the hopkinson tariff alternative to tou rates in the israel electric corporation

Copyright # 2002 John Wiley & Sons, Ltd.

MANAGERIAL AND DECISION ECONOMICS

Manage. Decis. Econ. 23: 9–19 (2002)

DOI: 10.1002/mde.1040

The Hopkinson Tariff Alternative to TOURates in the Israel Electric Corporation

C.K. Wooa,*, Brian Horiia and Ira Horowitzb

aEnergy and Environmental Economics, Inc., 353 Sacramento Street, Suite 1700,San Francisco, CA 94111, USA

bDecision and Information Sciences, Warrington College of Business Administration, University of Florida,Gainesville, FL 32611, USA

This paper determines three alternative Hopkinson tariffs to replace the Israel Electric

Corporation’s time-of-use (TOU) energy rate. The first apportions any system residual

revenue requirement between customer classes, based on their respective historic peakdemands. The second collects the same revenue as the current TOU rates. The third allocates

generation revenue requirements for base-load and peaking generation plants in proportion to

the base-load and peak-period energy consumption of each class, and allocates transmission

and distribution revenue requirements in proportion to the connected load of each class. Weshow that a Hopkinson tariff with demand subscription is an attractive alternative to TOU

rates, especially when quantity rationing is essential to maintaining a balance between the

provision of energy and the demand for it. Copyright # 2002 John Wiley & Sons, Ltd.

INTRODUCTION

Since the turn of the 20th century, the developedand developing nations of the world havedepended upon electricity as a primary source ofenergy. The electric utilities that provided thatenergy have commonly been spatial monopoliesthat owned the power-generation plants, thetransmission networks needed to transport thepower from its point of origin to one or morestrategically placed distribution facilities, and thedistribution facilities that direct the power to theend users.Integrated utilities achieve their monopolies as a

natural outgrowth of the employment of high-fixed-cost technologies in both generation andtransmission that go hand in hand with economiesof scale. Increasing returns in centralized genera-tion produce declining average costs with respectto output; even though recent technological

advances in combined cycle gas turbine havemade generation shed its natural monopolyimage. Nonetheless, increasing returns continuein transmitting power through the networks. Thistwo-pronged thrust has seen integrated utilitiesaccorded natural-monopoly status that warrantssome form of government regulation, including, atits most extreme end, nationalization. And,regulation typically implies some form of controlover price and the inevitable public debate over theappropriate form.In the electric utility industry the two principal

opposing sides in the debate have been thepractitioners commonly favoring a Hopkinsontariff, and the theoreticians commonly favoringtime-of-use (TOU) energy rates. The basic form ofa Hopkinson tariff consists of ‘separate energy andpeak demand charges’ (Munasinghe, 1990, p. 122).The simplest form of TOU pricing charges thecustomer marginal-cost-based hourly energy ratesfor peak and off-peak energy use. As detailed inSeeto et al. (1997, p. 171), it is always possible todetermine a Hopkinson tariff structure that willequate the customer’s bills to those that would

*Correspondence to: Energy and Environmental Economics,Inc., 353 Sacramento Street, Suite 1700, San Francisco, CA94111, USA. E-mail: [email protected] (C.K. Woo)

result under TOU pricing. Therefore, the choicebetween the two schemes ‘necessarily rests onconsiderations other than cost to the customer andthe ability to raise revenue for the utility’ (Seetoet al., 1997, p. 172).In recent years, electric utilities such as the Israel

Electric Corporation of Haifa (IEC) have usuallychosen to price their sales to large industrialcustomers at seasonal TOU energy rates. Thesecustomers, however, always have the option ofbypassing the utility, say by the on-site generationof their own power. The basic question thecustomer must answer is whether it is economicallyfeasible to do so, and the customer will notnecessarily answer the question correctly. Uneco-nomic bypass occurs when the utility would havebeen able to satisfy the customer’s energy needs atan incremental cost that is less than the cost of thealternative source upon which the customer haschosen to rely. Unfortunately, TOU pricing oftensends price signals that cause uneconomic bypassand may induce inefficient consumption.In principle, real-time pricing (RTP) with spot

energy rates that track marginal costs by time andspace is efficient (Bohn et al., 1984), but, inpractice, it is costly for a utility to implementand for its customers to use. These factors haveengendered a search by academics and practi-tioners alike for an alternative to both RTP andTOU energy rates. The alternative spawning thegreatest interest in recent years is a Hopkinsontariff with demand subscription (Spulber, 1992;Woo, 1990, 1991; Chao and Wilson, 1987; Wilson,1993).The Hopkinson tariff dominates TOU pricing in

four specific regards (Seeto et al., 1997). First, iteffects efficient rationing, and through demandsubscription charges helps to maintain reliabilityof service under a system capacity constraint.Those charges are based on marginal capacitycosts for generation, transmission and distribution(T&D). Second, it effects efficient consumptionthrough TOU rates set at marginal energy costs.Third, it collects fixed and sunk costs through a so-called connected load charge that applies to acustomer’s previously recorded maximum rate ofelectricity consumption (i.e. kilowatt (kW) de-mand). Finally, it prevents uneconomic bypassthrough marginal-cost-based charges for energyand demand subscription. Like RTP, however, thetariff is also complicated to implement. Whensurplus capacity enables the utility adopting a

