With the advent of the smart grid, state commissions throughout North America are showing increasing interest in dynamic pricing as a means of enhancing economic efficiency by reducing the need for expensive peaking capacity. But several barriers stand in the way of its rapid deployment.
As noted by MIT’s Paul Joskow in a recent discussion of the economics of climate change, “On the demand side there are relatively low-cost ways to reduce electricity consumption by increasing energy efficiency in building, lighting, heating, ventilating, air conditioning and other equipment. That’s why getting the retail price signals right is important and why muting them with regulation based on traditional cost- of-service models is inconsistent with promoting adoption of economical energy efficiency opportunities.”1
While the rate-design process (in conjunction with the revenue-requirements process) in principle results in utility recovery of all prudent costs, it doesn’t provide sufficient incentives to utilities to pursue energy efficiency and demand-response programs at a level commensurate with state and federal goals. A review of default rate designs across the continent reveals that prices paid by customers do not reflect the scarcity of capacity to produce energy at various times of day.
In fact, default rates embody a hedging or risk premium that insulates customers from price volatility and eliminates any incentive they otherwise would have for moving to dynamic-pricing tariffs. Additionally, customers lack the information to become smart shoppers. Policymakers have accepted a viewpoint espoused by defenders of the status quo that customers are averse to being placed on dynamic-pricing tariffs, since not only will they face price volatility but they also might pay higher bills. This is contradicted by evidence from 15 recent pilot programs on dynamic pricing, which clearly showed that once customers experienced a dynamic tariff, not only did they understand and respond to the price signals, they also overwhelmingly preferred dynamic tariffs to their conventional hedged rate form.2 The experiments also showed that a well-thought out customer education program is needed to sustain customer response.
In order to make a transition to dynamic pricing, a new framework is needed to develop innovative rate designs.
The fundamental premise is that that rate design should be driven by clearly articulated and feasible policy objectives.
For example, one rate structure might be designed to achieve simplicity in the important task of conveying the price of electricity to customers. Long, complicated bills over laden with fine print and impenetrable prose create a problem. A rate design that achieves this objective could be simply a flat volumetric charge (see Figure 1). Under this rate, a customer with 1,000 kWh of consumption per month would have a monthly bill of $190. The customer’s bill is calculated as follows: 1,000 kWh x $0.19/kWh = $190.
An even simpler pricing scheme is to charge the customer a flat fee per month of $190 plus a certain amount to capture the risk posed to utility earnings by month-to-month variation usage. However, completely decoupling electric costs from the rate of usage sends customers the wrong signal about the scarcity of the underlying resources required to supply electricity.
While this rate example achieves simplicity, it’s limited in its ability to achieve other objectives, such as promoting conservation and energy efficiency. One design to achieve this goal would charge customers a higher price the more they consumed (see Figure 2). The higher rate charged to consumption beyond 500 kWh per month would encourage customers to reduce their electricity consumption. At the same time, the rate for consumption up to 500 kWh is lower than the flat rate described in the previous example. This gives the customers the opportunity to save on their electricity bill if they are able to cut back on usage.
Without any price elasticity, the same customer’s bill would remain unchanged from the previous example: 500 kWh x $0.17/kWh + 500 kWh x $0.21/kWh = $190.
However, several studies have shown that customers do exhibit small but significant price elasticities, so in this example the customer likely would reduce consumption and achieve bill savings.3
While this example shows how rate design can encourage conservation, the design doesn’t provide any means for encouraging reductions in consumption during the peak (expensive) time of day. When customers reduce consumption during the peak hours, this allows the utility to improve its load shape and reduce the need for lightly used, and therefore very expensive, capacity. At the same time, this would achieve some of the conservation-related benefits that also would’ve resulted from the previous example. For such a time-of-use (TOU) rate, the peak period might last from 1 p.m. to 6 p.m. on weekdays, with all remaining hours being considered off-peak hours. Assuming the average customer in the previous examples consumes 250 kWh during the peak period and the remainder (750 kWh) during the off-peak period, the average bill would be calculated as follows: 250 kWh x $0.31/kWh + 750 kWh x $0.15/kWh = $190.
