Forecasting U.S. Bond Returns

Abstract

The author examines the ability of variables such as steepness of the yield curve, real bond yield, recent stock market performance, and bond market momentum to predict bond returns. He shows that these indicators are more effective at predicting returns collectively than individually. He uses historical data to back test static and active investment strategies tied to these predictors. Not all investors are willing to replicate these allocation methods, because the forecasts are incorrect 40 percent of the time and expected losses in the short term may be viewed as a high cost to bear for potential long-term gains.