ORDES
AN ECONOMIC ANALYSIS OF THE
ELECTRIC UTILITY SECTOR
IN THE
OHIO RIVER BASIN REGION
PHASE
OHIO RIVER DASIN ENERGY STUDY
-------
November 1979
AN ECONOMIC ANALYSIS OF THE
ELECTRIC UTILITY SECTOR
IN THE
OHIO RIVER BASIN REGION
by
Patrick C. Mann
Tom S. Witt
Bureau of Business Research and
Department of Economics
West Virginia University
Morgantown, West Virginia 26506
Prepared for
Ohio River Basin Energy Study (ORBES)
Subcontract Number R805588
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
-------
ABSTRACT
This research project had two major tasks. The first task involved a
review of the literature on the demand for electricity. In the review at-
tention was placed on the theory of electricity demand and the specification
of various electricity demand models. The results of a number of national
and regional studies are summarized in the report. Based upon this review
two sets of estimates of various elasticities of electricity demand in the
six state region centered on the Ohio River Basin were presented. The first
set of estimates is based upon the "best" estimates of the various elas-
ticities from national studies while the second set of estimates is derived
from a recent Oak Ridge National Laboratory study. Both sets of estimates
incorporate salient characteristics of electricity demand in the six state
region.
The second task involved an examination of the implications of various
regulatory pricing policies on the electricity utility sector in the Ohio
River Basin Energy Study (ORB'ES) region. The current regulatory environment
affecting electric utilities and historical patterns in electricity demand
and price are reviewed in the report. Next, attention is directed toward
three alternative regulatory policies regarding electricity pricing in-
cluding traditional average cost pricing, marginal cost pricing and time
differentiated (peak load) pricing. The potential effects of these dif-
ferent pricing mechanism on capacity requirements, load factors and fuel
costs are examined with particular attention placed on their implications
for the ORBES region.
11
-------
CONTENTS
Abstract ii
Tables iv
Acknowledgements vi
1. Introduction 1
2. Estimates of the Elasticities of Demand for Electricity in the
Six States Region 2
Definition and Interpretation of Demand Elasticities 2
The Theory of Demand for Electricity 5
Review of Elasticity Estimates: National 12
Review of Elasticity Estimates: Regional 21
Price and Income/Output Elasticities of Demand for the Six
States Region 30
Summary 39
3. The Impact of Regulation on Electricity Prices and Generation
Capacity in the Ohio River Basin 41
The Regulatory Environment 42
The Pricing Alternatives 44
The Potential Effects of Time Differentiated Pricing 55
The Implications for the ORBES Region 60
References 71
111
-------
TABLES
Number Page
1. Price and Income Elasticities for Electricity Demand-
Residential Sector 13
2. Price and Income/Output Elasticities for Electricity Demand-
Commercial and Industrial Sector 16
3. .Residential Electricity Demand Elasticities by Region-
Oak Ridge Model 26
4. Commercial Electricity Demand Elasticities by Region-
Oak Ridge Model 27
5. Industrial Electricity Demand Elasticities by Region
Oak Ridge Model 28
6. Projected Annual Growth Rates of Total Average Electricity Cost
(TOC) for Alternative Cost Scenarios and Selected States-
Oak Ridge Model 31
7. Forecasts of Annual Growth Rates (1974-1990) of KWH Demand by
Sector, State and Cost Scenarios-Oak Ridge Model 32
8. Forecasts of Annual Growth Rates (1974-1990) of Electricity
Price by Sector, State and Cost Scenario-Oak Ridge Model 34
9. Estimates of Six State Region Price and Income/Output
Elasticities of KWH Demand by Customer Category-Method 1 36
10. Percent of Six State Region KWH Consumption by Customer
Category and Region, 1974 37
11. Estimates of Six State Region Price and Income/Output
Elasticities of KWH Demand by Customer Category-Method II 38
12. Forecasts of Annual Growth Rates (1974-1990) of Electricity
Price and Demand by Sector in the Six State Region 40
13. Actual and Real Average Annual Growth Rates in United States and
ORBES Region Electricity Prices, 1961-1975 47
IV
-------
TABLES
Number Page
14. Projections of Real Average Electricity Prices (by User)
for the ORBES Region, 1975-2000 48
15. Projections of Real Average Electricity Prices (by State)
for the ORBES Region, 1975-2000 49
16. Potential MW Capacity Reduction Effects from Time
Differentiated Rates—The Six State Region 61
17. Potential MW Capacity Reduction Effects from Time
Differentiated Rates—The ORBES Region. . 64
18. Potential MW Load Reduction Effects from Time Differentiated
Rates—The Six State Region 66
19. Potential MW Load Reduction Effects from Time Differentiated
Rates—The ORBES Region 68
-------
ACKNOWLEDGEMENTS
We are indebted to several people in preparing this report. Randy
Holliday, a research assistant, provided valuable assistance throughout
the project. Pat Curtis, Mary Ann Albertazzie, and Deanna Jefferson in
the Bureau of Business Research provided assistance at critical points
during the project. Finally, Claire Noel typed the final report as well
as the various drafts preceding it. Her diligence and cheerfulness made
our research task immensely easier.
VI
-------
SECTION 1
INTRODUCTION
There are two research tasks discussed in this report. The first
involved reviewing the literature on the demand for electricity; particular
attention was directed at developing estimates of elasticities of electricity
demand for the six states region centered on the Ohio River Basin. The
second involved examining the implications of increasing electricity prices
as well as the regulatory restructuring of prices; particular attention was
directed at viewing the potential effects of marginal cost and peak load
pricing for the electric utility sector in the ORBES region.
In Section 2 of this report, the theory of the demand for electricity
is reviewed with attention to the development of models of electricity demand
and the specification of variables included in such models. Attention is
placed on the measurement of the elasticity of demand which is the respon-
siveness of the quantity demanded to changes in either prices or incomes.
This is followed by a summary of the elasticities of electricity demand de-
rived from numerous studies. Attention is devoted to both national and
regional estimates of price and income elasticities for kwh demand. Finally,
two sets of estimates of the elasticities of kwh demand for the six states
region are presented. The investigator having primary responsibility for
the research in Section 2 is Tom S. Witt.
In Section 3 of this report, the present state of electricity regulation
is examined. There is a review of the historical relationship between elec-
tricity prices and the several factors theoretically associated with elec-
tricity prices. Several alternative pricing methods are examined including
traditional average cost pricing (where prices are equal to the average cost
of electricity production), marginal cost pricing (where prices are equal to
the increment in the cost of producing additional electricity), and peak
load pricing (where prices are based on the time of day when the electricity
is demanded as well as either average or marginal costs of production). The
potential effects of time differentiated pricing in terms of utility cost
savings are also evaluated. Finally, capacity and load reduction effects
from time differentiated electricity rates are estimated for both the six
states region and the ORBES region. The investigator having primary re-
sponsibility for the research in Section 3 is Patrick C. Mann.
The research discussed in Section 2 and 3 can be viewed as comple-
mentary. The elasticity estimates provide valuable information for the
further task of attempting to determine the growth and patterns of electric-
ity consumption over the next several decades. Similarly, the information
on regulatory reform and the empirical, results from previous peak load pric-
ing experience provide insight into future electricity consumption patterns.
-------
SECTION 2
ESTIMATES OF THE ELASTICITIES OF DEMAND FOR
ELECTRICITY IN THE SIX STATES REGION
This task of the research project involved reviewing the literature on
the demand for electricity and generating estimates of various demand elas-
ticities for a six state region centered around the Ohio River. This region,
composed of the entire states of Illinois, Indiana, Ohio, Pennsylvania,
Kentucky and West Virginia, is used as an approximation to the ORBES region
since data for the former was more readily available.
In this section of the report we first present the various definitions
and interpretations of demand elasticities utilized in the remainder of the
report. This is followed by a review of the theory of electricity demand.
In this review attention is not only placed on the economic theory of the
demand for electricity but also the development of electricity demand models
and the specification of variables included in these models.
The next part of this section summarizes the estimates of the elas-
ticities of electricity demand from a number of national studies. Based on
this review we present our "best" estimates of the national short- and
long-run price and income elasticities of electricity demand. Following
this is a review of regional studies of the demand for electricity. A major
portion of this review emphasizes a recent Oak Ridge National Laboratory
study which generates demand elasticities for each census region in the
United States.
The last part of this section presents two set of estimates of demand
elasticities for the six state region. The first set of estimates is based
upon the "best" estimates of national elasticities while the second set of
estimates is based on the census region estimates from the Oak Ridge study.
Both sets incorporate salient characteristics of electricity demand in the
six state region. In addition, some implications from the Oak Ridge study
for growth rates in electricity demand, costs and prices are also presented
for comparison with other ORBES studies.
DEFINITION AND INTERPRETATION OF DEMAND ELASTICITIES
The general theory of demand for a commodity states that the quantity
demanded of the commodity during a particular time period depends on the
price of the commodity, the income of the potential purchaser and the prices
of all other goods, among other factors. Although there are potentially a
large number of factors affecting the demand for a commodity, prices and
incomes are the most important determinants of demand and, thus, are of
-------
interest to economists. Over time many of the other factors affecting
demand, such as tastes and preferences, are relatively stable. In large
part this interest arises due to the need to understand the responsiveness
of demand to changes in prices and incomes in the market place. For example,
by specifying a particular price for the commodity, a particular income for
the potential purchasers and specific prices of other goods, an economist
could predict the quantity of good demanded.
Economists utilize various elasticity definitions to measure the
responsiveness of demand to changes in prices and income. The price
elasticity of demand, n..» measures the percentage change in the quantity
demand of good i with respect to a one percent change in the price of good
j, all other things unchanged.
The general formula is
when Aq. = change in the quantity demanded of good i
Ap. = change in the price of good j
q. = quantity demanded of good i
p. = price of good j
The own price elasticity refers to a situation where i and j refer to the
same good; on the other hand, the cross price elasticity refers to a
situation where i and j are different goods. For example, if i and j refer
to gasoline, then a price elasticity of gasoline demand equal to -.2 means
the quantity of gasoline demanded will fall twenty percent in response to a
one hundred percent increase in the price of gasoline, all other things
unchanged.1 If i refers to gasoline and j refers to diesel, then a cross
elasticity of gasoline demand with respect to diesel fuel price of +.2 means
the quantity of gasoline demanded will increase 20 percent in response to a
100 percent increase in the price of diesel fuel, all other things unchanged,
The following terms are associated with specific own price elasticity
of demand magnitudes: perfectly elastic, n..=-°°; elastic, -ro
-------
(increase) in the firm's electricity sales revenues. In general economic
theory leads to an expectation of own price elasticities of zero or less;
in actual empirical work an own price elasticity greater than zero is an
indication of either an incorrectly specified model or errors in the data
utilized in the estimation of the model.
Another elasticity measure of interest is the income elasticity 'of
demand for good i, n. , which measures the percentage change in the quantity
of good i demanded with respect to a one percent change in the income y of
the consuming unit, all other things unchanged.
The general formula is
n. =(Aqi/Ay)-(y/q.j,)
where Aq. = change in the quantity demanded of good i
A = change in income of the consuming unit
q. = quantity demanded of good i
y = income of the consuming unit
If the income elasticity of demand is less than zero, then the good is said
to be an inferior good; on the other hand, if the income elasticity of
demand is greater than zero, then the good is said to be a normal good.
Since all energy sources are normal goods, then their income elasticities
will be positive. If the income elasticity of demand for gasoline is +0.6,
then a one hundred percent increase in income would be associated with a
sixty percent increase in the quantity of gasoline demanded.
In measuring the price and income elasticities of demand one must dis-
tinguish between short-run and long-run elasticities for certain types of
goods and services. The degree of demand responsiveness to price or income
changes varies directly with the amount of time a consuming-unit has in order
to make adjustments in its purchased bundle of goods and services in response
to a change in price or income. For example, the price elasticity of demand
for gasoline is relatively inelastic (i.e. less responsive) in the short-run
since it is difficult to adjust the stock of automobiles in a short period
of time in response to a change in the price of gasoline. The only adjust-
ments available in the short-run in response to an increase in the price of
gasoline is to reduce the level of usage of automobiles. In the long-run
the price elasticity of demand for gasoline is more elastic since consuming
units not only can alter their level of usage of automobiles in response to
an increase in the price of gasoline but they can also scrap existing in-
efficient automobiles and purchase more efficient automobiles in response
to an increase in the price of gasoline. In general, the long-run price
and income elasticities of demand are more elastic (i.e. greater in absolute
vnlue) than the corresponding short-run elasticities since consuming units
have more flexibility and alternative goods available in the long-run.
-------
These concepts of price and income elasticities of demand are utilized
extensively in the remainder of this study in describing the measurement and
interpretation of estimates of the elasticity of demand for electricity.
THE THEORY OF DEMAND FOR ELECTRICITY
The demand for electricity is a derived demand which depends on the de-
mand for the services provided from an electricity utilizing capital stock
(such as lights, refrigerators, motors, etc.). The short-run demand for
electricity assumes the stock of electricity utilizing capital stock to be
fixed and focuses on the factors which cause changes in the utilization rate
of this capital stock. The long-run demand for electricity focuses on the
factors which cause changes in the demand for the electricity utilizing
capital stock. Since the short-run demand does not allow for substitution
among capital stocks utilizing different energy sources, it is more inelastic
with respect to price and income than the long-run demand.
It should be noted that, in addition to the effects of the capital stock
on the demand for electricity, one must also consider the role of the price
of electricity, the prices of substitute energy sources, and the income of
the purchasing unit, among other factors, in fully specifying the demand for
electricity. In the next section the problems involved in the measurment
of the appropriate quantity, prices and incomes are discussed while in
subsequent sections we outline some of the explicit models of electricity
demand which have been utilized in the literature. It should be noted that
there have been a number of excellent reviews of this literature in recent
years, and, as a consequence, we only review the major issues in this section
(Taylor, 1975; National Economic Research Associates, Inc., 1977; J. W.
Wilson & Associates, Inc., 1978; Edmonds, 1978).
Measurement of Electricity Quantities and Prices
One specification of the quantity of electricity demand utilizes the
number of kilowatt hours (kwh) over a given time period. Although this
specification is widely used, it can result in an aggregation error in the
estimated models if consumers perceive kwh purchased at different time of
day and seasons of the year as different commodities (Electric Utility Rate
Design Study, 1977). The only instance when such aggregation error is not
present occurs when the marginal prices of electricity at different points
of time change in the same proportion. To the extent that such changes
occur but are not proportional, the estimated price and income elasticities
of kwh demand will be biased.
Models which utilize measures of kwh demand are useful in forecasting
revenues when prices or income change; however, they cannot be utilized to
predict either demand changes due to rate structure alterations or the
optimum scale of electricity generating plant required at some future time
period (J. W. Wilson & Associates, Inc., 1977).
-------
Another specification of the quantity of electricity demanded utilizes
the number of kilowatts (kw) purchased at a given point in time. For exam-
ple, one could have two utilities with the same kwh demand during a given
period of time but with substantially different kw demand patterns. In this
case one would say that the two utilities have substantially different load
factors. Models of kw demand can be used to forecast the optimum scale of
electricity generating plant required during some future time period.
In many applications where the interest focuses on changing the rate
structures to alter the quantity of electricity demanded, one should
examine the kw demand by time of day or season of the year. Such exam-
inations not only allow one to evaluate the demand effects of different
rate structures which incorporate peak load pricing of one form or another
(in which the price of electricity is higher at high kw demand than at low
kw demand) but also allow forecasts of revenue effects of such rate struc-
tures and electricity generation capacity requirements (National Economic
Research Associates, Inc., 1977). One way of examining this is to dis-
tinguish between the peak and off-peak demand for electricity. Un-
fortunately, data on peak and off-peak demands by customer class is not
readily available except by conducting specialized surveys National
Economic Research Associates, Inc., 1977).
In studies of the demand for electricity considerable attention has
been directed toward the specification of the price variables to be utilized
in the models. In theory the appropriate price should be the marginal price
which is the additional amount of money a customer must spend in order to
consume an additional unit of electricity. In actual practice electricity
customers face a declining block schedule in which the marginal price de-
clines as the quantity consumed increases from one block or level of con-
sumption to another.2 The theoretical implications of the declining block
schedule on the specification of the price variable in the demand schedule
has been extensively examined in the literature (Taylor, 1975; Taylor,
Blattenberger, and Verleger, 1977). The conclusions of this literature is
that both marginal and average price measures of electricity should be in-
cluded in the demand function. In both cases the price variables should be
calculated from the actual tariff schedules of the utility and not from ex
post measures (as it is commonly done in the literature). It has been
argued:
The marginal price should refer to the last block consumed in,
while the average price should refer to the average price per kwh of
the electricity consumed up to, but not including the final block.
Alternatively, the total expenditure on electricity up to the final
^Residential bills typically have two components: a fixed charge which
is independent of the amount of electricity consumed and a kwh change which
is a declining block charge based on the level of usage. In general, non-
residential customers also have a kwh charge which is based on the kw of
installed capacity utilized at the time of maximum demand by customer.
-------
block can be used in place of the average price. Whichever quantity
is used, the variable will measure the income effect arising from
intramarginal price changes, thus leaving the price effect to be
measured by the marginal price. The omission of one of these two
prices will lead to specification errors and biased estimates of
the coefficients of other variables which are correlated with the
omitted variable (Taylor, 1975, p. 80).
Although this is the proper specification of the price variable, few
studies of the demand for electricity have utilized this specification.
