The authors investigate how stock and corporate bond markets adjust to new information. They confirm the existence of substantial lags in price discovery between related markets. They also find that high-yield corporate bond markets are informationally efficient and use past bond returns to predict returns for corresponding stocks. Bond momentum is particularly effective when stocks are volatile or have recently increased in value.

In theory, new information spreads into markets simultaneously and should be incorporated into asset pricing immediately. But equity investors recognize new information with a lag as a result of constraints in equity markets.

The authors investigate the relationships between stock and bond markets to assess whether a price movement in corporate bonds can predict price changes for corresponding stocks. They show that stock markets react with a lag when information is negative. Equity returns are anticipated only by high-yield bond markets, whereas investment-grade bond markets seem to have no predictive power.

An abnormal price decline in a corporation’s most liquid bond over a month (i.e., bond past return in the lowest decile) is followed by an average abnormal stock price decline of −1.42%. This effect is statistically significant and larger for firms whose stocks previously increased in value (decline of −2.47%) or experienced high volatility. It is also larger for bonds with high coupons and shorter maturities.

The authors attribute the asymmetrical response to negative information to psychological factors and stock short-sale constraints that cause equity markets to delay price adjustments. They conclude that high-yield bonds have characteristics similar to equities and, in the absence of barriers, anticipate stock market changes.

The research is based on Trade Reporting and Compliance Engine (TRACE) bond trade data for 675 firms for July 2002 to December 2008. TRACE represents the activity of US bonds traded in the over-the-counter market. The daily bond returns include accrued interest obtained from the Mergent Fixed Income Securities Database. Corresponding stock returns are obtained from CRSP.

The authors adjust TRACE data by excluding all trades with values of less than $100,000 to remove volatility in the calculation of a daily trade-weighted price. They focus on firms’ most liquid bonds with a minimum of one trade per day and calculate equally weighted portfolio returns for bonds with similar credit ratings (i.e., investment grade, high yield, and not rated) and years to maturity.

Subsequently, the authors calculate abnormal bond returns as excess return over the average portfolio return in the bond’s category. They use regressions to model the current abnormal returns of stocks as a function of the corresponding one-month lagged abnormal bond returns and one-month lagged abnormal stock returns. An abnormal return factor is defined as the prior month’s return that is in the highest or lowest decile. Regression coefficients represent the bond and stock momentum factors that are used to predict abnormal stock movements.

Regression results are stratified by a bond’s years to maturity (i.e., default risk driver), coupon value, and the idiosyncratic volatility of stocks to measure the predictive power of each factor. To ensure that the conclusions are valid, the authors confirm the statistical significance of each coefficient using a t-test.

Contrary to common perception, high-yield corporate bonds seem to have an informational edge over the equity market when news is negative and stock returns are volatile. This finding also sheds new light on how the information is incorporated into asset prices in related markets. The study is useful for portfolio managers who may consider adding a lagged bond return variable to their stock valuation models. Further studies could look at the relationships among related markets. The authors’ study is comprehensive, but analysis of the most recent data to represent improved communication would make the findings even more relevant.