Evaluating Predictors within a Present-Value Framework

Abstract

Listen

Translate article

I impose functional-form restrictions on the time-series processes of expected returns and expected dividend growth rates to better estimate them in a small sample. The approach helps to aggregate information contained in the entire history of prices, dividend growth, and additional predictors without parameter proliferation. I find that both expected returns and expected dividend growth rates are substantially time-varying, positively correlated with each other, and covary with several macroeconomic variables. The estimated expectations of returns and dividend growth rates are strong predictors of realized returns and dividend growth rates, respectively, both in-sample and out-of-sample. Book-to-Market Ratio, Stock Variance, Consumption-Wealth-Income Ratio, and BAA-rated Corporate Bond Yield signi cantly improve the return and dividend forecasts of my present-value model.