Short-term stochastic movements of electricity prices and long-term investments in power generating technologies

Abstract

Modeling probability distributions for the long-term dynamics of electricity prices is of key importance to value long-term investments under uncertainty in the power sector, such as investments in new generating technologies. Starting from accurate modeling of the short-term behavior of electricity prices, we derive long-term stationary probability distributions. Then, investments in new baseload generating technologies, namely gas, coal and nuclear power, are discussed. In order to compute the stochastic Net Present Value of investments in new generating technologies, the revenues from selling electricity in power markets as well as the costs which come from buying fuels at uncertain market prices must be evaluated over very long time horizons, i.e., over the whole lifetime of the plants. Starting from accurate short-term stochastic models of fuel prices in addition to electricity prices, we provide long-run probability distributions which are used to compute revenues and costs incurring during the whole lifetime of the plants. Five sources of uncertainty are taken into account, namely electricity market prices, fossil fuel prices (natural gas and coal prices), nuclear fuel prices and CO2 prices. Our evaluation model is calibrated on empirical data to account for both historical market prices and macroeconomic views about future trends of electricity and fuel prices. The full probability density of the stochastic Net Present Value is thus determined for each generation technology considered in this study.