This article analyzes the Commodity Term Structure by using a two-factor model based on the spot price and a convenience yield in a regime switching framework. We use an extension of the Gibson-Schwartz model that may run through different regimes illustrating several equilibrium behaviors of the spot price and the convenience yield. It is assumed that noisy observations of future contracts are available in such a way that commodity markets dynamics
is represented by a regime switching Conditional Linear and Gaussian state space model. We
propose first a new algorithm to approximate the conditional distributions of the states (spot,
convenience yield and regime indicator) given a set of observations. These approximations are
obtained by combining Sequential Monte Carlo methods and Kalman smoothing techniques.
Then, model parameters are estimated with an Expectation Maximization based algorithm.
The performance of the proposed procedure is assessed with Chicago Mercantile Exchange
crude oil data.