With the introduction of retail competition in the electricity industry, regulatory authorities in many jurisdictions are now overseeing the purchase of electricity at wholesale by electric utilities for customers that do not otherwise obtain supply from independent retailers. There are two primary ways in which, under the supervision of regulatory authorities, electric utilities purchase electricity for these non-shopping customers: through simultaneous descending clock auctions or through fairly common sealed-bid auctions, commonly known as Requests for Proposals (RFPs).
Descending clock auctions have received a lot of coverage in the trade press and academic literature on electricity. In comparison, sealed-bid auctions have not received as much attention. This discrepancy may be explained by the myriad approaches to purchasing wholesale electricity that entail issuing RFPs that tend to obscure the fact that RFPs simply are sealed-bid auctions.
In states that have introduced retail competition into the electricity market, end-use customers (such as households and businesses) can purchase their power supplies from retailers that are independent of the local utility company. Many end-use customers, however, do not purchase their power from independent retailers and continue to rely on their local utility company. The regulatory authorities in these states oversee the purchase of electricity by electric utilities to supply such “non-shopping” customers1 and generally require that the electric utility purchase this power through a competitive process. In most cases, utilities use sealed-bid auctions carried out through RFPs to purchase these power supplies.
Competitive procurements of electricity for such utilities have been reviewed in recent years.2 The most common competitive procurement process is for the electric utility to solicit sealed-bid offers to provide full-requirements load-following wholesale electricity to these non-shopping customers.3 The RFPs in these solicitations share some characteristics, such as clearly defining the electricity product being purchased and the rules governing the purchase, but often differ in important ways, including the structure of the sealed-bid offers and the method of determining the winning bids. However, these “auction design” details—the structure of the sealed-bid offers and the method for determining the winning bids—play a crucial role in determining the price at which wholesale power supplies will be purchased and hence the rate that non-shopping customers will pay for electricity.
Three auction design elements—tranche size (i.e., the size of each unit of the power product being purchased), bid format, and auction clearing methodology—play a particularly important role in determining the price that these RFPs return. Sealed-bid auctions that allow for flexibility in the pricing structure can yield lower prices for electricity than other formats.
Many jurisdictions use sealed-bid auctions to purchase a particular type of default electricity product referred to as full-requirements load-following power supply.4 Take for example, sealed-bid auctions for full-requirements load-following power supply carried out by 10 utilities in seven states where retail competition has been introduced (see Table 1). In surveying these auctions, the details regarding the quantity of peak load being purchased, the bid format and offer pricing structures, and the auction-clearing methodology appear to be particularly important for understanding how these processes work.5
• Tranche Size: Full-requirements load-following products are purchased in discrete units called tranches. A tranche is a percentage of hourly load for a given grouping of non-shopping customers established by the electric utility.6 Tranches, as measured by the estimated amount of peak load that they represent, tend to be around 50 MW in size,7(see Table 1). However, in some instances tranche sizes are considerably larger. For example, NSTAR (in Massachusetts), Central Maine Power, and Connecticut Light & Power have tranche sizes that are estimated to exceed 100 MW.
Tranche size has important implications for the competitiveness of sealed-bid auctions and the prices that they return. Using relatively small tranche sizes can provide incentives for smaller suppliers to compete and often results in utilities obtaining supplies from more than a single supplier in any given solicitation. On the other hand, larger tranche sizes arguably can lower transaction costs as fewer suppliers ultimately are needed to meet customer loads. It’s difficult to determine with certainty the impact of tranche size on an auction’s competitiveness, given that only limited information on the competitiveness of these auctions is publicly released. However, where such auction details are reported, some anecdotal evidence suggests smaller tranche sizes seem correlated with more bidders. For example, Pepco Maryland has tranche sizes around 50 MW and its recent solicitations have had between seven and twelve bidders, while Connecticut Light & Power has tranche sizes around 400 MW and its recent solicitations have had between four and eight bidders.8 Thus, smaller tranche sizes may encourage greater supply-side competition.
