The Importance of Marginal Cost Electricity Pricing to the Success
of Greenhouse Gas Reduction Programs
By Lee S. Friedman*
Professor of Public Policy
Goldman School of Public Policy
University of California
Berkeley, CA 94720?7320
This paper identifies a previously unnoticed problem important for the efficient reduction of GHG
emissions: the pricing of off?peak electricity. There may be many opportunities to reduce emissions by
substituting relatively clean and inexpensive off?peak electricity in place of a more GHG?intensive fuel,
with the most important example being vehicle electrification. However, off?peak electricity for
residential consumers is currently priced at 331% above its marginal cost in the United States as a
whole. Even for the 1.3% of residences that are on some form of time?of?use rate schedule, the off?peak
rate is still almost three times higher than the marginal cost. A primary barrier to the increased use of
marginal?cost based TOU rates is that less than 5% of U.S. households have the “smart” meters in place
that can measure and record the time of consumption. A FERC report estimates the deployment rate of
these meters for residences by 2019 will reach only 40%; policies should be put in place so that this
deployment rate nears 100% by that time. Another important barrier is TOU rate design. I describe two
TOU rate designs (baseline and two?part tariff) that utilize marginal?cost based rates, ensure appropriate
cost recovery, and minimize bill changes from current rate structures. A final barrier is to get residences
on to these rates. I consider whether a marginal?cost based TOU rate design should remain as an
alternative for which residences could “opt?in,” or become the default choice, or become mandatory. I
conclude that it should become mandatory. Time?invariant rates are a historical anachronism that
subsidize very costly peak?period consumption and penalize off?peak usage to our environmental
detriment. There is no reason for such a system to continue.
This paper has been prepared for presentation on November 6, 2009 at the Annual Research Conference
of the Association for Public Policy Analysis and Management, Washington, DC. Comments welcome.
*I am grateful to Carson Christiano for excellent research assistance throughout this project. I am also grateful to
Eugene Smolensky and Judi Greenwald for their comments on a preliminary draft.
The Importance of Marginal Cost Electricity Pricing to the Success
of Greenhouse Gas Reduction Programs
By Lee S. Friedman
From the time perspective of 2009, most of the world’s science and policy communities have come to
accept both that the globe is warming and that this is largely caused by the emissions of greenhouse
gases (GHGs) as byproducts of human commerce. Furthermore, there is recognition that any increase of
more than 3 degrees centigrade by the next century poses a severe danger, perhaps catastrophic, to the
world’s ecosystems. Many nations are struggling to design and implement policies to prevent such an
increase. To achieve this, current thinking is that by 2050 something like a 50?80% global GHG reduction
from the 1990 level is necessary. The problem is greatly complicated by its nature as a global public
good: reductions in any one country do not work to cool that specific country’s climate, but are a
general cooling force throughout the globe. Thus the willingness of one nation to undertake reductions
depends on the willingness of others to contribute as well. I wish to focus here on the type of regulatory
systems that countries will use to achieve reductions, if they are willing to make them. The more
efficient the regulatory systems, the lower the reduction costs and the more likely global participation
can be negotiated and carried out.
A prevailing view from economics that focuses many of these policy efforts is that the GHG emissions
are negative externalities, and that therefore corrective policies should effectively internalize them.
Because the sources of these emissions are so varied, market?based policies like a carbon tax or a GHG
cap?and?trade program are typically centerpieces of the recommended corrective efforts. The European
Union has already implemented a GHG cap?and?trade program and Australia is about to do the same. In
North America, the Regional Greenhouse Gas Initiative (RGGI) is using cap?and?trade to limit emissions
from electric?generating plants in the northeastern U.S., California is implementing a cap?and?trade
program to reduce GHGs that will cover over 80% of its sources, and it is coordinating with other
western states and several Canadian provinces to form the Western Climate Initiative that plans to
operate a regional cap?and?trade program—a much larger trans?national version of the California
system. At the U.S. federal level, the Obama administration supports a national cap?and?trade program,
the Waxman?Markey bill that offers one has passed the House, and it is now up to the U.S. Senate to
take the next steps if successful legislation is to result in the current session.