This example again assumes no price elasticity. In reality, customers might be expected to reduce peak energy usage in response to the higher rate. They also might increase usage during off-peak periods because the price is relatively lower. Of course, some customers would respond more than others to the time-varying price signal, with some perhaps not responding at all. However, in the aggregate, substantial customer response likely would occur.
This progression of rate designs illustrates how various objectives can be met. However, the tradeoffs also are apparent. While the TOU rate is still a straightforward way of charging customers for electricity, it’s not as simple or as easy to explain as the original flat rate. In the past, as rates were designed to accomplish an increasing number of objectives, they have tended to become more complex. It’s important to keep this in mind when discussing ratemaking criteria, and particularly when discussing the strengths and weaknesses of alternative dynamic-pricing rates.
In addition to such rate-design objectives as simplicity, efficiency and load management, a utility or policymaker might want rates to achieve several other goals, from bill stability for customers to revenue stability for the utility. These were first codified in 1961 by Professor James Bonbright of Columbia University.4
After years of research, Bonbright formulated eight criteria for establishing the rate structure. The original list of eight was subsequently expanded to the following 10:
1) Effectiveness in yielding total revenue requirements under the fair-return standard without any socially undesirable expansion of the rate base or socially undesirable level of product quality and safety.
2) Revenue stability and predictability, with a minimum of unexpected changes that are seriously adverse to utility companies.
3) Stability and predictability of the rates themselves, with a minimum of unexpected changes that are seriously adverse to utility customers and that are intended to provide historical continuity.
4) Static efficiency, i.e., discouraging wasteful use of electricity in the aggregate as well as by time of use.
5) Reflect all present and future private and social costs in the provision of electricity (i.e., the internalization of all externalities).
6) Fairness in the allocation of costs among customers so that equals are treated equally.
7) Avoidance of undue discrimination in rate relationships so as to be, if possible, compensatory (free of subsidies).
8) Dynamic efficiency in promoting innovation and responding to changing demand-supply patterns.
9) Simplicity, certainty, convenience of payment, economy in collection, understandability, public acceptability, and feasibility of application.
10) Freedom from controversies as to proper interpretation.
These criteria have served as guiding principles in electricity ratemaking for the past half century. They are simple and comprehensive but somewhat duplicative and verbose, and thus can be collapsed into three broadly defined criteria without any loss of content: 1) efficiency, 2) equity, and 3) simplicity.
While the Bonbright criteria are a good starting point for designing today’s rates, they’re insufficient for meeting the changing needs of a smart-grid world. The advent of advanced metering infrastructure, coupled with the introduction of in-home displays and price-responsive appliances, are bringing about a revolution in how consumers approach electricity. It’s necessary to update the criteria that we use for designing electric rates.
It’s possible to conceive innovative rate designs that meet the requirements of the smart grid world by conforming to four criteria: promote energy efficiency; promote equity; facilitate customer choice; and clearly and simply communicate prices and costs.
• Promote economic efficiency: The desire to achieve economic efficiency has been one of the key drivers underlying the increasing complexity of electricity pricing in the last two to three decades. When consumers pay prices that reflect the marginal cost of supply, societal resources are employed optimally in the economy; everybody wins. Increasing block rates are designed to reflect the fact that the marginal cost of electricity supply now exceeds the average cost, and time-varying prices reflect the time-varying nature of electricity supply costs.
A tariff that incorporates both an increasing block structure and time-varying pricing can provide adequate price incentives for encouraging energy efficiency and demand response. On the other hand, challenges in implementing such economically efficient pricing have been a key driver in the use of incentives to promote economic efficiency and demand response (DR). Incentives for energy efficiency are designed to account for the market failure in setting correct electricity prices that incorporate social marginal costs. Incentives for DR options, such as direct load-control, reflect the historical perception that load-control technology is cheaper than time-of-use metering. They also reflect a perception that behavioral response to time-varying pricing is more unreliable than push-button technology options.
It’s not hard to design economically efficient rates, even ones that incorporate the inherent uncertainty in supply conditions. What is difficult is designing economically efficient rates that customers understand well, that overcome the political challenge of transitioning from longstanding cross-subsidies to more equitable and efficient cost allocation, and that can be implemented cost effectively. The interplay and tradeoffs between economic efficiency and the other criteria needs to be re-examined in future regulatory deliberations.
