Changing settlement rules might offer a fix for broken power markets.
Before consumers endure yet another summer of tenuous reliability and high electricity prices, perhaps it is time to consider changing the rules of the game. One such rule is the common practice in regional electricity spot markets whereby the last (highest) accepted supply bid sets the clearing price for all generators. Does this practice, known as a "uniform price auction," make spot markets more vulnerable to anti-competitive behavior and the abuse of market power?
In fact, some experts have suggested that the likelihood of market power abuse depends, to some degree, on the settlement rule used to determine payments in wholesale auctions. They suggest that the exercise of market power is likely when auctions adopt uniform pricing-i.e., "last bid sets price"-in a market with high demand and low supply.1# The analysis I offer here appears to confirm this theory. More importantly, my analysis suggests that a "discriminatory price auction"i.e., "pay according to bid"might offer a better approach for reducing prices in bulk power markets.
This article evaluates both types of auction settlement rules: (1) "last bid sets price," by which the last offer accepted determines the price paid to all participants; and (2) the "pay as bid" model, by which each participant is paid the amount bid by that party. By modeling the markets for energy and various ancillary services in a large control area, it is demonstrated that under conditions of market power, substantial revenues with commensurately high profits can be commanded under a "last bid sets price" price auction.
By contrast, by employing a "pay as bid" price rule, much of the impact of market power can be ameliorated. In addition, a "pay as bid" price auction, by virtue of requiring each bidder to explicitly state their desired revenue rate, is more transparent and more readily exposes attempts to make use of strategic pricing and market power. By discouraging the use of market power through greater price visibility, "pay as bid" price auctions also may reduce instances of strategic capacity withholding, which in turn, should enhance overall system reliability.
Settlement Rules: A Look At Theory
The trading of electricity in today's open market is coordinated by an entity that seeks to match buyers (i.e., loads) and sellers (i.e., generators). That entity may take the form of a power exchange, as in California, and/or an independent system operator (ISO), as found in many states. In addition to serving as matchmaker for electrical energy consumers and suppliers, the ISO also must ensure that the overall grid system remains stable, responsive, and reliable in the face of unexpected events such as an immediate plant outage (trip). To do so requires the commitment of generation assets that must stand ready to react to system changes. These activities, referred to as ancillary services,2 have been defined such that they, too, can be procured in a competitive market environment. Thus, suppliers have multiple, interdependent markets in which they can choose to participate.
In its simplest form, an auction is conducted for energy and each ancillary service in which suppliers offer an amount of capacity for a given period of time for a particular price (i.e., the bid price). The coordination entity (e.g., ISO) selects winning offers on the basis of rank-ordered bid price until the projected demand is met.
The most commonly used settlement rule is to pay winning participants at a rate equal to the last (i.e., highest) accepted bid price, irrespective of the participant's own bid price. This "last bid pays price" approach is referred to as a uniform price auction, and is well documented in economic literature (e.g., Feldman 1993). It is not clear, however, why this settlement rule has become the one most often applied to multi-unit, multi-dimensional markets for electricity. A number of researchers have begun to question the use of "last bid pays price" settlements in these markets, and suggest that there are other settlement rules that can produce a more efficient and cost-effective outcome (Mount 1999, Oren 2000).
The rule proposed by Mount (1999) and evaluated in this article is the discriminatory price auction. Under this method, the selection process is the same as described above, but, in this case, the winning participants are paid exactly what each bid. This method also has been called the "pay as bid" approach.
Mount (1999) provides a good theoretical basis for using a "pay as bid" price auction in electricity markets. My analysis evaluates the quantitative benefits to the consumer of using this approach, as compared to the uniform price approach for settlement of energy and ancillary service markets in a typical, large transmission control area.
A Market Simulation: Measuring the Impact of "Gaming"
In order to obtain cost results with realistic orders of magnitude, the Oak Ridge National Laboratory Electricity Market Model was used to simulate a large transmission control area having a peak annual energy demand of 50,000 megawatts. This multi-generator, multi-hour simulation model is a tool for better understanding, testing, and predicting the resulting prices, participation in, profits, and coverage of the interrelated, competitive electric energy and ancillary services markets.
