Testing the performance of electricity markets using POWERWEB has already shown that relatively inexperienced players can identify and exploit market power in load pockets. When transmission constraints are not binding, however, auctions with six players have been shown to be efficient. There is evidence from operating electricity markets that prices can be driven above competitive levels when the largest supplier controls less than 20% of total installed capacity. This is accomplished by causing price spikes to occur. In experiments, uncertainty about the actual load and paying standby costs regardless of whether or not a unit is actually dispatched contribute to volatile price behavior. The objective of this paper is to investigate characteristics of a market that affect price volatility. The tests consider three different sets of rules for setting price when there are capacity shortfalls, and the following four market structures: load is responsive to price; price forecasts are made before market settlement; a day-ahead market and a balancing market auction; and suppliers are paid actual offers (a discriminatory auction).

The high average prices and high volatility of prices in many restructured markets for electricity have raised concerns about the abuse of market power by generators. At the same time, information about the true costs of generation, that was readily available under regulation, is no longer disclosed by generators. Hence, it is becoming impractical to use a comparison of actual prices with competitive prices as the basis for identifying the use of market power. In this paper, an engineering procedure is proposed for a given pattern of dispatch to measure the potential for market power for all generators in a network. This procedure is equivalent to a set of factor demand equations in a standard neoclassical model of production. An optimal dispatch, for given sets of offers to sell and constraints on capacity, can be replicated exactly by resolving the dispatch using the optimal nodal prices as offers with no constraints on capacity. Market power exists when the degree of substitutability for power generated at a particular site is low. Withholding capacity and/or raising offers to sell at such a site would be one of the possible ways to exploit market power. Sensitivity of the results for any given pattern of dispatch can give some indication of how effectively prices have been raised by a generator. Under competitive conditions, submitting higher offers to sell at a site will result in a substantial reduction of the capacity dispatched, and the own-price elasticity of demand for generation at the site is very large (negative). As market power increases, the own-price elasticity gets smaller and approaches zero. These effects can be aggregated easily to deal with a specified group of generators, which could represent generators in a load pocket or ownership by a parent company. Use of these procedures is illustrated by evaluating the results of two experiments designed to test market power. These experiments were conducted with undergraduates and utility executives using POWERWEB, which simulates a full AC network with 30 buses, six of which are generators. The objective of the experiments was to determine whether two of the generators could discover that they were in a load pocket and raise their profits by exploiting market power. The authors are: Postdoctoral Associate, School of Electrical and Computer Engineering (ECE); Graduate Research Assistant, Department of Applied Economics and Management (AEM); Professor, (AEM); Professor, (ECE); and Senior Research Associate, (AEM) at Cornell University. For correspondence, use tdm2@cornell.edu (Tim Mount) or Applied Economics and Management, 214 Warren Hall, Cornell University, Ithaca, NY 14853-7801. This project was part of the research program of the Power Systems Engineering Research Center (PSERC) sponsored by the National Science Foundation. Additional support came from the Consortium for Electric Reliability Technology Solutions (CERTS) program in the U.S. Department of Energy and a project on complex interactive networks supported by the U.S. Department of Defence and the Electric Power Research Institute (EPRI).

Given the load profile of an electricity market and the capabilities of the set of generators supplying power to that market, it is likely that at any given point in time, available supply will exceed demand. If only a subset of generators is required, some method is required to commit and de-commit generators. In the past, system operators have employed a centralized method of unit commitment. Deregulation of the electricity industry throws doubt on the continued suitability of this method due to fairness issues and availability of accurate cost data. This paper examines the performance of decentralized unit commitment, where dispatch of generators is determined by offer curves submitted into a spot market by power producers.