The Electricity Journal 29 (2016) 31–41

Contents lists available at ScienceDirect

The Electricity Journal

journal homepage: www.elsevier.com/locate/electr

Optimizing prices for small-scale distributed generation resources: A

review of principles and design elements
Amparo Nieto
NERA Economic Consulting, 777 South Figueroa Street, LA 90017, USA

A R T I C L E I N F O A B S T R A C T

Article history:
Received 1 March 2016 Regulators worldwide are considering many proposals to improve pricing schemes for low-carbon
Accepted 7 March 2016 distributed electricity resources in order to achieve compensation that is fairer and better aligned with
Available online xxx the incremental benefits to the utility and society. The key concern is that under the current practice of
net metering, the expansion of renewables-based, small-scale distributed generation is at odds with the
Keywords: utility’s efforts to provide service on a reliable, efficient (least-cost) and equitable basis.
Energy ã 2016 Elsevier Inc. All rights reserved.
Tariffs
Marginal cost
Distributed generation
Solar PV
Storage

1. Introduction systems, the growth of renewables-based distributed generation

The U.S. has moved a long way towards the restructuring of its
energy sector. Today, the push for a cleaner environment and the
subsequent expansion of renewable energy generation resources is
changing the dynamics of both wholesale and retail electricity
markets, and is demonstrating the need for further regulatory
reforms. The intermittent, rather unpredictable output of renew-
able-based generation presents short term reliability challenges
and may dramatically affect long-term resource planning deci-
sions. At the wholesale level, new mechanisms to incentivize
“flexible” generation capacity resources are being developed to
address the sudden and potentially large fluctuations of aggregate
generation supply in order to ensure system stability.1 At the retail
level, regulators are developing new pricing measures for low-
carbon distributed electricity generation resources such as solar
Photovoltaics (PV) and wind, that allow for fair compensation, one
that better aligns with the incremental benefits of these resources
to the utility and society, so as to not unduly burdening other
customers. The main concern is that under the current pricing
(DG)2 is increasingly at odds with the utility’s efforts to provide
service on a reliable, efficient (least-cost), and equitable basis.
In the case of solar DG, the cost of integrating residential and
small commercial customers’ generation systems is often social-
ized among all customers. Among the integration challenges are
the increased ancillary services often required to balance the
excess output sent to the grid in real time, and the potential
acceleration of upgrades to distribution facilities in certain areas to
handle reverse flows, which raises the cost of delivery. More
notably, there may be severe adverse equity impacts due to the
implicit rate-funded subsidy mechanism employed in most states
to compensate for the output of small-scale solar or wind
generation systems. These facilities are subsidized by the practice
of net metering, along with ratepayer-funded rebates and federal
tax incentives. Currently, 44 states and the District of Columbia
have adopted formal net metering policies which typically is
available to residential and commercial solar systems of up to 1
MW of demand. Under this billing practice, the distribution utility
has the obligation to buy any exported generation from the

E-mail address: amparo.nieto@nera.com (A. Nieto).
1
This may include the need to acquiring ramping resources, and other
dispatchable generation required to integrate high levels of variable generation
reliably.
2
Distributed generation (DG), also known as embedded generation, is located at
or near the point of power consumption and typically meets a portion of the
customers' daily electricity needs or provides backup service to customers that
need highly reliable power.

http://dx.doi.org/10.1016/j.tej.2016.03.004
1040-6190/ ã 2016 Elsevier Inc. All rights reserved.
32 A. Nieto / The Electricity Journal 29 (2016) 31–41
Fig. 1. Average residential load before and after PV, SMUD July 2014.
customer’s generation facility at the prevailing retail per-kWh rate,
regardless of the value of that energy to the grid.3 The cumulative
excess energy delivered to the distribution grid in one month is
netted against the customer’s energy purchases from the grid the
same month, and any excess is carried over to offset consumption
in the next billing cycle. In essence, the electric meter runs
backwards when the customer system generation sends surplus
generation to the utility grid. Because the standard rate applicable
to residential and small commercial customers is generally
structured as a two-part rate, with a low monthly fixed charge
and a flat per-kWh charge that does not change by time of day, 4 the
tariff revenues from customers under net metering decline faster
than the cost savings realized by the utility or distribution owner.
The subsequent net revenue gap must be collected from all
customers in the rate class at the next general rate case to keep the
utility financially whole, thus leading to cross-subsidies.5
From its onset, the practice of net metering was known to
provide a subsidy to DG, funded by regular, full-service customers.
Installation costs for utility-scale solar PV projects and for behind
the meter residential and commercial rooftop solar have plum-
meted in recent years. As the costs of solar technologies fall and the
net metering share of the utility’s customer base expands, in some
cases, exponentially, the validity of net metering as an appropriate
tool to stimulate small-scale renewable DG has been called into
question. Not only are there equity concerns about the intra-class
cross-subsidies inherent in net metering, but also concerns that
other clean energy resources, in particular utility-scale solar and
3
Under net metering, a single bi-directional meter registers the energy exported
from the on-site system to the grid.
4
Rates that have an inverted kWh block structure may be particularly
suboptimal, providing a large credit to net metered usage. For example, SDG&E’s
residential and small solar commercial customers may receive for excess energy an
average of over 22 cents/kWh under net metering.
5
Cross-subsidies can be determined through comparing class revenue target
with actual tariff revenues plus avoided costs from DG.
wind turbines, do not receive comparable subsidies.6 The fact that
customers may not be acquiring the least-cost renewable energy
available raises controversy as to how much renewable DG is
desirable.
A number of states are now evaluating rate reforms that seek to
reduce or eliminate the subsidies to renewable DG customers
provided by net metering. However, regulators are finding it very
challenging to agree on a “fix” that is considered fair to all grid
users, and does not conflict with federal or state environmental
policy or energy conservation goals. In order to facilitate
consensus, first there needs to be a clear and widespread
understanding of the goals of regulatory energy pricing. The goal
behind a rate reform is not to hinder DG development, but to
promote a more optimal expansion of DG that benefits all
customers in the long term. Next, there needs to be a detailed
evaluation of the distortions embedded in the utility’s existing
pricing scheme for both bundled service rates and network rates,
and to what extent such distortions can be mitigated given the
available metering technology. Ensuring a system of tariffs that do
a better job at signaling the costs effectively imposed on the system
by load growth, or equivalently, the direct financial benefit to the
utility from load reductions, is critical to meet best practice rate
making goals. Any subsidy aiming to foster a target amount of
distributed solar generation on account of its clean attributes is
more appropriately channeled outside of the retail tariff or
unbundled distribution charges.
The right tariff revisions will not only promote efficient DG
development, but will also support the development of more
optimal amounts of other “distributed energy resources” (DER)
such as residential and commercial demand response, electric
vehicles (EV), behind-the-meter energy storage such as battery
6
A federal investment tax credit under the 2005 Energy Policy Act offers up to a
30% offset against all solar generator installation cost. Wind turbines receive a
production tax credit. These credits are expected to be gradually phased out by
2022.
 A. Nieto / The Electricity Journal 29 (2016) 31–41
systems or thermal ice storage, and microgrids. Rates that properly
reflect the underlying marginal costs of service will help ensure
that consumer decisions to install rooftop solar or wind turbines
are based on economically efficient incentives and do not lead to
uneconomic bypass7 of the utility system.
