factor that affect the succes regulatory to foster investment energi efficiency 2011

8/10/2019 Factor That Affect the Succes Regulatory to Foster Investment Energi Efficiency 2011

1/18

ORIGINAL RESEARCH

The many factors that affect the success of regulatory

mechanisms designed to foster investments in energy efficiencyJay Zarnikau

Received: 24 January 2011 / Accepted: 5 September 2011 / Published online: 18 September 2011# Springer Science+Business Media B.V. 2011

Abstract A utilitys profit-maximizing level of invest-

ment in energy efficiency or demand-side management

(DSM) programs and mix of programs is affected by

natural load growth, the frequency of rate cases,

program costs, and the structure of any mechanism

designed to either compensate the utility for foregone

profits or sever the link between sales and profits. Under

a range of reasonable assumptions, decoupling can

incent a utility to invest in DSM. However, a utility

experiencing high natural load growth and little inflation

is likely to resist the imposition of a decoupling

mechanism, as it would tend to lower profits. A utilitywith low growth in per-customer sales will tend to favor

decoupling, as it will tend to lead to higher profits than

under traditional regulation. The results presented here

are quite sensitive to the assumptions made regarding

natural load growth, regulatory lag, the frequency of

price changes, price elasticity of demand, and other

factors. This suggests that there is not a single approach

to promoting energy efficiency without penalizing

utility profits that will work in all situations for all

utilities.

Keywords Energy efficiency. Regulatorydecoupling . Electricity pricing . Conservation

Introduction

The National Action Plan for Energy Efficiency

prep ared by the U.S. Environmental Protection

Agency (2007) reports that Few energy efficiency

policy issues have generated as much debate as the

issue of the impact of energy efficiency programs on

utility margins.

The reduction in profitable salesresulting from energy efficiency poses a disincentive

to utility investments in energy efficiency. Regulatory

schemes have been devised to remove this disincen-

tive, including lost revenue adjustment mechanisms

(LRAMs), decoupling mechanisms, and shareholder

bonuses.

LRAMs compensate the utility for the margins or

earnings lost as a result of reduced sales. The impacts

of energy efficiency programs upon the utilitys sales

are estimated. The under-recovery of earnings or

margins resulting from energy efficiency programs isthen calculated. Adjustments or true-ups are under-

taken to compensate the utility for the earnings

foregone as a result of its investment in energy

efficiency. The implementation of this approach

requires credible estimates of the impacts of the

energy efficiency programs upon sales as well as the

utilitys fixed and variable costs.

Decoupling seeks to fully divorce earnings and

revenues from sales, thus removing the throughput

Energy Efficiency (2012) 5:393410

DOI 10.1007/s12053-011-9139-1

J. Zarnikau (*)

Frontier Associates LLC,1515 S. Capital of Texas Highway, Suite 110,Austin, TX 78746, USAe-mail: [email protected]

J. ZarnikauLBJ School of Public Affairs and Division

of Statistics of the College of Natural Sciences,The University of Texas at Austin,

Austin, TX 78713, USA


2/18

incentive. Allowed revenue or allowed revenue per

customer is calculated. Rates are periodically adjusted

through true-ups to ensure that the utility receives its

allowed revenues or margins. The revenue collection

allowed by the regulatory authority may be either

increased or decreased under decoupling. As a result,

actual utility revenues track projected revenuerequirements despite unexpected fluctuations in sales

(NARUC2007). The theory underlying decoupling is

explained in some detail by Eto et al. (1997). Full

decoupling not only addresses the recovery of

margins lost due to energy efficiency but also negates

the effects of other factors affecting projected sales

(Moskovitz et al.1992). Decoupling may also reduce

uncertainties associated with the use of volumetric

charges to recover the level of revenues approved by

a regulatory authority. This is of particular interest

since utilities tend to recover many fixed coststhrough volumetric changes, leaving the utilitys

earnings highly sensitive to unexpected swings in

the weather and economic conditions impacting sales.

Decoupling can remove much of this risk.

Performance bonuses can be designed to provide

an incentive for the utility to achieve higher levels of

energy efficiency. Shareholders receive a bonus when

a goal for energy efficiency is met or exceeded. The

bonus may be designed to be sufficiently large to

compensate shareholders for lost margins.

There are many variations on these themes.Different states of the USA have adopted slightly

different formulas when implementing LRAMs or

decoupling (Edison Foundation 2011; Hirst et al.

1994). At least seven states in the USA have

approved some type of decoupling mechanisms for

natural gas and/or electric utilities within their

regulatory jurisdiction, led by California (Kushler et

al. 2006). LRAMs have been introduced in Ohio,

Arkansas, North Carolina, and South Carolina. Texas

adopted a simple shareholder bonus approach.

These regulatory mechanisms have been endorsed byorganizations which favor expanded utility energy

efficiency programs. Consumer organizations tend to

oppose such mechanisms, either over concerns that such

mechanisms will increase short-term rates or over more-

general unease related to the growing involvement of

utilities in the delivery of energy efficiency services

(Anderson 2007). The Edison Electric Institute has

come to the defense of LRAMs while expressing

concern over full decoupling (Tempchin 1993).

Brennan (2010) notes that decoupling removes

incentives for utilities to operate in an efficient

manner. Kihm (2009) concludes that if a regulator

sets the utilitys rate of return in excess of its cost of

capital, then the utility will prefer sales growth that

will lead to greater supply-side investments and

decoupling will fail to remove the Averch

Johnsoneffect (Averch and Johnson 1962).

This paper complements recent analysis by Cappers

et al. (2009). As in Cappers et al. (2009), a prototypical

electric utility system is modeled. However, while

Cappers et al. (2009) simulates the financial effects on

a utility of a set of pre-determined demand-side

management (DSM) program portfolios under various

assumed regulatory mechanisms, this present analysis

determines the utilitys optimal profit-maximizing level

of DSM and the mix of DSM programs selected by the

utility under various ratemaking schemes. The focushere is to explore a utilitys motivation to pursue

energy efficiency investments under various market

conditions and regulatory incentive mechanisms.

Why is this a controversial topic? As shown here,

the impacts of each of these regulatory mechanisms

on utility profits are very sensitive to the economic

environment faced by the utility. A regulatory

mechanism structured in a particular manner for one

type of utility facing a particular set of circumstances

may not be fair if applied to other utilities facing

different economic environments. Further, the struc-ture of regulatory mechanisms may lead utilities to

pursue different types of programs. Certainly, a load

management program has a different impact on utility

rates, capacity needs, and power plant emissions than

a base load DSM program. Consequently, the mix

of programs pursued by a utility may be of interest

and is explored here.

