Forecasting gasoline prices using oil prices: New evidence based on the rocket and feather hypothesis

The rocket and feather hypothesis (RFH), which captures the asymmetrical reaction of gasoline prices to crude oil price movements, has received substantial interest. However, there has been limited successful exploration of its application in enhancing the predictability of gasoline prices. This paper addresses this gap by constructing an RFH-based prediction framework. The empirical results indicate that considering the asymmetry of crude oil prices exhibits statistically and economically significant in-sample and out-of-sample predictive performance for future gasoline prices, and the superiority holds using regularization models. Further analysis demonstrates that utilizing predictive models customized to different economic conditions leads to improved forecasting performance. Our findings can survive across a series of robustness tests. Moreover, this paper verifies the more prompt response of gasoline prices to increases in crude oil prices, thereby explaining the newfound evidence that incorporating RFH can significantly improve gasoline price predictability.