In recent years, the modernization of the electrical grid has taken center stage as an issue of importance to utility planners and stakeholders. It has been supported beyond the industry as well, in the form of stimulus grants from the federal government, positive guidance from the Department of Energy, and the ongoing development of “smart” appliances, meters, and other technologies by electronics manufacturers. Dozens of initiatives have already been undertaken by utilities throughout the country, and more are in the planning stage.
But these initiatives have not come without controversy. In California, Texas, and other states, electricity customers have resisted the introduction of “smart meters” into their homes, on the grounds that these constituted an invasion of privacy, presented health risks, and/or produced inaccurate results. And in some states, regulators, as well, have been unwilling to approve investments in infrastructure that would modernize the grid, on the grounds that the promised benefits to ratepayers were too uncertain.
In response to these challenges, utilities have taken a fourfold approach to make an effective case for grid modernization. They have developed communications plans which effectively rebut the unfounded fears surrounding smart meters. In their regulatory filings, they have introduced innovative strategies for sharing the risk associated with any grid infrastructure investment between shareholders and ratepayers. They have identified and, when possible, quantified all of the benefits – consumer as well as societal – that will stem from modernizing the grid. But they have also endeavored to ground their business cases on solid cost estimates, with only measurable, trackable, and verifiable benefits included.
Nowhere has the need for an irrefutable business case been more necessary than with smart meters, which have become something of a lightning rod in the controversy surrounding grid modernization. And while there are clear and measurable benefits associated with the implementation of these, in terms of operational savings and the deferred need for investment in generation and transmission capacity, one type of benefit which runs the risk of being overstated (and which therefore endangers the legitimacy of the entire business case) is the promise of consumer savings due to demand response. In the following discussion, the fallacy of equating savings with benefits will be illustrated, followed by an explanation of how to measure these benefits correctly.
The inherent promise in demand response is that, through peak/off-peak or real-time pricing (which is generally made possible through the use of smart meters), customers can save money on their electricity bills by reducing electricity usage during periods when prices are high and shifting all or part of that reduced load to other periods when prices are lower. The benefit to customers of doing so, it is often argued, is equivalent to the bill savings that are realized. Hence, if my electricity bill, for example, used to be $200 a month in the summer, but then drops to $150 a month after I go onto a time-of-use program and curtail my air conditioning load during peak hours, the benefit to me of being on the program is equivalent to this monthly summer savings of $50.
To understand the flaw in the net savings approach, consider the following example: Suppose that during the workweek, twice every morning, I visit my local coffee shop, and purchase a cup of coffee for $2 a cup. My expenditures for coffee each week then total $20.00, and I drink ten cups of coffee a week. But now suppose that – in order to manage the long lines due to heavy morning demand – the coffee shop raises the price of coffee to $3.00 per cup. Deciding that $30 a week is too much to spend on coffee, I cut my purchases in half, so that now I am only spending $15.00 in total each week. Now imagine how I would feel if this coffee shop starts to boast that it has done me (and other customers who behave in the same manner as I did) a big favor, by reducing my coffee expenditures by 25% - labeling these as “savings”! Am I better off as a result of this new arrangement, just because I am spending 25% less on coffee? I (and most persons who made the same choice as I did) would not think so. We’re spending less, but we’re getting much less of what we enjoyed and valued as well.
Defenders of the savings approach when applied to electricity sales might counter that the demand reductions occurring due to real-time electricity pricing are not necessarily lost electricity sales – all or part of them could be sales that just occur now at a different time of day (i.e., off-peak). So let’s take the coffee example one further step. Suppose that my local coffee shop decides to reduce the price of coffee in the evening hours, when demand slackens and its resources are underutilized, from $2.00 to $1.00. It then proudly announces that customers like me can still spend the same amount of money in total each week on coffee: all we need to do is change our buying patterns so that we buy one cup of coffee in the morning and one cup of coffee in the evening. Even better, they argue, if I choose to buy more coffee in the evening than in the morning, then I will be able to consume just as much as before and actually still save money. Am I as well off now as I was before? Hardly. While technically this is all true, I will not see much value in getting that caffeine jolt at 8:00 in the evening. I could be getting just as much coffee, and even spending less for it, but life just doesn’t seem as good to me.
Similarly, when discussing (and quantifying) the potential consumer benefits that could accrue from demand response, it is not correct to use dollar savings as an indicator of these benefits. To better appreciate the irrelevance of this measure, it will be helpful to revisit that bane of the introductory economics courses, the consumer surplus graph. Consider the following hypothetical example: An electric utility has been offering a standard flat rate energy charge for electricity of 4.07 cents per kWh. (Note that these rates do not include transmission, distribution, and customer service costs, all or part of which might appear in the customer’s bill as a fixed monthly fee.) For simplicity’s sake, we assume that the utility’s residential customers have exhibited two distinct consumption patterns: an on-peak usage averaging 1.327 kWh from the hours of 2:00 PM to 7:00 PM on weekdays, and an off-peak usage averaging 0.851 kWh during all other times. At these levels, the customer’s total energy charges for the year are $357.
