CFE 2015: Start Registration
View Submission - CFE
A0852
Title: Understanding expected returns Authors:  Andrea Tamoni - Rutgers Business School (United States) [presenting]
Daniele Bianchi - University of Warwick (United Kingdom)
Abstract: The predictive power of forecasting variables studied in the literature varies with the horizons. For example, the forecasting power of consumption-wealth ratio is particularly strong at short to intermediate horizons. On the other hand the dividend-price ratio tracks longer-term tendencies in asset markets. We show how to use the medium- and long-term information content of standard predictors to extract short-term expected returns. To combine information across horizons (or levels of resolution) we adopt a framework that relies on multi-scale models for time series. We show that our filtered expected return series exhibits aggregation properties which differs wildly from those of a standard autoregressive process of order one. In particular, our filtered monthly expected return series aggregated over the same temporal scale of the predictor used in the estimation, has an autocorrelation function which decays much slower than that obtained from aggregating a persistent, monthly AR(1) process. This result shows the importance of considering the scale at which the predictor conveys information about returns, as this may lead to different conclusions with respect to the high-frequency dynamics of the expected return process.