Structural credit risk models cannot adequately explain credit spread levels, credit spread changes, and observed defaults. On one hand, this empirical failure has prompted researchers to reconsider the role of idiosyncratic volatility and jumps to better describe credit risk. On the other hand, a growing number of studies argues that illiquidity might be an additional determinant of corporate bond prices. However, in informationally effcient markets, trading decisions are simultaneously a ffected by investors' assessment of credit risk (information side) and by the liquidity of the bond that they wish to trade (friction side).
In order to avoid attributing to liquidity what is in fact credit risk, I propose a stochastic friction model of corporate bond returns and liquidity developing on the idea that, while systematic and idiosyncratic credit risk variables aff ect expected returns, bond returns are observed only if there is enough valuable information to justify their liquidity costs. After obtaining several trading-based liquidity measures, I conduct a cross sectional regression analysis to assess the relative importance of illiquidity and credit risk variables in the determination of corporate yield spreads.
In recent applications of the standard friction model (Rosett (1959)) to stock and bond liquidity, systematic factors summarize all the relevant information and liquidity is constant (Lesmond, Ogden, and Trzcinka (1999), and Chen, Lesmond, and Wei (2007)). However, these assumptions are quite restrictive given that the contingent claim approach to structural credit risk modeling assigns an important role to idiosyncratic variables and given the evidence that corporate bond liquidity is time-varying.
I improve on the standard approach in several ways. First, I model bond returns as a function of individual factors such as equity returns and realized volatility. The use of high frequency measures, such as realized volatility, is better able to identify jumps, which carry more information than historical volatility estimated with daily data. Moreover, considering the relation between bond and equity returns of the same rm allows for a hedging interpretation, rather than a linear asset pricing model interpretation with all its caveats.