When electricity markets were originally designed, it was assumed that responsive demand would play an essential role. After nearly two decades of experimentation, most electricity market models have yet to incorporate demand participation.
Because most electricity markets evolved with inelastic demand as a foundation, market power became a real problem. Conversely, the control measures that were put in place to mitigate the exercise of market power have become an impediment to responsive demand. We can break this cycle now by implementing demand participation with today’s technologies.
Without demand participation, market power must be mitigated using control measures
While early market designs presumed a significant role for responsive demand, most electricity markets still lack true demand participation. When spot pricing of electricity was introduced in the 1980s, it was presumed that customers would be exposed to the real-time price of electricity and respond by reducing their consumption when prices increased significantly. (Schweppe, 1988) Although today’s electricity markets have implemented various measures to leverage demand-side resources for system reliability, most of them fall short of true demand participation. In some cases, attempts to establish “demand response” programs have even been counter-productive. (EPSA v. FERC, 2012)
The absence of demand participation has been cited as a primary factor enabling the exercise of market power in competitive electricity markets. Although there was no conclusive evidence that the high prices during the California crisis of 2000 were caused by physical withholding by power generators, (Harvey, 2002) it has been found that a 5% lowering of demand would have resulted in a 50% price reduction during peak hours. (Hirst, 2001) In a matter related to ISO-NE, FERC has stated that the lack of price-responsive demand was a major impediment to the competitiveness of electricity markets. (ISO-NE, 2001)
Price-responsive customers exhibit demand elasticity, thus directly countering the exercise of market power by producers. (Borenstein, 1999) In a market where customers could conveniently buy less electricity when prices were high, it would be difficult for power generators to increase prices without losing significant sales volume. Conversely, if customers were not aware of the price or needed electricity at any cost, producers could profitably raise prices to very high levels. In other words, power generators would only be incentivized to exercise market power if demand were inelastic enough for the reduction in sales volume to still be profitable. If demand were very elastic, then firms would have to collude illegally in joint output reductions for the exercise of market power to be advantageous.
The expectation of demand response by itself can effectively deter the exercise of market power. Price-responsive demand can mitigate the potential exercise of market power by producers even if demand response rarely occurs. Prospective elasticity in demand is a sufficient deterrent for market manipulation because the possibility of demand reductions in response to higher prices increases the risk to suppliers of incurring losses. (DOE, 2006)
Effective demand participation in electricity markets would limit the need for control measures such as offer caps. In order to foster competitive behavior on the supply side, markets require a well-functioning demand side. The demand side of most competitive electricity markets today is severely underdeveloped. This underdevelopment is among the primary reasons for the necessity to maintain control measures such as offer caps in PJM and other markets. (Monitoring Analytics, 2005)
Market power control measures repress prices and preclude responsive demand participation
Market power control measures such as offer caps, administrative actions, and out-of-market transactions tend to suppress real-time energy prices during peak times. This limits incentives to invest in both supply-side capacity and demand-side capabilities to meet peak demand and has created a long-term market failure referred to as the “missing money” problem. (Hogan, 2005) Although capacity auctions have been established to address the “missing money” problem, their usefulness is limited to getting “steel in the ground” for generation capacity. Capacity auctions are thus not sufficient to incentivize demand participation. (EMRF, 2014)
The essential starting point in addressing revenue reconciliation problems such as “missing money” is to improve real-time price signals by all available means. According to first principles, real-time prices should reflect current real system conditions and produce the most efficient outcome possible through the rational choices of market participants. (Hogan, 2001) Without adequate real-time price signals, out-of-market interventions such as capacity markets and uplift charges fail to improve system operations and lead to economically inefficient outcomes. (Centolella, 2006)
Lack of voluntary demand participation in electricity markets makes providing reliable electricity service much more expensive than necessary. Because involuntary loss of power is costly to the economy, the electricity system is overbuilt such that wholesale-level outages are limited to one event every ten years. (Brattle, 2013) ERCOT conducted an analysis of the Value-of-Lost-Load (VOLL) for involuntary curtailments and concluded that it ranges on average from less than $1,000 per MWh for residential loads to over $40,000 per MWh for small industrial loads that lack backup generation. (London Economics, 2013) Conversely, for loads that can be voluntarily curtailed or time-shifted, the VOLL has been shown to be less than $500 per MWh and often negligible. (PNNL, 2014)
Similarly to the “missing money” problem for generation capacity, investment in equipment for responsive demand capabilities necessitates adequate scarcity pricing. Assuming that there is no VOLL, in order for demand-side capability investments to have a five-year payback with 5% of total hours at high peak load the scarcity price would need to reach $1,222 per MWh for residential customers and $930 per MWh for small residential customers. [Appendix A] These required scarcity prices assume that the VOLL is zero. In practice, the VOLL would range from being very small to very large based on the specific load and timing. The spot price during times of scarcity would need to be enough to cover both the VOLL and the unamortized cost of equipment. Due to market power control measures, the spot price in markets such as PJM seldom reaches $500 per MWh. (PJM, 2013)
Incorporating operating reserves into the real-time spot market would improve the adequacy of scarcity prices and incentivize demand participation. (Hogan, 2013) The ensuing market price would combine energy and ancillary services in real-time through the use of an “operating reserve demand curve”. The price would thus more accurately reflect real-time system conditions at each location, particularly during times of scarcity. Scarcity hours would become more frequent, the price duration curve would become more linear, and the resulting scarcity prices would be able to exceed generator offer caps. This would provide better incentive signals to resources that can effectively provide operating reserves, both on the demand side and the supply side.
