Buyer's Remorse


The PJM complaint and the rising cost of electric reliability.

Fortnightly Magazine - September 2008

Who says ratepayers must accept the traditional measure of electric reliability—a single one-hour outage every ten years? If shown the bill ahead of time, might they decide otherwise; that such luxury is no longer affordable? Consumers are making similar decisions about gasoline and mortgages. Why not electricity?

Notice that Chairman Joseph Kelliher and his fellow commissioners at the FERC already are busy preparing consumers for this sobering comeuppance—that we as a nation can no longer afford our grandfathers’ grid. Listen to Commissioner Suedeen Kelly, commenting at FERC’s June 19th meeting on an eye-opening presentation from the commission staff (see Figure 1) on the rising cost of electric-generation facilities.

“We have a weak dollar,” warned Kelly. “We have an economic slowdown … We have a financial sector that has been battered by financial crises, and credit is tight.

“This,” she added, “makes the whole notion of building more electricity infrastructure a difficult one.”

And Kelliher added, “I think it is important that these hard realities be better understood by the general public and others.”

Such concerns are emerging in other regulatory proceedings. Exhibit A is the recent complaint against the PJM power grid, which was filed at FERC at the start of summer to dispute results of the first four base residual auctions conducted under the region’s new and controversial “reliability pricing model” (RPM). (See, Maryland PSC v. PJM, FERC Docket EL08-67, filed May 30, 2008.)

The RPM regime, analogous to the ICAP program in New York and New England’s forward capacity market (FCM), serves to recruit electric-generating capacity three years ahead of time, in order to keep the lights on. RPM includes market auctions built on complex mathematical algorithms aimed specifically at complying with the one-in-10 reliability standard. This new complaint, filed by a group of self-styled capacity “buyers”—including state utility commissions in Maryland, Delaware, New Jersey and Pennsylvania, rural cooperatives, and state-funded ratepayer advocates, plus a dozen or so additional parties, with various federal government agencies thrown in—marks a new turning point in the unsteady march of utility restructuring.

Now, for perhaps the first time, we see opponents confronting a market regime not because electricity is special and so cannot abide a market, or that that deregulation violates the laws of physics or the duty to serve. Rather, this new complaint attacks PJM’s capacity market because, in essence, it mandates too much reliability—a level of security that American ratepayers simply cannot afford.

A $26 Billion Mistake?

Put in place by a settlement agreement OK’d by FERC on December 22, 2006 (Docket ER05-1410, 117 FERC ¶61,331), PJM’s RPM regime features regular auctions in which suppliers offer for a given price to sell future availability of electric power resources to the RTO (be it plant output, demand response, or even space on a new transmission line), for physical delivery across a 12-month commitment period that does not ordinarily begin until three years later. In May 2008, for example, PJM conducted its fifth base residual auction (BRA), for the 2011-12 delivery year (June 1 to May 31). The three-year forward period echoes a similar design feature in the ISO-NE FCM plan. It’s designed to accommodate a typical plant-construction interval, so developers can bid to supply new resources, and then go out and build them in time for the assigned delivery year. Capacity cleared in PJM’s May 2008 BRA for delivery in June 2011 is set at $110 per megawatt-day—under the pricing metric that PJM uses. That’s equivalent to $40.15 per kilowatt-year, or $3.34 per kilowatt-month, as capacity prices are sometimes expressed in other regions.

The first four auctions, however, did not proceed in the same manner. PJM had designed its RPM regime to avert an impending crisis of possible reliability violations in New Jersey and Baltimore-Washington zones [Eastern and Southwest Mid-Atlantic Area Council (MAAC) respectively]. To get off the ball quickly, and foster familiarity among market players as soon as possible, PJM began with four so-called transitional auctions, each targeting a future twelve-month delivery period, but with forward intervals shorter than three years:

• BRA-1 (April 2007), for delivery 2007-08;
• BRA-2 (July 2007), for delivery 2008-09;
• BRA-3 (Oct. 2007), for delivery 2009-10; and
• BRA-4 (Jan. 2008), for delivery 2010-11.

