Electricity prices are rising. These price increases coincide with policy changes in many parts of the country that introduced into the electric industry greater reliance on market forces. Although today’s electricity prices still are relatively low in historical terms (about two-thirds of their 1980s levels when adjusted for inflation1) and rising electricity prices have been largely the result of movements in global markets for fossil fuels, these price increases nonetheless have placed pressure on policy makers in a number of recently restructured electricity markets to question whether power prices have increased due to the design of competitive markets. Some observers have begun to push for redesign of market rules or even a return to elements of traditional cost-of-service regulation in the electric industry.2
Among the proposed reforms are changes to the design of auction processes used in various wholesale electricity markets. These auctions involve offers to supply (and, possibly, bids to buy) power. The auction determines the identity of the winners—those competitive suppliers selected to supply power in any given time period—along with the price and other terms of the power sale. Typically, the buyer is some entity with responsibility to supply power to end-use consumers of electricity. Notable examples of such auctions in the power industry are the organized3 energy and capacity markets administered by various regional transmission organizations (RTOs). In addition, in some states with retail choice, distribution utilities have used auctions to obtain power supplies for their basic-service customers.
In response to rising prices in some of these markets, some observers advocate a switch from uniform-price auctions (also called single clearing price markets), relied on almost exclusively in all organized wholesale power markets in the United States, to pay-as-bid auctions.4 These alternatives differ in the way the price awarded to winning suppliers is determined. Under uniform-price auctions, all suppliers receive the same market-clearing price, set at the offer price of the most (or nearly most) expensive resource chosen to provide supply. In contrast, in a pay-as-bid auction, prices paid to winning suppliers are based on their actual bids, rather than the bid of the highest priced supplier. For this reason, pay-as-bid auctions also are known as “discriminatory auctions” because they pay winners a different price tied to the specific prices they offer into the auction.
At first blush, the intuitive appeal of a pay-as-bid auction is obvious. Why pay a higher price to a supplier than the price at which he or she offers to sell the product? Wouldn’t doing so simply impose higher-than-necessary costs on the buyer—and by extension, to the consumer? Wouldn’t switching to a pay-as-bid model produce lower electricity costs for consumers?
Although pay-as-bid auctions may appear like a quick fix for rising prices, switching to a pay-as-bid approach likely would produce just the opposite result. This counter-intuitive outcome stems from the propensity for strategic-bidding behavior, as well as the resulting inefficiencies in plant dispatch and capacity investment. A change in auction design would do little to other pressing market concerns, including ensuring adequate T&D resources and increasing the role of demand response. Further, continued changing of market rules creates regulatory uncertainty and fears of regulator opportunism that may discourage investment in new generation and transmission facilities.
Based on long-standing principles, even when it was composed of highly regulated utilities providing power under a government-authorized monopoly structure, the electric industry’s clear goal has been to provide reliable power as efficiently as possible.5 In theory, efficiently produced and supplied power supports the goal of providing electricity to consumers at the lowest price.
But what does “efficiency” mean for electricity markets? First, it’s important to distinguish between short-run (static) efficiency and long-run (dynamic) efficiency. Static efficiency means: 1) output is produced by the least-cost suppliers; 2) it is consumed by those most willing to pay for it, and; 3) the right amount is produced. Achieving short-run productive efficiency (for a given load) essentially has devolved into the problem of operating existing power resources (i.e., existing plants and dispatchable demand-side resources6) according to the principle of economic dispatch—that is, running plants in a dispatch order so the ones with lowest operating costs, or short-run marginal costs, are dispatched ahead of others with higher operating costs.
The nature of the electricity grid—including transmission constraints and reliability requirements, and resource constraints, such as plant start-up costs or ramping times, and demand-response notice periods for different facilities—adds many wrinkles to achieving least-cost dispatch, but doesn’t fundamentally change the problem. Solving this problem effectively depends on information that allows control operators to order plants and other resources from least-costly to most-costly—and a host of other things, such as the degree control operators can minimize system costs over a wide geographic grid, issues that are not greatly affected by choice of auction design.
Even as some parts of the country have shifted control of the grid from regulated utilities to RTOs, the basic process by which least-cost dispatch is implemented has not changed dramatically. In all regions, control-area operators minimize system costs given the security constraints of maintaining system reliability—that is, satisfying all demand for power. An important difference emerges, however, in the information used by control-area operators to minimize costs. While non-RTO regions rely upon cost data provided by the utilities, RTO regions rely on auction-based bids from suppliers. While there are many important differences between RTO and non-RTO models affecting economic dispatch—e.g., the effectiveness of resource “pooling” across states and utility service areas and access to transmission rights—an important issue differentiating alternative auction designs is their ability to elicit accurate information about the relative costs of operating power plants.
While economic dispatch is designed to produce power as efficiently as possible given the generating and other resources that already exist in the system, there is still the critical question of whether the market system provides the incentives for development of the most efficient generation, transmission and demand-side infrastructure. Achieving long-run, dynamic efficiency thus requires developing the right amount and type of resources, such as generation facilities, in the right locations. To create this infrastructure, the system must create price signals that encourage the right incremental investment. It also must provide incentives that promote a competitive market structure, and provide sufficient confidence that the system of pricing and compensation for power production will afford a reasonable opportunity for investment recovery given the capital intensity of, and multi-year lead times for, plant construction.