2. The case for rate restructuring
Revisions to retail and network tariff structures are urgently
needed to prevent growing and potentially unsustainable intra-
class subsidies. There is limited value to regulatory methods that
leave suboptimal rates in place, as observed in prior utilities’
attempts to handle revenue erosion challenges. A mechanism
called “revenue decoupling” has been used in the past to break the
link between a utility’s revenue and the amount of energy it sells to
the customer, in the context of energy efficiency and demand
response programs, and is now being proposed in some states to
deal with net revenue losses from DG. The decoupling surcharge is
meant to ensure that the utility collects its authorized fixed costs in
between general rate cases. Thus, this mechanism removes the
disincentive for a utility to encourage energy conservation or
distributed generation in any given period of time. Decoupling is
useful to effectively correct the short-term utility revenue gap as
energy sales decline, but, since it does not require changing the
basic rate structure, it does nothing to reduce the underlying cross-
subsidy problem, or to encourage the development of cost-
effective DG.
2.1. Time-differentiation: the key starting point
An optimal rate aims to minimize the social deadweight loss
created by prices that encourage increased usage when it is more
costly to serve (on-peak) and lower usage when it would be
cheaper to serve (off-peak). In general, many of the rate reform
proposals put before regulators to date have focused on addressing
the cross-subsidization issue via a “fixed vs. variable” cost split,
generally assuming that all “fixed” embedded costs must be
recovered through fixed charges. There is normally little consider-
ation given to whether volumetric charges will provide efficient
price signals that change with marginal cost changes throughout
the day. If not done correctly, a simplified ‘fixed-variable’ fix may
actually introduce another type of intra-class cross-subsidies by
further undermining incentives for efficient usage patterns.
The main culprit of the current distortions in retail energy
residential and small commercial rates is precisely the traditional
emphasis on overly simplified structures and prices that do not
vary by time of day. With no time of day charges, customers with a
higher percentage of their monthly electricity consumption in the
high cost (peak) hours relative to the average customer are
subsidized by those customers with a lower percentage of monthly
usage in the peak hours. The absence of time of day energy charges
means that capacity costs, along with energy costs, are recovered
uniformly across all hours of the day. In that case, a higher-than
class’ average load factor customer may subsidize customers with a
lower load factor, even though they all may be consuming the same
amount of energy at the time of system peak.
The presence of renewable generation such as solar PV
exacerbates these intra-class problems because solar customers
reduce sharply their load factor upon installation of solar panels.
The total energy that solar customers require from the utility or
competitive retailer drops, while their maximum, coincident peak
7
Uneconomic bypass may take place if the utility’s marginal cost of serving the
customer is lower than the per-kWh retail charge that the customer saves when
self-generating, overstating the actual avoided cost to the system.
33
demand is barely reduced as compared to pre-solar demand levels.
Thus, the customer contributes roughly the same to peak
infrastructure (generation capacity and network) needs as
compared to prior to installing solar. This is the case in winter-
peaking regions where the solar generation profile does not fit load
well, but also in summer peaking regions, where the lack of perfect
alignment of solar generation with the hours of system peak
largely disqualifies solar facilities as a firm peak-shaving resource.
Fig. 1 illustrates this phenomenon. The solid line shows a
residential customer’s average hourly load pattern in a summer
month in Sacramento Municipal District (SMUD). The solar PV
production is given by the dotted line, and the dashed line shows
the pattern of consumption that the utility sees as a result of solar
system installation. The excess solar output that flows back to the
grid, represented by the negative net energy usage in the chart,
begins to decline in the middle to late afternoon, corresponding
with SMUD system’s peak period when system generation costs
and delivery costs may be at their highest. By 4 pm, most solar
homes generate enough power to meet their own needs, but not
enough to export power back to the grid, precisely when excess
power would be most beneficial to the grid. The peak of customer’s
solar output in the summer generally takes place in the late
morning and early afternoon, which tend to be the off-peak hours
of the utility system. This is also when solar homes send the most
extra power to the grid.
The lack of time-differentiation means that DG customers have
little incentive to control their consumption in the late afternoon
hours when their production is tapering off but the utility’s costs
are still high. If solar PV output does not have a meaningful effect
on annual system peak demand, it does little to enhance system
reliability, or reduce expected annual loss of load probability
(LOLP) or shortage costs. Thus, it does not avoid transmission or
distribution network expansion, or lower the cost of energy and
capacity market purchases that the utility may need to meet
expected load growth. If small-scale solar customers were able to
access the wholesale market (directly or through third-party
aggregators) and bid an expected profile of excess generation into
day-ahead or real-time markets, the value of solar to the grid at a
given time, net of any ancillary services, would be immediately
monetized by solar owners. Lacking that ability, retail tariffs are
left to fulfill the market value-discovery role, if only averaged by
time of day within a season and over the years the rates will be in
effect. A retail tariff with time-differentiated charges will be more
likely to encourage more installation of solar panels with the best
possible orientation to maximize peak period production8 and
provide incentives to all customers to control their peak loads.
Consumers will make more efficient energy consumption and DG
investment decisions when the prices they face for electricity
reflect the marginal costs of using more or less electricity at any
given time. Efficient consumption patterns and levels are a
prerequisite for optimal system expansion and resource allocation,
ultimately enabling lower cost of service.
2.2. Marginal-cost based rate structures
A number of alternative rate designs can be adopted to improve
incentives for development of efficient DG and other DER. The final
structure will be influenced by regulatory priorities and con-
straints in each state, but the principles should not fundamentally
differ. To ensure that both utilities and consumers have the right
8
The orientation of the solar panels makes a difference. Most solar panels point
south since they maximize year-round solar power production, generating about
10–20% more than west-facing panels. However, west-facing panels are better
aligned with the later afternoon (system peak) hours.
34 A. Nieto / The Electricity Journal 29 (2016) 31–41
Table 1
Cost-reflective rate structure.
Rate components
Energy charge On-peak demand charge
(c/kWh) ($/ kW)b
Marginal cost componenta Peak Off-
Peak
Energy cost (LMP) U U
Generation capacity U
Transmission U
Distribution substation and primary U
lines
Local distribution facilities
Customer costs
a
A revenue reconciliation factor would be needed to meet total class' allocated costs.
b
It may be replaced with a “super-peak” per-kWh charge.
financial incentive to encourage or engage in DG or other DER, key
objectives to guide ratemaking reforms should be as follows:
(a) Rate structures should ensure that all customers see economi-
cally efficient price signals for usage and distributed generation
investment decisions, by time of day and season.
(b) Rates must provide enough revenue such that the utility can
earn a reasonable return on prudent investment over time. This
means that changes in tariff revenues track as much as possible
the on-going changes in costs of service.
(c) Rates should be equitable, meaning customers are placed in the
right rate class and are charged on the basis of the energy,
capacity and customer-related costs they impose to the system.
(d) Rates should be affordable, meaning that major rate changes
may need to be gradually phased in to avoid unacceptably large
bill impacts.
Ensuring stability in rate-setting methods and transparent
criteria to trigger tariff adjustments is often cited as another key
ratemaking goal, since it promotes trust in the regulatory system
and improves price predictability. Notwithstanding this principle,
major breakthroughs in technology such as those introduced by
smart meters and communications infrastructure may warrant a
tariff overhaul to maximize efficiency gains and improve fairness.
Preserving gradualism and regulatory trust in that context may
require grandfathering rates for existing DG owners, in the event
that a change in the rate setting method will significantly reduce
the compensation for energy exports. The specific term over which
existing rates will be grandfathered should be determined on a
case by case basis, to balance the goals of providing reasonable
payback certainty for PV investors and to align the compensation
for solar exports with the evolving value of distributed solar to the
system.