Theory

Incentives to invest in DSM under traditional

regulation

Under traditional regulation with annual rate adjust-

ments, a utility with no regulatory mandate to pursue

DSM will weigh certain benefits against certain costs

in deciding whether to invest in DSM. The utility sees

a benefit if it can avoid uneconomical salese.g.,

peak period sales of electricity for which the cost of

394 Energy Efficiency (2012) 5:393410


3/18

making the sale exceeds the retail price at which the

electricity may be sold.1 The value of electricity

obtained through a wholesale electricity market

during a price spike may be many times greater than

the retail prices set under regulation. This savings in

operating cost increases the utilitys profit, provided

prices are based on the utilitys cost of service in a

previous year (a historical test-year). Prices in the

subsequent years will be reduced to reflect the savings

in operating costs achieved by the utility in the

current year, so most of the benefit will be realized

in the first year that the new DSM investment has an

effect on reducing operating costs. This benefit to the

utility may be completely negated if there is a

subsequent true-up of costs or an automatic fuel cost

adjustment. This efficiency gain through DSM fo-

cused on reducing uneconomical sales has a second-

ary effect (even in the presence of true-ups orautomatic cost adjustments)the resulting reduction

in future retail prices may lead to higher future sales,

revenues, and profits through the price elasticity of

demand effect.

Under traditional regulation, there may be a

detrimental financial impact on the utility if the

DSM reduces the utilitys future rate base and thus

the future financial returns to the utility permitted by

the regulatory authority. This impact is dependent

upon the level of costs potentially avoidable as a

result of the utilitys investment in DSM, the rate ofreturn on rate base permitted by the regulatory

authority, and the utilitys cost of capital. Assuming

the regulatory authority permits the utility to fully

recover its DSM costs from ratepayers, the program

costs may nonetheless impact the utility. There may

be a lag in time between when DSM costs are

incurred and when they may be recovered from

consumers. And the increase in retail prices resulting

from the recovery of DSM costs shall reduce future

sales through price elasticity of demand impacts.

An investment in DSM in a particular year haseffects which last many years into the future. A one-

time payment by the utility to foster consumer

investments in energy efficient equipment may impact

demand on the utilitys system, the utilitys load

shape, operating costs, and the utilitys capacity needs

over the life of the equipment promoted through a

DSM program. And through price elasticity effects,

the impact on electricity rates of the recovery of DSM

costs will place future demand growth on a certain

trajectory.

Further complicating the analysis, there is consider-able feedback among key variables. DSM investments

affect retail prices, operating costs, and capital costs.

Operating costs and retail prices affect the utilitys

interest in pursuing DSM. Thus, the level and type of

DSM investment, retail prices, operating costs, and

capital costs should be treated endogenously.

If rate cases become less frequent under traditional

ratemaking, the utilitys interest in avoiding uneco-

nomical retail sales will increase. Thus, investments

in load management, for example, may grow. As rate

changes become less frequent, the utility may retainany profits from savings in operating costs for a

longer period of time, and any detrimental reductions

in allowed return on rate base will be recognized in

retail prices with a longer time lag.

In summary, a utility under traditional regulation

weighs various benefits and costs in order to

determine whether to invest in various types of

DSM. Investment in DSM requires that the profits

resulting from avoiding uneconomical sales exceed

foregone future returns on rate base. Price elasticity

effects resulting from efficiency gains and programcosts also play a role. The DSM decision is affected

by a variety of parameters, including the length of

time between rate adjustments, the costs and impact

of various types of DSM, the operating costs which

may be avoided through reliance upon DSM, the

supply-side investment costs potentially avoided by

the DSM, the utilitys discount rate, the price

elasticity of demand, and various regulatory practices.

Our analysis extends beyond a simple Utility Cost

Test (California Public Utilities Commission and

California Energy Commission2001) by consideringprice elasticity effects and the AverchJohnson effect.

The allegation that utilities under traditional regu-

lation have no incentive to invest in DSM absent

externally imposed goals or regulatory requirements

(Moskovitz et al.1992) overstates the situation. While

it is indeed unlikely that an investor-owned utility

would take actions to reduce sales upon which profits

can be earned, some types of retail sales are unlikely

to prove profitable once all the relevant long-term

1 Of course, non-economic motives may be considered in

addition to the economic benefits and costs considered here.These may include customer service goals and corporate

sustainability commitments.

Energy Efficiency (2012) 5:393410 395


4/18

revenues and costs are considered. Targeted DSM

programs can be used to reduce these types of sales.

Incentives to invest in DSM under decoupling

Under the form of decoupling considered here, the

regulatory authority seeks to assure the utility that itwill receive a certain level of margin or profit on

providing service to each customer. The per-customer

profit is calculated based on test-year costs and the

test-year levels of sales. The utilitys expected

recovery of this level of per-customer profit is

unaffected by the incremental impact of DSM on the

current years level of sales. Prices are adjusted each

year. Any under- or over-recovery of margins in

1 year that is attributable to deviations of actual sales

from the level of sales upon which rates are set is

reconciled through an adjustment in rates the followingyear.

Under decoupling, the utilitys incentive to achieve

efficiencies by reducing uneconomical sales is greatly

weakened. The reconciliation process prevents the

utility from keeping the profits resulting from effi-

ciency gains, although the utility may see a small

benefit from achieving such gains if the additional

profits are returned to consumer during the following

year without interest.

Under the form of decoupling considered here, the

utility has an incentive to invest in the types of DSMwhich will reduce the utilitys sales. Each year, retail

prices are adjusted based on the difference between

expected per-customer margins or profits before any

DSM investments and the margins recalculated based

on the actual level of sales taking into account the

effects of the DSM programs on sales. This adjustment

factor is:

PtQt1 EXPENSEt1

CUSTCOUNT

PtQt EXPENSEt1

CUSTCOUNT

1

which may be simplified to:

PtQt1 Qt

CUSTCOUNT 2

Qt1 is the test-year level of sales. This historical sales

level may be adjusted for known and measurable

changes (e.g., customer growth), but not for the

impacts of DSM programs. Qt is the actual sales

volume in year t after any investment in DSM.

EXPENSEt1 denotes the historical test-year level of

expenses, while CUSTCOUNT represents the number

of customers on the utility system.

Once multiplied by the number of customers, Pt(Qt1Qt) becomes an addition to the utilitys

revenue requirement. Note that the utility may

increase its revenues by increasing the gap betweenhistorical sales and actual sales. The higher revenues

will enhance profits.

It is sometimes alleged that decoupling cannot

encourage investments in DSM. It will merely

remove disincentives (Moskovitz et al.1992), so that

an additional external goal or explicit incentive is

necessary in order to motivate the utility to pursue

DSM investments. Yet decoupling can actually induce

DSMparticularly forms of DSM which would result

in large decreases in utility sales. The utility has a

financial incentive to increase the rate adjustment toits revenue requirement since a higher adjustment to

rates will lead to higher revenues and profits.

As is the case under traditional regulation, price

elasticity effects and the impact of DSM on opportu-

nities to receive future returns on investments in new

capacity which will expand its rate base complicate

the story. Relative to traditional regulation, the utility

has less interest in constraining its retail prices over

time. Any per-customer increase in profits from an

increase in sales which is induced by a price reduction

is returned to consumers. Nonetheless, there is a(typically, small) countervailing effect. Price reduc-

tions lead to increased future capacity needs, which

may in turn lead to greater returns for the utility.