The analysis of this paper is intended to identify a different problem that to my knowledge is not
currently being addressed by any nation, but is also crucial to the success of GHG reduction efforts. The
problem is the proper pricing of electricity, apart from how its price may be affected by cap?and?trade
programs.1 In the usual analysis of negative externality problems, the initial price of one activity
(emitting GHGs) is too low relative to all other prices. But the other prices are generally assumed to be
set appropriately. However, that is not the case with electricity. Electricity prices to its consumers are
almost never equal to the marginal costs of providing it, and actual prices are often orders of magnitude
1 The latter is also an important issue, but will not be the focus of this paper. Retail electricity prices should rise by
the marginal cost of GHG allowances necessary to generate it. However, most retail rates are set at average cost
rather than marginal cost, and if electricity distribution companies effectively receive revenue to offset allowance
costs (e.g. by free distribution of the allowances to them, or by earmarking some portion of auctioned allowance
revenue to them), then even their average costs may not rise by the full value of the allowances.
away from the appropriate marginal cost. Failure to fix this problem can seriously impair efforts to
reduce GHG emissions.
A simple example can suggest the seriousness of the mispricing. One of the most promising new
technologies for reducing GHGs is the electric vehicle. In California, it has been estimated that compared
to conventional gasoline, delivering the equivalent power to the wheel of a vehicle by electricity reduces
GHG emissions by 75%.2 Of course if the cost of using electricity is much more expensive than gasoline
(here assuming that each fuel must secure the necessary GHG allowances for its emissions and that
those are included in the cost), then few consumers would choose to use it (some who value the GHG
reduction highly might use it anyway, for the same reasons that some people voluntarily provide
support for other public goods). Based on current technology, several researchers have estimated
“break?even” prices—for a given gasoline price, the electricity price that makes the marginal cost per
mile the same. For example, Kammen et al (2008) estimate that gasoline at $5.00 per gallon is
equivalent to electricity at $.10 per kilowatt?hour (kwh), at current (high) battery prices of $1300/kwh.
However the actual marginal cost of using off?peak electricity to recharge a vehicle in California is only
on the order of $.03/kwh, based on recent off?peak market wholesale prices. Even with current battery
technology, consumers should be indifferent between using electricity and gasoline at a gasoline price of
$1.50 per gallon. Since gasoline is about $3.00 per gallon at the time of this writing, off?peak electricity
charging of vehicles should seem like a great bargain compared to gasoline. One problem is that virtually
all residential customers in California are on rate schedules in which prices are the same throughout the
day, and far above off?peak marginal cost. According to a FERC survey that we will discuss in more detail
later, less than 1% of California’s residential consumers are on rate schedules that vary prices within a
day (which we shall call time?of?use or TOU schedules, because we mean to encompass not just simple
time?of?day pricing that is the same from day to day but also more sophisticated, dynamic variants like
real?time pricing or critical peak pricing).3 While the rates that California residences face are tiered so
that they increase with consumption, the average residential rate in the year ending May 2009 was
about $.15 per kwh, five times greater than the marginal cost.4 Many of these customers face actual
marginal rates of well over $.30 per kwhr, or prices more than 10 times higher than a marginal cost
To put this in a pocketbook perspective, assume a compact plug?in hybrid vehicle is driven 1000 miles
per month and that half of that or 500 miles is powered by electricity. With current technology the PHEV
gets about 4 miles/kwh, or it will need 125 kwhs of electricity each month. On Tier 3 of our
representative rate schedule, this is $47.33 extra on the monthly electricity bill or $568 annually,
whereas the off?peak marginal cost of providing it is only $3.75 per month or $45 annually. The
2 The reduction would not be as great for jurisdictions with coal?fired electricity as the marginal source, but it
would still be significant.
