As electric restructuring spreads around the nation and the world, the idea of a "PoolCo" spot market (pool) gains credence. Pools already exist in England, Australia, Norway, Alberta, and Argentina. On December 20,1 the California Public Utilities Commission formally proposed a pool, called the California Power Exchange, to begin operation as of January 1, 1998. Still, questions remain.
Can a pool adequately compensate all the generators that the market needs, or provide correct price signals for market entry? To some, this question may appear trivial. After all, generators can raise prices until they receive adequate compensation, and new generators will come in if prices are high enough.
If a pool works as intended, generators will have an incentive to bid their short-run variable costs. That assumption is made here (em that generators will in fact bid at short-run variable cost. From there, examples test whether a pool can provide an adequate power supply, compensate generators fairly, and give correct signals for adding new capacity.
Bidding and Dispatch
The pool serves as a central energy clearinghouse to dispatch generation to meet load. The choice of which generation to run is accomplished by an auction in which generators compete by submitting bids. A bid takes a form such as, "I will supply 40 megawatts (Mw) so long as I can receive a price of at least $30/Mwh." The pool operator (often called the "Independent System Operator," or ISO) that manages the auction dispatches generators in order of their bids, lowest first, until load is met. All the generators that get dispatched are paid according to the bid of the last generator chosen, that being the market-clearing price for the period in question. The period would cover a single hour or perhaps half an hour. Pool proponents may envision this spot market as just one mechanism in a market that might feature bilateral contracts, options, and futures. Even so, the pool would occupy a central position, providing a reference point for the other markets.
Ideally, generators bidding into the pool would not collude, explicitly or tacitly. None of them individually would control enough of the market to hold out a reasonable expectation of setting the market-clearing price. In these circumstances, generators should try to bid as low as they can, because this strategy will best assure that they will be dispatched and operating during all hours when the market-clearing price rises high enough to yield some contribution, small or large, to fixed-cost recovery and profit. "As low as they can" would be variable cost or, for a unit that has multiple operating levels with varying efficiencies, marginal cost. Variable cost equals fuel cost plus variable operation and maintenance. Marginal cost represents the change in hourly operating cost divided by change in output, and is close or equal to variable cost when a unit is operating in its most efficient range. If all generators bid variable or marginal cost, the pool can
effectively perform one of its most important functions, economic dispatch (em i.e., meeting load with the least-cost combination of resources.
The pool could also accept bids from the demand side. A demand-side bid would take a form such as, "I'll take 40 Mw so long as the price is less than $80/Mwh."2 Demand-side bids can function like supply-side bids. The market-clearing price is set by: 1) the price bid by the highest-priced generator whose bid was accepted; or 2) the price at which demand drops off sufficiently that it matches the limited supply available below a certain price, or available at any price. For example, if the ISO had exhausted its supply bids or could obtain none below $90/Mwh, it could exercise the demand-side bid described above. In that case, the pool's market clearing price would become $80/Mwh. Load on the pool would drop by 40 Mw.
The first examples below will test a PoolCo market with only supply-side bids. Later examples will assume both supply- and demand-side bids.
Constructing the System
If we can assume a rationally constructed system, the first requirement calls for a load duration curve to represent demand. In this case let's select Pacific Gas and Electric Co. (PG&E) as a model.3 PG&E's curve shows a steep section in the range from 1 to 10 percent of the year; it then continues to 100 percent of the year at a fairly even slope (see Figure 1). For ease of computation, the curve in Figure 1 is smaller than PG&E's in numerical size, but the shape is similar.
Next we need resources to meet the load. Traditionally, resources fall into three categories: baseload, intermediate, and peaking. According to PG&E, baseload resources should meet levels of demand that occur at least 85 percent of the time, peaking resources should meet loads that occur about 25 percent of the time or less, and load levels that occur between 25 percent and 85 percent of the time should be met with intermediate resources. Starting with a rough idea of what peaking, intermediate, and baseload resources should cost to buy and run, we can construct a
plausible collection of resources that will minimize revenue requirements. Figure 2 illustrates the method.