The utilitys interest in earning returns from an

increasing rate base may remain strong, but factors

such as the cost of capital and avoided costs will

determine the strength of this motive. These factors

suggest that the type of DSM favored by a utility

under decoupling will be quite different from the type

of DSM pursued by a utility under traditional

regulation. Particularly if there is significant timebetween rate changes, the utility under traditional

regulation will favor the types of DSM which will

permit it to reduce uneconomical salese.g., load

management orpeak clippingprograms. In contrast,

the utility under decoupling is more likely to pursue

forms of DSM which will result in large reductions in

overall energy sales or off-peak sales. While the

utility under traditional regulation must weigh the

benefits of higher profits from reduced uneconomical



5/18

sales against reduction in long-term returns if load

management or peak clipping reduces future capacity

needs and the associated returns from rate base

investments (albeit, complicated by the other factors

noted here), the utility under decoupling sees the

reduction in long-term returns but sees fewer rewards

from reducing operating costs. As with the traditionalregulation case, the relative strength of all of these

effects depends upon the values assumed for price

elasticities, rates of return on rate base, the supply-

side investment costs potentially avoided by the

DSM, the utilitys discount rate, and other factors.

Incentives to invest in DSM under an LRAM

The LRAM mechanism considered here is the lost

contribution to fixed cost approach used in Arkansas

and the Carolinas in the USA. The LRAM iscalculated as the revenue loss associated with DSM

(i.e., the megawatt hour sales lost as a result of DSM

multiplied by the retail price) minus the operating

costs which are avoided as a result of the foregone

sales. This difference is divided by sales which are

adjusted for the impact of DSM. Using this factor, the

retail price is raised to compensate the utility for any

revenue loss resulting from DSM investments. The

prospect of this additional charge on retail bills seeks

to ensure that the utilitys profits are no lower than

under traditional regulation.A properly designed LRAM removes traditional

regulations disincentive for the utility to reduce

profitable types of retail sales (e.g., baseload or off-

peak sales). The remaining factors (e.g., price elasticity

effects and the ability to earn returns from future

rate base additions) determine whether an LRAM

can induce any DSM absent an external goal for

energy efficiency. The utilitys incentive to reduce

uneconomical retail sales (e.g., sales during peak

periods) is muted, since the LRAM mechanism

seeks to maintain the profit margins on retail saleswhich were established by the regulator.

Modeling approach

The typical regulated utility described in this analysis

selects an optimal level and mix of DSM programs so

as to maximize the present value of profits over T

years subject to a demand constraint, requiring it to

meet the energy needs of its ratepayers. A nonlinear

optimization model was developed using General

Algebraic Modeling Software (GAMS) to examine

the impacts of various regulatory incentive mecha-

nisms upon utility earnings and rates under various

scenarios.2 The base-case scenario will first be

described, providing calculations of a utilitys profits

and optimal DSM portfolio under the following

regulatory systems:

1. Traditional ratemaking with an annual adjustment

of prices

2. Traditional ratemaking with rate cases every

5 years

3. Decoupling of margins or revenues from sales

volumes with an annual adjustment of prices

4. Traditional ratemaking is adjusted using an

LRAM and new rates are set every year

5. Traditional ratemaking is adjusted using an

LRAM and new rates are set every 5 years

We then alter some base-case assumptions to demon-

strate how the results are sensitive to the natural

growth in the utilitys demand, DSM program costs,

the utilitys level of avoided capacity cost, the level of

any externally established goals for DSM, and other

factors. We assume that this is a vertically integrated

utility with fuel costs.

Base-case scenario: traditional ratemakingwith annual adjustment of prices

Total energy consumption is assumed to be 43,800 MWh

in yeart=0, corresponding to a utility system with a 10-

MW peak demand and a 50% load factor. Under base-

case assumptions, demand would grow at a 3% annual

rate in the absence of any DSM.

We define three different load levels: Low_Demand,

Medium_Demand, and Peak_Load. In the absence of

any DSM, total energy consumption is allocated to each

category of load using the allocators presented inTable1.

In yeart1, the utility faces a decision regarding the

appropriate quantity of three different types of

DSM in which to invest: Baseload_DSM (e.g.,

efficiency programs targeting end-uses with long

operating periods such as motors or refrigerators),

2 The GAMS code used in this analysis is available from the

author upon request.



6/18

Medium_DSM (e.g., programs that seek to reduce air

conditioning or space heating needs by promoting the

installation of insulation, other building shell meas-

ures, or HVAC equipment efficiency measures), and

Load_Management (e.g., the control of air condition-

ing equipment or curtailments to industrial operations

for a very limited number of hours per year). dis used

to denote a type of DSM. Each of these types of DSMis expected to have different impacts on the energy

consumption assigned to the three load level catego-

ries, denoted . For example, Load_Management is

assumed to impact energy consumption only during

the periods of the highest demand. The allocation of

the DSM program impact on future energy consump-

tion is provided in Table2.

DSM programs are assumed to impact current and

future sales and peak demand over a 10-year horizon.

QtQ

Baseline

t X

d ALLOCDSMd DSMd

h Pt1 Pt0 =Pt0 QBaselinet

8; t> t0

3

The pre-DSM level of sales within each load level

category denoted QBaselinet is adjusted to reflect

the energy savings achieved through the DSM

decision in year t1. Demand is further adjusted for

the price elasticity of demand, based on price changes

in the previous year.

3

The price elasticity of demandfor electricity, , is set to0.2.4 This results in a post-

DSM level of sales,Qt. The DSM impact is assumed

to persist over the 10-year life of the DSM measures

with no degradation of savings. It is assumed that

implementation of the DSM programs occurs gradu-

ally within the year of implementation, t1, such that

the impact of DSM on sales in t1 is one half of the

impact of the DSM in subsequent years.

The unit cost of the total savings (across all load levelcategories) of Baseload_DSM and Medium_DSM is

modeled as an increasing function of the level of DSM

investment, consistent with the concept of an upward-

sloping supply curve for DSM. An exponential function

is adopted here. The unit cost (in dollars per total

megawatt hour savings) is multiplied by the DSM

quantity (in megawatt hours) in order to calculate the

total DSM cost.

DSMCOSTd 2 e0:0295 DSM

DSMd

d Baseload DSM; Medium DSM4a

This DSM cost function provides a reasonable rela-

tionship between DSM investment and the unit cost of

DSM, as noted in Fig.1.

The cost of Load_Management is set at $1,141/MWh,

such that the cost is $50/kW or $50,000/MW divided by

the 43.8 h associated with the Peak_Load period.

DSMCOSTLoad Management

1; 141 DSMLoad Management 4b

A cap is placed on the amount of Load_Ma-

nagement that may be obtained by the utility, to

reflect limits to the market potential. This cap,

stated as a megawatt hour value, is equivalent to

1 MW multiplied by the 43.8 h in the Peak_Load

period.

DSMLoad Management 43:8 5

The utilitys variable operating cost varies for the three

different load level categories, per Table3. If we assume

a vertically integrated electric utility, operating costs

might include fuel or generation costs. Note that prices

during the short Peak_Load period may reach $1,000/

MWh.

DSM investments can result in the avoidance of

future capacity costs. Thus, the utilitys future rate

3 Efforts to permit demand to adjust to contemporaneous pricechanges resulted in convergence problems. Consequently, alagged response is modeled here.4 A wide range of long-term price elasticities are reportedin the literature. Espey and Espey (2004) report a range

from2.25 to 0.04.