Hopkinson tariff to maintain almost perfectreliability, it should initially choose one withTOU energy rates and a connected-load charge.If the surplus evaporates, the tariff can beaugmented with rate options that allow thecustomer to pay an additional subscription fee toassure a higher probability that its energy demandswill be met during those times when the system istaxed to capacity and beyond.Joskow (1997) reports a worldwide trend of

reform in electricity markets. Rate design by anintegrated utility, however, continues to be animportant issue that extends beyond the indus-trialized economies of the Western World and tothe developed and developing economies of Asia,Africa, Eastern Europe, and the Middle East.Prima facie evidence for this is that management inthe IEC has questioned whether it should replaceits industrial TOU energy rates with a Hopkinsontariff.With an installed capacity of approximately

8500 megawatts (MW), the IEC is a government-owned integrated utility that supplies all of theend-use loads in Israel. To provide input intomanagement’s decision-making process, we havedeveloped three Hopkinson tariffs as alternativesto replace the IEC’s industrial TOU energy rates.The first Hopkinson tariff apportions betweencustomer classes any system residual revenuerequirement to recover fixed and sunk costs inaccordance with the historic peak demand of eachclass. The residual revenue is the total revenuerequirement less energy sales priced at the TOUmarginal energy costs. The second is a status quotariff that collects the same revenue as the currentTOU rates. The third tariff seeks to allocate‘system capacity costs to different pricingperiods’ (Munasinghe, 1990, p. 122). Thus, itallocates generation revenue requirements for boththe base-load generation plants that operateduring all periods and the peaking generationplants that operate only in periods of high or peakdemand, in proportion to the base-load and peak-period energy consumption of each customer class.T&D revenue requirements are allocated inproportion to the connected load of each class,as detailed below.To preview our results, we conclude that RTP

based on marginal-cost pricing is an efficientprocedure with severe implementation and trans-action costs that render it impractical. TOUpricing with energy rates that reflect the utility’s

C.K. WOO ET AL.10

Copyright # 2002 John Wiley & Sons, Ltd. Manage. Decis. Econ. 23: 9–19 (2002)

expected marginal costs is inefficient, because thesystem is not perfectly reliable, the utility mustrecover its costs for financial viability, and becauseof varying and location-specific distribution costs.We show that a Hopkinson tariff with demandsubscription is an attractive alternative to thesetwo options, especially when quantity rationing isessential to maintaining a balance between theprovision of energy and the demand for it.

THE MARGINAL-COST PRICING

PRINCIPLE

Marginal-cost (MC) pricing is a fundamental tenetin electricity rate design. The MC price is amarket-clearing price that ‘signals to consumersthe costs of producing the last unit of the good,’and simultaneously ‘signals to competitive produ-cers the marginal benefits of the good to con-sumers’ (Spulber, 1989, p. 234). Electricitydemand, however, is time dependent, and so toomay be the capacity available to satisfy it. Thisimplies that the ‘price in each period must equalthe conditional expected marginal operating plusrationing costs’ (Crew and Kleindorfer, 1976,p. 214), and that marginal capacity costs must beappropriately apportioned (Vardi et al., 1977).TOU pricing, a staple of practitioners and

theoreticians alike in both England and Francein the 1960s, was ‘discovered’ by America’sintegrated regulated electric utilities in the 1970s(Joskow, 1976; Acton, 1982). An importantadvance was that efficient TOU energy ratesshould track the continuously varying MC in realtime to clear spot energy markets (Bohn et al.,1984). The resulting RTP scheme is efficient, as itmaintains the equality between MC and marginalbenefit through its market-clearing prices.There are, however, both theoretical and real-

world caveats that dim the luster of MC-basedRTP. The efficiencies of the MC pricing principlemight only be realized in a certainty world withoutexternalities, incomplete or asymmetric informa-tion, or potentially distorting government rulesand regulations (Graff, 1967; Spulber, 1989).Further, RTP has significant implementation andtransactions costs, including costly metering andaccounting systems for billing and settlementpurposes. It is costly to transmit the spot pricesto customers in real time, even if the real-time MC

