Debates on restructuring the U.S. electricityindustry are often about the degree to whichmarket relationships should replace transactionsthat formerly took place within regulated, verti-cally integrated utilities. Markets for the pur-chase of energy by vertically unintegrated distri-bution utilities are clearly feasible, but verticaldeintegration of existing systems may eliminatesome operational and reliability benefits that areimportant in light of the unique characteristicsof electricity.

Politicians and policy analysts have almosttotally disregarded a large body of academic lit-erature regarding the efficiencies that are gainedthrough vertical integration in the electricity sec-tor. At the same time, those parties have enthusi-astically embraced other studies that purport toestimate the benefits of switching to a so-calledrestructured regime consisting of independentgeneration and integrated transmission and dis-tribution. The result has been the passage of elec-

tricity utility restructuring laws that may createproduction inefficiencies that shrink the netbenefits of any move toward market provision ofpower supplies.

A review of the debate surrounding electricutility restructuring in California—the first stateto embrace restructuring—reveals that legislatorsand regulators regarded vertical integration pri-marily as a tool that incumbent utilities coulduse to perpetuate their market power. They thusdisregarded the benefits that might accrue fromvertical integration and used the force of regula-tion to encourage the sale of generating plants toindependent power producers. The idea was tocreate a competitive market structure in the elec-tricity generation sector. Unfortunately, the costsassociated with this experiment in Californiaand elsewhere have yet to be compared with ben-efits in any economically meaningful way.

A proper comparison of the two suggests thatrestructuring is presently off course.

Vertical Integration and the Restructuringof the U.S. Electricity Industry

by Robert J. Michaels

_____________________________________________________________________________________________________

Robert J. Michaels is a professor in the Department of Economics at California State University, Fullerton, and anadjunct scholar at the Cato Institute.

Executive Summary

No. 572 July 13, 2006�������

PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1PA Masthead.indd 1 2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM2/9/06 2:08:34 PM

Introduction: VerticalIntegration and Electricity

The past 30 years have transformed theeconomic theory of the business firm.1 In-stead of assuming that the scope of a firm’sactivities is fixed, economists now treat itsboundaries as matters of choice. The eco-nomics of organization asks such questionsas whether the firm should purchase its rawmaterials in markets or produce them in afacility that it owns and whether its productshould be sold by salaried employees or byindependent retailers. A rational decisionabout producing raw material or buying itfrom a third party requires that the firm con-sider alternative ways to hedge price uncer-tainty and ensure deliveries, its ability tocoordinate production and use of the rawmaterial in question, and its competence inmanaging the dissimilar activities of rawmaterial and output production.

Almost since their origin, electric utilitieshave been vertically integrated, with genera-tion, transmission, and distribution com-bined in a single firm. The operational ration-ale for vertical integration was largely relatedto the physics of electricity delivery. In orderfor electricity to be transmitted from the gen-erator to the consumer, electricity supply andelectricity demand must remain in precisebalance at every instant over a wide area. Thatchallenging task requires a central authorityto govern both the supply of and the demandfor electricity along the power grid.2

There were also economic incentives for ver-tical integration. Low-cost production requiresthe simultaneous optimization of both gener-ator dispatch and transmission capacity. Long-run efficiency requires the coordination ofinvestment decisions at all stages of the chainfrom generators to low-voltage distributionlines.

In the mid-1970s scholars first argued thatgeneration could be organized as a competi-tive market.3 Superficially, the case for verticaldeintegration is clear: changes in technologyhave turned generation into a potentially com-petitive market, and efficiency would be

enhanced if that market were allowed to oper-ate. Transmission and distribution, however,remain most efficiently organized as monopo-lies, and those activities should continue to beregulated.

In reality, the case for deintegration isproblematic. Its advocates often argue frominappropriate analogies with other industriesor nations and disregard a large body ofeconometric research on the efficiencies ofvertically integrated utilities.4 If both integra-tion and competitive markets have desirableeconomic properties, industry restructuringshould focus on facilitating the most effi-cient mix of the two. Unfortunately, the valueof integration between generation and trans-mission has been conspicuously unexplored,and thus restructuring threatens to produceinstitutions that foreclose the realization ofimportant efficiencies.

The economically efficient degree of dein-tegration is not obvious. Vertical deintegra-tion could remedy discrimination againstcompetitors by an integrated utility, but socould a policy that requires integrated utili-ties that transmit their own power to honorrequests from others to use their lines on thesame terms. Note that the latter remedy doesnot change the organizational structure ofthe company whereas the former does justthat. Of course, there may not be a problemto remedy in the first place—favoring genera-tors that one owns may be efficient.

Various gradations of deintegration havebeen proposed. The least extreme form man-dates a functional separation of generation,transmission, and distribution into differentadministrative divisions within the firm. A stepbeyond that lies structural separation, whichcreates subsidiaries that must deal at arm’slength and in a nondiscriminatory mannerwith each other. A step beyond that lies the pre-ferred policy of the Federal Energy RegulatoryCommission, which encourages an opera-tional separation of generation and transmis-sion services and a surrender of the control ofthe power grid to a nonprofit, public-privateindependent system operator (ISO) or regionaltransmission organization (RTO).5 The most

2

The case for deintegration is

problematic. Its advocates

often argue from inappropriateanalogies with

other industriesor nations and

disregard a large body ofeconometric

research on theefficiencies of

vertically integrated

utilities.

extreme deintegration breaks generation andtransmission into separate corporations, asoccurred in the United Kingdom.

To better understand vertical integrationand electricity markets, this paper summarizesthe economics of vertical integration and itsapplication to electricity. I then confront therecord of economic and legal thought onrestructuring with the econometric evidenceconcerning integration. That research almostunanimously concludes that vertical integra-tion is an efficient form of organization forelectric utilities. Research on the role of com-petitive markets in electricity has been lesscomplete and often less rigorous.

The Economics of VerticallyIntegrated Utilities

Economic activity can be organized inthree basic ways: markets, contracts, or verti-cal integration. The merits of each vary de-pending on the nature of the enterprise.

Markets are places or institutions wherebuyers and sellers compare their valuationsof goods. Prices are discovered as informa-tion about offers and other market condi-tions becomes public. The cost of using amarket instead of contracts or vertical inte-gration will be lower the easier it is (1) to con-tact potential counterparties, (2) to comparetheir offers, and (3) to perform the transac-tion, whose costs may include the determina-tion of product quality and buyer creditwor-thiness.

Buyers and sellers are more likely to usemarkets to exchange relatively standardizedgoods in situations in which informationabout their characteristics and the character-istics of counterparties is easy to obtain. Thecost of using markets is also affected by thecost associated with changing buyer-sellerrelationships. Markets characterized by sub-stitutable products and uncommitted buyersand sellers work smoothly. For instance, aseller who stops dealing with buyer A andstarts dealing with buyer B does not need tomake any investments specific to the rela-

tionship with B or lose any that were specificto the relationship with A.

Contracts will supersede markets, howev-er, when a nonstandardized product is par-ticularly valuable, when durable and specificinvestments are necessary to realize thatvalue, or when the allocation of risk the par-ties prefer cannot be obtained in the market.For instance, assume a buyer wants a fuelsupply with flexible deliveries, which requiresthat the supplier construct a specialized stor-age facility the cost of which is unrecoverableif the buyer stops taking fuel from the firm(there are no comparable buyers nearby). Thebuyer gets value only if the facility is built,and the seller builds it only if the buyer com-mits to a long relationship. A contractbetween them may prohibit the buyer fromprocuring fuel elsewhere and the supplierfrom selling it to others when the buyerexpects delivery.

Vertical integration is an efficient organi-zational choice if (1) assets are highly specificto a given use or location, (2) assets are uti-lized in activities that must be coordinated,and (3) if the best uses of an asset depend oncontingencies that are hard to predict.6

Whether governance of a relationship will beby integration, markets, or contracts dependson the benefits and costs of each option, possi-bly including the cost of changeover from oneto another of the three. Markets may becomemore attractive if they offer better alternativesthan the buyer could self-provide at the samecost or if the cost of using markets decreases(Internet access, for instance, allows quickworldwide shopping with lower risks of nonde-livery). The benefits of contracting may likewiserise (health insurance is more valuable to me,for instance, if medicine is more advanced) if itscosts become lower (without standardizedautomobile insurance, for instance, liabilityrisks are so high that I choose not to drive).Integration can become a more attractive orga-nizational form (if the market for a raw materi-al input, for example, becomes more unstableand the costs of writing contracts to managethat instability are prohibitive). It can alsobecome less attractive (for example, if growth of

3

Economic activitycan be organizedin three basicways: markets,contracts, or vertical integration. Themerits of eachvary dependingon the nature ofthe enterprise.

an industry implies that external specialist sup-pliers can make a component more cheaplythan users can if they do the job themselves).

ElectricityGeneration, transmission, and distribu-

tion of electricity are highly interdependent.With minor exceptions, power cannot bestored and must be produced the instant it isconsumed. Failure of generation to meetdemand will result in blackouts. The demandfor electricity has both random elements andpredictable hourly and seasonal characteris-tics. Efficient response to both predictableand unpredictable events requires centralizedoperation of generation and transmission.7

Electricity can be produced and deliveredeconomically only if highly specialized assetsare in place. Distribution lines must physicallyreach users, and transmission lines must coverthe distance between distribution lines andgeneration. For reliability, some generatorsmust be close to consumers while for econom-ical production others may be more distant.Investment in generation and transmission isa long and costly process, and, once in place,the equipment cannot be cheaply redeployedto some other location or use.

Such characteristics suggest contract orvertical integration as the likely industrialorganization for electricity. Contracts governsome vertical relationships in electricity, forexample, between a utility and an indepen-dent power producer or between a transmis-sion-owning utility and a small municipalutility that depends on the other’s lines fordeliveries. But even if a highly specific asset isunder contract, its owner may act oppor-tunistically (e.g., a generator may attempt toovercharge the utility if it knows that refusalto cooperate will cause a blackout). The utili-ty may, of course, sue the generator, but itsprobability of success will depend on how acourt interprets the details of a complex con-tract. A contract will be more difficult tonegotiate and enforce if there is uncertaintyabout when the utility will require powerfrom the generator and how much it willrequire.

Several attributes of electrical servicemake vertical integration an efficient organi-zational choice.8

• Vertical Integration and the Hold-Up Problem.The dedicated nature of electricity assetsimplies that generators need transmissionto get their product to consumers andthat transmission assets need generators.Thus either side can “hold up” the other.9

That is, once assets are in place and inde-pendently owned, transmitters mightrefuse to pay anything above a generatingplant’s marginal costs and generatingfirms might accept such demands.Generation requires transmission to reachconsumers, the power plant’s assets can-not be dedicated to other uses, and theplant itself cannot move to a more lucra-tive service territory. A solution to this pos-sibility is vertical integration, which endsthe fight between generation and trans-mission over the division of the economicsurplus from their interaction.

• Coordination of Investments in a ComplexSystem. Vertical integration facilitates thecoordination of highly specific and inter-dependent investments in generationand transmission. The two are substi-tutes in the production of bulk power(since transmission allows access to alarger universe of power plants) and com-plements in the delivery of electricityfrom generators to consumers. Any newfacility affects the economic value of allother facilities on the system, and anorganization that owns most such facili-ties may also be most likely to under-stand their interactions and invest opti-mally in them.

• Risk Reduction and Risk Management. A ver-tically integrated utility may have less riskthan one that operates under long-termcontracts with generators. The probabili-ty of a blackout will be lower with coordi-nated operation of a large system. Great-er certainty may lower the company’scost of capital, potentially important insuch a capital-intensive industry.

4

Its characteristicssuggest contract

or vertical integration as the

likely industrialorganization for

electricity.

Electricity’s Changing EnvironmentDo changes in the fundamental character-

istics of electricity production warrant a recon-sideration of vertical integration as the defaultorganizational design? To better understandvertical changes in the electricity sector, firstconsider the unchanging interface betweentransmission and distribution, where restruc-turing has had no substantial impact. Both arehighly specific assets, restricted as to locationand transferable to nonelectrical uses only athigh cost. Competitive duplication of either iscostly and sacrifices the scale economies anddiminished line losses of larger conductors.The process of transforming voltages acrossthe transmission-distribution interface is littlechanged, and second-by-second coordinationof flows across the grid remains necessary.Vertical integration between transmission anddistribution may in fact have become morevaluable if the emergence of wholesale mar-kets has increased uncertainty about fluctuat-ing flows across the interface.10

Vertical deintegration between generationand transmission is more economicallydefensible. FERC’s “open access” rules (seebelow) require transmission owners to carrythe output of independent power producers(IPPs) in a nondiscriminatory manner. IPPsnow make up 45 percent of U.S. generationcapacity.11 Although the assets are highly spe-cific and require coordination, other attrib-utes of electric energy may make marketsdesirable. It is a homogeneous commodity, itcan be centrally traded, and bilateral con-tracts are common between buyers and sell-ers who choose not to use the centralexchange. Market size is growing withFERC’s RTO initiatives, and the technologiesof long-distance transmission and wide-areasystem controls are improving. Finally, econ-omists and others have devised new marketinstitutions to facilitate trade. Some short-term markets operate under “two-settle-ment” systems for day-ahead and real-timetransactions; ancillary services (load follow-ing and reserve generation) can also be trad-ed, and some grids use Locational MarginalPricing of transmission.12

Such facts imply that markets are moredesirable today than in the past. They do not,however, by themselves, imply that verticaldeintegration is warranted because they donot consider its costs. Deintegration’s netvalue also depends on the benefits of integra-tion that will be forgone. So, the policy ques-tion is “What is the optimal degree of deinte-gration?” American restructuring, however,has not approached the problem this waydespite the availability of some relevantresearch findings.