3 The survey we are referring to is the December 2008 Assessment of Demand Response & Advanced Metering by
the Federal Energy Regulatory Commission. The California night marginal cost is estimated by using the average
off?peak price of $.028/kwh. reported by the California Independent System Operator (CAISO) for April?June 2009.
4 The California Energy Commission has a table on its website “Average Retail Price of Electricity to Ultimate
Customers by End?Use Sector, by State (EIA)” in which it is reported that California residential customers had an
average rate of $.1464 per kwh in the year ending May 2009.
5 For example, customers of PG&E on the most common residential rate schedules (E?1, EM, ES, ESR, ET) have an
average rate of $.17643, but those customers who are between 201?300% of the baseline quantity pay a marginal
rate of $.37866 and those 300% above or more pay a marginal rate of $.44098. These rates are from the schedule
on the PG&E website described as in effect from March 1, 2009 through the present (as of Sept. 2009).
difference in the attractiveness of these two amounts to consumers, and therefore their willingness to
consider purchasing plug?in hybrids, is obvious. It is also very important to note that I am not suggesting
here that anyone subsidize electricity; I only want it priced at its marginal cost.
A number of questions can and should be raised about this illustrative example. Is it true that the off?
peak marginal cost is only about $.03 per kwh? Wouldn’t consumers simply switch to a TOU rate
schedule? This paper attempts to give more detailed answers to these questions, although careful
answers remain as disturbing as the example. I review evidence about actual off?peak marginal costs
throughout the U.S., actual available rate structures, and obstacles to marginal cost off?peak pricing. I
conclude that, under status quo policies, there is substantial reason to be concerned about the gap
between actual consumer rates and off?peak marginal costs. I then consider how policies might address
the problems identified. I consider metering issues that impede the use of TOU rate schedules. I
consider better TOU rate design so that rates are closer to marginal costs. Finally, I consider as options
whether an appropriate marginal?cost based TOU rate structure should be an available alternative
should a residence choose to “opt?in”, the default alternative that would be chosen unless the residence
“opts out,” or mandatory.
II. Is There Really a U.S. Problem of Mispriced Off?Peak Electricity?
A. The Historical Argument for Marginal Cost Electricity Pricing
Long before global warming became a known issue, economists have pointed out the
inefficiency caused by the mismatch between electricity’s rates and the highly variable marginal cost of
providing it. The 20th century argument was usually in the context of rates set approximately at average
cost (under rate?of?return regulation), whereas actual marginal cost during peak periods was well above
this average, and actual marginal cost during off?peak periods well below the average. This rate
structure resulted in the building of many high marginal cost plants designed only to operate during
peak periods, with substantial unused capacity during the off?peak. There was inefficient
overconsumption of electricity during the peak, and inefficient underconsumption during the off?peak.
TOU prices, on the other hand, would give consumers incentive to reduce these inefficiencies by
conserving more during peak periods and shifting load to off?peak periods. The arguments for the more
sophisticated variants of simple TOU pricing, like real?time pricing where rates could vary from minute
to minute and day to day, were simply extensions of the same argument: efficiency requires that
customers always face rates equal to marginal costs, and that marginal costs during all or part of
unusual days (e.g. very hot summer days, very cold winter days, or days with unexpected major supply
interruptions) could be many multiples of marginal costs at a similar hour during a more typical day.
This argument met resistance from the regulated utility world on two practical grounds: existing
meters were not smart enough to distinguish time?of?day let alone real?time variations, and electricity
consumers preferred the simple system where they did not have to pay continual attention to the
timing of (and varying rates for) their electricity consumption. We have come to understand the
metering question as a transaction cost that had been assumed away in the early argument, and the
consumer behavior question as one for the rapidly?developing area of behavioral economics with
consumers characterized by bounds on (and often distaste for) calculating.