can be accurately computed, and customers mustbe able to understand and rationally respond toRTP rates in real time. Moreover, RTP does notfully collect the integrated utility’s fixed and sunkcosts when the marginal costs consistently fallbelow the average costs. Finally, electricity cannotbe economically stored, and must be supplied ondemand. The continuous balance between systemdemand and supply is absolutely critical to preventvoltage drop and the ensuing brownouts andblackouts. Given the inherent demand and supplyuncertainties, periodic quantity rationing is inevi-table for the network’s safe and reliable operation.Capacity rationing can only be made efficientwhen the electric utility knows customers’ will-ingness to pay (WTP) for electricity. RTP does notprovide the mechanism used by customers toreveal their private WTP information beforerationing becomes necessary. Nor does RTP definethe contractual obligation of the electric utility inproviding reliability to each individual customer.As will be shown later, a Hopkinson tariff withdemand subscription can be an attractive alter-native to RTP.The problems with RTP compound with the

global deregulation and restructuring that nowmarks the industry (Joskow, 1997). Deregulatedgeneration markets are conducive to decentralizedownership and control of plants, and therefore tocompetitively priced energy that reflects marginalgeneration costs. Their economies of scale, scope,and density mean that T&D tend to remainmonopolistic. Efficient dispatch by a systemoperator requires a transmission tariff to coordi-nate the decentralized decisions of buyers andsellers. This tariff can also be MC based,incorporating the marginal costs of generationand line losses, and the opportunity cost createdby network congestion (Hogan, 1992). The differ-ence in the spot prices between locations will bethe sum of the differences in their marginal linelosses and their marginal congestion costs (Bohnet al., 1984). Applied to consumption at thedistribution level, RTP that reflects TOU energyrates should account for the marginal distributioncapacity cost by TOU period. Orans et al. (1994)describes a process for quantifying that cost.In any event, the pervasive transactions and

implementation costs of RTP, and the risk ofdemand-and-supply imbalance cast doubt on itseconomic efficiency. A limited number of TOUrates based on system MC must be inefficient,

TARIFF ALTERNATIVE TO TOU RATES 11


because electricity systems are not perfectly reli-able and utilities need to recover large fixed andsunk costs that are not covered by MC pricingwhen MC is less than average total cost. Location-specific marginal distribution capacity costsfurther challenge the efficiency of TOU energyrates.The difficulties with RTP and TOU energy rates

have led to a number of alternative proposals.These include priority service and demand sub-scription service in generation (e.g. Chao, 1983;Wilson, 1993; Woo, 1990, 1991; Spulber, 1992),capacity reservation for transmission (e.g. Wooet al., 1998), and a Hopkinson tariff with demandsubscription for distribution (e.g. Seeto et al.,1997). Our purpose is to demonstrate the efficacyand construction of a Hopkinson tariff withdemand subscription that is specifically designedto solve the problems of excess capacity, highaverage costs and rates, and the expectation ofhigher future demands for energy. These areproblems faced by the IEC and other integratedelectric utilities globally.

THE HOPKINSON TARIFF WITH DEMAND

SUBSCRIPTION

A Hopkinson tariff with demand subscription is aforward contract that defines service curtailmentconditions and various demand and energycharges. The tariff has three basic components: amonthly connected-load payment, a monthlysubscription payment, and a per kilowatt-hour(kWh) charge for monthly energy consumption.The connected-load payment is designed to

collect the residual revenue requirements forgeneration and T&D. Let a denote the chargeper kW; D denotes the customer’s previouslyrecorded peak demand (in kW) over someparticular period of time (e.g. 12 months). Theconnected-load payment is, then, aD.The residual revenue for each function is its

revenue requirement less the revenue collectedthrough the other charges. When only used tocollect revenue, the connected-load charge doesnot vary by location. This charge is applicable to autility with a stable customer base and little loadgrowth. Equity considerations suggest that anynew load be assessed a connected-load paymentthat might be discounted if such encourages

economic development and reduces othercustomers’ fixed and sunk cost contributions.Connected-load payments of new customersshould be based on estimates of their connectedloads, thus reflecting their ‘fair’ (i.e. pro rata) shareof the utility’s fixed and sunk costs. By the sametoken, they should not be so large as to discouragecustomers from connecting to the network or as togreatly distort decisions about connected-load size.The subscription charge adds together a gen-

eration demand charge, PG, a transmissiondemand charge, PT, and a distribution demandcharge, PD. These charges are per kW-monthrates. The sum of the three rates,PG þ PT þ PD ¼ b, is the unit price for subscrib-ing to a firm service level (FSL) of y kW for onemonth. The total subscription bill is, then, by. Thesubscription demand charges can vary by location,but the generation charge only does so if themarginal energy costs do so.Finally, there is an hourly TOU energy rate that

is set at the marginal energy cost. Denote that rateby g. If total kWh consumption during the billingmonth is C, then the total energy charge is gC.Geographic differences in marginal line losses thatoccur when power is transmitted through thenetwork should reflect the voltage of service. Thesewill result in energy rates that differ by location.Hence, the customer will be billed a total of