Restructuring andEconomics

Until the 1960s U.S. courts often ruledthat vertical mergers by large firms couldextend market power at one stage of produc-tion into otherwise competitive stages.13 Inthe 1960s and 1970s economists came to theconclusion that the judicial view was general-ly incorrect.14 First, a monopolist in one stageof a vertical chain (diamonds) does not needto merge with or acquire other competitivebusinesses in the chain (jewelry stores) tocapture all the monopoly gains possible. Themore efficiently the diamonds are retailed,the higher the wholesale price the monopo-list can charge and the higher its profit.Second, if vertical mergers or restrictions can-not increase a seller’s market power, theirprobable purpose is to turn the firm into abetter competitor by reducing the transac-tion costs between stages of production.

Regulatory evasion provides a potentialexception to this benign view of vertical rela-tionships. In 1973 the Supreme Court decid-ed Otter Tail Power v. U.S., holding that a verti-cally integrated utility with market power intransmission had violated the antitrust lawsby refusing municipal distribution utilitiesthe use of its lines to deliver inexpensivepower they had purchased for themselves.15

Because the municipal utilities had no trans-mission alternatives, they had to take higher-cost service from Otter Tail. The Court con-cluded that the vertically integrated utility

5

The facts implythat markets aremore desirabletoday than in thepast. They donot, however, bythemselves, implythat vertical deintegration iswarrantedbecause they donot consider its costs.

company was attempting to monopolize dis-tribution in its area, when competition forfranchises was in fact possible. It further ruledthat the government could order Otter Tail totransmit power to the towns if necessary.

Guided by the Otter Tail ruling, scholarsbegan to make the case for the vertical deinte-gration of electricity. Over the 1970s and 1980slawyers and economists produced several pro-posals for deintegration, some still citedtoday.16 They differ in numerous details, but allbegin by considering long-distance transmis-sion and local distribution as natural monopo-lies. All of them want greater competitionbetween corporate utilities and local govern-ments for franchises to distribute power.

The reasons for encouraging franchisecompetition are unclear. Distribution is astandardized technology the costs of which,in most areas, are under 15 percent of thedelivered cost of power, and few if any realsavings would result if a small municipalitywere to take over the operation of lines with-in its boundaries.17 The authors of thesestudies also intended to facilitate the growthof energy markets by introducing competi-tion by nonutilities for contracts with dis-tributors and shared participation in newprojects. At the time of their writings, howev-er, continuing technological progress in largepower plants and other factors actually madegeneration an unpromising market.18

The radical deintegrations that theseauthors proposed were based on a belief thateven relatively small market benefits wereworth pursuing since they could be obtainedby the simple (so the authors believed) step ofbreaking up corporate utilities. In particular,they unanimously asserted without proof thatvertical deintegration would produce few if anyefficiency losses:

The reduction of competition at the dis-tribution stage might be acceptable ifvertical integration made utilities moreefficient. That, however, is not the case.Utilities strive to integrate forward toobtain a dependable supply of bulkpower. But vertical integration does not

significantly reduce the cost of opera-tion at any stage of the industry.19

Some tried to prove their cases by analogy:

[I]n other industries, production hasnot, for the most part, been integratedwith distribution. There is today nocompelling reason for such integrationin electric power either.20

References such as these continue to guidemany policymakers. They do so despite thefact that, shortly after these studies were pub-lished, economists began attempting to esti-mate the benefits of vertical integration.Almost uniformly, their findings would con-tradict the claims these studies made aboutdeintegration.

Econometric Studies of Vertical IntegrationThere are at least 11 published studies that

investigate the relationship between the verticalintegration of electricity generation, transmis-sion, and distribution and utilities’ costs.21

They cover the United States and Japan, both ofwhich are served by regulated, vertically inte-grated corporate utilities with assigned territo-ries. Their data cover subsets of years between1970 and 1997, all taken from utilities’ annualfilings with regulatory agencies under stan-dardized reporting systems.22

The Appendix to this paper summarizestheir methods and findings.23 The only studyto find that vertical integration worsenedeconomic efficiency is the most questionableon several grounds.24 Only one author findsno statistically significant economic advan-tages to integrated generation, transmission,and distribution.25 Because the authors uti-lize different samples and research tech-niques, it is impossible to compare theirnumerical estimates of the savings from inte-gration, but with the exceptions mentionedabove they are all significantly positive.26

Vertical Integration, Fuel Supplies, andGenerator Performance

Research on electricity generation, trans-

6

There are at least11 published

studies that investigate the

relationshipbetween the

vertical integra-tion of electricity

generation, transmission, and

distribution andutilities’ costs.

The only study tofind that vertical

integration worsened economic

efficiency is themost question-able on several

grounds.

mission, and distribution primarily concernsthe effects of vertical integration on produc-tion cost. Research on the integration of fuelsupplies and the outsourcing of generatorengineering and construction, by contrast, isabout the transactions costs associated withvertical integration.27 For example, MIT econ-omist Paul Joskow has examined the factorsthat may incline a coal-fired generator to enterinto long-term contracts for coal from miningcompanies rather than to purchase coal froma third party.28 The more particular the needsof the power plant or the coal producer, themore likely—in theory—that coal productionand consumption will be integrated withinthe same firm. Greater specificity (in either thegenerator or the coal supply) should be morelikely to entail integration between the mineand the utility. His findings are generally con-sistent with this theory:

• Only a small amount of coal is traded inspot markets, and trades are primarilyin the East, where there are more minesand more generators than in the West.29

• Mine-mouth plants are more oftendesigned to burn a specific type of coalthan are non-mine-mouth plants, andmore likely to be integrated with utility-owned mines.30

• Coal supply contracts are on average 12years longer for unintegrated mine-mouth plants than for plants not locat-ed there. Longer-term contracts tend toexist for the generator’s full require-ments and they contain complex mar-ket-based price adjustment terms.31

• Long-term contracts are more commonin the West, where a plant must burnlow-sulfur coal compatible with thedetails of a generator’s engineering, thanthey are in the East, which has numerousinterchangeable coal sources.32

Regulation can change the costs and bene-fits of the integration of mining and powergeneration. Economist John Filer found thatthe most important factor influencing thedecision to integrate is the presence of a fuel

cost adjustment provision in rates, whichmight allow utilities to opportunistically over-state the costs of mining to obtain higherrates.33 Likewise, economist John Gonzalesfound that productivity is lower in utility-owned coal mines than in independent ones.Some utility-owned mines, however, areunregulated, and their productivity is thesame as that of independent mines.34 By con-trast, economist Joe Kerkvliet found that verti-cally integrated mines were more technicallyefficient than unintegrated ones. With a givenmix of inputs, Kerkvliet argued, an integratedmine would produce more than an uninte-grated one, other things being equal.35

Generator performance provides a moreindirect but less conclusive test of the trans-action-cost model. Economists Paul Joskowand Richard Schmalensee examined theoperational heat rates and unit availability oflow-pressure “subcritical” and high-pressure“supercritical” coal-fired units. Their regres-sions included indicator variables for thefour utilities that were the largest owners ofthese plants, which performed their owndesign and engineering work; the other utili-ties outsourced those functions. For bothtypes of generators, two of the four integrat-ed owners enjoyed significantly better avail-ability and heat rates than average, while theother two companies were at the average.36

Vertical Integration and Reliability There are no publicly available studies that

estimate the actual or potential impact of ver-tical deintegration on reliability. Noteworthyoutages are rare in the United States, and reli-ability analysts are justifiably more interestedin their proximate causes (equipment mal-function, trees touching lines, etc.) than intheir relationships to changes in industrystructure.

Structural changes in the industry such asvertical deintegration, the formation of RTOs,the growth of existing wholesale markets, anddirect access of final customers to nonutilitysuppliers could affect system reliability. All ofthose changes make operations more complexand possibly riskier, but there is no clear way

7

There are no publicly availablestudies that estimate the actual or potential impactof vertical deintegration on reliability.

to apportion the causation of outages amongthem. Concerns have also been expressed thatmore extensive restructuring will adverselyaffect investment in transmission because costrecovery may be at risk if unforeseen marketchanges leave a new line underutilized. Theseeffects could worsen already-existing prob-lems that have been caused by 20 years of lowtransmission investment.37

The North American Electric ReliabilityCouncil has for some time been concernedabout the effects of restructuring on reliabil-ity.38 Its annual reliability assessments dis-cuss the consequences in general terms:

The responsibility for coordinating oper-ations between generating plants andtransmission systems traditionally hasbeen assigned to the utility transmissionsystem operators and system planners.Administrative separation [i.e., verticaldeintegration of generation and trans-mission] as well as the growing numberof [independent power producers] de-mands a more standardized and formalunderstanding of the bulk electric gridcontrol and reliability criteria by all.39

NERC also sees inefficiencies resulting fromuncoordinated planning and investmentdecisions:

The close coordination of generationand transmission planning is diminish-ing as vertically integrated utilities divesttheir generation assets and most newgeneration is being proposed and devel-oped by independent power producers.Once new generation is announced thenecessary transmission additions to sup-port it must still be designed, coordinat-ed with other generation and transmis-sion, and constructed. Since these activi-ties are no longer carried out within asingle organization, more time will needto be allowed to coordinate and performthese tasks to properly integrate the newgeneration to ensure reliability before itcan come into service.40

NERC’s concerns about operating diffi-culties may be justified, but its reports do notdiscuss any actual outages or operating crisesthat it believes were caused by vertical deinte-gration or increased reliance on markets. Theorganization’s data, however, show increasesof several hundred percent between 1998 and2004 in emergencies that required the use ofextraordinary procedures for redispatch andcurtailment known as Transmission LoadingRelief.41

The increase in TLR probably has multiplecauses. There has certainly been increasedstress on the transmission system due to lowinvestment in new and upgraded facilities.NERC also blames changes in the pattern ofgrid use, as systems designed for predictabletransfers between utility-owned generationand captive loads have been required toaccommodate the less predictable flow pat-terns that result from market transactions.42

Some observers worry that a vertically inte-grated utility can exercise market power if itcalls for TLR in a nonemergency situation.That’s because TLR protocols on capacityreservation and service curtailment can attimes give priority to the transmission owner’sown generation over transactions by competi-tors that use the same lines. Attorney DianaMoss concludes that determining whetheremergencies or market power explain TLRgrowth will require further research.43 If verti-cally integrated utilities actually do invokeTLR for strategic reasons, however, it willmore likely be as a consequence of the particu-lar TLR rules in effect than of vertical integra-tion itself.

Moss’s work more generally addressespotential conflicts between competition andreliability that may have been aggravated bydeintegration and market growth. She recog-nizes, however, that inefficiency and threatsto reliability can also result from the absenceof market forces. For example, if transmis-sion is sold at regulated rates that recoveraverage cost rather than priced in a market toreflect its scarcity, there may be little invest-ment in new lines, and those that are actual-ly built may be inefficiently located.44 By con-

8

The case for deintegration andrestructuring has

implicitly beenfounded on a

belief that thesavings and

other benefitsobtainable from

markets exceedthose that are

associated withvertical

integration.

trast, NERC appears to believe that engineer-ing standards should generally take prece-dence over market outcomes:

[Due to vertical deintegration] genera-tion additions cannot be planned in anintegrated fashion with transmissionexpansion, resulting in sub-optimaltransmission expansion in some areas.Generation is not locating close todemand centers, but rather is locatingclose to a fuel supply, adequate coolingwater, and a transmission line inter-connection.45

The interrelationship between investmentsin generation and transmission leads NERCto favor planning by utilities over reliance onmarkets. Beyond this statement, however,NERC provides no discussion about whichdecisions it thinks are best made in markets.In electricity, the choice between planning andmarkets is a matter of degree: vertical integra-tion and centralized planning yield reductionsin operating costs, but markets may at timesprovide other sources of cost reduction bene-fits that outweigh the losses from less com-prehensive planning.

An alternative perspective that empha-sizes the benefits of independent transmis-sion, however, has recently surfaced at FERC,which is considering several proposals toallow “participant funding” of additions toRTO grids by generators and others.46 One ofthe most important proposals was a 2002application to form SEtrans, an RTO in theSoutheast, which envisioned participantfunding as one of two types of transmissioninvestment.47 The SEtrans applicants arguedthat lines linking new generators to the gridshould generally be funded by the new par-ticipants in the system because the benefitsaccrued primarily to those parties. Someother lines (often planned by the RTO)would bring more general benefits in theforms of increased reliability and improvedaccess to markets. Their costs would be pro-rated (“rolled in”) according to agreed-uponformulas.

SEtrans had good reason to propose par-ticipant funding. Its area contained fuel sup-plies and generator sites that might producepower for distant consumers, but those gen-erators would add little to system reliability.Further, some people argue that mandatorycost sharing might allow inefficient trans-mission investments that would not be madeif beneficiaries had to bear their full costs.

One representative of a large utility inSEtrans saw the failures of past planning asfurther reason to institute participant fund-ing. In his view, the ability of grid planners tomake efficient long-run choices is doubtful.Seeing that today’s industry faces unprece-dented uncertainty about load growth, mar-ket development, new technologies, and fuelprices, he said, “We cannot optimally planthe transmission grid any longer, and weshould not try and pretend that we can.”48

That statement is well-founded. A centrallyplanned RTO must choose which lines tobuild or upgrade from numerous alterna-tives, each of which might be consistent withreliability. Participant funding leaves thosedecisions to the market, where pressure tomake efficient choices may be greater. Linesthat create benefits for the entire regionmight still best remain under the ownershipof vertically integrated utilities.

Harvard economist William Hogan, how-ever, has raised concerns about a participantfunding scheme. He observed that a “free-rider” problem might well arise if some lineswere funded by new participants while otherswere funded by all users as a group. That’sbecause an entity that would normally pro-pose a participant-funded line may prefer towait until its absence begins to affect reliabil-ity, at which time the RTO might authorizecollective funding.49 Thus far, the search for aclear distinction between lines that should beparticipant funded and those needed for reli-ability has produced no operational criteriafor making that distinction.50

SummaryThe movement to restructure electricity

began with generalities about the desirability

9

The NorthAmerican ElectricReliabilityCouncil has forsome time beenconcerned aboutthe effects ofrestructuring onreliability.

of markets coupled with claims that verticalintegration in utilities was either unimpor-tant or that its effects could easily be dupli-cated in markets. The econometric evidence,however, makes clear that there are substan-tial economic advantages associated with ver-tical integration.