This transitional wrinkle explains a key allegation in the recent complaint. The Maryland Public Service Commission and its compatriots charge that because the first four transitional BRAs featured such short forward intervals, there was no chance for new entry; i.e., no chance for developers to bid on supplying newly constructed projects. Instead, say the complainants, without competition from new projects, the four transitional auctions cleared at exorbitant price levels that produced windfalls for capacity resources already built.

All told, say the complainants, this unfortunate beginning to RPM will force PJM-region ratepayers to pay a hugely excessive bill—some $26 billion—to cover adequacy of electric-resource capacity for the delivery period June 2008 through end of May 2011, without really getting much new plant construction for the money.

Of course, PJM knew full well that implementing short forward intervals for the first few auctions was less than ideal, but stressed in the cover letter attached to its August, 2005 RPM application that the early start would “help build confidence in the new capacity market, discourage retirements or mothballing of plants that may be needed, and provide valuable experience in market behavior.”

The July auction, for example, yielded 1,300 MW of new resources, including 536 MW of demand response, plus 2,300 MW of generation added back to the market through the cancellation of planned retirements, or restarting of closed plants. Commenting on the auction, PJM’s vice-president of markets, Andrew Ott reported: “We had new generation enter the auction largely through upgrades to existing units to produce more power. These are the results we intended to see from RPM.”

The buyers also complain of many additional distorting factors they believe have produced high prices that will harm retail ratepayers when it comes time for utilities to pay the bill for all the capacity resources that cleared in the four transitional auctions:

Monopolistic Structure. Concentration of plant ownership in a few hands, allowing plant owners with large portfolios to withhold some resources from bidding to push up prices for other assets sold into the auction. Also, too steep a vertical slope for RPM’s administratively determined demand curve (the variable resource recovery curve, known as VRR), that helped to reward large-scale suppliers for withholding bid offers.

Suspicious Bidding. Bids in certain zones (Eastern & SW MAAC) that fluctuated widely from auction to auction, casting doubts on effectiveness of price mitigation by PJM’s market-monitoring unit (MMU), which purportedly capped all offers at avoidable cost.

Manipulation of Outage Rates. Faulting PJM’s special rule known as the EFFORd offer segment (equivalent forced outage rate-demand), which allows a supply bidder to hedge the risk of higher outage rates in the forward delivery year (increasing the must-off obligation) than in the year prior to auction.

Faulty Offer Caps. Inordinate emphasis on avoided costs of gas-fired, simple combustion turbines as proxy to calculate hypothetical CONE value (cost of new entry). Resources clearing the auctions included incumbent coal-fired plants and demand response. Also, use of outdated figures for the E&AS revenue offset in calculating net CONE for building a new simple gas turbine, net of yearly revenues for sales of energy and ancillary services. Plus, allowing suppliers to boost avoided costs by adding in long-run costs of certain capital expansions under RPM’s special APIR rule (avoidable project investment recovery rate).

Queue Delays. Slow turnaround in processing applications to connect new projects with the grid (as shown by the complaint filed by Dominion, FERC Docket EL08-36, filed Jan. 28, 2008), tending to boost auction prices by limiting the pool of new resources.

Restrictions on Self-Suppliers. Lack of flexibility in RPM rule that limits MW-capacity that can be bid into the RPM by utilities (such as American Electric Power) choosing to supply their own capacity requirement (or purchase bilaterally) instead of buying through RPM auctions, under the FRR option (fixed reserve requirement).

Excessive Reserve Margins. Unwarranted increase in target IRM to 116.5 percent (installed reserve margin, where price along the VRR curve equals 1.5 x Net CONE), in order to achieve a one-in-ten reliability. Also, unjustified goal of one-in-25 reliability standard for LDAs (local delivery areas), representing RPM auction pricing subzones within PJM, such as EMAAC or SWMAAC.

