The Political Economy of Market-Based Environmental ... - CiteSeerX

The Political Economy of Market-Based Environmental Policy: The U. S. Acid Rain Program Author(s): Paul L. Joskow and Richard Schmalensee Source: Journal of Law and Economics, Vol. 41, No. 1 (Apr., 1998), pp. 37-83 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/725659 Accessed: 13/01/2009 17:11 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ucpress. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Journal of Law and Economics.

http://www.jstor.org

THE POLITICALECONOMYOF MARKET-BASED ENVIRONMENTAL POLICY:THE U.S. ACID RAINPROGRAM* PAUL L. JOSKOWand RICHARDSCHMALENSEE MassachusettsInstituteof Technology ABSTRACT The U.S. acid rain program enacted in 1990 gave valuable tradable sulfur dioxide emissions permits-called "allowances"--to electric utilities. We examine the political economy of this allocation. Though no Senate or House votes were ever taken, hypothetical votes suggest that the actual allocation would have beaten plausible alternatives. While rent-seeking behavior is apparent, statistical analysis of differences between actual and benchmark allocations indicates that the legislative process was more complex than simple models suggest. The coalition of states that produced and burned high-sulfur coal both failed to block acid rain legislation in 1990 and received fewer allowances than in plausible benchmark allocations. Some of these states may have received additional allowances to cover 1995-99 emissions by giving up allowances in later years, and some major coal-producing states seem to have focused on benefits for miners and on sustaining demand for highsulfur coal. I.

INTRODUCTIONAND SUMMARY

DESPITEthe attractiveefficiencypropertiesof "market-based" approaches to internalizing environmental externalities, such as emissions taxes or tradable emissions permit systems, these approaches have rarely been used.1 United States environmental policy has relied instead on a vari* The authorsareindebtedto Amy Ando,PaulEllickson,and,especially,ElizabethBailey for excellentresearchassistanceandto the MITCenterfor EnergyandEnvironmental Policy Researchfor support.They are gratefulfor aid and guidanceto Bruce Braine,Rob Brenner, DennyEllerman,DennisEpple,RobertHahn,KarlHausker,BrianMcLean,LarryMontgomery, RichardNewell, RobertStavins,Max Stinchcombe,andparticipantsin seminarsat HarvardUniversity,Yale University,and the Universityof Texas, Austin.Despite all this help, only the authorsare responsiblefor the views expressedin this articleand any errorsit may contain.RichardSchmalenseeparticipatedin the preparationof the Bush administration's acid rain proposalsand in negotiationsbetween the administrationand the Senate in early 1990 on what becamethe CleanAir Act Amendmentsof 1990. Paul Joskowwas a member of the EPA Acid Rain AdvisoryCommittee,which workedwith the EnvironmentalProtection Agency to producethe regulationsthatimplementedthe acid rainprogramcalled for by that legislation. ' For discussionsof otherenvironmentalprogramsemployingeconomic instruments,see Thomas H. Tietenberg,Emissions Trading:An Exercise in ReformingPollution Policy (1985); RobertW. Hahn& GordonL. Hester,MarketablePermits:Lessons for Theoryand [Journalof Law and Economics,vol. XLI (April 1998)] ? 1998 by The Universityof Chicago.All rightsreserved.0022-2186/98/4101-0002$01.50 37

38

THE JOURNAL OF LAW AND ECONOMICS

ety of source-specific"commandandcontrol"regulationsthatspecify limits on emissions rates or mandateparticularcontroltechnologies.Title IV of the 1990 Clean Air Act Amendments(1990 CAAA, Public Law 101549) establishedthe firstlarge-scale,long-termU.S. environmentalprogram to rely on tradableemissionspermits(called "allowances"in the 1990 legislation)to controlemissions.Its targetwas electricutilityemissionsof sulfur dioxide (SO2),the majorprecursorof acid rain. Any tradablepermit scheme for controllingemissions must specify a quantitycap or emissionsceiling for each of the geographicemissionsmarkets within which emissions permitscan be traded.This emissions cap, in turn,definesthe totalendowmentof emissionspermitsthatwill be in circulation in each emissions market.Any tradablepermit scheme must also adopt a methodfor distributingthose permits-by giving them away, by selling themat auction,or by some othermeans.The choice necessarilyhas distributionalimplications,and, in the presence of transactionscosts or other barriersto trading,it has efficiency implicationsas well.2 Because emissions permitsare valuable and decisions about their distributionare madeby politicalinstitutions,these decisionsare likely to be highly politicized, reflectingrent-seekingbehaviorand interestgroup politics. In the U.S. acid rainprogram,the expectationwas thatallowanceswouldbe worth about$5 billion per year once the programwas fully operational. In this article, we examine how Congress,influencedby the executive branchand variousspecialinterests,distributedwhatwas essentially(as we discuss below) a fixed endowmentof SO2allowancesamongelectricutilities in the process of craftingacid rain legislation.The literaturecontains essentiallyno empiricalwork on the distributionalimplicationsof alternative market-basedcontrolmechanisms,largelybecausetherehave been few applicationsof such mechanisms.3In particular,little attentionhas been devoted to how interestgroup politics and associatedrent-seekingbehavior Practice,16 Ecology L. Rev. 361 (1989); andNationalEconomicResearchAssociates,Inc., Key Issues in the Design of NOXEmission TradingProgramsto Reduce Ground-Level Ozone,ch. 2 (1994). RogerG. Noll, EconomicPerspectiveson the Politicsof Regulation,in 2 Handbookof IndustrialOrganization1254 (R. Schmalensee& R. D. Willig eds. 1989), at 1275, discusses some of the politicalreasonssuch programsare rare. 2 On efficiencyimplicationsin this context,see RobertN. Stavins,TransactionsCosts and TradeablePermits,29 J. Envtl. Econ. & Mgmt. 133 (1995). 3 For relatedwork, see Noll, supra note 1; Bruce A. Ackerman& William T. Hassler, CleanCoal andDirtyAir (1981); RobertW. Crandall,An Acid Test for Congress?8 Regulation 21 (1984); RobertW. Hahn& Roger G. Noll, Barriersto ImplementingTradeableAir Pollution Permits:Problemsand RegulatoryInteractions,1 Yale J. Reg. 63 (1983); and B. PeterPashigian,The Effects of Environmental Regulationon OptimalPlantSize andFactor Shares,27 J. Law & Econ. 1 (1984), andEnvironmental Regulation:Whose Self-Interests Are Being Protected?23 Econ. Inquiry551 (1985).

MARKET-BASEDENVIRONMENTALPOLICY

39

affectthe allocationof permitsin a tradablepermitsystem.This is a serious gap in the literature.The political acceptabilityof market-basedmechanisms for internalizingenvironmentalextemnalities will dependheavily on theirdistributionalimplications.Whenevervaluablepropertyrightsare created by legislation,4the associated allocation decisions are likely to be highlypoliticizedin muchthe same way as tax legislationor appropriations bills.5Understandingbetterhow the politicalprocessdeals with such allocationalissues can help us to design environmentalprogramsthat are both economicallyefficientand politicallyacceptable. It is difficult to apply some of the tools of modem political economic analysisto complex legislationof this sort, particularlywhen there are no meaningfulvotes.6However,since allowancesarehomogeneousandcan be tradedandbanked,the distributiveimplicationsare easy to quantify.Inthis case,the allocationof allowancesis similarto the allocationof government fundsthroughthe legislativeappropriations process.7The availabilityof detaileddata on the initial allocationof allowances permitsanalysis of the incidenceof individuallegislativeprovisions,as well as analysisof winners andlosers underalternativeallowanceallocationschemes.8 Thenext sectionprovidesa brief overviewof the tradableSO2allowance programcreatedby the 1990 CAAA. Section III reviews importantaspects 4Technically,the SO2 allowancescreatedby the 1990 CAAA are not propertyrights, sinceCongresscan changethe numberof allowancesissuedor do away with themaltogether withoutraising a constitutionalclaim for compensation(see Section 403(f) of the 1990 CAAA).In all otherrespects,however,allowancesare treatedas propertyrights.They are freelytradeable,a varietyof marketmechanismsare mediatingtransactions,and the EPA consciouslyallocatedallowancesto eligible partiesfor yearsbeyond 2010 to provideconfidencethat they would be treatedas durablepropertyrights.All this would clearly make it difficultto alterallowanceallocationsin the futurein responseto new information politically about costs or benefitsof reducingSO2emissions. 5Related researchon congressionalspendingdecisions has been performedby Lisa J. Kiel& RichardB. McKenzie,The Impactof Tenureon the Flow of FederalBenefits to 41 Pub.Choice285 (1983); DavidP. Baron,DistributivePoliticsandthe Persistence SMSAs, ofAmtrak,52 J. Pol.883 (1989); and Steven D. Levitt & JamesM. Poterba,Congressional Distributive Politics and State EconomicPerformance,Pub. Choice (in press). 6See Noll, supra note 1, at 1270-72, for a general discussionof the use of empirical modelsto test interestgrouptheoriesof legislativepolitics;see also JosephP. Kalt & voting Mark Zupan,Captureand Ideologyin the EconomicTheoryof Politics, 74 Am. Econ. Rev. 243(1984); Pashigian,EnvironmentalRegulation,supra note 3; and Sam Peltzman,How Efficient Is the Voting Market?33 J.Law & Econ. 27 (1990). 7See, for example,Levitt & Poterba, supra note 5, and the literaturethey discuss. 8We have data on the allocationof allowancesto individualcombustionunits, each of which consistsof a combustiondevice (boileror turbine)used to powerone or moregeneratunits,each of which consists of a single electricgenerator.Most combustionunitspower ing asingle generatingunit, so our use of the term unit for the sake of brevityin what follows should cause no confusion.A generatingplant often houses severalgeneratingunits, which be of differentscales, vintages,or types. may