• Promote equity: Equity in ratemaking can mean different things to different people. For some, it means preserving cross-subsidies and making sure that no one is made worse off relative to their existing situation. Of course, for this to be good for society, one has to assume that the existing situation was good to begin with. If the existing situation consists of significant cross-subsidies, some individuals will be made worse off when those cross-subsidies are eliminated—even though all they are giving up are financial gains that weren’t theirs to begin with. Thus, the most conservative definition of optimality, attributed to the economist Vilfredo Pareto, rears its head and crimps forward progress.
A Pareto improvement is one in which at least one person is made better off by a change in policy, while no one is made worse off. Adherence to only Pareto-optimal changes makes it impossible to move to a better allocation of resources through more efficient pricing, even if people agree it’s ultimately the correct outcome. An alternative and less restrictive economic metric is the Hicks-Kaldor optimality criteria, which states that if, as the result of a price change, winners gain enough to be able to pay off losers, and still be ahead, then that constitutes a welfare improvement for a society as a whole—even if the winners are not required to make such compensation. Note that if the winners were required to make the compensation, we would be back to the Pareto Optimality criterion.
An alternative definition of equity means having lower rates for low-income consumers, as is the case with the California Alternative Rates for Energy (CARE) program that provides a discount of at least 20 percent for low-income users. While rate options such as these are common throughout the industry, most economists would argue they distort price signals and lead to excess electricity consumption.
A third definition of equity is accurate cost allocation—that is, setting prices so they vary across customer classes or segments in accordance with variation in the cost of supply to those classes or segments. An example is having higher average prices for households with central air conditioning, or time-varying prices that incorporate the higher cost of supply associated with air conditioning loads during peak periods. Equity in this context focuses on eliminating cross-subsidies that are inherent in average cost pricing.
According to the second definition, lifeline rates (based on the theory that low-income consumers are low users) and such explicit discounts as the CARE tariff are worthy of pursuit. Lifeline rates (sometimes called baseline rates) are designed to meet the critical or lifeline needs of all consumers by supplying power at subsidized rates for the first several hundred kilowatt hours of usage. They serve a laudable social goal but detract from the overriding goal of economic efficiency. Regulators considering such rate designs should quantify the loss in economic efficiency they will create.
For example, suppose the full cost of power is 10 cents per kWh and customers pay 7 cents/kWh on the first 300 kWh of usage that’s designated as lifeline usage. Customers are getting a price subsidy of 3 cents/kWh on 300 kWh, or $9. In the second step, the $9 subsidy would be converted into an income subsidy and the price on the first 300 kWh would be raised to its full marginal cost of 10 cents/kWh. Probably most consumers would spend a good portion of the $9 income subsidy on higher value necessities such as food, clothing and transportation and conserve a certain amount of electricity by turning off lights in occupied rooms, perhaps installing compact fluorescent lamps, weatherizing their homes, adjusting their thermostat settings, and so on. The amount of electric usage might come down by a few percentage points, which would promote the state’s goal of enhancing energy efficiency.
In addition, removing price subsidies would improve the financial position of the electric utility. The financial burden of subsidizing customers would be shifted back to state and federal governments, on whose shoulders ultimately it should rest.
This social goal could be achieved, without compromising the goal of basing prices on costs to achieve economic efficiency in the allocation of scarce resources, by expanding the federal government’s Low Income Home Energy Assistance Program (LIHEAP).
Commissions need to test hypotheses about the distributional impacts of various rate options on different customer segments, rather than basing them on supposition and conjecture. For example, do low-use customers have flatter load shapes than high-use customers? If so, they are likely to be made better off with TOU pricing and not worse off, as some consumer groups often contend. Many myths and preconceptions have grown around equity issues. The only way to slay the myths is to quantify and analyze the implicit hypotheses concerning which tariff will make which customer group worse off.