The case for deintegration and restructur-ing has implicitly been founded on a beliefthat the savings and other benefits obtain-able from markets exceed those that are asso-ciated with vertical integration. It is quitepossible that utilities invest or operate ineffi-ciently. Rate of return regulation may inducethem to overcapitalize or to extend them-selves excessively into unregulated business-es. As regulated monopolies, they may feelless pressure to cut costs than do firms incompetitive markets.

If rate of return regulation is retained dur-ing the process of vertical deintegration,however, deintegration by itself is unlikely toproduce more efficient operation or invest-ment. Performance-based or price cap regula-tions are less drastic alternatives to deinte-gration and have shown some success inpractice. Moreover, the latter reforms are eas-ily reversible—deintegration is much moredifficult to reverse.

If markets for generation are superior tovertically integrated operations, can their ben-efits be obtained by policies that also maintainthe benefits of integration? Such questionshave gone largely unasked as the U.S. industryrestructures.

Deintegration in California

The campaign for vertical deintegrationwent into high gear on April 20, 1994, whenthe California Public Utilities Commissioninstituted a rulemaking on electricity.51 Its rad-ical proposal to allow consumers “directaccess” to suppliers of their choice generatedvolumes of testimony from interest groups,most of which is no longer available on theInternet.52 The CPUC held hearings in 1994and early 1995 and issued its initial order in

December 1995. The California legislatureenacted that order into law in September 1996and established the California Power Exchange(PX)—the market in which wholesale powerwas to be traded—and the Independent SystemOperators to facilitate the new regulatoryregime. The newly created electricity marketopened for business on April 1, 1998.53

Perhaps the most frequently expressedopinion on vertical integration before theCPUC was that it was undesirable in a regu-lated world. In a deregulated system, verticalintegration was thought to facilitate the exer-cise of market power by utilities. Testifyingfor municipal utilities, economics professorWilliam Shepherd either rejected or wasunaware of the research discussed in the pre-vious section of this paper. He claimed that,in order to achieve economies of scale andscope, “[t]here may need to be separation ofthe core functions into distinct entities.”54

Shepherd provided no evidence, however,that integrated utilities failed to exhibiteconomies of scale and scope. Others pro-posed radical restructurings along the samelines, not necessarily restricted to California.They included energy law professor RichardPierce, who failed to mention any possiblecosts of deintegration in a scheme to separategeneration from transmission and transmis-sion from distribution.55 The trade press wasfull of similar arguments.56

The two founders of independent powerproducer AES attempted to make the quanti-tative case for vertical deintegration by citingthe postprivatization drop in UK generationcosts, but they failed to identify the key factorin that drop: lower fuel prices.57 They alsodescribed but failed to cite an “analysis [that]suggests divestiture of generation will loweroverall costs per kWh by 15 percent” and anunpublished consultants’ report that the sav-ing would be from 20 to 40 percent.58 Perhapsthe most surprising views were those of econ-omist Irwin Stelzer, the retired founder of aconsulting firm whose clients include manyintegrated utilities. He asserted that competi-tion was impossible as long as utilities werevertically integrated and proposed that utili-

10

If markets forgeneration are

superior to vertically

integrated operations, can

their benefits beobtained by

policies that alsomaintain the

benefits of integration?

Such questionshave gone largely

unasked as theU.S. industryrestructures.

ties deintegrate as a precondition for strandedcost recovery.59 None of those people evenbrought up the possibility that integrationcould also be beneficial.

A few experts were a bit less hostile to verti-cal integration. Two economists from the Uni-versity of California Energy Institute wrote:

If the vertically integrated utilitiesremain largely intact . . . their coordi-nation abilities could enhance reliabili-ty and reduce transaction costs.However, the utilities would also have acorrespondingly large capacity for theexercise of [horizontal] market power.If the utilities are dismantled along thelines of the UK model, then new mech-anisms for coordination would have tobe developed.60

MIT engineering professor Marja Ilic andher associates described the requirements foroperating methods and software that had yetto be developed if an ISO in a vertically dein-tegrated system was to operate a well-func-tioning set of wholesale and direct accessmarkets for both energy and ancillary ser-vices.61 And only two analyses by economistsprior to the opening of California’s marketsbrought up any of the econometric studies ofintegration discussed above.62 Both of themprovided cautionary discussions on the valueof integration, and one noted that prior todeintegration its advocates should show that“cost savings exceed foregone economies.”63

After more than a year of hearings and nego-tiations, the CPUC issued its initial decision inDecember 1995. The wholesale power marketand retail access aspects of that decision wouldbe altered before markets opened, but its gener-ation divestiture provisions would remain.They required that the state’s two largest cor-porate utilities divest themselves of 50 percentof their fossil fuel generating capacity located inCalifornia.64 A commission majority justifiedthis radical step by stating (without evidence)that “the vertically integrated electric utility isnot compatible with the institutions of a com-petitive market for electric services” and that

utility structure is “rooted in the past andincompatible with emerging markets.”65 Thedecision cited no testimony or other evidenceregarding the benefits of vertical integration orthe possible costs of a breakup.

The utilities accepted the decision primari-ly because it would guarantee revenue torecover the costs associated with the construc-tion of power plants that had yet to be recov-ered through the rate base (so-called strandedcosts) and allow the utilities to maintain somecompetitive advantages even after direct accessbegan.66 The 1996 legislation authorizing thePX, ISO, rate freeze, and stranded cost recov-ery imposed the same divestiture require-ments, again with no discussion of the costsand benefits of integration.67

California then applied to FERC forapproval of its market-based PX and ISO-managed transmission system. Testimonybefore FERC’s market-based rate proceedingsemphasized the ability of vertically integratedutilities to leverage market power from trans-mission to generation and distribution. Thestandards for market-based rates require anapplicant to delineate geographic markets forshort-term energy and capacity, and possiblyother commodities. The applicant must thenshow that it (in this case, California’s threelarge corporate utilities as a group) controls asmall enough part of the market and that itspower over price is minimal. The utilities,however, were unable to meet FERC’s stan-dards. Intervenors (protesting parties) com-pounded the problem with testimony claim-ing that the utilities’ horizontal dominance ofgeneration left them ideally suited to use theirtransmission to exercise vertical market power,and that even independent operation of thetransmission system (the ISO) might not suf-fice to neutralize it, at least prior to actualdivestiture of the generating plants. The utili-ties responded by proposing additional mar-ket power mitigation measures, including anindependent monitor and special contractsfor the pricing of generation required to oper-ate for reliability.68 The utilities were the onlyparties one would expect to defend verticalintegration, and in more normal circum-

11

Only two analyses by economists priorto the opening of California’s markets broughtup any of theeconometric studies of integration.

stances they might have done so. Here, howev-er, stranded costs were their prime concernand they would reluctantly accept verticaldeintegration as the price for recovering them.Thus the record at FERC is essentially devoidof any discussion of vertical integrationbeyond conjectures about market power.

In sum, prior to the opening of California’smarkets, most interested parties viewed verti-cal integration as a tool for the exercise of mar-ket power by utilities. The utilities also enjoyedhorizontal market power as owners of mostexisting generation. Regulators and othersbelieved that the combination of divestitureand an ISO might suffice to mitigate bothtypes of market power; particularly during thelimited time California gave its utilities torecover most of their stranded costs.69

The CPUC required the two largest utili-ties to divest half of their in-state gas-firedplants, but ultimately they chose to divest allof them to six different independent powerproducers and marketers.70 By FERC’s stan-dards for horizontal market power, the areawas now sufficiently competitive that theprices arising at the PX and ISO would not besubject to further regulation.71

The reforms initiated by the CPUC set thestage for the crisis that was to follow. Duringthe two years of the operation of the newCalifornia market, prices hovered near mar-ginal cost. By the spring of 2000, however, theyhad begun their rise to crisis levels. Numerousfactors contributed to the problem and arestill the subject of litigation and academicdebate.72 One possible factor that simple mea-sures of seller concentration could not predictwas market power exercised by the owners ofdivested generation. If generation is near itslimits, transmission is scarce, and demand ishighly inelastic, a single generator might movethe market price with a small change in out-put, and others would have reason to bidabove their marginal costs as well.73

Vertical Integration after the CaliforniaCollapse

Three years after California’s marketsbegan operating, its Power Exchange was

bankrupt and its utilities in disastrous shape.Only after their deintegration did econo-mists begin rethinking the relationshipbetween vertical integration and marketpower. This time their conclusions were quitedifferent.

New models showed that integrationcould actually constrain rather than enhancea generation owner’s market power. A gener-ator required to serve final demand has littlereason to cut the output of plants that itowns unless it can obtain power more cheap-ly from a market.74 Forward contracts thatcommit generators and users to fixed deliv-ery prices likewise diminish the incentive fora generator to exercise market power with itsuncommitted plants.75 Vertical deintegrationwas not solely responsible for California’sproblems, but a consensus arose that it facil-itated the exercise of market power by ownersof the divested plants in ways that would nothave happened if the utilities had remainedvertically integrated.76 As this was happening,the utilities began their long journey back tofinancial health and found themselves withan opportunity to vertically reintegrate.

Between 1998 and 2003 a binge of mer-chant power-plant construction had leftmany nonutility generators either bankruptor in poor financial health. The markets theyhad expected to materialize as states restruc-tured had largely failed to appear. Over thoseyears, total U.S. generation increased fromroughly 800 GW to 1,000 GW; 150 GW ofthat increase had been built by IndependentPower Producers.77

Over only 10 years, the ownership struc-ture of generation had changed dramatically.In the mid-1990s approximately 90 percent ofgenerating capacity was owned by utilities.Today, new plants and divestitures have leftonly 55 percent of the national total undercost-based regulation. Approximately 60 per-cent of the remainder is owned by unregulat-ed affiliates of utilities. Overoptimism fromall parties allowed IPPs (usually under projectfinance) to be funded largely by debt. By 2004,90 GW of them had been turned back tolenders, 23 GW had been bought by private

12

Prior to the opening of

California’s markets, most

interested partiesviewed vertical

integration as a tool for

the exercise ofmarket power

by utilities.

investors, and 10 GW had been purchased orrepurchased by regulated utilities.78 Thosechanges may be evidence that vertical integra-tion is returning to the industry.

As the finances of the IPP sector deterio-rate, the distressed assets have often beenpriced so attractively that purchase by utili-ties or their affiliates is clearly efficient.According to some observers, however, thesepurchases raise antitrust concerns becausethey needlessly reconcentrate suppliers inregional energy markets.79

Vertical integration is also being pursuedmore directly. Two of California’s three largeutilities are building new generation and thethird is applying to the CPUC for permissionto do the same. Under new state laws,California intends to reregulate and reverti-calize utilities in hopes of avoiding a repeti-tion of the 2000–01 crisis.80 Utilities mustnow file short-term and long-term resourceplans with state regulators, who approveindividual investments, set reserve require-ments, and impose “renewable” resourcequotas on them. California utilities are alsoattempting to slow the growth of distributedgeneration (very small facilities on end-usersites). They claim that restriction of its scopeis necessary for reliability, while others claimthat the utilities are trying to eliminate com-petitors.81

Lessons Learned about VerticalIntegration

There has been little pressure for reintegra-tion by either utilities or the public in thosestates where deintegration has been accompa-nied by relatively successful market outcomes(e.g., Texas, Massachusetts, and New York).These market outcomes may reflect no morethan temporarily advantageous supply anddemand situations, as California’s did duringits first two years. In particular, there are noavailable research findings about the effects ofeither deintegration or RTO membership onthe operating efficiency of utilities. One recentstudy has shown that fuel efficiency hasincreased for both divested generators andutility-owned units, with the largest improve-

ments occurring in those generators that weredivested to utility affiliates.82

It may be possible to perform studies com-paring utilities before and after they becamemembers of RTOs. The only available relatedstudy is by economists Magali Delmas andYesim Tokat, who found that deregulation ofretail access has a generally negative effect onutilities’ productive efficiency.83 Consistentwith the predictions of organization theory,they found that vertically integrated utilitiesthat supply the full requirements of their retailcustomers experience smaller efficiency lossesfrom the opening of retail markets, and so dothose that purchase their entire power sup-plies on wholesale markets. Utilities that mustmix market purchases with internal produc-tion suffer efficiency losses greater than thoseat the extremes.

California’s performance has brought ageneral agreement on the value of requiringtransitional contracts between utilities andthe owners of divested generation.84 A transi-tion from integration to unbundling gives riseto new price risks for both generators andretailers because generators sell at the whole-sale price while retail rates are usually fixed. Inan integrated utility, these cancel out, but adeintegrated system will probably require con-tracts to allocate the obligations and risks.85

Such contracts may be difficult to formulatebecause independent plants can obtain capitalmore cheaply if their contracts contain com-mitments for both prices and outputs, whileutilities prefer discretion about their econom-ic dispatchability under changing fuel pricesand system conditions. Utility CEO JohnRowe and his coauthors believe that a majordifference between California and Rowe’s util-ities in Philadelphia and Chicago was that reg-ulators in his states allowed divestitures to bedetermined by the utilities themselves, andthey also allowed contracts and hedging.86

Rowe also discusses the value of a properlyplanned transition. In Pennsylvania, the timepaths of stranded cost recovery were deter-mined in settlements with individual utilitiesand surcharges to their rates were set inadvance. Only one of Pennsylvania’s utilities

13

There has beenlittle pressure forreintegration byeither utilities or the public inthose stateswhere deintegra-tion has beenaccompanied by relatively successful marketoutcomes.

chose to divest.87 Instead of a discontinuousinstitutional break like California’s, theregional transmission organization known asthe PJM (so called because it is an intercon-nected system incorporating utility service inPennsylvania, New Jersey, and Maryland,although it also manages utility service inparts of Delaware, Illinois, Indiana, Kentucky,Michigan, North Carolina, Ohio, Tennessee,Virginia, West Virginia, and the District ofColumbia) imposed wholesale markets on a“tight” centrally dispatched regional powerpool that had operating and settlement mech-anisms in place. As a further safeguard, gener-ators were required to submit only cost-basedbids during the first year after PJM’s marketsopened. New York, however, offers a potentialcounterexample to Pennsylvania. Its regula-tors increased utilities’ uncertainty by requir-ing divestitures prior to formulating anystranded cost policy. They did, however, allow(but not require) contracts between utilitiesand owners of the divested plants.88 Most ofthose contracts will expire in the near future.