B ¼ aDþ byþ gC: ð1Þ

IMPLEMENTING THE TARIFF

Retail customers must pay the connected-loadpayment and make demand subscriptions prior totaking service. Customers taking transmissionservice must also subscribe for generation; thosewho take distribution service must also subscribefor distribution capacity. When subscribing, cus-tomers need to know the probability of curtail-ment and the demand charges ex ante. Woo et al.(1998) describe a posting process for achievingthis.The subscribed demand, or the FSL, cannot be

curtailed during a capacity shortage. The FSLdoes not vary by function, because users make nodistinction between the causes of a curtailment,and the FSL is only firm when sufficient capacity isreserved for all three functions. Subscriptions canbe tailored to the user’s desired degree of

C.K. WOO ET AL.12


reliability. Shortages do not affect users whoseFSLs equal their peak demands. Users with zeroFSLs risk losing their entire load during a short-age. Hence, customers with higher (lower) outagecosts subscribe at higher (lower) FSLs (Chao andWilson, 1987; Woo, 1990; Spulber, 1992) andreceive larger (lower) shares of the limited capa-city.Service begins after a customer makes the

connected-load payment and pays the demandsubscription fee. Capacity shortage being absentcustomers can use as much electricity as theydesire, so long as they are willing to pay the TOUenergy rates. With a capacity shortage, one’s loadcan be limited to the FSL, either at the utility’srequest, enhanced by a severe non-compliancepenalty, or through a remotely activated load-limiting mechanism. Since curtailing all customersto their subscribed demands ordinarily leads toexcess capacity, service might be curtailed either toa select few users (Chao and Wilson, 1987) or to allusers in proportion to their respective FSLs(Spulber, 1992).Reliable systems, such as that of the IEC, have

few locations and sizable marginal transmissioncongestion costs and marginal distribution capa-city costs. Therefore, it is unnecessary to develop afull-blown Hopkinson tariff with location-specificdemand subscription charges. To develop a readilyimplemented rate for the IEC we assume thatmarginal line losses do not vary significantly bylocation. This assumption allows us to use thesystem marginal energy cost in setting the TOUenergy rates.As in the above discussion of a basic three-

component Hopkinson tariff, the bill of customer iin consumer class k in location j begins with aconnected-load payment. That payment is equal tothe product of a network-wide per kW loadcharge, ak, and user i’s historic peak demand, Dikj.The load charge ak is the system revenue require-ment for class k, including a discount for itscurtailable load, net of kWh revenue collectedfrom the tariff’s TOU energy rates, divided by thetotal historic peak demand for this class. Thus,akDikj is that customer’s connected-load payment.The second component is the monthly subscrip-

tion charge per kW of FSL. Let E denote theexpectation operator. The first demand charge,that for generation, is PG=Min.{Monthly sum ofE [Max.(Hourly Outage Cost - Hourly MarginalEnergy Cost, 0)], Per kW-month Generation

Capacity Cost} (Woo, 1990; Chao and Wilson,1987; Spulber, 1992). This charge does not varygeographically if the hourly marginal energycost is constant across locations. The seconddemand charge, that for transmission, isPT=Min.{Monthly sum of E [Hourly MarginalCongestion Cost], Per kW-month TransmissionCapacity Cost Due to New Line Construction}(Nasser, 1999). This charge may vary geogra-phically. The third demand charge, that for dis-tribution, is PD=Min.{Monthly sum of E [Max.(Hourly Outage Costs - Hourly Marginal EnergyCosts, 0)], Per kW-month Distribution CapacityCost} (Ball et al., 1997). This charge, too, mayvary geographically. As previously, the subscrip-tion payment is, then, bikjyikj, wherebikj ¼ PGðikjÞ þ PTðikjÞ þ PDðikjÞ. In as much as bikjis location specific and must be developed on acase-by-case basis, it is not computed for theHopkinson tariffs developed for the IEC, whichare detailed below. Thus, we shall not discuss itfurther here. The details regarding its computa-tion, however, are given in the references cited.The third component, the energy charge, is equal

to the sum of the products of the per kWh rate forseason m, denoted gm, and energy consumptionduring that season, Cm. The most compact formfor the energy charge has three seasons: peak, mid-peak, and off-peak. Thus, the customer’s energycharge is SmgmCm.Ignoring the possibility of rationing, compar-

able to Equation (1) the final bill of customer i incustomer class k in location j will therefore be

Bikj ¼ akDikj þ bikjyikj þX

m

gmCm: ð2Þ

Our recommendation to IEC, however, was toallow a discount to customers whose peakdemands exceeded their FSLs. That discount, inthe amount of the subscription rate, is designed tocompensate the customer when the utility is unableto meet that customer’s energy needs. In short, theutility treats the difference between a customer’speak demand and self-chosen FSL as curtailableload, subject to service interruption at times ofcapacity shortage. Hence, the bill that we recom-mend is

Bikj ¼ a0kDikj � BikjðDikj � yikjÞ þX

m

gmCm: ð3Þ

Here, a0k is a revised connected-load charge thatreflects the increase in residual revenue caused bythe discounting policy.