Table 1 Baseline allocation of energy consumption amongload level categories

Allocation

Low_Demand 0.500

Medium_Demand 0.495

Peak_Load 0.005Total 1.000



7/18

base is affected by DSM investments in year t1. The

utilitys rate base in years t>1, after adjustments are

made to reflect the DSM investment, is modeled as:

ADJRBtRBt X

d

ALLOCDSMd;0Peak Load0

DSMd=43:8 ACC 8t> 1

6

RB is the pre-DSM projection of the utilitys rate

base, 43.8 is the number of hours in the Peak_

Load period and is used to convert the energy

conserved in that period into a megawatt value.

ACC represents the annual capacity cost which

can be avoided as a result of the DSM. The

utilitys rate base is set at $10 million in year t0and assumed to increase by the same rate of growth

as demand (3% under base-case assumptions) absent

any DSM investment. Annualized avoided capacity

costs are set at $100,000/MW.The utilitys non-DSM expenses are the sum of

variable operating costs, depreciation on rate base,

and interest payments to bondholders. None of

these non-DSM expenses is reconcilable. In t1 the

cost of the utilitys investment in DSM will also be

treated as an expenseas is common under tradi-

tional regulationbut is collected through a separate

rate rider.

EXPENSEt OpCost Qt dep ADJRBt

debtcost ADJRBtX

d

DSMCOSTd 8t 1

7a

EXPENSEt OpCost Qt dep ADJRBt

debtcost ADJRBt 8t> 1

7b

Operating costs for various load levels, OpCost, were

provided in Table3. The average annual depreciation

rate, dep, is set at 0.15, such that the annual

depreciation expense is calculated as 15% of the rate

base. Following the example provided by Weston

(2008), we assume a capital structure of 55% debt and

a cost of debt of 8%. Multiplication of the debt ratioby the cost of debt yields 0.044 for the weighted cost

of debt, debtcost.

For simplicity, we assume that the utility recovers

its revenue requirement (the sum of expenses and

return) through a volumetric charge. The revenue

requirement is established based upon historical test-

year revenue requirements (i.e., expenses, interest

paid to bondholders, and return) and historical test-

year sales. Taxes are ignored in order to facilitate

model convergence.5 Presently, we assume that prices

5 Attempts to explicitly model federal income taxes as afunction of profits (and, thus, implicitly, a function of all othervariables in the model) resulted in convergence problems.These problems persisted even when rates were set so as toinclude taxes paid in the previous year. Provided that profitsremain positive and the tax rate is a fixed percentage of profits,

after-tax profits will be proportional to pre-tax profits. Andoptimization based on pre-tax profits should yield the same

result as optimization based on after-tax profits.

Fig. 1 Relationship between level of DSM investment and costof conserved energy

Low_Demand Medium_Demand Peak_Load Total

Baseload_DSM 0.500 0.497 0.003 1.000

Medium_DSM 0.400 0.595 0.005 1.000

Load_Management 0.000 0.000 1.000 1.000

Table 2 Allocation of DSM

program impacts amongload level categories



8/18

are re-set each year. Later, we examine a common

situation where rates are changed less frequently.

Pt ADJRBt1 ROE EXPENSEt1 =Qt1 8t

8

The allowed rate of return on equity established by

the regulatory authority for ratemaking purposes,

ROE, is set at 0.0495. This is again based on the

example provided by Weston (2008), which assumes

a capital structure with 45% equity and a cost of

equity of 11%. Multiplication of the equity ratio by

the cost of equity yields 0.0495 for ROE.

Utility revenue is determined by multiplying the

current years price by the adjusted megawatt hour

consumption for the current year.

REVtPtQt 8t 9

While prices are determined by costs in the previous

year, actual utility profits are determined by costs in

the current year:

PROFITt REVt EXPENSEt 8t 10

The level of profit is affected by the rate of return on

equity and regulatory lag, i.e., the differences in costs

and sales in the historical test-year versus the actual

costs and sales which are realized.

Finally, the objective function is the maximization

of discounted profit over the 10-year modeling

horizon:

Max OBJXT10t1

PROFITt1

1 rt 11

A 9% discount rate forr is used in the calculation of

present values.

The solution to Eqs.1 through10 yields the profit-

maximizing level of DSM investment in t1 and the

resulting levels of annual profits, annual revenues,

annual operating cost, DSM cost, future rate base, annual

expenses, annual adjusted demand, and annual prices.

Base-case scenario: traditional ratemaking

with adjustment of prices every 5 years

Under traditional regulation, it is unusual for rates to

be adjusted every year. In fact, in many US states, it

takes far longer than 1 year for a regulatory authority

to process an application by a utility to change rates.Consequently, a system where prices are changed

every 5 years is also considered. Regulatory lag or the

time between when costs are incurred and the utility is

afforded a reasonable opportunity to recover its

reasonable and necessary costs tends to benefit a

utility with sales growth and low marginal costs, but

fixed retail prices over an extended period of time

may financially distress a utility facing a decline in

sales, cost inflation, or sales growth combined with

high marginal costs.

To model this regulatory system, Eq. 8is replacedwith:

Pt ADJRBt0 ROE EXPENSEt0 X

d

DSMCOSTd=Qt0

t t1

12a

Pt ADJRBt1 ROE EXPENSEt1X

d

DSMCOSTd=Qt1

8t2 t2 to t5

12b

Pt ADJRBt5 ROE EXPENSEt5 =Qt5 8t2 t6 to t10

12c

Thus, prices in years 2 through 5 are based on

expenses and return on rate base in year 1, while costs

in year 5 determine prices in years 6 through 10. The

price in year t=1 reflects the costs of the DSM

investment incurred in that year and the rate base andexpenses in year 0. Otherwise, the equations pre-

sented earlier are applied.

DSM expenses in t1 are fully recovered by the

utility in yeart1through a rate rider or other means.

They are a component of the price in t1. A full rate

case is not necessary in order to enable the utility to

recover these costs. These DSM costs are ignored

when base rates are later changed through rate

cases.

$/MWh

Low_Demand 30

Medium_Demand 40

Peak_Load 1,000

Table 3 Variable operating

cost



9/18

Base-case scenario: decoupling

To calculate the profit-maximizing level of DSM

investment selected by a utility under a decoupling

ratemaking scheme, the expected margin or profit

to be earned on each customer is calculated based

on test-year costs and test-year levels of sales. Theutilitys recovery of this level of profit for a set of

customers is ensured, regardless of any incremental

impact of DSM on the current years level of

sales. Prices are adjusted each year. Any under- or

o ve r- re co ve ry o f marg in s i n 1 y ea r t ha t i s

attributable to deviations of actual sales from the

l ev el of sa les u pon w hi ch r at es ar e s et is

reconciled through an adjustment in rates the

following year. Equations 8 and 9 are replaced by

the following.

The expected margin per customer is based ontest-year sales. CUSTCOUNT refers to the number

of customers and is set at 100. It is assumed that

there is no growth over time in the number of

customers.

ExpectedMarginPerCustomert PtQt1 EXPENSEt1 =

CUSTCOUNT 8t> 0

13a

The expected margin per customer is based on

actual sales, which may have been affected by theutilitys investment in DSM.

