The post-mortems on last summer's price spikes in the Midwest are in. At least three studies of the event diverge in their conclusions:
First, on Sept. 24 of last year, the staff of the Federal Energy Regulatory Commission found the root cause of the spikes in extreme weather and unexpected outages. It observed no direct evidence of market manipulation and concluded that the events were unlikely to recur.
Reacting to those findings, Judah Rose of ICF Kaiser International saw deficiencies in the FERC report and predicted more spikes as a shortage of generating capacity worsens in the near future. (See "Missed Opportunity: What's Right and Wrong in the FERC Staff Report on the Midwest Price Spikes," Public Utilities Fortnightly, Nov. 15, 1998, p. 46.) To cope with it, he recommends operational changes and quick deregulation, the latter to end uncertainty that discourages new investments.
Four days later, on November 19, the Public Utilities Commission of Ohio reported to the legislature that last summer's weather and outages might well recur. Policy interventions to improve market coordination were needed, said the PUC, since today's "immature" market could "evolve rapidly only under optimal conditions."
Today, several months later, is there anything left to say?
Certainly it is true that bulk power markets in the Midwest are relatively underdeveloped. However, our data show that they functioned quite effectively under extreme conditions. A relatively undeveloped market adapted quickly and efficiently to events never before seen. Buyers and sellers reacted rationally to those events in the face of great uncertainty and unfamiliar limits on their abilities to transact. Market forces were at work over much of the Eastern Interconnection, both at a macro level and hour by hour over the critical days. Observed prices offer evidence of a unifying market that joined many regions within the Eastern Interconnection - before, during and after the spikes. Transaction data shows that the choices of market participants were surprisingly rational, despite unprecedented restrictions on transmission availability.
Yet, these restrictions on transmission deserve a second look. Here we attempt to isolate their causative role.
The transmission restrictions flowed from new protocols introduced to relieve line overloading. They arose last year in mid-June, a few weeks ahead of the price spikes, when the FERC took notice of a request by the North American Electric Reliability Council to impose procedures for transmission loading relief. These new TLR rules would relieve the impact of parallel flows on grid systems not located directly on the contract path of the curtailed transaction. Though it took the matter under advisement, the FERC encouraged utilities to apply the TLR rules within the Eastern Interconnection pending a final decision.fn1
Did market turmoil in the Midwest flow directly from transmission restrictions? With the data presented here we cannot conclude with certainty that TLR procedures caused the price spikes. However, even if they did not, we find that these restrictions may have contributed to inefficient use of resources over a wide area. These new TLR rules curtailed substantial amounts of power and foreclosed valuable market transactions. Major utilities declared emergencies just short of rolling blackouts. Through it all, however, the lights stayed on.
Before and During: The Runup in Prices
Early in 1998, the North American Electric Reliability Council's summer forecast noted that the margin of generation available to meet peak summer loads in the Midwest was becoming quite thin. NERC was particularly concerned about systems under the East Central Area Reliability Coordination Agreement (ECAR, covering Ohio, Indiana, Lower Michigan, and adjacent areas) and the Mid-America Interconnected Network (MAIN, covering most of Illinois, Wisconsin, and adjacent areas). Ontario Hydro's unexpected shutdown of several nuclear plants heightened the anxiety because parallel flows along its lines also could affect reliability.
Summer arrived early in the Midwest, with some generators out of service for scheduled maintenance in June, prior to planned peak use in July and August. Adding scheduled and unscheduled outages, American Electric Power entered the days of the spikes with more than 20 percent of its capacity out. The company estimated the probability of these outages at 1.5 percent, and that of such extreme heat at 0.3 percent. The rest of ECAR and MAIN suffered from more outages than usual, aggravated by the loss of important transmission due to storms.
On-peak power prices began to rise above $100 per megawatt-hour as the week of June 22 began, and they reached $250/MWh to $400/MWh for delivery on June 24. Around that date, Federal Energy Sales, a small marketer, was alleged to have defaulted on some call options it had written to deliver power at $50/MWh. Utilities and marketers that had counted on deliveries from FES (or others dealing with that company) found themselves thrown unexpectedly into the short-term market. That market became less accessible as storms and loading restrictions cut available transmission and transactions.
Prices rose to the thousands of dollars per megawatt-hour throughout the Midwest and the Southeast, where producers were exporting power northward. As the week of June 22 ended and more normal conditions prevailed, prices returned to their pre-spike levels, generally less than $50/MWh, throughout the entire Eastern Interconnection. July's spikes were a less extreme replay of June's, due again to record loads and unexpected outages but without defaults.
Was There a Market?
In regions like New England and California, bulk power markets have functioned and grown for decades. Even before California's restructuring, some of its utilities regularly obtained more than half of their power from independent producers and utilities beyond the state's borders. In the Midwest and Southeast, greater self-reliance by utilities has been the rule. Operating units of some large holding company systems trade substantial amounts of power among themselves, but most utilities in those areas obtain relatively little of their requirements from outsiders. Last summer, however, some midwestern utilities had to go to market for power to augment their production. Prices responded by climbing to four-digit levels.