• Bid Format and Offer Pricing Structure: The format of bids being submitted in a sealed-bid auction dictates the pricing offered by suppliers for the products and terms sought by the utilities. Bidders typically input the details of their offers into a pre-formatted Excel spreadsheet that defines the number of tranches, contract duration, and pricing associated with their bids. While bid formats ostensibly are comparable across auctions, offer pricing structures can vary considerably.
An auction’s bid format can be as simple as requesting a single price per tranche that applies to a particular customer grouping for the entire term of the contract, or as complex as requesting several prices per tranche that apply to different customer rate schedules (e.g., within a customer grouping) and vary based on time of day and time of year. Examples of the simple bid formats can be found in the solicitations by the Maine PUC, Delaware’s Delmarva Power & Light, and PPL Electric. These solicitations request a fixed price for a given term for a particular group of customers; the fixed price that wins the sealed-bid auction forms the basis of the customers’ rates.9
Bid formats that request more refined offer pricing do so in two ways. First, prices can be requested for more specific time periods, such as on a monthly basis, for customer groupings. For example, Connecticut Light & Power, United Illuminating, and NSTAR each request pricing on a monthly basis. Connecticut Light & Power and United Illuminating go further and request that a bid’s prices be broken down into on- and off-peak periods. Second, prices can be requested for specific customer rate schedules within a customer grouping and further broken down by the actual diurnal time of use. For example, the utilities serving Maryland and District of Columbia use this type of pricing format. Specifically, these utilities request prices for a particular customer grouping (e.g, residential, small commercial and industrial), but within the grouping they solicit prices by rate schedule. Depending upon the rate schedule itself, pricing can be broken down by season (e.g, summer and non-summer), as well as time of day (e.g., peak, intermediate, and off-peak). Often the more refined prices in the winning bids for these solicitations can be used directly in customer rate schedules. However, the bid formats’ spreadsheets typically take more refined pricing and reduce it to a single price that then can be used to determine the winning bids.
Many sealed-bid auctions use bid formats that require bidders to make offers for a single tranche in each submitted bid form. If a bidder wants to make offers on multiple tranches, the bidder must submit a separate bid spreadsheet for each offer. Utilities in Connecticut use this type of bid format. Other bid formats allow bidders to make offers for multiple tranches in a single submitted bid spreadsheet. For example, the utilities in Maryland and the District of Columbia, as well as PPL Electric, use bid formats that provide for multi-tranche offers in a single bid spreadsheet.
In some limited instances, bid formats also can allow bidders to elect whether or not they’ll manage the costs associated with congestion. For example, Connecticut Light & Power and United Illuminating allow bidders to decide whether their offer pricing places congestion risk with the bidder or with the utility’s customers.10 NSTAR also allows bidders to submit offers that will place the risk of wholesale market charges associated with reliability costs on NSTAR’s customers as opposed to paying suppliers to manage this risk. Bid formats are designed to take into account these preferences of bidders separate from their offer pricing structure.
Although bid formats often appear similar (even though offer pricing may be more or less refined), subtle differences can affect the offers that suppliers submit. The utilities in Maryland and the District of Columbia request that a bidder provide a price per MWh and the number of tranches the bidder is willing to supply. The utility can take any number of tranches up to the number indicated by the bidder at the single price the bidder identifies. Here the bidder is unable to condition the utility’s acceptance of the bid on taking supply of all the tranches indicated. For example, suppose a bidder bids to supply three tranches at $85 per MWh; in this case, the utility can take one, two, or three tranches from this bidder at $85 per MWh.
In contrast, PPL Electric and United Illuminating allow bidders to make multi-tranche offers in which there’s some ability for the bidder to make the provision of one tranche contingent on the utility accepting other tranches.11 In the case of PPL Electric, bidders are required to submit an offer schedule that indicates the prices and number of tranches they are willing to supply. The offer schedule indicates separate prices for each different number of tranches the bidder offers to supply. Bidders offering more than one tranche must provide separate prices for each number of tranches (from a single tranche to the total number of tranches that the bidder is willing to provide). For example, a bidder offering a maximum of seven tranches must provide separate prices for supplying one, two, three, four, five, and six as well as seven tranches. In the case of United Illuminating, bidders can condition the acceptance of an offer for a particular tranche on the acceptance of different tranches. For example, a bidder can provide different prices for a given tranche depending on whether the utility also purchases a separate indicated tranche or tranches. That is, tranches can be offered separately at individual prices or offered at different prices where the prices are only valid if a given tranche is purchased with another particular tranche.