For maximum efficiency in price signals applicable to all
customers, time-differentiated per-kWh charges should be set as
close as possible to the underlying marginal unit cost in each
period. A marginal cost analysis seeks to estimate the impact on
costs experienced by the utility or the unbundled default service
supplier, and distribution owner, from a customer addition and
changes in customer usage at different times of day. A marginal
cost study unbundles different components of cost of service- i.e.,
generation, transmission, ancillary services, distribution substa-
tion and trunkline feeders,9 local distribution facilities and
customer services—according to what triggers these costs by
customer group and voltage level. As more customers install DG,
9
The trunkline feeder is the segment of the primary line from a distribution
substation to the point where it branches to create a primary tap line, to which
primary customer connects.
Per-connected or contract demand charge Monthly customer charge
($/max kW) ($/mo)
U
U
more unbundled rate components will help address the increas-
ingly divergent load patterns of customers within the class more
effectively, and thereby minimize cross-subsidies.
With these objectives in mind, an optimal rate requires a
multipart structure that follows the underlying structure of
marginal cost of service, as illustrated in Table 1.
2.2.1. Local facilities charges per kW of contract capacity
When unbundling the distribution cost, it is essential to
distinguish between the elements of the distribution infrastruc-
ture that are sized according to ongoing changes in a distribution
area’s coincident peak demand (distribution substations and
upstream primary feeders) versus the local facilities (local primary
lines, secondary lines, and transformers), which are sized to
specifically meet the total non-coincident demands of the
customers served from them. Substations and upstream high-
voltage distribution feeders are designed to serve diversified loads,
while the local facilities are typically sized to serve the maximum
expected demand (the design demand) of the customers expected
to be connected to it over the service life of the facility.10 The
typical per-kW cost of connection for a customer class will
generally vary by service area (rural vs. urban) number of
customers with access to natural gas for space heating versus
“all-electric” customers, and other variables such as presence of
electric vehicle charging. Because the utility does not expect to
expand the size of the transformer over time as the customer’s
demand grows, all connected customers, including those with DG,
should face a fixed monthly local facilities charge levied on an
estimate of their design demand.11 Ideally, the billing systems
would include a record of each customer’s design demand (or
contract kW). If there are no records of design demand, the billing
contract demand could be set at the highest metered non-
coincident demand of the customer over a sufficiently long recent
period (e.g., the past 18 months), as a proxy for the amount of load
that the distribution transformers must stand ready to withstand
at any given time.12 In general, if a customer’s maximum demand
exceeds the assumed design demand at any time, the billing
demand should be reset at the higher level. A customer’s non-
coincident demand may be larger than his demand at the time of
system peak.
10
The diversity of customers’ loads is taken into account when developing the
required size of the facility but diversity may not be a large factor if there are only a
few customers connected to the transformer.
11
The type of distribution connection policy in place will determine the local
facilities costs that are to be recovered in rates as opposed to up-front.
12
If metering is not available, the standard design demands used by planners for
buildings of specific sizes can be used as a proxy.
 A. Nieto / The Electricity Journal 29 (2016) 31–41
An important caveat regarding defining a DG customer’s design
demand is that DG customers should be treated the same, for
reasons of equity, as non-DG customers. Using a contract demand
charge based on the customer’s generation nameplate capacity is
not appropriate. A solar DG customer’s billing demand for a local
facilities charge should reflect the maximum load that the
customer may potentially place on the local distribution system,
because the transformer needs to be sized large enough to handle
the customer’s load when the DG system is not operating.13 This
requires looking at the sum of the energy delivered to the customer
from the grid and the energy generated by the solar system that is
consumed on-site. In addition, the local distribution system needs
to absorb the excess energy produced by solar panels and unused
locally. Thus, the billing determinant for local facilities charges in
rates specifically set for DG customers should take into account
whether the volume of export energy from their facilities is large
enough to add stress to the distribution facilities, after taking into
account the share of export power absorbed by other residents
served from the same transformer.
Temporary non-coincident demand reductions due to milder-
than-average weather conditions or installation of renewable,
intermittent self-generation do not free up space in the
transformer on a long-term basis. A customer may be entitled to
a lower than the class average design demand in cases where the
utility has installed a circuit breaker that limits access to the grid to
that amount, or if the customer adopts an automated control
system that enables a permanent lower maximum demand.
2.2.2. Fixed (customer) charge
Fixed charges should ideally be set to recover the costs that do
not change when a customer uses more or less electricity or
demand after connecting to the grid. These costs may include the
marginal customer-related expenses such as installing, operating
and maintaining the meter and service drop, meter reading and
billing activities, and the cost of maintaining customer commu-
nications. They may also include a fixed monthly local distribution
facility price component, based on the average class non-
coincident demand if there are relatively homogeneous customers
within the class. In practice, many utilities have residential fixed
charges that barely recover the direct customer-related costs. In
some states residential charges are as low as $5 per month, when in
fact marginal customer-related costs may be as high as $20 per
month or $40 once the monthly distribution facility cost is added.
Although marginal costs is the starting point, the fixed charge
may be adjusted to reconcile the difference between the revenue
that would result from setting all rate components at the estimated
marginal unit costs and the revenue requirement established for
the class. Reconciling the marginal-cost-based revenue gap by first
adjusting the fixed charge is the right design approach because
fixed charge changes do not typically affect electricity usage
decisions considerably. In general, keeping the volumetric charges
– those that vary with energy consumption and any metered (as-
used) peak demand charges – as close as possible to marginal costs
is more important to ensure that customers do not have an
overstated or understated incentive to conserve. The same
efficiency principle applies in the context of addressing rate
reforms in the presence of renewable DG. Accommodating cost-
effective DG may require raising fixed charges to bring volumetric
charges closer to the underlying time-differentiated marginal
costs.
13
Large co-generation systems owned by commercial and industrial customers
are generally assessed standby rates that reflect the costs of using the distribution
grid as a backup. Small-scale and renewable DG facilities are normally exempt from
this rate.
35
Customers should not be discouraged from using electricity
that would have more value to them than it actually costs the
utility to supply. There may be, however, situations where demand
growth has been unusually slow, and the ample capacity margins
in the region or service area drive both generation capacity and
grid-related marginal costs down. In some extreme scenarios, the
increase to the fixed charge needed to set usage charges at the
lower expected marginal cost over the upcoming rate period might
result in very large customer bill impacts that could drive some
customers to disconnect from the grid. It is inefficient to drive
customers off the system simply because of electricity bills that
exceed marginal costs. Thus, even if increasing the fixed charge to
meet the class revenue target would be optimal, there may be cases
where it is not possible to do so completely,without keeping
volumetric charges slightly above marginal costs or market-based
value.
Opponents to any increase to fixed charges argue that shifting
cost recovery away from volumetric components of the rate harms
the state policy goal of expanding solar DG, since it will reduce the
value of any net metering credits that small-scale DG systems
receive. As DG continues to expand, however, a fixed charge
increase may be needed to prevent growth of inequity between
customers with on-site solar generation and other residential
customers. Utilities across 18 states requested a cost recovery shift
to fixed charges for standard residential and small commercial
rates over the last two years,in large part motivated for expected
growth of distributed solar generation.14 However, it is not efficient
to simply shift the peak demand grid-related costs to a fixed
charge, because doing so precludes customers from having control
over their electricity bill, i.e., they could not directly receive a
reliability credit if they reduced their peak energy usage. Cross-
subsidies would remain, in this case those customers using more
than the class’ average peak demand would be subsidized by
customers using less than the average peak demand. Assessing the
suitability of the requested fixed charge increases in standard rates
should be done keeping in mind the extent to which they will
correct current cross-subsidies implicit in the existing rate
structure. When the fixed charge is set so low that costs unrelated
to usage are shifted to volumetric charges, higher-than average
usage residential customers – which may include customers with
limited access to natural gas for heating – end up paying more than
their fair share of fixed costs.