We can break this cycle now by implementing demand participation with today’s technologies
Before effective demand participation can occur, customers must be exposed to their locational spot price in real-time. Although over a third of US households are currently equipped with smart meters, less than 2% are buying the energy portion of their electric on a time-varying rate. Of this small percentage, most are exposed to time-of-use (TOU) rather than real-time prices (RTP). (Faruqui, 2014) Real-time prices are crucial for system reliability because extreme scarcity periods tend to be relatively short and voluntary load curtailment would be the most efficient means of stabilizing supply and demand. Because the overwhelming majority of customers pay a fixed retail rate, responsive demand has historically been obtained through “demand response” programs that act as incentive payments for curtailment rather than enabling demand to truly participate in real-time electricity markets. (Chao, 2011)
Devices that can automate customer choices and responses to price signals are essential for the successful implementation of demand participation. (CPUC, 2014) After decades of rapid progress, technological capabilities for communications, computation, and control continue to improve exponentially. (Moore, 1965) Although appliances with price-responsive capabilities are currently limited, industry research suggests that their adoption will increase dramatically after 2015. (Navigant, 2012) Developments in communication technologies such as LTE wireless broadband service, MPLS networking, and the IPv6 protocol are forming the backbone for an interconnected “smart grid”. (Budka, 2014) Because interconnected networks of smart devices take complexity to a scale where the devices can no longer be managed centrally, distributed computing and control architecture is also necessary. (Ambrosio, 2011) To achieve this, interoperability standards are required such that devices can communicate efficiently and understand each other’s language. (GWAC, 2008)
As a matter of open access, adequate information about past transactions and future prices should be available to all buyers and sellers. The growing abundance of information simplifies the convenience and reliability of demand participation for customers and service providers. For example, a variety of dynamic pricing pilots have suggested that demand response to price signals can be estimated confidently and accurately. (EPRI, 2008) FERC has a statutory mandate to facilitate price transparency and provide for the dissemination, on a timely basis, of information about prices to the public. (FERC, 2012) Although initiatives such as Green Button have been established, the applications are limited and detailed market information is still only available to utilities and ISOs. (NIST, 2013)
Retail competition and demand aggregation services would significantly improve both the value proposition for customers and the reliability of demand-side resources. Real-time pricing for retail customers is a sensitive topic and may require some business model innovation to transfer the risks from customers to for-profit companies. For example, the California State Senate has instated legislation prohibiting utilities from implementing mandatory or default real-time pricing for residential customers without bill protection. (CA Senate, 2009) In addition to offering bill protection services, retail energy providers could finance the initial cost of equipping a home or business with responsive demand technology. From the perspective of the system operator, there is tremendous value in making demand-side resources dispatchable. (NYISO, 2011) Furthermore, load aggregation diversifies heterogeneity and leverages second-order dynamics such as limited energy stocks, (Hogan, 2014) changing elasticities, (IEEE, 2013) staggered availability, and resource orchestration. (PNNL, 2013)
Forward transactions can make demand-responsive electricity markets more competitive, improve reliability, and offer buyers and sellers more control and certainty. In a Cournot context where firms are price takers and choose quantities, it has been found that when a forward market exists each firm will sell forward, thus impeding the exercise of market power and making customers better off than if the forward market did not exist. As the number of forward trading periods reaches infinity, the outcome approaches the fully competitive solution. (Allaz and Vila, 1993) As an example, CAISO has argued that forward contracting by load serving entities would have prevented the California crisis. (CAISO, 2000) More recently, market designs have been proposed that leverage highly liquid forward markets on continuous nested timescales for energy and transport where the vast majority of energy deliveries are transacted in advance. (Cazalet, 2011) One possible improvement to make demand more responsive and improve system reliability under such designs would be to impose an unhedged operating reserve price on all real-time energy deliveries during times of scarcity.