The more practical economists did not give up in the face of this resistance; rather they
attempted to address it. It was true that existing electromechanical meters could not be used in
conjunction with TOU rates, but it was not impossible to build meters that could be so used. They were
expensive, but the benefits would outweigh the costs for larger customers. Similarly, even if most
residential households did not want to be bothered with electricity plans that required more ongoing
attention from them, surely larger plants with energy management departments would be interested in
opportunities to reduce their electricity expenses through more efficient rate plans. So the use of TOU
rate plans began to spread, but limited primarily to the industrial and larger commercial electricity
consumers. There was also a response to the behavioral problem that customers do not like the threat
of very high real?time rates that are sometimes observed in wholesale market prices on unusual days.
Interruptible rate plans (now often called “demand response” programs) that offered consumers lower
normal rates in return for willingness to shed load during the unusual days also had some success in a
number of jurisdictions. The behavioral genius of this plan is to reward those who conserve during peak
periods, rather than to penalize those who do not conserve. Again, these were primarily used by larger
customers but also included some residences that agreed to reduce air conditioner usage.
If we fast forward to the 21st century, metering and control technology have advanced greatly.
Smart, reliable electricity meters that can tell time and can go not only forwards but backwards (if the
consumer has some on?site electricity generation like solar that produces more than the site is using,
the excess can be sold back into the electricity market) are available at no higher cost than the older?
style mechanical meters. Computer programs and other inexpensive control devices are available to
respond automatically to increases in electricity rates and adjust electricity usage in accordance with the
prior instructions of the user. Both the transaction cost and behavioral arguments against TOU rates
have been substantially weakened by these technological advances.
For many economists, the old 20th century argument updated with 21st century technological
advances is sufficient grounds for pushing anew for increased use of TOU rate structures, and preferably
the more sophisticated kind like real?time pricing or approximations of it. Borenstein (2005), for
example, shows that substantial welfare gains would arise from increased use of time?varying rates in
competitive electricity markets, even with quite small price elasticities of demand.
Even the updated version of the old argument emphasizes gains that come primarily from
reducing loads during the peak periods. This is because these demands can be so expensive to fulfill,
because they threaten the reliability of the grid for all users, and because demand reduction during
these periods mitigates the threat of harmful exercise of market power (a feature that might have made
a difference in the California electricity crisis of 2000?20016). These are all valid and important
arguments, but I wish to add another layer to them: the social gain from increased use of off?peak
electricity as a GHG?reducing response. That is, in some cases off?peak electricity can power
something—serving as a substitute for another power source—and do so at lower social cost precisely
because it reduces GHG emissions. This may help the “peak” problem of capacity that goes unused
much of the time in a surprising way, by expanding off?peak demand and thus reducing the differential
between peak and off?peak. However these opportunities will not be taken up to anywhere near the
extent that they should if off?peak electricity is priced substantially above its marginal social cost. I have
already given the important example of vehicle electrification, but there may be many other
opportunities for this type of substitution. Suppose, for example, inexpensive and abundant wind power
supplies become available at night, meaning relatively few GHG allowances must be used for this off?
peak supply. This further reduces the off?peak marginal cost relative to alternative fuel options
(including but not limited to on?peak electricity), and some activities now using alternative fuel sources
may shift to off?peak electricity if priced at its marginal cost.
B. What Exactly is the Marginal Cost of Off?Peak Electricity?
Earlier in a California example, I referred to an off?peak marginal cost of $.03 per kwh based upon
wholesale prices. The relevant short?run marginal cost is the extra cost necessary to deliver an
6 See Friedman (2009).
additional kilowatt?hour of electricity to a customer at a particular time and location. In many parts of
the country, where wholesale prices are set by competitive electricity markets, the Independent System
Operators (ISOs) keep track of these prices as delivered to particular points on their grids; these are
referred to as locational marginal prices (LMPs). These prices depend primarily on the fuel cost at the
marginal electricity plant generating the power (e.g. coal, natural gas, nuclear), and they vary around the
country at any single time as well as varying over time at any single location due to fuel price changes.