On the strength of these allegations, the buyers ask FERC for refunds reflecting a hypothetical do-over of the transitional auctions. As the buyers see it, an honest price for electric capacity in today’s world ought not to run any higher than $100/MW-day, which represents a rough approximation of the average capacity clearing price for the “rest of RTO” zone in BRA nos. 1-4. Starting from there, the buyers would set just and reasonable capacity prices for the four transitional auctions based on a graphic “straight-line interpolation,” beginning with $40.80 for 2007-08, and ending with $100/MW-day for 2011-12. (See Table 1, RPM Auctions; Selected Zones.)

Does all this really make for a $26 billion mistake? Opponents say the buyers are asking FERC to pull prices out of thin air. They point out, for example, that stakeholders designing New England’s FCM set the initial price floor at $4.50/kW-month ($148/MW-day). And FERC just recently allowed the New York ISO to boost the CONE value in its ICAP market by a 7.8 percent escalation factor for delivery years 2008-09 through 2010-11, reflecting a 5.1 percent rise in plant construction costs (Handy-Whitman Index) plus 2.7 percent for general inflation. (See, Docket ER08-283, Jan. 29, 2008, 122 FERC ¶61,064.)

In its 2007 State of the Market Report, PJM’s Market Monitoring Unit (MMU) reported that for the nine-year period immediately preceding RPM’s launch (1999-2007), average yearly net revenues from economic dispatch for power plants serving PJM fell way short of covering 20-year levelized fixed costs for power plants of all types: including not only simple-cycle gas turbines (CT) but also combined-cycle turbines (CC) and pulverized coal fossil units (PC):

• CT: $32,200 vs. $75,200;
• CC: $61,000 vs. $99,700; and
• PC: $165,000 vs. $231,700.

Robert Stoddard, who heads the energy and environment practice at CRA International (formerly Charles River Associates) followed up on that fact in testifying for the PJM Power Providers Group in comments filed opposing the complaint. Stoddard defends the $26 billion invoice as a pretty reasonable ballpark figure for the cost of resource adequacy in PJM. (See, Answer of PJM Power Providers Group, Attachment A, Affidavit of Robert B. Stoddard, FERC Docket EL08-67, filed July 10, 2008.) He writes:

“First, using a very conservative replacement cost of $800/kW, the value of the PJM capacity resources exceeds $100 billion; annual depreciation alone likely exceeds $4 billion; this figure includes no allowance for any return on equity or payment of debt costs.

“In this context, capacity charges of $4.3 billion for the 2007-08 delivery year, rising to $8.4 billion for the 2010-11 delivery year, appear economically reasonable as return on invested capital for the fleet.”

Money Already Spent

The smart money would expect FERC to deny the buyers’ complaint. As seen from industry comments filed at FERC through the end of July, the overwhelming sentiment sees the complaint as an unlawful collateral attack on PJM’s RPM market, since many of the complainants actually signed off on the final settlement approving the rates. The complaint appears also to violate the filed-rate doctrine; opponents cannot sign on to RPM and then later repudiate its auction-derived prices as excessive, since, in effect, it is the algorithm itself that stands as the rate—not the dollar prices that flow from the auction.

Many argue as well that the rule against retroactive ratemaking also bars the action, since the buyers filed their complaint after the four transitional auctions were completed. Legal precedent holds that the liability to pay the capacity prices became fixed when PJM established the auction parameters, even though the auction’s capacity buyers need not tender payment until the delivery year rolls around. In fact, FERC already has ruled that Duquesne Light could not escape liability for any RPM auction charges billed by PJM after May 31, 2008, the effective date for Duquesne’s requested withdrawal from PJM. (See, Docket ER08-194, Order issued Jan. 17, 2008, 122 FERC ¶61,039.)

Note also that Duquesne’s withdrawal of retail load from the fifth BRA auction, conducted in May, without at the same time withdrawing its supply offers, led arguably to lower-than-expected clearing prices for the 2011-12 delivery period. (See Table 1 and Figure 2.)

FERC on many occasions has made clear its reluctance to re-do market results.