Those studies which utilize this specification (Taylor, Blattenberger and
Verleger, 1977; Acton, Mitchell and Mowill, 1975) have confirmed the superi-
ority of this specification of price in the demand function. Alternatively,
many studies have utilized the average price as calculated ex post by divid-
ing total expenditures by quantity of electricity consumed (Fisher and
Kaysen, 1962; Baxter and Rees, 1968, Houthakker and Taylor, 1970; Mount,
Chapman and Tyrrell, 1973; Lymann, 1973; Griffin, 1974; Baughman and Joskow,
1974; Uri, 1975; Chern 1976; Chern, et al., 1978; and others). As Halvorsen
has shown, such a procedure leads to the introduction of simultaneity and
problems in the identification of the demand schedule (Halvorsen, 1978).
Given the demand schedule for electricity, the existence of the declining
block tariff implies that the customer faces a downward sloping supply
schedule since the average price of electricity declines as the quantity
of electricity supplied increases. Consequently, without prior restrictions
or additional information, one cannot disentangle the demand curve from the
supply curve.
One possible solution is to include in the supply and demand schedules
variables which are excluded from the other schedule. With this specifi-
cation some studies estimate the parameters in a simultaneous equation model
using average price and quantity of electricity as endogenous variables
(Wilder and Willenborg, 1975; Chern, et al., 1978; Halvorsen, 1978).
An additional justification for the use of average price is due to the
nature of the payment for electricity by customers. It can be argued that
customers are rarely aware of the true marginal price of electricity since
they pay at the end of some billing period and are never aware of the point
in time when they switch from one consumption block to another. In addition,
if one is interested in forecasting average price, the use of a marginal
price variable will require forecasting the entire rate schedule which is
more difficult than forecasting the average price.
One complication in this specification of the average price simultaneous
equation model is that the demand schedule is not a function but is simply
one point on the corresponding supply schedule. That is to say, a given rate
schedule only gives one equilibrium quantity and average price per customer.
A complete demand function in the usual sense, it is argued, requires a
family of rate schedules which rarely exists for a given utility (Taylor,
Blattenberger and Verleger, 1977).
-------
Another approach utilizing marginal prices involves calculation them
from records of the typical bills of customers for different levels of kwh
demand.^ Unfortunately, this procedure is invalid in that the kwh quan-
tities from the typical bills blocks are not the actual quantities consumed
by the customer and, thus, represent biased estimates of the true declining
block marginal prices of electricity. These prices also understate the
actual price to the customer since they exclude certain taxes collected by
utilities which vary among utilities (J. W. Wilson & Associates, Inc., 1977).
An additional complication is due to the inclusion of changes in customer
fixed charges and intramarginal rates in the calculated marginal price from
these bill (Taylor, Blattenberger and Verleger, 1977).
In conclusion, the theoretically superior price variable would include
both the marginal price and average price while in actual practice the use
of average price is justified within the context of a simultaneous equation
model. Elasticity estimates for these and similar prices are presented later
in this section.
Measurement of Income and Output
A less difficult problem in the specification of electricity demand
models has been the measurement of income. In the case of residential
customers most studies have utilized some measure of per capita or family
income which is readily available for the various observational units used
in the analysis. In the case of commercial customers, researchers have
generally used the same per capita or family income measures used in the
corresponding residential study. It should be noted that the linkage
between per capita or family income and the demand for electricity by
commercial customers is much more indirect than residential customers. An
alternative in the case of commercial customers would be the use of a value
added measure similar to that utilized for the industrial customers; however,
this has not been utilized for commercial customers due to inadequate data.
In the case of industrial users, the studies reviewed utilize some measure
of value added so that instead of an income elasticity of demand such studies
generated an output elasticity of demand. In conclusion, there are fewer
conceptual and empirical problems associated with the specification of the
income or output component in contrast to the price variable.
Empirical Models of Electricity Demand
The discussion thus far has focused on the specification of the major
factors examined in a variety of studies on the demand for electricity. In
these studies researchers have utilized four types of models of the demand
for electricity.
Statistics on the typical electric bills of different customer classes
for different demand levels are published yearly by the Federal Power
Commission.
-------
Underlying an autoregressive model that relates the dependent variable
in time t to its value in time t-1 and a set of independent variables is a
distributed lag model which relates the dependent variable at time t to the
independent variables at time t and all preceeding time periods. In this
distributed lag model which was popularized by Koyck (Maddala, 1977) the
coefficients of the independent variables decline geometrically with the
passage of time.
This type of autogressive model when applied to the estimation of the
demand for electricity has the desirable property of generating both short-
and long-run price and income elasticities of demand. The limitations of
this specific model have been summarized by Edmonds:
This type of model, and all of the lagged adjustment models
suffer one serious weakness. None is explicitly derived from a
theory of why adjustment costs exist in the first place. It has
also been criticized, however, because Koyck lag structures require
that the large response to any price change occurs in the first
period, while each succeeding period has smaller responses (Edmonds,
1978, pp- 65-66).
'In an attempt to circumvent the problems of geometrically declining co-
efficients in the Koyck distributed lag model, some studies have utilized
the polynomial lag structure (Griffin, 1974) which allows for a more flexible
albeit finite lag structure (in contrast to the Koyck infinite lag struc-
ture) .
A fourth type of model utilizes a production function as a basic for
examining the adjustment process for the factor inputs to the production
process in response to changes in prices of all inputs, among other variables
(Edmonds, 1978). This model not only makes explicit the interrelationships
among all inputs used in the production process but it also allows one to
examine the short- and long-run adjustment processes. Although this, method
is potentially promising it has only been used one (Halvorsen, 1976).
The models which we have briefly surveyed above represent some of the
major types of approaches used in modeling the demand for electricity. As
will be seen later in this report, some of the differences in the electricity
elasticity estimates encountered in the literature are attributable to the
differences in the specification of the models. In the interpretation of
the elasticities presented below one should remember that each elasticity
estimate assumes a different type of model and behavioral characteristics of
customers.
Other Considerations in the Estimation of the Demand For Electricity
In estimating the demand for el ec t r i c i. ty, one lias ,-i wealth of infor-
mation Available due to federal and slate electric utility report in>',
requirements; however, lliere are some other problems in the modeling of elec-
tricity demand. For example, our prior discussion above indicated the
t lieorel i c;il necessity of including marginal price in Ihe demand function;
in /ic Lua I i ty , such price measures are not available in the published
.10
-------
statistics and one must construct estimates. Problems also occur in the
(lt>f inition of other explanatory variables such as income and electricity
utilizing capital stock. In addition, the existence of: a substantial amount
of claLa means that the Individual investigators have to decide whether to
utilize cities, states, nations, households, firms, or utilities as units of
observations. The usage of more aggregative data raises the possibility of
introducing aggregation error which could lead to biased estimates of the
elasticities of interest (J. W. Wilson & Associates, Inc., 1977). In the
case of state data, additional problems occur in that a number of states are
served by more than one electric utility so there exists several different
rate structures and prices confronting customers in each of these states.
As will be seen later, a considerable amount of variation in the electricity
estimates is associated with the different levels of aggregation utilized
in the model estimation.
Another major question prior to estimation involves whether one should
utilize time-series or cross-section data. If one is interested in obtain-
ing estimates of the short-run elasticities of demand, it has been argued
that time-series data is more appropriate since it allows for the dynamic
adjustment process through time. In contrast, cross-section data has been
proposed for the estimation of long-run elasticities since it reflects in-
dividual observation units in economic equilibrim. In the latter case
differences among individual observation units reflect differences in long-
run equilibrium positions. To'the extent that one has both cross-section
and time-series data, one can, it has been argued, estimate both short-
and l.ong-run elasticities through the use of various pooled econometric
models. One should note, however, that there can be difficulties in the use
of one type of data, cross-section or time-series, exclusively. For example,
Taylor has argued that
...while the view that cross-sectional observations reflect steady
state variation has some limited validity as a general tendency, it
is not correct to say that cross-section data never reflect short-
term, dynamic adjustments. For the latter will be represented to
the extent that individual observation units (states, SMSA's etc.)
are not all at the same point of disequilibrium arising from recent
changes in income, prices, or other relevant factors. Since income
and prices, in general, do not change at the same time across cities,
states, and regions, differential disequilibria are almost certain
to be reflected in the data. If these differential disequilibria
are not allowed for explicitly-say, through the inclusion of appli-
ance stocks or last period's consumption as a predictor-then the
elasticities obtained will, in general, be downward biased estimates
of long-run elasticities (Taylor, 1975, p. 103).
It also should be noted that a variety of functional forms can -be used
in the estimation of the models. One of the more popular functional forms
is the double-log specification in which the quantity, price and income
variables are expressed in terms of logarithms. In this model the estimated
coefficients of price and income are the elasticities of interest. In this
11
-------
functional specification the elasticities are asumed constant over all
ranges of prices and income; for example, a given percentage change in price
wil.l have the same percentage impact on the quantity demanded i r regard 1 ess
of the level of price. In contrast the linear specification of a model leads
to elasticity estimates which vary with the price and income .levels.
Finally, a variety of econometric methods have been used in the esti-
mation of the demand models due to the variety of stochastic error terms,
number of equations, existence of distributed lags and lagged endogenous
variables, among others. Part of the variation in elasticity estimates can
also be 'attributable to the choice of econometric method and computer
program.
REVIEW OF ELASTICITY ESTIMATES: NATIONAL
Our discussion thus far indicates there are several approaches and
problems in the estimation of the demand for electricity; however, the
existence of several studies in this area suggests that many of these
problems are surmountable. In this part we review the estimates of the
price and income elasticities of kwh demand which have been generated in
national studies. By national studies we simply mean studies which are not
specific to a particular geographical area or utility service area. Such
studies generally utilize either time series data on the United States as
a whole or cross section data on states or SMSAs. The elasticity estimates
are generally measured at the mean and thus refer either to the national
average over a period of time or the average observational unit in the sample
at a particular point in time. After reviewing these estimates for resi-
dential, commercial and industrial categories, we summarize the conclusions
which have been drawn from these studies by other reviewers. Finally, we
present our judgement as the "best" national estimates of kwh demand elas-
ticities recognizing that these estimates are implicitly accompanied by a
variance reflecting uncertainty. It should be noted that no estimates of
kw demand elasticities are considered in this section.
Numerous studies have been made of the kwh demand in the residential,
commercial, and industrial sectors and many of these are summarized in
Tables 1 and 2.1* As mentioned earlier there is a considerable amount of
variation in the short- and long-run price and income elasticities esti-
mates due to the differences in models, data, variables, and estimation
methods, among other considerations. An examination of these estimates
'It should be noted that the studies summarized in Tables 1 and 2 ex-
clude some widely cited studies which .are cither of the United Kingdom
(HouthnkkcT, 1951; Baxter and Rees, 1968) or arc of specific regions (Nelson,
1965; Levy, 1973; Lacy and Street, 1975; AcLon, Mitchell and Mowill 1975;
Chern, et al., 1978).
12
-------
TABLE 1
PRICE AND INCOME ELASTICITIES FOR ELECTRICITY DEMAND-RESIDENTIAL SECTOR
Study
Fisher-Kaysen (1962)
Houthakker -Taylor
(1970)
Wilson (1971)
<-> Halvorsen (1972)
Anderson (1972)
Halvorsen (1973)
Mount -Chapman-
Tyrrell (1973)
Andersen (1973)
Lyman (1973)
Houthakker-Verleger-
Sheehan (1974)
Griffin (1974)
Data
Type3
CS-TS:
States
TS:USA
CS:
SMSA's
CS:
States
CS:USA
CS : USA
CS-TS:
USA
CS-TS:
States
CS:
States
CS-TS:
Area
Served by
Util ities
CS-TS:
States
TS:USA
b
Price Elasticity
Vintage Short-Run Long-Run
1947-57 0.00
1947-64 -0.13 -1.89
1960, '66 -2.00
1961 -1.16
1969 -0.84^-0.90
1969 -0.77
1947-69 -0.26 -2.11
1946-70 -0.14 -1.20
1960-70 -1.12
-0.90
1959, -0.90 -1.02
1965
1970
1951-70 -0.06 -0.52
(Continued)
h
Income Elasticity
Type0
Short-Run Long-Run of Price
0.10 Small A
0.13 1.94 A
0.00 A*
M*
A*
M*
0.02 0.20 A
0.80 A*
-0.20 A
0.14 1.64 M
0.06 0.88 A
-------
TABLE 1 (Continued)
Study
Baughman-Joskow (1974)
Wilder-Willenborg
(1975)
Uri (1975)
Halvorsen (1975)
Taylor-
Blattenberger-
Verleger (1976)
Chern (1976)
FEA (1976)
Halvorsen (1976)
Halvorsen (1978)
Data
*q
Type Vintage
CS: 1969
States
CSrlndi- 1973
viduals
TS:
Monthly
USA
CS-TS 1960-70
States
TS: 1956-72
States
CS-TS: 1971-'72
States
CS-TS: 1960-72
Census
Regions
CS: 1969
States
CS-TS: 1961-69
States
Price Elasticity Income Elasticity
c
Type
Short-Run Long-Run Short-Run Long-Run of Price
-0.53^-2. 08d A
-1.00 0.16 M*
-0.61 -1.66 0.44 0.12 A
-1.15 -1.52 0.51 1.52 M*
-0.07 -0.78 0.10 1.18 M
-1.446 0.82e A
-0.19 -0.30 1.10 A
-0.97 0.71 M*
-.58 -1.15 Negative 0.51 M*
(Continued)
-------
TABLE 1 (Continued)
a - TS refers to time-series data; CS to cross-sectional data; and CS-TS to pooled CS and TS data.
b - Elasticities listed between short-run and long-run columns are ambiguously defined in the
reference cited.
c - M refers to marginal price; M* to a theoretical model in which both average and marginal
price elasticities are identical (price data was, however, either A or A*); A to an average
price for electricity; and A* to an average price for a fixed amount of electricity.
d - These are "saturation" electricities and in general should be smaller than true price elasticities.
e - Combined residential and commercial sectors.
Sources: Individual Studies; Table 2 in Edmonds (1978); Electric Utility Rate Design Study (1977);
Halvorsen (1978).
-------
TABLE 2
PRICE AND INCOME/OUTPUT ELASTICITIES FOR ELECTRICITY DEMAND-COMMERCIAL AND INDUSTRAL SECTORS
Study
COMMERCIAL SECTOR
Mount-Chapman-
Tyrrell (1973)
Lyman (1973)
__i
JN
Hudson-Jorgenson
(1974
Uri (1975)
Tyrrell-Chern
(1975)
FEA (1976)
Halvorsen (1976)
Data
o
Type Vintage
CS:48 1946-70
States
CS-TS: 1959-68
Area
Served
by
Utili-
ties
TS:USA 1947-71
TS: Month-
ly aggre-
gate USA
CS:States
CS-TS: 1960-72
Census
regions
yearly
CS:
States
1969.
Price Elasticity Income/Output Elasticity
Short-Run Long-Run Short-Run Long-Run of Price
-0.17 -1.36 0.11 0.86 A
-2.10 .A
-0.36 1.00d A
-0.34 -0.85 0.79 1.98 A
-1.23 A
-0.24 -0.38 0.73 1.63 A
-0.92 1.25 M*
(Continued)
-------
TABLE 2 (Continued)
Data Price Elasticity Income/Output Elasticity
Study
INDUSTRIAL SECTOR
Fisher-Kaysen (1962)
Anderson (1971)
Mount-Chapman-
Tyrrell (1973)
Lyman (1973)
Griffin (1974)
Hudson-Jorgenson
(1974)
Uri (1975)
Type Vintage Short-Run Long-Run Short-run Long-Run
CS: 1946-57 -1.25
States
CS: 1958, '62 -1.94
States
CS-TS: 1947-70 -0.22 -1.82
States
CS-TS: 1959-68 -1.40
areas
served
by utili-
ties
TS: 1951-71 -0.04 -0.516
Aggre-
gate U.S. .
TS:USA 1947-71 -0.07 1.00d
TS: -0.35 -0.69 1.32 2.63
Monthly
Aggre-
gate US
Type0
of Price
A
A
A
A
A
M
A
(Continued)
-------
TABLE 2 (Continued)
Data
Price Elasticity Income/Output Elasticity
Study
Type Vintage Short-Run.
Type
Long-Run Short-Run Long-Run of Price
Baughman-Zerhoot
(1975)
Chern (1975)
i—>
oo
Tyrrell-Chern
FEA (1976)
Halvorsen (1976)
CS-TS:
48
States
and
Wash.
B.C.
CS-TS:
16 US
indus-
tries
CS:
States
CS-TS:
Census
regions
annual
CS:
States
CS:
States
1962-72 -0.11
1959-71 -0.61
1960-72 -0.15
1969
1971
-1..28
-1.28
-1.03
-1.24
-0.92
0.69
-1.98 0.30
1.00
0.97
iioo"
0.68
1.00^
A
A
M*
A
(Continued)
-------
TABLE 2 (Continued)
a — TS refers to time-series data; CS to cross-sectional data; and CS and TS to pooled Cs and TS data.
b - Elasticities listed between short-run and long-run columns are ambiguously defined in the
reference cited.
c. - M refers to marginal price; M* to a theoretical model in which both average and marginal price
M elasticities are identical (price data was, however, either A or A*); A to an average price for
electricity; and A* to an average price for a fixed amount of electricity.
Sources: Individual Studies; Table 2 in Edmonds (1978); Electric Utility Rate Design Study (1977).