A graph such as Figure 2 was made by adjusting estimates until the baseload plot intersected the intermediate plot at 80 percent, and the intermediate plot intersected the peaking plot at 25 percent. This method formed a continuous (although segmented) curve representing the combinations of the three resource types that minimize revenue requirements over the course of a year. The numbers represented by this curve are shaded in Table 1.
Assume that all Table 1 generators bid their variable costs. Thus, each peaking generator always bids $40/Mwh, each intermediate always bids $26/Mwh, and each baseload always bids $17.5/Mwh.
Table 2 (see page 30) shows revenues and costs for each category of resource through the course of the year, with all demand met and all generators paid according to market-clearing, variable-cost bids. Most of the baseload capacity is always dispatched. It is paid its own bid price during the 15 percent of the time when it is on the margin, and gets the intermediate or peaking price during the other hours. Baseload plants finish the year $26.4 million in deficit, because they fail to fully recover their fixed costs. Intermediate resources get higher average prices when they run, but their operating costs are higher and they run less often. They finish the year $15.9 million in deficit. Peaking resources run still less often, and never get paid more than their operating costs, so they finish the year with their full fixed-cost obligation of $13.8 million outstanding. Total losses for the year for the entire resource mix amount to $56 million. Note: this figure marks the precise cost of 1,000 Mw (the full peak load) of capacity, priced at $56/Kw-yr, or the fixed cost of a peaker. Table 2 indicates that the generators are all losing money, and suggests the reason: They are not being paid for capacity. Capacity is traditionally valued according to the fixed cost of peaking units.
At least in the example of Table 2, the stream of pool prices paid to generators falls well short of the revenue they would require to remain viable. Consumers enjoy low rates and get all the energy they want, but not on a sustainable basis.
Table 3 (page 31) presents a different case. It shows what happens when the system (em not the same system as in Table 2, but an alternative one (em allows the peak energy demand to go unserved. ("Peak" in this case denotes the top 5 percent of hours in terms of energy demand.) Energy buyers bid up the price of the available energy during those hours, such that a demand-side bid sets the market clearing price at $167.85/Mwh.
Instead of meeting the full peak demand of 1,000 Mw that would exist with energy priced at system marginal cost (as in Table 2), the system meets only 857 Mw. Peaking capacity of 143 Mw (1,000 Mw - 857 Mw) that was present in
the Table 2 system does not exist in this system and does not have to be paid for. Collection of $167.85/Mwh for the 438 unserved energy hours, plus collection of system marginal costs for all the other hours, allows all categories of generation resources to break even. In this example, they all break even exactly to the dollar, and this is not an accident. It indicates that the resource mix (Table 1 and Figure 2) used to meet demand was composed exactly of the optimal proportions of each resource type (optimal, that is, if PG&E's definitions of peaking, intermediate, and baseload represent an optimal system).
The first part of the question posed initially was whether a pool can adequately compensate all the generators that the public may require. Apparently, the answer is yes. With demand-side bidding, all categories of generators can receive full compensation and nothing more, as they do in the hypothetical case of Table 3. In this example, the public's demand for energy is fully met, although 14.3 percent of potential peak demand is priced out of the market. This portion of demand either forgoes the use of electricity or gets its electricity from a source other than the pool. Although consumers pay more for electricity in Table 3 than in Table 2, the Table 3 system is sustainable.
The second part of the question was whether a pool can provide generators with the correct price signals for market entry. The stated fixed costs of generators, shown on Table 1, should be considered to include the minimum required return on investment. If this return is not earned, generators would provide less capacity than the market is willing to buy. If it is met with any surplus, competition should be expected to arise in the form of new capacity that would bring prices down to equilibrium.4
The outcome observed in Table 3 occurs by solving for the demand-side bid which, if allowed to clear the market for 5 percent of the hours, will make all generators whole. The solution is easy, because the required bid equals the full revenue requirement of a peaking unit that runs only 5 percent of the year, or 438 hours. Table 1 shows the answer: $167.85/Mwh. In a real market, however, the level of market-clearing demand bids would determine the amount of capacity in the system, not the other way around.
The level of demand-side bids depends on the elasticity of demand (the response of demand to changes in price). Historically, electric resource planners have given little attention to elasticity. Instead, planners aimed to meet the full demand; the capacity costs were wrapped into bundled rates. The system modeled by Table 2, in which the marginal cost of generation always sets the pool price, would be maintainable at the full 1,000 Mw only with capacity charges assessed to the consumers at the rate of $56 million per year.
