Design of Wholesale Electricity Markets Hung-po Chao and Robert Wilson* Electric Power Research Institute Draft 990101 Note: when viewing or editing this draft document (cid:143) Turn on <Tools/Track_Changes/Highlight> so that additions and corrections can be tracked. (cid:143) Use <Insert/Comment> to insert remarks and questions. * [email protected] and [email protected] Chao and Wilson, Draft 2 Design of Wholesale Electricity Markets Hung-po Chao and Robert Wilson Contents An Overview of Wholesale Electricity Markets......................................................4 Introduction......................................................................................................................7 A Short History of Restructuring...................................................................................10 Developments Abroad................................................................................................10 Developments in the U.S............................................................................................11 Developments in Regulatory Policy...........................................................................13 Radical Designs..............................................................................................................13 Pools, Exchanges, and Bilateral Markets............................................................14 Contract and Market Solutions to Operational Problems..............................................18 Centralized Systems Based on Relational Contracts..................................................18 Decentralized Markets................................................................................................19 Bid Formats, Market Clearing, and Prices..........................................................21 Scope of Markets...........................................................................................................21 Procedural Rules............................................................................................................24 Other Designs.................................................................................................................26 Bids and Prices for Strips*.........................................................................................27 Competing Markets.............................................................................................28 Pricing and the Trading Process.....................................................................................29 Equilibrium Among Competing Markets......................................................................31 Market Power of Market Makers................................................................................32 Complexity of the Market Structure.....................................................................32 Transmission Management.................................................................................34 Congestion Pricing of Transmission ...................................................................38 The Adjustment Market.................................................................................................40 Derivation of Usage Charges.........................................................................................41 Summary........................................................................................................................42 Hedges and Insurance........................................................................................43 Energy............................................................................................................................43 Transmission..................................................................................................................44 Ancillary Services and Losses.......................................................................................45 Summary........................................................................................................................46 Pricing of Ancillary Services*..............................................................................46 The Process of Market Clearing and the Mode of Competition...........................48 Contract Commitments and Settlements............................................................52 Settlement Procedures.......................................................................................54 The Role of Intertemporal Operating Constraints..........................................................55 The Role of Incentives and Gaming..............................................................................56 Mitigating Factors..........................................................................................................57 Summary........................................................................................................................58 Chao and Wilson, Draft 3 Multiple Markets and Inter-Market Efficiency......................................................58 Parallel Markets.............................................................................................................58 Inter-Market Efficiency..................................................................................................59 Limitations of Time-Differentiated Prices............................................................62 Feasibility............................................................................................................64 Ancillary Services..........................................................................................................65 Transmission..................................................................................................................66 Energy............................................................................................................................67 Mitigation of Market Power.................................................................................69 Investments in Transmission Capacity................................................................72 Concluding Remarks ..........................................................................................73 Appendix A: Glossary of Acronyms ....................................................................75 Appendix B: Dictionary of Technical Terms........................................................78 Appendix C: Effects of Market Structure on Reliability Management..................85 Appendix D: Comparison with Gas and Pipeline Markets...................................88 What’s Different About Gas and Pipelines....................................................................88 