A comprehensive look at stock return predictability by oil prices using economic constraint approaches

This study investigates the predictability of oil return on stock market return using a series of economic constraints. We find that oil return has a more powerful and stable prediction ability than its asymmetric form using an unconstrained approach and three constraint approaches. A new constraint, regarding the three-sigma rule, can obtain a higher forecast accuracy than other methods. Moreover, compared to univariate macro models, incorporation of oil return can increase the average forecasting performance of 14 macroeconomic predictors. Finally, the predictive performance of oil returns varies during different periods linking to the business cycle, geopolitical risk, and financial crisis. The predictability source of oil returns can be explained from the discount rate channel and the sentiment channel.