Recently, for example, a surprisingly heavy influx of demand-side resources caused New England’s FCM auction to clear capacity at the minimum floor price, which triggered FCM rules requiring an automatic downward administrative adjustment to the benchmark CONE value, even as anecdotal and empirical evidence showed that gen-plant development and construction costs were rising. Market players in New England challenged this perverse result, required under the tariff, arguing that they “simply did not foresee the … precipitous decline in CONE that would result if a substantial number of new capacity resources elected to be price-takers,” as occurred in New England’s first FCM auction. Nevertheless, FERC held firm; it refused to block the automatic reset of CONE value, no matter how counter-intuitive, since it did not consider the auction results to constitute “new evidence” or “changed circumstances” that should warrant a market re-do. (See Docket ER08-633, order issued June 20, 2008, 123 FERC ¶61,290.)

Practical reasons stand in the way as well. States within the PJM footprint already have solicited energy supplies for retail consumers in reliance on the auction prices. As Constellation points out, both the New Jersey BGS auction (basic generation service) and the Maryland RFP for SOS energy (standard-offer service) tend to combine energy, capacity and ancillary services into a single bundled price, making it impossible to refigure retail costs if PJM should recalculate capacity costs at wholesale.

Investors, too have made commitments. For instance, it now appears lack of certainty of RPM auction revenues has prompted Sempra subsidiary Catoctin Power LLC to table plans to build a 600-MW plant in Frederick County, Md. Catoctin had obtained a certificate of convenience from the Maryland PSC in April 2005, but started dragging its feet recently after FERC decided this past April to deny PJM’s bid to boost the benchmark CONE value reflected in the RPM VRR curve. Catoctin now appears willing to allow certain state commitments on environmental permits to lapse. (See, Maryland PSC Case No. 8997, Order No. 82113, issued July 8, 2008.)

Crunching the Numbers

One particular accusation deserves a special mention. In a complaint, and in an affidavit and white paper prepared for the American Public Power Association and filed at FERC by James F. Wilson, an LECG consultant (see “Raising the Stakes on Capacity Incentives: PJM’s Reliability Pricing Model,” March 14, 2008, available at, the buyers allege that wide fluctuations in generation-supply bids offered in the SWMAAC zone between the first and second transitional auctions, for delivery in 2007-08 and 2008-09, versus the third auction, for 2009-10, either must prove that sellers wielded market power by withholding capacity and manipulating prices, or that PJM’s market-mitigating offer cap, purportedly set at avoidable cost, was not working. Nevertheless, it appears that PJM’s defenders have come up with explanations that dispel these claims.

As the complaint alleges, suppliers bidding in SWMAAC BRA 1&2 offered only 300 MW at prices exceeding $125, but in BRA 3 offered some 3,600 MW at $125 or higher, with a third of such bids coming in at prices above $225, which helped push the clearing price to $237.33/MW-day, along with the fact that offered capacity actually declined by 750 MW in SWMAAC in BRA 3, a fact the buyers ascribe to “deratings of existing units, rising outage rates at existing units, and only 32 MW of new capacity.”

The buyers state further that in BRA 4 (for delivery 2010-11) the same capacity cleared “with offers at or below the RTO-wide clearing price of $174.29. The buyers then argue that if the PJM MMU had capped offers at avoidable cost, as it was required to do, then “those same generators could not have made their much lower offers for 2010-11 without risking significant losses.” The buyers claim that generator manipulation of outage rates (affecting availability factors and the must-offer quantity), plus artificial inflation of avoided costs through an unwarranted inclusion of long-term capital investments, also played a role in distorting bids and clearing prices.

These allegations bear some explaining.

The RPM rules provide a market screen that requires mitigation by capping bid offers at avoidable cost under several different scenarios, such as if any three suppliers taken together are pivotal, meaning that their resources are needed to clear the market. And, since this three-pivotal-supplier (3PS) test was failed in each auction and in each zone, RPM mitigation rules kicked into to cap all offers at avoidable costs—defined in the tariff as “the costs that a generation owner would not incur of the unit did not operate for one year.”

(Note, by the way, that FERC recently opened an investigation to consider possible alternatives to the 3PS test, after finding that it can be triggered “too frequently” to be just and reasonable. See Docket EL08-34, Order issued May 16, 2008, 123 FERC ¶61,169.)

