On the rationality of professional forecasts of corporate bond yield spreads

Abstract

In evaluating the multiperiod forecasts of the corporate bond yield spread from the Survey of Professional Forecasters (SPF), it is shown that the consensus forecasts are generally unbiased and consistently outperform the comparable ARIMA forecasts and are thus, at least, weakly rational.

Notes

The SPF forecasts are available on the website of the Federal Reserve Bank of Philadelphia. The use of the median response instead of the individual forecasts is consistent with Zarnowitz (1985), who shows that the consensus forecast generally tends to be more supportive of the expectational rationality than the individual forecasts.

The time plot of (Aaa–TB), not reported here, shows an upward trend in the spread. This may imply that corporate bonds have become riskier as investors have continually demanded higher returns (see Lamdin, 2004, on this and some other related issues).

For a detailed description of this procedure, which combines Hansen's (1982) and White's (1980) methods, see Fair and Shiller (1990, p. 378). Note that the covariance matrix of Equations 3 and 4 for f = 0 is corrected only for heteroscedasticity.