Application of stochastic dominance tests to utility resource planning under uncertainty

Abstract

Uncertainty and risk have become critical elements of electric-utility planning. Planning has become more complex because of uncertainties about future growth in electricity demand, cost and performance of new generating resources, performance of demand-side resources that increase end-use efficiencies, and overall structure and competition within electricity markets. Current methodologies for incorporating risk and uncertainty into planning are limited by their potential lack of consistency with economic theory, or the necessity to specify expected utility functions that are difficult to defend. In this paper, we describe these limitations and suggest a new approach to utility planning under uncertainty, called stochastic dominance, which can augment or, in some cases, replace existing methods to evaluate uncertainty. Stochastic dominance permits comparison of resource plans with uncertain performance in a theoretically consistent manner, without having to assess or justify individual expected utility functions. Stochastic dominance tests can also be implemented with little additional effort over current decision methodologies, especially those methodologies that rely on Monte-Carlo models as a planning tool. Stochastic dominance tests can incorporate alternative risk attitudes of decision makers more easily than existing empirical methods and therefore provide more robust solutions to decision making under uncertainty. The paper concludes with a demonstration of stochastic dominance applied to a resource-planning decision for a hypothetical utility.