Utilities’ requests for higher fixed charges to address the impact
of DG are sometimes rejected on the grounds that they have a
negative impact on conservation goals, since they reduce the per-
kWh price signal, and/or a negative large impact on low-use
customers’ bills, which may include low-income residential users.
However, these effects are not by themselves a reason to
perpetuate an inefficiently low customer charge or a poor rate
structure. In fact, these arguments presume that existing residen-
tial rates facilitate the right conservation incentives or low-
income-related subsidies. In most cases this is not the case, and an
increase in customer charges actually corrects a distortion built in
the prior design. First, flat seasonal volumetric prices incentivize
customers to reduce usage below optimal levels in off-peak
periods, leading to an inefficient use of the capacity that is available
to them, and do not encourage enough conservation in peak hours.
Excessive usage charges in the bulk of off-peak hours may also lead
to self-generation, or uneconomic bypass. Second, keeping fixed
charges unreasonably below cost helps residential customers with
below the class average usage, but generally a share of these
customers are not low-income. Low-income customers often have
14
Utilities in 12 states recently received regulatory approval to implement fixed
charge increases for all residential rates.
36 A. Nieto / The Electricity Journal 29 (2016) 31–41
higher electricity consumption than the average customer in the
class and are actually disadvantaged by high variable rates. Most
solar customers are not low-income users and in fact lower income
users tend to disproportionally shoulder the subsidies to solar
customers when volumetric charges are above cost.
In principle, providing support to low-income citizens and
correcting for uneven social income distribution is more efficiently
achieved through fiscal policy, i.e., outside of the electricity tariffs.
However if electricity rates are to continue funding low-income
programs, subsidies should specifically target qualifying low-
income customers. An appropriate mechanism would be through a
fixed low-income credit directly applied on their electricity bills, so
as not to distort price signals in the basic rates for all consumers.
2.2.3. Super-peak per-kWh charges
Increases in load may trigger investments in transmission
infrastructure in hours when capacity is strained, generally around
the time of system peak. Meeting generation resource adequacy
targets is also linked to changes in the utility’s coincident system
peak. Investments in high-voltage distribution infrastructure may
involve adding or upgrading distribution substations or trunkline
feeders generally driven by expectations of growth in the
substation area peak load. Thus, incremental investments or
peak-related procurement costs are in principle suitable for
recovery in on-peak per-kWh charges, or peak demand charges.
Time-differentiated peak kWh charges require a forecast of hourly
locational marginal energy prices (LMPs), time-differentiated
regional generation capacity market prices,15 an estimate of
marginal transmission costs beyond the congestion costs that
are captured by the LMPs ancillary services, and high-voltage
distribution costs.16 Estimating marginal delivery costs require
examining the utility’s planning processes and, in particular, any
investment decisions aimed to meet load growth or maintain
reliability over the planning horizon, as opposed to hypothetical
plant and load additions. A long-run marginal cost estimate is
generally not useful for purposes of setting electricity rates that
may be changed every few years, since they do not send
meaningful price signals to customers when current and expected
system conditions, including reserve margins, differ from opti-
mal.17 Rates based on long-run marginal costs do a poor job at
reflecting the ongoing cost to the utility of meeting peak load
growth over the upcoming planning period (or the avoided cost
associated with load reductions from demand response, energy
efficiency or DG over the same period).
When time of use (TOU) rates are not properly designed, they
fail to preserve intra-class cross equity or produce efficient
conservation incentives. A peak period should ideally be designed
to include only the highest marginal-cost hours in the day, to
promote the right conservation incentives. The annualized
marginal generation capacity, transmission, high-voltate distribu-
tion feeders and distribution substation costs are allocated to the
hours in each daytype and month, based on each hour’s relative
probability of peak, or the relative share of loss-of-load hours
(LOLH) in the case of generation. The combination of hourly
marginal costs is the key input to define the right time-of-day and
seasonal periods and price differentials. However, today the few
examples of residential or small commercial TOU rates existing in
15
The marginal generation capacity cost for a utility operating in a well-
interconnected market is a function of loads and capacity resources region-wide.
16
In a restructured market, estimating marginal transmission costs requires a
financial approach, identifying a forecast of open access transmission tariffs and the
hours that trigger a higher transmission bill from the ISO.
17
A long-run marginal cost analysis assumes the utility has sufficient time to
adjust all inputs to the most efficient system configuration by employing capital and
operating cost measures to precisely match load growth.
the U.S. are generally not sufficiently cost-reflective. When
residential customers can opt for TOU rates, the diurnal peak
period definitions are often too broadly defined covering eight or
ten hours – even when system planners may predictably expect the
highest probability of peak and highest cost of service to occur in a
particular subset of those hours. Too long peak periods dilute the
peak per-kWh price signal. In addition, a peak season should be set
to include the core peak months, where peak load is statistically
more likely to drive system expansion, particularly in the context
of reducing the size of cost shifting associated with net metering
but also for improving incentives for conservation. This again is not
always the current practice in today’s rate making.
2.2.4. (As-used) coincident peak demand charges
In some cases, peak chasing concerns may prevent implement-
ing a three or four-hour “super peak” energy charge that would
reflect the typical high marginal costs in the critical hours of the
day, particularly if applied to all residential customers. In that case,
in order to maintain a clear peak price signal, and provided that the
right meters are in place, it may be advisable to adopt an “as-used”,
on-peak demand charge that would apply to the customer’s
maximum demand as metered within a reasonable number of
hours around system peak period. The demand charge would
recover coincident peak-related marginal capacity costs, thereby
reducing the per-kWh charges to only reflect marginal energy costs
by time of day. Demand charges that are based on customer’s
highest demand during the peak period are important for all
customers but especially relevant for solar net metering customers
who consistently rely on energy imports on the highest cost hours
of the day, typically the evening hours. An on-peak demand charge
gives solar customers an incentive to reduce consumption in the
few hours when their solar system generation is ramping down but
the utility system is experiencing peak demands.
With a properly designed on-peak demand charge, if the
customer reduces peak usage, he receives bill savings equal to the
estimated full avoided capacity cost. As-used peak demand charges
that reflect the underlying marginal capacity costs will therefore
ensure that decisions to invest in demand-limited devices, stagger
the use of domestic electric appliances, or adopt behind-the meter
energy storage are based on economically-efficient incentives, i.e.,
without incurring in uneconomic bypass. A caveat of coincident
peak demand charges is that they assume the customer’s
maximum monthly on-peak demand takes place the same day
and hour that the system peaks, and that the entire monthly
capacity cost responsibility lies in that hour. This limitation is a
necessary compromise in absence of “dynamic” rates that allow
the utility to predict the highest demand hours in the month. As
with any rate design reforms, the level of the peak demand charge
may need to be tempered slightly after a careful analysis of
distributional bill impacts. Since solar customers’ electricity
consumption has a tendency to peak closer to (or at the time
of) the system peak, it makes sense to set separate rates with a
peak demand charge calculated explicitly for them. While a solar
customer will rely more on the utility in cloudy days, imports are
unlikely to peak early in the afternoon (outside of the core peak
hours). This is because solar panels will still be producing energy in
overcast days, albeit at a lower rate than normal. In general, more
granularity in rate periods may be required if a time-differentiated
demand charge is adopted for the entire residential rate class. For
example, a lower demand charge may be required for a subset of
the early afternoon (shoulder) hours, along with the peak demand
charge, for the residential rate.
