Corporate Credit Spreads and the Sovereign Ceiling in Latin America

ABSTRACT

We exploit a panel of 72 US dollar-denominated bonds issued by Latin American publicly listed firms between 1996 and 2004, a period of regional financial crises, to answer the following three questions: (1) Is sovereign risk a statistically and economically significant determinant of the corporate credit spread, controlling for firm- and bond-specific characteristics? (2) If yes, do market participants apply the sovereign ceiling rule adopted by rating agencies in the pricing of our bond market data? And (3) how do market views compare with the rating agencies ceiling policy for each corporate bond? We find strong evidence of an economically and statistically significant effect of sovereign risk on corporate spreads across different panel econometric specifications and bonds. Moreover, markets do not apply the ceiling rule in 77–90% of the bonds we sample and these findings are consistent with rating agencies’ policies toward the latter for about 50% of the firms. These results are robust to the inclusion of firm- and bond-specific variables derived from the structural approach to credit risk and to the business cycle in each country.

KEY WORDS:

• Corporate bond spreads
• credit risk models
• emerging bond markets
• Latin America
• sovereign ceiling

JEL Codes:

• F34
• G0
• G1
• G2.

Acknowledgments

For invaluable comments or suggestions, the authors are grateful to Alberto Musalem, Silvina Vatnick, Daniel Marx, Marcelo Dabós as well as to participants at the LACEA 2008 and 2007 annual meetings, the 2009 annual meeting of AAEP and participants or discussants at conferences held at the 2011 World Finance Conference in Rhodes, Greece, Trinity College of Dublin, Université de Paris Dauphine, the Chinese Academy of Finance and Development, Nankai University, Universidad del CEMA, Pontificia Universidad Catòlica Argentina, the UN Business Forum, Ecole Normale Superieure (Lyon), and the Center for Financial Stability of Argentina.

Notes

^1. This article will use interchangeably the terms “corporate bond spreads,” “(corporate) credit risk,” “credit yield spreads,” and “corporate default premium.”

^2. For a survey of this literature, we refer the reader to Elton et al (2001) and Cossin and Pirotte (2001).

^3. In a later unpublished version of their paper, Grandes and Peter (2006) are able to prove the robustness of their findings to the inclusion of global factors, namely the UST 10-year bond yield, the CBOE VIX measure of risk aversion, and the volatility of global (MSCI) equity.

^4. Other theoretical frameworks are (1) the classical or actuarial (for a survey of these methods, see for instance Caouette, Altman, and Narayanan (1998), and (2) the reduced-form, statistical, or intensity-based approach. Readers interested in reduced-form models are referred to the works of Pye (1974), Litterman and Iben (1991), Fons (1994), Das and Tufano (1996), Jarrow and Turnbull (1995), Jarrow, Lando, and Turnbull (1997), Lando (1998), Madan and Unal (1998), Duffie and Singleton (1999), Collin-Dufresne and Solnik (2001), and Duffie and Lando (2001), most of which are surveyed and nicely put into a broader context by Cossin and Pirotte (2001), and Bielecki and Rutkowski (2002). We choose the structural approach because the classical approach is both too subjective and too backward looking and the reduced-form approach is atheoretical with respect to the determinants of default risk.

^5. Shimko, Tejima, and Van Deventer (1993) assume that the short-term risk-free interest rates follow a stationary Ornstein–Uhlenbeck process of the form $dr=\alpha (\gamma -r)dt+{\sigma }_{r}d{Z}_{2,t}$,where γ is the long-run mean which the short-term interest rate r is reverting to, α > 0 is the speed at which this convergence occurs, ${\sigma }_r$ is the instantaneous variance (volatility) of the interest rate, and $d{Z}_{2,t}={\epsilon }_2\sqrt{dt}$is a second standard Gauss–Wiener process, whose correlation with the stochastic firm value factor, $d{Z}_{1,t}$, is equal to $\rho$, i.e., $d{Z}_{1,t}.d{Z}_{2,t}=\rho dt$

^6. Shimko, Tejima, and Van Deventer (1993) determine the signs of $\mathrm{\partial }s/\mathrm{\partial }d$, $\mathrm{\partial }s/\mathrm{\partial }{\sigma }_V$, $\mathrm{\partial }s/\mathrm{\partial }\tau$, and $\mathrm{\partial }s/\mathrm{\partial }{\sigma }_r$ through simulations.

^7. Shimko, Tejima, and Van Deventer (1993), p. 59.

^8. Also see Helwege and Turner (1999), who demonstrate through an experiment the existence of a positively sloped credit spread-to-maturity curve for speculative grade borrowers.

^9. These factors are dealt with in the literature on corporate default risk in mature markets, in particular the US corporate bond market. See, for instance, Athanassakos and Carayannopoulos (2001).

^10. In Thomson Financial Datastream, we found 171 firms having issued at least one bond. However, many of these bonds did not display yield to maturity and price data over the relevant period.

^11. Elton et al. (2001) argue that one should use spreads calculated as the difference between yield to maturity on a zero-coupon corporate bond (called corporate spot rate) and the yield to maturity on a zero-coupon government bond of the same maturity (government spot rate) rather than as the difference between the yield to maturity on a coupon-paying corporate bond and the yield to maturity on a coupon-paying risk-free bonds.

^12. Available at http://www.treas.gov/offices/domestic-finance/debt-management/interest-rate/yield.shtml

^13. The econometric specification we applied is $y_i={\beta }_1+{\beta }_{2}log(t_i)+{\beta }_3{t_i}^2+{\epsilon }_i$, where y denotes each bond yield and t denotes time to maturity. The specification fits well to the US Treasury estimation.

^14. A methodological note discussing in detail the operationalization and measurement of these determinants can be obtained from the authors upon request.

^15. We do not control for global risk aversion because the short-term volatility in US interest rates is highly correlated with the former. Moreover, global risk aversion as for instance measured by the VIX index and UST bond yields at both ends of the curve have been found to be significant determinants of sovereign spreads. Therefore, their potential effect on corporate bond spreads should already be captured by those sovereign spreads.

^16. Following Hammersley and Atkinson (1983), we can state that what is involved in methodological triangulation is not the combination of different types of methodologies per se, but to correct the potential weaknesses that may limit the validity of the analysis. For Fielding and Fielding (1986), the conventional idea of triangulation is that if diverse types of data or methods sustain the same conclusion, the trustworthiness of the results is increased.

^17. We make this choice despite rejecting the null of the Hausman test, which favors the FE estimator to RE as the latter is inconsistent but efficient under the alternative hypothesis, The Hausman’s test may not be reliable under certain conditions. Given that GLS-RE remains an efficient and unbiased estimator and corrects for both serial correlation and heteroskedasticity, we prefer to retain this estimator.

^18. Note that among the industrials firms, Braskem (Brazil) appears as the only inconsistent case. Although its credit rating pierced the sovereign ceiling from June 2003 until November 2003, market views seem to reflect the opposite, as we accept the null hypothesis that market participants apply the sovereign ceiling rule to the Braskem´s bond.

^19. These firms are YPF SA (Argentina) from 1997 to 2004, Telenorte (Brazil) from June 2003 until November 2003, Televisa Group (Mexico) from June 2004 until January 2005, Kimberley Clark (Mexico) from July 1999 until November 2005, America Movil (Mexico) from August 2002 until January 2005, Coca Cola Femsa (Mexico) from October 1996 to date, and CEMEX (Mexico) from November 1997 until January 2005.