These prices also take into account any congestion and line losses that arise between the generator and
the receiving electricity distributor.
Before going further, it is important to clarify that the retailer’s prices to its consumers must fully
recover all of its costs for the operation to be viable. However, substantial portions of the cost are not
marginal—e.g they include reimbursement for the sunk costs of investments made long ago, like plant
construction costs. Allowed cost recovery also may depend on contracts signed long ago for delivery of
electricity at fixed prices, also a sunk cost. Because these sunk costs have historically been a large
portion of the total allowed cost, recovery of them by pricing at average cost rather than marginal cost
has characterized pricing in regulated sectors. But these prices do not give the correct signals to
consumers about the cost of additional consumption. This has been a major bone of contention
between economists seeking more efficient prices (closer to marginal costs) and managers of the
regulatory practice wanting an easy?to?administer system.
Much work has been done by economists to devise workable pricing plans for regulated settings that
allow for cost recovery but have prices closer to marginal costs. Higher revenue during peak periods
tends to offset the lower revenue during off?peak periods, but in general they will not balance out
simply by charging marginal cost and some adjustment is needed so that the overall revenue
requirement is met. While I will not review all of the various pricing plans that achieve this, I will
mention the one that I consider most promising for practical applications including the off?peak problem
that is the focus here. The method is that of the “two?part tariff,” although in reality it would be a multi?
part tariff.7 With this method, all rates (think both peak and off?peak) are set at short?run marginal
costs, and the residual amount still necessary for full cost recovery is assessed as a fixed fee (spread over
all consumers). Payment of this fixed fee is mandatory for using the system, but it need not be the same
for all customers. I have shown elsewhere (Friedman and Weare 1993) that this fixed fee can be set by
dividing customers of a certain class (e.g. residential) into 5?6 consumption groups from low to high
levels, each group with its own fixed fee representing its share of the non?marginal costs, and that these
fees can be set so that virtually all customers experience little to no aggregate bill changes at their
current consumption levels.
Assuming for now that this method can be used with marginal cost prices to meet the overall revenue
requirement, are there are other costs besides the LMP that may be additional short?run off?peak
marginal cost components? There are basically two categories here: ancillary services used to ensure
grid reliability and balance, and any marginal distribution costs incurred to move the electricity from the
door of the receiving utility (where the LMP ends) to the ultimate consumer. I discuss these in turn.
The cost of ancillary services is also reported by the ISOs, although it is not clear how much if any of
these costs are incurred by incremental increases in off?peak demand (e.g. the cost of maintaining
reserve margins). For example, data from the New England ISO wholesale cost report for April?June
7 One of the first proponents of this idea was Coase (1946).
2009 in Connecticut shows that the average off?peak LMP was $.0321/kwh.8 The same table shows a
total wholesale cost of $.0401. However, almost all of the difference is due to a capacity payment that
averages to $.0078/kwh but is clearly described as Connecticut’s share of a monthly payment based on
the system?wide peak from a year earlier. None of this should be billed or attributed to off?peak hours.9
With the capacity payment removed as nonmarginal, the marginal cost including all ancillary services is
$.0323/kwh (barely distinguishable from the off?peak LMP). In general for the New England ISO, these
off?peak ancillary costs are usually below 3% of the LMP, so we will add 3% to approximate them here.
Similarly, the marginal cost at the retail level is simply the marginal wholesale cost plus marginal
distribution charges. However, most of the distribution expenses are nonmarginal; they are the fixed
costs of the low?voltage wire system, its maintenance, and the administrative costs of meter reading,
billing, and other service changes. The full average of all distribution expenses is on the order of $.017
nationwide.10 Perhaps the only clearly identifiable off?peak marginal cost is the line loss in going from
entry to exit of this system, which varies but is usually on the order of 6?7%, or $.002/kwh (ISO New
England 2009). To be conservative, let us assume that 10% of the LMP ($.003/kwh in this case) can be
considered marginal distribution expenses. This brings our estimated off?peak marginal cost for
Connecticut to $.0363, having added 13% to the LMP to account for marginal ancillary services and
Table 1 column (1) presents estimates of the off?peak marginal costs for April?June 2009 in each of the
50 states. For those 34 states served by ISOs, we found data analogous to that used in the example
above: the off?peak LMP, plus 13% to approximate the marginal ancillary services and distribution cost.