2.2.5. Dynamic rates
As is the case for conservation, promotion of cost-effective DG is
best achieved with marginal cost-based dynamic pricing. The
 A. Nieto / The Electricity Journal 29 (2016) 31–41
optimal dynamic rate design would be a real-time pricing (RTP)
rate, but RTP rates are not well accepted in the residential segment.
Other dynamic rates only modify the kWh charge on system’s
critical days and during specific critical events called by the utility,
and thus may be easier to implement, particularly if coupled with
smart programmable thermostats. These include critical peak
pricing (CPP) or peak-time rebate (PTR). Any CPP or PTR critical
peak rate need to include a marginal cost component for
generation capacity, transmission, and high-voltage distribution
in the price for those critical peak hours. While they are more
efficient than regular TOU rates, they may not be easily adopted as
a mandatory residential rate.
3. Redefining cost of service by customer class
3.1. Treating DG customers as a separate class
A separate rate class for net metered DG customers will be
necessary if the key features for reforming standard residential
rates into a multi-part, time-differentiated structure are not
politically acceptable. Initially a rate for net metering customers
may need to be set as revenue neutral to the otherwise applicable
tariff. Hourly and sub-hourly load data from net metering
customers facilitated by smart meters or advanced metering
infrastructure (AMI) will inform the cost to be allocated to the net
metering class. AMI data can improve load forecasting dramati-
cally, since there is no need to extrapolate results from a sample of
load research profiles. As discussed, currently intermittent DG may
not provide substantial system wide or local reliability benefits to
the system, but to the extent that DG customers are placed in a
separate cost-reflective rate and collectively begin to change usage
patterns in a way that reduces going-forward costs of the utility,
the residential DG class should see a corresponding reduction in its
revenue target. That reduction should ideally match the expected
avoided costs over the period the rates are going to be in effect. This
requires that network planners become adept at accounting for
forward-looking DG and DER system impacts in general.18
Otherwise, a reduction in the peak demand-related cost allocator
for net metering customers would merely lead to a cost shift to
other customer classes in the cost of service study. Likewise,
additional costs imposed by DG may be better tracked through a
separate net metering class.
The specific connection policy also matters. If the connection or
line extension policy is not properly designed, placing DG
customers in a separate rate class would be required to mitigate
subsidies to high interconnection costs. Rates do not need to
socialize any higher connection costs associated with the bi-
directional electricity flow of DG customers when customers are
asked to pay for any needed reinforcement costs in an up-front
connection fee or customer-specific separate monthly fee.
3.2. Test year
A critical input to pricing is the allocation of the overall revenue
requirement to the various rate categories and customer classes.
Recognizing the impact of DG customers on utility costs going-
forward may require revisiting the traditional cost of service
approaches. There is often no detailed examination on the methods
to allocate cost of infrastructure and social public policies to
customer classes, even though they are the premise to establish the
level of non-bypassable costs. At a minimum, determining a tariff
18
Utility load forecasts generally rely on historical trends, and to that extent they
implicitly capture the ongoing impact of DG, but not sufficiently the impact of future
DG.
37
revenue requirement should involve looking at a future test year’s
costs and weather-normalized class hourly load profiles. Despite
AMI, load forecasting is challenging due to the intermittent nature of
DG and the potential for errors in the expected DG penetration by
area. Utility rates will never perfectly match utility cash flow and
therefore there will always be over- or under recovery of costs in a
given year. However, if the utility incorporates an estimate of future
customer reaction to new rate charges or specific demand response
programs when setting rates, the mismatch between revenue
collected and revenue target will be largely reduced. Cost allocations
will still need to be revisited at the time of a rate case and adjusted if
customers within the class demonstrate a change in load patterns.
3.3. Choice of cost allocation method
A thorough tariff reform should include an analysis of the legacy
cost allocation methods, to determine if changes are warranted. As
a result of growing renewable DG, more attention should be paid to
the specific method to classify and allocate costs to set class
revenue requirements, particularly if customers with DG can be
separated into a standby alone class. There are two main methods
to allocate costs to customer classes. One is based on embedded
costs, also referred to as “average cost” approach. The other
approach is based on class marginal cost of service. These days,
many utilities use embedded costs to allocate revenue requirement
to customer classes, as well as to set rate levels. Other utilities use
embededd cost studies to set class revenue targets but rely on
marginal cost analysis to set the actual tariff design. Setting
bundled standard tariffs or network charges with volumetric
charges that reflect marginal costs by time of day can be done even
if class revenue targets are established based on an embedded cost
allocation method. However, fully adhering to the goal of achieving
an efficient size of network system requires setting revenue targets
as a function of classes’ marginal cost contributions relative to total
marginal cost of service. At the very least, elements of marginal
costs should be used in embedded cost allocation studies.
A marginal cost study generally allows more granularity at the
time of assessing forward-looking costs of service. For example,
embedded cost studies rarely differentiate local connection costs
versus upstream distribution costs. All distribution costs may
treated as being driven by system wide non-coincident demands.
For generation costs, embedded cost of service studies also exhibit
limited granularity, with little time differentiation. A number of
studies classify the bulk of fixed costs of existing power plants as
peak demand-related, even though mid and base-load power
plants were built mostly to provide energy cost savings. This
approach favors high-load factor customers. In contrast, a marginal
cost analysis is a bottom-up approach that uses classes’ hourly load
profiles, relies on hourly marginal unit cost data for both energy
and capacity related costs, thus taking into account class usage in a
broader range of hours when defining classes’ cost-responsibility.
When marginal costs are used to set each class revenue
requirement, each class’ marginal cost revenue (the revenue that
would be obtained from setting all rate components equal to
marginal cost) is adjusted as necessary to create a set of tariffs that
will yield the total allowed revenue requirement. A marginal-cost
based revenue requirement allocation seeks to assign costs to
classes in a manner that takes into account not only cost causation
but also customer response to price changes, to minimize
efficiency distortions that result from setting prices different from
marginal costs.19 Using this approach, average rates for each
19
This process steems from the inverse price elasticity rule formulated by Ramsey
pricing, which guides the appropriate relative difference between rates and
marginal costs.
38 A. Nieto / The Electricity Journal 29 (2016) 31–41
customer class are marked up (or down) with respect to marginal
costs in inverse proportion to the customer group’s elasticity of
demand, e.g., how likely they are to react to price changes by
relocating, switching to an alternative energy source, installing
self-generation, or acquiring more efficient electricity appliances.
In practice, many states that rely on marginal costs for cost
allocation use an implicit assumption that each customer class has
the same price elasticity and adjust each class’ marginal cost
revenues proportionally (a method known as equi-proportional
marginal cost, or EPMC) to close the gap between marginal cost
revenues and total revenue requirement. Improvements to this
method can be made by applying different adjustments depending
on the potential for self-generation in particular customer classes.
This is relevant in the context of DG, since customers who adopt
solar panels are generally showing higher price elasticity than
those who do not.
3.4. Environmental cost analysis
While there is little doubt that solar and wind DG resources
have a positive impact as clean energy resources, their environ-
mental benefits may not be inherently quantifiable in a traditional
cost-of-service study, since they do not represent a direct benefit
or cost saving to the utility or load-serving entity. An environmen-
tal cost component may be implicitly captured in the marginal
energy cost analysis, in hours when fossil fuel units that have
internalized the cost of complying with state or federally
established CO2 emission reductions are dispatched at the margin.