-------
leads to the general conclusion that the short-run price and income elas-
ticities of kwh demand are more inelastic, on average, than the corres-
ponding long-run elasticities. In addition, the kwh demand response to a
change in the marginal price appears in general to be less than the kwh
demand response to a change in the average price.
In an extensive review of some of these studies Edmonds concluded:
These studies indicate a larger long-run price elasticity than short-
run elasticity, and in fact suggest that short-run elasticities lie
somewhere between zero and -.25. Long-run elasticities are larger,
however, and there seems rather unanimous agreement that the long-run
average price elasticity of demand for residential electricity has an
absolute value larger than 1. A glance at the results for the other
sectors shows that this conclusion holds for the commercial and in-
dustrial sectors as well. Income elasticities also seem to follow
this general trend: inelastic in the short run while elastic in the
long run (Edmonds, 1978, pp. 10-11).
A similar review by a task force of the Electric Utility Rate Design Study
concluded that the long-run price elasticity for each class was -1.3
(Electric Utility Rate Design Study, 1977).
National Economic Research Associates, Inc., as part of their tasks
for the Electric Utility Rate Design Study, intensively reviewed several
of the studies reported in Tables 1 and 2 and concluded:
First, absent (or holding constant) interfuel substitution effects,
direct price elasticities of the demand for electricity are all
roughly -0.5. Second, price elasticities pertaining to choice of
electricity as against fossil fuels in a specific application are
generally larger than this (in absolute terms). Thus, combining
these two effects, measured total price elasticity appears to be
-1.0 or slightly higher....
A brief look at short-run elasticities indicates the following
kinds of estimates. For the residential class, electricity price
elasticity is generally around -.2. For the commercial class, it
may be slightly higher. But is also seems to be around -.2 for the
industrial class (National Economic Research Associates, Inc.,
1977, p. xii).
These conclusions and our review of the literature lead us to the
following "best" estimates of the various United States elasticities of kwh
demand. The following long-run average price elasticities of kwh demand
allow for interfuel substitution: residential, -1.0; commercial, -1.0; and
industrail, -1.1. Industrial kwh demand is estimated as more elastic due to
greater possibilities for self-generation of electricity for this customer
category. The range of uncertainty regarding these estimates is larger for
commercial and industrial customers since there have been fewer studies of
these categories in comparison to residential customers.
20
-------
The estimate of the long-run marginal price elasticity of kwh demand
with allowance for interfuel substitution is -0.8 in the case of residential
customers. Although there are a few studies of industrial and commercial
customers using marginal price, we do not provide estimates of their corre-
sponding marginal price elasticities of kwh demand due to the very large
variance in reported results.
The following short-run average price elasticities of kwh demand do not
allow for interfuel substitution effects: residential, -0.2; commercial,
-0.25; and industrial, -0.2.
Although estimates of the income/output elasticity of kwh demand are
presented in Tables 1 and 2, most reviewers do not establish a "best" esti-
mate of income elasticities of kwh demand. Our review leads to an estimates
of the average long-run residential income elasticity of kwh demand of +1.0
assuming the use of average price and allowing for interfuel substitution.
The corresponding average short-run residential income elasticity of kwh
demand is estimated as +0.1 assuming the use of average price and no inter-
fuel substitution. The long-run commercial output elasticity of kwh demand
is estimated as +1.6 assuming average price and interfuel substitution while
the corresponding short-run output elasticity of kwh demand is estimated as
+0.75 assuming the use of average price and no interfuel substitution. Due
to relatively few studies and diversity of results, we cannot provide any
"best" estimate of the output elasticities of kwh demand for the industrial
sector at this time; however, we note that several studies assume this
elasticity is +1.0 in both the short- and long-run.
REVIEW OF ELASTICITY ESTIMATES: REGIONAL
Several studies have developed estimates of the price and income/output
elasticities of kwh demand for geographical areas such as states, SMSAs,
cities and utility service areas (Nelson, 1965; Anderson, 1972; Mount,
Chapman and Tyrrell, 1973; Acton, Mitchell and Mowill, 1975; Lacy and Street
1975; Chern, et al., 1978). A survey of these studies shows the same
diversity in models, data, and other characteristics as was found in the
previous survey of national studies. No extensive discussion of these
studies is presented in this section since they are specific to a particular
region; in addition, there is no table reporting all of the elasticity esti-
mates encountered in this review. The sizable number and diversity of re-
gional studies has led several reviewers to pessimistic conclusions regarding
our knowledge of regional differences in electricity demand elasticities.
For example, Edmonds has concluded:
Having looked at these papers, which have sought to illuminate
the area of regional... energy-price responsiveness, we must ask, How
much do we really know about regional elasticity differences...? In
the area of regional elasticities, the answer would seem to be, very
little...The numerous regional papers do tell us some things; however,
21
-------
because they differ with respect to their data, theoretical specifi-
cations, and econometric techniques, it is impossible to distinguish
how much elasticity differences are due to differences in regional
responses and how much to econometric differences (Edmonds, 1978,
pp. 60-61).
We, however, are not as pessismistic as Edmonds in that there exists a
recently completed regional model developed at Oak Ridge National Laboratory
(Chern, et al., 1978) which was not available to Edmonds at the time of his
study. In our opinion this particular regional model surmounts many of the
difficulties encountered in prior modeling efforts at the regional level.
Although the principal usage of the Oak Ridge model is to forecast regional
demand for electricity to the year 1990, it in addition generates short-
and long-run estimates of the price and income elasticities of kwh demand
by major census regions which can be used to estimate the corresponding six
state region elasticities. We first survey the salient features of the Oak
Ridge model and its relationship to some of the other models discussed
earlier in this report. We then present the estimated price and income elas-
ticities of kwh demand by census region and the forecast growth rates of kwh
demand and average electricity price to the year 1990 by customer category
and state. The elasticity estimates are utilized in a later part of this
section to generate corresponding estimates for the six state region while
the growth rates provide a basis for comparison with other ORBES studies.
The Oak Ridge model is basically a simultaneous equation model which
has submodels for the residential, commercial and industrial sectors. Within
each submodel are demand and price equations which imply that kwh demand and
average price per kwh are endogenous to the system. This specification
eliminates the identification and estimation problems associated with the
use of average price within a declining-block rate structure. Thus, this
model is seen as a variant of the Halvorsen type approach to the specifi-
cation of the electricity demand equation.
All of the demand equations in the three sectors have the same general
dynamic structure. These equations are autoregressive with the lagged
endogenous variable as a predetermined variable in each demand equation.
As was noted earlier this specification is derived from a state adjustment
model (Houthakker and Taylor, .1970) and has the advantage that no explicit
capital stock variable needs to be used in the estimation of the model.
In addition, this specification allows for the calculation of both short-
and long-run price and income elasticities.
The residential demand equation was specified as
In ERS. =a + a In ERS..^ + a ln(PER/CLI). + a InX. + a D. + a A.
+ u it
where i = state
t = year
22
-------
ERS = residential sales of electricity measured in kwh
PER/CLI = average price of electricity in the residential sector
deflated by the cost of living index
X = set of explanatory variables
D = set of state and shift dummy variables
A = set of dummy variables for reclassif ication of customers and other
shifts in historical trends of residential sales
The set of explanatory variables used in the estimation of the residential
demand equation includes the average natural gas price in the residential
sector (deflated by the cost of living index) , the average retail price of
No. 2 fuel oil (deflated by the cost of living index), the number of resi-
dential customers, per capita personal income (deflated by the cost of liv-
ing index), population, heating degree-days, cooling degree-days, and the
number of natural gas customers in the residential sector. The set of
dummy variables D includes state dummies as well as a dummy variable to
investigate possible structural shifts between the periods of declining
real electricity price and increasing real electricity prices. In addition
it includes variables, to measure the effects of natural gas availability
on the quantity of electricity demanded. The set of dummy variables A are
included to account for distortions in historical trends in electricity
sales due to the reclassif ication of customers from one category to another.
Detailed discussion of the variables' construction can be found in the
original study.
The residential electricity price equation was specified as
PER.t-TOC.t=B0 + ^(ERS^/CR^) + B^ERS^/CR./ 4- g^ + g^ + f^
+ V it
where TOC = average total cost of generating and distributing electricity
period.
The other variables are defined in the residential demand equation.
The commercial demand equation was specified as
lnECS.t=Y0 + Y^nECS.^ + Y2ln(PEC/CLI)lt + Y
where ECS = kwh of commercial sales of electricity
PER/CLI = average price of electricity in the commercial sector
deflated by the cost of living index
M = set of explanatory variables
23
-------
D = set of state and shift dummy variables
B = set of dummy variables for reclassif ication of customers
The set of explanatory variables includes population, real per capita
personal income, real fuel oil price, heating degree-days, and cooling
degree-days as previous defined in the residential demand equation.
Furthermore, the equation includes as explanatory variables the average
natural gas price in the commercial sector (deflated by the cost of living
index) and the number of natural gas customers in the commercial sector.
The set of dummy variables B and D are as previously defined. except that
B includes some additional dummies to account for the reclassif ication
of commercial and industrial customers over the years.
The commercial electricity price equation was specified as
PEC.t-TOC.t=60 + 6l(ECSit/CC.t) + S^ECS^/CC^) + 6^ + 6^ + 6
+ v .
it
where CC = the number of commercial electricity customers period.
The other variables are as previously defined.
The industrial demand equation was specified as
lnEIS.t=eo + e^nEIS.^ + 62ln(PEI/WPI)lt + 93lnN.t + 6^ + 6^ + u.
where EIS = quantity of industrial sales of electricity (kwh)
PEI/WPI = average price of electricity in the industrial sector
deflated by the wholesale price index
The other variables are as previously defined. The set of explanatory
variables used in the estimation of the industrial sector includes value
added in manufacturing (deflated by the wholesale price index of manu-
facturing), average natural gas price in the industrial sector (deflated
by the WPI), wholesale price of No. 6 fuel oil (deflated by the WPI),
average price of coal (deflated by the WPI, and the number of natural gas
customers in the industrial sector.
The industrial electricity price equation was specified as
where DI = number of industrial electricity customers.
The other variables have been defined previously.
24
-------
The six structural equations have the price and quantity variables as
endogenous variables whereas the remaining variables are predetermined. Al-
though the system of equations is nonlinear, the model is estimated using two
stage least squares and three stage least squares after treating the non-
linear variables as new variables. The estimation assumes no serially
correlated errors. After the equations were estimated, a nonrigorous exami-
nation of the residuals from the demand equations indicated no apparent
serial correlation problems in these equations. The final equations present-
ed in the report (which are not presented here) have some of the predetermin-
ed variables excluded from individual equations due to incorrect signs of the
estimated coefficients. As the report indicates, this can lead to
misspecification of the model if the excluded variables actually belong in
the individual equations. Since the three stage least square coefficient
estimates are the most sensitive to misspecification of the individual
equations, we believe attention should have focused on their two stage least
squares coefficients estimates; however, we note that the authors utilize
the three stage least squares coefficient estimates in estimating elas-
ticities and deriving forecasts of the growth rate in demand and prices to
the year 1990. The authors justify this procedure on the basis that the
three stage least squares coefficient estimates are similar in magnitude to
the two stage least squares coefficient estimates and the former are more
efficient than the latter.
Tables 3-5 present the regional estimates of the short- and long-run
price and income elasticities of kwh demand by customer category which were
calculated in the Oak Ridge study.5 It should be noted that, in the case
of the industrial category, an industrial output elasticity of kwh demand
was calculated instead of an income elasticity for reasons discussed earlier.
The following are general conclusions from the examination of these re-
sults. First, there is a considerable amount of variation among regions in
the estimates of the elasticities. In part this may be attributable to
differences in end use consumption, in types of appliance stocks, climatic
conditions and industrial composition not reflected in the model. Second,
the short- and long-run estimates are comparable to national estimates with
few exceptions; the major exception are the lower short- and long-run income
elasticities in the East. North Central region compared to elsewhere. Third,
the commercial and residential customers have more elastic average price
effects than the industrial customers, particularly in the long-run esti-
mates. Finally, the study reinforces national studies which find the average
price of electricity has a very important negative effect on the quantity of
electricity demanded.
The Oak Ridge simultaneous equations model is then used to forecast
the kwh sales and average electricity price growth rates to 1990 for
consumer sector and state. In making these forecasts the study required
5These elasticities are calculated from the estimated coefficients of
the various electricity demand equations. For details of this procedure
see Chern, et al., 1978. Since the estimated models are double log models
the elasticities are assumed to be constant over all observable ranges of
quantity, price and income.
25
-------
TABLE 3
RESIDENTIAL ELECTRICITY DEMAND ELASTICITIES BY REGION
OAK RIDGE MODEL
Region'
Electricity Price
Short-run Long-run
Income
Short-run Long-run
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
-0.33
-0.22
-0.35
-0.27
-0.31
-0.47
-0.57
-0.19
-0.08
-1.50
-0.60
-1.22
-0.73
-1.12
-0.95
-1.07
-0.43
-0.37
0.07
0.34
0.06
-0.01
0.21
0.30
0.27
0.45
0.01
0.32
0.91
0.19
-0.02
0.77
0.61
0.51
1.03
0.04
The states which make up the regions are as follows:
New England:
Middle Atlantic:
East North Central;
West North Central:
South Atlantic:
East South Central:
West South Central:
Mountain:
Pacific:
Maine, New Hampshire, Vermont, Massachusetts,
Rhode Island, Connecticut
New York, New Jersey, Pennsylvania
Ohio, Indiana, Illinois, Michigan, Wisconsin
Minnesota, Iowa, Missouri, North Dakota,
South Dakota, Nebraska, Kansas
Delaware, Maryland, Virginia, West Virginia,
North Carolina, South Carolina, Georgia, Florida
Kentucky, Tennessee, Alabama, Mississippi
Arkansas, Louisiana, Oklahoma, Texas
Montana, Idaho, Wyoming, Colorado, New Mexico
Arizona, Utah, Nevada
Washington, Oregon, California
Source: W. S. Chern, R. E. Just, B. D. Holcomb and H. D. Nguyen, Regional
Econometric Model for Forecasting Electricity Demand by Sector
and by State, Oak Ridge, Tennessee: Oak Ridge National Laboratory,
(NUREG/CR-0250), October 1978, Table 5.1.
26
-------
TABLE 4
COMMERCIAL ELECTRICITY DEMAND ELASTICITIES BY REGION
OAK RIDGE MODEL
Region
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
Electricity
Short-run
-0.47
-0.33
-0.43
-0.09
-0.39
-0.66
-0.25
-0.48
-0.40
Price
Long-run
-1.31
-0.51
-1.60
-1.02
-1.27
-1.29
-1.60
-0.90
-0.66
Income
Short-run
0.25
1.22
0.20
NE3
0.33
0.33
0.03
NE
0.31
Long-run
0.70
1.88
0.76
NE
1.09
0.65
0.20
NE
0.52
NE-Not estimated
Source: Ibid, Table 5.2.
27
-------
TABLE 5
INDUSTRIAL ELECTRICITY DEMAND ELASTICITIES BY REGION
OAK RIDGE MODEL
Region
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
Electricity
Short-run
-0.06
-0.02
-0.32
-0.26
-0.15
-0.28
-0.10
-0.19
-0.03
Price
Long-run
-0.16
-0.04
-0.54
-0.87
-0.71
-0.55
-0.62
-0.39
-0.09
Industrial
Short-run
0.50
1.01
0.74
0.25
0.21
0.48
0.17
0.38
0.32
Output
Long-run
1.41
1.55
1.28
0.83
1.03
0.96
1.03
0.80
• 0.90
Source: Ibid, Table 5.3.
28
-------
forecasts of all of the predetermined variables over the same time period.
The assumptions used in projecting values for all of the predetermined
variables except electricity generation and distribution costs are presented
in detail in the study. Basically they utilized generally accepted forecasts
of population, real per capita income, value added, number of customers,
heating degree-days and cooling degree days and general price level changes
which have been prepared and utilized by both governmental agencies as well
as private consulting firms.
Three cases of assumptions were made in the Oak Ridge report regarding
the electricity generation and distribution costs to 1990. Because of the
importance of these in the forecasts the following are details of their
procedure:
For the base case we took the Hudson and Jorgenson projections
of the price in current dollars of natural gas, refined petroleum
products, and coal. The fuel prices in real terms are, of course,
easily obtainable using the projected cost-of-living index and
wholesale price index. To derive the estimate for the overall
average of the costs of generation, transmission, and distribution
is more complicated. Basically, we decomposed the overall average
costs to two components: fuel costs and other costs (operation,
maintenance, capital, taxes, etc.). The cost of fuels depends on .
the shares of various fuels used by electric utilities. This com-
position of fuels varies from the state to state. We took the
1974 data and derived the exact relationships between composite
fuel costs and prices of fuels used by utilities for each state.
The cost of fuels are then projected based on assumed prices of
natural gas, petroleum products, and coal. For the operating and
maintenance cost component, we assume that it will increase
slightly more than the increases in the wholesale price index
(6.1% for 1974-1980, 4.4% for 1980-1985, and 3.7% for 1985-1990).
The projected total electricity costs are the weighted average of
projected fuel costs and operating and maintenance costs. The
percentage of these two cost components in 1974 were used as
weighting factors.
In the low-price case, we assume that all fuel prices in the
residential and commercial sectors will increase at the same rate
as the cost-of-living index. All prices of fuels in the industrial
sector will increase at the same rate as the wholesale price index.