Did a functioning market emerge in the Midwest?
A market is often defined by convergence of prices to a common value, net of costs of transportation. Convergence may be stymied when trading is too thin, because of the high cost of arranging transactions or obtaining information about alternatives. Economists have not yet agreed on a numerical criterion for concluding that two regions lie within the same market, particularly when each is affected by a similar set of external forces. When analyzing the price spikes, a graph of daily power prices in different regions would be dominated by a handful of observations of four-digit prices, obscuring inter-regional linkages that might have occurred during times when prices were at more normal levels.
It is easier to see the similarity of price movements among regions by plotting their natural logarithms, as we do in figure 1 for daily index prices between June 1 and Aug. 7, as published by Power Markets Week. Logarithms facilitate comparison of proportions, since a 30-percent range on a spike day is the same vertical distance as a 30-percent range on a normal day. The individual blue lines show deliveries of on-peak power into various Midwestern and Southeastern regions and utility territories. They include ECAR, MAIN, the Southeastern Electric Reliability Council (SERC), the Florida-Georgia border (FLA/GA), Tennessee Valley Authority (TVA), Southwest Power Pool (SPP), and Entergy Corporation. They move together so closely that individual labeling is virtually impossible. The percentage range of prices among the regions differs little between spike days and more ordinary ones.
Negotiated prices for bulk power and substantial available transmission capacity link the Midwest and Southeast into a single trading area. Prices in nearby regions that transact only small volumes with the Midwest and South behave quite differently, as shown by the white lines in figure 1. Those regions include (1) the Mid-Continent Area Power Pool (MAPP) in the upper Midwest, which has little transmission capacity into ECAR and MAIN; (2) the Electricity Reliability Council of Texas (ERCOT), an electrically isolated area that can only export small amounts of power; and (3) the Pennsylvania-New Jersey-Maryland Interconnection (PJM), a centrally dispatched pool into which utilities must offer power at cost-based prices; and (4) the New England Power Pool, at a greater distance to the Northeast.
Hour by Hour:
Spikes Thinly Traded
The demand for electrical energy fluctuates predictably during a normal day, reaching its peak(s) in the late afternoon or evening, and its bottom in the predawn hours. In the absence of such constraints as minimum load conditions, generators with high avoidable costs only will be efficient to dispatch during hours of high demand. If production is determined by market bids, high-cost plants will offer their power only during peak hours, when the market-clearing price is high enough to cover their incremental costs. A utility that generates most of its own power will usually face market prices for supplementary or economy purchases that also vary predictably during the day. If prices do not follow load in this way, the market will be operating inefficiently, producing its hourly requirements at higher cost than necessary.
During the June spike days, the variation of hourly prices was consistent with economic efficiency in generation. Power Markets Week reported off-peak power prices in ECAR in the $12/MWh to $14/MWh range throughout the week of June 22, the same as in the preceding and following weeks.
Figure 2 shows the results of a confidential survey of Midwestern marketers by Tabors Caramanis and Associates (TCA) of Cambridge, Mass.fn2 It shows hourly variation in the highest and lowest reported prices per MWh of short-term power over June 24 and 25. Both high and low prices peak in the midday hours and bottom late at night, consistent with (but certainly not proving) efficiency in the allocation of generation. The gap between high and low reported prices is substantial. It may reflect both the difficulties of shopping in this type of market and the uncertainty that comes with an unprecedented situation. Utilities must make commitments to meet demand in upcoming hours, and currently available transmission paths may become unavailable if not claimed quickly.
Power purchased at spike prices was a small fraction of total power use during the critical days. To estimate the amounts transacted, TCA combined volume data from its survey with reports filed by marketers at FERC for June 24 and 25. Of 92 Midwestern transactions that took place at prices exceeding $100/MWh, 61 were for volumes of 200 or fewer MWh, mostly to flow for two hours or less. These figures are not exhaustive (reports to FERC are sometimes incomplete), but if provisionally accepted they show that high-price power flowed at an average of 695 megawatts per hour over the Midwest on June 24 and 25.
Since peak 1997 summer demand in the region was 168,000 MW, unless the data are wildly incomplete they show how unimportant spike-priced power actually was in the overall picture. That power went to satisfy the last increments of demand and was purchased in the most extreme of market conditions. With most utility-owned generation committed to serve at regulated retail rates, spike prices had a minor impact on overall power costs and the bills of most customers.
Transmission: TLR Rules Significant
Prior to last summer, a utility experiencing line-loading problems generally redispatched its system to meet the problem as economically as possible. A utility undertaking redispatch could adversely affect nearby systems as a byproduct of its operations by increasing their line loadings. Last summer, FERC permitted utilities in the Eastern Interconnection to use new TLR procedures proposed by NERC, whose members are largely transmission-owning utilities. The new procedures deal with overloading on a region-wide basis, and can require that transactions far from the affected area be curtailed on the basis of non-economic criteria.