Other charges and bills are compatible with abasic Hopkinson format, but the one proposed forthe IEC has several virtues. First, the basic unitcharges of ak and gm do not differ by location andconstitute a standard two-part tariff with aconnected load and TOU energy charges. Second,the format is a minor modification of a tariff thatuses only TOU rates, as holds in the IEC. Third,the IEC can selectively decide the locations to begiven a curtailable-load discount. With a highlyreliable system there will be very few of these. Theresulting tariff has few location-specific chargesand can be quite simple. Last, designating a loadto be curtailed through demand subscription canbe viewed as a rate option to the basic tariff.Customers desiring the status quo need notsubscribe.

THE RATE-DESIGN PROCESS

We now want to determine the class revenuerequirements for industrial users, the class energyrates, the residual revenue requirement, and theper kW-month connected-load charge. The designprocess is straightforward, given the relevant data.Specifically, the class energy revenue is the productof the class kWh sales by TOU and TOU ratesequal to the marginal energy costs. The residualrevenue requirement is the difference between theclass revenue requirement and the class energyrevenue. The per kW-month charge is determinedin two steps: (1) divide the residual revenuerequirement by the class total of customer-specifichistoric peak demands, as recorded in the mostrecent 12 months; and (2) divide the result fromthe first step by 12 months. What complicates thisotherwise straightforward process is that there arethree credible alternatives for determining the classrevenue requirement.The alternative we proposed for the IEC uses a

basic Hopkinson tariff. Let c denote the historicpeak demand for the industrial class as a fractionof the system’s total historic peak demands. Thisalternative sets the industrial class revenue require-ment equal to the product of c and the excess ofthe system revenue requirement for generation andT&D over the system kWh revenue.The second alternative, the status quo approach,

sets the class revenue requirement equal to theclass TOU energy sales by TOU at current TOU

energy rates. To apply this approach, one mustfirst determine both the system hourly marginalcosts of generation and T&D, and the MC-basedclass revenue as the annual sum of the products ofthe class hourly kWh sales and the system hourlyMC. The industrial class revenue requirement isthe product of the utility’s total revenue require-ment and the proportion of the total MC-basedclass revenue accounted for by the industrial class.The third alternative is based on the utility’s

capacity planning and recognizes that fixed andsunk costs are incurred to provide reliable servicein the face of growing demand. Since T&Dcapacity is added to serve peak demand growth,it seems reasonable to allocate the T&D require-ment based on a class’s peak demand (Orans et al.,1994). Optimal generation capacity planning,however, dictates that base-load capacity, ratherthan peaking capacity, be added if the additionalbase-load capacity yields sufficient fuel-cost sav-ings to offset the difference in the capacityinstallation costs (Chao, 1983; Wilson, 1993).Since base-load capacity reflects the base-loaddemand that the IEC serves, the revenue require-ment for the base-load generation plants and thepeak-demand generation plants can be appor-tioned to each customer class based on the kWhconsumption of that class. To implement thisapproach, the revenue requirements by functionare determined first. Then, the T&D revenuerequirements are allocated to industrial and non-industrial classes in proportion to their respectivepeak demands. Next, the class-specific base-loadenergy is computed as the product of the minimumload for the class multiplied by the 8760 hcomprising a year. The peak-load energy for theclass is computed as its class total sales for the yearless the base-load energy. Third, the revenuerequirements for the base-load and peak-loadgeneration plants are allocated in proportion tothe base-load and peak-load energies for eachclass. The sum of the revenue requirementsdetermined in the first and third steps is theclass-specific total revenue requirement.

SAMPLE REVENUE ALLOCATION AND

TARIFFS

Table 1 reports the revenue requirements (inmillion of New Israeli Shekels (NIS) with

C.K. WOO ET AL.14


US$1=4 NIS)) for each IEC customer class undereach of the three allocation alternatives. Theunderlying computations are quite complex andrequire detailed accounting data. We thereforelimit our discussion of the individual figures to theconnected-load allocation for residential custo-mers (3570 NIS) and the base-load-peaker alloca-tion for TOU for low voltage (1776 NIS).In the former regard, the connected-load alloca-

tion starts with marginal energy costs for eachclass, calculated as the sum of class energy byTOU period times the marginal energy cost byTOU period. For residential users, this amountedto 1354.4 NIS.The total marginal energy cost for all classes is

less than the generation and transmissions (G&T)revenue requirement of 6673.3 NIS. The residualG&T revenue requirement of 2116.8 NIS isallocated to classes based on their shares of total

non-coincident (NC) demand. At the G&T level,residential NC demand is 50.41% of the total, soresidential is allocated 0.5041(2116.8)=1067.1NIS.The revenue requirements for the other func-

tions are 915.6 NIS for the high-voltage (HV)system and 1051.9 NIS for the low-voltage (LV)system. These requirements are also allocatedbased on shares of total NC demand, with therespective shares being 52.37% and 65.15%.Hence, the residential shares are0.5237(915.6)=479.5 NIS for HV and0.6515(1051.9)=685.3 NIS for LV.Finally, there was a credit for other income that

was allocated to classes based on their overallusage of energy, which amounted to 16.3 NIS forresidential users. Thus, the total allocation toresidential users is 1354.4+1067.1+479.5+685.3�16.3=3570 NIS.