Furthermore, the costs of fuel and operating and maintenance will
increase at the same rate as the wholesale price index. In other
words, it is assumed in this case that the real prices of fuels and
the real costs of electricity generation, transmission, and dis-
tribution will remain at the 1974 level.
In the high-price case, we assume that the growth rates of all
price and cost components in the bnsc case will be doubled in real
terms (Chern, et a.l . , 1978, pp. 7-4 through 7-5).
29
-------
The projected annual growth rates of total electricity generation and
distribution cost under these three cost cases from the Oak Ridge report
are presented in Table 6. The associated forecast by Oak Ridge of annual
growth rates in kwh demand by sector and state under these cost cases
and the forecast values of the other predetermined variables are presented
in Table 7.6 An examination of kwh demand growth rates shows some variation
amo:ig states during the forecast period; however, there is less variation
for the ORBES states than in the remainder of the states. For the states
listed in Table 7 there appears to be no definite pattern in the relative
growth rates by customer category, i.e., no one category has consistently
higher or lower growth rates compared to another category. These growth
rates are utilized to estimate the growth rates for the six state region
later in this section.
The Oak Ridge report also generated forecasts of the average electricity
prices in nominal terms by customer category and state which is reported in
Table 8. The forecast growth rates in average prices were consistent with
the scenarios regarding the growth rates in total average electricity cost.
It should be noted that the Oak Ridge model forecasts different growth rates
in average electricity prices by sector among the states in the ORBES region,
PRICE AND INCOME/OUTPUT ELASTICITIES OF DEMAND FOR THE SIX STATE REGION
In this section we provide two methods of estimating the short- and
long-run and income/output elasticities of kwh demand in the six states
region. The first method will utilize essentially the same estimates of
these elasticities as were found in the review of the national studies. The
second method will generate estimates of these elasticities from the Oak
Ridge report. It should be noted that both methods only provide estimates
of the ORBES region elasticities. If one had access to appropriate data
for the ORBES region one could develop estimates from an estimated
econometric model; however, due to data, time, and budget limitations no
such model was estimated.
The first set of estimates for the six state region are based on the
"best" estimates of the short- and long-run price and income/output elas-
ticities of kwh demand for the United States from studies completed during
the period 1968-1975. The six state estimates are assumed to be equivalent
to the United States estimates in each customer cateogry. In making this
assumption one need not assume that the characteristics of the electricity
customers and their tastes and preferences are the same between the six
state region and the rest of the United States. In estimating the price
and income/output elasticities reported in Table 1 and 2, the studies also
examined simultaneously other predetermined variables including the effects
of climate, appliance stocks, prices or other energy sources, trends, and
6Two sets of industrial growth rates were calculated to allow in one
case for projected growth in the demand for electricity by Department of
Energy uranium enrichment plants.
30
-------
TABLE 6
PROJECTED ANNUAL GROWTH RATES OF TOTAL AVERAGE
ELECTRICITY COST (TOC) FOR ALTERNATIVE COST SCENARIOS AND SELECTED STATES
OAK RIDGE MODEL
Scenario and
Period
Projected Annual Growth Rates (Percent) of Total
Average Electricity Cost (TOC)
Pennsylvania Ohio Illinois West Virginia Kentucky
Indiana
Base Case
1974-80
1980-85
1985-90
Low-price Case
1974-80
1980-85
1985-90
High-price Case
1974-80
1980-85
1985-90
6.49
5.42
4.71
5.10
3.40
2.70 .
7.90
7.46
6.72
6.66
5.37
4.62
5.10
3.40
2.70
8.18
7.34
6.54
6.43
5.12
4.41
5.10
3.40
2.70
7.77
6.81
6.10
6.76
5.71
4.92
5.10
3.40
2.70
8.42
8.03
7.13
7.12
6.08
5.27
5.10
3.40
2.70
8.13
7.08
6.27
6.50
5.12
4.39
5.10
3.40
2.70
7.31
6.30
6.07
Source: Ibid, Table C-6.
31
-------
TABLE 7
FORECASTS OF ANNUAL GROWTH RATES (1974-1990) OF KWH DEMAND
BY SECTOR, STATE AND COST SCENARIO
OAK RIDGE MODEL
Forecast of Annual Growth Rates
of KWH Demand by Sector
(Percent)
State Cost
Pennsylvania
Ohio
Indiana
Illinois
West Virginia
Kentucky
Scenario
B
L
H
B
L
H
B
L
H
B
L
H
B
L
H
B
L
H
Residential
4.2
4.6
3.7
4.2
5.7
2.7
6.1
7.4
5.1
4.6
6.0
3.2
4.4
5.2
3.4
5.6
6.4
5.3
Commercial
6.4
6.4
6.4
3.9
4.4
3.3
6.1
6.4
5.9
4.1
4.4
3.6
4.7
5.5
3.7
3.5
3.8
3.6
Industrial
5.7
5.5
6.4
4.5(4.7)]'
5.0(5.1)°
3.9(4.2)
5.3
5.8
5..1
3.4
3.9
3.0
3.4
3.8
2.9
4.9(4.9)°
5.6(5.3)°
4.8(4.8)
Total
5'. 4
5.4
5.4
4.3(4.4)^
5.1(5.2)^
3.5(3.7)
5.7
6.4
5.2
4.0
4.7
3.2
3.9
4.5
3.2
4.8(4.8)°
5.5(5.0)°
4.7(4.7)°
(Continued)
32
-------
TABLE 7 (Continued)
o
B=base case, L=low price case and H=high case.
Includes the projected consumption of the DOE uranium enrichment plant in
Portsmouth, Ohio
Q
Includes the projected consumption of the DOE uranium enrichment plant in
Paducah, Kentucky.
Source: Ibid, Tables 7,3, 7.4, 7.6.and 7.7.
33
-------
TABLE 8
FORECASTS OF ANNUAL GROWTH RATES (1974-1990) OF AVERAGE
ELECTRICITY PRICE BY SECTOR, STATE AND COST SCENARIO
OAK RIDGE MODEL
Forecast of Annual Growth Rates
of Average Electricity Price by Sector
(Percent)
State Cost
Pennsylvania
Ohio
Indiana
Illinois
West Virginia
Kentucky
Scenario
B
1
H
B
L
H
B
L
H
B
L
H
B
L
H
B
L
H
Residential
5.2
4.4
6.2
3.3
1.9
4.8
2.8
1.5
3.8
3.3
1.9
4.6
4.1
3.1
5.4
4.0
2.7
4.6
Commercial
6.0
4.9
7.2
4.3
3.4
5.4
3.9
3.1
4.7.
4.7
3.9
5.7
4.2
3.1
5.6
7.2
6.3
8.0
Industrial
5.4
3.6
7.3
5.7
3.9
7.4
4.8
3.1
6.0
6.2
4.6
7.8
5.8
3.8
7.8
5.0
2.8
6.0
Total
5.4
4.2
6.8
4.5
3.1
6.0
3.9
2.6
4.9
4.7
3.4
6.0
5.0
3.6
6.6
5.2
3.6
5.9
Source: Ibid, Appendix D.
34
-------
demographic characteristics, among others. As a result the reported elas-
ticities assume the effects of other variables are held constant in the
estimated models. The necessary assumption for the national "best" esti-
mates to be good estimates of the six state region is that there be
independence between the elasticity measures and the other variables in the
model. Implicit here is the assumption that the econometric models are well
specified and that they contain no specification errors which involve
omitted variables correlated with the included variables.
Table 9 presents the estimates of the six state region average price
and income/output elasticities of kwh demand by customer category using
the first method. Estimates of these regional elasticities are derived by
weighting the customer category elasticities by the percent of total kwh
consumption in 1974 in the six state region by customer category. The
short-run average price and income/output elasticity estimates are more
inelastic than the corresponding long-run estimates.
The second set of estimates for the six state region are based on the
regional estimates of the short- and long-run average price and
income/output elasticities of kwh demand from the Oak Ridge study. For
example an estimate of the six state region residential short-run price
elasticity of kwh demand is derived by constructing a weighted average of
the residential price elasticities in relevant census regions multiplied by
the percent of six state region residential kwh consumption occuring in
states within the census region. The latter consumption weights are
presented in Table 10. The small divergence in percentages across customer
categories is attributable to the small differences among the regions in
the composition of kwh demand by customer category.
The estimates of the short- and long-run price and income elasticities
of kwh demand using this second method are presented in Table 11. A
comparison of the estimated six state price elasticities from the two
methods shows that the average price elasticities derived from the Oak Ridge
study are more elastic in the short-run than the national estimates. In
contrast, the Oak Ridge derived estimates of the total long-run average
price elasticity are more inelastic than the national estimates. The
latter finding is attributable to the inelastic average price elasticity
of kwh demand estimated for the six state industrial category from the Oak
Ridge study.
A comparison of the total short- and long-run income/output elas-
ticities of kwh demand shows a more inelastic demand in the Oak Ridge
derived estimates; in contrast, there is considerable variation in the
comparison of customer category estimates between the two methods. Overall
we have more confidence in the Oak Ridge estimates as reflecting the types
of patterns of elasticities which would be expected. For example, it is
possible that type of industries located in the six state region have more
inelastic demands for electricity than industries located elsewhere in the
35
-------
TABLE 9
ESTIMATES OF SIX STATE REGION PRICE AND INCOME/OUTPUT ELASTICITIES
OF KWH DEMAND BY CUSTOMER CATEGORY-METHOD I
Estimate of Elasticity by
Customer Category in Six State Regipn
Elasticity Measure
Residential Commercial Industrial
Total
Short-run Average Price
Long-run Average Price
Short-run Income/Output
Long-run Income/Output
-.2
-1.0
0.1
1.0
-.25
-1.0
.75
1.6
-.2
-1.1
1.0b
i.ob
-.21
-1.05
.68
1.12
Derived by weighting the customer category elasticity by the following
weights: residential, 30.1; commercial, 20.8; and industrial, 49.1.
These weights are the percent of total kwh consumption in 1974 in the
six state region by customer category.
Assumed to be equal to 1; actual studies show too much diversity for
actual estimate.
Source: Based on subjective review of national studies listed on Table
1 and 2.
-------
TABLE 11
ESTIMATES OF SIX STATE REGION PRICE AND INCOME/OUTPUT
ELASTICITIES OF KWH DEMAND BY CUSTOMER CATEGORY-METHOD II
Customer Category
Elasticity Measure
Short-run Average Price
Long-run Average Price
Short-run Income/Output
Long-run Income/Output
Residential
(Percent)
-0.33a
-1.04
0.16
0.43
Commercial
(Percent)
-0.43
-1.31
0.45
0.96
Industrial
(Percent)
-0.23
-0.41
0.77
1.31
Total
(Percent)
-0.33b
-0.79
0.52
0.97
Individual customer category elasticities calculated by weighting the
customer category elasticities in Table 3-5 by the corresponding weights
in Table 10.
Derived by weighting the customer category elasticity by the following
weights: residential, 30.1; commercial, 20.8; and industrial, 49.1.
These weights are the percent of total kwh consumption in 1974 in the
six state region by customer category.
38
-------
United States. In addition, our review of the Oak Ridge study gave us a
high degree of confidence in this approach to modeling the kwh demand for
electricity. Of course, one must recognize that our estimates of elas-
ticities from both methods are implicitly accompanied by a variance
reflecting our uncertainty as to the "correct" magnitude of these
elasticities.
Finally we utilize the projected average price and kwh demand growth
rates to 1990 from the Oak.Ridge report (Tables 7 and 8) together with
the consumption weights (Table 10) to generate estimates of the six state
region's average price and kwh demand growth rates to 1990 by customer
category. These growth rates are reported in Table 12. The estimated
total kwh demand annual growth rate in the base cost scenario is 4.8
percent while the corresponding average price growth rate was 4.9 percent.
The low price cost scenario results in a higher total kwh demand growth
rate and a lower average price growth rate in comparison to the base case;
the reverse is found in a comparison of the growth rate in the high price
cost scenario compared to the base cost scenario.
SUMMARY
This task of the research project reviewed the literature on the demand
for electricity and generated estimates of the short- and long-run price and
income elasticities of demand for electricity in the six state region.
Since there have been a considerable number of studies in this area and
several economists have recently reviewed this literature, we presented
a summary of the results of these studies. In addition we summarized
the results from a recent study of the demand for electricity by region
which generated elasticity estimates by customer category and cenus region.
Our study generated two sets of elasticity estimates for the.six state
region using our "best" estimates of national elasticities and the regional
estimates from the Oak Ridge model. Of the two sets of estimates presented
we have more confidence in the- derived 0;ik Ridge estimates. In nddi I ion, since
our estimates of the elasticities will be used in simulating demand growth
in the Teknekron model by the year 200, we presented independent estimates
of electricity demand growth in 1990 derived from the Oak Ridge study.
39
-------
TABLE 12
FORECASTS OF ANNUAL GROWTH RATES (1974-1990) OF ELECTRICITY PRICE AND
DEMAND BY SECTOR IN THE SIX STATE REGION
Variable
Growth Rates (Percent) by Customer Category
Cost Scenario Residential Commercial Industrial Total
KWH Demand
Average Electricity
Base
Low Price
High Price
Base
Low Price
High Price
4.7b
5.8
3.7
3.8
2.8
5.0
4.9
5.1
4.5
5.2
4.2
6.1
4.7
5.1
4.5
5.6
4.5
8.4
4.8
5.3
4.3
4.9
4.1
6.1
Same as utilized in Table 6-8 as well as in the text.
Derived by weighting growth rates in Table 7-8 using weights given in
Table 10.
40
-------
SECTION 3
THE IMPACT OF REGULATION ON ELECTRICITY PRICES AND
•GENERATION CAPACITY IN THE OHIO RIVER BASIN.
Electric utilities in the United States are presently in the process of
reevaluating planned additional generating capacity in the context of de-
clining rates of growth in kilowatt-hour (kwh) demand. As a result, some
new capacity increments have been delayed while other planned additions to
generating capacity have been either postponed indefinitely or permanently
cancelled (Wall Street Journal, February 7, 1979). For example, Potomac
Electric Power Company has estimated it will need only 1.2 million kilowatts
(kw) of additional capacity by 1982 in contrast to its original estimate of
4.4 million kw. The power firm's annual kwh growth rate has declined from
9.0 percent in 1973 to approximately 3.0 percent at present (compared to the
national average of 3.5 percent). The decline in demand growth can be at-
tributed to several factors including utility conservation programs, in-
creased energy efficiency in production processes and appliances, and most
important, increased price. The average kwh price has nearly doubled since
1973. This has induced residential consumers to conserve (e.g., reduce
unnecessary use and insulate residences) and has induced commercial and
industrial consumers to make capital investments necessary to conserve ener-
gy. The improvements in the efficiency of production processes and ap-
pliances are expected to continue along with increases in the price of elec-
tricity. As a result, most projections indicate that the United States
annual growth rate in kwh demand will average approximately 4.0 percent
through 1990, significantly less than the historical growth rate of 7.0
percent.
The purpose of this section is to examine the implications of declining
growth in electricity demand as well as the implications of increasing prices
and regulatory restructuring of prices for the Ohio River Basin Energy Study
(ORBES) region. Several regulatory alternatives are examined as to their
pricing and capacity requirements implications for the ORBES region.
One part (The Regulatory Environment) examines the present state of
electricity regulation in the United States, particularly in the context of
increasing pressure for regulatory reform (a function of high rates of in-
flation, increasing consumer militancy, and declining supplies of low cost
fuel). The historical relationship between electricity prices and the
several factors theoretically and empirically associated with these prices
arc briefly reviewed.
41
-------
A second segment (The Pricing Alternatives) examines some alternative
pricing situations. Three basic cases are analyzed. The first is tradi-
tional average cost pricing with no time differentiation, i.e., prices based
on unit or average accounting cost. In the traditional case there is little
justification for presuming any significant alteration in the load patterns
faced by electric utilities. The second is marginal cost pricing with no
time differentiation, i.e., prices based on incremental or marginal cost.
The effect of marginal cost rates is generally higher prices of electricity
than those from average cost rates, particularly since most marginal cost
estimates are derived from projected future costs rather than historical or
imbedded accounting costs. However, marginal cost pricing without time
differentiation (i.e., without incorporation of peak and off-peak
rates), similar to average cost pricing without time differentiation, has
minimal effects on .electric utility load patterns. The third case is some
variation of time differentiated or peak load pricing, i.e., prices varying
with peak and off-peak consumption. It is presumed that load patterns con-
fronting electric utilities are significantly altered under peak load pric-
ing schemes regardless of whether based on marginal cost or average cost.
A third part examines the potential effects of time differentiated prices
regarding required capacity, load factors, and fuel costs (The Potential
Effects of Time Differentiated Pricing). A final segment (The Implications
for the ORBES Region) examines the potential capacity reduction effects from
time differentiated rates for the ORBES region.
THE REGULATORY ENVIRONMENT
Regulation per se appears to have had minimal influences on electricity
rates. In their classic study, Stigler and Friedland (1962) found that av-
erage electricity rate levels were influenced primarily by market factors
(market size and population density) and fuel costs rather than rate regu-
lation itself. Jackson (1969) produced similar conclusions indicating that
the cost per kwh has been a function of fuel costs, utility scale, and
availability of cheap hydroelectric power, and not commission regulation.
