Three essays on the application of conditional heteroskedasticity models to the deregulation of the wholesale electricity market

Deregulation in the wholesale electricity industry creates new circumstances which must be incorporated into the industry players = decision making process. Even if a general trend towards a competitive price exists, it is still possible that deregulation can generate price instability. Understanding the impacts of deregulation on the wholesale electricity market requires empirical tools that can estimate volatility and the interdependence of markets. Underestimating volatility or market dynamics could result in poor forecasts, the cornerstone of decision-making processes. Therefore, it may be appropriate to apply conditional heteroskedasticity models to the wholesale electricity market, particularly when estimating dynamic relationships between markets.;This dissertation is comprised of three interrelated papers that investigate wholesale spot market prices across the U.S. during the period from 1994 through 1998. The first paper considers the impact of deregulation on market prices by examining the level of market integration, which can partly assess whether the industry is becoming more competitive. Also, model structural stability is tested to provide insight into the empirical modeling needs of the industry. The second paper examines the forecasting abilities of the conditional heteroskedasticity models by comparing out-of-sample forecasts for linear and non-linear models. The third paper examines dynamic interactions between spot markets in the fore-mentioned time period.;Cumulative sums of squares suggest market stability is most often obtained with the use of nonlinear conditional heteroskedasticity models. The ability of nonlinear models to estimate the market is tested via out-of-sample forecasting. Out-of-sample forecasts confirm the applicability of generalized autoregressive conditional heteroskedasticity (GARCH) equations to the wholesale market because GARCH outperforms a vector autoregressive (VAR) model in nearly every case.;Impulse response functions (IRF) and forecast error variance decompositions (FEVD), elicited from a system of multivariate GARCH equations, show that all of the markets studied exhibit some degree of endogeneity. Endogeneity, or the linkage of market prices, suggests the potential of market integration and the dissemination of price information between markets. A directed acyclic graph, which provides causal ordering used in the orthogonalization of the innovations, confirms endogenous links between the markets. Time varying coefficients also suggest regional market integration, but very little integration between regions.