Now the utility would like to introduce a time-of-use rate. Based upon load research, it has estimated that during peak hours, customer usage is price sensitive, with a price elasticity of -0.6 (i.e., for every 1% increase in the energy price, usage would decline by 0.6%). During the off-peak hours, however, customer load is relatively insensitive to price, with an elasticity of -0.1. This should be the case, as consumption during the off-peak hours will tend to be dominated by electrical appliances that run constantly and/or are considered necessities, such as computers and lighting. With similar information derived about the sensitivity of electricity supply to price, the utility is able to derive supply and demand curves, as illustrated in Figure 1. (For simplicity, all supply and demand curves are assumed to be straight lines, with the assumed elasticities occurring in the regions where price changes will be occurring due to adoption of the time-of-use rate.)
The utility introduces a time-of-use tariff with a peak energy rate of 4.94 cents/kWh and an off-peak rate of 3.67 cents/kWh. When the new rate is instituted, we see a 13% drop in peak-time usage from 1.327 kW to 1.157 kW during each hour of the peak (and a 1% increase in usage during off-peak hours). Unlike the coffee example, this change to higher pricing during the peak will not result in a net savings. In each hour of the peak, the consumer will be paying an incremental amount of:
(1.157 kWh x $.0494/kWh) – (1.327 kWh x $.0407/kWh) = $0.0031, (1)
which, with 1,304 hours of peak demand in each year, totals $4.10. However, this extra cost is more than made up for during the off-peak hours, where savings will occur:
(0.859 kWh x $.0367/kWh) – (0.851 kWh x $.0407/kWh) = ($0.0031). (2)
Although the hourly savings rate off-peak is roughly equivalent to the extra hourly cost that is incurred on-peak, there are many more off-peak hours in the year than on-peak hours, totaling, in this example, 7,456. Total off-peak savings will then be $23.19, leading to a net annual energy savings of $19.09 per year. But is this the actual net benefit to the consumer which would be obtained by switching from flat to peak/off-peak pricing? A closer look at what is happening, in terms of change in consumer surplus, will tell the real story.
Figure 2 illustrates what is happening on-peak.
Customers are using less but paying more on a per kilowatt-hour basis. As indicated in the earlier calculation, in spite of the reduced usage, the higher rate will cause customers to pay more in total during the on-peak period with the new rate. In terms of the labeled rectangles in Figure 2, this net change in expenditures is equivalent to:
(A + B) - (B + D) = A-D, (3)
where Rectangle A is the incremental expenditure from purchasing electricity at the new, higher rate, while Rectangle D is the change in expenditures associated with the change in usage. The problem with this calculation is that it is not truly measuring the change in net consumer surplus. Rectangle A should be included, as it was in the savings calculation, because paying a higher price for the electricity that is still being purchased is a direct loss. But the impact of reduced electricity usage on net consumer surplus is trickier. The total value, from the consumer perspective, of the electricity no longer being purchased is equivalent to the area under the demand curve, bounded by the old and new usage levels, which is represented as the sum of Triangle C and Rectangle D. Foregoing these electricity sales does produce a cost savings, equivalent to Rectangle D, but this cost savings is exceeded by the total value of the electricity no longer consumed:
D – (C + D) = -C. (4)
Hence, the actual loss, from the consumer perspective, is not equivalent to the net increase in expenditure (A-D), but rather the increased expenditure for electricity still being consumed (Rectangle A) plus the net loss in value of the electricity no longer being consumed (Triangle C).
In numeric terms the net consumer surplus calculation is represented as:
(Old Price – New Price) x (New Hourly Consumption) [Rectangle A]
+ ½ (Old Price – New Price) x (Change in Consumption) [Triangle C] (5)
($0.0407/kWh - $0.0494/kWh) x (1.157 kWh)
+ ½ x ($0.0407/kWh - $0.0494/kWh) x (1.327 kWh – 1.157 kWh)
When multiplied by the total number of on-peak hours (1,304), the net loss in consumer surplus totals $14.09, which is much higher than what the savings calculation suggested for this period.
However, as with the savings calculation, there is a positive change to consumer surplus in the off-peak hours, where the electricity rate has been lowered (Figure 3).