The Transmission Markets.............................................................................................89 The Effects of Monopoly Power....................................................................................91 Implications for Restructuring Electricity Markets.......................................................92 Appendix E: Electricity Restructuring Facts & Figures........................................94 United States..................................................................................................................94 California....................................................................................................................94 New England (ISO-NE)..............................................................................................95 New York....................................................................................................................97 Pennsylvania—New Jersey—Maryland (PJM).........................................................98 Other States and some General Resources...............................................................100 Other Countries............................................................................................................101 Argentina..................................................................................................................101 Australia....................................................................................................................103 Brazil........................................................................................................................105 Canada--Alberta........................................................................................................105 Canada--Ontario.......................................................................................................107 New Zealand.............................................................................................................108 Scandinavia (NordPool)..........................................................................................108 Spain.........................................................................................................................109 United Kingdom.......................................................................................................111 Appendix F: Bibliography*.................................................................................113 • An asterisk * indicates an unfinished section. • Appendix E needs to add: (a) Chile, (b) new Amsterdam Bourse for electricity trading, (c) implications of new European Common Market requirements for each country to deregulate/restructure its infra-structure markets, (d) expansion of NordPool to include Denmark and Finland, (e) dramatic new developments in UK, see July 1998 reports by Offer and DTI. Chao and Wilson, Draft 4 Design of Wholesale Electricity Markets Hung-po Chao and Robert Wilson An Overview of Wholesale Electricity Markets The purpose of this chapter is to provide a functional description of the tasks accomplished by wholesale electricity markets. These markets are organized differently in various jurisdictions, and procedural rules tend to differ markedly even between systems that are decentralized to the same degree, but the basic tasks remain the same. The description here supposes a decentralized design in order to isolate the separate components. To simplify we focus mainly on the supply side, with the understanding that many tasks have analogs on the demand side; e.g., the analog of incremental generation is demand curtailment. We omit the longer-term issues of planning investments in new generation and transmission capacity. The technology of electricity generation and transmission requires that the primary task is physical control of the system. Control has three main components. (cid:143) Energy. Allocation of energy generation among suppliers and among demanders. (cid:143) Transmission. Allocation of transmission capacity among suppliers and demanders. (cid:143) Reserves. Provision of reserve capacities for generation and transmission to meet contingencies and to ensure the reliability and security of the system. The provision of reserves is often called ancillary services. Each component has an important time dimension that differentiates between forward planning and real-time operations. Forward plans distinguish between long-term and short-term, with the short-term identified by “day-ahead” plans for the next day’s operating cycle and “hour-ahead” plans for the next few hours – often 1, 2, or 4 hours. Real-time operations are conducted on a time frame that ranges from 1 hour down to 5 minutes ahead of actual dispatch, and even 1 second in the case of following variations in the load. The ultimate authority for physical control is invariably assigned to a system operator (SO) charged with responsibility for managing the transmission system. This authority includes invoking supplementary generation and reserves as needed to follow variations in the load and to maintain the stability of the transmission system. The SO’s authority is virtually complete in real-time operations, but jurisdictions assign different degrees of authority over forward planning. More latitude is allowed for decentralized markets to establish plans as the time horizon increases. In highly centralized systems the SO optimizes the allocations of generation and transmission, whereas in decentralized systems the allocation is done by markets for energy and transmission that may be managed by the SO or by independent entities. In most jurisdictions the SO acquires reserves in markets that it conducts itself, although some allow participants in energy Chao and Wilson, Draft 5 markets to self-provide reserves. Some of the SO’s operating procedures regarding scheduling, and its standards for reserves and other ingredients affecting system security and reliability, are specified by regional coordinating councils (e.g., WSCC for the western states) and the North American Electric Reliability Council (NERC). It is important to realize that the markets for transmission and reserves are both extensions of the energy market. As will be described later, excess demand for transmission is eliminated by altering the spatial pattern of generation; similarly, reserves are provided by allocating a portion of generation capacity to stand-by status or to the job of load following on a short time scale. Transmission and reserves can also be interpreted as derived demands. A scarcity of transmission capacity is revealed by initial plans for energy generation, which