40


of earlierdebates on legislation affecting SO2emissions. Section IV discusses the political developmentof the 1990 acid rain program.Section V examinesthe comparativelysimple allocationof allowancesin Phase I of that program,coveringcalendaryears 1995 through1999. Section VI discusses in moredetailthe provisionsfor allocatingallowancesin the first 10 years of Phase II, 2000-2009. Section VII comparesdistributionalaspects and otherfeaturesof the resultingallocationand of benchmarkalternative allocationschemes. Section VIII presentsthe resultsof hypotheticalvotes by bothHouses of CongressbetweenactualandbenchmarkPhaseII allocation patterns.Section IX employs regressionanalysis to relate differences between actual and benchmarkpatternsto a varietyof variablesdesigned to capturethe influenceof importantinterestgroups,congressionalleadership and committeeinfluence,and stateand nationalelectoralpolitics. Section X presentsa few concludingcomments. II. THE1990 ACIDRAINPROGRAM IN BRIEF Acid rain(or, moreproperly,acid deposition)occurswhen sulfurdioxide (SO2)and nitrogenoxides (NOx)react in the atmosphereto form sulfuric and nitric acids, respectively.9These acids then fall to earth, sometimes hundredsof miles downwindfrom their source,in eitherwet or dry form. In North America, acid rain is a concernmainly in the northeastUnited States,particularlyin the Adirondacksand New England,and in southeast Canada.It has been arguedthat in those areas acid rain damagesaquatic life and harmstrees in sensitive forest areas. The dominantprecursorof acid rainin the UnitedStatesis SO2from coal-firedand, to a much smaller extent,oil-firedpowerplants.These emissionsare the focus of the tradable allowanceprogramadoptedin 1990.1? Title IV of the 1990 CAAA representsa fundamentalchangein the regu9 U.S. GeneralAccountingOffice (GAO), Allowance TradingOffers an Opportunityto ReduceEmissionsat Less Cost, at 13 (ReportGAO/RCED-95-30,December1994). '0 Electricutilitiesaccountedfor about70 percentof 1985 U.S. SO2emissions:coal-fired units accountedfor 96 percentof this total, and oil-firedunits accountedfor the remainder; U.S. EnvironmentalProtectionAgency (EPA), National Air PollutantEmission Trends, 1900-1993 (ReportEPA-454/R-94-027,October1994). The other30 percentof emissions is accountedfor by a wide varietyof industrial,commercial,and residentialboilersand process sources(includingsmeltersandpaperfacilities),as well as by the use of diesel fuel for Aside fromcertainvoluntaryopt-inprovisionscontainedin the 1990 CAAA, transportation. includingthese othersourcesin the allowanceprogramwas not given seriousconsideration. These sources are generallyindividuallymuch smaller than utility sources and are much more diverse. Moreover,there were no systematic"baseline" emissions data availablefor these sourcesto providea basis for allocatingallowancesto incumbents.On this issue, see Nancy Kete, The Politics of Markets:The Acid Rain ControlPolicy in the 1990 Clean Air Act Amendments,217-21 (unpublishedPh.D. dissertation,JohnsHopkinsUniv., 1992).


41

latoryframeworkgoverningair pollutionin the United States.Previousair pollutionregulationsfocused on individualsources, their emission rates, and the applicationof specific control technologies to individualsources with certainattributes.The 1990 acid rain law, on the otherhand,focuses on aggregateemission levels ratherthanindividualsources,deals with the emissionsof SO2ratherthanemissionrates,places an aggregatecap on SO2 emissions,and gives pollutersextensiveflexibilityin choosingwhetherand how to reduceemissions at specific sources.The importanceof this change of approachgoes well beyondthe introductionof interutilitytrading.In particular,the 1990 law gave utilitieswith multiplefossil-firedgeneratingunits enormousand unprecedentedflexibilityin complyingwith emissionslimits even if they tradedno allowancesat all with otherutilities. Title IV of the 1990 CAAA was advertisedas requiringa 10 million ton per year reductionin SO2 emissions from 1980 levels by the year 2000. To achieve this goal, the law createda cap on SO2emissions from electric generatingplants of roughly 9 million tons per year, effective in the year 2000 and beyond. This emissions cap was to be achieved in two phases. During Phase I (1995 through1999), the 261 dirtiestgeneratingunits (in 110 generatingplants)were requiredto reducetheir emissions by roughly 3.5 million tons per year beginningin 1995.11In Phase II (2000 and beyond),virtuallyall fossil-fueledelectricgeneratingplantsbecome subjectto the nationalcap on aggregateannualSO2emissions.(All stateshadPhaseII unitsexceptIdaho,which had no fossil-fueledgeneratingunits,Alaska,and Hawaii.) The Phase I reductionsand the Phase II cap were enforcedthroughthe annualissuanceof tradableemissionsallowances,each of which permitsits holder to emit 1 ton of SO2in a particularyear or any subsequentyear.12 Each unit has 30 days after the end of each year to deliver to EPA valid allowances sufficientto cover its emissions duringthe year. At that time the EPA cancels the allowancesneeded to cover emissions.Failureto produce the necessary allowances subjectsthe utility to substantialfinancial penaltiesand the need to make additionalfutureemissions reductions.Allowancesgood in any particularyearbut not neededto cover SO2emissions in that year may be "banked" for futureuse. Ownersof individualunits 11 These units are simply listed in table A of the 1990 law; they correspond to 263 combustion units. These units were selected because they had an emissions rate (ER) greater than 2.5 pounds of SO2 per million Btus of heat input and a nameplate capacity of at least 100 megawatts. 12 In fact, these allowances are like checking account deposits; they exist only as records in the EPA's computer-based allowance tracking system. This system contains accounts for all affected generating units and for any other parties that want to hold allowances. It can be used to transfer allowances from one account to another.

