Valuation of a Combined Cycle Gas Turbine: under price uncertainty and operational constraints
Abstract
In this thesis we combine multivariate time series modelling with real options theory to value a combined cycle gas turbine. We propose a novel price model with co-integrated power, gas and carbon prices, with multivariate stochastic volatility and MNIG distributed errors. The estimated model is found to outperform competing specications in terms of higher likelihood and lower information criteria. We implement a Least Squares Monte Carlo (LSM) simulation to value the plant, incorporating ramp times, startup costs and variable plant eciency. We take into account that day-ahead prices are settled the day before prices take eect, which is often overlooked in related literature. We nd that ignoring this leads to suboptimal choices and a lower value estimate. An analysis of the regressions in the LSM algorithm reveals that the choice of basis functions has a signicant eect on the estimated value of the plant. Particularly, for a low-eciency plant, a regression on the spark spread underestimates the value by 20% compared to a regression on both the electricity price and the fuel cost components. This implies that in spread option valuations where the LSM is applicable, simulating all asset or commodity prices may be advantageous over simulating the spread alone.