How Electricity “Capacity” Markets Work

There has been a lively discussion in the comments threads of recent Bacon’s Rebellion posts about what lessons Virginia can learn from the near-collapse of Texas’ electric grid. A key difference between the two states is that Texas maintains its own reliability council, ERCOT, while Virginia belongs to an interstate compact, PJM. Both organizations administer auctions to sell electricity in near-real time. Unlike Texas, PJM maintains a market for future electricity “capacity.” The role of capacity markets is hard for most people (including me) to wrap their heads around. But reader Allen Barringer (Acbar), a retired utility regulatory lawyer, gives it a shot. — JAB

The concept of reliability in electricity grids is probabilistic. There is no such thing as absolute certainty of reliability. In general, the acceptable risk of an outage is defined by the North American Electric Reliability Corporation (NERC), a standards-setting organization regulated by the Federal Energy Regulatory Commission (FERC), which sits atop around a dozen regional reliability councils whose members are the utilities and Independent System Operators (ISOs) that run the electric grid. The reliability criterion is that consumers should not lose electric service as the result of problems on the “bulk power” electric grid more often than one day in 10 years.

State regulatory authorities such as Virginia’s State Corporation Commission (SCC) don’t regulate the bulk power grid; they focus on local reliability issues like distribution line outages. But the states also regulate retail electric rates and, in Virginia, the SCC reviews the “integrated resource plans” (IRPs) of the retail electric companies.  

If you know how much risk of involuntary outage to customers you can tolerate, how do you translate that into generation and transmission policy? That’s what electric utility planning and its regulatory spin-offs like the Integrated Resource Plan are all about.

The planning used to be entirely up to the utilities and their state regulators. FERC entered the picture mainly when it came to the reliability of wholesale supply to distribution cooperatives and municipals. That changed in a big way with the advent of independently owned generation companies, which put the competitive squeeze on utility-owned generation built on a cost-plus basis. The independents demanded equal (“common carrier”) treatment by transmission owners.

How we got to where we are. Legislation back in the Carter years discouraged the use of natural gas, which was scarce in those days, and encouraged “cogeneration” by little independent generators and DIY homeowner solar based on “avoided cost” compensation. That led eventually to the deregulation of generation by the states and the establishment by FERC of wholesale electricity markets bound together by common-carrier transmission facilities and managed by independent system operators — all the changes of the 1980s and ’90s.

PJM now has an independent role in transmission planning and in planning capacity connections to the grid and capacity adequacy for Load Serving Entities (commonly referred to as LSEs, a term for electric companies). PJM has a broader, more interstate (regional) view than a single utility. It can’t order new facilities to be built — only FERC or the state commissions can do that — but it can blow the whistle early on problems, and its tariff provisions for LSEs give incentives (e.g., capacity market payments) for unregulated generators to cooperate.  State regulators now regularly call in PJM for its views on things like IRP filings by their regulated utilities.

A formerly fully-integrated Virginia Electric & Power, responsible for its own reliability and reviewed only by the SCC, is now broken down into jurisdictional layers:  a collection of Dominion generators selling to PJM on the same terms as the independents, essentially unregulated except through markets regulated by the FERC; a set of bulk power transmission facilities now comprising the “Dominion Zone” of PJM and operated under PJM direction but still owned and maintained by Vepco, all regulated by FERC; a set of distribution facilities owned, operated and maintained entirely by Vepco and regulated by Virginia (I’m leaving North Carolina out of this simplified discussion), and retail sales of electricity to customers by its load-serving entity or “LSE” unit, with rates and service regulated by Virginia (again I’m leaving out N.C. and non-jurisdictional/ military customers here). In addition to the investor-owned utilities (IOUs), Virginia’s electric coops and municipal utilities have an LSE function — in their case that’s about all they do, except as co-owners of generation through ODEC.  

Modern utility ratemaking has always distinguished between fixed and variable cost recovery. The calculation of avoided cost also has a fixed cost and a variable cost component. The power pools that pre-dated the ISOs dispatched their generation based on variable costs, so it was a natural evolution to base the new energy markets on variable costs and create “capacity” markets for transferring entitlements based on fixed costs. People got used to thinking about generation costs that way.

Economists will tell you that separate energy and capacity (variable and fixed cost) markets are unnecessary; you can in theory recover all your costs through a “pure” auction-based energy market with bids for sell and buy prices that could go sky-high in a shortfall — and that’s what Texas’ ERCOT tried to establish. FERC, however, bought into the nascent Northeast/Mid-Atlantic/Midwest ISO arguments for a separate capacity marketplace, partly for market price stability (which the LSEs and state regulators wanted) but mainly as a way to provide the system operator with sufficient resources contractually rather than point fingers after the fact. This in turn promoted reliability planning and early warnings in advance of reliability problems, which all the regulatory bodies liked.