ActualMarginPerCustomert PtQt EXPENSEt1 =

CUSTCOUNT 8t> 0

13b

The difference between expected and actual margins

per customer is then calculated.

RateAdjustmentt

ExpectedMarginPerCustomert ActualMarginPerCustomert

CUSTCOUNT 8t> 0

14

Note that this formulation is algebraically equiv-

alent to the revenue per customer decoupling

approach that has been implemented in some

jurisdictions, since expenses in the previous

period appear in both Eqs. 13a and 14 and it is

assumed that the customer count has not changed

over time.

Any under-recovery of margins from the previous

year is added to the price charged in the current year

to calculate the price under decoupling.

PDECOUPLEDt

ADJRBt1 ROE EXPENSEt1 RateAdjustmentt1

=Qt1 8t> 0

16

Utility revenues are now determined by multiplying

the price established by the decoupling scheme by the

adjusted megawatt hour sales.

REVtPDECOUPLEDt Qt 8t 17

Base-case scenario: LRAM with annual rate cases

and rate cases every 5 years

This regulatory approach is a hybrid between

traditional regulation and decoupling. Unlike tradi-

tional ratemaking, lost revenues are calculated and

added to the utilitys revenue requirement. Rates

may be adjusted each year or every 5 years. If

rates are set every year, this would be similar todecoupling where the adjustments for lost margins

or lost revenues would occur every year. If rates

are set every 5 years, these calculations are made

only twice during the 10-year modeling horizon,

similar to the case of traditional ratemaking with

regulatory lag. Equations3 through7aand7band9

through11are retained from the model of traditional

regulation.

Prices are adjusted in order to compensate the

utility for lost contribution to fixed costs attributable

to DSM. Fixed costs are all costs other than variableoperating costs and DSM costs. The rate adjustment is

calculated as:

RateAdjustmentt

Pd

DSMdPtP

OpCostQBaselinet Qt

P

Qt

8t> 0

18



10/18

If rates are re-set every year, prices are set based

on the prices set under traditional ratemaking with

annual rate case, plus the rate adjustment for lost

revenues:

Pt ADJRBt1 ROE EXPENSEt1 =Qt1

RateAdjustmentt 8t> 0

19

For the setting of prices every 5 years, we use:


d

DSMCOSTd=Qt0

RateAdjustmentt t t1

20a


d

DSMCOSTd=Qt1

RateAdjustmentt 8t2 t2 to t5

20b

Pt ADJRBt5 ROE EXPENSEt5 =Qt5 RateAdjustmentt

8t2 t6 to t10

20c

Again we assume that DSM expenses int1are fullyrecovered by the utility in year t1 through a rate rider

or other means. They are a component of the price in

t1. A full rate case is not necessary in order to enable

the utility to recover these costs. These DSM costs are

ignored when base rates are later changed through

rate cases.

In practice, the treatment of load forecast error

would also distinguish the LRAM approach from

decoupling. However, errors in load forecasts are not

considered here.

As noted earlier, various regulatory authoritieshave adopted different variations upon these themes

when implementing LRAMs or decoupling mecha-

nisms. A number of simplifying assumptions have

been made here in hopes of keeping these calculations

somewhat transparent. These simplifications have

been balanced against the need to retain sufficient

detail in order to allow an examination of utility

motivations to invest in DSM under various economic

conditions and ratemaking schemes.

Base-case results

Table 4 summarizes the results under the base-case

assumptions described above. Under traditional rate-

making with annual price changes, the utility would

decline to invest in DSM. Under traditional regulation

and regulatory lag, the utility invests in a maximumlevel of load management but declines to invest in

other forms of DSM. Load management becomes an

attractive investment, since it lowers the utilitys

operating expenses and the utility can retain the

resulting profits if prices are changed infrequently.

As anticipated earlier in this paper, decoupling leads

the utility toward investments in energy conservation

(i.e., Baseload_DSM and Medium_DSM). The

LRAM cases also lead to utility investments in

conservation but at lower levels of DSM investment

than under decoupling.Because load grows at a higher rate than costs

under the base case, regulatory lag leads to higher

profits for the utility under traditional regulation. And

the more regulatory lag, the better, from the utilitys

perspective. Up to 5 years of regulatory lag is

preferred (i.e., leads to higher earnings for the utility)

to traditional regulation with 1 year of lag.

The shift in investment from load management under

traditional regulation with infrequent rate cases to base

load and medium load DSM under decoupling and

LRAMs would seem consistent with the explanationoffered by Brennan (2010). Load management reduces

the utilitys average operating costs and thus leads to

some short-term efficiencies. Decoupling and LRAMs

blunt the utilitys interest in pursuing these efficiencies.

It is assumed that base rates are based on historical

test-year costs, unlike the projected rate year costs

which are applied in California and some other states

for ratemaking purposes. This introduces an incentive

for the utility to utilize DSM in order to change its

level of sales and expenses from test-year levels if

such changes will lead to higher profits.Since RateAdjustmentt in the decoupling case is

essentially Pt (Qt1Qt), the utility may increase its

revenues by increasing the gap between historical

sales and actual sales, as noted earlier. In this

particular case, the utility increases its profits from

$2.88 million to $2.93 million by investing in DSM in

order to reduce sales.

As can be seen from Fig.2, prices are perfectly flat

under traditional ratemaking with annual rate cases.



11/18

Sales growth equals the growth in the utilitys rate

base. Per megawatt hour operating costs are assumed

to be fixed over time. Since the utility makes no

investment in DSM under this scenario, there is no

need to recover any DSM program costs through

prices, and the proportion of sales within each of the

load levels remains constant.Under traditional ratemaking with regulatory lag,

there is an initial increase in price to enable the utility

to recover its investment in load management. Over

time, this investment in load management leads to

lower prices to consumers.

Decoupling and the LRAM cases lead to higher

prices. The utility invests in DSM and the cost of

DSM is recovered from ratepayers in t1. The invest-

ments in base load and medium DSM results in lower

sales, and the spreading of the utilitys fixed costs

among fewer billing determinants (i.e., sales) leads toprices which are higher than under the scenarios

assuming traditional regulation (and minimal DSM

investment).

Are these robust results which would hold over a

variety of utilities and economic environments?

Unfortunately, the answer is no. As shown below,

these results are sensitive to the utilitys natural load

growth, cost structure, DSM costs, price elasticity of

demand, return on equity, and regulatory constraints

faced by the utility. The approach favored by a utility

may change under different economic environments.

Changes in avoided capacity costs

To explore the AverchJohnson effect, the model was

solved under various assumed values for avoided

capacity costs. Avoided capacity cost of $0 might

reflect a utility with sufficient transmission and

distribution capacity that relied entirely on purchased

power (e.g., a distribution cooperative or a small

municipal system) or that was not involved in power

generation (e.g., a transmission and distribution utility

in an unbundled electricity market). When avoided

capacity costs are reduced, DSM would have a

diminished effect on reducing the future rate base

upon which the utility earns future returns. When theyare increased, DSM has a bigger impact on the

erosion of future returns on rate base. Results for

avoided capacity costs of $0 and two times the base-

case assumption are reported in Tables 5and6.6

A reduction in avoided capacity costs motivates

the utility to now invest in load management under

both traditional regulation and decoupling. This type

of DSM will reduce the utilitys operating costs by an

amount more than it will decrease the utilitys

revenues, andgiven the assumption of no avoided

capacity costit will not adversely impact theutilitys future rate base and return on rate base.