A major objective of industry restructuring has been to improve the price signals for efficient investment, along with improving the incentives for cost-effective implementation of investments (and cost-effective investments in existing plant operations). In principle, differences in the potential level of static and dynamic efficiency emerge for each of these regulatory models (as illustrated conceptually in Figure 1).
Setting aside the question of whether restructuring has increased electric-market efficiency, several important factors influence an analysis of auction-design alternatives.7 First, there is an important difference in how investors are compensated under the new RTO markets. While independent power producers (IPP) in RTO markets rely upon revenues from energy, ancillary services and evolving capacity markets, they do not—unlike generators under cost-of-service regulation—receive compensation for fixed costs (including return on investment) through rate base or through depreciation charges in base rates.8 Second, the price signals created by alternative auction approaches yield important consequences for the efficiency of resulting investment incentives. Uniform-price auctions develop these price signals differently than pay-as-bid auctions do.
How, then, do these such differently structured wholesale markets provide for supporting a power plant’s combined investment and operating costs? Does one overpay suppliers compared to another? These questions can be answered by comparing single-clearing-price auctions and pay-as-bid auctions, and considering the implications for prices and investment.
Pay-as-bid auctions are unlikely to have any significant effect on market prices for electricity, and more likely than not will raise prices rather than lower them.
To compare auction models, it’s easiest to start with a uniform-price auction. Suppliers bidding in a competitive uniform auction have the incentive to bid their opportunity cost of selling electricity into the RTO market, which could reflect the cost of producing electricity or the lost opportunity to sell electricity at another time or into another market. A supplier’s bidding alternatives for an individual plant illustrate this incentive to bid at cost. Bidding above costs decreases the likelihood the plant will be selected to supply electricity, without affecting the price they will receive, while plant owners might lose money if they bid below their marginal costs. Bids may be affected by the opportunity to raise prices for other plants owned by the supplier, but that doesn’t dramatically change the intuitive conclusion. In a pay-as-bid auction, suppliers who would bid their marginal costs under the uniform price auction instead bid based on their individual forecasts of the market clearing price.
While suppliers’ changes in bidding strategy leads to a set of offers that are higher overall, it is not immediately apparent whether the price level is affected by the auction choice. In fact, most analyses suggest that, under competitive conditions, market prices are unlikely to differ materially between uniform-price and pay-as-bid auction formats (see Figure 2). While small differences may emerge due to factors particular to the market setting, structure, and design, economic analysis fails to support the conclusion that a pay-as-bid auction will lower overall prices.
Even so, many observers point to the potentially large differences between supplier bids and prices in a single-clearing-price auction as an inherent flaw in that market design (if not in wholesale markets themselves) resulting in excess payments to suppliers. Many observers view these payments as profits, and see these margins as unacceptably high for the suppliers with low marginal costs.
This viewpoint entirely overlooks the fact that in this market design, the margin is part of the compensation to suppliers to recover their fixed costs—that is, all those “other” but real expenses necessary to recoup the investment made to construct the power plant, to employ people to work in it, and maintain and operate it in safe working order. Because these expenses typically do not vary with the output of the plant, they are not a part of the marginal costs that figure into an economically rational bid in a market with a single-clearing price auction. While such fixed costs are collected in rates under traditional cost-of-service regulation—through return on investment in utility rate base, depreciation expenses, labor expenses, and so forth—the mechanisms for fixed capacity payments in some of today’s RTO markets clearly are not intended to fully cover these fixed costs. Instead, they are intended to make up for the fact that revenue streams from current energy markets—even given the current margins—provide insufficient income needed to support new generation investment (the missing-money problem).
Thus, limiting supplier payments to their variable generation costs would erode further investment incentives and thus exacerbate concerns about resource adequacy with risks for security of supply and increases in opportunities for market withholding. Further, changes in supplier-bidding strategy under a pay-as-bid auction would foil efforts to limit payments to them. Thus, a switch to a pay-as-bid auction does little to address concerns about either the affordability of electricity rates or the reliability of power supplies. The hoped-for outcomes are illusory.
Some analysts have suggested the pay-as-bid approach may reduce short-term price volatility compared to a uniform-price approach. But while a pay-as-bid approach may reduce short-term volatility—that is, volatility across hours or days—such volatility can be hedged through financial instruments. And that volatility even might promote investment in peaking technologies that rely upon price spikes in a few hours to justify initial capital investments. However, pay-as-bid auctions offer no panacea for reducing longer to medium-term price uncertainties that may act as a deterrent to new generation investments or raise risks for consumer budgets.
A switch to a pay-as-bid auction, while having little impact on prices—and consumer payments—in wholesale electricity markets, likely would have an adverse impact on efficiency of economic plant dispatch. Uniform-price auctions generally achieve efficient economic dispatch. Because of the tendency for plant owners to bid in relation to their marginal costs, the order of plants based on supplier bids is the same order as that based on marginal costs (see Figure 2).