Market energy prices will fail to reflect the cost of negative
externalities associated with conventional carbon-based genera-
tion resources in the absence of a CO2 cap-and trade program.20 In
that case, the only direct quantifiable environmental benefit of
renewable-based DG exports to the grid is the market value of the
renewable energy certificates (RECs), to the extent that the utility
can use the renewable energy received from the DG facility
towards its renewable portfolio standard (RPS) requirement.21
Alternatively, the customer may be able to retain the REC,
obtaining a supplemental source of funding for DG. The market
value of RECs may be low in states with significantly more REC-
producing resources than needed to meet the prevailing RPS. Any
additional environmental value attributable to low-carbon DG that
is not currently internalized by market prices can only be
estimated as a shadow price. Subsidies to renewable DG deemed
necessary on account on its environmental value are best provided
through a transparent mechanism outside of the standard tariff
setting, such as via utility-run auctions targeting specific amounts
of renewable DG, so as to avoid excessive cross-subsidies.
4. Alternatives to standard retail rate restructuring
Regulators may reject any sort of marginal cost-based demand
charge or mandatory time-differentiation, at least until a general
rate case may take place, or due to policy, legislative, or technical
constraints. In the meantime, other approaches may be used to
attempt to mitigate cross-subsidy concerns under net metering,
particularly with regard to small-scale solar PV. Ad-hoc grid access
20
A cap-and-trade program offers opportunities for the most cost-effective
emissions reductions, if well designed. States may be a part of a cap-and-trade
program aimed at pricing emissions, such as the Regional Greenhouse Gas Initiative
(RGGI).
21
Utilities and other load-serving entities may satisfy RPS through the purchase of
third-party RECs of by procuring renewable electricity, typically at a premium over
conventional generation contracts. REC is a tradable certificate. One REC is issued
for every 1 MWh of electricity placed on the grid and its value represents all the
clean energy benefits of energy generated from a renewable facility.
charges (either upfront or ongoing) may be combined with
traditional net metering or with a fixed payment for excess solar
generation. Regulators may also adopt a buy-sell arrangement that
relies on social avoided cost compensation. There is substantial
controversy about emerging proposals as to what costs are avoided
and what is the amount of net unrecovered cost that should be
shifted back to net metering customers.
4.1. Solar access charges
One of the proposed mechanisms to recover the “non-
bypassable” costs involves preserving net metering but ensuring
that DG customers pays the same amount for all the energy that
they would have used before the installation of solar panels, except
for the avoided costs directly achieved by the utility. The access
charge aims to recover the “excessive compensation” the customer
receives through net metering for both reduced usage and
exported energy. Thus the DG customer continues to pay the
normal residential rate for purchases from the utility, under a net
metering arrangement, but also pays a solar access charge which
may be defined as a per-kWh surcharge, as a specific solar demand
charge, or as a charge per kW of the involved solar system
nameplate capacity. The access charges proposed to date generally
assume that the utility will continue to procure the same amount
of generation and transmission capacity to serve that customer,
and plan for the same amount of distribution capacity as if the solar
panels had not been installed; in other words, it reflects the
amount of “fixed costs” attributable to the average solar customer,
effectively charging back the customer for the portion of fixed
transmission and distribution costs, and social policy-related non-
bypassable costs embedded in the kWh rate that would have been
collected if there had been no on-site generation.
Distribution grid access charges implemented this way are
equivalent to adopting a fixed charge, assessed exclusively to net
metering customers. Because the grid access charge is designed to
avoid impacting the customer’s otherwise applicable tariff, only
costs directly avoided by the utility are used in its calculation.
Generally, the solar access charge acknowledges estimated
benefits (avoided costs) associated with the customer’s solar
system output such as reduction in energy and losses. The actual
implementation details vary widely among the proposals put
forward to date. The decision to establish a cost recovery
mechanism through an ‘ad-hoc’ grid access charge is fundamen-
tally different from a rate restructuring exercise, one that sets up
the base rate components equal to marginal costs. To estimate a
grid access charge, the analyst generally looks at an average or
typical customer. The charge is based on the assumed solar
generation profile and a pre-defined assumption on the amount of
energy consumption offset by the solar output and exported to the
grid. Thus, the grid access charge may not be completely equitable,
since it is not based on each customer’s maximum demand or
actual net energy volumes. Also, because the grid access charge is
set as a residual amount to recover the non-bypassable costs, the
customer does not face any price signal about the value of a
permanent reduction in peak demand.
4.2. Avoided cost-based compensation for export energy
If neither the required rate structure reform, nor moving
customers to a separate, net metering class is possible, a second
best solution to reducing cross-subsidization to DG customers is to
eliminate net metering of exports and compensate excess
generation at avoided costs. The avoided-cost compensation can
be set to reflect the time-differentiated utility’s opportunity cost,
i.e., the expected avoided incremental energy costs adjusted by
losses, plus a component of avoided capacity costs if the solar
 A. Nieto / The Electricity Journal 29 (2016) 31–41
generation is paired with storage and helps offset coincident peak
load. The avoided unit cost may also in principle, be designed to
include an environmental avoided cost, such as the REC value. In
cases where the distributed solar generation does not directly
qualify as an RPS resource, the export payment may still include a
small green energy premium, to reflect the fact that the solar
generation exported to the grid offsets loads of other utility
customers and hence lowers the overall retail load that needs to be
served from utility-scale contracted generation resources. In some
jurisdictions, utilities already provide avoided cost payments to
commercial or industrial customers for their excess generation,
and they are generally limited to avoided loss-adjusted energy
marginal costs. Using an export avoided-cost approach does not
fully remove the cost-shifting to non-solar customers, since the
solar customer would continue benefiting from large bill savings
for the share of on-site electricity usage that is offset by self-
generation at a rate that exceeds the value of load reductions.
Attempts to remove net metering have taken place in California
and other states. In August 2015, the three California investor-
owned utilities (IOUs) proposed eliminating net metering and
adopting a system of compensation of solar excess generation at
the utility’s levelized avoided costs. The proposals differed by
utility but they would all significantly reduce the profitability of
the customer-sited solar systems, compared to their value under
existing net-metering practice. For example, Southern California
Edison (SCE) proposed an export rate of $0.08/kWh, based on
avoided energy cost plus a $0.01 per-kWh REC premium,, along
with a monthly grid access charge of $3 per kW of nameplate
capacity of the solar system that would recover the estimated non-
avoided transmission, distribution, and social costs associated with
the energy displaced by the output of a typical residential PV
system.22 In January 2016, the California Public Utilities Commis-
sion (CPUC) issued a decision that preserved net energy metering
and determined the parameters of a successor net energy metering
rate (NEM 2.0) for all three IOUs, subject to further review in 2019.
CPUC also ordered that net metering customers with systems less
than 1 MW in size will be charged a one time interconnection fee,
based on the utility company’s actual costs but excluding
distribution upgrades, and non-bypassable charges ranging
between 3 and 2 cents per kWh to cover the cost of social
programs such as low-income and energy efficiency subsidies, for
all their metered imports from the grid, before netting for exports.
Finally, the CPUC decided that all residential customers will be
moved to a mandatory TOU rate by 2019.23 NEM 2.0 customers will
have their accounts grandfathered on this successor tariff for 20
years. By contrast, within approximately the same time frame, the
Nevada Public Service Commission approved the end of net
metering in the state, moving all NV Energy's DG customers to a
separate rate with increased fixed charges and lower kWh charges
effective in January 2016.24 Instead of net metering, solar
customers in Nevada will be compensated at the estimated
avoided wholesale energy cost of roughly 2.6 cents per kWh for the
excess solar power delivered to the grid, i.e., assuming no reliability
or environmental value is provided by solar exports.
22
This grid access charge was estimated assuming non-avoidable costs were equal
to 10.8 cents per kWh of energy offset by the customer solar system, assuming a
typical solar panel system size of 6 kW.
