market and the magnitude of sensitivity increases with the decrease in the creditworthiness of the reference entity. Literature Review Related findings Santiago Forte, Lidija Lovreta © ESADE

36. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Data set - descriptive statistics Santiago Forte, Lidija Lovreta © ESADE

37. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Measures of contribution to the price discovery process The loadings 1 and 2 Gonzalo and Granger (1995) Hasbrouck (1995) Market contribution - a proportion of the innovation variance that can be attributed to that market. Market contribution – the ratio between two factor loadings. Determine the short-term adjustment dynamics of the less efficient market. Santiago Forte, Lidija Lovreta © ESADE

38. Credit Risk Discovery in the Stock and CDS Market: Who, When and Why Leads? Measures of contribution to the price discovery process The loadings 1 and 2 1 is significantly negative for 11 companies; 2 is significantly positive for 11 companies; One-way price adjustment is evident in 6 cases. 1 is significantly negative for 11 companies; 2 is significantly positive for 11 companies; One-way price adjustment is evident in 6 cases. Gonzalo and Granger (1995) Hasbrouck (1995) GG>0.5 for 6 companies; GG<0.5 for 11 companies; Average GG for 2002-2004 period equals 0.44. GG>0.5 for 6 companies; GG<0.5 for 11 companies; Average GG for 2002-2004 period equals 0.44. HM>0.5 for 10 companies. HM<0.5 for 7 companies; Average HM for 2002-2004 period equals 0.55. HM>0.5 for 10 companies. HM<0.5 for 7 companies; Average HM for 2002-2004 period equals 0.55. It seems that both markets almost equally contribute to price discovery. Santiago Forte, Lidija Lovreta © ESADE