The Term Structure of Credit Spreads and the Cross-Section of Stock Returns by SSRN
University of Toronto, Rotman School of Management
University of California, Los Angeles (UCLA) – Finance Area; Centre for International Finance and Regulation (CIFR)
Florida State University, College of Business, Department of Finance
February 22, 2015
Rotman School of Management Working Paper No. 2560693
Abstract:
We explore the link between credit and equity markets by considering the informational content of the term structure of credit spreads. A shallower credit term structure predicts decreases in default risk, increases in future profitability, as well as favorable earnings surprises, and vice versa. Further, the slope of the credit term structure negatively predicts future stock returns, and this result does not arise from a premium for default risk. Rather, limited attention and arbitrage costs play important roles: return predictability from the credit spread slope holds mainly for stocks with low institutional ownership, analyst coverage, and stock liquidity.
The Term Structure Of Credit Spreads And The Cross-Section Of Stock Returns – Introduction
Firms finance their capital needs using a combination of debt and equity. To what extent are markets for these claims linked to each other and to other economic fundamentals? This question has long preoccupied scholars. For example, Fama and French (1993) find that excess returns on U.S. stocks and corporate bonds are positively related to the slope of the yield curve.1 Researchers have also shown that the aggregate credit spreads – the difference between corporate and treasury yields – forecasts economic activity such as output and investment growth (e.g., Stock and Watson (1989), Lettau and Ludvigson (2002)) as well as future stock market returns (e.g., Keim and Stambaugh (1986), and Fama and French (1993)). In contrast to the preceding studies, which consider equity and debt markets at the aggregate level, we study the informational content of the term structure of credit spreads at the individual firm level and examine their predictive power for the cross-section of stock returns. Our underlying economic reasoning is that credit markets contain important information about the future financial health of the firm, which, in turn, could influence future stock price movements, provided the two markets are at least partially segmented. We provide evidence that indeed, credit spreads do forecast firms’ fundamentals as well as equity market returns in the cross-section.
To measure the term structure of credit spreads for individual firms, we use data from the credit default swap (CDS) market, which has grown tremendously and has become increasingly liquid during recent years.2 In our data, we observe CDS spreads each day on the same set of maturities (ranging from 1 to 10 years). Barring arbitrage, the CDS spread for a given maturity should be equal to the credit spread, that is, the difference in yield to maturity between a corporate bond and a U.S. Treasury bond with the same maturity.
The CDS data have important advantages over the corporate bond data. First, unlike credit spreads based on bond yields, CDS spreads are not subject to the specification of a benchmark risk-free yield curve. Second, CDS contracts are much more liquid than corporate bonds. CDS contracts are traded on a daily frequency while corporate bonds are usually held to maturity and may not trade even once in a month. Compared to credit spreads, CDS spreads are less contaminated by nondefault risk components (Longstaff, Mithal, and Neis (2005) and Ericsson, Reneby, and Wang (2006)). Third, CDS prices lead credit spreads in the price discovery process (e.g., Blanco, Brennan, and Marsh (2005)). Fourth, the contract terms are standard and easily comparable across firms, making the CDS data more suitable for cross-sectional study.
We find that the slope of the term structure of CDS spreads, defined as the difference between a 5-year and a 1-year CDS spread, significantly and negatively predicts cross-sectional stock returns. Stocks ranked in the bottom decile by CDS slope on average outperform those ranked in the top decile by more than 1.20% per month (14.40% annualized). The negative relation between a CDS slope and an average future stock return is robust to weighting schemes. It holds with Fama-MacBeth regressions and is robust to controlling for stock characteristics known to be related to the cross section of a stock return. The Fama-French factors and the momentum factor explain little of the significant positive average return of the portfolio that buys low CDS slope stocks and shorts high CDS slope stocks. These results indicate that the slope of the CDS term structure contains useful fundamental information about the firm, but such information gets incorporated only gradually into stock prices.
Compensation for the probability of default is not able to account for our result. The predictive power of the CDS slope for stock returns is robust to controlling for various default risk measures such as Moody’s KMV default metric, and the measure constructed by Campbell, Hilscher, and Szilagyi (2008).3
1 For evidence that the term structure of interest rates can forecast aggregate stock returns, see also Campbell (1987), Boudoukh and Richardson (1993), Zhou (1998) etc.
2 According to BIS, CDS markets grew from $0.6 trillion in notional amount outstanding in 2001 to $62 trillion in 2007, but dropped to $26.5 trillion in July 2009 because of the recent financial crisis. The total amount of CDS contracts outstanding at end of June 2013 was $24.5 trillion.
3 Some studies find a positive cross-sectional relationship between expected stock returns and default risk (e.g., Vassalou and Xing (2004) and Chava and Purnenandam (2010)). However, other studies document that returns are actually lower for firms with high financial distress risk (see, e.g., Griffin and Lemmon (2002), Campbell, Hilscher, and Szilagyi (2008)). In our sample, there is no significant relation between these measures of default risk and average stock returns.
See full PDF below.
