WHITE PAPER Achieving Decentralized Coordination In the Electric Power Industry Using dynamic pricing and digital technology to move from centralized control to decentralized coordination. www.UtilitiesProject.com/10827 F or the past century, the dominant business and regulatory paradigms in the electric power industry have been centralized economic and physical control. The ideas presented here and in my forthcoming book, Deregulation, Innovation, and Market Liberalization: Electricity Restructuring in a Constantly Evolving Environment (Routledge, 2008), comprise a different paradigm – decen tralized economic and physical coordi nation – which will be achieved through contracts, transactions, price signals and integrated intertemporal wholesale and retail markets. Digital communica tion technologies – which are becom ing ever more pervasive and afford able – are what make this decentralized coordination possible. In contrast to the “distributed control” concept often invoked by power systems engineers (in which distributed technology is used to enhance centralized control of a system), “decentralized coordination” repre sents a paradigm in which distributed agents themselves control part of the system, and in aggregate, their actions produce order: emergent order. [1] Dynamic retail pricing, retail product differentiation and complementary end use technologies provide the foundation for achieving decentralized coordination in the electric power industry. They bring timely information to consumers and enable them to participate in retail market processes; they also enable retailers to discover and satisfy the heterogeneous preferences of consumers, all of whom have private knowledge that’s unavailable to firms and regulators in the absence of such market processes. Institutions that facilitate this discovery through dynamic pricing and technology are crucial for achieving decentralized coordination. Thus, retail restructuring that allows dynamic pricing and product differentia tion, doesn’t stifle the adoption of digital technology and reduces retail entry bar riers is necessary if this value-creating decentralized coordination is to happen. This paper presents a case study – the “GridWise Olympic Peninsula Testbed Demonstration Project” – that illustrates how digital end-use technology and dynamic pricing combine to provide value to residential customers while increasing network reliability and reducing required infrastructure investments through decentralized coordination. The availabil ity (and increasing cost-effectiveness) of digital technologies enabling consum ers to monitor and control their energy use and to see transparent price signals has made existing retail rate regulation WRITTEN BY Lynne Kiesling, Northwestern University Lynne Kiesling is a senior lecturer in the Department of Economics and the Kellogg School of Man agement at Northwestern University, a faculty member at the Northwestern Institute on Complex Systems and a faculty affiliate at the Center for the Study of Industrial Organization. She is also a member of the GridWise Architecture Council and has written or co-written numerous academic journal articles and policy studies. obsolete. Instead, the policy recommen dation that this analysis implies is that regulators should reduce entry barriers in retail markets and allow for dynamic pricing and product differentiation, which are the keys to achieving decentralized coordination. THE KEYS: DYNAMIC PRICING, DIGITAL TECHNOLOGY Dynamic pricing provides price signals that reflect variations in the actual costs and benefits of providing electricity at different times of the day. Some of the more sophisticated forms of dynamic pric ing harness the dramatic improvements in information technology of the past 20 years to communicate these price signals to consumers. These same technological developments also give consumers a tool for managing their energy use, in either manual or automated form. Currently, with almost all U.S. consumers (even industrial and commercial ones) paying average prices, there’s little incentive for consumers to manage their consumption and shift it away from peak hours. This inelastic demand leads to more capital investment in power plants and transmis sion and distribution facilities than would occur if consumers could make choices based on their preferences and in the face of dynamic pricing. Retail price regulation stifles the eco nomic processes that lead to both static and dynamic efficiency. Keeping retail prices fixed truncates the information flow between wholesale and retail markets, and leads to inefficiency, price spikes and price volatility. Fixed retail rates for elec tric power service mean that the prices www.UtilitiesProject.com U8_Ch2_final.indd 79 p79 5/2/08 9:12:21 AM CHAPTER 2 OPERATIONS individual consumers pay bear little or no relation to the marginal cost of provid ing power in any given hour. Moreover, because retail prices don’t fluctuate, con sumers are given no incentive to change their consumption as the marginal cost of producing electricity changes. This severing of incentives leads to inefficient energy consumption in the short run and also causes inappropriate investment in generation, transmission and