Assessing the Effectiveness of Competition 
in the Electricity Generation Industry

ey to the restructuring of the US electricity industry is allowing generators of  electricity to compete for residential, commercial, and industrial customers. But in certain  regions or at certain times, might the number of electricity providers be so small that  competition is limited, enabling a few providers to exercise "market power," raising their  prices above their costs without losing sales? One means of examining that question is to consider the geographic area over which providers can compete. A broad geographic market suggests more competitors and less opportunity for market power. To explore the extent of the geographic market for electricity-generation services, an Energy Laboratory researcher studied trading between providers in the already-operating "wholesale" electricity market, where utilities sell electricity to one another at competitive prices and transmit it using today's transmission system. Using new econometric techniques, the researcher analyzed data on daily wholesale electricity prices to determine the extent to which utilities in the five subregions of the western United States traded with one another during an 18-month period. She found that fully 80% of the time the wholesale market included potential competitors throughout the entire western area. The rest of the time the market was considerably smaller, largely due to congestion on the transmission system. While these findings cannot be directly applied to the emerging retail market, they do suggest conditions under which the number of effective competitors may dwindle and market power may be an issue.

Examining whether the emerging retail market for power generation will be geographically broad, thus including many buyers and sellers, and the conditions under which it may narrow has been the focus of a study by graduate student Elizabeth M. Bailey, under the supervision of her faculty advisor, Paul L. Joskow. Because restructured retail markets are just beginning to operate, no historical data are available for analysis. Ms. Bailey looked instead at the long-running wholesale market--the market in which utilities trade electricity with one another. In making wholesale deals, utilities negotiate prices, quantities, and other terms of trade with one another and then transmit the electricity over the transmission lines connecting them. Thus, the existing wholesale market has many features expected to characterize the new retail market. Whether the wholesale market is geographically broad may provide insight into whether the retail market will be sufficiently broad for competition to be effective.

To perform her analysis, Ms. Bailey focused on a wholesale market well suited to this study--the wholesale electric power market in the western United States. The western region has five natural subdivisions: the Pacific northwest, northern California, southern California, the inland southwest, and the central Rockies. Each subregion has distinctive generating and demand characteristics, for example, the primary fuels used and the season in which demand peaks. Many large transmission lines interconnect utilities in the west, so it is possible for the entire region to behave as a single wholesale market. In addition, the market is fairly isolated from activities in the east because only small transmission lines cross the Rocky Mountains. While buying and selling electricity is common among the five subregions, at times transmission problems or other factors can limit trading opportunities between various subregions. Under those circumstances, subregions become isolated markets including fewer utilities, thereby increasing the likelihood that market power can be exercised.

The extent to which subregions were able to trade with one another could be determined from historical data on transfers of electricity from one subregion to another. However, such data are not publicly available. Ms. Bailey therefore used econometric techniques to infer the extent to which subregions could trade from published daily prices for wholesale electricity within each of the western subregions.

To understand how price data can be used to infer the potential for trading, assume that the cost of generating a kilowatt-hour (kWh) of electricity is higher in one subregion than in another. (Cost and price are roughly equivalent under the conditions of this study.) The high-cost subregion will pay less by buying a kWh of electricity from the low-cost subregion than by generating it itself--if the price of transmitting the electricity is less than the difference between the generating costs in the two subregions. As the low-cost subregion starts generating more electricity to sell, it will begin running its more expensive generating units, and its cost per kWh will go up. At some point, that cost plus the price of transmission will exceed the cost of generation in the high-cost subregion. The high-cost subregion will use its own generators to fulfill the balance of its demand. Thus, when the market settles, the costs of generating a kWh of electricity in the two subregions will differ by the price of transmission between them--as long as those subregions are able to trade.

In her analyses, Ms. Bailey used daily price data provided by Economic Insights, Inc., a  firm that continuously surveys electricity market transactions in the western United States. The data cover June 1995 to December 1996 and include separate price observations for "peak" and "off-peak" periods. (Peak periods include daytime weekday hours, when demand tends to be high. Off-peak periods are nighttime hours and all hours on weekends and national holidays.) The study drew on a total of 948 daily price observations for each subregion as well as information on levels of supply and demand, the status of the transmission system, and other supply and demand conditions that might affect trading.