One variation on sealed-bid auctions that allows bidders to condition the provision of one tranche on the acceptance of other tranches can be found in Connecticut Light & Power’s auction design. Connecticut Light & Power allows bidders to enter bids for tranches and specify the contract terms the bidders would like the utility to consider.12 For example, if Connecticut Light & Power is evaluating offers for a six-month term (e.g., the second half of 2008), and three future one-year terms (e.g., 2009, 2010, and 2011), the utility allows bidders to make offers for any combination of these terms on a per-tranche basis through the submission of separate offers. This structure allows bidders to link the prices for a given tranche across the various time periods. However, this conditioning does not link across offers for different tranches.
Auction designs that allow bidders to condition their offers provide pricing flexibility for bidders. Utilities can benefit from this flexibility because it provides them with additional pricing information and can improve the efficiency of the auction’s results.
• Auction Clearing Methodology: The methods of determining the winning bids, and thus the resulting prices, in these sealed-bid auctions vary considerably. Auction-clearing methodologies range from a simple process of finding the lowest price per tranche to an evaluation that takes into account price and non-price terms. In some instances, the auction clearing methodology is very transparent and easy to understand, while in other instances there’s a degree of judgment exercised in selecting the winning bids.
Generally, the auction clearing methodology focuses on using the prices that bidders offer (via confidential transmission of bid sheets, i.e., bids on Excel spreadsheets, to the auctioneer at a pre-specified time) as the primary input for determining winning bidders. Bidders typically are required to pre-qualify in order to submit their bids. Thus, the primary task of the auctioneer is to assemble the bidders’ offers into a format where they easily can be evaluated. In all of these sealed-bid auctions, the bid sheets appear to be structured to facilitate the evaluation of offers and “clear the market.”
In instances where the bid format allows (or requires) prices to be submitted with seasonal and daily variation by customer grouping or rate schedule, the bid sheets are designed to take these prices and convert them into a single price per tranche. Converting prices typically requires the use of various weighting factors that are based on projections of the amount of electrical energy that will be supplied to each customer grouping.13 Once bids have been converted into single prices per tranche (or tranches) the auction winners can be identified.
Determining the winning bids in these sealed-bid auctions depends on the way the bids are formatted. The utilities in Washington, D.C., and Maryland use a straightforward methodology in which the prices that bidders have offered are ranked from lowest to highest. The winning bids are the lowest-priced bids necessary to provide the supply being solicited. Because these utilities use sealed-bid auctions that specify that bidders must be willing to sell any whole number of tranches up to the total number offered at the single price the bidder offers, the utilities can buy, for example, two tranches of a four-tranche offer, and acquire the exact number of tranches being solicited.14 In other auctions the analysis necessary to clear the market is not so straightforward.
When bid formats allow for bidders to condition their offers, the auction clearing methodology can become more complicated. For example, Connecticut Light & Power and United Illuminating use bid formats in which bidders can submit offers with, and without, bearing congestion costs. This bid format gives the bidder flexibility, but requires that the auctioneer evaluate more information. In addition, determining the winning bids becomes more complex because these utilities solicit offers for several products with overlapping terms and subsequently decide which products to purchase at any particular time. In the case of United Illuminating, bidders can condition an offer for one tranche of a product with a particular term on the purchase of another offer for a separate tranche of a different product with a different term. Such flexibility creates more dimensions along which the auctioneer must evaluate bids. When the bid format allows bidders to condition their offers, the auction clearing methodology must consider all the possible combinations of offers that can be purchased to meet the utility’s needs to determine what combination best suits the utility.