Table 1. Revenue Allocation Impact on Total Group Revenues* in Millions New Israeli (NIS) Shekels(US$1=4 NIS)

Allocation summary Current Connected load allocation Status quo allocation Base-load – peaker allocation

Revenues Revenue % Change Revenue % Change Revenue % Change

Residential 2960 3570 20.6 2960 0.0 2317 �21.7Street lighting 114 170 49.1 114 0.0 252 120.7Small commercial 897 1038 15.7 897 0.0 1160 29.3Bulk (low voltage) 30 19 �36.8 30 0.0 17 �42.8Bulk (high voltage) 416 326 �21.6 416 0.0 328 �21.1TOU for low voltage 1513 1259 �16.8 1513 0.0 1776 17.4TOU for high voltage 2175 1838 �15.5 2175 0.0 2132 �2.0TOU for extra-high voltage 490 368 �24.8 490 0.0 606 23.7

*Excluding consumer revenue.

Table 2. Hopkinson Tariffs by Revenue Allocation Scheme with Energy Rates in Agorot/kWh andConnected-Load Demand Charges in NIS/kW (100 Agorot=1 NIS)

Smr Pk Smr Sh Smr Off Wtr Pk Wtr Sh Wtr Off Spr Pk Spr Sh Spr Off Demandcharge

Energy rates under residual allocation using connected loadTOU for low voltage 18.63 15.18 8.46 19.35 11.87 8.42 16.48 14.24 8.28 16.64TOU for high voltage 18.25 14.87 8.29 18.95 11.63 8.25 16.14 13.95 8.12 11.39TOU for extra-high voltage18.04 14.70 8.20 18.73 11.49 8.15 15.96 13.79 8.02 7.69

Energy rates with no change in group revenue allocationTOU for low voltage 18.63 15.18 8.46 19.35 11.87 8.42 16.48 14.24 8.28 25.93TOU for high voltage 18.25 14.87 8.29 18.95 11.63 8.25 16.14 13.95 8.12 19.11TOU for extra-high voltage18.04 14.70 8.20 18.73 11.49 8.15 15.96 13.79 8.02 20.10

Energy rates under base-load-peaker allocationTOU for low voltage 18.63 15.18 8.46 19.35 11.87 8.42 16.48 14.24 8.28 35.52TOU for high voltage 18.25 14.87 8.29 18.95 11.63 8.25 16.14 13.95 8.12 18.12TOU for extra-high voltage18.04 14.70 8.20 18.73 11.49 8.15 15.96 13.79 8.02 31.97

Notes: Smr=summer; Wtr=winter; Spr=spring and autumn;Pk=peak; Sh=shoulder peak; Off=off peak.



Table

3.

Exam

ple

sofH

opkin

son

Tariffs

(Base

don

the

Follow

ing

Exch

ange

Rate

sper

US$:A

ust

ralia}

1.5

637;C

anada}

1.5

;H

ong

Kong}

7.7

5;

Isra

el}

4.0

8;South

Afr

ica}

5.8

35)

Serviceterritory

Utilityname

Tariff

Demandcharge

(US$/kW-month)

Energycharge

(US$/kWh)

Connected

Load/Ratchet

demandprovision

TOU

Definition

Israel

IEC

Option1:highvoltage

2.79

0.0203–0.0456

Yes

Threeseasons,andthree

TOUperiodperseason

Israel

IEC

Option2:highvoltage

4.68

0.0203–0.0456

Yes


TOUperiodperseason

Israel

IEC

Option3:highvoltage

4.44

0.0203–0.456

Yes


TOUperiodperseason

Highvoltage

0.0810Peak

Peak:7–9a.m;5–8p.m.,

M-F,non-holiday.

Timeofuse

0.0685Shoulder

Shoulder:9a.m.to5p.m.;

8–10p.m.,M-F.

non-holiday

NSWAustralia

ACETW

Demand

4.1perkVa

0.0418Offpeak

No

SouthAfrica

ESKOM

Largecustomer

standard

7.083

0.0128

No

HongKong

ChinaLight&Power

LargepowerTariff

14.45PeakperkVa

0.0657Peak

0.056Offpeak

Yes,50%forsummer

duringprior12months

SummerisMayto

OctoberOffpeakis

11p.m.to9a.m.and

alldaySundayOn-peak

isallotherhours

BritishColumbia,

Canada

B.C.Hydro

1821

2.94perkVa

0.0173

Yes,75%ofprior

NovembertoFebruary,

or50%ofcontract

demand

Kentucky,USA

AmericanElectric

Power

C.I.P.–T.O.D

(Primary)

8.60Peak

2.02Offpeak

0.0122

Yes.60%ofprior

11months

Peak:7a.m.to9a.m.