The generating units available to meet the load were modeled after plants located in the PJM Interconnection control area. Recent actual plant data served as a guide in developing model inputs. Though similar in size and composition to the PJM market, this analysis and associated simulations are not intended to portray the PJM area specifically, but rather, to provide results that are representative in behavior and magnitude of large, multi-plant control areas having a broad mix of generation types.
As mentioned earlier, aberrant price behavior occurs frequently when a system experiences high demand. For this analysis, a peak hour was simulated in which the system load factor was 94 percent. In addition to an energy demand of 48,700 MW, ancillary service demands included 500 MW of regulation, 1,150 MW of spinning reserve, 1,000 MW of load following, and 500 MW of non-spinning reserve. Bids for each plant were based on the expected marginal price for that hour. In practice, suppliers do not know with certainty what the total demand for a given hour will be. Historical data coupled with weather predictions can provide fairly accurate estimates, however. Similarly, bidders cannot predict with certainty the last accepted bid price, so bidders likely would bid their expectation of the marginal price or perhaps slightly less to improve their selection chance.3
For the given hourly load in this study, the expected marginal price was assumed to be between $200 and $250 per megawatt-hour, with bids varying accordingly. For the few plants whose production cost was greater than $200 per megawatt-hour (e.g., small, oil-fired internal combustion units), the energy bid for that plant was set to its production cost. Ancillary service bids were related to the energy bid, adjusted upward by 5 percent to 15 percent to recognize the incremental cost impacts of varying (and/or lower) output operation.
If that were all that was done to produce plant bids, the economic impact of "last bid sets price" and "pay as bid" price auctions on the consumer would not differ greatly. Thus far, bids have relied on the assumption that the bid price for the marginal unit will be based on its costs with, at most, some allowance for a fair profit. In a competitive market with many potential suppliers, that is not an unreasonable assumption. However, what has been experienced on occasion in actual markets is an exercise of market power in which inadequate supply has caused exorbitant prices to be bid under the logic of charging what the market will or must bear. Whether the shortage is due to true resource shortfalls or intentional supplier withholding, the result is that during that period, suppliers can be the tail that wags the dog. It is in these circumstances that "pay as bid" price rules can protect the consumer by limiting profits that accrue to participating suppliers.
The Electricity Market Simulation Model
In order to understand the economic impacts of alternative market mechanisms in the scheduling and procurement of electric energy, a means of simulating the behavior of markets under various rules and conditions is needed. As part of Oak Ridge National Laboratory's support to the Office of Power Technologies, a multi-generator, multi-hour simulation model has been developed to facilitate analysis of various market arrangements. The ORNL Electricity Market Model (OREMM) serves as a tool to better understand, test, and predict the resulting prices, participation in, profits, and coverage of the interrelated, competitive electric energy and ancillary services markets.
The PC-based simulation model has been designed with the following attributes:
- Capable of multiple, diverse generation units and fuel types;
- Sequential hourly analysis of energy and individual ancillary service demands;
- User-provided bids for energy and ancillary services by unit;
- Tracking of sales, actual cost, and profit by unit; and
- Capable of modeling different market rules and behaviors.
The model simulates the auction process in which hourly energy demand is satisfied by bid-ordered generation. After an initial energy assignment has been made, the various ancillary services are considered in order of their required response time (i.e., regulation, then spinning reserve, load following,* and non-spinning reserve). Each of these services is limited by the user-defined quantity available and unit ramp rate considerations. In a manner similar to the energy assignments, each ancillary demand is matched to generation resources, ranked by bid price.
For each market, the model calculates the amount paid to each unit, the unit's internal variable cost (based on heat rate, fuel cost, and other variable operating costs), and the resulting profit from the transaction. These amounts are summed across all units to obtain a total-market payment, cost, and profit for each service. -R.H.
*Although not universally defined as an ancillary service, load following is included in the model to reflect the capacity margins that must be available for intra-hour load changes.
To quantify the effect of these two pricing approaches, it was assumed for the simulation that market power was exercised in the spinning and non-spinning reserve markets. In these markets, a few bidders (including the marginal bidder) offered reserve at 50 times their plant's normal production cost, resulting in bid prices greater than $1,000 per megawatt-hour. In comparing the impact of the two settlement rules, identical input data (e.g., bids, loads, plant parameters) were used.
