Real-Time Demand in U.S. Natural Gas Price Forecasting: The Role of Temperature Data

Abstract

This paper provides evidence of the pivotal role temperature data can play in forecasting natural gas prices at the Henry Hub in real time. Considering a newly constructed temperature index as an additional exogenous variable in a Bayesian vector autoregressive (BVAR) framework significantly increases forecast accuracy at horizons of up to 12 months. Our novel approach to energy price forecasting simultaneously considers both supply and demand and incorporates temperature data as a proxy of real-time demand for natural gas.

Keywords: energy prices, natural gas, Bayesian VAR, price forecasting, real-time data, temperature

JEL Classification: C11, C32, C53, Q41, Q47