42


are free to decide whatmix of emissionsreductionsand allowancetransactions they will employ to meet each year's allowanceconstraint,andessentially no restrictionsare placedon emissionsreductiontechniques.Thereis also no restrictionon who may buy or sell allowances.Brokershave acquiredsome in hopes of futureprice increases,for instance,and environmentalistshave acquiredsome in orderto reduceemissions more than the law requires. The units subjectto PhaseI reductionswere issued a total of roughly5.7 million allowancesfor each of the 5 years includedin Phase I. The basic allocationformulafor each unit in Phase I involved multiplyingan emissions rate (ER) of 2.5 poundsof SO2per million Btus of heat input times baseline heat input (generallythe unit's 1985-87 average).As we discuss in Section V, however,therewere significantdeparturesfrom this formula in the finalbill, the most importantof which was designedto favorthe use of easternhigh-sulfurcoal. Phase I obligationscannotbe met by shifting electricityproductionfrom a Phase I unit to units not affectedby Phase I. DuringPhase II, each utility generatingunit is allocateda specific number of SO2allowancesper year out of the roughly9 million per year made availablefor the entirecountry.(When variousbonuses are takeninto account, about9.4 million allowancesare availableannuallyfrom 2000 until 2009, and 8.95 million tons are availableannuallythereafter.)The allocation rules for the years 2000-2009, which we analyzein detail below, are specifiedin about30 statutoryprovisions.The provisionsfor 2010 and subsequentyears are only slightly less complex. DuringPhase II, utilities can cover their emissions with the allowancesthey were allocated,can buy allowances, sell allowances,or bankallowancesfor futureuse. Any individual or firmis free to buy and sell allowancesas well. In additionto giving allowancesto each generatingunit, the EPA has conductedsmall annualrevenue-neutral allowanceauctionssince 1993. The auctionedallowancesare acquiredby the EPA by holdingback 2.8 percent of the allowancesissued to each unit, and each unit in turnreceives a pro rata shareof the proceedsof the auction.The auctionprovisionwas a response to concernsby independentpower producersand rapidlygrowing utilities that an active marketfor allowanceswould not emerge, concerns strengthenedby assertionsduringdebateson the 1990 CAAA thatutilities would hoardtheirinitial allocationsand refuse to sell at any price.13 13 Until recently, some allowances were also held back for sale at a fixed price (which turned out to be well above market prices); any excess supply was later auctioned. Karl Hausker, The Politics and Economics of Auction Design in the Market for Sulfur Dioxide Pollution, 11 J. Pol'y Anal. & Mgmt. 553 (1992), discusses the political economy of these institutions; Paul L. Joskow, Richard Schmalensee, & Elizabeth M. Bailey, The Market for Sulfur Dioxide Emissions, Am. Econ. Rev (in press).


43

This type of flexiblecompliancemechanismrequiresan accuratemethod for measuringemissionsand trackingallowances.Title IV requiresutilities to install continuousemissionsmonitoringequipment,and the EPA's regulationscontainpowerfulfinancialincentivesto ensurethat these monitors areoperatingaccurately. Afterthe 1990 legislationwas passed,EPA set up an Acid RainAdvisory Committeeto assist it in developingthe regulationsrequiredto implement Title IV and to provideadvice on interpretingthe statutorylanguage.The EPAalso createdthreeinternalteamsto come to a consensusinterpretation of the complex and interrelatedallowanceallocationprovisionscontained in Title IV. In orderto recordand defend its decisions, EPA documented thePhase II allocationmethodsin detail and producedthe NationalAllowanceData Base (NADB) and SupplementalData File (SDF).14The NADB andSDF areessentiallylargespreadsheetsthatdisplaythe calculationsused to allocate allowancesto each of 3,842 existing and plannedelectric units and,in orderto do this, provide a good deal of unit-specificinformation fromwhich allocations under alternativerules can be computed.These are the main sourceof datafor the analysisthatfollows. spreadsheets III. HISTORICAL BACKGROUND ON FEDERAL CONTROL OFSO2 EMISSIONS

Thestructureof the 1990 acid rain programcannotbe understoodapart fromthe historyof federaleffortsto limit electricutility emissionsof SO2. The1970 CleanAir Act Amendments,the firstsignificantfederalair pollutionlegislation,led to the establishmentof nationalmaximumstandardsfor ambientconcentrationsof SO2.Stateswere largelyresponsiblefor meeting thesestandardsin each local area.The 1970 Amendmentsalso imposed a new source performancestandard(NSPS) applicableonly to emissions fromnew powerplants,which took effect in 1971. Accordingto the NSPS, theemissionsrate (ER) for newcoal plantscould not exceed 1.2 poundsof sulfurdioxide per million Btus of fuel burned(0.8 pounds/mmBtufor oil). Theseregulationscreateda significantgap between the emissions rates of manyexisting plantsand the ratespermittedfor new plants,thus providing astrongincentiveto extend the lives of old, dirtyplants.15Furthermore, in order to help meet the local ambientSO2standards,the statesrequiredsome 14U.S. Environmental Protection Agency (EPA), Acid Rain Division, Technical Documentationfor Phase II Allowance Allocations (March 1993). Pre-1970 plants were still subject to controls under State Implementation Plans (SIPs) 15 requiredby the Clean Air Act to ensure that each state came into compliance with national ambientair quality standards. There was wide variation among the states in the aggressivenessof their SIPs and how they affected existing plants.

44


existing and new powerplantsto have high smokestacksto disperseemissions over a widerarea.By keepingSO2in the atmospherelonger,however, tall stacks may increase ambientconcentrationsat other locations. They also generallyencouragethe formationof sulfatesandsulfuricacid andthus increasethe total amountof acid deposition,which may affect geographic areashundredsof miles away. Congressnext amendedthe Clean Air Act in 1977. Ambientconcentrations of SO2were again the focus of attention;acid rain was still not an issue. The politicalsolutionthatemergedfromthe 1977 legislationandsubsequentEPA rule makingsatisfiedenvironmentalists, high-sulfurcoal interIt requiredcoal-firedplantsbuilt after 1978 ests, andMidwesternutilities.16 both to meet the ER- 1.2 constraintand either (a) to remove 90 percent of potentialSO2emissions (as determinedby the sulfurcontentof the fuel burned)or (b) to remove70 percentof potentialSO2emissionsandto operate with ER < 0.6. This "percentreduction"standardrequiredall new coal plantsto operatewith flue gas desulfurizationfacilities-generally referred to as "scrubbers"-even if they burnedlow-sulfurcoal.17This provision significantlyreducedthe advantageof low-sulfurcoal as a means of compliance and effectively imposed a lower emissions rate on new sourcesin the West thanin the East.As AckermanandHasslerhave stressed,this provision gave environmentaliststhe tighterNSPS they sought, but it raised the costs of SO2control and may well have dirtiedthe air on balanceby encouragingutilities to burn high-sulfurcoal and by strengtheningincentives to extend the lives of old, dirty plants.18It is also generallyviewed as a victoryfor high-sulfurcoal producersand miners,since the scrubbing provisionsreducedwhat would otherwisehave been a very significanteconomic disadvantagefor high-sulfurcoal. Conversely,of course, Appalachianand,to a lesser extent,Westernproducersof low-sulfurcoal lost. This legislationwas viewed as a victory for most Midwesternand Northeastern coal-burningutilities and their customers,since old plants generally remained relatively lightly controlled,19and slow economic growth meant 16 For more on this episode, see Ackerman & Hassler, supra note 3, Kete, supra note 10, at 158-59; Richard E. Cohen, Washington at Work: Back Rooms and Clean Air (1992), ch. 2; and Paul L. Joskow & Richard Schmalensee, The Political Economy of Market-Based Environmental Policy: The U.S. Acid Rain Program (MIT Ctr. for Energy & Environmental Policy Res., Working Paper 96-003, March 1996), Section 3. 17 Oil-fired units built after 1978 also had to meet the 1971 ER constraint (ER ' 0.8) and to remove 90 percent of potential emissions; they faced no percent reduction requirement if ER < 0.2, however. To avoid simply sending SO2 emissions long distances downwind, the 1977 legislation sharply limited the use of tall smokestacks as a compliance strategy. 18 Ackerman & Hassler, supra note 3. 19 Controls had been imposed on some old plants by state environmental agencies in order to meet ambient SO2 standards. The stringency of these controls varied greatly, however, and