Pike (1967) provided evidence that residential electric rates were determined
more by power system scale, fuel costs, and access to relatively inexpensive
public power than by state regulation. Mann (1974) found that electricity
rates for specific residential consumption levels were influenced by genera-
tion costs, distribution costs, and system scale; political factors impacted
on rates for publicly-owned electric firms but not on rates for investor-
owned electric utilities. There is a dissenting voice regarding the impact
(or lack thereof) of regulation on electricity prices. Moore (1975) pro-
duced empirical results that indicated utility regulation increased elec-
tricity prices, i.e., increased regulatory effort as measured by funds
expended and time consumed was associated with increased average kwh prices
for residential, commercial, and industrial users.
The historical situation regarding utility regulation appears to be
changing, at .least in the electric power sector. Inflation coupled with
surging fuel prices has changed a very passive regulatory process into an
active and continuous review process (Joskow, 1974). The kwh price of elec-
tricity tended to decline through the mid-1.960's; it was relatively stable
42
-------
in the latter part of the 1960's. The average kwh price began to rise in the
early 1970's. Confronted with increased rate hike applications and rate
challenges, the nature of the regulatory process in the United States has
changed substantially since 1970. That is, prior to 1970, electric utilities
could maintain adequate profit rates without resorting to price increases
since technological change and scale economies more than offset input price
increases. Shepherd (1976) has noted that 1965-1975 is the period when
electric utility cost functions turned upward, i.e., some electric utilities
in their entirety and other electric utilities in specific components (e.g.,
distribution) began to experience diseconomies of scale. In brief, electric
utilities recently have been confronted with increasing unit cost due to both
shifts upward in cost functions (e.g., the effect of inflation) and dis-
economies of scale (Wilson and Uhler, 1976).
The rising average cost of electricity production is a function of
numerous factors (Berlin, Cicchetti, and Gillen, 1974). One, inflation has
meant increasing input prices (e.g., fuel and labor). Two, the technological
impact which historically decreased unit costs has lessened in recent years;
in addition, technological change in the areas of safety and environmental
requirements has increased unit costs. Three, economies of scale appear to
have been exhausted in some dimensions of electric power (e.g., generation
appears to have reached the threshold of either constant or rising unit costs
with increasing capacity). Four, financial, capital equipment, and construc-
tion costs have increased substantially, as well as land values for gener-
ation plant sites. In brief, the effect of inflation, increasing fuel costs,
and environmental protection have tended to exceed the gains from techno-
logical change and increasing output. As a result, the declining block rate
schedule for electricity, which historically has been justified by both load
factor improvements and long-run economies of scale, may no longer be justi-
fied by the latter (Cicchetti, 1974).
Mann and Witt (1977) indicated that the specific effects of inflation
and rapidly increasing energy prices have been several. One, rate structures
for both publicly-owned and privately-owned electric utilities have become
flatter, i.e., since 1967, there has been a trend toward less disparity in
rates across customer classes. Two, the recent rising costs have been re-
flected significantly more in industrial rates than in commercial and res-
idential electricity rates.
The traditional objectives of utility regulation have been the control
of monopoly earnings, the prevention of excessive price discrimination, and
the assurance of adequate service on a continuous basis to all user classes.
Pressures have emerged in the 1970's for regulatory reform and broadened
regulatory objectives. These external pressures include high inflation
rates, increasing consumer militance, diminishing supplies of low cost fuel,
increasing environmental concern, and technological change (Trebing, 1977).
As a result, there has been regulatory reaction to each external pressure.
For example, the general response to diminishing supplies of low cost fuel
and inflation has been reform and experimentation in electricity rate
structures.
43
-------
THE PRICING ALTERNATIVES
The traditional pricing system for electricity has .involved declining
marginal (and average rates) with increasing use and demand. For small
users (largely residential), there are declining bloqk schedules for kwh
usage. For large users (commercial and industrial), there are declining
energy (kwh) charges coupled with declining capacity (kw) charges, i.e.,
rates that decline with increased kw capacity. The latter are generally
utilized at the time of maximum demand. Thus, electricity price has es-
sentially two components: energy or kwh rates which decline with kwh con-
sumption; and demand or kw rates which decline with maximum kw demand.
The conventional method of pricing, rates based on average or unit
accounting cost, focuses on three kinds of costs in establishing electricity
prices: customer costs (a function of customer number), operating costs
(a function of kwh usage), and demand costs (a function of maximum kw demand).
The latter can be categorized as to generation capacity costs, transmission
capacity costs, and distribution capacity costs. Similarly, prices based on
marginal cost are the sum of marginal generation costs (capacity and
operating), marginal transmission costs (capacity and operating), marginal
distribution costs (capacity and opetating), and marginal customer costs.
The Options
The electricity rate options involve choices of costing method and
whether or not to apply time differentiation. There are four basic options
(Uhler, 1977):
1. Non-time differentiated average cost (NTDAC). This option is
the traditional regulatory approach involving electricity rates
based on some variation of fully distributed, imbedded, or
historical accounting cost. Rates do not vary with time of usage.
2. Non-time differentiated marginal cost (NTDMC). This option involves
electricity rates based on some variation of incremental or marginal
cost (e.g., long-run incremental costs). Rates do not vary with
time of consumption.
3. Time-differentiated average cost '(TDAC). This option involves
electricity rates based on average accounting cost but with
adjustments for time of usage. The peak load pricing can take
the form of either seasonal, time-of-day rate variations, or both.
The critical aspect is that rates vary with time of consumption
i.e., rates vary with peak and off-peak periods.
4. Time-differentiated marginal cost (TDMC). This option involves
electricity rates based on marginal cost adjusted for time of
usage. The peak load pricing can take the form of either seasonal,
time-of-day rate variations, or both. The critical aspect is that
rates vary with time of consumption, i.e., rates vary with peak and
off-peak periods..
44
-------
This categorization indicates that time-differentiated or peak load
pricing can be based on average cost as well as marginal cost. Recent ex-
perimentation with peak load pricing has involved peak and off-peak rates
based on both average and marginal cost. As indicated below, peak load
pricing based on average cost can be viewed as a practical compromise which .
minimizes the potential problem of excess revenue generation (i.e., revenues
in excess of operation and other relevant costs including an adequate rate
of return) occurring under peak load pricing based on marginal cost (Uhler,
1977). In brief, marginal cost pricing is not synonymous with peak load
pricing. Electricity rate levels can be based on either average or marginal
costs. Electricity rate structures can be based on average cost, marginal
cost, or some other consideration (e.g., value of service); electricity rate
structures can also be time differentiated.
Trends in Average Prices
It is instructive to examine past pricing trends under conventional
average cost pricing. The price measure employed is cost per kwh, or
electricity revenues divided by kwh sales. The data sources include the
Federal Power Commission publications: Statistics of Privately Owned
Electric Utilities in the United States; Statistics of Publicly Owned
Electric Utilities in the United States; and the Rural Electrification
Administration publication, Annual Statistical Report, Rural Electric
Borrowers.
Viewing a composite of investor-owned electric utilities, publicly-owned
electric utilities, and rural electric cooperatives, one finds average
electricity prices (in. actual terms) in the United States to be:
Residential Commercial-Industrial Aggregate
1961 2.47C 1.48C 1.80
-------
Table 13 provides the average annual growth rates in electricity
prices for the United States and for an ORBES region sample. The growth
rates for the United States are categorized by both ownership type and by
user class. The growth rates for the ORBES region sample are categorized
by both state and by user class. The price trends are calculated for three
different historical periods: 1961-1975, 1965-1975, and 1970-1975. The
annual growth rates are expressed in actual terms and, when applicable, in
real terms. Given the more recent actual price trends (i.e., 1965-1975,
1970-1975), it is reasonable to conclude that 5.0 percent establishs a
somewhat conservative estimate of future per annum increases in actual
electricity rates; in contrast, 11.0 percent establishs more liberal estimate
of future per annum increases in actual electricity rates.
Tables 14 and 15 present real average price projections. The real price
projections employ actual 1975 kwh prices at their base. Prices are esti-
mated at 5-year intervals through the year 2000. The real price projections
are made under three different annual growth rate assumptions, i.e., 5.0
percent increase in actual prices (the conservative benchmark), 8.0 percent
increase in actual prices (a midpoint estimate), and 11.0 percent increase
in actual prices (the liberal benchmark). The annual inflation rate in-
corporated in the real electricity price projections is 5.5 percent. Thus,
the projected price reflect only real increases in electricity prices.
Table 14 exhibits the price projections for customer categories while Table
15 exhibits the real price projections for the six states represented in
the ORBES region sample.
The price projections in Table 14 can be viewed as illustrative of what
may happen to real electricity prices for the various user categories in the
next several decades; however, they do not take into account regulatory
effects such as industrial users bearing a greater burden of the real rate
increase than commercial and residential users (Mann and Witt, 1977).
Similarly, the price projections in Table 15 can be viewed as illustrative
of what may happen to real electricity prices in the six states represented
in the ORBES region, however, the projections do not take into account ef-
fects such as interstate differences in regulation and differences in acces-
sibility to relatively low cost fuel. The projections in Table 14 and 15
indicate a wide range of possible real average electricity prices by the
year 2000, given different assumptions regarding the rate of increase in
actual electricity prices.
A Comparison of Marginal Cost and Average Cost Pricing
There are important conceptual differences between marginal cost and
average cost pricing as applied to the public utility sector. For example,
the relationship between average and marginal cost on a historical basis is
not identical to the relationship between average and marginal cost on a
current basis (Cicchetti, Gillen, and Smolensky, 1977). That is, marginal
cost estimation by definition is forward-looking (it generally incorporates
a future time horizon of some specified duration). Since the actual costs
46
-------
TABLE 13
ACTUAL AND REAL AVERAGE ANNUAL GROWTH RATE IN UNITED STATES AND
ORBES REGION ELECTRICITY PRICES, 1961-19751
1961-1975
Actual Real
1965-1975
Actual Real
1970-1975
Actual Real
I.
United States
Investor-Owned Electrics 3.4% -1.0%
Publicly-Owned Electrics 2.6 -1.1
Rural Electric Cooperatives 1.1 -3.3
Residential
Commercial-Industrial
Total
2.1 -2.2
4.0 -0.5
3.2 -1.1
5.6% -0.2%
4.3 -0.7
2.8 -3.2
3.9 -1.5
6.3 0.2
5.4 -0.4
III. ORBES Sample
11.8% 3.2%
10.9 2.3
7.8 0.4
9.5 2.5
13.1 3.2
11.7 3.1
. Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Residential
Commercial-
Industrial
Total
1
1
1
3
4
4
1
2
4
3
.5%
.7
.9
.0
.5
.5
.8
.2
.1
.1
-2
-3
1
-2
0
0
-2
-2
-0
-1
.3%
.8
.3
.0
.3
.0
.3
.1
.2
.2
3
3
3
6
6
7
3
3
7
5
.1%
.3
.9
.2
.5
.2
.3
.9
.3
.5
-1
-3
1
0
1
1
-1
_i
1
0
.8%
.9
.6
.8
.3
.2
.8
.8
.6
.0
7.
7.
8.
12.
13.
14.
8.
9.
14.
11.
7%
9
2
7
7
7
6
4
6
7
0.8%
-2.4
-0.1
2.8
5.6
2.7
1.9
-1.0
-1.6
2.7
The real growth rates are a result of adjustments for inflation (as measured
by the Consumer Price and Wholesale Price Indices) during the period,
1961-1975.
7
"Based on a sample of 39 investor-owned electric utilities located in the
ORBES region. The sample accounts for 90.1 percent of 1975 kwh sales in
the six states region.
47
-------
TABLE 14
PROJECTIONS OF REAL AVERAGE ELECTRICITY PRICES (BY USER)
FOR THE ORBES REGION, 1975-20001-
Actual
1975 1980
1985 1990 1995
2000
I. 5 Percent Annual Growth
Residential
Commercial
Industrial
Total
II. 8 Percent Annual Growth
Residential
Commercial
Industrial
Total
III. 11 Percent Annual Growth
3.35C 3.27C 3.20C 3.12c
3.24 3.17 3.09 3.02
2.04 1.99 1.94 1.90
2.69 2.62 2.56 2.50 .
3.35C 3.76C 4.23C 4.76c
3.24 3.64 4.09 4.60
2.04 2.30 2.58 2.90
2.69 3.02 3.40 3.82
3.05C 2.97C
2.95 2.88
1.85 1.81
2.45 2.39
5.35C 6
5.18 5.82
3.26 3.66
4.30 4.83
Residential
Commercial
Industrial
Total
3
3
2
2
.35c
.24
.04
.69
4
4
2
3
.32c
.18
.63
.47
5
5
3
4
• 57c
.39
.39
.47
7
6
4
5
.180
.94
.37
.76
9.26C
8.95
5.64
7.43
11.
11.
7.
9.
93C
54
27
58
Price is measured in cents per kwh. The price projections are based on
1975 data for a sample of 39 investor-owned electric utilities located
in the ORBES region. The real electricity price projections incorporate
an annual inflation rate of 5.5 percent.
48
-------
TABLE 15
PROJECTIONS OF REAL AVERAGE ELECTRICITY PRICES (BY STATE)
FOR THE ORBES REGION, 1975-200Q1
Actual
1975
I. 5 Percent Annual Growth
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
II. 8 Percent Annual Growth
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
III. 11 Percent Annual Growth
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
2
2
2
2
3
2
2
2
2
2
3
2
2
2
2
2
3
2
.710
.28
.31
.75
.04
.54
.710
.28
.31
.75
.04
.54
.710
.28
.31
.75
.04
.54
1980
2
2
2
2
2
2
3
2
2
3
3
2
3
2
2
3
3
3
.650
.23
.26
.69
.97
.48
.050
.56
.59
.09
.42
.85
.500
.94
.98
.54
.92
.27
1985
2
2
2
2
2
2
3
2
2
3
3
3
4
3
3
4
5
4
.580
.17
.20
.62
.90
.42
.42c
.88
.92
.48
.84
.21
.500
.79
.86
.57
.05
.22
1990
2
2
2
2
2
2
3
3
3
3
4
3
5
4
4
5
6
5
.520
.12
.15
.56
.83
.37
.850
.24
.28
.91
.32
.61
.810
.89
.95
.89
.52
.44
1995
2.46o
2.07
2.10
2.50
2.77
2.31
4.330
3.64
3.69
4.39
4.86
4.06
7.490
6.30
6.38 .
7.60
8.40
7.02
2000
2
2
2
2
2
2
4
4
4
4
5
4
9
8
8
9
10
9
.410
.02
.05
.44
.70
.26
.870
.09
.15
.94
.46
.56
.660
.12
.23
.80
.83
.05
Price is measured in cents per kwh. The price projections are based on
1975 data for a sample of 39 investor owned electric utilities located
in the ORBES region. The real electricity price projections incorporate
an annual inflation rate of 5.5 percent.
-------
of generation plant and fuel have been increasing rapidly, one anticipates
that the difference between current average and marginal cost (adjusted for
price level changes) is significantly less than the difference between
historical average cost and marginal cost.
Boiteaux (1960) argued that the difference between current average cost
(AC ) and historical average cost (AC ) for French electricity production in
c . a
the late 1950's was similar to the difference between current average and
marginal cost (MC). That is, AC > AC , to the same degree as AC > MC;
c a c
therefore AC tended to approximate marginal cost. However, for electric
3.
utilities in the United States in the 1970's, it is reasonable to presume
that MC > AC > AC • Disregarding the difficulties of determining the
C 3.
extent to which MC > AC and MC > AC (or conversely, AC > AC ), conser-
c a c a
vative estimates of effects of marginal cost replacing average cost as the
pricing base can be generated by focusing on the difference between MC and
AC ; liberal estimates can be generated by focusing on MC and AC .
C 3
Morton (1976) indicated that since long-run average cost (LRAC) is the
average cost of electricity from all existing plants priced at historical
cost, it will be lower than both current average cost and long-run incremental
cost (LRIC). LRAC incorporates both the cost of new output as well as the
cost of old output; LRIC focuses only on the cost of producing new output.
And, as emphasized previously, peak load pricing can be adopted under either
LRAC or LRIC standards, e.g., a LRAC price of 4c per kwh can have deviations
for peak and off-peak; a LRIC price of 5£ per kwh can have variances for
peak and off-peak.
The regulatory process tends to minimize the differences between
electricity prices ."based" on marginal cost and electricity prices "based"
on average or imbedded cost. Since accounting costs have been (and will
probably continue to be) the dominant consideration in the determination of
the revenue requirements for electric utilities in the regulatory process,
and since marginal cost tends to be higher than historical average cost;
meeting the revenue requirement constraint has involved setting some rates
less than marginal cost (Electric Utility Rate Design Study, 1977). In
constraining marginal cost rates, the result is that they tend to converge
toward average cost rates. Thus, marginal cost pricing in practice deviates
significantly from marginal cost pricing in theory. In brief, a set of
marginal cost rates must meet the regulatory revenue requirements standard
determined by accounting costs. Given the substantial differences between
marginal and actual average cost, the former must be adjusted downward to
yield the required revenues. In practice, marginal cost rates are not
identical to marginal cost but are instead "based" on marginal cost subject
to the revenue requirement constraint (Electric Utility Rate Design Study,
1977).
50
-------
Marginal Cost Pricing
Marginal cost is the specific cost of producing and/or selling a single
incremental unit; in electricity, it can be expressed in terms of either a
kwh or kw increment. That is, the marginal cost of electricity is the change
in total cost by providing additional electricity. It has two primary com-
ponents: short-run marginal cost (SRMC) which is essentially the change in
operating costs by changing the utilization rate of existing capacity; and
long-run marginal cost (LRMC) which is essentially the change in operating
costs and the incremental capacity costs that ensue from capacity expansion.