M.F,excludingholidays

Ohio,USA

OhioPowerCompany

GS-4(primary)

11.470

0.0030

Yes,100%ofprior

11months

California,USA

SouthernCalifornia

Edison

GS-2

5.4Ratchet

7.75Max

fortheMonth

0.0439

Yes,50%ofprior

11months

Michigan,USA

IndianaMichigan

Power

Company

L.G.S(Secondary)

8.920

0.04583Peak

0.03027

Offpeak

Yes,60%to25%

ofcontractcapacity

Peak7a.m.to9p.m.M-F

WestVirginia,USA

AmericanElectric

Power

I.P.(Secondary)

12.51Peak

5.5Offpeak

excess

0.0151

Yes,60%of

(contractcapacity

orprior11months)

Indiana,USA

IndianaMichigan

Power

Company

I.P.(Secondary)

17.159perkVa

0.0114

Yes,60%ofMax

(contract

capacityorprior

11months)

C.K. WOO ET AL.16


Alabama,USA

AlabamaPower

(SouthernCompany)

HLF

10,000

0.0046

Yes,90%ofcontract

capacityorprior

11months

Portland,USA

PortlandGeneral

Electric

Company

32

4.3Peak

0.03755Peak

0.03255Offpeak

Yes,averageof2

highestincurrentand

11priormonths

Peak:6a.m.to

10p.m.M-F

NorthCarolina,

USA

DukePower

13.600

0.0329

Yes,contractminimumor

60%orprior11months

Tennessee,USA

KingsportPower

Company(AEP)

I.P.(Primary)

8.7Peak

2.57Offpeak

excess

0.0230

Yes,60%ofMax

(contractcapacityor

prior11months)

forPeak

Caigary,Canada

Enmax

Directaccesstariff

(temporary)

2.00perkVa

0.012Peak

0.002Offpeak

Yes,100%penalty

forusageabove

contractdemand,

andresetcontractdemand

tohighervalue.

Notethisproposed

ratedoesnotinclude

energyprocurement

payments

Alberta,Canada

TransAlta

6300

2.633

Winter

Summer

0.02290.0199Peak

0.02220.0190

Shoulder0.0193

0.0149Offpeak

Yes,100%ofprior

11months



Residual revenue allocation using the connectedload increases the bills for small users by 16–49%,and decreases the bills to large users by 16–37%.The status quo allocation simply collects theexisting class revenues. To allocate base-load andpeak-load costs based on the base-load andresidual energy use for each class, it was assumedthat 75% of fuel, operating and maintenance costs,and 10% of generation costs, are related to base-load units. All non-generation costs are allocatedbased on peak energy use.For base-load-peak allocation the revenue

requirements were separated into variablegeneration (2959.5 NIS), other generation(2933.3 NIS), and all other revenue requirements(2695.4 NIS). Hence, the base-load share is0.75(2959.5)+0.10(2933.3)+0.0(2694.5)=2512.9NIS and the peak-load share is 6075.2 NIS. Base-load consumption is calculated for each as thelowest average demand in any of the nine TOUperiods (see Table 2). For TOU for LV,this totals 4029 kWh. For all classes the totalbase-load energy is 28167 kWh. Thus, the LVshare is 14.304% of the 2512.9 NIS; or,0.14304(2512.9)=359.5 NIS. The residual energyis 2473 kWh for LV and 10604 kWh for all classes.Thus, the TOU LV share of residual energy is100(2473/10604)=23.321%, and the peak-loadrevenue requirement is 0.23321(6075.2)=1416.8NIS. The total TOU revenue requirement for LVis therefore 359.5+1416.8=1776 NIS.Wide changes in revenue responsibility result.

The bulk LV users are clear winners, and the smallusers are clear losers whose bills increase from29% to 121%. Table 2 reports three sets of sampletariffs corresponding to the choice of class-revenueallocation method. Under the status quo, thedemand charge for each class recovers the residualrevenue not recovered by its TOU energy rates setat the marginal energy costs of the IEC.Table 3 compares the three IEC-based high-

voltage Hopkinson tariff options of Table 2,translated into $US, with a convenience sampleof readily accessible Hopkinson tariffs outsideof Israel. This is not necessarily a representativesample, but except for ESKOM and ACETWthe data show that many utilities use some formof connected-load charge based on historicpeak demand. These data make it apparent thatthe Hopkinson tariffs that we propose for the IECare not at odds with what is taking placeelsewhere.