distribu tion capacity in the long run. It has also stifled the implementation of technologies that enable customers to make active consumption decisions, even though com munication technologies have become ubiquitous, affordable and user-friendly. Dynamic pricing can include time-of-use (TOU) rates, which are different prices in blocks over a day (based on expected wholesale prices), or real-time pricing (RTP) in which actual market prices are transmitted to consumers, generally in increments of an hour or less. A TOU rate The 2007 Galvin Electricity Initiative report “The Path to Perfect Power: New Technologies Advance Consumer Control” catalogs a variety of end-user technolo gies (from price-responsive appliances to wireless home automation systems) that can communicate electricity price signals to consumers, retain data on their con sumption and be programmed to respond automatically to trigger prices that the consumer chooses based on his or her preferences. [2] Moreover, the two-way communication advanced metering infra structure (AMI) that enables a retailer and consumer to have that data transparency is also proliferating (albeit slowly) and declining in price. Dynamic pricing and the digital tech nology that enables communication of price information are symbiotic. Dynamic pricing in the absence of enabling technol ogy is meaningless. Likewise, technology without economic signals to respond to is extremely limited in its ability to coor ances that can be programmed to behave differently based on changes in the retail price of electricity, these products and services provide customers with an oppor tunity to make better choices with more precision than ever before. In aggregate, these choices lead to better capacity utili zation and better fuel resource utilization, and provide incentives for innovation to meet customers’ needs and capture their imaginations. In this sense, technological innovation and dynamic retail electricity pricing are at the heart of decentralized coordination in the electric power network. EVIDENCE Led by the Pacific Northwest National Laboratory (PNNL), the Olympic Peninsula GridWise Testbed Project served as a dem onstration project to test a residential net work with highly distributed intelligence and market-based dynamic pricing. [4] Washington’s Olympic Peninsula is an area of great scenic beauty, with population FIXED RETAIL RATES for electric power service mean that the prices consumers pay bear little or no relation to the marginal cost of providing power in any given hour. typically applies predetermined prices to specific time periods by day and by season. RTP differs from TOU mainly because RTP exposes consumers to unexpected varia tions (positive and negative) due to demand conditions, weather and other factors. In a sense, fixed retail rates and RTP are the end points of a continuum of how much price variability the consumer sees, and different types of TOU systems are points on that continuum. Thus, RTP is but one example of dynamic pricing. Both RTP and TOU provide better price signals to customers than current regulated aver age prices do. They also enable companies to sell, and customers to purchase, electric power service as a differentiated product. TECHNOLOGY’S ROLE IN RETAIL CHOICE Digital technologies are becoming increas ingly available to reduce the cost of send ing prices to people and their devices. p80 U8_Ch2_final.indd 80 dinate buyers and sellers in a way that optimizes network quality and resource use. [3] The combination of dynamic pric ing and enabling technology changes the value proposition for the consumer from “I flip the switch, and the light comes on” to a more diverse and consumer-focused set of value-added services. These diverse value-added services empower consumers and enable them to control their electricity choices with more granularity and precision than the envi ronment in which they think solely of the total amount of electricity they consume. Digital metering and end-user devices also decrease transaction costs between buyers and sellers, lowering barriers to exchange and to the formation of particu lar markets and products. Whether they take the form of building control systems that enable the consumer to see the amount of power used by each function performed in a building or appli centers concentrated on the northern edge. The peninsula’s electricity distribu tion network is connected to the rest of the network through a single distribution substation. While the peninsula is experi encing economic growth and associated growth in electricity demand, the natural beauty of the area and other environ mental concerns served as an impetus for area residents to explore options beyond simply building generation capacity on the peninsula or adding transmission capacity. Thus, this project tested how the com bination of enabling technologies and market-based dynamic pricing affected utilization of existing capacity, deferral of capital investment and the ability of distributed demand-side and supply-side resources to create system reliability. Two questions were of primary interest: 1) What dynamic pricing contracts do consumers find attractive, and how does enabling technology affect that choice? Innovating