The auction clearing methodology can involve a certain level of subjectivity. For example, Connecticut Light & Power and United Illuminating reserve the right not to purchase all of the supplies they solicit. Thus, when determining the least costly offers, the companies also decide how much supply to purchase. Deciding how much supply to purchase within the context of the sealed-bid auction allows the utilities to determine whether the current solicitation is the best time to make the purchase or whether purchasing in the future would be more beneficial. As a result of the auction clearing methodology that these utilities employ, the decision-making process surrounding the procurement of power is less transparent to the public.
Other auction clearing methodologies more explicitly incorporate non-price criteria. For example, the Maine PUC asks potential bidders to first submit indicative bids that provide the commission with initial information as to bidder’s pricing and potential exceptions to contractual terms and conditions. The Maine PUC then will meet with bidders and negotiate non-price terms and conditions with the goal of learning whether particular terms and conditions push up prices.15 The Maine PUC then may adjust particular terms and conditions after these negotiations. A short time thereafter, bidders submit final bids.
To determine the winning bids, the Maine PUC evaluates which offers provide the greatest value to customers considering the prices that bidders offered. However, this process can be complicated by the fact that bidders are allowed concomitantly to offer to purchase power entitlements from the very utilities they are offering to supply. Thus, in Maine, identifying the winning bidders may require considering the linkage between the offer to supply power and the offer to purchase power entitlements. The Maine PUC provides procurement reports that provide some insight into how this process works.
The pricing structure that bidders can offer is dictated by the design of each jurisdiction’s sealed-bid auction. In most jurisdictions, the pricing structure prevents suppliers from making conditional bids. A conditional bid is “a bid that is valid only if a specified set of requirements are met, such as package, minimum delivery, or payment require-ment.”16 Pricing structures that allow suppliers to make conditional bids can increase the efficiency of the auction.
The efficiency that can be gained by allowing bidders to make conditional bids can be illustrated by comparing the pricing structure under RFPs used by utilities in Maryland with the price structure under PPL Electric’s RFP.17 In Maryland, bidders are asked to submit the maximum number of tranches the bidder is willing to provide and a single price per tranche at which it’s willing provide any number of tranches up to that maximum. For example, a bidder in Maryland might bid to provide a maximum of five tranches for $85 per MWh. The electric utility can purchase any number of tranches—up to five tranches—from this bidder at a price of $85 per MWh. Under the Maryland RFPs, the bidder is unable to condition the price at which it’s willing to provide the first tranche, second tranche, etc.
Thus, with Maryland RFPs, the bidder in this example could issue five separate bids for one tranche each. In this case, the electric utility incrementally would pay for each tranche that it wants to purchase starting with the least expensive. If the first tranche is more costly for the bidder to provide than the second tranche,18 there’s no way for the bidder to reflect this cost difference in its bid. Thus, such a bidder is forced to increase the price at which it’s willing to provide the second tranche above its marginal cost to be sure to cover its marginal cost of supplying the first tranche (see Figure 1).19 Basic economics tells us that pricing above marginal cost can lead to inefficiencies.20
In contrast to the utilities in Maryland, PPL Electric’s RFP explicitly allows a bidder to submit a bid that is conditional on the number of tranches that the bidder is awarded. In PPL Electric’s RFP, bidders are asked to submit prices for each total number of tranches they’re willing to provide. For example, a supplier in this RFP might be willing to provide a maximum of five tranches. This bidder would have to provide five separate prices, i.e., a price for providing one tranche, two tranches, three tranches, four tranches, and five tranches. In this manner, the bidder’s bid is conditional on the number of tranches it’s awarded. If the first tranche is more costly for the bidder to provide than the second tranche, the bidder can reflect this cost difference in its bid by lowering the price of providing two tranches relative to one tranche. Such a bidder can offer prices at which it’s willing to provide each total number of tranches equal to its marginal cost (see Figure 2).
The type of conditional bidding allowed under PPL Electric’s RFP clearly can increase efficiency compared to Maryland RFPs, but it requires a different auction clearing mechanism. With conditional bidding, determining the winning set of bids becomes more complex and, thus, identifying the winning bidders requires the auction manager to use a different solution technique. For the Maryland utilities, the process of identifying the winning bids is rather simple and straightforward. Bids are ranked from least expensive to most expensive. The utility purchases the least expensive tranches that fulfill the supply it was seeking to obtain.