45

therewas little need to build new plants the NSPS.20The big losers were those states, mainly in the West, meeting that were using nearbylow-sulfur coaland growingrapidly.Scrubbing effectively requiredthese statesto engagein costly cleanupof what was alreadyclean coal and to beara disproportionateshareof cleanupcosts because they were buildinga disproportionateshareof new fossil-fueled capacity. TotalU.S. emissions of SO2 peaked in the early 1970s and declined steadilyduringthe 1980s. The focus of Clean Air Act regulationon new generating units,however,servedto extendthe economic lives of old, dirty plantsthat were not burdenedwith significant control costs. As a consequenceof this "new-sourcebias," by 1985 83 percentof powerplantSO2 emissionscame from generatingunits not the 1971 NSPS, and 63 meeting percent were from units with ER ' 2.5.21By 1990, over rain precursorsemittedby powerplantswere emitted two-thirdsof acid by plantsconstructed before 1970.22 Table 1 shows that there were huge interstatedifferencesin aggregate and per capitaemissions in the mid-1980s. Differencesin per capitaemissions reflecteddifferencesin the amountof electricitygenerationper capita (largely reflectingdifferencesin economic and industrial structures),in the use of coal of varioustypes (reflectingaccidentsof geographyandhistory), and in the vintagesof generatingplantsin use. Per capitaemissionstended to be highest in Midwesternstates that had grown relatively little since 1970 and that are locatednearhigh-sulfurcoal deposits.Emissionstended to be lowest in states that had new power plants and had made relatively little use of coal. IV. FEDERALACID RAIN

LEGISLATION23

Acid rain graduallyemergedas a serious environmentaland politicalissue only after1977 becauseof pressuresfrom environmentalgroups,Northeastern states,and, especially,Canadianobjectionsto transborder pollution flows, arisingfrom concernsaboutthe effects of acidic depositionon propwere rarelyif ever as strictas the NSPS they standard.Nonetheless,in partbecauseof these controls, utility SO2emissions declinedsteadilyafterthe mid-1970s,despite increasedcoal consumption. Inlate 1970s, the technologyof the 20 choice for meetingincremental generatingcapacity needs in the East, the South, and portionsof the Midwestwas nuclearpower. These statistics were calculatedusing the 22 National Allowance Data Base, described above. As noted above, Phase I covered only units with ER -> 2.5. large Kete, supra note 22 Joskow &

23

10, at 118.

Schmalensee,supra note 16, treatthis historyin more

depth.

46

THE JOURNAL OF LAW AND ECONOMICS TABLE 1 HIGHEST AND LOWEST BASELINE SO2 EMISSIONSPER CAPITA

Pounds/ Thousands of Tons Capita 1,029.5 550.9 548.4 439.3 427.8 425.6 381.7 347.4 341.0 281.0 221.8 197.9 177.3 155.5 143.0

962.5 1,519.7 126.7 806.6 138.2 2,303.1 957.4 1,037.1 807.2 557.5 69.9 1,174.7 1,013.2 373.1 72.6

State West Virginia Indiana Wyoming* Kentucky North Dakota Ohio Missouri Georgia Tennessee Alabama Delaware Pennsylvania Illinois Wisconsin New Hampshire

Pounds/ Thousands of Tons Capita 53.6 53.5 52.6 44.6 41.8 38.1 32.5 27.9 25.9 20.7 6.8 4.3 .7 .6 .5

42.7 82.7 151.3 396.8 16.6 60.9 68.5 62.8 97.8 12.2 3.3 1.3 .9 .2 6.8

State Utah* Colorado* Virginia New York Montana* Connecticut Louisiana* Washington New Jersey Maine Rhode Island* District of Columbia Oregon Vermont* California*

NOTE.-Emissionsper capitaare baselinesulfurdioxide emissionsin pounds(fromthe NationalAllowanceDataBase) dividedby the averageof 1980 and 1990population.Baselineemissionsfor generating units operatingin 1985 are generallythe productof each unit's 1985 emissionrate and its average 1985-87 fuel consumption.Emissionsin thousandsof tons are baseline generatingunit sulfurdioxide emissionsin thousandsof tons (from the NationalAllowanceData Base). All states with emissions of 500,000 tons or moreare shownexcept for Florida(635.2) andTexas (641.5). The only statewith emissions of 50,000 tons or less not shownis SouthDakota(25.8). * Designatedas a "CleanState"undersec. 406 of the 1990CAAAbecausebaselineaverageemissions rate(ER) fromfossil fuel-firedsteamgeneratingunitsdid not exceed 0.8 poundsper millionBtu. "Clean States"not shown,withpercapitaemissionsin parentheses,arethe following:Arizona(70.50),Arkansas (63.55), Nevada(112.92), New Mexico (101.99), Oklahoma(60.66), and Texas (82.81).

erty, trees, and aquaticlife.24Many acid rainbills were proposedby Western and Northeasternsenatorsand representativesduringthe 1980s.25This 24On the early historyof this issue, see Ackerman& Hassler,supra note 3, at 66; U.S. GeneralAccountingOffice (GAO),An Analysisof Issues concerning"Acid Rain" (Report GAO/RCED-85-13,December1984); and JosephA. Davis, Acid Rain to Get Attentionas ReaganChangesCourse,Cong. Q., March22, 1986, at 675. To help resolve scientificdisputes aboutthe damagescausedby acid rain,Congresscreatedthe NationalAcid Precipitation AssessmentProgram(NAPAP),which spentabout$600 millionthroughthe end of 1990 (NAPAP, 1989 AnnualReportto the Presidentand Congress7 (June 1990)). Its work had no visible effect on the 1990 legislativedebates,however;see Leslie Roberts,Learningfrom an Acid Rain Program,251 Science 1392 (1991), and Acid Rain Program:Mixed Review, 252 Science 1302 (1991). Among the reasonsoffered for NAPAP's lack of impactare its focus on "good science" insteadof policy-relevantanalysisand its lack of politicalsupport from the environmentalcommunity. 25See Cohen, supra note 15, at 36-44; RobertHanley, TurningOff Acid Rain at the Source,N.Y. Times, December11, 1983, at A12; RobertW. Crandall,Air Pollution,Environmentalists,and the Coal Lobby, in The Political Economy of Deregulation:Interest Groupsin the RegulatoryProcess (Roger G. Noll & Bruce M. Owen eds. 1983); and the following pages in the indicatedannualnumbersof the CongressionalQuarterlyAlmanac: 1982, at 425-34; 1983, at 340-41; 1984, at 340-42; 1986, at 137; 1987, at 299-301; 1988, at 142-48.


47

legislationgenerallycalled for reductionsof from 6 to 12 million tons of SO2emissions per year from 1980 levels, targetedthe dirtiestgenerating units for cleanup,and often involved some variantof the ER ' 1.2 constraintthathad been appliedto new units since 1971. In partbecausecosts of cleanupvariedconsiderablyamongexisting units, these proposalsoften providedfor more flexibilitythantraditionalcommandand controlregulationby, for instance,applyingemissionslimits at the statelevel ratherthan unitby unit.26Becausethe costs of these controlstrategieswould have been heavilyconcentratedin a few Midwesternstates,and projectionssuggested thatelectricratestherewould have to risesignificantlyto cover those costs, someof the proposalsincludeda nationalelectricitytax to help to pay for cleanupcosts andto "sharethe pain." Some proposalsincludedmandatory scrubbing,while othersdid not. Duringthe 1980s, Midwesternand Appalachianhigh-sulfurcoal-producingstatesgenerallyopposedany new acid raincontrols,while Westernand Northeasternstates opposed both a nationalelectricitytax and any additionalscrubbingrequirements.Acid rainlegislationwas effectivelyblocked inthe Houseby JohnDingell (D-Mich.),who becamechairmanof the powerfulHouse Energyand CommerceCommitteein 1981. His main concern wasthat any legislationamendingthe Clean Air Act would likely tighten autoemission standardssignificantly,and he accordinglyblocked all such In the Senate,acid rain legislationwas effectively blockedby legislation.27 themajorityleader,RobertByrd (D-W.Va.).West Virginia,with high per capitaemissions of SO2and high productionof high-sulfurcoal burnedin otherstates,was potentiallya big loser from acid rainlegislation.Completingthe constellationof major"JustSay No!" forces on acid rainwas PresidentRonald Reagan, who opposed environmentalregulationgenerally.28 Thepoliticalstrengthof the environmentalmovementgrew dramatically asthe decadeof the 1980s proceeded,fueled in partby the Reaganadministration'sapparentintransigence.29 Populationcontinuedto shift to the West andSouth, and the numberof high-sulfurcoal miners dwindled as high26E. H. Pechan& Associates,Comparisonof Acid Rain ControlBills (EPA ContractNo. 68-WA-0038,Work Assignments94 and 116, OTA ContractLs-5480.0, November1989), six contemporaryacid rainproposals. compare 27 See, for instance,Cohen,supra note 16, at 29-32. 28 Crandall,supra note 3, discusses other obstaclesto assemblinga winningpro-control coalition duringthe 1980s. The SierraClub's membershipincreasedmore than sixfold between 1980 and 1990 29 (personal communicationwith Clubofficials),andthe shareof respondentsagreeingwith the followingstatementincreasedfrom 45 percentto 80 percentbetween 1981 and 1989 (RoSuro,Concernfor the Environment,N.Y. Times, July 2, 1989, at A1): "Protectingthe berto environment is so importantthatrequirementsand standardscannotbe too high, andcontinuingenvironmentalimprovementsmust be maderegardlessof cost."

