Forecasting the Equity Premium with State Variables Estimated from the Yield Curve

Abstract

I propose a novel method to estimate a state variable that summarises the information content of the yield curve, and find that unexpected changes in the state variable significantly predict the expected equity premium on the next trading day. This predictor outperforms changes in the short rate, and the term and credit spreads, especially during recessions. The state variable changes also have much more predictive ability than do changes in the level, slope and curvature factors commonly used to summarise the information content of the yield curve. The two step state variable estimation procedure is intuitive: first I estimate the parameters of the equation of motion governing the evolution of the state variable for a chosen term structure model. The yield for a given maturity is then a known function of the state variable and the parameters estimated in the first stage; inverting this function using the observed yield curve gives the estimate.