Modeling gasoline price volatility

Abstract

We investigate the asymmetry in gasoline price volatility using a new pseudo leverage heterogeneous autoregressive (P-LHAR) model. The model introduces a common leverage factor derived through principal component regression, replacing the traditional individual leverage factor. We apply this model to forecast the volatility of RBOB gasoline prices. Our results show that the P-LHAR model outperforms the traditional HAR, LHAR and combination models, especially in turbulent periods, on weekly and monthly horizons. In-sample estimates indicate the relevance of the common leverage factor, which further strengthened by the superior out-of-sample predictive performance across various horizons. Robustness checks confirm the model's reliability.