For each electricity service (i.e., energy, regulation, spin), the electricity market model determined the degree of utilization, cost, revenue, and profit for each plant submitting a bid. A listing of the outcome for each plant is too lengthy to include in this article, but a summary of the results under both types of price rules appears in the table, "'Last Bid Sets Price' vs. 'Pay as Bid' Pricing." (See p. 48.)
Of particular interest are the results for the spin and non-spin services, where market power has been exercised. Both revenue and related profits received under a "last bid sets price" rule are considerably higher than under the "pay as bid" price rule. These findings are shown in the figure, "Average Electricity Revenue by Auction Rule." A graph of average profits would be nearly identical.
These results reveal that the "last bid sets price" auction rules indeed allow market power in the form of limited competitive bids to influence the entire market. The "pay as bid" price rules, by contrast, limit the impact of such market power by not applying the marginal price to all inframarginal units. Further, by requiring the submission of individual "to be paid" bids, this auction permits greater scrutiny of bidding behavior. And by discouraging the use of market power through greater transparency, "pay as bid" auctions also may reduce strategic capacity withholding, thereby enhancing system reliability.
Back, K., J. F. Zender, "Auctions of Divisible Goods: On the Rationale for the Treasury Experiment." , Vol. 6, 733-764, 1993.
Feldman, Robert A., Rajnish Mehra, "Auctions: A sampling of techniques." , Vol. 30, Issue 3:32, September 1993.
Mount, Timothy, "Market Power and Price Volatility in Restructured Markets for Electricity." , January 5-8, 1999. Maui, Hawaii.
Oren, Shmuel S., "Design of Ancillary Service Markets." Working paper. August 2000. University of California, Berkeley.
Smith, Rebecca, "Northeast Faces Electricity-Price Surge-Costly Oil-Fired Plants May Drive Summer Rates." , Eastern Edition. March 20, 2000. A2.
Smith, Vernon L., "Experimental Studies of Discrimination Versus Competition in Sealed-Bid Auction Markets." , 40:1, 56-84, January 1967.
U.S. Department of Energy, "Horizontal Market Power in Restructured Electricity Markets." DOE/PO-0060, Office of Policy, USDOE, March 2000.
Wilson, R., "Auctions of Shares." , Vol. 93, 675-689, 1979.
Wolak, Frank A., Severin Borenstein, James Bushnell, "Diagnosing Market Power in California's Restructured Wholesale Electricity Market." Stanford University, www.stanford.edu/~wolak/, April 2000.
Wolfram, Catherine D., "Strategic bidding in a multiunit auction: an empirical analysis of bids to supply electricity in England and Wales." , Vol. 29, No. 4. Winter 1998.
1 In his seminal work on market auctions, Smith (1967) proposed that in shallow market conditions, bid values and related revenues would be higher in "last bid sets price" price auctions than in "pay as bid" auctions. Subsequent research appears to confirm Smith's findings (Wilson 1979, Back and Zender, 1993, Wolfram 1998). Wolfram (1998) suggests that the incentives for high-priced, strategic bids "would not exist if the electricity auction were discriminatory." DOE (2000) states that a recent analysis by Wolak (2000) indicates that exercise of market power in California during the summers of 1998 and 1999 resulted in more than $800 million in payments above competitive levels to generators. More recently, the "last bid sets price" method has garnered the attention of the broader media as a cause of high electricity prices in the Northeast (Smith 2000).
2 The Federal Energy Regulatory Commission has defined ancillary services as those "necessary to support the transmission of electric power from seller to purchaser given the obligations of control areas and transmitting utilities within those control areas to maintain reliable operations of the interconnected transmission system." Ancillary services include regulation (maintaining minute-to-minute generation/load balancing), spinning reserve (capacity that is synchronized to the grid that can respond immediately to grid disturbances), and non-spinning reserve (capacity not connected to the system but capable of being brought on-line and serving demand quickly, e.g., 10 minutes).
3 Under uniform price auctions, bidders are not compelled to bid the expected marginal cost, as only one plant will determine the payment value for all successful bidders. As a result, zero price bids have been observed in uniform price auctions in order to guarantee selection. In a discriminatory price auction, each bidder must submit a more carefully considered bid, as the revenue obtained will be directly related to the bid amount.
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.