Ironically, this increase in DSM investment would

coincide with a situation where DSM (particularly,

load management) may not pass some of the usual

cost-effectiveness tests for screening DSM programs

because of the zero avoided capacity cost (California

Public Utilities Commission and California Energy

Commission2001). This change in avoided costs has

little effect on profits under any form of ratemaking.

There is a small effect on investments in base load

and medium DSM by the utility under decoupling.As shown in Table 6, there is investment in load

management in only the case of traditional regulation

with regulatory lag when avoided costs are doubled

relative to the base-case assumption. Figure3further

6 Because utility investments in infrastructure can have very

long accounting lives and the model presented here examinesonly a 10-year planning horizon, some of the results presented

here may be understated.

Ratemaking

scheme

Frequency of

rate changes

Utility pre-tax earnings ($ millions

in present value over 10 years)

DSM investment (MWh)

BaseloadDSM

MediumDSM

Loadmanagement

Traditional Annual 3.67 0 0 0

Traditional Every 5 years 3.87 0 0 43.8

Decoupling Annual 2.94 999 1,008 0

LRAM Annual 4.01 786 776 0

LRAM Every 5 years 3.91 576 576 0

Table 4 Results under base-

case assumptions



12/18

displays the relationship between avoided cost and the

utilitys investment in DSM. As the avoided cost

increases, investment decreases gradually under the

decoupling case. Changes in avoided costs appear to

have little impact on DSM investment under theLRAM cases.

External DSM goal

As a variation from the base-case scenario, we

consider a situation where a regulatory or govern-

mental authority has set a minimum goal for DSM.

This has become a common practice in US states and

in nations which seek to meet targets for reductions in

greenhouse gasses. This becomes an additionalconstraint to the model:

DSM MinDSM 21

MinDSM is set at 2,000 MWh. All other equations of

the model are retained.

Of course, the impacts upon the utility will depend

upon how high the external goal is set. The external

goal would have no effect on a utility which already

planned to exceed the level of energy conservation

imposed by the goal, as in the case of the utility under

decoupling here. The level of avoided cost and the

relative costs of the DSM options will also affect the

preferred source of the energy savings. Results are

provided in Table7.

Variations in load growth

It is frequently observed that decoupling is more

popular among natural gas utilities, which often face

declining per-capita natural gas consumption in North

America and among electric utilities in regions where

there is little or no growth in electrical demand (e.g.,

California). By eliminating all load growth in the

model, this can be demonstrated.As suggested by Table 8, utility profits are

significantly reduced under traditional regulation if

there is no natural load growth. The utility fails to

realize a benefit from the regulatory lag inherent in

traditional ratemaking if there is no load growth.

While it is the least-preferred ratemaking treatment

for the growing utility, decoupling yields profits

similar to the other approaches in a system with no

load growth. Profits under an LRAM decline if there

$80

$85

$90

$95

$100

$105

t1 t2 t3 t4 t5 t6 t7 t8 t9 t10

Dollars

PerMWhPrice

Traditional Ratemaking withAnnual Price Changes

Traditional Ratemaking withRegulatory Lag

Decoupling with Annual PriceChanges

LRAM with Regulatory Lag

LRAM with Annual RateChanges

Fig. 2 Prices under fiveratemaking approaches

Ratemakingscheme

Frequency ofrate changes

Utility profit ($ millions inpresent value over 10 years)


Baseload

Mediumload

Loadmanagement

Traditional Annual 3.70 0 0 43.8


Decoupling Annual 2.96 1,023 1,042 43.8

LRAM Annual 4.01 786 777 0


Table 5 Results whenavoided capacity cost isreduced to $0



13/18

is no load growth, although the LRAM remains an

attractive option to the utility. The decoupling

mechanism may be slightly preferred to traditional

regulation if an external goal for energy savings is

imposed on the utility and there is no load growth.

When 10% annual decreases in sales are assumed,

decoupling becomes favored over traditional regula-tion, as is evident from Table 9.

DSM investments under higher returns on equity

and decoupling

Kihm (2009) finds that if a regulator sets the utilitys

rate of return in excess of its cost of capital, then

decoupling will fail. The utility will prefer sales growth

that will lead to greater supply-side investments and

higher dollar returns on rate base. While the utility willcontinue to invest in DSM under decoupling as the ROE

increases in this model, the utilitys overall interest in

DSM does indeed decline as the ROE is increased. This

is depicted in Fig. 4. As the ROE increases, DSM

investments shift away from load management, which

has a high impact on the need for additional supply-

side capacity (and rate base) and toward base load

DSM. The base load and medium load DSM invest-

ments have a lesser impact on the utilitys rate base and

profits. A potential topic of further study is whether

this effect could be negated if DSM expenses were

capitalized, as permitted in some states.

DSM investments under higher price elasticity

of demand

The price elasticity of demand also affects DSM

investments. As indicated in Table10, when the price

elasticity of demand is raised (in absolute value)

from 0.2 to 0.3, the utility under traditional regulation

and annual rate cases invests in load management. This

investment reduces the utilitys operating cost, reduces

prices, and increases sales and profits. Thus, ironically,

load management may lead to a higher level of sales.With the higher elasticity, the AverchJohnson effect is

somewhat overcome. The more-elastic demand curve

results in higher levels of investment in baseload DSM

and medium DSM under decoupling, as reported in

Table10. Yet, DSM investment declines slightly under

the LRAM mechanism, as demand becomes more

responsive to price changes.

0

500

1,000

1,500

2,000

2,500

$0

$25,

000

$50,00

0

$75,

000

$10,00

0

$125

,000

$150

,000

$175

,000

$200

,000

$225

,000

MWhofDSMInvestment

Traditional Ratemaking withAnnual Price Changes

Traditional Ratemaking with

Regulatory Lag

Decoupling with Annual Price

Changes

LRAM with Regulatory Lag

LRAM with Annual Rate

Changes

Fig. 3 Relationship between

avoided cost (dollar permegawatt) and total DSM

investment (annual megawatthours)

Ratemakingscheme

Frequency ofrate changes

Utility profit ($ millions inpresent value over 10 years)


Baseload

Mediumload

Loadmanagement



Decoupling Annual 2.93 981 982 0LRAM Annual 4.00 785 775 0


Table 6 Results whenavoided capacity cost isdoubled



14/18

Decouplings effects under further changes

in assumptions

This hypothetical utility continues to invest in DSM

under decoupling under a wide range of assumptions

regarding the cost of DSM, as can be seen from

Table11. Base-case results were reported earlier. With

a 500% increase in the cost of Baseload_DSM and

Medium_DSM with no change in the cost of load

management, there is a modest reduction in the

utilitys investment in DSM. If the cost of load

management is reduced by 75% and the base-case

costs of the other two types of DSM are retained, thenthe utility invests in load management under decou-

pling. Under a ten-fold increase in the costs of all

types of DSM, the decoupled utility continues to

invest in Baseload_DSM but drops any investment in

other types of DSM.

