Multivariate Modelling of Energy Markets

University dissertation from Department of Economics, Lund Universtiy

Abstract: This thesis contributes to the empirical energy finance literature and consists of three research papers. The common denominator for all papers is the multivariate modelling approach, placing the electricity market at the core and delving into its interdependencies on fundamentally related markets and factors.In the first paper, we focus on the multivariate modelling of the return series of electrical power, natural gas, coal and carbon emission allowances in the German energy market. We pay special attention to selecting an appropriate volatility model allowing for cross-commodity effects, coupled with a flexible skew-Student distribution for the error terms. We discuss the relationship between the discovered volatility spillover effects and the fundamental developments in the energy markets.The second paper develops a comprehensive analysis of the transmission of independent shocks from the gas, coal and carbon markets to the power market, building on the estimation results from the first paper and employing the novel Volatility Impulse Response Function (VIRF) methodology. We find that spillover effects show significant time variation and are substantial in size.The third paper presents a model for the joint dynamics of the Nordic system spot power price together with its major demand-side factor, the outdoor temperature, and its major supply-side factor, the hydrological balance. We demonstrate how the model can be utilized in meteorological scenario analysis.