Fortunately, new databases now exist that quantify the response of customers to alternative rate designs, making such analysis possible. A good example is the individual customer data generated by dynamic-pricing experiments across the continent. In most cases, published analyses focus on the behavior of the average customer. However, the databases are a fertile source of empirical information on customer response to rates that can be harnessed to test—and resolve— some of these distributional impacts that continue to be debated ad nauseum. The experimental data differ from the myriad datasets generated as part of ongoing load-research activities such as cost-of-service studies, load forecasting, and direct access compliance. Those datasets include information on hourly (and half-hourly) load shapes on a representative sample of customers. They usually don’t include information on customer characteristics (such as size and type of dwelling, saturation of end uses, and sociodemographic factors) nor do they include information on customer price responsiveness, both of which are richly represented in the experimental datasets.
• Facilitate customer choice: One of the objectives of power market restructuring activities initiated in the mid-1990s was to provide more choices to customers. Initially, policymakers thought the best way to accomplish this goal was by providing choice of power supplier. They hoped competitive power suppliers also would provide choice of pricing products and services to customers. The former avenue has not been found to work for mass market customers—at least not in California, the largest state to attempt customer choice. Even in places such as Baltimore and the District of Columbia, customer-switching rates for mass-market customers are very low. Thus, a way must be found for pursuing the latter through the incumbent utility provider. This is not as difficult as it seems. The incumbent provider can design and market a variety of pricing products for customers. These would be differentiated along the risk-sharing spectrum and represent different ways of allocating risks between customers and suppliers. A middle-of-the-road option, such as critical-peak pricing, can be made the default option, and customers would have the option to switch over to any of the other options that better match their risk-taking preferences.
• Clearly and simply communicate prices and costs: Retail rate structures vary widely from state to state, with some vastly more complex than others. California’s residential electricity tariffs, for example, with their increasing block structure, subsidies and surcharges, and unbundled cost structure, are among the most convoluted tariffs in the continent. And that’s before California’s landmark pricing experiment, the Statewide Pricing Pilot (SPP), incorporated time-varying surcharges, credits and dynamic price variation.5
Indeed, research conducted during the SPP indicated that many customers didn’t understand even the basic characteristics of their standard rates, let alone the nuances of how average and marginal prices move across rate tiers and time periods. On the other hand, the SPP showed that many customers did understand that prices were much higher during peak periods on critical days. Also, the SPP showed that time-varying prices can produce considerable peak demand reductions, even in a world of significantly increasing block tariffs and rate complexity. In other words, the SPP showed that even complex rates can produce demand response. What the SPP didn’t show, however, is whether significantly greater reductions could be achieved if rates could be simplified, or how best to achieve such simplification while reflecting sufficiently the key underlying economics of electricity supply.
There are many ways to make tariffs simpler and to help customers better understand and respond to price signals. A useful start would be to just simplify bill presentation by creating a simple summary sheet at the top of the bill and placing into a backup document the large amount of extraneous information contained in current bills (e.g., all of the unbundled bill amounts). Of course, one could seek to simplify tariffs themselves. For example, a simpler dynamic rate is what some refer to as a pure critical-peak pricing (CPP) rate, which preserves the dynamic nature of CPP without the burden and confusion of facing a time-varying rate every weekday. A pure CPP rate, with a high price on a limited number of emergency days and a single low price on all other days with no increasing block structure, would be fairly simple for customers to understand. On the other hand, such a rate focuses only on DR and not on energy efficiency. Alternatively, one could use a more complex, cost-reflective tariff that incorporates time variation and increasing block pricing, and rely on technology to automate response to price changes or to translate the complex tariff into more understandable information through, for example, in-house displays that report cumulative and incremental bill amounts.
None of the above examples, however, address the most fundamental challenge of electricity pricing, namely, the fact that no matter how simple the tariff, customers don’t know what a kilowatt-hour is or how much it costs them to do a load of laundry or a load of dishes or to run their refrigerator for a day. That is, customers don’t know whether the simplest tariff, say 10 cents/kWh, means that it costs 5¢, 25¢, or 50¢ to wash a load of dishes, or that a 5-degree change in a thermostat translates into a $1 savings on a typical summer day or a $2 savings on a really hot summer day. Consequently, the industry needs to explore the feasibility of service level pricing, that is, pricing based on the end-use services consumed.
Would such an outside-the-box pricing strategy improve consumer decision-making? Could one be designed to reflect accurately the underlying economics of electricity supply? An example might be 10 cents for a load of wash done after 8 p.m. at night, but 30 cents for the same load done between noon and 8 p.m. during summer weekdays. While implementing such an approach might present many practical challenges, the potential benefits could be huge. Indeed, this might be the only sure way to significantly improve customer decision-making.