For example, if the utility seeks to buy a total of five tranches, it purchases the five least expensive tranches (see Table 2). For PPL Electric, the process of identifying the winning bids requires considering which combination of bids yields the desired number of tranches at the least cost. This process requires more computation than the simple ranking of bids necessary in Maryland because the cost of each combination must be assessed. For example, if the utility seeks to buy a total of five tranches, it must consider all combinations of five tranches that can be obtained from the available bidders’ bids in order to determine the unique set of low-cost suppliers (see Table 3).
Conditional bidding gives rise to the need for a different solution technique like that used in PPL Electric’s RFP. Suppose the electric utility wants to buy a total of 10 tranches and that each bidder is willing to provide a maximum of five tranches. Under the pricing structure in Maryland’s RFP, the utility only needs to rank each bidder’s bid from least expensive to most expensive and identify the two least-costly bidders. The complexity of this problem increases almost linearly with the number of bidders (see Table 4). Under the pricing structure in PPL Electric’s RFP, the utility needs to consider all combinations of bids that yield 10 tranches and identify the least-costly combination. The complexity of this problem increases much faster with the number of bidders (see Table 4).
It might appear that the increase in complexity that accompanies conditional bidding makes such pricing structures in electricity auctions unrealistic. However, this is not the case. While auctions with conditional bidding cannot necessarily be solved using a spreadsheet program like Microsoft Excel, computer software quickly can solve the combinatorial mathematics presented by conditional bidding. For example, determining the winning bidders needed to obtain 10 tranches when 15 bidders each bid to provide a maximum of eight tranches requires assessing more than 3 million combinations; the most basic computer algorithms can identify the least costly combination of bids in about 45 seconds on a common office computer. Advances in computational economics have made it possible to solve much more complicated pricing structures associated with conditional bidding.
In addition, in recent years it has become more common for auctions to be carried out over the internet. Electricity supplies are being procured using internet-based platforms in some states, and many large businesses use such systems as a means of procuring various goods and services. The submission and analysis of offers can be simplified using these internet-based systems. Moreover, the more complex computational challenges easily can be evaluated using computer systems specifically designed to process auction information.
The sealed-bid auctions used to procure electricity to supply non-shopping customers range from the simple to the more complex. However, many of these auctions use structures that constrain types of offers that bidders can make. Appropriately finding ways to make these auctions more flexible, using advances in computational economics, will yield better pricing outcomes for ratepayers. In the process, regulators also should consider implementing these auctions through internet-based platforms. It’s not difficult to improve these existing auctions by focusing on the tranche size, bid format and offer pricing structure, and the auction clearing methodology.
1. There are two notable exceptions: 1) In Maine, the Maine Public Utilities Commission (ME PUC) issues RFPs for wholesale power supply and selects suppliers as opposed to the two local utilities, Bangor Hydroelectric (BHE) and Central Maine Power (CMP), soliciting wholesale power supplies with oversight from ME PUC; and, 2) In Massachusetts, the Department of Public Utilities does not oversee wholesale power procurements by utilities if they are purchasing power products with a term of one year or less.
2. See, for example, Tierney, Susan F., and Todd Schatzki, Competitive Procurement of Retail Electricity Supply: Recent Trends in State Policies and Utility Practices, for the National Association of Regulatory Utility Commissioners, July 2008; and LaCasse, Chantale, and Thomas Wininger, “Maryland versus New Jersey: Is There a ‘Best’ Competitive Bid Process?,” The Electricity Journal, Volume 20, Issue 3, April 2007. These researchers suggest that the most important consideration for ensuring a competitive procurement is that the process be fair and transparent. The Federal Energy Regulatory Commission has also established similar guidelines for conducting successful competitive power procurements with the focus on preventing affiliate abuse [see, generally, Allegheny Energy Supply Company, LLC, 108 FERC ¶ 61,082 (2004)].