48


sulfur coal productionfell and productivityimproveddramatically.30 The 1988 presidentialelection was won by GeorgeBush, who had promisedto be "the EnvironmentalPresident"andhad advocatedlooking "to the marketplace for innovative solutions" to environmentalproblems. George Mitchell (D-Maine),an ardentproponentof acid rain controls,succeeded RobertByrd as Senate majorityleader.Even before Bush's inauguration, staff at EPA, in the vice president'soffice, and elsewherein the executive branchbegan work on a set of proposedamendmentsto the Clean Air Act that would deal with acid rain, as well as toxic air pollutants,urbansmog, and otherair qualityissues.31Workon acid rain was heavily influencedby an emissions-tradingproposalthat had been circulatedduring1988 by the EnvironmentalDefense Fund (EDF). Though there were concerns about both the workabilityof the EDF proposaland the size of the emissionsreductionsit required(12 million tons from 1980 levels), relyingon tradable permitsto controlacid rain would respondto Bush's call to look "to the marketplace"and could reducecontrolcosts. Moreover,it was hoped that EDF's supportwould provide protectionagainst knee-jerkantimarketattacks from otherenvironmentalgroups.While some EPA staff clearlypreferredtraditionalcommandand controlmethods,strongsupportdeveloped within the agency, and the basic idea of using tradablepermitsto control acid rain was adoptedby the Bush administrationwithout much internal warfare. The administration'sCleanAir proposalwas announcedin generalterms in June 1989, and draft legislation was released the following month. In the House, the administration'sbill went to the Committeeon Energyand Commerce,still chairedby JohnDingell (D-Mich.).The acid rainTitle was sent to the Subcommitteeon Energy and Power, chairedby Philip Sharp (D-Ind.).Indianahadlargeemissionsfromold dirtyplants,andwhile Sharp had earlierjoined in supportingsome modest acid rain controlproposals, he opposedstringentcontrolstargetedat existing plantswithoutsignificant cost sharingwith other regions. He had advocatedpaying for acid rain abatementthrougha nationalelectricitytax. On the Senate side, the administration'sbill went to the Subcommittee on EnvironmentalProtection,chairedby Max Baucus (D-Mont.), of the Committeeon Environmentand PublicWorks,chairedby QuentinBurdick (D-N.Dak.). Fifteen of the 16 membersof the full committeewere also 30Between 1980 and 1990, the averagedaily employmentof miners in Easternmines (both high-sulfurand low-sulfur)fell from 202,039 to 115,216 (Coal Data (NationalCoal Association,variousyears)). 31For a contemporary view of this process,see MargaretE. Kriz,Politicsin the Air, Nat'l J., May 6, 1989, at 1098.


49

membersof the EnvironmentalProtectionSubcommittee,and Burdickwas not much interestedin environmentalpolicy, so all the action was in the Baucussubcommittee.Montana,a state that both producesand uses lowsulfurcoal, was one of the losers in the 1977 amendments. TheSenateandHouse committeesdiffered substantiallyin regionalcompositionand supportfor environmentallegislation.32 Five of 16 senatorson Environment and Public Works were from New England,where concerns aboutacid rain were high. On the House side, however, New wereoutnumbered41 to 2 on Energy and Commerce.WhileEnglanders the Senate committeehad representationfrom neitherthe states with the highest SO: emissionsnor the largest Easterncoal-producingstates, these states were wellrepresentedon EnergyandCommerce.Only 31 percentof the senators onEnvironmentand PublicWorkswere from stateswith Iplants,as comparedto 56 percentof representativeson (old, dirty)Phase Energyand Commerce.As one mightexpect, the Senatecommitteehad members who were more inclined to supportenvironmentallegislationthan their substantially in the House. counterparts Inmid-November, afterfour days of debateand one day of markup,the SenateCommitteeon Environmentand Public Works approvedClean Air legislation writtenby its staff by a 15-1 vote.33The presidentthreatened to vetothe committee'sbill unless its costs were reduced and thebill was poorly received on the Senate floor. In an substantially, attemptto produce acceptable legislation,the majorityleader,SenatorMitchell,conveneda set ofclosed-doorsessions involving senatorsand administrationofficials bein early February.These meetings were open to all ginning senatorsand their staffs, and stateswith large stakesin the acid rainTitle were well represented when it was discussed. TheSenate negotiatorsmodified the administration's rules for determiningPhase II allowanceendowments. relatively simple They also brought forward the startingdates of both phasesby a year, thus producinggreater Joskow& Schmalensee,supra note 16, providemoreon the 32 points in this paragraph. 33Atthe insistence of Alan Simpson (R-Wyo.),representinga state that both produced and burnedlargequantitiesof low-sulfurcoal, the committeebill to repealthe "percentreduction"provisionof the 1977 containeda provisionpurporting This provision was retainedin the finallegislation,and the EPA was given 3 amendments. As of January1998, however,it had not yet done so. The yearsto producea new NSPS. provisionrequiresthatany new NSPS allow no unit to emit more than it would haverepeal been allowed to emit underthe 1978 NSPS. But this requirementis a "Catch-22," since the 1978 that emissions be less than the sulfur contentof the coal burned NSPS always requires essence of "percent Thus the only way to ensurethata unit emits no more(the removal"). thanit would have been allowed to emit underthe 1978 NSPS is to install a In addition,we are told that state regulationseffectivelyrequirescrubbingin areasscrubber! wherenew coal-firedplantshave been so thattherehas not been strongindustrypressureto revise built, the 1978 NSPS.

50


emissionsreductionsin 1995 and 2000 thanthe administration's proposal.34 In responseto efforts by RobertByrd (D-W.Va.) and other senatorsfrom states producinghigh-sulfurcoal, the incremental1995 reductionswere given back as "bonus" allowances for utilities that installed scrubbers ratherthanswitchingto low-sulfurcoal in Phase I. The incrementalreductions in 2000 were given back over the 2000-2009 periodthrougha number of provisions. A majorissue in these negotiationsand, afteran agreementbetweenthe administration andthe Senateleadershipwas unveiledin earlyMarch1990, on the Senate floor was the so-called Byrd Amendment.This provision would have provided generous financial aid to high-sulfurcoal miners whose jobs were eliminatedby Clean Air legislation. The administration and the Senateleadershipopposedthis amendmentand prevailedby a single vote.35

The bill subsequentlydevelopedin the House also modifiedthe administration'ssimple allocationrules, but it retainedthe originaladministration startdates and ceilings for Phase I and Phase II.36Like the Senate bill, it providedincentivesfor scrubbing.A provisionauthorizingunemployment and job-trainingbenefits for displacedworkerswas added on the House floor;it was not restrictedto minersand had a much smallerprice tag than the Byrd Amendment. The acid rain Title producedby the conferencecommitteewas based 34 See Kete, supra note 10, at 210.