The average estimated off?peak marginal cost for this group is .02767 cents/kwh. For the 17 states not
served by ISOs, we had to use somewhat rougher data. The North American Electric Reliability
Corporation (NERC) has average wholesale price data for each of its nine regions, but not broken into
peak and off?peak. For each ISO state, we calculated the percent that the (previously calculated) off?
peak marginal cost is to that state’s NERC wholesale data using its relevant NERC region; the average of
these is 51.1 percent. We then estimated off?peak marginal costs for the non?ISO states by applying this
percentage to the available NERC wholesale data for each state’s region. This yielded an average U.S.
off?peak marginal cost estimate of .02794.11 These averages of course contain considerable variation,
8 This is the simple average of the April, May and June off?peak LMPs in $/mwh shown in Table 3.3.4 as $31.48,
$33.35, and $31.46.
9 ISO New England states that its figures do not represent billing figures. Some of the non?LMP charges are monthly
charges that it simply divides by the number of mwhs reported for that month in order to produce cost/mwh
numbers. This procedure does not attempt to distinguish what costs are appropriately charged to peak or off?peak
10 The Energy Information Administration (1997) reported that all distribution expenses were 19 percent or $.013
of the $.071 full average cost of electricity (p. 11). Updating this from 1997 to 2008 by the Consumer Price Index
yields an estimate of $.017.
11 We also tried an alternate method for estimating off?peak marginal costs for the non?ISO states shown in
column (2). This did not substantially change the results. We used 2009 data from the Intercontinental Exchange
(ICE), calculating the simple average of peak and off?peak prices for 6 hubs having both available. Compared to the
NERC data for the same six states, wholesale prices have declined by 1.935 cents/kwh from 2007. We used this
number to adjust the 2007 NERC data to be estimated 2009 wholesale prices. Then we again calculated the ratio
for ISO states of off?peak marginal costs to these estimated 2009 prices, which was 79.7%. We used this
percentage to estimate off?peak marginal costs for the non?ISO states, and derived an overall average U.S. off?peak
marginal cost of 2.844 cents/kwh.
with a high of .03627 in Florida and New York, and a low of .02076 in the states served by the Midwest
We note again that these estimates are heavily dependent upon fuel prices that have declined from
unusually high levels in 2008. For comparison purposes, we present in Table 2 the 2003?2008 annual
average off?peak marginal cost estimates for residences in the PENELEC zone (western Pennsylvania) of
the PJM system. These are the simple average off?peak LMP prices for this zone, plus 13% as before.
Over the past six years, the average off?peak marginal cost has been between $.03 and $.05 per kwh,
except when it rose to $.06 in 2008 before returning in 2009 to the lower part of its historically more
The point of this section is that appropriate off?peak rates for electricity should be based upon its
marginal costs. In 2009, these marginal costs are around $.03 per kwh in most of the country, although
they are as low as $.02 and as high as $.04 depending upon location. How do they compare to the rates
C. How Close Are Actual Rates to the Off?Peak Marginal Costs?
The next part of our task is to see how close actual retail off?peak rates are to the off?peak marginal
costs. There are two distinct parts to this task. One is to compare the actual rates that residential
consumers now face with the marginal costs. However, this comparison takes as a given the rate
structures that consumers are on at the current time. Almost all of these consumers are on the standard
rate structures that do not vary with time, and they could if motivated switch to a TOU schedule that
would have lower off?peak rates. However there are numerous obstacles and barriers reviewed below
that deter residential consumers from using available TOU rate structures and make such a switch far
more problematic than it sounds. These include meter availability, disincentives due to regulation
(including areas in which parts of the industry have been restructured), and consumer resistance due to
behavioral economics factors. We refer to the effects of these as the obstacles gap. Nevertheless, a
second comparison is with available TOU rates to our estimated marginal costs. We define any gap in
which available TOU rates exceed marginal costs as the rate gap and later consider policies that might
change the rate structure to reduce any such gaps.