Economies of scale in DSM program costs

Earlier, it was assumed that the utility faced an

upward-sloping supply curve for DSM. That is,

there were diminishing returns to additional DSM

investments and the cost of achieving additional

energy savings was increase on a dollar per

megawatt hour basis. It has sometimes been

reported that the cost of achieving additionalDSM may decline across relevant program scales,

such that there are economies of scale of running

such programs.

Ratemaking

scheme

Frequency

of rate changes

Goal? DSM investment (MWh)

Base load Medium load Load management

Traditional Annual No goal 0 0 0

Goal 972 984 43.8

Traditional Every 5 years No goal 0 0 43.8

Goal 940 1,017 43.8

Decoupling Annual No goal 999 1,008 0

Goal 999 1,008 0

LRAM Annual No goal 786 776 0

Goal 1,000 1,000 0

LRAM Every 5 years No goal 576 576 0

Goal 1,000 1,000 0

Table 7 External DSMgoal placed upon utility

Ratemaking scheme Frequency of rate changes Goal? Utility profit ($ millions in presentvalue over 10 years)

3% growth No growth

Traditional Annual No goal 3.67 3.19

Goal 3.56 3.06

Traditional Every 5 years No goal 3.87 3.40

Goal 3.67 3.16

Decoupling Annual No goal 2.94 3.22

Goal 2.94 3.22

LRAM Annual No goal 4.01 3.50

Goal 4.05 3.54

LRAM Every 5 years No goal 3.91 3.73

Goal 3.74 3.17

Table 8 Utility profits under3% growth versus nogrowth



15/18

To model this situation, Eq. 4a is replaced with

Eq.22:

DSMCOSTd 0:06 DSMd 0:000004 DSMd2

d Baseload DSM;Medium DSM

22

Under this scenario, the utility would invest in load

management under all regulatory mechanisms, includ-

ing traditional regulation with minimal regulatory lag

and decoupling. As in the base case, decoupling would

continue to be the only case which would elicit

investments in medium load and baseload DSM, and

the level of investment in those forms of DSM would be

over four times greater (i.e., 4,660 MWh of baseload

DSM and 4,748 MWh of medium load DSM).

Alternative growth rates for the rate base

It was earlier assumed that the rate baseprior for any

adjustments for the effects of DSM on the utilitys

capacity needswould rise at the rate of demand growth

(3%). If it is instead assumed that RBt has no growth,

then results identical to those presented in Table 4 for

DSM investments continue to hold for the traditional

regulation cases. The results for an LRAM with rate

cases every 5 years are virtually the same as base-caseresults. However, a slightly lower investment in DSM

under the decoupling case is found than under the base-

case assumptions. Under an LRAM with annual rate

adjustments, investments in DSM decline by about 4%.

If we instead assume a 10% annual growth for RB t,

then the results presented in Table 12 arise. The

ratemaking treatment involving up to 5 years of

regulatory lag hinders the utilitys ability to quickly

change prices in order to recover its rising costs.

Slightly higher levels of DSM investment are under-

taken under decoupling and the LRAMs.

Lower ROE under decoupling

Regulatory authorities have sometimes lowered a

regulated utilitys allowed rate of return on equity

0.04

0.05

0.06

0.07

0.08

0.09 0.

10.11

0.12

Return on Equity Permitted by Regulator

MWhofDSM

Base Load DSM

Medium Load DSM

Load Management

0

200

400

600

800

1000

1200Fig. 4 Under decoupling,DSM investments declineas return on equity increases

Ratemaking scheme Frequency of rate changes Goal? Utility profit ($ millions in present

value over 10 years)

3% growth Negative 10% growth

Traditional Annual No goal 3.67 2.05

Goal 3.56 1.85

Traditional Every 5 years No goal 3.87 2.29

Goal 3.67 1.94

Decoupling Annual No goal 2.94 3.51

Goal 2.94 3.49

Table 9 Utility profits under

3% growth versus 10%annual negative growth



16/18

for a utility under a decoupling scheme, under the

theory that risks of sales and revenue volatility havebeen reduced for the utility and shifted to ratepayers.

Consequently, the utility may be able to attract capital

at a lower cost.

Table13compares base-case results to a model run

where the allowed ROE used for ratemaking purposes

was reduced from 0.0495% to 0.0395%. Base-case

assumptions pertaining to the cost of debt are

retained. With the lower ROE, the utilitys profits

are understandably lower. The lower ROE leads the

utility to slightly higher DSM investments, including

some load management.

Cost function for load management

Under base-case assumptions, load management was

modeled using a constant per megawatt hour cost and

the quantity of load management that could be

selected was limited. Upward-sloping unit cost supplycurves for load management were also developed,

mimicking the approach used for the other two types

of DSM. When simple linear functions were tested,

the results were similar to the base-case findings

load management was selected by a utility under

traditional regulation, but not selected by utilities

under decoupling or an LRAM (unless a goal was

imposed upon the utility and load management was

sufficiently inexpensive). However, the quantities of

load management procured by the utility under

traditional regulation were often unrealistically highand the inclusion in the model of realistic supply

curves with polynomial or exponential forms tended

to lead to convergence problems.

Further considerations

The set of variables examined above to show the

economics of regulatory mechanism designed to

foster energy efficiency is not exhaustive. The scope

of the utilitys operations, the presence of automaticpass-through clauses for fuel, and political consider-

ations will also have effects.

The utility modeled in this paper was assumed to

be a vertically integrated electric utility. Yet, in some

competitive markets such as the Electric Reliability

Council of Texas (ERCOT), unbundled regulated

transmission and distribution utilities provide ratepay-

er funded energy efficiency programs. Transmission

and distribution services providers have a more-

Table 11 Results for decoupling under alternative assumptions

Scenario DSM investment (MWh)

BaseloadDSM

MediumDSM

Loadmanagement

Base case 999 1,008 0

500% increase in cost of|baseload and mediumload DSM

591 600 0

75% decrease in cost ofload management

1,001 1,010 43.8

10-fold increase in cost ofall types of DSM

1,616 0 0

Ratemaking

scheme

Frequency

of rate changes

Price elasticity

of demand




0.3 0 0 43.8


0.3 0 0 43.8

Decoupling Annual 0.2 999 1,008 0

0.3 1,071 1,077 0

LRAM Annual 0.2 786 776 0

0.3 765 756 0


0.3 558 559 0

Table 10 Variation in priceelasticity of demand (and noexternal DSM goals)



17/18

limited scope of operations. The models constructed

here would need to be substantially revised to reflect

their operations and economic incentives. Natural gas

utilities which primarily provide a pipeline function

might face different incentives and disincentives.

Generation-only entities (e.g., independent power

producers) or retail-only entities (e.g., retail electricproviders in the ERCOT market) pose other distinct

cases for which the analysis presented here would

probably not apply.