for the Future 5/2/08 9:12:23 AM WHITE PAPER 2) To what extent will consumers choose to automate energy use decisions? The project – which ran from April 2006 through March 2007 – included 130 broadband-enabled households with electric heating. Each household received a programmable communicating ther mostat (PCT) with a visual user interface that allowed the consumer to program the thermostat for the home – specifically to respond to price signals, if desired. Households also received water heaters equipped with a GridFriendly appliance (GFA) controller chip developed at PNNL that enables the water heater to receive price signals and be programmed to respond automatically to those price sig nals. Consumers could control the sensi tivity of the water heater through the PCT settings. These households also participated in a market field experiment involving dynamic pricing. While they continued to purchase energy from their local utility at a fixed, discounted price, they also received a cash account with a predetermined bal ance, which was replenished quarterly. The energy use decisions they made would determine their overall bill, which was deducted from their cash account, and they were able to keep any difference as profit. The worst a household could do was a zero balance, so they were no worse off than if they had not participated in the experiment. At any time customers could log in to a secure website to see their cur rent balances and determine the effective ness of their energy use strategies. On signing up for the project, the households received extensive informa tion and education about the technologies available to them and the kinds of energy use strategies facilitated by these technol ogies. They were then asked to choose a retail pricing contract from three options: a fixed price contract (with an embed ded price risk premium), a TOU contract with a variable critical peak price (CPP) component that could be called in periods of tight capacity or an RTP contract that would reflect a wholesale market-clearing price in five-minute intervals. The RTP was determined using a uniform price double auction in which buyers (households and commercial) submit bids and sellers sub mit offers simultaneously. This project represented the first instance in which a double auction retail market design was tested in electric power. The households ranked the contracts and were then divided fairly evenly among the three types, along with a control group that received the enabling technologies and had their energy use monitored but did not participate in the dynamic pric ing market experiment. All households received either their first or second choice; interestingly, more than two-thirds of the households ranked RTP as their first choice. This result counters the received wisdom that residential customers want only reliable service at low, stable prices. According to the 2007 report on the project by D.J. Hammerstrom (and others), on average participants saved 10 percent on their electricity bills. [5] That report also includes the following findings about the project: Result 1. For the RTP group, peak con sumption decreased by 15 to 17 percent relative to what the peak would have been in the absence of the dynamic pric ing – even though their overall energy consumption increased by approximately 4 percent. This flattening of the load dura tion curve indicates shifting some peak demand to nonpeak hours. Such shifting increases the system’s load factor, improv ing capacity utilization and reducing the need to invest in additional capacity, for a given level of demand. A 15 to 17 percent reduction is substantial and is similar in magnitude to the reductions seen in other dynamic pricing pilots. After controlling for price response, weather effects and weekend days, the RTP group’s overall energy consumption was 4 percent higher than that of the fixed price group. This result, in combination with the load duration effect noted above, indicates that the overall effect of RTP dynamic pricing is to smooth consumption over time, not decrease it. Result 2. The TOU group achieved both a large price elasticity of demand (-0.17), based on hourly data, and an overall energy reduction of approximately 20 per cent relative to the fixed price group. After controlling for price response, weather effects and weekend days, the TOU group’s overall energy con sumption was 20 percent lower than that of the fixed price group. This result indicates that the TOU (with occasional critical peaks) pricing induced overall conservation – a result consistent with the results of the California SPP proj ect. The estimated price elasticity of demand in the TOU group was -0.17, which is high relative to that observed in other projects. This elasticity suggests that the pricing coupled with the enabl ing end-use technology amplifies the price responsiveness of even small resi dential consumers. Despite these results, dynamic pricing and enabling technologies are proliferat ing slowly in the electricity industry. Proliferation requires a combination of formal and informal institutional change to overcome a variety of barriers. And while formal institutional change (pri marily in the form of federal legislation) is