How does a capacity market work? PJM is presumed to be reliable overall if its LSEs bring sufficient capacity commitments to the table contractually (including by ownership) to meet their load plus reserves. A capacity commitment means that the generator involved must deliver when called-upon (“dispatched”) by PJM, or the LSE will have to pay an extremely stiff penalty.  The generator’s location is irrelevant as long as it can deliver to the PJM backbone transmission; “bottled” generation in excess of local demand is downrated.  The generator is excused from responding if its reasonable pre-negotiated limitations on response time from cold start or “spinning-reserve” status are not met, or if it is out of service due to scheduled maintenance approved in advance by PJM, or forced out of service due to a forced outage meeting (in PJM’s judgment) reasonable criteria for such things, or unable to participate due to a transmission limitation. Overall, PJM’s reserve requirement should cover these exceptions.

The main requirement is the long-term “base” capacity market, in which LSEs have to buy delivery commitments three years out, to meet their peak forecast load (plus the PJM-assigned reserve requirement) for the delivery year. LSEs that fall short due to load adjustments or generator retirements as the planning year approaches can trade or buy in the “base residual” capacity market to adjust their capacity commitments accordingly. There are “incremental” capacity auctions for fine-tuning these amounts of available capacity as late as the day ahead of performance. But woe to the LSE that is short of capacity on the day of performance. I am summarizing, of course, but the details are easy to find here.

The PJM energy marketplace allows the generator to bid any price. Too high a bid and it won’t be called upon to run. Too low a bid and it will run but lose money. The sweet spot is to bid roughly the unit’s variable cost in the energy market and recover as much fixed cost as possible from capacity sales. All units that are dispatched are paid the market’s marginal dispatch rate, which is at least what they bid but usually more than that. Therefore, there is an income stream from energy sales above incremental Operations & Maintenance most of the time, which supplements the negotiated price generators get for their capacity commitment.

If the sum of those two sources of revenue above Operations and Maintenance (O&M) doesn’t exceed fixed costs, then the generator’s owners ought to shut it down and exit the marketplace. (Of course, if the integrated utility bypasses the market’s discipline by having its state commission include a generator’s fixed costs in the affiliated LSE’s retail rate base, rather than requiring that generation subsidiary to negotiate arms-length contracts with buyers in the capacity marketplace, there is no competition, and its regulators won’t know if the ratepayers got a good deal or a bad one.)

How about renewables? All this works pretty well for regular, fossil-fueled and nuclear generators. How can a solar or windmill unit deliver on a capacity commitment when there’s no sun or wind, or a hydro unit when there’s no water? They can’t. They are non-dispatchable. So, originally, they were excluded from capacity markets. These generator owners, however, argued successfully to FERC that they did have some value for reliability as there was a measurable probability, if not certainty, that they would be there when needed. To make a long, complicated story short, PJM now credits commitments from such units with a partial capacity value if the total commitment is not in excess of certain percentages of an LSE’s total generating capacity. It’s not much, and there’s no way under current rules that an LSE could present only renewables capacity to meet its total capacity requirement, but it’s something.

There is an ongoing FERC effort to reform capacity market pricing to deal with subsidizing low-carbon-emissions generation. This is not feel-good tokenism but a serious effort to address a more subtle problem: nuclear retirements. Nuclear fuel is no longer the bargain it used to be, and old nuclear plants coming up on life-extension expenses force a choice:  retire the unit immediately and build new, cheaper $/MWH, gas-fired generation — or ask for a subsidy of some kind from ratepayers to keep the nukes in service. Everyone who loves clean, environmentally friendly nuclear power fails to reckon with the current economics of these dinosaurs. The utilities — most nukes are owned by utilities, not independent generators — have become expert at shaking down retail regulators for payments — a good example being Dominion’s recent deal with the Connecticut public utility commission to keep its Millstone nuke in service rather than retire it.

The problem is, these subsidized nuclear units then bid into the marketplace at “artificially low” prices displacing other generators who say the regulated nukes aren’t playing by the same competition-based rules. FERC would like such payments to be recognized somehow in how these units are treated in the wholesale capacity and energy markets — but due to differences of opinion on the FERC, there has been regulatory deadlock and no progress on this front.

A similar competition-distorting issue is the “subsidy” (as some see it) that renewable generators receive from their sale of renewable energy credits (RECs) in the state-run REC markets. Those sales, of course, are driven by state mandates for ever higher percentages of renewable resource generation in an LSE’s retail sales under state jurisdiction. If in fact we ever develop a national carbon emissions trading methodology with teeth in it, then monetizing the low-carbon value of renewable resources in the wholesale capacity markets will also become a priority for the FERC and the regional transmission organizations. In short, capacity is no longer bought and sold purely on price competition, but the markets haven’t caught up. This is a big problem for the FERC right now; it has had competing proposals before it for a couple of years now (including a complicated compromise proposal from PJM) but there’s no hint where it plans to go with this issue.

That’s enough for now