The overriding objective of dynamic-pricing programs is to create a financial incentive for electric customers to curb load on the grid when the electric system peaks, either by conserving energy, shifting use to off-peak periods or generating electricity on-site. Such action by customers helps enhance reliability and curb the potential for market-power abuse by generators; most important, it obviates the need to build central generation and delivery capacity to serve load that occurs in only a small number of hours. Indeed, experts believe most of the customer benefit from dynamic-pricing programs comes from avoided generation, transmission and distribution capacity costs rather than avoided energy costs.1
Some also believe that a wide-spread implementation of dynamic-pricing programs ultimately would eliminate the need for subsidized demand-response (DR) programs, in which all customers pay some customers to drop load during event hours called by the system operator during system peaks.2
But in New York state, the hourly prices that utilities pass through to customers in their mandatory hourly pricing (MHP) programs do not send strong signals to upstate New York customers during peak demand periods (see Figure 1). Electric load consistently peaks in the summer months. New York state consistently uses about 10 percent of its generating capacity (~3,200 MW) in just 100 summer-time hours. In contrast, driven by spikes in the price of natural gas, day-ahead electricity prices in upstate New York generally peak during the late afternoon and evening hours of the winter months. Day-ahead price signals in much of downstate New York also are more muted than one might expect at the time of the system peak.
The lack of strong price signals at the time of the system peak might not be so surprising, especially to those who follow developments in wholesale energy markets. What is, however, surprising is the growth of support for advanced-metering and dynamic-pricing programs in New York state when the day-ahead price signals that help to substantiate these programs seem so muted. The lack of strong price signals reduces the financial incentive for customers to make their operations more flexible and to install the on-site generation needed to curb grid-supplied demand during peak periods. The lack of strong price signals, therefore, undermines a key objective of implementing hourly pricing programs, weakening the case for installing advanced metering infrastructure (AMI) and providing customers load-curtailment services predicated on strong price signals. Indeed, a recent study found that the incremental annual savings available to the average mass market customer in National Grid’s upstate New York service territory under an hourly pricing program is quite small, in large part due to the lack of strong price signals at the time of the system peak.3
The most obvious explanation for muted price signals at the time of the system peak is a surplus of electric generating capacity. This makes sense, since under deregulation, policy makers have maintained the same generation adequacy standard (one day of outage every 10 years) that existed under regulation. To implement this standard in New York, all load-serving entities (LSEs) are required to purchase enough unforced capacity to cover their forecast peak load plus a minimum margin. Capacity payments are designed to help generators cover the fixed costs they don’t recover through the energy market. To receive capacity payments, generators also must agree to bid into the day-ahead market for energy. Because more generation than actually is needed bids into the day-ahead market, the possibility of scarcity prices is sharply reduced.
The amount of excess generating capacity that bids into the day-ahead market has increased over the past five years. The minimum reserve margin for unforced capacity (UCAP), set in November 2002 at 12 percent, was raised in May 2003 to 14 percent. However, the amount of capacity in excess of the forecast peak expanded well beyond this level in the middle of 2003 when the administrative demand curve (demand curve) for spot capacity was implemented by the New York Independent System Operator (NYISO). The minimum reserve margin has been lowered over time and was set at 8.4 percent through October 2008. But, the amount of capacity in excess of the forecast peak still is close to 20 percent, primarily due to the increased demand fostered by the administrative demand curve (see Figure 2).
The tendency for energy prices to spike at the time of the system peak has mirrored changes in the amount of excess capacity. When the margin of excess unforced capacity was quite low, between 2000 and 2002, hourly day-ahead prices even in upstate New York were much more prone to price spikes in the summer. For example, the day-ahead hourly price in the Albany capital region spiked to more than $1.40 a kWh in the summer of 2000 and rose near the newly imposed $1/kWh price cap during the summer of 2001. The hourly day-ahead price also was quite volatile during the summer of 2002. The more muted behavior since 2003 by hourly prices during summer peak hours coincides with the jump in excess unforced capacity to the 20-percent level. The excess in generating capacity state-wide has had a smaller damping impact on the hourly day-ahead price for electricity in New York City and on Long Island, because those markets have location restrictions that reduce the number of generators qualified to participate. This increases the potential for scarcity prices in those markets.
Ultimately the most straightforward way to solve the problem of missing price signals is to eliminate the generation adequacy standard, capacity payments and the administrative demand curve for capacity. But policy makers are reluctant to abandon these mechanisms until they are confident that demand-side resources are as reliable as generation resources and will fully participate in the day-ahead price discovery process.
Boosting this confidence will require more than installing interval meters and extending MHP programs. It also will require clear evidence that a sufficient proportion of customers have made their operations flexible enough to curb use at the time of the system peak. Such investments only can be justified and likely to occur, if the hourly retail prices that customers face are sufficiently high for a sufficient number of hours. This creates a Catch-22: Hourly pricing customers are not likely to invest in the equipment and processes needed to enable demand response unless they are subject to significant hourly price signals. But significant hourly price signals at the time of the system peak are not likely to be forthcoming until policy makers dial back the minimum capacity requirement and the capacity demand curve. And for this to happen there would have to be evidence that a significant portion of customer load has the operating flexibility to reliably respond to hourly price signals.