CONCLUSIONS

TOU energy rates can cause uneconomicbypass and do not convey the price signals toinduce efficient electricity consumption. Theinefficiency of TOU rates is compounded whena utility cannot impose a demand-curtailingprice premium during periods of excess demand,and instead must ration the limited capacity.The capacity allocation can be made efficientonly when customers have revealed theirpreferences for system reliability, but theirresponses to the TOU rates are uninformative inthis regard.The IEC faces precisely this problem, and

the solution that we have proposed is a Hopkinsontariff with demand subscription. We have ex-plored three variations on the Hopkinson theme.Our recommended approach apportions thesystem residual revenue requirement betweencustomer groups in accordance with theirhistoric peak demands. The procedure we havedetailed for determining the tariff under eachvariation is readily implemented, and the resultingtariffs are comparable with those elsewhere in theworld.Indeed, the IEC management adopted the first

Hopkinson tariff with the primary goals of stablecollection of fixed costs, marginal-cost pricing ofenergy, efficient rationing of limited capacity, andpreventing uneconomic bypass. The same manage-ment rejected the other two Hopkinson tariffsbecause of their relatively high demand chargesthat cause significant bill changes for somecustomers with relatively low-energy consumptionbut high peak demands.If also adopted by the Israel Public Utilities

Authority (PUA), the Hopkinson tariff proposalof the IEC would have introduced various demandcharges for the large industrial users. The PUA,however, rejected the proposal, because of itsdesire to promote on-site generation owned bylarge users and its misguided belief, which we andthe IEC management would reject, that the TOUtariff structure is in some sense superior to that ofthe Hopkinson tariff. Nevertheless, it is our hopeand belief that the procedure and the results will beuseful to other utilities that are weighing aHopkinson tariff alternative to their currentenergy rate packages, and that they will encouragestill other utilities to consider the Hopkinsonoption.

C.K. WOO ET AL.18


Acknowledgements

The authors gratefully acknowledge the cooperation andsupport of the Israel Electric Corporation (IEC), and itswillingness to release the data contained herein for publicconsumption. Horowitz acknowledges the support of a SummerFaculty Research Grant from the Public Utility ResearchCenter (PURC) of the Warrington College of BusinessAdministration of the University of Florida. The positionstaken in the paper are those of the authors and do notnecessarily reflect the views of either the IEC or PURC.

REFERENCES

Acton JP. 1982. An evaluation of economists’ influenceon electric utility rate reforms. American EconomicReview 72: 114–119.

Ball G, Lloyd-Zannetti D, Horii B, Birch D, Hicks RE,Lively H. 1997. Integrated local transmission anddistribution planning using customer outage costs.Energy Journal, DR Special Issue 137–160.

Bohn RE, Caramanis MC, Schweppe FC. 1984. Optimalpricing in electrical networks over space and time.Rand Journal of Economics 15: 360–376.

Chao HP. 1983. Peak load pricing and capacity planningwith demand and supply uncertainty. Bell Journal ofEconomics 14: 179–109.

Chao HP, Wilson R. 1987. Priority services: pricing,investment and market organization. American Eco-nomic Review 77: 899–916.

Crew MA and Kliendorfer PR. 1976. Peak load pricingwith a diverse technology. Bell Journal of Economics 7:207–231.

Graff J. 1967. Theoretical Welfare Economics. Cam-bridge University Press: Cambridge.

Hogan WW. 1992. Contract networks for electric powertransmission. Journal of Regulatory Economics 4: 211–242.

Joskow PL. 1976. Contributions to the theory ofmarginal cost pricing. Bell Journal of Economics 7:197–206.

Joskow PL. 1997. Restructuring, competition andregulatory in the US electricity sector. Journal ofEconomic Perspectives 11: 119–138.

Munasinghe M. 1990. Electric Power Economics. Butter-worths: London.

Nasser TO. 1999. Regulation of a power transmissioncompany. Working paper. McKinsey and Company:Washington, DC.

Orans R, Woo CK, Horii B. 1994. Targeting demandside management for electricity transmission anddistribution benefits. Managerial and Decision Eco-nomics 15: 169–175.

Seeto D, Woo CK, Horowitz I. 1997. Time-of-use ratesvs. Hopkinson tariff redux: An analysis of the choiceof rate structures in a regulated electricity distributioncompany. Energy Economics 19: 169–185.

Spulber DF. 1992. Optimal nonlinear pricing andcontingent contracts. International Economic Review33: 747–772.

Spulber DF. 1989. Regulation and Markets. The MITPress: Cambridge MA.

Vardi J, Zahavi J, Avi-Itzhak B. 1977. Variable loadpricing in the face of loss of load probability. BellJournal of Economics 8: 270–288.

Wilson R. 1993. Nonlinear Pricing. Oxford UniversityPress: New York City.

Woo CK. 1991. Capacity rationing and fixed costcollection. Energy Journal 12: 153–164.

Woo CK. 1990. Efficient electricity pricing with self-rationing. Journal of Regulatory Economics 2: 69–81.

Woo CK, Horowitz I, Martin J. 1998. Reliabilitydifferentiation of electricity transmission. Journal ofRegulatory Economics 13: 277–292.



the hopkinson tariff alternative to tou rates in the israel electric corporation

Documents