reducing some of these barriers, it remains an incremental process. The traditional rate structure, fixed by state regulation and slow to change, presents a substantial barrier. Predetermined load profiles inhibit market-based pricing by ignoring individual customer variation and the information that customers can communicate through choices in response to price signals. Furthermore, the per sistence of standard offer service at a discounted rate (that is, a rate that does not reflect the financial cost of insurance against price risk) stifles any incentive customers might have to pursue other pricing options. The most significant – yet also most intangible and difficult-to-overcome – obstacle to dynamic pricing and enabling technologies is inertia. All of the primary stakeholders in the industry – utilities, regulators and customers – harbor status quo bias. Incumbent utilities face incen tives to maintain the regulated status quo as much as possible (given the economic, technological and demographic changes surrounding them) – and thus far, they’ve been successful in using the political pro cess to achieve this objective. www.UtilitiesProject.com U8_Ch2_final.indd 81 p81 5/2/08 9:12:23 AM CHAPTER 2 OPERATIONS Customer inertia also runs deep because consumers have not had to think about their consumption of electricity or the price they pay for it – a bias consumer advocates generally reinforce by arguing that low, stable prices for highly reliable power are an entitlement. Regulators and customers value the stability and predict ability that have arisen from this vertically integrated, historically supply-oriented and reliability-focused environment; however, what is unseen and unaccounted for is the opportunity cost of such predictability – the foregone value creation in innovative services, empowerment of customers to manage their own energy use and use of including consumer surplus? In aggregate, do these distributed actions create emer gent order in the form of system reliability? Answering these questions requires thinking about the diffuse and private nature of the knowledge embedded in the network, and the extent to which such a network becomes a complex adaptive sys tem. Technology helps determine whether decentralized coordination and emergent order are possible; the dramatic transfor mation of digital technology in the past few decades has decreased transaction costs and increased the extent of feasible decentralized coordination in this industry. Institutions – which structure and shape ENDNOTES 1. Order can take many forms in a complex sys tem like electricity – for example, keeping the lights on (short-term reliability), achieving economic efficiency, optimizing transmission congestion, longer-term resource adequacy and so on. 2. Roger W. Gale, Jean-Louis Poirier, Lynne Kiesling and David Bodde, “The Path to Perfect Power: New Technologies Advance Consumer Control,” Galvin Electricity Initiative report (2007). www.galvinpower.org/resources/ galvin.php?id=88 3. The exception to this claim is the TOU con tract, where the rate structure is known in advance. However, even on such a simple DYNAMIC PRICING in the absence of enabling technology is meaningless. Likewise, technology without economic signals to respond to is extremely limited in its ability to coordinate buyers and sellers in a way that optimizes net work quality and resource use. double-sided markets to enhance market efficiency and network reliability. Compare this unseen potential with the value cre ation in telecommunications, where even young adults can understand and adapt to cell phone-pricing plans and benefit from the stream of innovations in the industry. CONCLUSION The potential for a highly distributed, decentralized network of devices auto mated to respond to price signals creates new policy and research questions. Do individuals automate sending prices to devices? If so, do they adjust settings, and how? Does the combination of price effects and innovation increase total surplus, p82 U8_Ch2_final.indd 82 the contexts in which such processes occur – provide a means for creating this coordination. And finally, regulatory insti tutions affect whether or not this coordi nation can occur. For this reason, effective regulation should focus not on allocation but rather on decentralized coordination and how to bring it about. This in turn means a focus on market processes, which are adaptive institutions that evolve along with tech nological change. Regulatory institutions should also be adaptive, and policymakers should view regulatory policy as work in progress so that the institutions can adapt to unknown and changing conditions and enable decentralized coordination. ■ dynamic pricing contract, devices that allow customers to see their consumption and expenditure in real time instead of waiting for their bill can change behavior. 4. D.J. Hammerstrom et. al, “Pacific Northwest GridWise Testbed Demonstration Projects, volume I: The Olympic Peninsula Project” (2007). http://gridwise.pnl.gov/docs/op_ project_final_report_pnnl17167.pdf 5. Ibid. WEBLINK >>>>>> More information and additional material can be found online at www.UtilitiesProject.com/10827 Innovating for the Future 5/2/08 9:12:25 AM