Some suggest the answer is to simply pass through the NYISO’s hourly real-time price to customers, since real-time prices considerably are more volatile than day-ahead prices. Indeed, the hourly real-time price in the Albany region peaked at $1.04 in 2006 and $1.29 in 2007. But despite the volatility, the real-time price is no more likely to send a strong signal at the time of the system peak than is the day-ahead price (see Figure 3). For example, the annual peak in the hourly real-time price in 2007 occurred at 5 p.m. on October 18, 2007, when the electric system load was far below the summer peak. More important, the hourly real-time price is posted at the end of each hour, so whatever signal the real-time price contains occurs after customers already have consumed energy. Studies show that customers subject to an hourly real-time price for default commodity service are more likely to seek out hedged commodity service to escape the volatility and stochastic nature of real-time prices.4 In contrast, customers subject to day-ahead hourly price for their default service are more likely to remain on this service, and modify usage in response to price signals, because they know in advance what cost they will avoid by curtailing or shifting load.
If we accept the premise that capacity requirements, the capacity market and the administrative demand curve are here to stay, there is still a promising way to send strong and persistent signals to customers at the time of the system-peak. Utilities can collect capacity costs in their retail commodity rates in a way that more accurately reflects when these capacity costs are incurred, and that reflects the role that customers play in driving capacity costs higher. To quote Paul Joskow, a leading figure in wholesale market design “It matters how capacity payments are reflected in retail prices.”5
In large part, capacity costs in New York are determined by the amount of electricity that customers jointly demand during the hour of the system peak, since each LSE must purchase enough unforced capacity for the next year to cover the peak demand of their customers in that hour plus a minimum reserve margin.6 LSEs can collect these costs from customers however they see fit. They generally choose to do so in a manner that ensures that capacity inflows from customers match monthly capacity payments to generators in order to mitigate the risk of non-recovery. Establishing a rate design that sends a strong price signal to customers at the time of the system peak appears to be less of a concern. But it seems apparent that each utility simply could modify its existing approach to address this situation, and in so doing help to realize the ultimate goal of dynamic-pricing programs, which is curbing load at the time of the system peak.
New York State Electric and Gas (NYSEG) and Rochester Gas and Electric (RGE) collect capacity costs from MHP customers based on each customer’s capacity tag or, equivalently, the customer’s demand at the hour of the previous year’s New York system peak. Many energy services companies (ESCos) bill retail customers for capacity using a similar approach. This method helps the LSE ensure that it collects the amount of revenue needed to meet its capacity obligation to generators each month. In addition, each MHP customer is held directly responsible for its role in contributing to the prior year’s New York system peak. Customers with higher tags will have higher monthly capacity bills. The separate charge for capacity on each customer’s monthly bill also helps customers focus on ways to reduce this charge—especially since customers must live with their tag for an entire year.
The capacity tag approach does not send a dynamic signal to customers, but the charge associated with the capacity tag on each customer’s bill changes monthly and could be adjusted to send a stronger signal to customers during the summer months. Even if the monthly capacity charge is fixed each month, customers still have an incentive to reduce their peak usage in the summer as a way to reduce their capacity costs. Customers can either guess when the peak will occur or simply trim demand during summer time hours from 3-5 p.m. when the system is most likely to peak. The main drawback to the capacity tag approach is that it only works for customers who have interval meters.
National Grid collects capacity costs from its retail commodity customers in upstate New York by adding a class-specific kWh capacity charge to the NYISO’s hourly DALBMP each week-day from 12-8 p.m. The capacity adder creates a year-round financial incentive for hourly pricing customers to shift load away from week-day afternoon hours to mornings, evenings and week-ends. But the existing capacity adder does not send a strong incremental incentive to customers to modify use at the time of the system peak.
One way for National Grid to create such a signal would be to modify the capacity adder for each hour based on the relative probability that hour will be the system peak. One can see what National Grid’s hourly retail prices would have been in 2007 if its capacity adder had been designed using the 90 percent probability of peak (POP) approach (see Figure 4). Under this rate design a capacity adder would be imposed only during the summer week-day hours from 10 a.m. to 10 p.m. The adder would vary by hour based on the probability of peak and the highest adders would occur between 3 p.m. and 5 p.m., the hours most likely to be within 90 percent of system peak. The hourly retail price would have been above 15 cents/kWh during the weekday hours between 12 noon and 5 p.m. in the summer of 2007 and 4 percent lower in the other three seasons, so the average customer’s annual commodity bill would have been about the same.