Nevertheless, these claims appear to wither under the testimony of Jonathan Lesser (an economist with Bates White, LLC), submitted on behalf of Constellation. Lesser points out that the apparent deratings and capital infusions followed directly from passage of the Maryland Healthy Air Act. As Lesser explains, the HAA is a “do-or-die” law that does not permit covered facilities to mitigate emissions through a trading of allowances, but that will require actual physical reductions of mercury, SO2 and NOx, effective Jan. 1, 2010. Thus, as Lesser reports, Constellation and Mirant (each owning 3 plants covered by HAA) have announced spending programs of about $1 billion each to achieve compliance.

RPM rules allow gen-plant owners to include capital expenditures in “avoidable costs,” on the theory that plants would have to be mothballed without such investments. Moreover, such additions appear to justify raising the offer cap, and also a derating of capacity output, that would affect the must-offer quantity.

As Lesser notes, installations of FGD equipment (flue-gas desulphurization) using LSFO technology (limestone forced oxidization) “typically have parasitic loads of around 2 percent of nameplate capacity.”

He continues: “If the costs of pollution control equipment are necessary to keep an existing plant in operation and the alternative is to retire … then the costs are avoidable… [T]his is consistent with the Commission’s order concerning capacity market mitigation measures for New York City.” (See, Protest of Constellation, Affidavit of Jonathan Lesser, ¶¶ 13-24, FERC Docket EL08-67, filed July 11, 2008.)

The One-in-Ten Test

When PJM designed its capacity market, it discovered that its traditional installed reserve margin (IRM) of 15 percent (IRM would produce an LOLE (loss of load expectation) of only one outage every nine years. To achieve one-in-ten LOLE, it found it needed to boost IRM to 15.5 percent. Adding in a 1-percent buffer to hedge against the risk that resources might not show up at the tail end of a three-year forward period led PJM eventually to modify its RPM parameters and the shape and plot of the VRR demand curve so that the desired equilibrium point would coincide with an IRM of 16.5 percent.

Moreover, PJM thought it needed to achieve a much stricter standard of one outage per 25 years in each individual LDA pricing zone (such as SWMAAC), in order to keep the outage odds at one-in-ten or lower throughout the entire RTO, as it contains many separate zones. PJM enforced this new standard by adjustments to the zone-specific CETO (capacity energy transfer objective), the amount of transmission capacity needed into each LDA zone to assure compliance with this new one-in-25 test.

To understand why PJM defends a one-in-25 test for separate subzones, consider probability theory. A coin flip carries a 50 percent or one-in-two chance of coming up heads or avoiding tails. Yet if one must try three flips in a row to assure avoiding tails in any single flip, one must multiply the odds, so the chances of avoiding a tail across three successive flips are only one-in-eight, or one-half cubed.

Citing these factors, the buyers complaint points to these oddities and alleges that PJM’s RPM capacity market is actually too reliable—that RPM in effect has led to a certain upward creep in reliability standards, and has raised costs for consumers in a similar fashion. And here the buyers might at last be onto something concrete.

Several months ago, responding to earlier complaints about rising prices, FERC had directed PJM to conduct an internal review of its RPM capacity market. (See, Docket ER05-1410, order issued Apr. 17, 2008, 123 FERC ¶61,037.) Thereafter, on June 30, PJM filed with FERC in the same docket a report conducted by the Brattle Group, which largely endorses RPM as effective, but which in fact also recommends some spots for possible improvement.

Overall, the Battle Report finds that prices in the four transition auctions “have largely followed the pattern set by reserve margins and [have] moved toward the price required to sustain new entry.” The Brattle Report also finds that prices in the SWMAAC and EMAAC zones now have tended to fall back to the norm, even as surplus capacity levels in the Rest-of-RTO zone have fallen back as well between the first and fourth BRAs. This is the sort of convergence one would expect to see in a successful market.

At the same time, however, the Brattle Report questions whether PJM is applying the one-in-ten LOLE standard appropriately throughout the RTO region, and whether the one-in-25 standard for individual LDAs is justified.

Above all, the report questions how much reliability the nation really can afford—“Whether the same target level of reliability should be maintained as the cost of capacity increases.”