23
Before the Public Utilities Commission of the State of California, “Decision
Adopting Successor to the Net Energy Metering Tariff”. Decision 16-01-044, January
28, 2016. Rulemaking 14-07-002.
24
The fixed charge for new solar customes will increase from $12.75 per month to
$17.90 a month in 2016, then will increase by $5.15/month each year until it reaches
$38.51 a month in 2020. The rate that solar customers will pay for grid-supplied
power effectively in January 2016 declines slightly from 10.8 cents per kWh in 2016
to 9.9 cents for Southern Nevada. Before the Public Utilities Commission of Nevada,
Modified Order February 12, 2016. Dockets No. 15-0741 and 15-0742.
39
Moving away from full net metering and towards avoided-cost
based payments for solar energy exports is a step in the right
direction and it is preferable than continuing the practice of net
metering under sub-optimal rate structures. However, ideally the
export price should be time-differentiated, and locational, in order
to ensure a fair compensation. In addition, enforcing ad hoc
surcharges to recover an estimated average of non-bypassable
costs may be problematic because it often does not involve a
transparent and easily replicable computation and because again,
the surcharge calculation is a function of the level of the kWh retail
charge and a predetermined assumption on the monthly energy
displaced by an average solar system, thus it is less efficient than
adopting full 'gross metering' and simply letting the customer
receive an avoided-cost based credit for all of its solar output.
4.3. Value of solar, or ‘sell all, buy all’ approach
A number of states, particularly where the retail residential per-
kWh rate does not stimulate as much solar DG as compared to
states with higher rates, have adopted or are considering adopting
a value of solar (VOS) rate, or sell-all, buy-all payment format.
Under such arrangements, the compensation for DG output is
decoupled entirely from the energy tariff. A DG customer pays the
standard retail rate for all of its energy requirements (regardless of
whether they are met with imports from the grid or from the
customer on-site generation) and sells all energy generated by the
solar panel (i.e., not just the excess energy) to the utility, at a price
that represents the “social” avoided cost associated with that
generation. In essence, a VOS rate is equivalent to compensating for
export generation at avoided cost, along with a non-bypassable
grid access charge, but including avoided environmental exter-
nalities in addition to the direct financial benefits to the utility. The
VOS rate is generally a flat, non-time-differentiated value, and set
higher than the average retail per-kWh rate.
VOS agreements can emulate a long-term contract by
guaranteeing a levelized price for the entire output of the solar
panel throughout its useful life, if the payment is fixed throughout
a guaranteed period of 20–25 years, thus providing high
predictability of investment returns, or can be designed so that
the payment is updated annually or every few years. Austin Energy
in Texas pioneered the adoption of a VOS rate in 2012. Its levelized
value is recalculated annually, so as to reflect current utility costs
and prevent overpayments when system costs fall. Effective Jan. 1,
2016, the VOS rate is 10.9 cents per kWh. A similar method was
approved as optional for utilities in Minnesota, with the difference
that the VOS rate is fixed for 20 years once the customer registers
for the rate. VOS is also under consideration in Maine and at several
other jurisdictions.
The VOS gross-metering approach aims to ensure that the
utility recovers the full cost of serving the solar customer, including
the costs of social programs that may have been allocated to the
rate class the customer belongs to. This is in principle, analogous to
the idea of paying demand response participants the market price,
or LMP, less the generation component of the rate. Just like the grid
access charge notion, the VOS rate intends to make customers pay
for the non-avoided share of generation capacity, transmission,
and distribution costs embedded in the retail rate, but using the
actual customer usage, as opposed to the class average, to
determine the appropriate recovery of non-bypassable costs. Since
the VOS is generally set higher than the retail rate the customer
receives a positive value for the energy consumed on-site. A VOS
rate ensures that the customer benefits by exactly the value of
estimated avoided or marginal costs, in addition to any expected
environmental benefit. In that sense, it seeks to mimic the
efficiency properties of compensation that solar customers would
receive with rates that were set based on marginal costs plus an
40 A. Nieto / The Electricity Journal 29 (2016) 31–41
adder for externalities under net metering. However, the lack of
time-differentiation in VOS rate approaches to date means that
solar customers are compensated on the basis of an average solar
generation profile and so are not entirely equitable. A VOS rate is
also a second best solution to the extent that it does not correct for
the cross-subsidies that occur among customers within a class
when local distribution facilities costs are recovered in per-kWh
charges.
The VOS components in the proposals seen or approved to date
include long-term estimated cost savings in energy, some savings
in generation capacity, transmission and distribution losses, plus
other potential benefits from renewable energy development. The
range varies with Maine having estimated the highest VOS to date,
at 33 cents per kWh, about two times the retail per-kWh rate level.
The main controversy with the VOS approach is the size of the
environmental adder. The environmental value may be limited to
the avoided cost to the utility to comply with environmental
regulations such as meeting the state RPS or the federal Clean
Power Plan (CPP) requirements, or buying pollution emission
credits. In some cases, such as in Minnesota, the approved
methodology also includes the value of additional externalities
created by solar such as local job creation, and avoided future
adverse climate change effects such as flooding.25 The use of
environmental externality values is common in benefit-cost
analyses of energy efficiency and demand-side-management
(DSM) programs in integrated resource planning. They have not
been used explicitly as part of an actual retail pricing mechanism,
other than implicitly in “feed-in tariffs” that provide a rate of
return high enough to attract investment in an infant industry. A
VOS payment structure is still a cross-subsidy to the extent that the
externality values exceed the current utility avoided compliance
costs, and the gap must be met by rate revenues from non-solar
customers. Despite its limitations, VOS compensation is neverthe-
less a more transparent method to provide a subsidy to clean
energy resources than net metering, and can potentially improve
with more granularity in the VOS price-setting methodology, for
example by distinguishing between regular solar customers and
customers with behind the meter storage.
5. DG as a ‘non-wires’ alternative
5.1. The role of energy storage
The concept of distributed energy resources that may serve as
“non-wires” alternatives and thus partially or fully defer network
investment is gaining traction as a regulatory policy goal. The
regulator should provide opportunities for customers who may
adopt storage technologies paired with their DG facility, or who
engage in automated demand response, to participate in the utility
transmission and/or distribution planning process as local “non-
wires” alternatives, i.e., in lieu of new capacity.26 This may be
achieved through the right combination of rate design and
incentive regulatory approaches. California, New York, and other
states are paying particular attention to both grid-connected and
behind-the-meter energy storage. Distributed energy storage will
add flexibility to the distribution grid, enabling DG customers to
store part or all of the energy generated in excess of their own
needs for their own use or for exports to the grid at a later time,
25
The VOS rate approved in Minnesota includes about 3 cents/kWh as an
environmental adder, based on the federal social cost of carbon ($37 per metric ton)
developed by the Environmental Protection Agency (EPA).
26
Currently, homeowners are unlikely to save money by shifting their power
consumption from the grid to a battery-backed PV system, particularly one large
enough to make going off-grid feasible.
namely at peak hours, thereby transforming the way the
distribution grid is planned. In New York, under the PSC’s initiative
“Reforming the Energy Vision” (REV), the distribution utility will
assume the role of “Distribution System Platform” (DSP), promot-
ing and coordinating interconnection of DERsthat may provide a
wide array of services to the grid.