Because almost all residential consumers are on time?invariant rates, average electricity rates in each
state are a good proxy for the distance between the price per kwh that they would be charged today
relative to the off?peak marginal cost. We show the average residential rates in effect for June 2009 in
column (3) of Table 1, and have calculated in column (4) the percent that these average rates exceed the
off?peak marginal cost. These are also shown in Figure 1 along with the off?peak marginal cost data, with
the data arranged from least to greatest off?peak marginal cost. The average residential rate for the U.S.
as a whole is 12.05 cents per kwh, which is 331 % above the average off?peak marginal cost of 2.79 cents
per kwh.12 There is considerable variation in these “mark?ups” by state, with our estimates showing the
lowest mark?up of 145% in Washington DC and the highest mark?up of 766% in Michigan (followed
closely by Idaho’s 756%).
12 12.05 cents per kwh is the simple average of the rates given for each of the states. The average weighted by
each state’s kwhs is almost the same, 11.91 cents per kwh (only 1% difference). The weights necessary to compute
weighted averages are not always available for our calculations.
Again, I am not saying that electricity sellers are collecting too much revenue in total; these rates are
almost always the result of a process (whether competitive or regulatory) intended to provide an
appropriate overall rate?of?return. However, I am saying that today’s consumers face highly
inappropriate rates and incentives for consuming off?peak electricity, and that such rates would be
inappropriately discouraging to anyone thinking of buying an electric vehicle like a plug?in hybrid and
recharging it during off?peak hours. Of course, if these consumers were able to switch to a TOU rate
schedule, the story might be quite different.
Column (5) of appendix Table 1 lists the off?peak rates in effect in June 2009 at 50 utilities or retailers,
chosen largely from those who had participated in the 2008 Demand Response and Advanced Metering
survey undertaken every two years by the Federal Electricity Regulatory Commission (FERC). However,
we did wish to focus on entities that served suburban populations, and that offered residential TOU
rates. For a few states in which utilities in the FERC survey reported no TOU programs, we looked for
other utilities in the same state that did have such programs. However, in 7 states neither FERC nor we
could identify utilities with residential TOU rates available.13 In one case (Maryland), FERC’s only utility
with residential TOU served a rural population, and we substituted a utility serving a suburban
population. When there was more than one utility in a state with TOU programs, we chose the one near
the largest city that included a suburban residential area in its service territory. It would be difficult for
this sample to be very different from the U.S. residential population as a whole, but it does give
somewhat higher weight to the suburban areas that should be prime targets for electric vehicles (many
vehicle commuters, many garages convenient for night recharging of electric vehicles). Five utilities in
our sample had block TOU rates (decreasing in Iowa, Oklahoma, Pennsylvania and Wyoming, increasing
in Oregon), and in those cases we report the second block which represented consumption in the 600?
800 kwh monthly range for residences.
The result of this exercise is that our estimate of the average off?peak residential TOU rate for the U.S.
as a whole is 7.452 cents/kwh, almost three times higher than our 2.794 cents/kwh estimate of the off?
peak marginal cost for the same period. This is not just a problem that rates may lag behind marginal
costs: it is also substantially higher than average off?peak marginal costs for any year from 2003?2009, if
our earlier estimates of 3?5 cents from PJM are representative. Of course there is considerable variation
by jurisdiction. Figure 2 shows these rates graphically, again along with the corresponding off?peak
marginal cost estimate from that state. Indiana, Tennessee, West Virginia and Wyoming all had utilities
with rates that were within 15% of our current estimated off?peak marginal cost for them; they are the
winners for good pricing. Of the other 40 jurisdictions, another 10 were within 100% of estimated
marginal cost. That leaves 30 with rates that are at least two times higher than marginal cost. California,
still suffering consequences from its 2000?2001 electricity crisis, is the worst at 741% above marginal
cost. It is followed by Connecticut (425%), Washington DC (384%), Hawaii (380%), and Maine (306%).