then requires adjustments in the spatial distribution to alleviate congestion. The quantity of reserves is usually specified as a percentage of the load, differentiated by function and time frame; e.g., • A percentage (e.g., 3.5% in California) is required as “spinning” reserve, defined as generation capacity that can be fully available (subject to limitations on ramping rates for thermal generators) within ten minutes and sustained for two hours. This is interpreted as incremental generation, but some systems also require a reserve of generation that can be decremented. • An additional percentage (again 3.5% in California) is required as “non-spinning” reserve available within thirty minutes, and a further percentage is required as “replacement” reserve that can be activated to replace spinning and non-spinning reserves as these are used up. • Additional generation capacity is required to have “black start” capabilities so that the system can be revived after a collapse. The category of ancillary services also includes two key functions called “regulation.” One is “VAR?” that provides reactive energy at key locations to maintain the phase angle of alternating current systems. The other is automatic generation control (AGC) provided by units equipped with electronic control devices that increase or decrease generation on a second-by-second basis to follow the load. The distinction between energy and transmission markets disappears as the time scale shrinks. In real-time operations the SO relies first on units assigned to VAR and AGC services to meet small variations in load, then markets for supplementary energy to meet larger variations on longer time scales, and when these are exhausted, draws on spinning and non-spinning reserves. A fully decentralized market for supplementary energy consists of offers to increment or decrement generation, called “incs” and “decs.” The SO calls on these as needed, based first on the required location, and then accepting those offered at the lowest prices. In contrast, the SO running a centralized system issues instructions for these alternations in generation patterns. On the time scale of day-ahead planning the distinctions among energy, transmission, and reserves is stronger. In a fully centralized system the SO optimizes these components in a single joint plan, but decentralized systems tend to address them in sequence. Typically there are markets for long-term bilateral contracts as well as day-ahead markets for trades of energy deliverable in each of the hours of the next day. The aggregate of these transactions identifies the derived demands for transmission and reserves. If the demand Chao and Wilson, Draft 6 exceeds the available transmission capacity (as specified by the SO) on any major line or interface then the line is said to be congested. When there is congestion the SO conducts a market to allocate the scarce transmission capacity among the users competing for access, and in any case the SO conducts auction markets in which it purchases reserves sufficient to maintain its reliability standards. As mentioned, the transmission market is an extension of the energy market in which the spatial pattern of generation is altered to alleviate congestion. Some systems do not price transmission access on a spot basis, relying instead on directives from the SO that specify which generators must increment or decrement their output (called “constrained on” or “constrained off”). Those that use “congestion pricing” of transmission establish a usage charge based on the marginal cost of alleviating congestion, either on a nodal basis specific to each location, or on a zonal basis in which only congestion between major zones is priced explicitly. In a zonal system the SO uses offered bids for increments and decrements to compute the least costly way of alleviating congestion. For example, if transmission is congested between an export zone and an import zone then the SO sets the usage charge for transmission between them as the difference between its marginal cost of incrementing generation in the import zone and its marginal revenue from decrementing generation in the export zone. (An incremental bid requires payment from the SO whereas a decremental bid requires payment from the supplier.) The SO’s revenue from usage charges is usually conveyed to the owners of the transmission assets. The market for reserves is also an extension of the energy market. For example, those generation units assigned to AGC are among those whose bids were accepted in the energy market and therefore are scheduled to operate during the next day. Spinning and non-spinning reserves are typically provided by units whose bids accepted in the energy market did not fully exhaust their productive capacity. Some sources, such as hydro and fast-start turbines, can provide spinning reserves even if not scheduled beforehand. The auction markets for reserves are unusual in their reliance on two bid prices, one for reserved capacity and one for actual generation when activated by the SO. The simplest scheme, say for spinning reserve, accepts bids solely on the basis of the offered price for reserving capacity, and pays for actual generation at the real-time spot price, using the offered energy bid only as a reservation price indicating the least price at which the owner wants to be called to generate. The forward markets for energy and transmission are best interpreted as financial markets. The physical aspect is important, because the planned scheduling of units is based on the transactions in these markets, but in practice the physical commitment of resources is indicative rather than binding. In fact some systems settle all transactions at the real-time price, and even in those that settle day-ahead trades at the day-ahead prices it is still true that a supplier can deviate from its day-ahead schedule by paying the hour- ahead or real-time energy price for the deviation. The purely financial aspect is especially strong in the case of bilateral contracts, since often these are structured as “contracts for differences” in which the parties insure each other against differences between the real-time price and their agreed-on price, and physical differences are settled at the real-time price. Chao and Wilson, Draft 7 In some jurisdictions, such as Alberta and the U.K., participation in the market conducted by the SO is mandatory. Even so, bilateral trades can be accomplished privately via contracts for differences from the SO’s price. In thoroughly decentralized markets the only mandatory aspect is the requirement that the SO is advised of schedules. Energy trading takes many forms, ranging from bilateral contracting mediated by brokers