According to the savings calculation, the net benefit here is represented by:
(A + B) – (B + D) = (A – D) (6)
where Rectangle A represents the savings obtained by using the original level of off-peak electricity usage at a lower rate, and Rectangle D represents the extra expenditures arising from buying more electricity because of the rate change. But again, net consumer surplus calculations tell a different story. As with the on-peak case, both methods of calculation are in agreement that the savings (or losses) which stem from a price change for electricity still used represent a direct change in consumer surplus. But unlike the on-peak case, now more electricity is being used as compared to when flat pricing was used, because of the reduced off-peak rate. The total value of this electricity, as before, is equivalent to the area under the demand curve, bounded by the old and new usage levels, which is represented as the sum of Triangle C and Rectangle D. While the consumer is spending more to make these additional purchases, the incremental expenditure is exceeded by the value of the electricity, yielding a net benefit:
(C + D) – D = C (7)
and so this time the net change in consumer surplus is equivalent to the sum of Rectangle A (the savings from purchasing the original off-peak level of electricity at a new, lower price) and Triangle C (the net benefit of buying additional electricity at this lower price).
In numeric terms, this net consumer surplus calculation is represented as:
(Old Price – New Price) x (Old Hourly Consumption) [Rectangle A]
+ ½ (Old Price – New Price) x (Change in Consumption) [Triangle C] (8)
($0.0407/kWh - $0.0367/kWh) x (0.851 kWh)
+ ½ x ($0.0407/kWh - $0.0367/kWh) x (0.859 kWh – 0.851 kWh)
When multiplied by the total number of off-peak hours, the net gain in consumer surplus for the year is $25.50.
Combining the net loss in consumer surplus which is incurred in the on-peak period with the net gain in consumer surplus incurred in the off-peak period yields a total of $11.41. So while there is a net benefit from a consumer perspective in using a time-of-day pricing mechanism, this net benefit is not as high as a simple savings calculation would indicate. In fact, in this example the savings calculation overstated the benefit by over 67%. While the magnitude of the actual numbers seem small in this example, it must be remembered that this is a per customer estimate, and when multiplied by thousands if not hundreds of thousands of customers, the incorrect method of benefits estimation could overstate aggregate consumer benefits by millions of dollars.
Defenders of the savings method might concede its incorrectness, but counter that it is a simple and direct approach to getting some type of estimate of how the change in electricity consumption behavior will produce a benefit. But the calculations outlined above are actually very straightforward, and require little or no additional information than that used to calculate net savings. All that must be known is the original flat price, the new time-of-use prices, and total usage within each pricing period (e.g., peak and off-peak) before and after the new pricing was introduced. A final objection might be that these equations assume, as the diagrams suggest, linear supply and demand curves, which do not reflect the actual response of supply and demand to pricing. But the same calculations could be performed using curves rather than lines, with just a little bit of higher mathematics involving the calculations of areas under curves. And the equations used above only assume that linear – or near linear – behavior is occurring in the region of the curves where the prices are changing. This is a simplifying assumption, but the estimate produced will still be far superior to one based on a savings calculation, which can produce results that are off by orders of magnitude, or – even worse – results that indicate a net consumer benefit where none has been realized (as in the coffee example), because the savings calculation is fundamentally wrong to begin with.
(It should also be mentioned that there may be other stakeholders that will be impacted by a change in retail electricity pricing practices, such as producers and/or wholesale sellers of electricity. For these, calculations of changes in producer surplus can be derived in a very similar manner to the calculations for consumer surplus above.)
When developing a business case for investment in grid modernization, it is essential that the case be grounded on solid economic principles, for the benefit of the utility and its ratepayers, as well as the regulators (and interveners) who will review it. The preceding discussion is intended to illustrate the error in equating benefits with the savings arising from changes in hourly electricity usage patterns as a result of dynamic pricing. However, while the example provided in this paper clearly illustrates the risks of benefits overestimation, it should not cause one to jump to conclusions about the value of dynamic pricing and investments in smart meters. As mentioned at the beginning of this paper, there are other benefits that arise from the implementation of smart meters, including operational savings (from, for example, reduced meter reading costs, fewer losses from meter error or energy theft, and lower call volumes) and a reduced need to add generation and/or transmission capacity to meet future growth in peak demand. These benefits alone could easily eclipse any direct benefits associated with changes in energy usage. It should be noted as well that the example above is just that, and may not represent the actual extent to which residential customers will change their hourly consumption if real-time or peak/off-peak rates are introduced. A greater level of demand response (i.e., higher price elasticity) might result in benefits that are significantly higher than the ten or twenty dollars per year that was produced in the example. But even if the direct benefits from changes in hourly usage are small, this only makes it more imperative that they are calculated correctly. Because any business case, when scrutinized by potentially critical third parties, is only as strong as its weakest link, and with so many large benefits clearly arising from smart meters and demand response from other sources, it would be tragic to ruin the credibility of an entire business case by miscalculating this particular element of it.
* Director of Economics, Edison Electric Institute 701 Pennsylvania Avenue, NW, Washington, DC 20004 - firstname.lastname@example.org