Abstract
In order to fully understand the changes in gas component prices and, more
importantly, to predict future prices and their effect on both production
and investment decisions, it is vital that we model them appropriately. The
approach shown in this paper that uses a time series with unobservable components
is employed with a stochastic underlying trend and seasonality, using
monthly data (January 1995 to 2006 November) for the propane, butane and
naphtha traded in the north European market. We test the predictive power
of fitted models using various hold-out samples. The in-sample and out-ofsample
results indicate that gas component prices follow stochastic processes,
with levels and slopes shifting continuously and unpredictably over time while
seasonal patterns seem to be fixed. This suggests a random walk with fixed
seasonal parameters for the time series. Predicting gas component prices for
horizons relevant for gas processing production and investment decisions will
therefore be a very challenging task.
importantly, to predict future prices and their effect on both production
and investment decisions, it is vital that we model them appropriately. The
approach shown in this paper that uses a time series with unobservable components
is employed with a stochastic underlying trend and seasonality, using
monthly data (January 1995 to 2006 November) for the propane, butane and
naphtha traded in the north European market. We test the predictive power
of fitted models using various hold-out samples. The in-sample and out-ofsample
results indicate that gas component prices follow stochastic processes,
with levels and slopes shifting continuously and unpredictably over time while
seasonal patterns seem to be fixed. This suggests a random walk with fixed
seasonal parameters for the time series. Predicting gas component prices for
horizons relevant for gas processing production and investment decisions will
therefore be a very challenging task.