A1258
Title: Analyzing the EPU effect on the risk of extreme events in the oil market: A MIDAS touch to dynamic POT models
Authors: Katarzyna Bien-Barkowska - Poznan University of Economics and Business (Poland) [presenting]
Agata Kliber - Poznan University of Economics and Business (Poland)
Rodrigo Herrera - Universidad de Talca (Chile)
Abstract: The potential of the economic policy uncertainty (EPU) index is analysed to enhance the forecast of tail risk dynamics in the oil market. Specifically, a novel approach is proposed to harness the temporal clustering of such extreme events and their intensity in oil market returns by using a dynamic peaks-over-threshold (POT) model in discrete time that incorporates the EPU index within a MIDAS framework. The proposed model, referred to as the autoregressive conditional MIDAS-POT (AC-MIDAS-POT), combines a dynamic specification for inter-exceedance times and exceedance magnitudes, both of which are supplemented with MIDAS components to integrate macroeconomic information. Applied to Brent and WTI data, the findings reveal that increases in policy uncertainty significantly elevate the frequency and magnitude of extreme events, especially in the right tail of the distribution for Brent and, to a lesser extent, for WTI. The AC-MIDAS-POT model demonstrated superior risk forecasting performance compared to traditional GARCH-MIDAS models using Gaussian and Student t distributions, both in-sample and out-of-sample. This framework provides a robust tool for forecasting tail risk in oil markets and offers valuable insights for investors and policymakers to anticipate and mitigate the adverse effects of extreme price fluctuations.