Partial vertical integration may be a soundstrategy for utilities that expect to serve sub-stantial amounts of load that have chosen notto leave regulated service.89 In the future, manyutilities will have some customers who obtaintheir own power supplies and others who areeither “captives” legally prohibited from usingthe market or who choose not to do so.Economist Jamie Read observes that their“provider of last resort” functions are nolonger expected to be transitional, and verti-cally deintegrated utilities must design effi-cient procurement plans for their core cus-tomers.90 Utilities that have sold their powerplants and lost their safe monopolies will havelower-quality credit, which will affect theirdecisions to build generation or buy energy.Read sees reverticalization by asset ownershipat one extreme, providing the hedge that onlyphysical assets can provide but also invitingregulatory scrutiny about prudence. Thatscrutiny will be more likely if the load servedby these assets chooses to depart.

An alternative to reintegration is a portfo-lio model in which the utility holds nothing

but contracts and uses the spot market toprovide for any excess load or to dispose ofexcess contracted power.91 California’s utili-ties are in transition toward an intermediatemix between the two, but one that will beheavily weighted in favor of utility-ownedassets and longer-term contracts. Utilitieswill have a continuing interest in well-func-tioning bulk power markets, although thedegree of interest may depend on whetherexisting customers can also depart and usethose markets.

SummarySome retrospective studies have asked why

restructuring attracted so much support,given its goal of moving a vital industry intolargely unknown territory. A slower openingof markets to direct access by large customerswould certainly have been feasible. As the dif-ficulties of administering the limited marketwere resolved, transactions could have beenopened to smaller customers. The market’sscope would have been market determinedrather than regulator imposed.92 A few econo-mists even question whether markets shouldhave been opened at all. Richard Rosen hasattempted to make a qualitative showing thatthe cost of creating and using markets in adeintegrated system is probably not worth theeconomies of integration that were sacri-ficed.93 Rosen believes that many industryanalysts were blinded to the costs of massiverestructuring by a long-standing and some-times justifiable dissatisfaction with the per-formance of regulation.94 Other economistsargue that restructuring has been a success inmost states and nations that have carried itout. Lynne Kiesling believes that deintegrationitself can and should be market driven:

The encouragement of restructuring ofutilities created substantial flexibilityin Pennsylvania’s electricity market.Divestiture is likely to occur to someextent as a part of restructuring, whenutilities refine their “core competen-cies.” Allowing retention of at leastsome generation capacity enables com-

14

Some retrospectivestudies have

asked whyrestructuring

attracted so muchsupport, given itsgoal of moving a

vital industry intolargely unknown

territory.

panies and consumers to reap the ben-efits of vertical integration where theyexist.95

The California restructuring process couldhave been a forum for reasoned discussion onthe future of vertically integrated utilities. Theold view held that integration was an obstacleto competition and the coming of energy mar-kets would allow regulators to specialize inwhat they allegedly did well—controlling nat-ural monopolies. The market could be left todo what regulators probably did poorly—applying competitive pressure to produce andinvest efficiently.

The newer view holds that the continuedexistence of vertical integration is evidence ofits efficiency. The fact that generation wastechnologically separable from other aspectsof power delivery did not imply that separa-tion was economically desirable.

Economists had a great deal to say aboutthe efficient design of energy markets duringthe restructuring, but the design of utility orga-nizations has been primarily a political ques-tion. In California the utilities’ prime interestslay in recovering stranded costs and position-ing themselves for post-transition competi-tion. After they made the bargains thatbrought the PX, ISO, and divestiture, therewere no parties interested in undoing the polit-ical compromise by attempting to make thecase that some degree of continuing verticalintegration might in reality be efficient.

During the 2000–01 crisis, energy pricesin the California spot markets tracked short-term energy prices at other locations in theWest quite closely. The major difference wasthat California’s utilities had a far greaterexposure to this market than utilities thatremained integrated, and the CPUC did notallow them to use other risk managementtools. Utilities elsewhere in the West wouldappear in the short-term markets as eitherbuyers or sellers depending on the day’s oper-ating conditions, but California’s utilitieswould always be massive buyers.

The state’s disastrous transition was afailed gamble by utilities that for the next five

years demand would not catch up with thestate’s largely unchanged generation capaci-ty. At the peak of the crisis, the state govern-ment signed long-term contracts for nearlyall of the power that its insolvent unitiescould not generate from resources that theystill owned. A few weeks after the signing,supply and demand conditions changed andenergy prices fell below those in the con-tracts, but by then California’s utilities werein effect reintegrated.

Over three years, California regulators weregiven two lessons on the hazards of thought-less decisions about integration: a quickdivestiture aggravated the effects of depen-dence on highly volatile energy markets, and apanic-driven reintegration through state con-tracts brought very high but stable prices.Those lessons about integration went eitherunlearned or misinterpreted, and new lawscontinue to expand the scope of state activityin utility planning.

Transmission Operations ina Restructured Industry

The regulatory restructuring undertakenin the electricity sector has been more ambi-tious than the regulatory restructuringsundertaken in other industrial sectors. That’sbecause, in the case of electricity, both marketinstitutions and governance institutions havebeen subject to politically induced change.96

The previous section described the changesin industrial organization in California. Thissection describes changes in the transmissionsystem’s governance institutions. Robustwholesale markets require that buyers andsellers have access to a wide region, but accesshad historically been obstructed by both utili-ties and regulators. Utilities preferred monop-olies in their service territories and externaltransactions only with other utilities, andprior to 1992 FERC had no powers to orderthem to transmit for eligible third parties.

Regulated transmission rates also stood inthe way. When two transmission-owning utili-ties traded power, a fictitious “contract path”

15

The regulatoryrestructuringundertaken inthe electricity sector has beenmore ambitiousthan the regulatoryrestructuringsundertaken inother industrialsectors.

between them would determine the allocationof transmission charges. In reality, the powerflowed everywhere in the region, but as long astransactions were few and excess transmissioncapacity was common, they could neglect theconsequences of power flows along the grid.Regulators set transmission charges on anaverage cost basis, and principles of nondis-crimination treated utilities on the contractpath symmetrically. If utility A sent power toutility C on a contract path that went throughB, C would be expected to pay both A’s and B’sfiled transmission charges. From a regionalstandpoint, this was only a slight alteration inpower flows, but under contract path ratemak-ing the cost of transmission over multiple sys-tems was a barrier to the growth of markets.

In late 1995 FERC began to study openaccess transmission policy. FERC embodied itsfindings in Order 888 of 1996.97 It expressedthe commission’s preference for ISOs that metcertain standards of independence but did notcompel their formation. FERC would considerproposals for both nonprofit and for-profitISOs, but stated that the latter could not beclosely affiliated with generation. FERC thenheld technical conferences at which corporateutilities envisioned ISOs as regulated corpora-tions while public entities preferred that theytake the form of nonprofit entities.

Economists were quick to weigh in. MITeconomist Paul Joskow envisioned a nonprof-it joint venture whose board of directorswould contain representatives of utilities,nonutility generators, regulators, and “othersrepresenting the public interest.”98 The appro-priate functions of an ISO were also debated,with Harvard’s William Hogan favoring loca-tional marginal pricing of transmission andfull integration with a PX. Some attendees,however, were concerned that they were plan-ning the details of an institution that hadnever before existed, and that once such aninstitution was in place it could not adjust tochanges in technology and markets.99

Transmission in CaliforniaAt the same time that FERC was formu-

lating Order 888, the CPUC released its

December 1995 decision on restructuring.100

Virtually all interested parties, includingcompetitive producers and traders, agreedwith its plans for an ISO. They believed thatif the utilities continued to operate transmis-sion they would schedule the flow of electric-ity on the grid in order to advantage them-selves against competitors. The ISO, on theother hand, would take no market positionsand have no interests in load or generation. Aseparate institution, the Power Exchange,would administer the energy markets, andbilateral transactions outside the PX werepossible for all parties other than the utilities.The ISO would integrate PX and bilateraltransactions and administer a zonal pricingsystem for transmission. The decision tookno position on whether it should be a regu-lated corporation, a nonprofit, or a govern-mental operation.101

After having helped to create the design ofthe PX and ISO, in mid-1996 California’sutilities applied to FERC for market-basedrates and argued that those markets were suf-ficiently competitive that their prices wouldsatisfy its “just and reasonable” legal stan-dard. The PX and ISO would both be non-profit institutions, governed by boards ofinterest group representatives.102

Economists on all sides had much to sayabout California’s market designs and trans-mission pricing, but none questioned the insti-tutional structures being proposed.103 Onlyone economist, Dennis Carlton of theUniversity of Chicago, testified as to the gover-nance rules and independence of ISOs. TheSacramento Municipal Utility District retainedCarlton to argue that transmission-owningutilities would dominate the ISO (their per-sonnel were in some cases the only ones knowl-edgeable enough to operate it) and that theywould use that knowledge to advantage theirown generation. Acting as planners at the ISO,the utilities would not want to build transmis-sion that would decrease the value of their gen-erators, many of which were “must-run” unitswhose operation was at times required for reli-ability. Carlton also questioned the voting rulethat required a two-thirds majority, since it

16

Economists on all sides had

much to sayabout California’s

market designsand transmissionpricing, but none

questioned theinstitutional

structures beingproposed.

would allow utilities to form coalitions withallies to veto proposals beneficial to a majorityof the board.104 Municipal utilities includingSacramento’s also protested that in the “col-laborative” process to form the ISO and PX,the only parties allowed to vote were the threecorporate utilities.105

Shortly after the ISO began operation, thepresident of the CPUC told a trade journalthat the CPUC actually believed that trans-mission divestiture and the formation of asingle transmission-only corporation wouldhave been a superior alternative to ISOs.“Political reality,” however, stood in the waybecause a divestiture would have been legallydifficult and would have required three tofive years and extensive financing.106 Thisepisode further points up the difficulty ofdesigning rational economic institutions in apolitical setting. At the time, there were noprospective transmission-only firms in exis-tence to offer expert testimony favoring sucha structure.107 Ten years later, a few transmis-sion-only companies exist, but they operateunder ISOs whose governance is heavilyinfluenced by the remaining integrated utili-ties.108

Cost/Benefit Studies and Order 2000Two years after the formation of ISOs in

California and the Northeast, only one otherISO had opened, in Texas. FERC’s interest inregional coordination remained strong, butits legal ability to compel membership inRTOs is still in doubt.

On December 17, 1999, the commissionissued Order 2000, which offered additionalinducements to join RTOs. Still faced withresistance, FERC next proposed a set ofregional RTOs, and in 2001 it commissioned acost/benefit study of ISOs and the marketsthat would result. The study estimated thatthe RTO markets would create benefits in theform of lowered production costs with a pres-ent value of $40.9 billion between 2002 and2021, approximately a 2 percent annual savingover their base case.109 Critics quickly deter-mined that the model’s assumptions abouttechnology, as opposed to markets, drove

most of its results. Approximately 85 percentof the alleged benefits came from its assump-tions about the increased efficiency of newgeneration. Some of the remainder was due toassumptions that reserve margins coulddecline from 15 to 13 percent and that trans-mission transfer capability would increase by 5percent per year at no cost.110

One of the most important flaws in thisand most later studies was the lack of anydescription of the trading institutions thatwere being assumed and how those tradinginstitutions might affect the calculation. Thebenefit estimate in the FERC study, forinstance, was the solution of a linear pro-gramming problem, derived from a model ofleast-cost dispatch rather than a model of theoperating practices that might occur in actu-al markets. Moreover, the study’s authorsdetermined that the cost of forming RTOswould be between $1 billion and $5.75 bil-lion.111 If 85 percent of their projected bene-fits are in fact due to improved generator effi-ciency, this implies that RTOs may not beworth forming. In practice, those costs haveproven quite high and have been increasingover time. Between 2000 and 2003, the opera-tion and maintenance costs of RTOs andISOs in California rose by approximately 35percent; in New York they rose by 100 percent;and in the PJM they rose by 250 percent. Thecorresponding figures per megawatt-hour ofelectricity were 23 percent, 73 percent, and181 percent. All of those ISOs had initiatedtheir basic market operations before or dur-ing 2000.112 Their setup costs ranged from$250 million to $500 million.113

A substantial number of other studies haveused methods similar to FERC’s. In 1996 agroup of pro-market organizations examineda least-cost dispatch model for wholesalepower markets and estimated a saving of up to40 percent.114 The U.S. Energy InformationAdministration estimated savings of 8 to 15percent from competitive markets, again onthe basis of dispatch algorithms.115 A numberof others exist, most of little individual inter-est.116 Their complex modeling techniquesand large data requirements make it extreme-

17

Two years afterthe formation of ISOs inCalifornia andthe Northeast,only one otherISO had opened,in Texas.

ly difficult to pinpoint the reasons for theirdiffering results.117

Even if we accept the calculations as accu-rate, many of their treatments of economicefficiency are theoretically questionable.Often they identify increased efficiency withdecreased customer bills, but some (possiblymuch) of that decrease must be nettedagainst the loss of wealth by generators,whose incomes will be lower. The only studythat adequately accounts for the transfer iseconomist Ellen Wolfe’s work on the pro-posed RTO West. She estimates a 2004reduction in marginal costs of $1.3 billion,from which lowered generation revenues of$900 million must be subtracted. The reportis also noteworthy because unlike others itanalyzes the situation with and without aspecific institutional innovation—the RTO’sproposed locational pricing system for trans-mission.118

Profits, Voting, and Monitors FERC Orders 888 and 2000 state that

FERC will consider applications by bothnonprofit and profit-seeking ISOs or RTOs.The original ISO proposals (made at a FERCtechnical conference) by economists WilliamHogan and Paul Joskow envisioned a non-profit organization with representative gov-ernment. They said little about the difficul-ties in governance such an organizationmight actually encounter (and whichCalifornia’s ISO did see during the 2000–01crisis). Neither they nor FERC gave notice-able weight to economists’ findings of moreefficient operations by profit-seeking firmsin other industries that contained a mix ofthem and nonprofit organizations (e.g., hos-pitals).119 On the other side, supporters ofnonprofit organizations largely disregard theefficiency findings and conclude that aninvestor-owned transmission company witheven minimal interests in generation will actmonopolistically. This author’s work is theonly work to examine the nonprofit contro-versy in the light of recent developments inorganizational and financial economics.120

My case for the efficiency of investor-owned

transmission firms is based on several appli-cations of principal/agent theory and theeconomic theory of voting.