35For discussionsof this episode,see RichardE. Cohen,WhenTitansClashon CleanAir, Nat'l J., April 7, 1990, at 849; and Phil Kuntz& George Hager,Showdownon Clean-Air Bill: SenateSays "No" to Byrd,Cong. Q., March31, 1990, at 983. SenatorByrd called on longstandingrelationshipswith his Democraticcolleaguesand on his power as chairmanof the Appropriations Committee,andDemocratsvotedwith him (andagainstthe Senateleadership)38 to 16. In addition,all RepublicansfromMidwesternandAppalachiancoal-producing states voted for the Byrd Amendment,except for SenatorWarner(R-Va.), despite strong administrationand Republicanleadershipopposition.Finally, SenatorsCochrane(R-Miss.) and McClure(R-Idaho)voted for the Byrd Amendment,and thus againstboth the White House and the RepublicanSenate leadership,even thoughthey representedno high-sulfur coal miners.SenatorSymms(R-Idaho)was talkedout of doing likewise only in the last minute of voting. Given that these three senatorswere in the bottom 20 in termsof the AFLCIO's evaluationsof lifetime voting records,it seems unlikely that they were casting prolabor votes for ideologicalreasons.Strategicmotives are suggestedby the facts that these threesenatorswere in the bottom10 in termsof the Leagueof ConservationVoters'ratings of 1989-90 voting recordsand thatthe presidenthad threatenedto veto any legislationcontainingthe Byrd Amendment.It is most plausiblethatthese senatorshoped thatpassageof the Byrd Amendmentwould force the presidentto carryout his veto threatand thus likely kill new clean air legislation. 36For a detailedcomparisonof the acid rainprovisionsof the House and Senatebills, see ICFResources,Inc. (ICF),Comparisonof the EconomicImpactsof the Acid RainProvisions of the SenateBill (S. 1630) and the House Bill (S. 1630) (draftreportpreparedfor the U.S. EnvironmentalProtectionAgency, July 1990).


51

mainlyon the Senatebill, while the House prevailedon most of the rest of the legislation.37A relativelysmall provisionfor aiding displacedworkers and the Senate'sprovisions (Title XI) based on the House bill was added,38 for allocatingallowanceswere modified. The provisions of the law allocating allowances in Phase I remained fairly simple, but 8 dense pages of about30 complex and convolutedprovisions were developed to govern Phase II allocations.In orderto ensure that the intendedconstraintson total Phase II emissions were satisfiedin the face of a risingtide of proposedspecialinterestprovisions,work on the Senate/administrationbill had quickly incorporated an overarching "ratchet"provision.This provision,which was not controversialand was retainedin the finallegislation,in effect said thatat the end of the day total allocationsunder(almost) all otherprovisionswould be scaled down to a specified total.39 This provision had the effect of making negotiations about

allowance allocationsinto a zero-sumgame. It also implied that, all else equal,benefitsfromrules changesthatwould decrease somebodyelse's allowanceswould be widely shared.Thus, at least in the negotiationsthatled up to the passage of the Senate bill, debates about allowance allocations were conductedprimarilyin termsof proposedrules for increasingparticular allocations,which were typicallysupportedby argumentsaboutfairness underautarchy. V.

PHASEI ALLOWANCE ALLOCATIONS

Table A of the 1990 CAAA lists the Phase I units and specifies allowances to be allocatedto each. Eight of these units were addedto table A before the Senate bill was passed in April. Though these additionswere justified by technicalcorrectionsto earlierwork, it is interestingthat five of the eight units, accountingfor 84 percentof the allowancesallocatedto these units,were locatedin Minnesota,New York,andWisconsin.The two Wisconsinunits had been retiredin 1988, so that addingthem to Phase I 37 See Cohen, supra note 15, ch. 10; and Alyson Pytte, A Decade's AcrimonyLifted in the Glow of CleanAir, Cong. Q., October27, 1990, at 3587. 38Of the total authorizationof $250 million, less than$29 million had been spent for all displacedworkersas of May 1995 (telephoneinterview,EmploymentandTrainingAdministration,U.S. Departmentof Labor). 39When the EPA figuredout the allocationsrequiredby the statutein 1992, the ratchet's operationreducedPhaseII allowancesby about9.6 percentfrom the total impliedby strict applicationof the otherallocationprovisionsin the bill. The "ratchet"had the effect of reducingannualPhaseII "basic" allowancesfrom9.876 to 8.90 milliontons andthus,if "bonus" allowancesare takenas fixed, of reducingtotal annualallowancesallocatedin the first 10 yearsof PhaseII from 10.115 to 9.139 milliontons (EPA, supranote 14, at 5). The large size of the "ratchet"announcedin early 1992 was a greatsurpriseto those involved in the process,most of whom had expecteda ratchetof less than5 percent.

52


clearly made their owners better off by the value of the allowances they were given. Since all three of these states had significant acid rain legislation on the books by 1990, adding the Minnesota and New York units to Phase I and using their 1985 emissions rates to determine their allowance allocations probably made the owners of these units better off as well. Most commentators describe the annual table A allowance allocations as equal to emissions (in tons) from baseline fuel use and an ER of 2.5 pounds per million Btus.40The EPA's NADB reveals that this formula is only approximately correct. Moreover, table A does not fully describe the Phase I allocations. The table A allocations differ by more than 1 percent from those produced by applying this formula to the NADB data, which we will call the Basic Rule, in 44 cases. The absolute value of the difference exceeded 1,000 allowances per year for 17 units.41Most, but not all, of these differences reflect departuresfrom the administration's original table A proposal. In large part, at least, these differences reflect the fact that the NADB contains more recent data than those employed in computations underlying table A. Table 2 shows that at the state level, the table A allocations of Wisconsin, Indiana, and Missouri are well above those implied by the 2.5 ton/ mmBtu Basic Rule, while Pennsylvania's is noticeably lower. The third column of Table 2 also shows the effects of two "bonus" provisions: Section 404(h), which affected one unit in Iowa, and Section 404(a)(3), which affected all units in Illinois, Indiana, and Ohio except for three plants that sell mainly to Department of Energy uranium-processing plants.42This latter provision was added in response to Midwestern pressures for some form of cost sharing-with care taken not to allocate valuable allowances to plants that sold electricity under cost plus arrangements back to federal facilities. It is worth noting as well that representatives from Indiana and Illinois were chairman and ranking member, respectively, of key House subcommittees. Ohio had the highest total emissions of any state in the country, and two representatives on the House Energy and Power Subcommittee, one of 40 See, for instance, Reinier Lock & Dennis P. Harkawik, eds., The New Clean Air Act: Compliance and Opportunity, at 24 (1991). 41 This does not count large, almost exactly offsetting differences for units 1 and 2 of Georgia Power's Bowen plant. These differences seem almost certain to reflect some sort of error or the correction of another sort of error. 42 Hausker, supra note 13, at 567, notes that the Midwest bonus provision, added late in conference, was the single breach in the zero-sum barrier imposed by the ratchet provision. A third bonus provision, Section 404(e), would provide Phase I allowances to Union Electric in Missouri and Phase II allowances to both Union Electric and Duke Power. Because of pending litigation, however, no allowances had been issued under this provision by late 1995. Only about 4,000 allowances had been allocated under a final bonus provision, Section 404(f), which rewards using conservation or renewable energy to reduce emissions.

TABLE 2 PHASE I ALLOWANCE ALLOCATIONSBY STATE

State Alabama Florida Georgia Illinois Indiana Iowa Kansas Kentucky Maryland Michigan Minnesota Mississippi Missouri New Hampshire New Jersey New York Ohio Pennsylvania Tennessee West Virginia Wisconsin Total

Basic Rule

Table A Difference

230,947 129,792 581,599 353,191 640,855 37,555 4,226 278,637 140,066 42,334 4,409 54,609 345,101 32,207 20,811 147,393 863,191 536,121 386,183 496,528 130,004 5,455,761

Bonuses

Final Allocation

-7 3,338 1 4,709 9,485 2,735 -6 -387 -526 6 -139 1 7,889 -17 -31 3,587 89 -1,981 247 1,342 13,376

0 0 0 36,356 66,724 1,350 0 0 0 0 0 0 0 0 0 0 96,920 0 0 0 0

230,940 133,130 581,600 394,256 717,064 41,640 4,220 278,250 139,540 42,340 4,270 54,610 352,990 32,190 20,780 150,980 960,200 534,140 386,430 497,870 143,380

43,709

201,350

5,700,820

NOTE.-Basic Rule allocationsare computedby multiplyingan emissionsrateof 2.5 times baselinefuel consum differencebetweenthe table A and Basic Rule allocations.Bonuses are from sec. 404(h) for the GeorgeNeal Nort units in Illinois, Indiana,and Ohio. The second to last column is the final allocationminus (Basic Rule rescaledt percentageof rescaledBasic Rule. The final column gives annualaveragePhase 1 extensionallowancesreceived, utilities.