3. Numerous fair and transparent RFPs successfully have been carried out over the last several years. See, e.g., approvals of procurements issued by regulatory agencies in Connecticut, Maryland, Delaware, and Pennsylvania, at: http://depsc.delaware.gov/orders/7405.pdf (accessed Dec. 11, 2008); http://webapp.psc.state.md.us/Intranet/Casenum/CaseAction_new.cfm?CaseNumber=9064 (accessed Dec. 12, 2008); http://www.dpuc.state.ct.us/dockcurr.nsf/f7de85eded62752a8525655d005653b8/e3a3481f152dc47a852573cc007093a2?OpenDocument (accessed Dec.12, 2008).
4. Full-requirements load-following power refers to a wholesale electricity product where the buyer pays a single fixed price in exchange for the seller’s commitment to obtain all of the services necessary to meet the utility’s customer loads. In this regard, “full-requirements” typically refers to energy, capacity, ancillary services, congestion management, transmission and distribution losses, risk management, renewable energy credits, and any other services necessary to completely supply the utility’s customer load; “load-following” refers to the seller’s commitment to supply a specified percentage (i.e., “tranche”) of the utility’s hourly load.
5. The contract terms for the solicitations shown in Table 1 vary from three months to three years.
6. Non-shopping customers often are grouped by the classification of their load (e.g., residential, or small, medium or large commercial and industrial). In many instances, utilities can use rate schedules as a means of creating these customer groupings. In other instances, customers can be grouped by peak demand level; however, this may result in customers on the same rate schedules being assigned to different groupings for the purposes of establishing energy supply rates.
7. The tranche size reported in Table 1 is the approximate amount of supply that a supplier will have to provide at the time of peak load. However, suppliers ultimately are responsible for a percentage of load such that the tranche quantity varies hour to hour based on actual hourly customer load. Suppliers can estimate hour-to-hour load based on historical data and expected future market conditions.
8. Pepco Holdings Inc., Maryland SOS Public Disclosure of Information Web site (accessed Dec. 12, 2008).
9. Given that solicitations are laddered over time to prevent purchasing all of the power needed to supply customers at any one time, customer rates typically are based on the average of the prices in the winning bids across several different solicitations. Where the utilities have different prices for different rate schedules, utilities typically employ specific formulas to determine the rates from the average of the winning prices across these solicitations.
10. United Illuminating and Connecticut Light & Power also allow to varying degrees the ability for suppliers to make “contingent” offers so that suppliers can submit linked offers and also limit their sales to a specified maximum total quantity.
11. The Maine PUC allows bidders to link the sale of wholesale power to a utility to a same-term purchase of power supply from the utility (with Maine PUC oversight). In Maine, utilities have historical entitlements to power supplies (i.e., non-utility generators). These supplies are sold at the same time wholesale suppliers are solicited. The prices for both purchases and sales can be linked by the bidder. However, this approach is not the same as allowing a bidder to condition acceptance of one tranche of its supply offer on the utility’s purchase of other tranches.
12. United Illuminating’s offering structure also allows for linking of different term offers; however, each unique product term is defined as a separate tranche.
13. Utilities in Connecticut, Washington, D.C., and Maryland convert these prices on the bid sheet so that bidders can see the price per tranche they are offering when the bid sheet is completed. In contrast, utilities in Massachusetts do not provide the weighting calculation that is used by bidders.
14. There can be situations in these auctions where offers are rejected by the utility commission if the prices are above a level considered reasonable.
15. For example, there could be a particular aspect of the credit requirements or default provisions of the proposed contract that a supplier may indicate are unnecessarily costly.
16. Combinatorial Auctions, 2006, edited by Peter Cramton, Yoav Shoham, and Richard Steinberg, at 616.
17. A similar comparison can be made to United Illuminating’s auction structure instead of PPL Electric’s format.
18. Such cost structures can arise from economies of scale and scope in production and distribution technologies, as well as the supplier’s approach to managing wholesale supply and delivery.
19. Given the bid format, the price of the subsequent tranches cannot decline, but either remain the same or increase.
20. See, e.g., Baumol, William J., and Alan S. Blinder, Economics: Principles and Policy, 1988, at 605-606.