Economists and political scientists haveextensively analyzed rules for collectivechoice.121 Their work has shown the innateimperfections of nearly all voting systems inaggregating individual preferences and thegeneral impossibility of controlling strategicvoting. That work, however, has also shownthat some decisionmaking mechanisms aresuperior to others in important ways, such asthe ability of the person who sets the agendato influence results by choosing a sequenceof votes.

The quality of the decisions that an ISO’sgoverning board makes will be critical to thesuccess of the markets it operates, but noexperts on voting or committee structuresprovided input during proceedings on ISOdesigns. The literature, however, suggeststhat the constellation of interest groups onan ISO board may render it relatively vulner-able to manipulation by strategic voting.122

“Nonprofit” ISOs may show no profit intheir books, but the votes of their governorsaffect the wealth of the interests they repre-sent. Some of California’s difficulties in2000–01 stemmed from the growing inabili-ty of its ISO’s governors to reach decisions,which ultimately led FERC to order a recon-stitution of the board, which has since beenruled an impermissible extension of the com-mission’s authority.123

Economists with an understanding ofcorporate organization and collective choicecould have usefully contributed to the RTOdebate in a third area. Order 2000 requiresthat all RTO applicants include a descriptionof their proposed market monitoring institu-tions (MMIs). These institutions use marketdata to detect activities believed to be exercis-es of market power, have further powers ofinvestigation, and are also charged withpointing out any flaws they might find inmarket design. Existing MMIs have pro-duced numerous reports and testimonies ofvarying quality, a discussion of which isbeyond the scope of this paper.

18

The quality of thedecisions that anISO’s governing

board makes willbe critical to

the success of the markets it

operates, but noexperts on voting

or committeestructures provid-

ed input duringproceedings on

ISO designs.

MMIs are both political and economicinstitutions. They were not suggested byFERC or by consumer groups. Instead, theywere originally proposed by the Californiautilities as amendments to their PX and ISOapplications after FERC ruled them ineligiblefor market-based rates. In some cases MMIsare staffed by RTO employees and in othersby appointed committees of external experts.Their functions are at least in part political.

Economists often disagree over whethercertain behavior is anti-competitive, but allMMI reports on record have been unani-mous.124 California’s MMIs reported someseller scheduling practices as anti-competi-tive attempts to raise price by submittingbids that did not reflect their true demands.They made no similar reports about attemptsby utilities, however, to submit false sched-ules whose effect would be to lower prices.125

In another vein, arbitrage between the day-ahead and real-time markets known as virtu-al bidding (simultaneous buy and sell ordersin the two markets) is a generally desirableand efficient practice. PJM’s monitors werenot under pressure from utilities to keepprices artificially low, and they encouragedvirtual bidding. California’s monitors wereunder such pressure, understood that virtualtrading would interfere with attempts tomanipulate prices downward, and made thepractice illegal. Economists have yet to per-form an impartial study of the costs and ben-efits of alternative methods of monitoringthe competitive behavior of markets. Hadthey done so, market monitoring might beless politicized than it is today.

SummaryEconomists have provided significant

input on the details of RTO market design,and their contributions have undoubtedlyimproved market performance.126 Whether byaccident or intent, their contributions to thedesign of RTOs and their governance wereminimal. Over the past 40 years there havebeen significant advances in the analysis oforganizations, transaction costs, and collec-tive choice that were directly applicable to the

design of ISOs and RTOs. That new learninghas convinced much of the economics profes-sion that the design of institutions is asimportant as the design of markets them-selves, and that economics offers insights thatcould not have been obtained from any otherdiscipline. It might have been quite useful atthe outset for economists to simply remindFERC and others that rational persons innonprofit organizations will seek to advan-tage themselves just as they would in for-prof-it ones. Instead, much of U.S. electricity isnow governed by organizations for whichthere are no precedents in any industrial con-text as important as electricity. Where thoseorganizations have been stressed, as inCalifornia, they have failed to produce coher-ent policy.

Conclusion

The analysis of vertical integration becamean integral part of economic theory only quiterecently. As this happened, economists cameto understand that vertical integration oftenhad desirable effects on economic efficiencybecause it reduced the costs of coordinatingeconomic activities relative to the alternativesof markets or contracts. Vertical integrationbecame a common organizational form inelectricity because of technological require-ments that supply equal demand at all timeseverywhere on a network. In addition, theindustry’s specialized plants were less vulner-able to opportunistic conduct if they wereowned by the same organization rather thanunder contract.

The old economic view saw vertical inte-gration as a tool that a monopolist could useto extract profit from competitive activities.Modern theorists discredited that argumentin unregulated situations, but it might stillapply to regulated ones. In the 1970s and1980s lawyers and economists created a liter-ature that made the case for vertical deinte-gration of utilities by simply assuming thatintegration served no useful function. If true,the separation of generation from transmis-

19

Much of U.S.electricity is nowgoverned by organizations forwhich there areno precedents inany industrialcontext as important as electricity. Where thoseorganizationshave beenstressed, as inCalifornia, theyhave failed toproduce coherentpolicy.

sion could bring the benefits of competitionat no cost in efficiency. Econometric researchproved that this was not so. Studies in the1980s and 1990s almost invariably conclud-ed that vertical integration produced effi-ciencies that would be lost in a breakup.These economies of integration applied toboth the generation-transmission interfaceand to the ownership of generators and fuelsupplies.

This scholarship was almost totally for-gotten as California and other states began torestructure their power industries in the mid-1990s. A few economists argued that therewere both costs and benefits to vertical inte-gration and a rule of reason was needed.Many others simply chose to assert that inte-gration was worse than useless. If not con-strained, transmission monopolists integrat-ed into power production could destroy thebenefits of competitive generation. The ISOcame into being as a midway point betweenfull integration and full deintegration.

In California ISOs were supplemented bydivestiture. Generators that would often setmarket prices at the PX and ISO were sold offwithout contracts that would have given theutilities security of supply and prices. Twoyears after California’s markets began opera-tion, the growing imbalance of supply anddemand combined with a constellation ofother forces to bring about a pricing crisis. Inits aftermath, utilities and regulators areinvestigating the possibilities for reverticaliz-ing utilities, possibly with a separation ofcore and noncore customers.

The ISO and RTO were envisioned asinstitutions that could operate and priceregional power flows efficiently. They werealso charged with administering markets forportions of that power. A series of question-able quantitative studies estimated that largebenefits would be forthcoming, but the stud-ies were calculations of optimum dispatchrather than projections of the behavior ofmarkets. Numerous interested parties wereconcerned about discrimination by transmis-sion owners, but the ISO concept was formu-lated without an adequate appreciation of

the economic incentives of the institution’smanagers and clients.

Some economic experts displayed thesame naiveté as noneconomists in theirexpectation that nonprofit organizationswould operate benignly and efficiently. Therewas never a real debate over whether RTOsshould be for-profit or nonprofit, in largepart because the nonprofit ISO was a politi-cally expedient compromise rather than athoughtfully planned institution. As econo-mists would have predicted, ISOs governedby representatives of interest groups have attimes had difficulty reaching coherent deci-sions and have instituted highly imperfectand politicized monitoring procedures.

If economists and others had betterunderstood the significance of vertical inte-gration in the industry, restructuring wouldhave produced better policies and betterinstitutions. Contracts and vertical integra-tion are substitutes, but California left itsutilities to divest their plants and rely onshort-term markets without any hedgingpossibilities. Markets have virtues, but thequestion of whether or not to rely on them isreally a question about the costs and benefitsof vertical integration. Economists have agreat deal of useful knowledge in this area,but they have played at best a peripheral rolein the design of the institutions that willdetermine the industry’s future.

Ultimately, the question of how best toorganize the electricity industry is a questionthat should be answered through trial anderror by market actors, rather than be decid-ed by politics. Unfortunately, while longer-term power contract markets arose almostspontaneously during the 1970s and 1980s,short-term electricity markets that balancehourly supply and demand require someplanning and design prior to the start ofoperations, which renders some governmen-tal role unavoidable. Decades of scholarshipon vertical integration were largely ignored inthe restructuring and market design process,and both producers and consumers will payfor that neglect for some time to come. Asstates and regions reorganized their institu-

20

Politics saw to itthat the most

important thingfor the market todecide would beoff limits—whatthe market itself

would actuallylook like.

tions, almost every interested party went onrecord as favoring one or another type ofmarket, and regulators often heard that theyshould “let the market decide.” But politics

saw to it that the most important thing forthe market to decide would be off limits—what the market itself would actually looklike.

21

Appendix: Summary of Vertical Integration Studies

Author Sample and(date) Data Year Method Findings Comments

Henderson 160 U.S. utilities, Marginal cost of steam, hydro, and Estimate of model that Only addresses effect(1985)127 most vertically purchased power is used as energy excludes produced power of generation costs on

integrated, 1970 transfer price in estimate of trans- yields, downwardly biased transmission/log cost function that includes labor, estimate of scale distributioncapital, and energy; tests coefficients economies; concludes and not reversefor separability costs are not separable

due to vertical economiesRoberts 65 U.S. electric- Estimates translog cost function for Coefficient restrictions implied Article primarily about

(1986)128 only utilities, no distribution to examine effects of by separability of distribution effects of territorial sizeholding company territory size and density, tests for and generation/transmission and customer densityunits, 1978 separability of distribution from costs are rejected (author on distribution cost;

generation and transmission notes this in passing since does not contain infor-study was intended to esti- mation for numericalmate effects of service area estimate of integrationdensity) effect

Eftekhari 61 U.S. non- Estimates multioutput translog cost Finds very few economies of Variables said to measure(1989)129 nuclear utilities, function with labor, capital, fuel scale but substantial output include number

1986 inputs diseconomies of joint pro- of ultimate customers,duction; concludes that fraction of generated smaller, vertically deintegrated power they buy, andutilities would be more efficient statistically unreliable

measure of interchange Kaserman & 74 U.S. electric- Estimates quadratic multiproduct cost Finds 12 % cost savings from Estimates of scope

Mayo only utilities, function that allows tests of vertical integration for average- economies require use(1991)130 vertically economies of scope between sized utilities; extremely small of a sample containing

integrated, 1981 generation and transmission/ utilities are the only ones not to distribution-onlydistribution benefit from it. utilities

Gilsdorf 72 U.S. vertically Estimates translog cost function for Performs Evans-Heckman Author notes that failure (1995)131 integrated utilities, generation and transmission/ subadditivity test for those to pass subadditivity test

1985 distribution [combined], with fuel, utilities whose location on need not support a capital, and labor costs, also customer estimated function has normal divestiture policy, sincedensity, capacity utilization, and percent economic properties [20 were there may be economiesof sales to ultimate customers excluded]; fails to reject null of scope between stages

hypothesis of additivity for any without subadditivityutilities; also finds stage-specific economies of scale

Continued on next page

22

Summary of Vertical Integration Studies (Continued)

Author Sample and(date) Data Year Method Findings Comments

Lee (1995)132 70 U.S. “electric Estimates translog production functions Tests for complete separability of Also estimates efficiencyutility firms,” for generation, transmission, generation, transmission, and losses from various 1990 distribution; also estimates final output distribution, and for separability forms of deintegration

as function of all variables of generation and distribution between 4.1 and 18.6alone; all null hypotheses of no percentseparability rejected

Hayashi 50 U.S. electric Estimates translog cost functions for Rejects null hypothesis of cost Estimates economies of et al. utilities, annual generation and transmission/ separability; also finds that both vertical integration for(1997)133 data 1983–1987 distribution, and for total large and small firms operate in firms ranging from 9.2

range of scale economies in percent to 24.2 percentgeneration

Thompson 83–85 U.S. “all Estimates translog cost function with Rejects separability of either Finds that in later years(1997)134 major investor- input prices and number of distribution or power supply the difference between

owned utilities” customers, territory size, and sales from remaining utility services unrestricted and restrict-1977, 1982, 1987, at different voltages over entire time period ed estimates is smaller1992 but remains significant

Goto & 9 Japanese vertically Estimates shadow cost and input Finds that generation enters Method also allowsNemoto integrated electric demands from Symmetric Generalized transmission/distribution cost estimation of allocative(1999)135 utilities, annual McFadden (SGM) function, inputs function positively in distortions in input mix;

data 1980–1997 include purchased power; tests for unintegrated case; concludes finds that average per-effect of generation capital on trans- that unintegrated costs are centage that costs couldmission/distribution costs and esti- higher because of overinvest- be reduced over samplemates allocative distortions ment in generation relative to period ranges from

integrated firms 0.13 % to 2.97 % forindividual utilities

Kwoka 147 U.S. corporate Estimates quadratic cost function in Negative interaction term Concludes that most (2002)136 utilities, some generation and distribution to test between generation and utilities have chosen to

unintegrated, 1989 for economies of scope distribution cost is evidence operate where they canof complementarity; comparison best realize these with standalone costs indicates economies, with that only very small utilities generation close to butshow diseconomies of vertical less than distributionintegration output

Nemoto 9 vertically Estimates SGM for variable and fixed Compares variable costs for Authors note questions & Goto integrated Japanese costs on assumption that capital is integrated and standalone pro- about their additive (2004)137 utilities, annual data incompletely adjusted to optimum duction of stages; finds average allocation of capital

1980–1999 economies of integration over between stages, state period for individual companies need to verify thatrange from 4.5 % to 13.9 % observed cost structures

are sufficient for natural monopoly

Notes1. The traditional topics of efficient input choice,profitable output choice, and optimal competitivestrategy are now subsumed in a more general theoryof economic organizations. A contemporary text-book that follows this approach is James A. Brickley,Clifford Smith Jr., and Jerold Zimmerman, Manage-rial Economics and Organizational Architecture, 3d ed.(New York: McGraw-Hill Irwin, 2002).

