The Predictability Implied by Consumption-Based Asset Pricing Models: A Review of the Theory and Empirical Evidence

Abstract

The consumption-based models have a lack of predictive power for explaining variability of stock returns. This paper examines two well-known models, Campbell and Cochrane (1999)’s habit model and Bansal and Yaron (2004)’s long-run risks model, to see whether they produce a significant power of return predictability. For the habit model, empirical tests reveal that the state variable, the surplus consumption ratio, explains counter-cyclical time-varying expected returns. The long-run risks model also proves to explain that main sources of volatility in price-dividend ratio are a persistent and predictable consumption growth rate and fluctuating economic uncertainty. The models are also tested by following the work of Kirby (1998) whether they can explain the observed return predictability. Both models fail to generate any significant predictive power. The habit model is relatively strong in volatility, which implies that variation in expected excess return is largely attributable to the time-varying risk aversion.