Under a pay-as-bid auction, suppliers’ bids reflect their estimates of expected market prices, not marginal costs. However, because supplier forecasts are likely to differ due to a range of factors unrelated to the underlying costs of generating electricity, the plants with the lowest bids may not reflect the plants with the lowest production costs. For example, a plant with low variable costs and an overly optimistic forecast of clearing prices inadvertently may bid above the eventual market-clearing price. As a result, this low-cost plant would not be selected to supply electricity while other plants with higher costs would be chosen and dispatched.
Figure 2(1B) illustrates this inefficiency: The lowest cost plant (Plant A) is not a winning bidder and does not supply electricity because its price forecast (and its offer price) was above that of the eventual market-clearing plant (Plant F). In its place, higher cost plants—Plant E and a portion of Plants F—are run that would not run on economic merit alone. Consumers pay no less in this example, but society’s resources have been used less efficiently than under uniform-pricing (Figure 2-1A), where only the least-cost resources are selected to generate power. While not costing consumers in the short-run, such inefficient use of resources eventually will lead to higher consumer costs in the long-run.
Given unique circumstances in electricity markets that greatly increase the complexity and uncertainty of price forecasting, differences between supplier forecasts may be large. Bidders interact in a repeated bidding environment, involving multiple markets that are inter-related over time, space and across commodities, including markets for generation, transmission, capacity, emission allowances, and ancillary services. Idiosyncrasies in power production, such as the start-up costs, create interactions between an individual supplier’s bid across the day or week. Because electricity cannot be stored, unanticipated shifts in supply, demand, or the operations of other suppliers dramatically can shift prices from day to day, or even hour to hour.
These dispatch problems that may be introduced by pay-as-bid auctions were not present under the traditional cost-of-service model. In fact, as long as data provided by the vertically- regulated utility were accurate, power pools were designed to minimize costs across utility service areas, and all suppliers had non-discriminatory access to transmission, the traditional cost-of-service model did a fairly good job of minimizing overall power production costs, at least for the generating resources available to the dispatcher—typically those in the supply mix of the utility.
Pay-as-bid auctions also might affect incentives for investment in new generation facilities. For example, additional market forecasting costs, necessary to compete in a pay-as-bid market, might reduce incentives for investment in new generation facilities. More important, pay-as-bid auctions inadvertently may discourage investments in baseload and some renewable generation technologies. Under pay-as-bid auctions, technologies with low variable costs likely will shade their bids below their true expectation of the market price to reduce the risk they inadvertently bid above the eventual clearing price. Because low-cost plants earn large margins when they are selected to produce power, these plants face a greater opportunity cost if their bids inadvertently are above the eventual clearing price. By shading bids, however, expected financial returns to investments in these technologies are reduced, which biases the mix of investments toward shoulder and peak generation technologies—such as natural gas-fired combined cycle and combustion turbines. Such a shift would increase the long-run cost of electricity generation and, as a result, likely would lead to corresponding increases in prices.
The design of wholesale RTO electricity markets continues to evolve. These changes are motivated by the continued desire for more reliable and efficient electricity production, as well as for more affordable electricity. Changes in wholesale market design to a pay-as-bid approach are unlikely to reduce prices, and may make matters worse by potentially leading to price increases, creating dispatch inefficiencies and decreasing incentives for baseload facilities.
Indeed, there is a risk that needed efforts and resources to improve facility siting, promote demand response, encourage and determine appropriate forward contracting, and refine capacity market design—to address the missing-money problem—might become diverted by an effort to switch to an auction format that does not address any of these needs.
1. EIA, January 2007, Monthly Energy Review.
2. See, for example, Burkhart, Lori A., “Regulator’s Forum: Restructuring Rollback, Public Utilities Fortnightly, November 2007.
3. The so-called organized wholesale markets are those that rely principally on centralized energy spot markets.
4. Laffer, Arthur B. and Patrick N. Giordano, “Exelon Rex, Will power deregulation in Illinois benefit consumers or utilities?” Wall Street Journal, Dec. 1, 2005.
5. See, for example, Morin, Roger A., New Regulatory Finance, Public Utilities Reports, Inc., 2006, p. 1, including reference to Bonbright, James, Principles of Public Utility Rates, New York: Columbia University Press, 1966, p. 3.
6. Dispatchable resources—whether from available generating units or demand-side resources—are the tools available to the dispatcher to meet electricity demand at least-cost at any point in time. For simplicity, in the rest of this paper we discuss the dispatch of power plants, while recognizing that many of these same discussion points apply to the dispatch of certain demand-side resources and even of certain controllable direct-current transmission facilities capable of making certain remote generation available for dispatch to a system.
7. Note that Figure 1 is only intended to provide insight into the orientation—but not necessary the relative magnitude—of economic efficiency under alternative regulatory/market models.
8. Capacity markets that are part of some organized wholesale markets provide investors with some fixed payments outside of energy and ancillary services markets, although the levels of and financial risks associated with such revenue streams differ substantially from those provided by regulated returns on investments in rate base.