2. Demand is managed centrally and automati-cally by relays that cut off (black out) customersin defined geographic areas when imbalancesbetween supply and demand occur that cannot beremedied from backup generation.

3. See, e.g., Leonard W. Weiss, “Antitrust in theElectric Power Industry,” in Promoting Competition inRegulated Markets, ed. Almarin Phillips (Washington:Brookings Institution, 1975), pp. 138–73.

4. Regarding other nations, deintegration is some-times posited as an explanation for the fall in powercosts after the formation of the United Kingdom’smarkets; see, for instance, Roger Sant and RogerNaill, “Let’s Make Electricity Generation MoreCompetitive,” Electricity Journal 7 (October 1994):49–72. Shortly after the UK market was organized,real fuel prices decreased by 20 percent (coal) and 45percent (gas) while labor productivity doubled.Increases in productivity are more likely a conse-quence of privatization than of deintegration.David Newbery, “Privatisation and Liberalisation ofNetwork Utilities, European Economic Review 41(1997): 374.

5. RTOs have superseded ISOs in FERC’s termi-nology. Although their legal definitions differ, thetext uses them interchangeably.

6. Oliver Williamson, “The Vertical Integration ofProduction: Market Failure Considerations,”American Economic Review 61 (May 1971): 112–23;and Brickley, Smith, and Zimmerman, p. 531.

7. Small on-site “distributed generation,” howev-er, can be scheduled by its users under certainconditions.

8. John Landon, “Theories of Vertical Integrationand Their Application to the Electric UtilityIndustry,” Antitrust Bulletin 28 (Spring 1983): 101–30.For application of transaction-cost economics to therestructuring of other energy industries (and alsoelectricity), see Samuel Van Vactor, Flipping the Switch:The Transformation of Energy Markets (Ph.D.Dissertation, University of Cambridge, 2004).

9. Oliver E. Williamson, “Franchise Bidding for

Natural Monopolies—In General and with Respectto CATV,” Bell Journal of Economics and ManagementScience 7, no. 1 (1976): 73–104.

10. In the United States, the lines on the two sidesof the interface between a large transmission-owning utility and a small municipal distributionutility are separately owned. Power deliveries areusually under an all-requirements contract. If themunicipal system owns generation elsewhere, thetransmission operator integrates its output intothe regional system and accounts for it in theprice of deliveries to the city. The contracts gov-erning this relationship limit the options of bothparties with effects similar to those of verticalintegration.

11. U.S. Federal Energy Regulatory Commission(FERC), “Minutes of Technical Conference onPublic Utilities’ Acquisition and Disposition ofMerchant Generating Assets,” Docket No. PL04-9-000, June 10, 2004, p. 5, available in eLibrary atwww.ferc.gov.

12. LMP is a computer algorithm that assignsprices to locations on the grid that correspond todifferences in the marginal cost of producingpower at those locations. If at point A the cost is 5cents per kilowatt-hour and at B it is 11 cents,then the implied value of additional transmissioncapacity between them is 6 cents. If productioncapacity were available and transmission wereunconstrained, then B would get all of its powerfrom the cheaper plant at A and save 6 cents onevery kWh delivered. The 6-cent difference intheir LMPs is the value per kWh of increasing thecapacity of that link. There are financial instru-ments known as firm transmission rights (FTR)or congestion revenue rights (CRR) that allowtheir holders to hedge against unpredictablechanges in LMP as system conditions change.

13. The key case is Brown Shoe Co. v. U.S., 310 U.S.294 (1962). There the Supreme Court held that ashoe manufacturer’s attempt to purchase a chainof retail stores was an attempt to use its marketpower in manufacturing to monopolize retailing.

14. Richard Posner, Antitrust Law: An EconomicPerspective (Chicago: University of Chicago Press,1976), pp. 147–211.

15. 410 U.S. 366. It appears that the Court disre-garded numerous facts that might have led it to adifferent decision. See Andrew Kleit and RobertMichaels, “Antitrust, Regulation, and Rent-Seeking: The Past and Future of Otter Tail,” AntitrustBulletin 39 (Fall 1994): 689–725.

16. Edward Berlin, Charles Cicchetti, and WilliamGillen, “Restructuring the Electric Power Industry,”

23

in Electric Power Reform: The Alternatives for Michigan,ed. William H. Shaker and Wilbert Steffy (AnnArbor: University of Michigan, Institute of Scienceand Technology, 1976), pp. 231–35; MatthewCohen, “Efficiency and Competition in the Electric-Power Industry,” Yale Law Journal 88 (1979):1511–49; Philip Fanara Jr., James Suelflow, andRoman Draba, “Energy and Competition: The Sagaof Electric Power,” Antitrust Bulletin 25 (Spring1980): 125–42; John Landon and David Huettner,“Restructuring the Electric Utility Industry: AModest Proposal,” in Electric Power Reform: theAlternatives for Michigan, pp. 217–30; James Meeks,“Concentration in the Electric Power Industry: TheImpact of Antitrust Policy,” Columbia Law Review72, no. 1 (1972): 64–130; Richard Pierce, “AProposal to Deregulate the Market for Bulk Power,”Virginia Law Review 72 (October 1986): 1183–1235;and Weiss.

17. There are, however, opportunities for cities totake advantage of certain legal provisions.Municipal debt in the United States is largely tax-exempt, and municipal utilities have priority overcorporate utilities in the allocation of inexpensivepower from federal dams. The latter fact motivat-ed the requests for transmission service fromOtter Tail. Municipal utilities do not pay taxes,but most of them contribute fractions of theirgross revenue (usually 10 percent or less) to citybudgets and provide local government with freepower. Partisans and opponents of public powerdiffer over whether these contributions are largeror smaller than the taxes that a corporate utilitywould pay.

18. Economies of scale in coal-fired plants werenear their highest point, nuclear facilities werestill feasible, hydroelectric sites were becomingscarce, and natural gas was in shortage because ofprice controls. The technologies and laws thatallowed independent power production to thrivewere not operative at the time of most of thesewritings.

19. Cohen, p. 1524. His footnoted references areto Meeks, who also provided no useful sources;Weiss, who acknowledged that studies were need-ed; and prepared testimony by an economist, whostill testifies today on behalf of municipal utilitiesat FERC.

20. Meeks, p. 82. His evidence was to note the exis-tence of power contracts between utilities andbetween utilities and the federal government.

21. In the United States, a utility’s vertical inte-gration can be quantified as its degree of self-suf-ficiency in generation. Some companies own gen-eration in excess of their own loads, others arepurchasing some power at all times, and still oth-

ers are the operating units of holding companiesthat control several utilities. There are a few unin-tegrated utilities that generate only for wholesalesales or distribute only purchased power. Severalof the researchers exclude the unintegrated sys-tems from their data sets. Some of the samplestreat a holding company as a single observation,while others include each of their operating com-panies. This paper does not discuss some otherforms of integration examined by economists.They include cost comparisons between utilitiesthat sell only electricity and those that sell elec-tricity and gas. See John Mayo, “MultiproductMonopoly, Regulation and Firm Costs,” SouthernEconomic Journal 51 (July 1984): 208–18; andRaymond Hartman, “The Efficiency Effects ofElectric Utility Mergers: Lessons from StatisticalCost Analysis,” Energy Law Journal 17, no. 2(1996): 425–57. This paper also does not discussestimates of economies of scope due to servingseveral types of customer. See Douglas Gegax andKenneth Nowotny, “Competition and the ElectricUtility Industry: An Evaluation,” Yale Journal onRegulation 10 (Winter 1993): 63–87.

22. This is the case in the United States. Theauthors of the Japanese studies do not commenton the consistency or accuracy of their data,which may mean that they, too, have few suchproblems.

23. One remaining study is not directly compara-ble to those on the Appendix. Economist FayeSteiner uses 1986–96 annual data from 19 OECDcountries to examine the effects of restructurings.She attempts to explain variation in capacity uti-lization, deviations of actual from optimal(assumed 15 percent) reserve margins, prices toindustrial users, and the ratio of industrial to res-idential prices, using random effects regressionsthat include measures of restructuring and priva-tization. Vertical deintegration is associated withsignificantly higher rates of generator capacityutilization and smaller deviations of actual fromideal reserves, as is her measure of privatization.She finds that prices to industrial users are notsignificantly associated with vertical deintegra-tion, but the ratio of industrial to residential priceis significantly lower in nations that have unbun-dled generation and transmission or that have apower pool. Results like these are almost surelysensitive to regression specification, particularlywith international data. Her only publishedresults, however, are summaries of single regres-sions for each of the four performance measures.Faye Steiner, “Regulation, Industry Structure andPerformance in the Electricity Supply Industry,”OECD Economics Department Working Papersno. 238, April 2000.

24. That study, by economist Hossein Eftekhari,

24

defines some variables in unorthodox ways. Hismeasure of interconnection activities includes thealgebraic sum of interchanges into and out of a util-ity’s territory, which could be zero for a large trader.One of his output variables is sales to ultimate cus-tomers as a fraction of total sales, rather than anamount. In any case, his estimated cost functioncarries the implication that utilities should alwayseither specialize completely in retail sales or in salesof power to other systems, rather than any mix ofthe two. Hossein Eftekhari, “Vertical Integrationand Power Generation in the United States,” Journalof Economics 15, no. 1 (1989): 25–31.

25. The author found that there was no cost com-plementarity to be found in vertical integration.“Cost complementarity” means that the marginalcost of producing one good decreases when outputof the other is increased. Keith Gilsdorf’s findingsof no cost complementarity are still potentiallyconsistent with economies of scope and economiesof vertical integration, and his estimates showunexploited returns to scale in generation, trans-mission, and distribution. Keith Gilsdorf, “VerticalIntegration Efficiencies and Electric Utilities: ACost Complementarity Perspective,” QuarterlyReview of Economics and Finance 34 (Fall 1994):261–82; and Keith Gilsdorf, “Testing for Subaddi-tivity of Vertically Integrated Electric Utilities,”Southern Economic Journal 62 (July 1995): 126–39.

26. All of the studies use variants of two basicstrategies to estimate vertical economies. The firstis to estimate a cost function (usually translog,otherwise quadratic) on the assumption that theoutput of each stage (generation, transmission,and distribution) is from a multiproduct firm.The sizes and signs of the coefficients of theirinteraction terms then provide evidence oneconomies of vertical integration. Some formula-tions allow tests for economies of scope (i.e.,whether the sum of costs of standalone firms pro-ducing each of the stages exceeds the cost of finaloutput in an integrated firm) and invariably findthem. The second strategy estimates cost or pro-duction functions for each stage and then testsfor vertical separability by examining whetheroutput of an earlier stage significantly lowers thecosts of a later one. If it does, vertical effects arepresent and the production process is not separa-ble. It is possible but not likely that these resultsare tainted by selectivity bias. Perhaps integratedutilities have been formed by merger or are toler-ated by regulators because of higher efficiency, assuggested in Michael Pollitt, Ownership andPerformance in Electric Utilities (Oxford: OxfordUniversity Press, 1995), p. 33. The implication isthat these estimates should include unobservedcharacteristics of individual firms that lead someof them to vertically integrate and others not to.Most if not all U.S. utilities have been vertically

integrated since their formation, instead of beingcreated by mergers of generation and distributionoperators.

27. In addition to the works discussed below, onestudy details the range of data on utility operationsrequired to optimize and evaluate a demand-man-agement program and makes clear that a verticallyintegrated utility minimizes difficulties in obtainingand analyzing those data. See Ren Orans, Chi-Keung Woo, and Brian Horii, “Case Study: Target-ing Demand-Side Management for ElectricityTransmission and Distribution Benefits,” Manage-rial and Decision Economics 15 (1994): 169–75.

28. Paul Joskow, “Vertical Integration and Long-Term Contracts: The Case of Coal-Burning ElectricGenerating Plants,” Journal of Law, Economics, andOrganization 1 (Spring 1985): 33–80.

29. Ibid., p. 51; and Paul Joskow, “ContractDuration and Relationship-Specific Investments:Evidence from Coal Markets,” American EconomicReview 77 (March 1987): 172. Most generators inthe eastern United States operate with pollutioncontrol technologies that allow them to burn coalwith a range of sulfur content. Those in the Westare more often engineered to use low-sulfur coalfrom a particular mine.

30. Joskow “Vertical Integration and Long-TermContracts,” p. 65.

31. Ibid., p. 54.

32. Keith Crocker and Scott Masten, “Regulationand Administered Contracts Revisited: Lessons ofTransaction-Cost Economics for Public UtilityRegulation,” Journal of Regulatory Economics,January 1996, pp. 5–39, citing Joskow, “ContractDuration and Relationship-Specific Investments.”

33. John Filer, “Impact of Regulation on VerticalIntegration in the Electric Industry,” Review ofIndustrial Organization 1 (Fall 1984): 219.

34. John Gonzales, “Efficiency Aspects of ElectricUtility Coal Operations,” Energy Economics 4(April 1982): 131. He also finds that productivityis lower when a regulated mine operates under acost-plus contract with the buyer. He cautionsreaders that his findings do not by themselvesmake a case for deintegration, since he has notstudied the possible benefits of integrated mines.

35. Joe Kerkvliet, “Efficiency and Vertical Integration:The Case of Mine-Mouth Electric GeneratingPlants,” Journal of Industrial Economics 39 (September1991): 467–82.

