| Restructuring of the electricity industry in the past two decades has led to higher volatilities of power prices in competitive wholesale markets. Some electricity market crises, such as the one in California, inspired the use of risk management in newly restructured power markets. This thesis studies high price risks measured by Value at Risk (VaR) in the tails of the underlying distributions of electricity prices. Compared with traditional normal distribution modeling and historical simulation modeling, extreme value theory (EVT) has obvious advantages in estimating VaR by directly dealing with the tails of the underlying distributions and allowing fat-tailed and asymmetric behavior. This argument is confirmed by the empirical data analysis using Alberta Power Pool prices. Back-testing results show that at a high quantile level, VaR estimates from EVT are the most accurate among the above three methods. The estimation results of VaR can be applied in evaluating hedging call options as an insurance against extraordinarily high prices in a wholesale spot power market. |
