Recent calibration studies suggest that a sizable component of corporate bond spreads cannot be explained by credit risk (Jones, Mason and Rosenfeld, 1984; Elton, Gruber, Agrawal and Mann, 2001; Eom, Helwege and Huang, 2004), especially for high-rung investment-grade bonds at the short maturity (Huang and Huang, 2003). While existing empirical studies link this nondefault component of bond spreads to bond liquidity (Longstaff, Mithal and Neis, 2005; Ericsson, Reneby and Wang, 2006; Nashikkar and Subrahmanyam, 2006), most of the evidence suggests that only speculative-grade bonds have a significant exposure to the liquidity risk factors (see Chen et al. (2005), Chacko (2005), de Jong and Driessen (2004), and Downing, Underwood and Xing (2005), among others).
This disagreement between existing calibration results and empirical findings may be caused by the correlation between liquidity risk and credit risk (see, e.g., Liu, Longstaff and Mandell (2004) for a reduced-form model and Ericsson and Renault (2006) for a structural model on the correlation). Lacking either appropriate, especially time-varying, bond-level liquidity measures or adequate control for the firm-specific credit risk, the existing empirical studies face serious challenges of identifying the liquidity effect on bond spreads.
In this paper, we construct various stochastic measures for corporate bond liquidity using their intraday transactions data. These measures contrast to the conventional liquidity proxies based on bond characteristics which are either constant or deterministic with the passage of time. In addition, to better control for credit risk, we use firm-specific term structures of CDS spreads to estimate the default component of bond spreads. In doing so, we address both maturity mismatch and coupon effects issues that were only partially controlled for in the existing studies. Moreover, our new methodology gives us enough degree of freedom to apply the fixed-effect method to account for the unobservable firm heterogeneity that have biased existing estimations.
We find a positive and statistically significant relationship between the illiquidity of intraday trading and the nondefault bond spreads. Importantly, our evidence across bond ratings and maturities is consistent with the so-called credit spread puzzle in that high-rung investment-grade bonds with short maturities have relatively larger nondefault spreads. This finding contrasts to those in the existing empirical studies that the liquidity impact on bond spreads is either generally marginal or only significant for the speculate-grade bonds. In the rest of the introduction, we discuss in more details about these key methodological innovations and new empirical findings.