On June 24 and 25 three separate invocations of TLR drastically curtailed power flows over the East and Southeast. A June 25 overload of 30 MW between Minnesota and Wisconsin ultimately led to curtailment of at least 1,900 MW of transactions throughout the Midwest. Generation outages had clearly strained the transmission system, but curtailments of this volume and scope were unprecedented.
Figure 3 plots aggregate weekly volume data (daily data are unavailable) against a weighted weekly index of prices over the Eastern Interconnection. At a time of heavy demand due to weather and generation outages, we would expect the volume of transactions to rise along with prices. In reality, during the weeks of the price spikes, power flows fell relative to surrounding periods. The week of June 22 had more hours with TLR curtailments in effect than any other of the summer, but the week of July 20 had fewer such hours than the summer's average. We cannot with certainty conclude that TLR rules caused the spikes, but even if not, they may have contributed to inefficient use of resources over a wide area.
An administrative decision in PJM also contributed to the Midwestern price spikes. In June 1998, the pool announced that 2,500 to 4,000 MW of power exports could be subject to curtailment. This meant the power had to stay in PJM, instead of being exported to the Midwest, even if users in the Midwest valued the power more highly than users in PJM. The system operator's action kept more low-cost power in PJM, but it also reduced the amount of power exported to the Midwest. As with most non-price rationing schemes, the costs imposed on those who could not bid for the power in all likelihood exceeded the benefits conferred on the lucky buyers who got it.
Distorted By Regulation
We agree with the FERC staff report that the spikes were odd phenomena in a relatively undeveloped market and, like FERC staff, we found no evidence that they resulted from the exercise of market power by owners of generation or transmission. Prices accurately reflected underlying supply and demand conditions. However, those conditions had been distorted by regulation and industry practices that rendered the Midwest and South more likely candidates for extreme price movements than other regions.
As discussed, utilities in both regions generate most of their electricity in large plants and rely less on the market for power supplies than utilities in the Northeast and Southwest. The default of a Northwestern marketer in late June had a smaller, shorter-lived impact than the one in the Midwest, in part because the Western market was more extensive, with more sources of power and transmission for those caught short.
Michigan is the Midwest's leader in independent power, ranking tenth among the states but with only 21 percent of California's capacity. Excluding recent utility divestitures, Ohio, Indiana and Illinois combined contain less than 1,000 MW of independent capacity, a total that is 20 percent of Michigan's. California, on the other hand, recently gave non-utility producers under contract to utilities expanded rights to market their surplus power. This move was intended to ease some bottlenecks that have raised the prices of ancillary services. A small number of outages in the Midwest will thus have a greater impact, particularly if imports are constrained.
On the demand side, Midwestern utilities typically serve at rates that insulate a high proportion of their customers from market price movements, and utilities have fewer opportunities to curtail loads or call on industrial generation to meet capacity shortfalls. California's flexibility exceeds the Midwest's on all these dimensions. All California customers with demands of more than 1 MW must take time-of-use rates, and their 13 percent of load is only part of the state's total on market-sensitive rates. Interruptible load and callable generation account for 6 percent of California's peak. Illinois, by contrast, first considered offering payments for callable generation and compensation for interruption in response to the 1998 spikes.
The Midwestern price spikes would have been both less severe and less likely if markets were fully competitive and power could flow unimpeded to buyers who value it most highly. For this reason, we believe the following policies could reduce the risk or severity of price spikes in the future:
n Don't shoot the messenger. Price controls in the bulk power market "prevent" price spikes the same way general wage and price controls "fight" inflation(by creating shortages. Controls are a recipe for blackouts and inefficiency, not a solution to price spikes.
n Move on retail competition. Full retail competition would allow all customers to choose their degree of exposure to market prices, and innovative competitors could offer price breaks to customers who agree to limit their usage when prices peak.
n Don't hamper risk management. Financial instruments reallocate risk to parties most willing to bear it. They allow power buyers to shield themselves from price spikes, and encourage investment by allowing power producers to shield themselves from precipitous drops. Regulation should not stand in the way of mutually beneficial reallocation of risks.
n Let prices allocate transmission and generation. Midwestern prices spiked in part because administrative decisions rather than market processes allocated access to critical resources.
n Allow multiple trading options. Bulk power trades in the Midwest take place through bilateral deals. A mandatory, centralized bulk power market would force market participants to trade only a few types of standardized electricity contracts, preventing them from making deals tailored to their individual preferences and constraints. If centralized power exchanges are created in the Midwest, participation should be voluntary.
Robert J. Michaels is professor of economics at California State University, Fullerton, and Jerry Ellig is senior research fellow at the Mercatus Center at George Mason University, Fairfax, Va. This article summarizes a lengthier research report available from the Mercatus Center. Funding was provided by Americans for Affordable Electricity, Electric Power Supply Association, Electricity Consumers Resource Council and Enron Corp.
1 The FERC has formally approved the TLR rules, ordering utilities to amend their pro forma open-access tariffs. Docket No. EL98-52-000, ER98-3709-000, Dec. 16, 1998, 85 FERC ¶61,353.
2 The survey was undertaken for Enron Corp. as part of its filing in FERC Docket EL98-52, investigating the price spikes.
Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.