54

JOURNAL OF LAW AND ECONOMICS THE

whom(ThomasLuken,D) had been heavily involved in debatesover acid rainlegislationproposedearlierin the 1980s. Thesecond to last column in Table 2 shows a comparisonof the Final Allocationin the 1990 law with the Basic Rule allocationscaled up to a totalof 5,700,820 tons. None of the differencesexceed 10 percentof the BasicRule benchmark,but at an expectedprice of $200/ton, a thousandtonannualdifferencecorrespondsto a million dollarsa year. The positive differences(gains) are much more concentratedthan the negative differences (losses):the top threegainers (Ohio,Indiana,and Illinois) accounted for93 percentof totalgains, while the top threelosers (Georgia,Pennsylvania,and West Virginia)accountedfor 52 percentof the losses. Only four stateshave positive differencesin excess of 5,000 tons per year,while eight stateshave negativedifferencesof that magnitude. Onehypothesisthat explains some of these differencesis that the states thatburnedmore coal thanthey produced,includingIndianaand Ohio, focusedtheir attentionon acquiringadditionalallowances,while the states thatproducedmorethanthey burned,includingPennsylvaniaandWest Virginia,focused theirattentionon providingincentivesfor scrubbingand directfinancialbenefitsfor displacedcoal miners.However,Georgia,which producedno coal, and Illinois, which producedabouttwice as much as it burnedin 1990, conspicuouslyfail to fit this pattern.43 Thelast columnin Table 2 shows the effects of a final importantPhase Iprovision.In responseto pressurefromhigh-sulfurcoal states,3.5 million allowances(the "gain" from moving the start of Phase I from 1996 to 1995) were set aside to encourage the use of "technology" (that is, as an emissionsreductiontechniquein preferenceto fuel switchscrubbers) ing.The mainbeneficiariesof this provisionwere utilitiesin Ohio, Tennessee, Indiana,and West Virginiaand the high-sulfurcoal interestsin these andnearbystates.44 43Both states hadhigh totalemissionsandhigh emissionsrates.Georgiahadno representationon the SenateCommitteeon Environmentand PublicWorksor the Senateleadership. Onthe House side it had nobodyon the House Energyand Power Subcommitteeand only onejunior member(J. Roy Rowland,D) on the full Energyand CommerceCommittee.It wasrepresentedon the House leadershipby the MinorityWhip (Newt Gingrich,R). Illinois on the Senateside, butit had the rankingmemberof also hadno well-placedrepresentation Subcommittee (EdwardMadigan,R), a memberof both inEnvironment and Health the volved subcommittees(TerryBruce,D), and a thirdmemberof the full committee(Cardiss Collins,D). As we discuss below, Georgiafaredpoorlyin both PhaseI and PhaseII, while Illinoisfaredwell in both phases.Illinois would have appearedto have done even betterin PhaseI if IllinoisPowerhad appliedfor scrubberextensionallowances,as hadbeen expected whenTitle IV was being debated. 44Again, Georgia got nothingout of the scrubberbonus allowances,but it also had no from the scrubbingincentives, high-sulfurcoal minersto protect.Illinois minersbenefited and it was generallyexpectedin 1990 that Illinois utilities would apply for a large number


55

Even thoughthe Phase I allowanceallocationshave generallybeen describedas following a simplerule, it is clearthatthe actualallocationswere significantlyinfluencedby special interestrent seeking. In additionto the differencesbetweenthe table A allocationsand those impliedby the Basic Rule, large special allocationsof allowanceswere given to threeof the five stateswith the highest SO2emissions(see Table 1): Ohio, Indiana,and Illinois. All threealso had substantialhigh-sulfurcoal mininginterests.Pennsylvania,West Virginia,and Kentucky,which had both relativelyhigh aggregate emissions and more importanthigh-sulfurmining interests,were not coveredby this special provision.However,these six states plus Tennessee (with emissionsjust above Kentucky'sbut few high-sulfurminers) acquiredalmost all the bonus allowancesmade availableto Phase I units that chose to reduce emissions by scrubbing.Georgia,number4 in emissions, benefitedfrom neitherthe special allocationnor the scrubberbonus. This patternsuggeststhatPhaseI allowanceswere used partiallyto compensatethreeof the high emissions statesthatwere well representedon the key committees.Phase I allowanceswere also used to subsidizescrubbers in response to high-sulfurcoal-mininginterests.There is some evidence that the states with importantcoal-mininginterestsfocused more on increasing scrubber allowances than on securing earmarkedallowances. Georgia,whichwas not representedon the relevantcommittees,did particularly badly overall in the Phase I allowanceallocationprocess. VI.

STATUTORY PROVISIONS FORPHASEII ALLOCATIONS

Calculationsof a generatingunit's Phase II allowanceallocationsgenerally begin with "baseline" emissions, determinedby the recordedemissions rate for 1985 and averageheat inputfrom fuel burnedduring198587.45The simple rule at the core of Phase II allocatesallowancesequal to each unit's baseline fuel use times the lesser of its actual 1985 emissions rate and 1.2 poundsof SO2per million Btu, expressedin tons. The statute contains over 30 individualprovisionsthat specify exactly how Phase II allowancesareto be allocated.These rulesfall into threegeneralcategories. The first category contains provisions that specify variationsfrom the simple rule above based on fuel type, unit age, unit capacity,and capacity of extensionallowances.Illinois Power eventuallydecidednot to install scrubbersbecause its regulatorsrefused to preapprovesuch investmentsfor rate-makingpurposes. Illinois Powerthen became an importantearly purchaserof allowances. 45 Special provisionswere includedfor units that were not in operationin 1985 or were still underconstructionin 1990 when the Act was passed. Emissionsrates for 1985 were used in all othercases becauseNAPAP(see note 24 supra)hadconstructedparticularlygood emissions datafor that year.

56


utilizationduringthe base period.These allocationrules were generallyadvertisedas dealingwith various"technicalissues" associatedwith the fuel and operatingattributesof units in these categoriesduringthe base period. These rules includespecial provisionsfor units thatoperatedat low capacity factorsduringthe baselineperioddue to mechanicalproblemsor unusually low demandalong with special allocationsfor small coal plants for whichcontroloptionswereparticularlylimitedandcostly. Other"technical arguments"supporting,for example,a special allocationfor units thathappenedto burnlots of gas duringthe baselineperiodbecausegas priceswere unusuallylow duringthattime, are more difficultto acceptas being "nonpolitical."As we show below, the allocationrulesin this firstcategorygenerally shift allowancesfrom relativelydirtystatesto relativelyclean states, especiallythose with oil/gas generatingunits. The second categoryof allocationrules consists of those rules that are narrowlyfocusedon specialinterests-either individualstatesor individual utilities.Table 3 providesthe clearestexamples.This table was developed by categorizingall Phase II units by applicableallocationrules and then searchingfor rules that appearedto be narrowlyfocused on a single state or a small numberof generatingunits. Table 3 should remove any doubt thatinterestgrouppolitics was at workin the developmentof the U.S. acid rain program.Some of these provisionsare clearly the work of influential of the Committeeon Enlegislators.SenatorBurdickused his chairmanship in NorthDakota to that his constituents vironmentandPublicWorks ensure for the units that generateelectricity lignite-fired got special allocations there by inserting Section 405(b)(3). In addition, CongressmanDingell seems to have providedregulatoryrelief for DetroitEdisonthroughSection 405(I)(2). It is moredifficultto relatesome of the otherprovisionsin Table3 directly to well-positionedcongressmenfrom the states that benefitedfrom them. Floridawas not representedin the leadershipof eitherHouse or Senate,and SenatorBob Graham(D-Fla.) and CongressmanMichaelBilirakis(R-Fla.) were the only Floridianson the relevantcommittees.Nonetheless,Senator Grahammanagedto securethousandsof incrementalallowancesfor Florida throughSection405(I)(1).46Section405(c)(3) originatedin the House, even thoughSpringfield,Missouri,was representedby a first-termRepublicannot 46 See Kete, supra note 10, at 207-10. The impact of this provision is capped in the statute at 40,000 allowances annually. Florida may have been treated well in part because it was a large state with competitive races for both senator and governor in prospect for the fall of 1990. (See Table 8 below for the definition of "competitive" used here.) At least one of the other Florida-specific provisions in Table 3 was added to the Senate bill at the insistence of the Republican leadership to give Florida's other senator, Connie Mack (R), something for which he could also claim credit.