36. Paul Joskow and Richard Schmalensee, “The

25

Performance of Coal-Burning Electric GeneratingUnits in the United States: 1960–1980,” Journal ofApplied Econometrics 2 (April 1987): 85–109.

37. Eric Hirst, U.S. Transmission Capacity: PresentStatus and Future Prospects, Report prepared forEdison Electric Institute, Washington, 2004,http://www.ehirst.com/PDF/TransmissionCapacityFinal.pdf; and North American Electric ReliabilityCouncil (NERC), Reliability Assessment, 1998, p. 7,http://www.nerc.com/~filez/rasreports.html. Bymost measures, a construction boom in the 1960sand 1970s allowed the industry to enter the 1980swith significant unused transmission capacity. Bythe early 1990s most of that capacity was in use,thanks to the growth in industry size and thegrowth of wholesale markets that began in the1980s. There was, however, no increase in newinvestments during the 1990s. Since 2003 trans-mission investment has increased in most parts ofthe country.

38. NERC is the coordinating agency for 10regional electric reliability councils that covermost of the continent. Members of those councilsinclude corporate utilities, independent powerproducers, governmental utilities, and coopera-tives. NERC, Reliability Assessment, 1997, p. 3, http://www.nerc.com/~filez/rasreports.html.

39. NERC, Reliability Assessment, 1998, p. 38.

40. Ibid., p. 7.

41. The current TLR procedures have been in placesince 1997. There are five different levels of emer-gency. The figures in the text refer to the three mostserious ones, whose growth rates have all beenhigh. A graph and source data are available atftp://www.nerc.com/pub/sys/all_updl/oc/scs/logs/trends.htm.

42. NERC, Reliability Assessment, 2001, p. 25,http://www.nerc.com/~filez/rasreports.html.

43. Diana Moss, “Competition or Reliability inElectricity? What the Coming Policy Shift Meansfor Restructuring,” Electricity Journal 17 (March2004): 25. A small number of transmission-own-ing utilities have been responsible for a large per-centage of TLR incidents. This, however, canreflect either the weakness of their grids or theirabundant opportunities to exercise marketpower.

44. Ibid., p. 17.

45. NERC, Reliability Assessment, 2001, p. 25.

46. Participant funding is also embodied in recent-ly issued rules for generator interconnections. See

Standardization of Generator InterconnectionProcedures and Agreements, 106 FERC & 61,220,2004.

47. Cleco Power LLC et al., Order GrantingPetition for Declaratory Order, Docket No. EL02-101-000, October 10, 2002. SEtrans withdrew itsapplication in 2003 because of conflictingdemands of state regulators and FERC.

48. Bruce Edelston, director of policy and plan-ning, Southern Company, quoted in BruceRadford, “The Laws of Physics,” Public UtilitiesFortnightly, April 4, 2003, pp. 22–23.

49. William Hogan, “Transmission Market Design,”Presentation graphics, April 4, 2003, http://ksghome.harvard.edu/~whogan/trans_mkt_design_040403.pdf.

50. A recent unpublished paper proposes use of ademand-revealing mechanism to circumvent free-rider problems. Robert Michaels, “The Economics ofParticipant-Funded Electrical Transmission,” Paperpresented at Rutgers University 14th AnnualAdvanced Workshop on Regulation, San Diego, June2004, http://www.business.fullerton.edu/economics/rmichaels/workingPapers/040920%20pf.pdf.

51. California Public Utilities Commission(CPUC), “Order Instituting Rulemaking on theCommission’s Proposed Policies Governing Re-structuring California’s Electric Services Industryand Reforming Regulation,” R.94-04-031, April20, 1994 (unavailable on Internet). This documentcame to be known as the “Blue Book,” from thecolor of its cover.

52. In 2002 the CPUC decided to remove all ofthese testimonies and the Blue Book itself fromits website for reasons that it has not made pub-lic. They are still accessible, however, at the com-mission’s offices.

53. FERC ratified the California experiment byapproving market-based rates in Californiathrough 1997 and beyond. FERC had to act toenable the California experiment to go forwardbecause the agency has the statutory obligation toregulate “just and reasonable” rates in wholesaletransactions. Prior to the coming of markets, thisrequired comparisons of proposed prices and pro-duction costs. In the 1980s FERC began allowingrates to be set by market prices in areas where sup-pliers were unconcentrated enough (according tocriteria set by the commission) that competitiveconditions would neutralize any market power oneof them might try to exert. In the months after theCalifornia filings, FERC began to process applica-tions to form exchanges in other parts of the coun-try, particularly the Northeast.

26

54. William Shepherd, “Reviving Regulation andAntitrust,” Electricity Journal 7 (June 1994): 23. Hedid not cite any of the research discussed abovebut warned that existing utilities would claimthat vertical separation “will cause large ineffi-ciencies, even when those claims are false.”

55. Richard Pierce, “The Advantages of De-inte-grating the Electricity Industry,” Electricity Journal7 (November 1994): 16–21. His earlier writings(Pierce, “A Proposal to Deregulate”) did describethe possible benefits of vertical integration butasserted without evidence that deintegrationwould be worth this cost.

56. For instance, environmental economistsDavid Moskovitz and Douglas Foy proposed tosolve the stranded cost problem with a deintegra-tion that included a sale of transmission at pre-mium prices to pay off the stranded costs. DavidMoskovitz and Douglas Foy, “Looking for Peacein the Middle of a Nervous Breakdown,” ElectricityJournal 7 (November 1994): 22–33. Blank, Gilliam,and Wellinghoff likewise suggested vertical dein-tegration of corporate utilities and the foundingof a nonprofit transmission company in order topay the utilities’ stranded costs and obtain taxadvantages. Eric Blank, Rick Gilliam, and JohnWellinghoff, “Breaking Up Is Not So Hard to Do:A Disaggregation Proposal,” Electricity Journal 9(May 1996): 46–55.

57. See Weiss.

58. Sant and Naill, p. 51. The probable source of the15 percent figure is Naill and Dudley, whose item-ization of savings yields a range of estimates between5 and 15 percent. Roger Naill and William Dudley,“IPP Leveraged Financing: Unfair Advantage,” PublicUtilities Fortnightly, January 15, 1992, pp. 15–20.

59. Irwin Stelzer, “Vertically Integrated Utilities:The Regulators’ Poisoned Chalice,” ElectricityJournal 10 (April 1997): 20–29.

60. Carl Blumstein and James Bushnell, “A Guideto the Blue Book: Issues in California’s ElectricIndustry Restructuring and Reform,” ElectricityJournal 7 (September 1994): 19. At the time oftheir writing, the concept of an ISO had not yetbeen developed.

61. Marja Ilic et al., “A Framework for Operationsin the Competitive Open Access Environment,”Electricity Journal 9 (April 1996): 61–69. Problemslike those she describes complicated operations inthe early years of the ISO and PX.

62. Gegax and Nowotny; and Lawrence Hill, “IsPolicy Leading Analysis in Electricity Restructuring?”Electricity Journal 10 (July 1997): 50–61. The integra-

tion studies are also mentioned in a report by theConsumer Federation of America, a political advoca-cy group usually sympathetic to regulation.Consumer Federation of America, All Pain, No Gain:Restructuring and Deregulation in the Interstate ElectricityMarket, 2002, http://www.consumerfed.org/allpain.pdf.

63. Hill, p. 53. I have encountered no subsequentcitations to this article.

64. CPUC, Decision D.95-12-063, December 20,1995, p. 98, http://www.cpuc.ca.gov/static/industry/electric/electric+markets/historical+information. The units in question were gas-fired genera-tors and under normal conditions would set pricein the new markets. Ultimately these two utilitieschose to sell all of their in-state gas-fired capacityto independent power producers.

65. CPUC, Decision D.95-12-063, pp. 10, 90.

66. One economist from a utility, however, com-mented, “The record in the CPUC case provided noevidence of a market power problem that needs tobe resolved through divestiture.” “CPUC PowerExchange Tops List of Latest State RestructuringPlans,” Inside FERC, December 25, 1995, p. 1. ASouthern California Edison vice-president wrotethat requiring the divestiture of generation“reduces competition” because it removes a com-petitor from the market. (It also adds new competi-tors who buy the units.) Vikram Budhraja, “PolicyChoices on the Road to a Competitive ElectricityMarket,” Electricity Journal 9 (May 1996): 60.

67. Currently in California Public Utilities Code.The law is still commonly known as Assembly Bill(AB) 1890.

68. FERC, Order Conditionally AuthorizingLimited Operation of an Independent SystemOperator and Power Exchange, Docket Nos. EC96-19-001 (Oct. 30, 1997); and “California’s ThreeMajor IOUs Submit Market Power MitigationStrategies,” Foster Electric Report, April 16, 1997, p. 8

69. The law allowed utilities to recover theirstranded costs in the difference between frozenretail rates and market-determined wholesaleenergy costs prior to 2002. Most market powerstudies submitted to FERC were concerned withmonopolistically high prices, but the law’s provi-sions made utilities more interested in low mar-ket prices. Some intervenors did express concernsabout monopsony (market power of a buyer) andpredatory pricing. The law also required utilitiesto apply any premia between the sales prices andbook values of divested plants to stranded costs.

70. The third-largest utility, San Diego Gas &

27

Electric, also divested its gas-fired plants as a con-dition imposed on its later merger with SouthernCalifornia Gas to form Sempra Energy.

71. FERC utilizes critical values of theHerfindahl-Hirschman Index of supplier concen-tration, a standard tool of antitrust analysis equalto the sum of squares of the market shares of allcompetitors. In some models of oligopoly it pre-dicts that increased concentration will lead tohigher prices, but in others it does not. See RobertMichaels, “Market Power in Electric UtilityMergers: Access, Energy, and the Guidelines,”Energy Law Journal 17, no. 2 (1996): 401–24.

72. For an overview of the crisis, see Jerry Taylorand Peter Van Doren, “The California ElectricityCrisis: What’s Going On, Who’s to Blame, andWhat to Do?” Cato Institute Policy Analysis no.406, July 3, 2001.

73. Timothy P. Duane, “Regulation’s Rationale:Learning from the California Energy Crisis,” YaleJournal on Regulation 19 (Summer 2002): 508. TheCPUC documents he cites are no longer availableon the Internet.

74. Frank Wolak, “Measuring Unilateral MarketPower in Wholesale Electricity Markets: TheCalifornia Electricity Market 1998–2000,” AmericanEconomic Review 93 (May 2003a): 425–31; and FrankWolak, “Diagnosing the California ElectricityCrisis,” Electricity Journal 16 (August 2003b): 11–37.

75. Frank Wolak, “An Empirical Analysis of theImpact of Hedge Contracts on Bidding Behavior ina Competitive Electricity Market,” InternationalEconomic Journal 14 (Summer 2000): 1–39; andJames Bushnell and Celeste Saravia, “An EmpiricalAssessment of the Competitiveness of the NewEngland Electricity Market,” University of CaliforniaEnergy Institute Working Paper CSEM-WP101,2002, http://www.ucei.berkeley.edu/PDF/csemwp101.pdf. This argument contains an unstatedassumption that makes it empirically questionable.It assumes that forward contracts are for some rea-son usually priced below the spot prices that willactually prevail in the future.

76. See Paul Joskow and Edward Kahn, “AQuantitative Analysis of Pricing Behavior inCalifornia’s Wholesale Electricity Market duringSummer 2000,” Energy Journal 23, no. 4 (2002):1–35; Severin Borenstein, James Bushnell, andFrank Wolak, “Measuring Market Inefficiencies inCalifornia’s Restructured Wholesale ElectricityMarket,” American Economic Review 92, no. 5(December 2002): 1376–1405; and RamteenSioshansi and Shmuel Oren, “How Good AreSupply Function Equilibrium Models? An Empiri-cal Analysis of the ERCOT Balancing Market,”

Working Paper, University of California EnergyInstitute, April 2006, http://www.ucei.berkeley.edu/PDF/EPE_017.pdf. For dissenting views, seeScott Harvey, William Hogan, and Todd Schatzki,“A Hazard Rate Analysis of Mirant’s GeneratingPlant Outages in California,” LECG LLC, Cam-bridge, MA, January 2, 2004, http://ksghome.harvard.edu/~whogan/Harvey_Hogan_Schatzki_Toulouse_010204.pdf; and Tim Brennan, “Question-ing the Conventional Wisdom,” Regulation 24, no. 3(Fall 2001): 63–69.

77. FERC, “Minutes of Technical Conference,” p. 5.

78. Figures are from testimony by Jone-Lin Wangof Cambridge Energy Research Associates at aFERC technical conference. FERC, “Minutes ofTechnical Conference,” pp. 5–7.

79. See testimonies of Peter Esposito and DianaMoss (antitrust concerns) and Christine Tezak(few antitrust concerns) in FERC, “Minutes ofTechnical Conference.”

80. “California’s Electric Utilities File 20-YearPlans with CPUC,” Foster Electric Report, April 23,2003, p. 10.

81. Richard Stavros, “Last Big Battle for StateRegulators? California Again Is the ProvingGround,” Public Utilities Fortnightly, October 15, 1999,p. 34.

82. James Bushnell and Catherine Wolfram, “Owner-ship Change, Incentives and Plant Efficiency,” Centerfor Study of Energy Markets Working Paper 140,March 2005.

83. Magali Delmas and Yesim Tokat, “DeregulationProcess, Governance Structures and Efficiency: TheU.S. Electric Utility Sector,” University of CaliforniaEnergy Institute, Energy Policy and Economics 004,March 2003.

84. Erin Mansur, “Vertical Integration in Restruc-tured Electricity Markets: Measuring MarketEfficiency and Firm Conduct,” Yale School ofManagement, Working Paper Series ES, no. 32,2003, p. 36. He also notes: “These results do notimply that divesting powerplants was a poor deci-sion. However, it does caution regulators that, if theydo require divestiture, then they also enable firms tosign contracts that will limit incentives to distort themarket.”