This exercise shows that it is not enough simply to get consumers off of the standard plans and on to
TOU plans. In answer to the question asked by this section: Yes, there is a problem of mispriced off?peak
electricity, and it is not a minor one when prices are routinely above competitive (marginal cost) levels
by more than 100%. The TOU plans themselves need to be reformed so that off?peak rates are kept
much closer to actual off?peak marginal costs. But what problems might there be in getting consumers
on to TOU plans?
III. The Obstacles Gap
13 The 7 states are Louisiana, Mississippi, Nebraska, North Dakota, Rhode Island, Utah, and Washington.
In this section, we begin with the results of the FERC survey and discuss physical metering issues that
prevent TOU rates from being offered. We then discuss regulatory obstacles that further retard or deter
the use of TOU rates. Finally, we discuss consumer behavior issues that lead to resistance to TOU rates.
A. The 2008 FERC Assessment of Demand Response and Advanced Metering
The U.S. Energy Policy Act of 2005 requires that FERC publish an annual report assessing U.S. electricity
demand response resources, including penetration rates of advanced meters and time?based rate
programs. FERC implements this by conducting a survey every other year, and reviewing industry
activities and regulatory actions during the non?survey years. For purposes of the survey, FERC defines
an advanced meter as one “…that records customer consumption (and possibly other parameters)
hourly or more frequently and provides for daily or more frequent transmittal of measurements over a
communication network….” These meters are sometimes referred to as “smart meters” or “advanced
metering infrastructure” (AMI) meters. While the penetration rate of these meters is increasing, in 2008
it was only 4.7 percent of residential meters (up from .6 percent in 2006). Furthermore, less than 30% of
these AMI meters are used for any kind of “price responsive demand response” program. Rather, they
are being used to improve customer service in other ways like faster outage detection and restoration.
Almost all households are metered by the older electromechanical meters that do not measure when
electricity is consumed. Some households are metered by “automated meter reading” (AMR) meters
that preceded the age of smart meters and offered some limited advances that do not necessarily
include time of day capability. Some AMR meters, for example, are read by having the meter reader
wave an electronic wand near them, which increases accuracy and speeds up the time it takes to read
the meters. The life expectancy of actual AMI meters is in excess of 20 years, and their costs are not very
different from the other meters; the Electric Power Research Institute reported that they averaged $75
per meter during 2005?06.14 However, to make full use of these advanced meters usually requires a new
communications infrastructure, and there are also one?time installation costs, project management, and
other information technology integration costs. According to EPRI and FERC(2006), this typically adds
$125?$150 to the average cost of an AMI upgrade per meter (or $200?$225 per meter in total
Given the very limited number of meters in place that are even capable of measuring consumption by
TOU, it is not surprising that very few residences are on TOU programs. We have used the FERC survey
data to calculate, for each state, the percent of residences that are on any type of TOU program
(including the more sophisticated variants like critical peak pricing and real?time pricing). For the U.S. as
a whole in 2008, only 1.3 percent of residences are on time?varying plans. While there is of course
variation by state (Table 3), the only state that exceeds 5% in the FERC data is Arizona with 23%.
The FERC survey, while the best available evidence on these developments, does have some reliability
issues and was not designed to produce accurate numerical results by state. While it requests data from
virtually all U.S. entities (n = 3407) providing electricity service, only 55% of these responded to the
demand response portion of the survey (and 60% to the AMI section). Nevertheless, FERC reports that
the respondents cover 91% of all electricity meters in the U.S., and that it finds no evidence of selection
bias in the results (e.g. little regional discrepancy in response rates). Still, we found certain anomalies in
14 EPRI (2007).