Although the POP method to collect capacity costs would not create a dynamic price signal like that provided by the DALBMP, it would help to send a strong and persistent signal to customers at the time of system peak so they can better justify investing in operational flexibility to reduce load on the grid at the summer peak. The POP method also would send a stronger signal during the peak summer months to customers who do not have interval meters and who currently are billed for commodity based on an average of hourly retail commodity prices.
Con Edison and Orange and Rockland use a monthly demand charge to collect capacity costs from retail customers. Again, like the approaches used by National Grid and NYSEG/RGE, this approach does not send a dynamic-price signal to customers at the time of the system peak. Instead, it creates an incentive for customers to reduce peak use every month, including the month in which the electric system peaks. The demand charge is based on prices from the NYISO’s six-month strip auction. For New York city the six-month strip price for capacity has been markedly higher during the summer capability period, so customers get a much stronger price signal to curb peak usage during the months from May to October.7 And for customers with interval meters, Con Edison also narrows the range of hours subject to the demand charge in summer to encourage customers to reduce load during hours that are more likely to be the summer peak. This would seem to have a similar impact as would a capacity adder based on the POP model described above for National Grid.
In order to realize the potential of dynamic pricing programs, customers should be billed for the electric commodity based on their actual hourly electric use and an hourly price that approximates the opportunity cost of consuming electricity in that hour. The easiest and most efficient way to accomplish this would be to abandon the generation reserve requirement, capacity payments and the administrative demand curve for capacity, and pass through market-determined hourly prices that include scarcity rents in those hours when generation is scarce. Policy makers understandably are reluctant to abandon these mechanisms until they are confident that demand-side resources are as reliable as generation resources, and will fully participate in the day-ahead price discovery process. However, hourly-pricing customers are not likely to invest in the equipment and processes necessary to enable demand response unless they are subject to significant hourly price signals.
One possible way out of this dilemma is to consider changing the way that LSEs are billed for capacity. If load-serving utilities were billed for capacity mainly during the summer months, they would be more apt to collect capacity costs from commodity customers in a way that sends a strong price signal to customers at the time of the system peak. These signals do not necessarily need to be dynamic in order to achieve the most important goals of dynamic-pricing programs—clipping peak load. Indeed, it seems that a strong and persistent signal at the time most likely to be the system peak would create an even stronger financial incentive for customers to modify their operations or to invest in the on-site generation equipment required to curb load on the system at the time of the peak. If customers responded to these signals, the need for system-wide capacity would be tempered, the capacity requirement would be lower and customers over time would pay less for capacity.
As a next step in this direction, utilities might consider running dynamic pricing pilot programs to evaluate customer response to alternative commodity rate designs to restore the missing price signals at the time of the system peak. The results from these pilots would help policy makers to better evaluate the potential advantages of AMI, dynamic pricing programs, and changes required to implement effective commodity rate designs at the retail level. Indeed, pilot programs that delivered significant price signals to customers at the time of the system peak were an important first step in developing and implementing more broad-based AMI/dynamic pricing programs in California.
In the mean time, utilities can continue to invest in, and promote, energy-efficiency programs. Installing more energy-efficient equipment (better air conditioning systems or raising thermostats on existing equipment, for instance) saves more kilowatt hours during the peak hours of the day when customers use these devices most, and therefore has the potential to significantly reduce summer peak loads. Other measures that promote energy efficiency would have similar built-in demand response potential. It seems like centrally administered subsidized demand-response programs also would continue playing a useful role.
1. Faruqui, Ahmad and Stephen George, “The Value of Dynamic Pricing in Mass Market,” Electricity Journal, July 2002.
2. Some believe that dynamic-pricing programs also help to lower wholesale market prices, but this depends on how the demand side of the market participates in the wholesale price discovery process. More important, lower wholesale prices simply transfer income from producers to consumers and results in no net savings to society as a whole.
3. See McDonough, Catherine, and Robert Kraus, “Does Dynamic Pricing Make Sense for Mass Market Customers?” Electricity Journal, August-September 2007.
4. Barbose, Galen, et al., A Survey of Utility Experience with Real Time Pricing, December 2004.
5. Joskow, Paul, Competitive Electricity Markets and Investment in New Generating Capacity, April 2006.
6. Because an LSE’s capacity requirement can change from month to month as customers migrate among commodity suppliers, LSEs must perform monthly true-ups by buying or selling capacity in the NYISO’s monthly balancing auction. But these transactions do not change the total amount of capacity required from a societal point of view. LSEs also might be required to purchase additional capacity each month to stabilize the spot price of capacity as dictated by the administrative demand curve. The administrative demand curve might induce an increase in the demand for capacity from a societal point of view, but this demand is not driven by customer behavior.
7 . The six-month strip price during the summer capability period in New York City is $11 to $12 per kW/mo and $6 to $7 per kW/mo in the winter capability period. In contrast, six-month strip prices in upstate New York range between $0.50 kW/mo and $2.50 per kW/mo. Prices during the summer capability period are somewhat higher than in the winter capability period, but the difference is much less extreme than in New York City.