Modeling the Price of Natural Gas with Temperature and Oil Price as Exogenous Factors

Abstract

The literature on stochastic models for the spot market of gas is dominated by purely stochastic approaches. In contrast to these models, Stoll and Wiebauer [14] propose a fundamental model with temperature as an exogenous factor. A model containing only deterministic, temperature-dependent and purely stochastic components, however, still seems not able to capture economic influences on the price. In order to improve the model of Stoll and Wiebauer [14], we include the oil price as another exogenous factor. There are at least two fundamental reasons why this should improve the model. First, the oil price can be considered as a proxy for the general state of the world economy. Furthermore, pricing formulas in oil price indexed gas import contracts in Central Europe are covered by the oil price component. It is shown that the new model can explain price movements of the last few years much better than previous models. The inclusion of oil price and temperature in the regression of a least squares Monte Carlo method leads to more realistic valuation results for gas storages and swing options.

A joint model for temperature and natural gas with an application to the US market

“…Recently Stoll and Wiebauer (2010) have proposed a model that relates the natural gas price at a European hub, the Title Transfer Facility in the Netherlands, to the cumulated Heating Degree Days (HDD) measured in a German meteorological station, the one located in Hannover. Their model incorporates in gas price a temperature component via HDD, an average cumulated temperature (below a threshold) over a winter (beginning on the 1 st of October), as a proxy of the lling level of gas storage; their approach has been generalized by Müller et al (2015), who include also oil price as another exogenous factor in a model for Central Europe gas prices.…”

Gas storage valuation with regime switching

“…It has been shown in [27,17,10] that a one-factor mean-reverting model does not capture the dynamics of a typical gas forward curve. Hence there is on the one hand the possibility to use multifactor models, like in [8,6,19,22] or to introduce regime-switches, see e.g. [10].…”