85. Newbery, p. 6. The British contracts ran forthree years. He also makes the interesting pointthat although many electricity industries havebeen restructured successfully, they all startedwith substantial spare capacity (p. 10). Californiabegan with enough excess capacity that for its

28

first two years many generators could not covertheir full costs at market prices. A rare constella-tion of events destroyed that excess more quicklythan the state’s utilities expected it would.

86. John Rowe, Janet Szczypinski, and PeterThornton, “Competition without Chaos,” AEI-Brookings Joint Center for Regulatory StudiesWorking Paper no. 01-07, June 2001, http://ssrn.com/abstract=286415. Rowe’s Chicago utilitydivested its fossil and nuclear plants, while thePhiladelphia company divested only nuclear.Along these lines, Green and Newbery supporteddeintegration for large British suppliers but not forsmall utilities in Scotland. Richard Green andDavid M. Newbery, “Competition in the ElectricityIndustry in England and Wales,” Oxford Review ofEconomic Policy 13, no. 1 (1997): 27–46. As inCalifornia, retail rates in Pennsylvania were capped.

87. That utility, GPU, encountered financial prob-lems when wholesale rates rose and customers inits area began abandoning direct access to return toits capped retail rates. State regulators refused togrant the company relief, saying that GPU shouldlive with the consequences of divestiture andrefusal to hedge. “To Avoid California Experience,GPU Wants to Collect over Rate Caps,” ElectricUtility Week, January 22, 2001, p. 16.

88. “New York Rebuts Idea of Bad Summer,”Power Markets Week, January 29, 2001, p. 16.

89. Lynne Kiesling, “Getting Electricity Deregu-lation Right: How Other States and Nations HaveAvoided California’s Mistakes,” Reason PublicPolicy Institute Study no. 281, May, 2001, p. 23,http://www.rppi.org/ps281.pdf.

90. Jamie Read, “Re-Verticalizing Electricity,”Presentation graphics to Harvard ElectricityPolicy Group, June 4, 2004, http://www.ksg.harvard.edu/hepg/Papers/Read.Reverticalizing.Electricity.060404.pdf.

91. Read also suggests that a utility could auctionthe right to serve its residual load to an indepen-dent organization.

92. Experts initially viewed the breakup ofAmerican Telephone and Telegraph as valuable toonly a handful of large businesses with extensivetelecommunications requirements. Within twoyears, new service providers were selling to indi-vidual residences.

93. Richard Rosen, “Can Electric Utility Restruc-turing Meet the Challenges It Has Created?” TellusInstitute, Boston, 2000, p. 32, http://www.tellus.org/energy/publications/restructchallenge.pdf. Healso notes some potentially harmful externalities

that have not yet been realized. For example, hequestions whether the choice of new generationinvestments should be in the hands of parties whodo not bear the risks of excessive reliance on natur-al gas.

94. Ibid., p. 112.

95. Kiesling, p. 23.

96. Robert Michaels, “The Governance ofTransmission Operators,” Energy Law Journal 20,no. 2 (1999): 233–62.

97. FERC, Order No. 888, FERC Statutes andRegulations & 31,036, 1996.

98. “FERC Wrestles with Implementation ofIndependent System Operators,” Electric UtilityWeek, January 29, 1996, p. 7. State regulators wouldnot appear on boards, but many other interestswould. The trade press (a transcript of the confer-ence is unavailable) does not discuss the reasoningbehind Joskow’s choice of a nonprofit.

99. “Most Industry Participants Voice StrongSupport for ISOs,” Foster Electric Report, February7, 1996, p. 12.

100. CPUC, Decision D.95-12-063, December 20,1995, http://www.cpuc.ca.gov/static/industry/electric/electric+markets/historical+information/d9512063/index.htm.

101. Ibid., p. 60.

102. “California’s Three Major IOUs SubmitMarket Power Mitigation Strategies,” FosterElectric Report, April 16, 1997, p. 8. Above the PXand ISO would be a newly created ElectricityOversight Board, whose jurisdictional conflictswith FERC were generally resolved in the latter’sfavor and which ceased to have many meaningfulfunctions as the market crisis grew.

103. Della Valle gives a fuller discussion of thelegal and financial issues in divestiture, as well asa taxonomy of the forms it might take. AnnaDella Valle, “Separating Transmission fromGeneration: What’s Required and Why,” ElectricityJournal 10 (March 1997): 83–90.

104. Dennis W. Carlton, Prepared statement onbehalf of the Sacramento Municipal UtilityDistrict, FERC Dockets ER96-1663-000 et al.,filed September 13, 1996. See also “FERC: Calif.Must Run vs. Market Power,” Electricity Daily,September 13, 1996; and “Most California Utility-Owned Thermal Plants Deemed ‘Must-Run’ byISO Board,” Electric Utility Week, July 21, 1997, p.11. Must-run units would be a continuing prob-

29

lem for the ISO, even after the utilities agreed oncontracts to set the price of their power. In 1997the ISO governing board classified 14,500 MW(one-third of the state’s power supply) as must-run, a figure which has since fallen.

105. “Various Parties Protest the California IOU’sISO and Power Exchange Proposals,” FosterElectric Report, June 26, 1996, p. 1.

106. “California PUC’s Conlon Urges Transmis-sion Sales, Not ISOs, as Market Power Cure,”Electric Utility Week, July 28, 1997, p. 10. Note thatCalifornia’s political reality became FERC’s pre-ferred institutional form. FERC, however, probablydoes not have the power to order divestitures.

107. Enron Capital and Trade Resources, a mar-keter, sponsored testimony at the CPUC byRichard Tabors proposing a transmission-onlyentity. The research underlying that testimonyappears in Chitru Fernando et al., “Unbundling theU.S. Electric Power Industry: A Blueprint forChange,” Risk Management and Decision Proc-esses Center Working Paper 95-03-05, WhartonSchool, University of Pennsylvania, 1995, http://www.tca-us.com/Publications/RUEI.pdf.

108. They include American TransmissionCompany in the Midwest, http://www.atcllc.com,and Trans-Elect, which operates regional systemsin Michigan and Canada and is prime contractorfor the expansion of Path 15 between northern andsouthern California, http://www.trans-elect.com.

109. ICF Consulting, Economic Assessment of RTOPolicy, Report to FERC, February 26, 2002, p. 7,http://www.ferc.gov/industries/electric/indus-act/rto/cost/02-26-02-report.pdf. The assumeddiscount rate was 6.97 percent. Several other sce-narios were posited, all of which provided annualbenefits ranging from 0 to 3 percent.

110. Thomas Lenard, “FERC’s Flawed Assessmentof the Benefits and Costs of Regional Transmis-sion Organizations,” Electricity Journal 15 (May2002): 74–78.

111. ICF Consulting, p. 79.

112. Margot Lutzenheiser, “Comparative Analysisof ISO/RTO Costs,” Presentation graphics fromAmerican Public Power Association NationalConference, June 21, 2004, pp. 3–4, available fromauthor at www.ppcpdx.org. The total percentageswere calculated from figures on a graph. Data aregiven for the per mWh costs, but these alsoinclude maintenance while the others are only“operating costs.” In unpublished correspon-dence, PJM has argued that Lutzenhiser’s figuresare too high since they include extraordinary

expenses in connection with expansion of mem-bership that should not be included in operatingcosts.

113. Vito Stagliano, “The Life and Death ofRegional Transmission Organizations,” ElectricityJournal 14 (December 2001): 23.

114. Michael Maloney, Robert McCormick, andRaymond Sauer, “Consumer Choice, ConsumerValue: An Analysis of Retail Competition inAmerica’s Electric Industry,” Citizens for a SoundEconomy Foundation, 1997, http://www.cse.org.

115. U.S. Energy Information Administration,Electricity Prices in a Competitive Environment: MarginalCost Pricing of Generation Services and Financial Status ofElectric Utilities, A Preliminary Analysis through 2015,DOE/EIA-0614, 1997, p. ix, http://tonto.eia.doe.gov/FTPROOT/electricity/0614.pdf. The reportestimated that prices could fall by as much as 24percent under conditions of “intense competition”with sellers aggressively cutting prices.

116. One recent study has shown largely negativeconsequences of an RTO. It was prepared by ICFConsulting for the three corporate utilities thatwould be operating under the proposed Grid-Florida operator. See Cost-Benefit Study of theProposed GridFlorida RTO, December 12, 2005, http://www.icfconsulting.com/Markets/Energy/doc_files/gridflorida-rto-report.pdf.

117. John Clapp and Margaret McGrath, “Compar-ing Apples and Oranges: RTO Cost-Benefit StudiesAre Difficult to Reconcile,” Public Utilities Fortnightly,September 15, 2002, pp. 32–37.

118. Ellen Wolfe, “RTO West Benefit/Cost Study,Final Report Presented to RTO West Filing Utilities,”Tabors Caramanis & Associates, 2002, p. vii. Thisreport is also the only one of its kind that estimatesthe spillover benefits to other regions that will resultfrom the changeover in RTO West’s territory.

119. A summary of research on for-profit and non-profit institutions appears in Henry Hansmann,The Ownership of Enterprise (Cambridge, MA:Harvard University Press, 1996). In one of manysimilar articles, lawyers Angle and Cannon assert intheir text that for-profit institutions will watchcosts more closely and be more innovative thannonprofits. The only authorities they cite are twoFERC commissioners, neither of whom was aneconomist. Stephen Angle and George Cannon Jr.,“Independent Transmission Companies: The For-Profit Alternative in Competitive Electric Markets,”Energy Law Journal 19, no. 2 (1998): 229–79.

120. Michaels, “The Governance of TransmissionOperators.”

30

121. James Buchanan and Gordon Tullock, TheCalculus of Consent (Ann Arbor: University ofMichigan Press, 1965); William Riker, The Theory ofPolitical Coalitions (New Haven, CT: Yale UniversityPress, 1962); and Donald Saari, The Geometry ofVoting (New York and Berlin: Springer-Verlag,1994).

122. Michaels, “The Governance of TransmissionOperators.”

123. California Independent System Operator v. FERC,372 F.3d 395 (2004)

124. Robert Michaels, “Watching the Watchers: CanRTO Market Monitors Really Be Independent?”Public Utilities Fortnightly, July 15, 2003, pp. 35–38.

125. In one of their reports the California PX’smonitors went so far as to explain how utilitiescould modify their bidding strategies to improvetheir chances of success in lowering market prices.

126. For a summary, see Robert Wilson,“Architecture of Power Markets,” Econometrica 70(July 2002): 1299–1340.

127. Stephen Henderson, “Cost Estimation forVertically Integrated Firms: The Case of Electricity,”in Analyzing the Impact of Regulatory Change in PublicUtilities, ed. Michael Crew (Lexington, MA: LexingtonBooks, 1985), pp. 75–94. This table was originallypublished in Robert J. Michaels, “Vertical Integra-tion: The Economics that Electricity Forgot,”Electricity Journal 17 (December 2004): 16–17.

128. Mark Roberts, “Economies of Density andSize in the Production and Delivery of ElectricPower,” Land Economics 62 (November 1986):378–86.

129. Hossein Eftekhari, “Vertical Integration and

Power Generation in the United States,” Journal ofEconomics 15, no. 1 (1989): 25–31.

130. David Kaserman and John Mayo, “TheMeasurement of Vertical Economies and theEfficient Structure of the Electric Utility Industry,”Journal of Industrial Economics 39 (September 1991):483–502.

131. Keith Gilsdorf, “Testing for Subadditivity ofVertically Integrated Electric Utilities,” SouthernEconomic Journal 62 (July 1995): 126–39.

132. Byung-Joo Lee, “Separability Test for theElectricity Supply Industry,” Journal of AppliedEconometrics 10 (January–March 1995): 49–60.

133. Paul Hayashi et al., “Vertical Economies: TheCase of the U.S. Electric Industry, 1983–87,” SouthernEconomic Journal 63 (January 1997): 710–25.

134. Herbert Thompson Jr., “Cost Efficiency inPower Procurement and Delivery Service in theElectric Utility Industry,” Land Economics 73, no. 3(1997): 287–301.

135. Mika Goto and Jiro Nemoto, "Analysis ofCost Structure by Multi-Product SymmetricGeneralized McFadden Cost Function: Economiesof Vertical Integration of Japanese Electric PowerCompanies," (in Japanese) Denryoku Keizai Kenkyu(Electricity Economic Studies) 42 (1999): 1–13.

136. John Kwoka, “Vertical Economies in ElectricPower: Evidence on Integration and Its Altern-atives,” International Journal of Industrial Organization20 (2002): 653–71.

137. Jiro Nemoto and Mika Goto, “TechnologicalExternalities and Economies of Vertical Integrationin the Electric Utility Industry,” International Journalof Industrial Organization 22 (2004): 676–81.

31

OTHER STUDIES IN THE POLICY ANALYSIS SERIES

571. Reappraising Nuclear Security Strategy by Rensselaer Lee (June 14, 2006)

570. The Federal Marriage Amendment: Unnecessary, Anti-Federalist, and Anti-Democratic by Dale Carpenter (June 1, 2006)

569. Health Savings Accounts: Do the Critics Have a Point? by Michael F. Cannon (May 30, 2006)

568. A Seismic Shift: How Canada’s Supreme Court Sparked a Patients’ Rights Revolution by Jacques Chaoulli (May 8, 2006)

567. Amateur-to-Amateur: The Rise of a New Creative Culture by F. Gregory Lastowka and Dan Hunter (April 26, 2006)

566. Two Normal Countries: Rethinking the U.S.-Japan Strategic Relationship by Christopher Preble (April 18, 2006)

565. Individual Mandates for Health Insurance: Slippery Slope to National Health Care by Michael Tanner (April 5, 2006)

564. Circumventing Competition: The Perverse Consequences of the Digital Millennium Copyright Act by Timothy B. Lee (March 21, 2006)

563. Against the New Paternalism: Internalities and the Economics of Self-Control by Glen Whitman (February 22, 2006)

PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2PA Masthead.indd 2 2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM2/9/06 2:08:35 PM