Under base-case assumptions, only DSM expenses

were reconcilable. Operating costswhich were

assumed to include fuel costswere not necessarily

reconcilable. Thus, regulatory lag could result in an

under-recovery or an over-recovery of the utilitys

fuel costs in the ratemaking process. However, many

utilities have automatic fuel adjustment clauses. The

presence of such clauses will tend to reduce theimpact of regulatory lag and would certainly affect the

results presented here. Similarly, a natural gas

distribution utility permitted to simply pass-through

the commodity cost of natural gas should be treated in

a different manner.

Finally, it should be noted that political factors

might have as much to do with a utilitys interest in

energy efficiency as strictly economic incentives.

Goals and expectations may be set by local or

regional governments with little attention to what

levels of DSM may prove economical. DSM spendingoften becomes a bargaining chip in rate case

settlement negotiations. Corporate sustainability com-

mitments and customer service goals may also affect a

utilitys motives.

Conclusions

Using a nonlinear optimization model of the DSM

investment decision of a hypothetical electric utility,

we find that regulatory mechanisms designed to

promote energy efficiency and economic conditions

affect not only the level of a utilitys DSM investment

but also the types of DSM programs preferred by the

utility. Under the assumptions made here, decoupling

will indeed lead the utility to greater investment in

energy efficiency than traditional regulation over a

broad range of alternative assumptions. Decoupling

may be particularly effective in promoting DSMfocused on energy conservation, rather than peak

d e ma n d r e du c ti o n ( e .g . , B a se l oa d _D S M o r

Medium_DSM) although the relative costs of differ-

ent types of DSM programs impact this. An LRAM

mechanism may also lead a utility to conservation

investments. But in only one case (i.e., exceptionally

high growth in the utilitys rate base) was the optimal

level of DSM investment higher under an LRAM than

under decoupling.

The type of DSM investments preferred by the utility

is sensitive to avoided costs, program costs, and othervariables. Load management is attractive to a utility

Ratemaking

scheme

Frequency of

rate changes

Utility pre-tax earnings ($ millions

in present value over 10 years)


BaseloadDSM

MediumDSM

Loadmanagement

Traditional Annual 3.35 0 0 43.8


Decoupling Annual 2.76 1,032 1,042 43.8

LRAM Annual 3.56 1,087 1,081 0


Table 12 Results with 10%

growth in rate base beforeDSM

Ratemakingscheme

Utility profit($ millions in presentvalue over 10 years)



Decoupling: base case 2.94 999 1,008 0

Decoupling: lower ROE 2.21 1,034 1,048 43.8

Table 13 Decoupling leadsto a lower ROE



18/18

under traditional regulation and infrequent rate cases.

This utility reduces its operating costs through the load

management and retains the resulting profits until rates

are re-set at the time of a subsequent rate case. At a

sufficiently high price elasticity of demand, load

management becomes attractive under traditional regu-

lation with annual rate changes, as well.Under the assumptions examined here, decoupling

and LRAMs would promote DSM investments with low

load factors (i.e., programs focused on energy conser-

vation throughout the year, rather than peak load

reduction). But this can change if load management

becomes sufficiently cheap.

In an environment of strong growth in energy

demand, a utility is likely to favor traditional regulation

with its built-in regulatory lag. A fast-growing utility

may rightfully see decoupling as a threat to its profits

and oppose any policy changes in that direction. In anenvironment of negative growth, a decoupling mecha-

nism is likely to be preferred over traditional regulation

by the utility. These results conform to the observation

that natural gas utilities experiencing declining sales per

customer and electric utilities in areas with little load

growth tend to favor decoupling.

Beware of those who claim that a single rate-

making approach is the universally best policy to

promote energy efficiency without damaging the

profitability of regulated utilities. The number of

possible situations is limitless and generalizationsare difficult. There is an enormous diversity among

electric utilities with respect to their operating

environment, cost structure, and available DSM

opportunities, and it is not possible to model every

utility or every economic condition. A regulatory

approach that yields some level and mix of DSM

consistent with policy goals in one situation may

yield different results in another environment. And

there are myriad ways to implement decoupling and

LRAM schemes with subtle differences from the

formulas examined here. Nonetheless, some of thefactors likely to affect the utilitys profits under

various ratemaking schemes and the optimal level

and mix of a utilitys DSM investments under various

conditions can be readily identified.

Acknowledgments The author benefited from earlierexchanges with C.K. Woo and I. Horowitz on this topic.

Michele Chait, Parviz Adib, and George Mentrup providedvaluable comments on earlier versions. Three referees provided

exceptionally detailed and invaluable suggestions on the initialsubmission, resulting in significant improvements to this paper.

References

Anderson, J. (2007). ELCON opposes utility revenue decoupling

as a device to open doors to energy efficiency.Platts PowerMarkets Week. March 2.

Averch, H., & Johnson, L. (1962). Behavior of the firm underregulatory constraint. American Economic Review, 52,10521069.

Brennan, T. (2010). Decoupling in electric utilities. Journal ofRegulatory Economics, 38, 4969.

California Public Utilities Commission and California EnergyCommission. (2001).California standard practice manual:

Economic analysis of demand-side programs and projects.

San Francisco: California Public Utilities Commission andCalifornia Energy Commission.

Cappers,P., Goldman, C., Chait, M., Edgar,G., Schlegel, J., Shirley,

W. (2009). Financial analysis of incentive mechanisms topromote energy efficiency: Case study of a prototypicalsouthwest utility. LBNL-1598E. March.

Edison Foundation. (2011). State electric efficiency regulatoryframeworks. Washington, DC: Edison Foundation.

Espey, J., & Espey, M. (2004). Turning on the lights: A meta-analysis of residential electricity demand elasticities.

Journal of Agricultural and Applied Economics, 36(1),6581.

Eto, J., Stoft, S., & Belden, T. (1997). The theory and practiceof decoupling utility revenues from sales. Utilities Policy,6(1), 4355.

Hirst, E., Blank, E., & Moskovitz, D. (1994). Alternative ways

to decouple electric utility revenues from sales. The

Electricity Journal, 7, 5466.Kihm, S. (2009). When revenue decoupling will work. . . and

when it won't. Electricity Journal, 22(8), 1928.Kushler, M., York, D., & Witte, P. (2006). Aligning utility

interests with energy efficiency objectives: A review of

recent efforts at decoupling and performance incentives.

Report no. U061. Washington, DC: American Council foran Energy Efficient Economy.

Moskovitz, D., Harrington, C., & Austin, T. (1992). Weighingdecoupling vs. lost revenues: Regulatory considerations.The Electricity Journal, 5, 5863.

National Association of Regulatory Utility Commissioners.(2007).Decoupling for electric and gas utilities: Frequentlyasked questions. Washington, DC: National Association of

Regulatory Utility Commissioners.Tempchin, R. (1993). Decoupling? Not so fast. The Electricity

Journal, 6, 2.U.S. Environmental Protection Agency. (2007). National Action

Plan for Energy Efficiency. Aligning utility incentives with

investment in energy efficiency (p. ES-3). Washington, DC:U.S. Environmental Protection Agency.

Weston, F. (2008). Customer-sited resources and utilityprofits: Aligning incentives with public policy goals.US EPA Webinar presentation. Regulatory Assistance

Project.


factor that affect the succes regulatory to foster investment energi efficiency 2011

Documents