and dealers, to exchanges that establish clearing prices to match supply and demand. Parallel markets compete for customers based on the comparative advantages of different contracts and price-formation procedures. Those exchanges organized as public-benefit corporations are limited by their charters to market-clearing, whereas other market makers are private profit-making enterprises with considerable latitude to Competition among market makers: Supplementary markets: FTRs, hedges, options, ETCs, entitlements [what next?] Anything on governance, regulation (PUC, FERC), market power The Introduction lays out some background and issues that motivate the subsequent discussion. The following sections consider the general architecture of wholesale markets for electricity. The first examines the choice among forms of organization, such as bilateral contracting or multilateral trading, and in the latter, the choice between a market-clearing exchange or a tight pool with centrally optimized scheduling. The second examines the transmission market in some detail, and the third examines the energy market similarly. The final two sections examine linkages among multiple markets in decentralized designs, focusing on the role of contractual commitments and the requirements for inter-market efficiency. Introduction To establish a point of departure: the current restructuring of electricity markets is consistent with the analysis by Joskow and Schmalensee in Markets for Power, 1983. They foresaw competitive markets for generation, transmission facilities operated on an open-access nondiscriminatory common-carrier basis, and retail competition among power marketers that rely on regulated utility distribution companies for delivery. Regulation of the wholesale and retail energy markets would be reduced to structural requirements and operational guidelines and monitoring, while retaining substantial regulation of the “wires” markets for transmission and distribution. These changes entail unbundling energy from T&D, thereby reversing the vertical integration of utilities. The current issues that we address here concern mainly the organization of the wholesale markets for energy and transmission, interpreted as including ancillary services and other requirements for system reliability and security. The examination of these issues in U.S. jurisdictions can benefit from the history of restructuring in the Canadian provinces, such as Alberta and Ontario, other countries such as Britain, Australia, New Zealand, and Norway, newly implemented designs in countries such as Spain, and current developments in several states in the U.S that have already implemented new market designs.1 1 Two useful surveys are those prepared by Putnam, Hayes and Bartlett for the Ontario Market Design Committee, and by London Economics for the California Trust for Power Industry Restructuring. Chao and Wilson, Draft 8 The peculiar features of the electricity industry that must be considered include temporal and stochastic variability of demands and supplies, accentuated by the non-storability of power, multiple technologies with varying sensitivities to capital and fuel costs and environmental and siting restrictions, and dependence on a reliable and secure transmission system. The economic problems include substantial non-convexities (immobility of generation and transmission facilities, scale economies in generation, non- linearities in transmission), and externalities (mainly in transmission). As regards generation these problems have eased sufficiently in recent decades to enable competitive energy markets, but they remain important considerations in designing these markets. The criteria for selecting among market designs include efficiency over the long term, including incentives for investment in facilities for generation and transmission. However, our exposition focuses on short-term efficiency, since this is the immediate concrete problem, and it is required for long-term efficiency. We emphasize the implications of the general principles of market design based on ideas from economics and game theory. To motivate the subsequent sections, we describe three parts of the overall problem of market design. The basic design choice is the architecture of the market. There are many contending options.2 The market can be centralized or decentralized; it can be based on bilateral contracting, a centralized exchange, or a tightly controlled pool; trades can be physical or financial obligations, and they can be forward or spot contracts; the market can include financial hedges or not; the “official” market can be mandatory or optional, and encourage or discourage secondary markets. As will be evident, our opinion is that on most dimensions, the purported advantage of one extreme or the other is illusory. We favor designs that mix the two extremes to capture some of the advantages of each from parallel operations. For instance, for the three time frames of long-term, day-ahead, and real-time, there are corresponding advantages from bilateral contracting, a central exchange, and tightly controlled dispatch. After the market architecture is established, a host of details must be specified. We do not address operational aspects here. Procedural rules must be constructed carefully to suppress gaming and promote efficiency. It is not only a matter of closing all loopholes; rather, the procedural rules must solve some basic economic problems, such as effective price discovery that enables more efficient decisions by suppliers. All this presupposes that the market will be sufficiently competitive to produce an efficient outcome, so if not, then further measures are required to diminish the market power of dominant incumbents and to promote entry by newcomers. The fact that we focus on the market architecture as the basic structural decision does not mean that it should be decided first. Parallel consideration of several designs and their implementation is useful in the early stages so that their merits can be compared in light of stakeholders’ interests. 