Any move to default dynamic-pricing rates is bound to create some anxieties, because it will make some customers better off and some other customers worse off. Of course, all customers will have the opportunity to become better off by lowering peak demands, especially on critical days, but they might be unwilling to try this out unless their concerns are addressed.
One or more off the following steps might be useful in easing the transition.
• Creating customer buy in: Customers need to be educated on why a century-old practice of ratemaking is being changed. They have to be shown how dynamic pricing can lower energy costs for society as a whole, help them lower their monthly utility bills, improve system reliability, prevent an energy crisis, and lead to a cleaner environment.
• Offering tools: Energy management tools should allow customers to get the most out of dynamic pricing. At the simplest level, such tools should provide information on how much of the customer’s utility bill comes from various end-uses such as lighting, laundry and air conditioning and what actions will have the largest effect on their bill. At the next level, real-time in-home displays would disaggregate the customer’s power consumption and explain how much they are paying by the hour. Finally, these tools would include enabling technologies such as programmable communicating thermostats. Similar examples can be constructed for commercial and industrial customers.
• Designing two-part rates: The first part would allow customers to buy a predetermined amount of power at a known rate (analogous to how they buy all their consumption today) and the second part would provide access to dynamic pricing and allow customers to manage their energy costs by modifying the timing of their consumption. They could be allowed to pick their predetermined amount, or it could be based on consumption during a baseline period.
• Providing bill protection: This would ensure the customer’s utility bill would be no higher than what it would have been on the otherwise applicable tariff, but would not preclude it from being lower based on the dynamic-pricing tariff. Customers would simply pay the lower of the two amounts. In later years, such bill protection could be phased out. For example, in year one, the customer’s bill would be fully protected and would be no higher than it would have been otherwise; in year two, it would increase by no more than 5 percent; in year three, no more than 10 percent; in year four, no more than 15 percent; and in year five, no more than 20 percent. In the sixth year and beyond, bill protection would be provided for a fee.
• Crediting customers for the hedging premium: Existing fixed-price rates are very costly for suppliers to service, since they transfer all price and volume risk from the customers to the suppliers. In addition, the supplier takes all volume risk. In order to stay in business, the supplier has to hedge against the price and volume risk embodied in such an open-ended fixed-price contract. The supplier can do so by estimating the magnitude of the risk and charging customers for it through an insurance premium. The risk depends on the volatility of wholesale prices, the volatility of customer loads, and the correlation between the two. Theoretical simulations and empirical work suggest this risk premium ranges between 5 percent and 30 percent of the cost of a fixed rate, being higher when the existing rate is fixed and time-invariant, and smaller when the existing rate is time-varying or partly dynamic. So customers who move to dynamic-pricing rates should be credited for the insurance premium.
• Giving customers a choice: Dynamic-pricing rates, even with a full range of protections and features, still might be too risky for some customers. Thus, they should have the option of migrating to other time-varying rates, perhaps with varying lengths of the peak period and with varying numbers of pricing periods. If the critical-peak pricing rate (combined with a TOU rate) becomes the default rate, risk-averse customers should have the opportunity to migrate to a fixed TOU rate, and risk-taking customers should have the opportunity to migrate to a one-part or two-part real-time pricing rate.
The benefits of dynamic pricing are well established and increasingly within reach as advanced metering infrastructure and other smart-grid technologies are deployed throughout the continent. What stands in the way of progress is a misplaced concern about price volatility, and a fear of dealing with the push back that might come from those who would lose the subsidies they’ve enjoyed under existing rates. To ease the transition to dynamic pricing, commissions and utilities can use several methods individually or jointly. But unless the transition is accomplished, society as a whole will continue to suffer from the well-known inefficiencies of uniform, static pricing.
3. Ahmad Faruqui, “Inclining Toward Efficiency,” Public Utilities Fortnightly, August 2008.
4. The first edition of his canon, Principles of Public Utility Rates, was issued in 1961 and influenced the thinking of many generations of rate designers.
5. Ahmad Faruqui and Stephen S. George, “Quantifying customer response to dynamic pricing,” The Electricity Journal, May 2007.