Conclusion: demand participation has already begun to revolutionize electricity markets
Demand-side resources are a relatively inexpensive way to improve the capabilities of our electricity infrastructure. Estimates by the Brattle Group and NREL have found that technology-enabled price-responsive demand could reduce peak energy demand by over 20% (Faruqui, 2013) as well as provide 33% of frequency regulation, 19% of frequency reserves, and 85% of flexibility reserves. (NREL, 2013) Demand-side resources could also be a viable alternative to transmission expansion and new generation for system reliability and resource adequacy. As with other “market resource alternatives” such as distributed generation and storage, demand-side resources provide a dependable and economic alternative to more “lumpy” capacity expansion projects. (London Economics, 2014)
The transition to a demand-responsive electricity system faces challenges in implementation and opposition from incumbent stakeholders. Incremental improvements to the system have to explicitly consider the need for an efficient, workable market. Beginning with the Energy Policy Act of 1992, it took over 15 years for deregulated markets to adopt a relatively standardized market design. (FERC, 2007) Initially, market design discussions were dominated by concerns over the allocation of costs for stranded generation assets. (Hogan, 1994) The mounting debate over demand participation in electricity markets has shown signs of a similar pattern of opposition and concern for sunken investments. (EPSA v. FERC, 2014)
There is a growing body of demonstration projects and real-world experience proving the feasibility and value of demand participation in electricity markets. In 2007, the Pacific Northwest National Laboratory (PNNL) conducted a 112 household pilot project dubbed the Olympic Peninsula Project that saved customers approximately 10 percent on their electricity bills and reduced peal demand by 15 percent. (PNNL, 2007) The DOE-affiliated national laboratory subsequently got $173 million in funding though the American Recovery and Reinvestment Act of 2009 (ARRA) to conduct a 60,000 household smart grid demonstration project ending in 2015. American Electric Power (AEP) also conducted a DOE-funded pilot project in Ohio that achieved a 10 peak feeder load reduction and 75% customer satisfaction. (PNNL, 2014) In New York, an ambitious regulatory reforming dubbed Reforming the Energy Vision (REV) is underway that promises to revolutionize electricity distribution by establishing Distributed Service Platform Providers (DSPP), overhauling electricity ratemaking, and significantly increasing third-party competition. (NY PSC, 2014)
Appendix A: Scarcity price calculations [go back]
- Discount rate is 10%
- Repayment period is 5 years
- Curtailment is voluntary with a VOLL of $0
- Scarcity pricing occurs for 5% of hours annually
- Loads curtail 30% of their demand during scarcity
- Total AMI installed cost per meter averages $220
- Wi-Fi smart thermostats retail for $230 and can be installed by the user
- The incremental cost for a smart washer, dryer, and dishwasher is $300
- The average US household consumes 10.8 MWh of electricity annually
- Installing a building automation system (BAS) for a 20,000 square-foot commercial building with annual power consumption of 350 MWh costs $18,500
- The calculations assume that 5% of hours equates to 5% of energy consumed and that no other benefits of installing “smart grid” devices are being captured
- Scarcity Price = Annuity / [Total MWh × Scarcity Hours × Scarcity Curtailment]
- Annuity is the annual scarcity rent where PV(Discount rate, N periods, Annuity) = ΣCost0
- Residential annuity: PV(10%,5,Annuity) = 220 + 230 + 300 ⇒ Annuity = 198
- Residential scarcity price = 198 / [10.8 × 5% × 30%]
- Commercial assumptions are 18,500 total cost = 4,880 annuity and 350 MWh