2 A balanced and useful view of these options is presented in Issue Paper 3 of the Ontario Market Design Committee, March 1998. Chao and Wilson, Draft 9 Our perspective is conditioned by an emphasis on strategic behavior. This seems paradoxical, since the aim is to construct a design that suppresses gaming or renders it ineffective in favor of greater efficiency. The principle, however, is to treat the market design as establishing a mode of competition among the traders. The key is to select a mode of competition that is most effective in realizing the potential gains from trade. To illustrate, we describe a common fallacy. It is deceptively easy to conclude that a mandatory pool based on a centralized optimization of all generation, transmission, ancillary services, etc. – as in the U.K. or PJM – can realize the full productive potential of the system. This view does not recognize that the schedules derived from an optimization program are no better than its inputs. In fact, suppliers can and do treat the program as a device whose outputs can be manipulated by the inputs they provide in the form of purported cost functions, availabilities, etc.3 Thus, the mode of competition consists of contending efforts to influence the “bottom-line” results from the program, such as dispatched quantities and prices for energy, transmission, and ancillary services. In terms of economic theory: reliance on an optimization affects the form and strength of traders’ incentives at various points in the process, but it does not obviate the role of incentives. A central design problem is to identify the best locus of incentives and competitive forces. In addition to our strategic perspective, we appreciate that traders have practical motivations that are not included in standard economic theory. For instance, suppliers are typically skeptical of designs that make their financial viability dependent on prices derived as shadow prices on system constraints included in the formulation of an optimization program, and centrally planned operating schedules that are several steps removed from the cost data they submit. They prefer market-clearing prices derived directly from the terms they offer, and they prefer to devise their own operating schedules to fulfill offers accepted in the market. Similarly, they are leery of intrusions by the transmission system operator (SO) into the energy markets, fearing that the SO’s extraordinary powers could bias the competitive process. We see two sources of these preferences. One is informational: submitted cost data is never sufficient to describe the full range of considerations relevant to a supplier. The other pertains to governance: the SO is usually described as the ISO, emphasizing its independence and adherence to operating standards derived from principles of power engineering, but few designs address the basic problem of incentives for the SO. For example, the SO is not liable for the financial consequences to traders of strict security standards that are motivated more by avoidance of any chance of mishap than an economic tradeoff between reliability and energy costs. Current designs rely on standards of transmission management inherited from the era when it was internalized within utilities who owned and operated transmission facilities for their native loads, but as this inheritance decays it will be useful to re-examine the issues of governance and incentives for the SO. 3 Expositions that address these issues include Mark Armstrong, Simon Cowan, and John Vickers, Regulatory Reform: Economic Analysis and British Experience (MIT Press, 1994, Chapter 9); Michael Einhorn (ed.), From Regulation to Competition: New Frontiers in Electricity Markets (Kluwer, 1994, Chapters 2-7); and Nils-Henrik von der Fehr and David Harbord, "Competition in Electricity Spot Markets: Economic Theory and International Experience" (ISBN 82-570-9166-9, Economics Dept., University of Oslo, Norway, January 1998). Chao and Wilson, Draft 10 A Short History of Restructuring As late as 1978 when the Public Utilities Regulatory Policies Act (PURPA) was passed the central role of regulated utilities was taken for granted. Even though this legislation enabled entry of independent power producers (IPP) and encouraged co-generation and renewable energy sources by imposing obligations on utilities to procure supplies from qualified facilities (QFs), it also assumed continuation of the utilities’ vertical integration and monopoly franchises. Besides the evident motivations for centralized control of transmission and distribution, generation was assumed to be effectively a natural monopoly because at the time the efficient scale of a coal-fired generator, the economical technology at the time, was on the order of 1000 MW. Virtually no attention was given to retail competition as a viable option, even though a major effect of the Act was to enable major industrial customers to bypass their local utilities by purchasing from IPPs and especially from co-generators. Five years later, Joskow and Schmalensee’s book, Markets for Power, envisioned competitive markets for electricity, both wholesale and retail, by separating these businesses from the “wires” businesses of transmission and distribution. They foresaw presciently that the era in which scale economics justified utility monopolies in generation was passing. In the ensuing years the minimum scale of gas-fired plants declined to approximately 200 MW and efficiency increased greatly, while natural gas prices declined and wellhead prices were deregulated. From an economic viewpoint, the justification for vertical integration disappeared. On the other hand, progress on designs of efficient markets for electricity proceeded slowly and fitfully. The motive for creation of market mechanisms was strongest in those jurisdictions with high retail prices due to utilities’ heavy obligations for payment of interest and principle on debt incurred in the 1970s for large nuclear and coal-fired plants, and purchases of energy from QFs under long-term contracts at prices above subsequent market prices. In California in particular the wholesale price at the California-Oregon border was usually under $20/MWh in a period when retail prices averaged $120 and even industrial sales exceeded $65. Developments Abroad The first changes occurred elsewhere, initially in Chile and Argentina [?], and most importantly in 1989 when the U.K. implemented an energy market for its England-Wales system. These privatizations of government enterprises were modeled on extensions of existing power pools, similar to those already in operation in various U.S. jurisdictions such as New England (NEPool), NY, and PJM. Their central features were a day-ahead optimization of unit commitments and scheduling, subject to system constraints such as transmission (although transmission was not priced on a congestion basis), based on detailed cost data submitted daily, and a uniform price for energy transactions. The U.K. design was imitated in Alberta, but not in others initiated soon after in Norway (NordPool) and Victoria (VicPool). Norway’s system had evolved slowly over two decades. Because generation came almost entirely from hydro sources there were no deep concerns about unit commitments and scheduling, nor ancillary services, so the focus was on transmission: the important