Batteries are beginning to be a cost-effective option for large
commercial and industrial customers, and it is conceivable that as
rates are structured to reflect the higher costs of service in peak
hours and the cost of batteries keeps declining, smaller customers
will also eventually find that battery systems are cost effective.26
Energy storage will particularly play a key role in the development
of renewable-based microgrids, a single controllable entity serving
one or multiple customers within a community embedded within
the distribution system. Smart microgrids can deploy, in addition
to co-generation, renewable generation, thermal and electric
storage, and advance controls that allow the system to automati-
cally disconnect from the grid and work in “islanding” mode during
planned or unplanned outages. Customers in a microgrid can
adjust their internal usage during normal grid conditions and sell
excess power to the grid when it is economic to do so, provided
that the right price signals are in place, thus adding resilience value
to the grid. Even outside of a microgrid or solar community,
advanced technology such as smart inverters with communication
capabilities will allow individual DG to be a resource to the utility
to maintain grid reliability as load grows, by making the
distributed generation available to the macrogrid to maintain
system stability or avoid curtailments. They may be directly
dispatched by the utility or respond as expected to relieve local
overloading in the distribution circuit, and may potentially do so at
a lower cost than the utility. Network prices or standby tariffs
applicable to microgrids should be set so that the microgrid
continues to pay for the use of the distribution facilities while
connected to the grid, to the extent that they need supplementary
energy from the grid or if they need extra transformer capacity to
sell excess power.
5.2. The value of ‘D’
As part of this process, a strong interest has been placed in
quantifying the “distribution value” to the utility, for purposes of
compensating DER resources for distribution grid services.
Providing distribution services and receiving a fair compensation
for it, in addition to any market-based revenue that they may
receive through aggregation bids in wholesale energy and capacity
markets, may help behind-the-meter storage or microgrids be
financially viable. In the past, any attempts to adopt locational
tariffs have been rejected by regulators on the grounds that it
would be inequitable. The administrative burden of having
separate rates for multiple distribution locations is likely another
factor. If the unbundled distribution component in rates is based
on a system wide average, the signal may be too low to serve as an
incentive for DG locational decisions. System-wide marginal cost
estimates are useful for pricing purposes when most areas of the
system have similar reserve margins but are less likely to induce
the right response from DERs when load growth or expansion has
been uneven across the various distribution service areas. To a
large extent, a detailed marginal cost analysis by substation area
can help with this exercise. The value of DG, storage or a microgrid
is in part dependent on the current and expected level of capacity
margins in a given distribution feeder or substation. The
computational method should be granular enough to identify
where expansion of distribution stations or lines is needed to meet
expected load growth and the level of investment, as an annualized
cost per kilowatt, that local DERs can provide when offsetting the
specific area's peak load Even if LMPs were defined at the
 A. Nieto / The Electricity Journal 29 (2016) 31–41
distribution node level, meaning that distribution losses and
congestion were captured by distribution nodal prices, their role
would be limited as a locational incentive to foster development of
microgrids. The congestion costs between nodes embedded in
those nodal prices would not necessarily or distribution planner
fully signal when it is efficient to invest and how much. This is
because reliability failures come at too high a cost, and the utility
will likely step in before the congestion costs around a distribution
load pocket or the risk of curtailment is high enough to incentivize
non-wires alternatives.
Initially, meeting the regulatory task of encouraging optimal
location of DER in the right places solely through geographically
differentiated distribution rates will be challenging. Among the
potential obstacles are:
 Distribution marginal cost estimates are typically not estimated
by sub-areas of service or feeder due to the lack of detailed
hourly substation load data. A DER in a particular area may have a
significant distribution value at a time when no other system
wide benefits are created because local distribution networks
peak at times different from the rest of the state or region where
they operate.Eventually detailed data on reserve margins by
feeder will become available, but load forecasting at the feeder
level may be challenging in the initial stages.
 A key element to determine proper signals in distribution rates
has to do with the actual cost study relied upon to set locational
rates; the cost basis used for the rate may not reflect short-term,
but rather a longer-term marginal cost that averages several
cycles of build and then growth into the new capacity.
Even if the DG facility has adopted storage and is optimally sited
to relieve capacity constraints in specific areas of the system, the
utilities might act conservatively and be reluctant to account for it
in its capacity expansion plan, in order to preserve reliability, at
least until experience demonstrates the minimum amount of
equivalent "wires service" that the DER provides under several
contingency scenarios. Thus, in the near term, a certain amount of
redundancy in grid investments will prevail. Specific payments or
rebates in exchange for grid-provided services may be necessary to
provide effective incentives to DG with behind-the-meter storage
capabilities, or microgrids, if standard distribution rates fail to
effectively signal locational differences in costs. A contract or
locational rebate system with penalty enforcement mechanisms
may do the job of signaling the incremental value of load reduction
in the specific area of constraint and at stimulating the use of DER
as non-wires alternatives.
Detailed estimates of near-term locational avoided costs,
including deferred investment or reduced cost of ancillary services,
such as reactive power resources, voltage control and other, over
the upcoming distribution planning cycle should be made
transparent to all customers, to allow higher predictability of
the value of D. Such cost estimates may serve to establish a cap
price for DER solicitations, or as a starting point to negotiate the
terms of contracts with DER owners that commit to serve as non-
wires alternatives. The ability of a specific DER to maximize
avoided costs to the utility will depend on dispatchability levels, or
41
committed response and availability, which is a function of the
deployed control systems. In the case of microgrids, the utility may
negotiate dispatchability and other operating arrangements
through short-term or long-term service agreements, ideally after
a competitive solicitation process that spurs efficient competition
between a range of non-wires alternatives and investments by
distribution facilities. The microgrid may receive a partial
distribution credit if using a control mechanism that ensures no
more than an agreed-upon amount of energy is taken from the grid
at critical times, effectively agreeing to reducing the standby
service needs. or receive a direct avoided capacity cost payment if it
can send energy to the distribution grid in hours where the
transmission system has experienced an outage. Location-specific
reliability-based rebates can be provided up front, tied to expected
contribution to capacity cost savings over the next cycle of
investment. Alternatively, payments can be provided upon verified
local load relief.
6. Conclusion
Regulators and utilities have an opportunity to promote clean
small-scale distributed resources in a way that serves environ-
mental goals without unduly burdening non-participant custom-
ers. Getting the right rate structure in place and improving cost
allocation methods as the market gets ready to embrace DG will be
critical in ensuring that outcome. In determining a rate reform, the
goal should not be to pick winners and losers, but rather to adopt a
time-of-use, marginal cost-based rate design that gives appropri-
ate and equitable price signals. Lacking the ability to optimize all
retail residential and commercial rates, a separate, cost-reflective
rate specifically designed for solar customers may be required to
preserve best practice ratemaking goals under net metering. An
alternative is to provide an avoided cost payment to compensate
DG customers for excess generation. This second-best approach
allows for value-based compensation but only partially mitigates
cost shifting if the base rate structures remain suboptimal. The
same principles that govern decisions on retail rates and net
metering can in principle be applied to reforms of standby rates for
larger DG, or microgrids, as well as any pricing mechanisms
intended to compensate for services provided by DER to the
system. Additional compensation for the positive environmental
attributes of renewable-based DG not accounted for in market
prices for energy or RECs is better handled outside of the standard
retail tariff system, via well designed buy-all, sell-all agreements or
feed-in tariffs. Finally, undertaking updates to cost of service
studies as more DG customers are added to the grid will be key in
ensuring that any incremental benefits from distributed resources
are captured appropriately in class revenue targets. This will be
particularly important as customers begin to adopt behind-the-
meter storage or join smart microgrids, ultimately enabling them
to serve as “non-wires” alternatives.
Amparo Nieto is a Vice President with NERA Economic Consulting, is an economist
specializing in the electricity industry with over 20 years of experience advising
utilities, regulators and system operators on wholesale energy and capacity market
design, electricity network pricing, regulatory rate reviews, and cost modeling.