Estimation of the Number of Factors in a Multi-Factorial Heath-Jarrow-Morton Model in Power Markets

We study the calibration of specific multi-factorial Heath-Jarrow-Morton models to power market prices, with a focus on the estimation of the optimal number of Gaussian factors. We describe a common statistical procedure based on likelihood maximisation and Akaike/Bayesian information criteria, in the case of a joint calibration on both spot and futures prices. We perform a detailed analysis on three national markets within Europe: Belgium, France, and Germany. The results show a lot of similarities among all the markets we consider, especially on the optimal number of factors and on the behaviour of the different factors.