Modeling natural gas markets in Europe under oligopolistic competition: the GaMMES model and some applications.

Summary This thesis studies the evolution of natural gas markets in Europe until 2035 using modeling tools. The proposed model, GaMMES, is based on an oligopolistic description of the markets and its main advantages are: a high level of detail of the economic structure of the gas chain and an endogenous consideration of long-term contracts upstream as well as substitution with oil products and coal on the demand side. First, we study the issue of gas supply security in Europe and the conditions for regulating markets vulnerable to the risk of supply disruption, especially from Russia. Three case studies are proposed according to the degree of dependence and the nature of regulation in place: the German market of the 1980s and the current markets of Bulgaria and Spain. In particular, we study the evolution of market characteristics as a function of the risk of disruption and the type of regulation to be put in place to ensure social welfare optimality. Then, we propose a dynamic systems model to take into account the energy substitution between coal, oil and natural gas. Our approach allows us to estimate a new functional form of the demand function for natural gas, which includes both energy substitution and consumption inertia due to end-user investments. In a third step, we use this demand function in a partial equilibrium model of natural gas markets in Europe. The GaMMES model, written as a complementarity problem, represents the main players in the natural gas industry by considering their strategic interactions and market powers. It has been applied to the natural gas market in North-East Europe in order to study the evolution, until 2035, of consumption, spot prices, long-term prices and volumes, production and dependence on foreign imports. Finally, we propose a stochastic extension of the GaMMES model in order to analyze the impact of the strong fluctuation of the Brent price on the gas markets. An econometric study was conducted to calculate the probability law of the oil price, when modeled as a random variable, in order to build and weight the scenario tree. The results allow us to understand how the randomness modifies the strategic behaviors of the actors, in particular at the level of long-term contracts. Finally, the value of the stochastic solution is calculated in order to quantify the importance of taking into account oil price fluctuations for each actor in the chain.