In sum, marginal cost is simply the cost (or savings) incurred in providing
more (or less) electricity.
The marginal cost of electricity is affected by multiple factors
(Cicchetti, Gilleh, and Smolensky, 1977). Marginal cost varies with voltage
(reflecting differences in transmission-distribution losses at different
voltage levels) at which the consumer receives service, with time (hours,
days and seasons) of usage, quantity of use, and consumer density in the
service area.
The recent experience with marginal cost pricing in the United States
has involved both peak load pricing for electricity based on marginalist
principles as well as electricity pricing (without time differentiation)
based on marginal cost (Joskow, 1976). In the context of peak load pricing,
the application of marginalist principles has generally involved peak users
paying marginal operating plus marginal capacity costs; in theory, off-peak
users pay only marginal operating costs. However, Wenders (1976) criticized
this traditional marginalist approach, particularly under conditions of
different generation technology being employed to produce electricity demand
of different durations. That is, he advocated off-peak prices including
an element of marginal capacity cost, thus reducing the peak marginal cost
price.
Shepherd (1966) distinguished between marginal cost and marginalist
pricing. Marginal cost pricing was defined as setting peak and off-peak
rates to reflect user contributions to peak demand. In contrast, marginalist
pricing was defined as having rate differentials encouraging off-peak usage
and discouraging peak consumption but with the rates having little resem-
blance to marginal cost (e.g., average cost based rates). Non-marginalist
pricing was defined as having prices ignoring cost differentials. Anti-
marginalist pricing was defined as having price differentials contrary to
actual cost differentials. In Shepherd's framework, marginal cost pricing
incorporates time differentiated (peak and off-peak) rates. Marginalist
pricing is oriented toward increasing off-peak consumption rather than
toward economizing on capacity, i.e., marginalist pricing does not involve
structuring of rates in accordance with marginal cost differentials.
The calculation of marginal cost generally involves the projection of
operating and capacity costs for a specified time frame (e.g., ten years).
By definition, the projections exclude historical or imbedded costs, focusing
instead on the change in electricity cost over time with capacity expansion
and demand increments. As indicated by the Electric Utility Rate Design
51
-------
Study (1977), marginal cost estimation has three components: marginal
customer costs, marginal energy costs, and marginal capacity costs. It may
be argued that the conversion from average cost rates to marginal cost rates
would alter significantly the demand forecasts upon which the marginal cost
rates are based. The result could be a revision of projected demand, ca-
pacity requirements, and associated costs. In practice, however, the
regulatory process ensures slow price adjustments; and new prices require
long time periods to have effect since electricity demand is linked to
appliance and equipment stocks (Turvey and Anderson, 1977). Therefore, while
feedback effects from the adoption of marginal cost pricing on capacity re-
quirements and electricity costs are important, these effects are sufficient-
ly lagged that rate-setters can wait until the new prices have had an effect
on demand forecasts, and then price adjustments can be made.
It may be instructive to examine some marginal cost estimates for elec-
tricity provision. One can obtain a comparison of historical average cost
and projected marginal cost from a Environmental Protection Agency study
(1974) of five electric utilities. One of the electric utilities examined
was Potomac Electric Power Company. The firm's 1972 costs (in cents per
kwh) were estimated to be:
(off-peak costs) (peak costs)
Average Cost SRMC LRMC
Residential 2.48C -70c 8.13c
Commercial 2.19 .70 2.50
Industrial 1.43 .70 2.12
The cost estimates indicate that average cost is below long-run marginal
cost (LRMC) and exceeds short-run marginal cost (SRMC). In the Potomac
Electric Power case, SRMC ranged from 29 to 49 percent of average accounting
cost; LRMC exceeded, average cost by 1.1 to 3.3 times. In similar estimates
for Duke Power Company, SRMC across user categories ranged from 16 to 49
percent of .average accounting cost; LRMC exceeded average cost by 1.2 to
2.5 times.
Scherer (1976) estimated the incremental costs for a thermal electric
power system (New York State Electric and Gas Corporation) under different
pollution emmission constraints. The study focused on estimating actual
marginal cost during several time periods within a demand cycle when a power
sytem operates subject to specified pollution limits. Marginal cost was
defined, as the increment to total cost when one additional kw is used in
the system during a specific time period; average cost was defined as total
annualized costs divided by the aggregate energy delivered at all load
centers during all time periods within one year. The average cost calcu-
lations were on a current basis, i.e., there was a fixed cost factor in-
volving insurance and taxes but not including depreciation or other sunk
costs.
52
-------
The summary results in the Scherer analysis indicated that peak marginal
cost, in certain locations within the system, can be 3-4 times average cost,
and can exceed 10 times base load marginal cost. Also, marginal cost for
base, low, intermediate, and. peak periods can vary significantly across load
centers within a specific power system. The marginal cost estimates
(expressed in cents per kwh and in 1970 dollars) for coal-fired steam plants
for seven different load centers were:
Base Period 0.59c 0.56c 0.50c 0.48c 0.54c 0.49C 0.59e
Low 0.59 0.56 0.54 0.52 0.56 0.54 0.62
Intermediate 0.74 0.70 0.67 0.65 0.69 0.67 0.76
Peak 3.42 3.13 3.11 2.94 3.23 3.13 3.78
The current average cost estimate was approximately 0.850- All estimates
included both operating and capacity costs. Peak marginal cost varied from
3.4 to 4.4 times current average cost. Peak marginal cost was generally
4-5 times intermediate demand marginal cost; peak marginal cost was generally
5-6 times low demand marginal cost; and peak marginal was generally 6-7 times
base period marginal cost. Peak marginal cost exceeded average cost, how-
ever average cost in turn exceeded the marginal cost associated with the
intermediate, low, and base demand periods.
Cicchetti, Gillen, and Smolensky (1977) provided marginal cost estimates
for three electric utilities: Wisconsin Power and Light Company (investor-
owned), Sacramento Municipal Utility District (municipally-owned), and Los
Angeles Department of Water and Power (municipally-owned). We focus on the
cost estimates for Wisconsin Power and Light since, similar to the electric
utilities in the ORBES region, it is primarily coal-fired (72 percent in
1974). The other two utilities are heavily dependent on hydro, oil, and
nuclear. The cost estimates are expressed in cents per kwh and in 1975
dollars.
high voltage primary voltage low voltage
winter weekday peak
energy 2.30C 2.38C 2.46c
capacity .95 1.23 1.46
3.25C 3.61C 4.12c
summer weekday peak
energy 1.77c 1.84c 1.90c
capacity .95 1.23 1.65
2.72C 3.16C 3.55C
weekend off-peak 1.30c 1.34c 1.38c
nighttime off-peak .93c -95c .98c
53
-------
The case study indicates that both marginal energy and capacity cost
vary with voltage and with time of consumption. The capacity cost estimates.
incorporated generation, distribution, and transmission capacity change
through 1983; the energy cost estimates were for 1975. The Wisconsin Power
and Light estimates can be viewed as representative of the overall electric
utility industry, i.e., a medium sized utility, diverse customer mix, diverse
generating equipment, and kw demand increasing at a growth rate similar to
the national average (Cicchetti, Gillen, and Smolensky, 1977).
Time Differentiated Pricing
A number of different rate structures can be employed in peak load
pricing. Wenders and'Taylor (1976) noted several variations. As noted
'previously (Uhler, 1977), time differentiated rates may or may not be based
on incremental cost. Furthermore, although marginal cost tends to vary with
peak and off peak usage, there are numerous exceptions (Cicchetti, Gillen,
and Smolensky, 1977). And it may make practical (but not economic) sense
to vary prices with time or peak-off peak when incremental cost does not
vary.
Time differentiated pricing is essentially prices varying with either
kwh usage, kw demand, or both over a daily and/or seasonal cycle. It is
generally based on marginalist principles which postulate that an electric
utility should expand output as long as consumers are willing to pay the
incremental cost of additional production. Its primary purpose has been to
improve system load factors. It offers a solution to the problem of supply-
ing peak demands, but again it is not in practice equivalent to marginal
cost pricing. For example, De Salvia (1969) estimated that is some cases
the ratio of peak marginal cost/off-peak marginal cost to be in excess of
37:1 and the ratio of peak marginal cost/intermediate demand marginal cost
to be in excess of 24:1. In practice, price differentials of this magnitude
would not be allowed since at the very mininum, drastic changes in usage
patterns would result (thus causing new peaks). That is, customers would
seek to avoid peak consumption and thus would shift usage to the original
off-peak periods.
In addition, price differentials of the above magnitude would not be
tolerated in a regulatory environment due to the potential excess revenue
problem. As indicated previously (Electric Utility Rate Design Study, 1977),
meeting the revenue standard in the regulatory process generally leads to
setting some rates below marginal cos.f. The practical consequence is to
constrain rates based on marginal cost to yield total revenues that match
total accounting cost including the permitted rate of return. The end
results is "marginal cost" rates that diverge significantly from actual
marginal cost.
54
-------
THE POTENTIAL EFFECTS OF TIME DIFFERENTIATED PRICING
The arguments for peak load or time differentiated pricing are generally
phrased in terms of capacity investment savings. That is, unless price elas-
ticity of demand is zero, time differentiated pricing will mean less capacity
required to meet peak demands than under uniform rates over time, in
addition to the savings on deferred capacity, there are also potential sav-
ings in fuel costs, particularly in the context of rapidly increasing fuel
prices. 'The mix of capacity cost and fuel cost savings will vary with the
plant mix across electric utilities. For example, a typical power system
consists of a specific mix of plants to serve different types of loads:
peak loads, intermediate loads, and base loads. Each type of load or demand
involves a different ratio of capital to fuel costs, e.g., peak loads are
generally met with plants having relatively low capital costs and high fuel
costs while base loads tend to require plants which are fuel efficient but
involve relatively high capital costs. While deferring some peaking capacity
and deferring some generation capacity whose criterion for installation is
to provide reserve margin or reliability, peak load pricing simultaneously
will tend to increase the capacity requirements for base load generation.
A theme that prevails herein is that the effect on capacity requirements
and associated fuel costs from peak load pricing is difficult to determine,
i.e., "hard" data on capital investment deferment and fuel cost savings are
difficult to obtain. A particular difficulty in evaluating the potential
benefits from time differentiated pricing is determining the alteration of
load curves. No extensive experience in the United States has been reported
upon which to base these determinations (Federal Energy Administration,
1977). However, preliminary analyses show potential and substantial long-
term savings from time differentiated pricing. In brief, since we are on
the threshold of peak load pricing in the United States it is much too early
to conclusively measure the impact on capacity utilization, capacity re-
quirements, etc. (Joint Economic Committee, 1974). Since we have had little
prior experience in the United States with differentiated rates, the results
of consumption shifts with peak load pricing are essentially conjectural
(Berlin, Cicchetti, and Gillen, 1974). However, as experience and data are
accumulated, the effects will become easier to estimate.
The United Kingdom Experience
A pricing experiment (under the auspices of the Electricity Council)
was conducted in England during the period of 1966-1972; the experiment
combined seasonal and time-of-day rate structures for residential users.
The average load factor without the experimental rates was 50 percent; with
the three experimental rates, the annual or average day load factor increased
to 51, 57, and 60 percent (Boggis, 1976; Joint Economic Committee, 1974).
Under a combined seasonal and time-of-day rate schedule in which the peak
price was 300 percent of standard price, intermediate price was 80 percent
of standard, and off-peak price was 40 percent of standard; load factors
experienced a 14 percent increase to 57 percent. Under a seasonal rate'
schedule to which peak price was 150 percent of standard price and off-peak
price was 70 percent of standard, load factors experienced an increase of 20
percent to 60 percent.
55
-------
There'are little data available in the English experiment as to the
effect on capacity requirements from the peak load (marginal cost based)
pricing. It is possible that the period 1966-1972 was insufficient in
duration to-ascertain the long-run impact on usage, operating costs, and
capital investment costs as compared to the French experience of over 20
years. In this context, Wenders and Taylor (1976) noted that the load factor
results reflected more the elevation of consumption valleys than the shaving
of consumption peaks. That is, the peak load data (hourly and seasonal)
indicated little decrease in peak demands. Therefore the capital savings
effect from this rate 'experiment may have been minimal. National Economic
Research Associates (1977) provided an estimate of 3 percent decrease in
peak load megawatt demand as one of the effects of the experiment.
The French Experience
The French initiated their "green tariff," a marginal cost based peak
]oad pricing system, in 1956. There was a transition period of adoption from
1957 through 1963. The peak load pricing system applied only to industrial
consumers and incorporated peak and off-peak energy charges as well as peak
and off-peak kw capacity charges. The statistics indicate that the effects
on load factors for the period 1954-1974 included: an annual percentage
decline in energy consumed of 0.4 at the hourly peak; a 0.6 annual percent-
age decline in the winter daily peak; and a 2.0 percent annual decline in
the summer daily peak (Balasko, 1976). The capital investment savings are
estimated to be 6,500 megawatts of capacity by 1980, i.e., it is estimated
that without the peak load pricing scheme, Electricite de France (EDF) would
need 6,500 megawatts of capacity in addition to the 47,000 megawatts of
capacity planned for 1980. In brief, after 24 years of peak load pricing,
the estimated savings is nearly 14 percent of actual megawatt capacity.
National Economic Research Associates (1977) reported a steady flatten-
ing of both daily and annual load curves over the period 1958-1975. For
example, the winter load factor increased from 72 percent in 1952 to 87
percent in 1975. Average winter peaks relative to summer off-peak consump-
tion decreased from 2.0 to 1.6 in the period of 1958-1975. In brief, the
French experience with industrial peak load pricing over several decades
indicates both substantial improvements in load factors and substantial
savings in capacity requirements.
The United States Experience
The Federal Energy Administration (1977) investigated the effects of
selected load management strategies (including time-of-day pricing) on two
electric utilities. The first electric utility was coal fired with .a large
industrial load; the second electric utility was oil fired with a large
residential load. The analysis involved cost simulations based on his-
torical accounting data for 1974. The operating benefits from time-of-day
pricing were an estimated 4 percent annual decline in operating costs for
both electric utilities and an estimated 7 percent decline in peak demand.
The elapsed time necessary for the substantial decline in peak demand was
not specified.
56
-------
Acton, Manning, and Mitchell (1977) reported on a time-of-day pricing
experiment in Los Angeles. However, the experiment was not initiated until
1977, therefore, unlike the British and French experience, the time elapsed
is insufficient to determine with any accuracy the effects on daily and
annual load factors, operating (fuel) costs and capacity requirements. For
example, 30 months data are certainly not comparable with the French experi-
ence with peak load pricing of over two decades. The same can be concluded
about a combined time-of-day and seasonal pricing experiment in Arkansas
(Kehler, 1977). The latter experiment was initiated in late 1975. No
significant shifts occurred in the early stages of the experiment. Again,
the relatively short duration of the experiment does not allow for accurate
measurement of the long-run effects on usage, load factors, operating costs,
and capacity requirements. Holeman (1978) reported some tentative results
from the Arkansas experiment. One, the time-of-day rates produced sub-
stantial differences between peak and off-peak kw demand for the time-of-day
sample and the control sample. Two, the time-of-day rates have reduced kwh
monthly usage 10 to 20 percent in the early stages of the experiment.
In general, implementation of time-of-use rates for commercial-indus-
trial consumers in the United States is a very recent phenomenon. Therefore,
detailed analyses of effects have not been published (Malko and Simpson,
1977). Long Island Lighting Company implemented time differentiated pricing
for its largest 175 customers in February 1977; Central Hudson Gas and
Electric (New York) implemented peak load pricing for its 50 largest indus-
trial users in March 1978. Other peak load pricing schemes include Wisconsin
Power and Light (130 largest commercial and industrial users, January 1977);
Commonwealth Edison (700 industrial users, November 1977); Detroit Edison
(2,100 industrial users, March 1976); and Consumers Power (3,000 commercial
and industrial users, April 1976).
An up to date summary of various regulatory pricing reforms and experi-
ments is provided by the United States Department of Energy (1979). As
stated previously, in most cases, the time elapsed since the initiation of
the reforms and experiments has been insufficient to ascertain both their
short-run and long-run impact. Pacific Gas and Electric Company implemented
time-of-day and seasonal rates for large users in February, 1977; preliminary
results indicate a 3 percent kw shift from peak to off-peak periods.
Southern California Edison Company and San Diego Gas and Electric Company
report similar preliminary results with time differentiated pricing. The
Los Angeles Department of Water and Power initiated voluntary peak load
pricing in December 1978; total participation is estimated to be 5,000
customers. Virginia Electric and Power Company initiated voluntary time-
of-use rates for residential consumers in 1977; this is to be extended an a
mandatory basis to 20,000 residential customers in 1980. Long Island
Lighting Company in early 1980 will implement mandatory time-of-use pricing
for 1,100 large residential users. Wisconsin Electric Power Company in July
1978 implemented mandatory time-of-use pricing for 580 large residential
consumers. Commonwealth Edison Company initiated a residential pricing
experiment for over 500 customers in late 1978.
57
-------
Uhler (1978), in a recent survey, indicated that substantial rate
structure reform has taken place in the United States. Since January 1976,
approximately 25 states have approved time differentiated electricity rates;
approximately 30 states have initiated voluntary or mandatory load management
experiments; and more than 35 state regulatory commissions have issued orders
or decisions relating to time differentiated pricing. -Forty-one states now
have some form of seasonal peak load pricing while 26 states have some form
of time-of-day peak load pricing. Uhler (1978) noted that the empirical data
is too meagre to provide generalizations as to the load factor, cost, and
capacity requirements effects of rate structure reform. In brief, the
current advocacy of time-of-use pricing is not based on extensive empirical
results.