In Table 3, consumers do not pay for capacity as such, but they choose to forgo the 143 Mw of generation that they would demand only 5 percent of the time, if priced as in Table 2. Perhaps real consumers would not be willing to forgo so much as 14.3 percent of demand. If so, they could indicate their preference for more capacity by bidding up the market-clearing price. The Table 3 consumers could bid it to a level higher than $167.85/Mwh during the peak hours.5 This rise in price would indicate a lower elasticity of demand. Generators supplying the system would then earn excess (above minimum) profits, signalling entrepreneurs of a potentially profitable opportunity to build a new power plant or plants for the system.
Should the market clear below $167.85/Mwh during the peak hours, all capacity resources of the Table 3 system would fall short of their revenue requirements. For example, if 169 Mw of peak demand were unwilling to pay so much as $103.93/Mwh, the system would support only 831 Mw (see Table 1).6 If generators are not receiving their revenue requirements, consumers apparently would rather suffer greater curtailments than pay for more, or even existing, capacity.
Pool proponents usually emphasize supply-side bidding, and when they mention demand-side bidding it is most often in the context of keeping pool prices down rather than allowing them to rise above system marginal cost. Other ways of paying for capacity have been suggested. In Britain, for example, where a pool system has been in operation for a few years, generators are paid for capacity according to a formula that considers loss of load probability (LOLP) and value of lost load (VOLL). Ancillary service payments, for such functions as spinning reserve and voltage support, could also help to meet revenue requirement for generators.
A pool that pays all generators according to the market clearing bid, supply-side or demand-side, can provide revenue streams sufficient to return the full fixed and variable costs of generation, including whatever returns on capital may be required. This conclusion proves true even if generators are constrained to bidding their variable costs. Such a pool can also give the appropriate signals for adding new capacity resources.
At least in theory, a pool does not favor the interests of either generators or consumers, because profits for generators are limited by competitive forces and consumers still must pay the full fixed and variable costs of generation. However, the pool creates a mechanism by which consumers can choose how much capacity they are willing to pay for. This feature should eliminate the costs of excess capacity and lead to a more efficient system. t
Robert D. Grow is employed as an electricity specialist for the California Energy Commission (CEC), working mainly on restructuring issues. He holds a B.S. in Business Administration from the University of California, Berkeley, an M.B.A. from California State Univ., Sacramento, and a J.D. from Northwestern California Univ. School of Law. His views are not necessarily accepted or endorsed by the CEC.
1. Order Instituting Rulemaking on the Commission's Proposed Policies Governing Restructuring California's Electric Services Industry and Reforming Regulation, Decision 95-12-063, R.94-04-031, I.94-04-032, Dec. 20, 1995, as modified, Decision 96-01-009, Jan. 10, 1996 (Cal.P.U.C.), published at 166 PUR4th 1.
2. The California Public Utility Commission's restructuring plan, released Dec. 20, 1995 (see note 1), includes a Power Exchange that would accept demand-side bids, but the form and content of such bids are not described.
3. Pacific Gas and Electric Co., Resource, An Encyclopedia of Energy Terms, Second Edition, 1992, p. 265.
4. Judah Rose and Charles Mann, "Unbundling the Electric Capacity Price in a Deregulated Commodity Market," Public Utilities Fortnightly, December 1995, p. 20.
5. The market modeled by Table 3 shows a radical price difference between the top five percent of the hours and all the rest. Intuitively, such a jump seems improbable. In a real market, smoother transitions should be expected. One reason they might be smoother is that the amount of capacity participating in the market will vary over time. If a system has 10,000 Mw of active capacity on a July afternoon, it might have only 6,500 Mw active on an October morning. Generators will not try to operate if they perceive little likelihood of running profitably. Thus, demand-side bids could clear the market at times other than system peak.
6. Table 1 gives the answers because of the way the system is constructed. Intermediate and baseload resources are included in the system only to the extent that they can reduce energy costs, so capacity cost net of energy savings are the same for all three categories. Thus a price that will allow all the peakers to meet their revenue requirements will also fulfill the revenue requirements of the intermediates and baseloads.
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.