Title: Crudeo oil return predictability revisited
Authors: Thomas Conlon - University College Dublin (Ireland) [presenting]
Abstract: Out-of-sample crude oil return predictability is re-examine by using a large set of predictors and finds that previous studies overstate the evidence of predictability. This overstatement comes from the significantly high first-order autocorrelation in monthly average crude oil returns, computed from monthly averages of daily prices of West Texas Intermediate crude oil spot provided by the U.S. Energy Information Administration. Monthly average returns would not have been available to the forecaster in real time. Following the convention in the stock, bond, currency, and commodity return predictability literature, we compute crude oil returns using end-of-month prices. Using this return series, we find no evidence of crude oil return predictability, reversing the conclusion of previous studies. More specifically, individual predictive and combination model forecasts of crude oil returns generated using popular economic and technical indicator variables fail to outperform the simple historical average return forecast.