The Electric Utility Rate Design Study
An important part of the Electric Utility Rate Design Study (1977) was
the employment of several simulation models to estimate the capacity and
cost effects of time differentiated pricing. The computer models simulated
the operation of an electric utility over a specified time period comparing,
among other things, the effects of non-time differentiated versus time
differentiated rates. Each of the two simulation models employed data from
a single electric utility for evaluating the benefits from .load shifts. In
doing so, the models simulated capacity expansion plans for 10-20 years.
Each simulation model employed a base case (non-time differentiated pricing
without load shifting) and several load shifting cases. The load shifting
cases were presumed to be representative of the type of load changes to be
anticipated from time differentiated pricing. Finally, the differences
were calculated between the alternative simulation runs and the base case
thus determining the estimated cost effects that can be attributed to time
differentiated pricing.
The methodology of the simulation analyses presumed the potential bene-
fits from the implementation of peak load pricing to be determined by
specific system characteristics. That is, the potential savings from time
differentiated pricing arc enhanced by general system (supply) character-^
istics such as: a wide range of fuel costs per kwh, low capacity utilization
rates on generating units having low fuel costs, the wholesaling of off-peak
energy at low kwh rates, and generation expansion plans that include high
fuel cost units. In addition, the potential savings from time differentiated
pricing are enhanced by load (demand) characteristics such as: low annual
load factors, low daily load factors particularly on peak days, highly
seasonal peak load patterns, and a high proportion of loads whose demands
are relatively price elastic. In sum, cost savings are presumed to be
influenced by both demand conditions (e.g., load served, peaking conditions)
and supplyconditions (e.g., generation system characteristics, planned
capacity) confronting each electric utility.
One electric utility analysed was Northern States Power Company (Power
Technologies, Inc., 1977). Thi-s electric utility experiences summer peaks,
has a one-third residential consumer mix, and has a generation mix of 90
percent coal-nuclear. The firm had an annual load factor of 54 percent in
1975 with a peak load of 4206 megawatts. The simulation exercise
58
-------
incorporated a planning period of 1976-1990 with time differentiated rates
being imposed in 1977. The simulation focused on all customer categories.
The base case projected the original load patterns without time differenti-
ated pricing. The three simulated cases incorporated different assumptions
as to peak load shifting but incorporated the common premise that 100 per-
cent of the load shift was recovered in off-peak periods. That is, the
three simulated cases varied as to peak periods and recovery patterns.
In the Northern States Power Company simulations, the base case annual
peak load was projected to be 9935 mw in 1990. One simulation under time
differentiated pricing projected a peak load capacity in 1990 of 9278 mw
(a 5.6 percent decrease from the base case) with a 1.4 percent decline in
operation costs and a 7.1 percent decline in total costs. A second simu-
lation projected 1990 peak load capacity to be 9579 mw (a 3.6 percent from
the base case) with a 1.4 percent decrease in operating costs and a 4.6
percent decrease in total costs. A final simulation projected 1990 peak
load capacity to be 9087 mw (an 8.5 percent decrease from the base case)
with a 1.5 percent decrease in operating costs and a 6.4 percent decrease
in total costs. In brief, these time differentiated pricing simulations
indicated capacity savings of approximately 6.0 percent over a 14-year
period (1977-1990) with total costs savings of approximately 6.0 percent.
The second electric utility analysed was Southern California Edison
(Systems Control, Inc., 1977). This utility experiences summer peaks, has
a generation mix of 50 percent oil-gas, and had a annual load factor of 61
percent in 1975 with a peak load of 11,081 mw. Again, the simulation
exercise incorporated a planning period of 1976-1990 with time differentiated
rates commencing in 1977. This simulation focused on industrial users only.
The base case projected £he original load pattern without time differentiated
pricing. The three simulation cases varied as to peak patterns and recovery
periods but incorporated the common premise that energy shifted from peak is
reconstituted at off-peak (total energy delivered remains constant).
In the Southern California Edison simulations, there were three sets of
results. One simulation under time differentiated pricing projected peak
load capacity to decline 10.6 percent from the base case by 1990 with a 4.2
percent decline in operating costs and a 4.1 percent decline in total costs.
A second simulation projected peak load capacity to decrease 5.2 percent
from the base case by 1990 with a 2.6 percent decrease in operating costs
and a 2.1 percent decrease in total costs. A final simulation projected
peak load capacity to decrease 5.2 percent from the base case by 1990 with a
2.6 percent decline in operating costs and a 3.3 percent decline in total
costs. In brief, these time differentiated pricing simulations indicated
capacity savings of approximately 7.0 percent over a 14-year period with
total cost savings of approximately 3.2 percent.
The two simulations indicate that time differentiated pricing can
generate capacity savings ranging from 4.0 to 11.0 percent, operation cost
savings ranging from 2.0 to 4.0 percent, and total cost savings ranging from
2.0 to 7.0 percent. These are estimates for n time interval of 14-years
from the adoption of time differentiated pricing.
59
-------
THE IMPLICATIONS FOR THE ORBES REGION
The simulation results previously discussed were dependent on assump-
tions regarding load shifting. The assumptions concerning changing load
shapes from time differentiated pricing can be viewed as reasonable con-
jectures, however, they do lack extensive empirical support. Therefore,
the simulation results must be viewed as highly tentative.
The assumptions regarding load shifting were necessitated by the
situation of no reliable body of data being available at present concerning
consumer response to time differentiated pricing (Electric Utility Rate Design
Study, 1977).
"At this point in time there does not appear to be a reliable body
of data concerning expected consumer response.... Overall energy
consumption price elasticity data may be satisfactory foundation
for the forecasting of total consumption as a function of average
price, but these data do not indicate if consumers will shift their
consumption (load) pattern and, if so, in what manner (p. 102)."
In brief, the simulations did not incorporate price elasticity effects; they
did not account for the potential change in electricity demand due to the
rate restructuring. The simulations only accounted for declines in peak
loads under specified load shifts. In sum, price elasticity data for.elec-
tricity demand can provide the basis for forecasting energy usage in response
to changes in average prices. The price elasticity data cannot be relied on
to determine whether or not consumers will alter consumption patterns (as
a results of time differentiated rates), to determine the exact nature of the
shift in usage patterns, and to determine the time period necessary for the
consumption changes.
Electricity generation capacity (in megawatts) in the six states
encompassing the ORBES region experienced an average annual growth rate of
A.I percent for 1961-1975; however, capacity increased at the lesser annual
rate of 3.8 percent for 1970-1975. We now examine the potential capacity
reduction effects from time differentiated pricing. This time differentiated
pricing is presumed to be based on marginal cost, although time differentiat-
ed pricing based on average cost will possibly produce similar results. The
effects are analyzed under different average electricity demand (and
corresponding capacity requirements) growth conditions, e.g., annual growth
rates of 3, 4, and 5 percent. The effects are also analyzed under different
peak load capacity reductions, e.g., 10, 15, and 20 percent. The capacity
effects are predicted for the year 2000 assuming the immediate implemen-
tation of time differentiated pricing for electricity.
Table 16 exhibits estimates of capacity reduction effects from time
differentiated rates for the six states region. The electric utility
simulations indicated that peak load capacity can be reduced by as much as
10 percent by time differentiated pricing (based on marginal cost) over a
14-year period. The French experience over a somewhat longer time period
indicated capacity savings of nearly 15 percent. T;ibJc 16 exhibits
estimated reductions in required capacity from time differentiated pricing
60
-------
TABLE 16
POTENTIAL MW CAPACITY REDUCTION EFFECTS FROM
TIME DIFFERENTIATED RATES—THE SIX STATES REGION
MW Capacity Projected MW Time Differentiated Rate Capacity Reduction Effect
19741 Capacity in 2000 10 Percent 15 Percent 20 Percent
Case 1=3 Percent
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Annual Growth Rate
25625
12517
11472
22945
22793
12347
Region 107699
Adjusted Annual Growth Rate in Peak
Case 2=4 Percent
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Annual Growth Rate
25625
12517
11472
22945
22793
12347
Region 107699
Adjusted Annual Growth Rate in Peak
in Average
55262.6
26994.0
24740.4
49483.0
49155.2
26627.4
232262.6
Demand (%)
in Average
71044.5
34703.0
31805.8
63614.3
63192.9
34231.7
298592.2
Demand (%)
Electricity Demand
49736.3
24294.6
22266.4
44534.7
44239.7
23964.7
209036.4
2.57
Electricity Demand
63940.0
31232.7
28625.2
57252.9
56873.6
30808.5
268732.9
3.56
46973.2
22944.9
21029.3
42060.6
41781.9
22633.3
197423.2
2.33
60387.8
29497.6
27034.9
54072.2
53714.0
29096.9
253803.4
3.33
44210.1
21595.2
19792.3
39586.4
39324.2
21301.9
185810.1
2.08
56835.6
27762.4
25444.6
50891.4
50554.3
27385.4
238873.7
3.08
(Continued)
-------
TABLE 16 (Continued)
MW Capacity
19741
Case 3=5 Percent Annual Growth
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Region
Adjusted Annual
25625
12517
11472
22945
22793
12347
107699
Growth Rate in
Projected MW
Capacity in 2000
Rate in Average
91114.1
44506.4
40790.7
81584.9
81044.4
43901.9
382942.4
Peak Demand (%)
Time Differentiated
10 Percent
Electricity Demand
82002.7
40055.8
36711.6
73426.4
72940.0
39511.7
344648.2
4.56
Rate Capacity
15 Percent
77447.0
37830.4
34672.1
69347.2
68887.7
37316.6
325501.0
4.32
Reduction Effect
20 Percent
72891.3
35605.1
32632.6
65267.9
64835.5
35121.5
306353.9
4.07
The 1974 generation capacity data is derived from Exhibit 6 in a memorandum (April 16, 1979) from
Owen Lentz, Executive Manager, East Central Area Reliability Coordination Agreement, to Dr. Walter
Page, Associate Professor Economics, West Virginia University.
-------
by the year 2000 to range from 23,000 mw (the 10 percent reduction case)
to 46,400 mw (the 20 percent reduction case) given a 3 percent annual growth
rate for average demand. Under the assumption of 4 percent annual growth in
average demand, the estimated reductions in required capacity range from
29,900 mw to 59,800 mw. Under the assumption of 5 percent annual growth in
average demand, the estimated reductions in required capacity range from
38,300 mw to 76,600 mw. As indicated by the tabular material, the estimated
reductions in peak load capacity can be translated into lesser annual growth
rates for peak electricity demand. For example, a 20 percent reduction in
capacity requirements by the year 2000 in the case of an annual growth rate
of 3 percent indicates that peak demand is increasing at an annual rate of
approximately 2.08 percent.
Table 17 exhibits estimates of capacity reduction effects from time
differentiated rates for the ORBES region. The estimated reductions in
required capacity by the year 2000 range from 16,600 mw to 33,200 mw given
3 percent annual growth in average demand. Under the assumption of 4 percent
annual growth in average demand, the estimated reductions in required
capacity range from 21,400 mw to 42,800 mw. Under the assumption of 5 per-
cent annual growth in average demand, the estimated reductions in required
capacity range from 27,400 mw to 54,800 mw. The reductions in peak load
capacity in the 3, 4, and 5 percent annual growth cases can be translated
into lesser annual growth rates for peak electricity demand in the ORBES
region.
Table 18 exhibits estimates of load reduction effects (Table 16 and
17 focus on capacity reduction effects) from time differentiated pricing
for the six states region. The estimated reductions in load by the year
2000 range from 17,000 mw to 34,000 mw given 3 percent annual growth in
average demand. Under the assumption of 4 percent annual growth in average
demand, the estimated load reductions range from 21,800 mw to 43,600 mw.
Under the assumption of 5 percent annual growth in average demand, the
estimated load reductions range from 28,000 mw to 56,000 mw. The load re-
ductions in the 3, 4, and 5 percent annual growth cases can be translated
into lesser annual growth rates for peak electricity demand in the six
states region.
In Table 19 are exhibited estimates of load reduction effects from time
differentiated rates for the ORBES region. The estimated reductions in load
by the year 2000 range from 9,900 mw to 19,800 mw given 3 percent annual
growth in average demand. Under the assumption of 4 percent annual growth
in average demand, the estimated load reductions range from 12,800 mw to
25,600 mw. Under the assumption of 5 percent annual growth in average de-
mand, the estimated load reductions range from 16,400 mw to 32,800 mw.
Similar to the other tabular material, the load reductions in the 3, 4, and
5 percent annual growth cases can be translated into lesser annual growth
rates for peak electricity demand in the ORBES region.
63
-------
TABLE 17
POTENTIAL MW CAPACITY REDUCTION EFFECTS FROM
TIME DIFFERENTIATED RATES—THE ORBES REGION
MW Capacity
19741
Projected MW Time Differentiated Rate. Capacity Reduction Effect
Capacity in 2000 10 Percent 15 Percent 20 Percent
Case 1=3 Percent Annual Growth Rate in Average Electricity Demand
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Region
Adjusted Annual
14432
10420
11472
17568
10890
12347
77129
Growth Rate
31123.9
22471.7
24740.4
37887.0
23485.3
26627.4
166335.7
in Peak Demand (%)
28011.5
20224.5
22266.4
34098.3
21136.8
23964.7
149702.2
2.57
26455.3
19100.9
21029.3
32203.9
19962.5
22633.3
141385.2
2.33
24899.1
17977.4
19792.3
30309.6
18788.2
21301.9
. 133068.5
2.08
Case 2=4 Percent Annual Growth Rate in Average Electricity Demand
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Region
Adjusted Annual
14432
10420
11472
17568
10890
12347
77129
Growth Rate
40012.3
28889.1
31805.8
48706.7
30192.2
34231.7
213837.8
in Peak Demand (%)
36011.1
26000.2
28625.2
43836.0
27173.0
30808.5
192454.0
3.56
34010.5
24555.7
27034.9
41400.7
25663.4
29096.9
181762.1
3.33
32009.8
23111.3
25444.6
38965.4
24153.8
27385.4
171070.3
3.08
(Continued)
-------
TABLE 17 (Continued)
• MW Capacity
19741
Projected MW Time Differentiated Rate Capacity Reduction Effect
Capacity in 2000 10 Percent 15 Percent 20 Percent
Case 3=5 Percent Annual Growth
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
"Region
Adjusted Annual
14432
10420
11472
17568
10890
12347
77129
Growth Rate in
Rate in Average
51315.5
37050.1
40790.7
62466.1
38721.3
43901.9
274245.6
Peak Demand (%)
Electricity Demand
46183.9
33345.1
36711.6
56219.5
34849.2
39511.7
246821.0
4.56
43618.2
31492.6
34672.1
53096.2
32913.1
37316.6
233108.8
4.32
41052.4
29640.1
32632.6
49972.9
30977.0
35121.5
219396.5
4.07
The 197- generation capacity data is derived from Exhibit 6 in a memorandum (April 16, 1979) from
Owen Lentz, Executive Manager, East Central Area Reliability Coordination Agreement, to Dr. Walter
Page, Associate Professor Economics, West Virginia University.
-------
TABLE 18
POTENTIAL MW LOAD REDUCTION EFFECTS FROM
TIME DIFFERENTIATED RATES—THE SIX STATES REGION
MW Load
19741
Case 1=3 Percent Annual
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Growth Rate
18585
10387
8360
21228
16790
3303
Region . 78653
Adjusted Annual Growth Rate in Peak
Case 2=4 Percent Annual
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Growth Rate
18585
10387
8360
21228
16790
3303
Region 78653
Adjusted Annual Growth Rate in Peak
Projected MW Time Differentiated Rate Load Reduction Effect
Load in 2000 10 Percent 15 Percent 20 Percent
in Average
40080.2
22400.5
18029.1
45780.1
36209.2
7123.2
169622.3
Demand (%)
in Average
51526.3
28797.6
23177.8
58854.0
46549.8
9157.5
218063.0
Demand (%)
Electricity Demand
36072.2
20160.4
16226.2
41202.1
32588.3
6410.9
152660.1
2.57
Electricity Demand
46373.7
25917.8
20860.0
52968.6
41894.8
8241.7
196256.6
3.56
34068.2 .
19040.4
15324.7
38902.1
30777.8
6054.7
144178.9
2.33
43797.4
24478.0
19701.1
50025.9
39567.3
7783.9
185353.6
3.33
32064.2
17920.4
14423.3
36624.1
28967.4
5698.6
135698.0
2.08
41221.0
23038.1
18542.2
47083.2
37239.8
7326.0
174450.3
3.08
(Continued)
-------
TABLE 18 (Continued)
MW Load
19741
Projected MW
Load in 2000
Time Differentiated
10 Percent 15
Rate Load
Percent
Reduction Effect
20 Percent
Case 3=5 Percent Annual Growth Rate in Average Electricity Demand
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Region
Adjusted Annual
18585
10387
8360
21228
16790
3303
78653
Growth Rate
66082.2
36932.8
29725.4
75479.8
59699.7
11744.4
279664.3
in Peak Demand (%)
59474.0
33239.5
26752.9
67931.8
53729.7
10570.0
251697.9
4.56
56169.9
31392.9
25266.6
64157.8
50744.7
9982.7
237714.6
4.32
52865.8
29546.2
23780.3
60383.8
47759.8
9395.5
223731.4
4.07
The 1974 generation load data is derived from Exhibit 6 in a memorandum (April 16, 1979) from
Owen Lentz, Executive Manager, East Central Area Reliability Coordination Agreement, to Dr. Walter
Page, Associate Professor of Economics, West Virginia University.
