Out-of-Sample Density Prediction of the End-of-Month Price of Crude Oil and the U.S. Economic Policy Uncertainty Index

Abstract Results reported in the recent studies of Conlon et al. (2022. “The Illusion of Oil Return Predictability: The Choice of Data Matters.” Journal of Banking & Finance 134: 106331) and Ellwanger and Snudden (2023. “Forecasts of the Real Price of Oil Revisited: Do They Beat the Random Walk?” Journal of Banking & Finance 154: 106962) have created a good deal of debate and controversy regarding the nature and extent of in- and out-of-sample crude oil price predictability when the end-of-month price is predicted instead of the monthly average price. At the same time, the newspaper-based U.S. economic policy uncertainty (EPU) index suggested in the study of Baker et al. (2016. “Measuring Economic Policy Uncertainty.” Quarterly Journal of Economics 131: 1593–636) has shown itself to be a very useful predictor of economic variables, such as equity returns and volatility. I assess the predictive impact of the log-EPU index on the end-of-month price of crude oil as opposed to the monthly average price. While the log-EPU index does not improve the relative accuracy of crude oil price return predictions, it has a statistically significant in- and out-of-sample predictive impact on the conditional distribution of the end-of-month log-crude oil price returns through the conditional volatility channel. The statistical evidence of relative predictability also translates into economic gains.