Credit risk premiums estimated from credit default swap spreads are shown to be directly related to equity returns and seem to contain economic information that is not reflected in the related physical or default probabilities alone.
In a quantitative study, the authors analyze the relationship between a firm’s stock returns and credit risk. They confirm Robert Merton’s theory that the market pricing of risk is consistent across all contingent claims on assets. Estimating risk premiums from firms’ credit default swap (CDS) forward curves, they identify a significant direct correspondence between these premiums and firms’ equity returns and Sharpe ratios. The modeled CDS spread term structure includes information relevant to stock pricing not captured by traditional risk factors. Finally, addressing the “distress puzzle” anomaly, they show an inverse relationship between physical/risk-neutral default probabilities and equity returns.
How Is This Research Useful to Practitioners?
Identifying predictable pricing links between different financial instruments for the same underlying asset provides investors with potential arbitrage and risk management opportunities. The authors’ assessment of CDS spreads adds another item to the asset management toolkit, with resultant gains in functionality as a real benefit for portfolios with exposure to associated holdings. A rewarding long–short investment possibility of buying high and selling low credit risk premium firms is noted, with such portfolios generating positive alpha after controlling for standard risk factors, such as size, liquidity, and distress risk.
But as with any tool, applications that are both careful and appropriate are essential. In this regard, the authors offer some prudential guidance. First, naive use of a firm’s CDS spreads from direct observation of either market prices or estimates of the firm’s physical default probability alone is insufficient to draw accurate inferences of the firm’s equity returns. Informative modeling must encompass both sources of information. Second, CDS spreads are a preferred metric for estimating risk premiums rather than corporate bond yield spreads because they are timelier and less contaminated by tax and liquidity effects. Third, although context can have tangible effects, such as with the potential returns being more extreme for the examined portfolios during the recent financial crisis, the overall findings are shown to be equally valid across the entire time period studied.
How Did the Authors Conduct This Research?
The authors’ empirical strategy is to estimate risk premiums and then relate them to subsequent equity excess returns of the studied US firms from 2001 to 2010, deriving firm-specific credit risk measures from the CDS forward curve. Data from Markit cover the daily CDS spreads for 675 US dollar–denominated contracts in five yearly maturities (1, 3, 5, 7, and 10). They also incorporate the standard payment protection terms applicable before and after the implementation of the CDS “Big Bang” Protocol (the incorporation of auction settlement terms into standard CDS documentation) in April 2009. For the equity return match, daily data from CRSP and monthly firm fundamentals and credit ratings from Compustat are used to compile approximately 839,000 joint observations for the 491 firms evaluated.
These measures are compared with the risk premiums applied by affine term-structure models. The authors sort firms into monthly portfolios differentiated by their estimated risk premiums and then compared with their related equity returns. The portfolios are checked for monotonic patterns related to firms’ size, book-to-market, risk-neutral, or physical default probabilities; CDS contract liquidity; or conditional co-skewness with the market. No such patterns are found. Firm characteristics are tested for effects, with excess returns shown to be highest for small firms, firms priced as value stocks, and firms with high default probabilities. The authors also test industry effects by excluding financial and utility firms from the sample (with the results remaining the same as before the exclusion). Parameters are estimated for both full-sample and out-of-sample data.
Given how financial theory can influence investor actions, an academic demonstration of a sound empirical basis for underlying principles is an effort worth making. The authors acknowledge that their work is only a starting point, potentially enhanced by more sophisticated econometric techniques, additional theorizing on the economics behind the CDS risk factor, and better understanding of the interplay between the affected markets’ institutional and regulatory features. The question of how closely CDS market mispricing during the financial crisis was reflected in real time in the equity markets, and potential predictive functionality, is not addressed.