-------
oo
TABLE 19
POTENTIAL MW LOAD REDUCTION EFFECTS FROM
TIME DIFFERENTIATED RATES—THE ORBES REGION
MW Load . Projected MW
19741 Load in 2000.
Case 1=3 Percent
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Annual
6800
7509
8360
14744
5631
3076
Growth Rate in Average
14664.8
16193.8
18029.1
31796.8
12143.8
6633.7
Region 46120 99462.0
Adjusted Annual Growth Rate in Peak Demand (%)
Case 2=4 Percent
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Annual
6800
7509
8360
14744
5631
3076
Region 46120
Adjust Annual Growth Rate
Growth Rate in Average
18852.8
20818.5
23177.8
40877.3
15611.8
8528.1
127866.3
in Peak Demand (%)
Time Differentiated Rate Load Reduction Effect
10 Percent 15 Percent 20 Percent
Electricity Demand
13198.3
14574.4
16226.2
28617.1
10929.4
5970.3
89515.7
2.57
Electricity Demand
16967.5
18736.6
20860.0
36789.6
14050.6
7675.3
115079.6
3.56
12465.1
13764.7
15324.7
27027.3
10322.2
5638.6
84542.6
2.33
16024.9
17695.7
19701.1
34745.7
13270.0
7248.9
108686.3
3.33
11731.8
12995.0
14423.3
25437.4
9715.0
5307.0
79569.5
2.08
15082.2
16654.8
18543.2
32701.8
12489.4
6822.5
102292.9
3.08
(Continued)
-------
TABLE 19 (Continued)
MW Load Projected MW
19741 Load in 2000
Time Differentiated Rate Load Reduction Effect
10 Percent 15 Percent 20 Percent
Case 3=5 Percent Annual Growth Rate in Average Electricity Demand
Illinois
Indiana
Kentucky
Ohio
Pennsylvania
West Virginia
Region
Adjusted Annual
6800
7509
8360
14744
5631
3076
46120
Growth Rate
24178.6
26699.6
29725.4
52424.8
20022.0
10937.2
163987.6
in Peak Demand (%)
21760.7
24029.6
26752.9
47182.3
18019.8
9843.5
147588.8
4.56
20551.8
22694.7
25266.6
44561.1
17018.7
9296.6
139389.5
4.32
19342.9
21359.7
23780.3
41939.8
16017.6
8749.8
131190.1
4.07
The 1974 generation load data is derived from Exhibit 6 in a memorandum (April 16, 1979) from Owen
Lentz, Executive Manager, East Central Area Reliability Coordination Agreement, to Dr. Walter Page,
Associate Professor of Economics, West Virginia University.
-------
The raw capacity and mw load data for both the six states regions and
the ORBES region indicate the potential for significant reductions 'in load
and required capacity from time differentiated pricing. The data indicate
that reasonable estimates of potential load and required capacity reductions
from peak load pricing for electricity are not, by any measurement, negli-
gible. However, the effect, in the short-run on peak demands and capacity
requirements from time differentiated pricing may be minimal, particularly
in the context of slow adjustment in usage patterns of electricity users.
In addition, for the ORBES region, the potential reductions in load capacity
from time differentiated pricing have a much higher probability associated
with them than the potential reductions in generating capacity. This is due
to the potential for the ORBES region becoming a more important exporter of
electric power by the year 2000 than it is at present.
The focus in this section has been on the effects of peak load pricing
on electric utility generating capacity. Other regulatory reforms are
taking place, however, in general these reforms have direct consequences
primarily for utility finance and user class revenues and probably have
minimal consequences for capacity requirements and associated costs. For
example, the imposition of lifeline rates generally affects the distribution
of revenues across customer classes. The adoption of automatic cost ad-
justment clauses generally affects the speed with which rate increases can
be implemented. In brief, this report has focused on regulatory reform
concerning rate level and rate structure determination, with particular
emphasis on issues such as time differentiation and the cost basis for rates.
These particular reforms will tend to have impact only on variables such as
load factors, load patterns, and capacity requirements. Regulatory reforms
concerning administrative procedure, and changes in the very fabric of the
regulatory process have been ignored. These latter types of regulatory
reform obviously could have affects .on distribution of wealth-income,
utility profits, and utility financing.
The regulatory reform of time differentiated pricing will have its
primary impact on capacity requirements and associated costs. It is
difficult to perceive any significant impact that peak load pricing will
have on variables such as profits and utility revenues. Future electric
utility profits will continue to be a product of regulatory constraints,
i.e., rates of return will be adequate to attract capital and compensate
existing investors. The rate of annual increase in average electricity
prices will be dampened somewhat by regulatory reforms such as peak load
pricing, however, the critical determinants of per annun increases in
electricity prices will continue to be inflation rates and the increasing
cost of energy.
70
-------
REFERENCES (Continued)
Chern, W. S., R. E. Just, B. D. Holcomb, and H. D. Nguyen. Regional
Econometric Model for Forecasting Electricity Demand by Sector and by
State. Oak Ridge, Tennessee: Oak Ridge National Laboratory,
NUREG/CR-0250, October 1978.
Cicchetti, Charles J. "Electricity Price Regulation: Critical Crossroads
or New Group Participation Sport." Public Utilities Fortnightly,
Vol. 94, August 29, 1974, pp. 3-8.
Cicchetti, Charles J., William J. Gillen, and Paul Smolensky. The Marginal
Cost and Pricing of Electricity: An Applied Approach. Cambridge,
Massachusetts: Ballinger Publishing Company, 1977.
De Salvia, Donald N. "An Application of Peak Load Pricing." Journal of
Business, Vol. 42, October 1969, pp. 458-576.
Edmonds, J. A. A Guide to Price Elasticities of Demand for Energy: Studies
and Methodologies. Oak Ridge, Tennesee: Oak Ridge Associated
Universities, ORAU/IEA-78-15 (R), August 1978.
Electric Utility Rate Design Study. Potential Cost Advantages of Load
Management. Report prepared by Task Force No. 6, March 1977.
Electric Utility Rate Design Study. Rate Design and Load Control; Issues
and Directions. Palo Alto, California: Report Prepared for the
National Association of Regulatory Utility Commissioners, November 1977.
Ellison, E. G., et al. Elasticity of Demand. Palo Alto, California: Report
prepared for the Electric Utility Rate Design Study, January 1977.
Fisher, F..and C. A. Kaysen. A Study in Econometrics: The Demand for
Electricity in the United States. Amsterdam: North Holland Publishing
Company, 1962.
Griffin, J. M. "The Effects of Higher Prices on Electricity Consumption."
Bell Journal of Economics and Management Science, Vol. 5, Autumn
1974, pp. 515-539.
Halvorsen, R. Residential Demand for Electricity. Cambridge, Massachusetts:
Ph. D. Dissertation, Harvard University, 1972.
Halvorsen, R. "Residential Demand for Electric Energy." Review of Economics
and Statistics, Vol. 57, February 1975, pp. 12-18.
Halvorsen, R. "Demand for Electric Energy in the United States." Southern
Economic Journal, Vol. 42, April 1976, pp. 610-625.
Halvorsen, R. Econometric Models of United States Energy Demand, Lexington,
Massachusetts: Lexington Books, 1978.
72
-------
REFERENCES (Continued)
Holeman, 0. V. "Mandatory Time-of-Day Rates: The Arkansas Power and Light
Company." Assessing New Pricing Concepts in Public Utilities, Harry
M. Trebing, Editor.. East Lansing, Michigan: Institute of Public
Utilities, 1978, pp. 430-441.
Houthakker, H. S. "Some Calculations of Electricity Consumption in Great
Britain." Journal of the Royal Statistical Society, Series B,
Vol. 114, 1951, pp. 351-371.
Houthakker, H. S. and L. D. Taylor. Consumer Demand in the United States.
2nd Edition. Cambridge, Massachusetts: Harvard University Press, 1979.
Houthakker, H. S., P. K. Verleger, and D. P. Sheehan. "Dynamic Demand
Analysis for Gasoline and Residential Electricity." American Journal
of Agricultural Economics, Vol. 56, May 1974, pp. 412-418.
Hudson, E. A. and D. W. Jorgenson. "United States Energy Policy and Economic
Growth, 1975-2000." Bell Journal of Economics and Management Science.
Vol. 5, Autumn 1974, pp. 461-514.
Jackson, Raymond. "Regulation and Electric Utility Rate Levels." Land
Economics, Vol. 45, August 1969, pp. 372-376.
Joskow, Paul L. "Inflation and Environmental Concern: Structural Change in
the Process of Public Utility Price Regulation." Journal of Law and
Economics. Vol. 17, October 1974, pp. 291-327.
Joskow, Paul L. "Contributions to the Theory of Marginal Cost Pricing."
Bell Journal of Economics, Vol. 7, Spring 1976, pp. 197-206.
Kehler, Larry L. ."Time of Day Pricing: An Empirical Study." Adapting
Regulation to Shortages, Curtailment^ and Inflation, John L. O'Donnell,
Editor. East Lansing, Michigan: Institute of Public Utilities,
Michigan State University, 1977, pp. 199-208.
Lacy, A. W. and D. R. Street. An Econometric Analysis of Residential
Electricity Demand for the Alabama Power Company. Auburn, Alabama.
Report prepared for the Alabama Power Company, September 1975.
Levy, P. F. The Residential Demand for Electricity in New England, Cambridge,
Massachusetts: Massachusetts Institute of Technology, MIT-EL-017, 1973.
Lin, W., E. Hirst, and S. Cohn. Fuel Choices in the Household Sector.
Oak Ridge, Tennessee: Oak Ridge National Laboratory, ORNLICON-3,
October 1976.
Lyman, R. A. Price Elasticities in the Electric Power Industry. Tucson,
Arizona: Department of Economics, University of Arizona, unpublished
p.-ipt-.r, 1973.
73
-------
REFERENCES (Continued)
Maddala, G. S. Econometrics. New York, New York: McGraw-Hill, 1977.
Malko, J. Robert, and James Simpson. "Time-of-Use Pricing in Practice: Some
Recent Regulatory Actions." Assessing New Pricing Concepts in Public
Utilities, Harry M< Trebing, Editor. East Lansing, Michigan: Institute
of Public Utilities,. 1978, pp. 389-419.
Mann, Patrick C. "User Power and Electricity Rates." Journal of Law and
Economics, Vol. 17, October 1974, pp. 433-443.
Mann, Patrick C., and Tom S. Witt. "The Impact on Electricity Rate
Structures from Increasing Energy Prices." Proceedings of the American
Statistical Assoication-Business and Economics Section, 1977,
pp. 566-571.
Mitchell, B. M. Selected Econometric Studies of the Demand for Electricity;
Review and Discussion. Santa Monica, California: Rand Corporation,
P-5544, November 1975.
Moore, Charles C. "Has Electricity Regulation Resulted in Higher Prices?;
An Econometric Evaluation Utilizing a Calibrated Regulatory Input
Variable." Economic Inquiry, Vol. 13, June 1975, pp. 207-230.
Morton, Walter A. "Long-Run Incremental Cost and Pricing of Electricity."
Public Utilities Fortnightly. Vol. 97, March 11, 1976, pp. 34-39;
Vol. 97, March 25, 1976, pp. 25-30.
\
Mount, T. D., D. Chapman, and T. Tyrrell. Electricity Demand in the United
States: An Econometric Analysis. Oak Ridge, Tennessee: Oak Ridge
National Laboratory, ORNL-NSF-EP-49, May 1973.
National Economic Research Associates. Analysis of Electricity Pricing in
France and Great Britain. New York, New York: National Economic
Research Associates, Report prepared for the Electric Utility Rate
Design Study, January 1977.
National Economic Research Associates. Considerations of the Price
Elasticity of Demand for Electricity. New York, New York: Report
prepared for the Electric Utility Rate Design Study, January 1977.
Nelson, D. C. "A Study of the Elasticity of Demand for Electricity by
Residential Consumers: Sample Markets in Nebraska." Land Economics,
February 1965, pp. 92-96.
Nissen, D. H. and D. H. Knapp. A Regional Model of Interfuel Substitution.
Presented at the SIMS Research Application Conference on Energy,
Alta, Utah, July 7-11, 1975.
Pike, John. "Residential Electric Rates and Regulation." Quarterly Review
of Economics and Business, Vol. 7, Summer 1967, pp. 45-52.
74
-------
REFERENCES (Continued)
Power Technologies, Inc. Potential Cost Advantages of Peak Load Pricing.
Schnectady, New York: Power Technologies, Inc., Report prepared for
the Electric Utility Rate Design Study, February 1977.
Scherer, Charles R. "Estimating Peak and Off-Peak Marginal Costs for an
Electric Power System: An Ex Ante Approach." Bell Journal of
Economics, Vol. 7, Autumn 1976, pp. 575-601.
Shepherd, William G. "Marginal Cost Pricing in American Utilities."
Southern Economics Journal, Vol. 33, July 1966, pp. 58-70.
Shepherd, William G. "Price Structure, Social Efficiency, and Equity."
New Dimensions in Public Utility Pricing, Harry M. Trebing, Editor.
East Lansing, Michigan: Institute of Public Utilities, Michigan
State University, 1976, pp. 125-143.
Stigler, George J. and Claire Friedland. "What Can Regulators Regulate?:
The Case of Electricity." Journal of Law and Economics, Vol. 5,
October 1962, pp. 1-16.
Systems Control, Inc. Estimating the Benefits of Peak Load Pricing for
Electric Utilities. Palo Alto, California: Systems Control, Inc.,
Report prepared for the Electric Utility Rate Design Study,
February 1977.
Taylor, L. D. "Demand for Electricity: A Survey" Bell Journal of Economics,
Vol. 6, Spring 1975, pp. 74-110.
Taylor, L. D. "Demand for Energy: A Survey of Price and Income
Elasticities" in W. D. Nordhaus, Editor, International Studies of the
Demand for Energy. Amsterdam: North Holland Publishing Company, 1977.
Taylor, L. D., G. R. Blattenberger, and P. K. Verleger. The Residential
Demand for Energy. Lexington, Massachusetts: Data Resources
Incorporated, EPRI EA-235, Volume 1, January 1977.
Trebing, Harry M. "Braodening the Objectives of Public Utilities
Regulation." Land Economics, Vol. 53, February 1977, pp. 106-120.
Turvey, Ralph and Dennis Anderson. Electricity Economics. Baltimore,
Maryland: Johns Hopkins University Press, 1977.
Uhler, Robert G. "Rate Making Objectives and Rate Design Options."
Adapting Regulation to Shortages, Curtailment, and Inflation,
John O'Donnell, Editor. East Lansing, Michigan: Institute of Public
Utilities, Michigan State University, 1977, pp. 209-220.
75
-------
REFERENCES (Continued)
Uhler, Robert G. "New Electricity Rate Structures: Some Empirical
Findings." Assessing New Pricing Concepts in Public Utilities, Harry
M. Trebing, Editor. East Lansing, Michigan: Institute of Public
Utilities, 1978, pp. 3-11.
United States Department of Energy. Electric and Gas Utility Topics:
Current Documents. Washington, D.C.: Office of Utility Systems
Federal Energy Administration, July 1979.
United States Environmental Protection Agency. The Economic and
Environmental Benefits from Improving Electrical Rate Structures.
Washington, D.C.: Office of Research and Development, November 1974.
United States Federal Energy Administration. 1976 National Energy Outlook.
Washington, D.C.: Federal Energy Administration, 1976.
United States Federal Energy Administration. Impact of Load Management
Strategies Upon Electric Utility Costs and Fuel Consumption.
Washington, D.C.: Office of Conservation, Federal Energy
Administration, June 1977.
United States Joint Economic Committee. The Public Utility Industry,
Washington, D.C. Hearing before the Joint Economic Committee,
93rd Congress, Second Session, December 4, 1974.
Uri, N. D. A Dynamic Demand Analysis for Electrical Energy by Class of
Consumer. Washington, D.C.: Bureau of Labor Statistics, United
States Department of Labor, Working Paper 34, January 1975.
Wenders, John T. and Lester D. Taylor. "Experiments in Seasonal Time-of-Day
Pricing of Electricity to Residential Users." Bell Journal of
Economics, Vol. 7, Autumn 1976, pp. 531-552.
Wenders, John T. "Peak Load Pricing in the Electric Utility Industry."
Bell Journal of Economics, Vol. 7, Spring 1976, pp. 232-241.
Wilder, R. P. and J. F. Willenborg, "Residential Demand for Electricity:
A Consumer Panel Approach." Southern Economic Journal, Vol. 42,
October 1975, pp. 212-217.
Wilson, J. W. "Residential Demand for Electricity." Quarterly Review of
Economics and Business, Vol. 11, Spring 1971, pp. 7-22.
J. W. Wilson and Associates, Elasticity of Demand. Washington, D.C.:
Report prepared for the Electric Utility Rate Design Study,
February 1977.
76
-------
REFERENCES (Continued)
Wilson, John W. and Robert G. Uhler. "Inverted Electric Utility Rate
Structures: An Empirical Analysis." New Dimensions in Public
Utility Pricing, Harry M. Trebing, Editor. East Lansing,
Michigan: Institute of Public Utilities, Michigan State University,
1976, pp. 163-204.
77
------- |