Ms. Bailey began her analysis by determining whether prices in pairs of subregions differed by a constant amount over time. Again, if two subregions are able to trade without constraint, their prices will differ by the price of transmission. Therefore, if price goes up in one of those subregions, price will also go up in the other. If subregions cannot trade, a change in price in one subregion will not affect price in the other.

As the figure below shows, wholesale electricity prices in all subregions generally move together, following the same seasonal trends. But a closer look reveals times when prices go up in some subregions and not in others. To quantify those differences over time, Ms. Bailey determined "price correlations" for each pair of subregions. Price correlations assess the
extent to which variations in the value of one price are associated with variations in the value of the other.

The analysis showed high price correlations under normal, uncongested conditions. Lower price correlations occurred during periods when transmission lines were full due to high demand from one subregion and when lines were downed by storms or other incidents or "derated" to carry less than their normal capacity due to damage or maintenance. Those findings confirm that the condition of the transmission system and levels of demand can constrain trading between subregions, narrowing the geographic expanse of the market and thus reducing the number of potential competitors.

But price correlations do not provide insight into how frequently congestion occurs and trading is constrained--critical information because the more often subregions can trade without constraint, the broader the market and the less likely market power can be exercised. Ms. Bailey inferred how frequently trading was constrained or unconstrained by examining how frequently the difference between prices in pairs of subregions was equal to, greater than, and less than the prevailing price of transmission. When the price difference equals the transmission price, the two subregions are able to trade fully with one another. When the prices in the subregions differ by more than the price of transmission, both subregions could benefit economically from trading; therefore, constraints on the transmission system must be preventing the transfer of electricity from the low-priced to the high-priced subregion. And when the prices differ by less than the price of transmission, the high-priced subregion is not interested in trading because it would have to pay more to buy and transmit the needed electricity than to generate the electricity itself.

Because data on transmission prices are not available, Ms. Bailey developed an empirical technique for estimating the implicit prices for transmission services between pairs of subregions. She focused on price pairs for two subregions, southern California and the northwest. She performed separate analyses for peak and off-peak observations and used her new technique to sort the price pairs according to whether the difference between the two prices was equal to, greater than, or less than the price of transmission.

The analysis suggests that trading between southern California and the northwest was, by and large, unconstrained during the 18-month study period, even during times of peak demand. Fully 80% of the price pairs for the peak period differ by roughly the price of transmission, indicating that trading was not restricted by transmission constraints. Thus, on most days during the period studied, the wholesale electricity market extended across both subregions, and utilities could transact with other utilities dispersed over the entire geographic area. In about 19% of the price pairs, the difference between the prices is greater than the price of transmitting the electricity, indicating that trading was constrained. The geographic size of the market narrowed, and the effective number of competing firms fell. In just 1% of the price pairs, the difference between prices in the two subregions was less than the price of transmission, showing that trading was not economically advantageous.

Ms. Bailey cautions against transferring those findings directly to the competitive retail market that is now developing. After all, the structure of the wholesale market does differ from that of the restructured retail electricity market envisioned. In particular, the current wholesale electricity market is overlaid on a closely regulated retail market. Ms. Bailey also warns that the geographic size of a market may imply nothing about the extent to which competition prevails. Even a geographically small market may contain enough generating companies for competition to thrive.

Nevertheless, the study does suggest that markets for buying and selling electricity can be geographically broad. Moreover, the study identifies particular conditions under which the market for generation services may narrow--conditions that may warrant public policy actions to prevent generating companies from raising their prices substantially above their costs. Perhaps most important, as price data become available from the new retail market, investigators can use the analytical techniques developed in this research to assess whether a broad geographic market for retail electricity has emerged.

Elizabeth M. Bailey is a PhD candidate in the Department of Economics. Paul L. Joskow is the Elizabeth and James Killian Professor of Economics and Management and head of the Department of Economics. This research was sponsored by the National Science Foundation and MIT's Center for Energy and Environmental Policy Research. Further information can be found in references.