In any balanced portfolio, investors need to assess the relative attractiveness of equities and bonds, the usual asset classes `competing' for funds. A widely used tool in asset allocation decisions is the so-called FED. The critique of the FED model has not always been fair and this paper, therefore, presents a behavioral defense of the FED model. By combining the FED model with the Capital Asset Pricing Model (CAPM), it becomes evident that the FED model is able to detect time variation in the equity risk premium and behavioral biases in long-term earnings growth expectations. Assuming that share prices are the sum of a fundamental value element and a noise/sentiment element, then the use of statistical tools such as confidence intervals will reduce potential decision biases caused by noise/sentiment and thereby improve the predictive power of the FED model. The results in this paper suggest that the FED model does a better job at predicting relative returns of stocks versus bonds than at predicting absolute stock returns. By basing decisions only on data points outside a predetermined confidence interval, the predictive power is increased manifold, enhancing potential gross returns and reducing transaction costs. The optimal prediction horizon for the FED model appears to be 12-36 months, somewhat shorter than the 5-10 year horizon found for the P/E mean reversion model proposed by Campbell-Shiller. Thus, the FED model and the long-term P/E mean reversion model are complementary models of return prediction, not competing models.

This paper is aimed at the practitioner and will combine insight from Behavioral Finance with statistical analysis in order to develop a simple market timing tool. A defense of the FED model and its usefulness as a prediction model of future relative returns between stocks and bonds is presented in this paper. Many previous tests of the FED model's predictive ability have focused on absolute stock returns over the long run (5, 10 and 20 years) and have used all data points (months) in the prediction and in general found low predictive power. My approach focuses on the relative returns between stocks and bonds and shows that the FED model works its best at the short and medium term (up to 3 years) when using only data points (months) outside a predetermined confidence interval. Thereby, the FED model is complementary to the P/E mean reversion model proposed by Campbell-Shiller, which works best over the longer-term (5-10 years). By combining the FED model with the CAPM, it is observed that the FED model is able to detect time variation in the equity risk premium and behavioral biases in earnings expectations around bull and bear market extremes.

A Behavioral Defense of the FED Model,predictive, relative, confidence, biases, stocks, element, expectations, equity, horizon, interval, market, noisesentiment, predicting, premium, behavioral, decisions, statistical, enhancing, equities, ability, fundamental, increased, investors