57

TABLE3 II (2000-2009) PROVISIONS PHASE OFSELECTED SPECIAL INCIDENCE

Section 404(h)

Coverage Phase I units 1990 ER < 1.0, -60% ER drop since 1980; system ER
1.2 in a state with no nonattainment areas 4 Florida(TampaElectric) State has >30 million KW capacity; unit barredfrom oil use, switched to coal between 1/1/80 and 12/31/85 2 Missouri(City of Springfield) Small unit, ER 2 1.2, on line before 12/31/65;system fossil steam capacity>250 MW and 20% scrubbed,rely on small Power) units, have large units expensive to scrub 1 Florida (City of Tallahassee) Oil/gas units awardeda clean coal technologygrantas of 1/1/91 48 New York (Consolidated EdiOperatedby a utilityprovidingelecson, Power Authority of tricity,steam,and naturalgas to a the State of New York) city and one contiguouscounty;or state authorityservingsame area 3 Arizona (Tucson Electric), Units covertedfrom gas to coal New York (Orange & between 1/1/85 and 12/31/87 with Rockland Utilities) proposedor final prohibitionorder States with >25% populationgrowth 134 Florida 1980-88 and 1988 electricgenerating capacity>30 million KW 6 Florida (Florida Power ComLargeunits with reducedactualor allowableemissionsmeetingfive pany), Michigan (Detroit conditionson emissions and Edison) growth

on Energyand Commerce.Section 405(g)(5) was broadenedin conference to includeTucsonElectric,even thoughArizonawas not representedon the conferencecommittee.47 Finally,Section404(h)originatedin theHouse,even thoughthe only Iowanon Energyand Commerce,Tom Tauke(R), was not on Energyand Power and was campaigningvigorously(thoughultimately unsuccessfully)againstan incumbentDemocraticsenator. 47MorrisUdall (D) of Arizonawas appointedto the conference,but specificallyto deal with issues otherthanacid rain;Cong. Record,June6, 1990, at S-7541.

58


These examplesmake it clear thatthe abilityof a utilityto obtainfavorable PhaseII allocationprovisionsin the statutedid not necessarilydepend on having one or more of the membersof its state's congressionaldelegation on a key committeeor in the leadership.States like Floridawere of "partisan"politicalimportancebecause of the presenceof close races for the Senateor governoror theirexpectedimportancein the next presidential election. Utilities could also gain influence with influentialmembers of Congressrepresentingother areas throughtheir trade associations,PACs, and political contributions.The existence of these alternativepathways throughwhich legislatorscan be influencedis consistentwith the difficulty scholarshave had in finding strong empiricallinkages between congresand the concentrationof interestgroupsin particular sional appropriations states and the seniorityand committeeassignmentsof theirrepresentatives in Congress.48 We encountersimilardifficultyin the regressionanalysisdiscussed below. As comparedwith legislationin otherareas,we do not believe thatthere is anythingunusualaboutthe provisionsin Table 3. The EPA data simply makeit easierto identifybeneficiariesof these rules thanof, say, functionally equivalentprovisionsin the tax code. Nor do we believe these are the only "special interest"allocationrules includedin Title IV-just the most obvious. For example, Section 405(f)(1) provides special bonuses for oil/ gas units with very low emissionsratesduringthe baselineperiod.Units in over 30 statesget some benefitfromthis provision,but the bulk of the benefits are concentratedin California,Florida,and New York. The thirdcategoryof Phase II allocationrules providesfor generalallocations of bonus allowancesto units located in groups of states that fall neatly into the "clean" and "dirty" camps. As we discussedabove, Section 405(a)(3) allocates50,000 additionalallowanceseach year to Phase I units located in 10 "dirty" Midwesternstates. Section 406 made 125,000 allowancesper yearavailableto unitsin "clean states,"whichthe governor of any of these states could access at his option in lieu of acceptingother bonus allowancesto which the units were entitled. (See Table 1 for the definitionand list of "clean states.") These allocationsclearly reflect efforts to "buy off" two well-organizedgroupsof stateswith utilities at opposite ends of the dirty/cleanspectrum. VII.

ALTERNATIVE ALLOCATION RULES AND THE DISTRIBUTION OF PHASE II ALLOWANCES

Given the numberand complexityof Phase II allocationrules, interactions betweenthem, and the global ratchet,it does not appeareitherpracti48 See Levitt & Poterba, supra note 5, and the references they cite.

MARKET-BASED ENVIRONMENTAL

POLICY

59

TABLE 4 INITIALPHASEII ALLOWANCE ALLOCATIONS CONSIDERED CORRELATIONWITH STATES' FINAL ALLOCATIONS

ALLOCATION

CODE

Total

Per Capita

Proportional Reduction

PR

.882

.811

Simple Rule

SR

.989

.985

Base Case

BC

.996

.991

Final Allocation

FA

Cost Minimization

CM

DESCRIPTION Baseline emissions ratcheted down by 42.3% to equal total Phase II allowances (i.e., the total in the Final Allocation) (1) Units on line before 1986 receive (1985 heat input) x Max (1985 ER, 1.2), expressed in tons; (2) units on line in 1986 or later receive unratcheted basic allowances per Section 405(g); (3) allocations are ratcheted up by 8.5% to equal total Phase II allowances (1) Allowances are allocated using basic provisions in the law that distinguish units by baseline emissions rate, fuel type, and vintage (for units on line in 1986 or later) as described in note 51; (2) allocations are ratcheted down by 1.4% to equal total Phase II allowances Actual allocation of Phase II allowances, as provided for in the law

.956

.887

Allocation of allowances that minimizes estimated total compliance cost in 2005 on the assumption that transactions costs rule out interstate trading; linear state-level marginal cost curves estimated from table A-16 in ICF, supra note 36, assuming intercepts are $115, as described in note 52 infra

cal or interestingto use the EPA datato try to sort out the effects of each individualprovisionas we did for PhaseI in Table 2. Nor is thereany simple, systematicway to tie these provisionsto specificinterestgroupsor legislators,since there are no votes to observeeitheron individualprovisions or on the acid rainTitle itself in isolationfromthe rest of the 1990 Amendments.Instead,we have elected to structureour analysisaroundthe allocation patternsproducedby the statuteand by the four benchmarkalternative allocationrules (PR, SR, BC, and CM) definedin Table4 and discussedin

60


more detail in the next severalparagraphs.We performa varietyof direct comparisonsin this section andthen use hypotheticalvoting andregression techniquesfor furtheranalysisin Sections VIII and IX, respectively. A. AlternativeAllocations The ProportionalReduction(PR) allocationis a naturalstartingpointfor most academicdiscussions,thoughit has been foundto lack attractivedistributionalpropertiesin several contexts.49The PR allocationimplies that in the absenceof interstatetrading,all states would reducetheiremissions by the sameproportionto achievethe PhaseII emissionscap. This rule implicitly ignoresthe fact thatsome stateswere alreadyclean, andthese states generallyfaced relativelyhigh abatementcosts.50 The SimpleRule(SR) resemblesthe core rule of the initialadministration bill as well as some earlierproposals.Like those bills, it reflectsthe maximumemissionsratefor new coal sources(ignoringthe "percentreduction" requirement)in effect since the 1970 CleanAir Act Amendments.Eachunit operatingin 1985 is initially allocatedallowancesequalto its baselinefuel use times the lesser of its actualemissionsrate and 1.2, expressedin tons. This allocationrule leads to significantlylower aggregateallowancesthan is providedfor by Title IV. Thus,these initialallocationsare thenratcheted up by 8.5 percentso thattotal allowancesunderSR equal the actualPhase II cap. The basic idea is to bring old generatingunits, which accountfor the bulk of SO2emissions, into conformitywith the 1971 NSPS in aggregate. Becausethe ratchetup from this basic principleto the actualPhaseII allocationsis so large,following this rulemakesit possibleto meet the statutory emissions limit by allowing all existing coal units to operatewith emissions rates substantiallyabove 1.2 poundsper million Btu on average and to provideall otherunits with allowanceswell in excess of theirbaseline emissions. The Base Case (BC) was producedby using the six basic provisionsin the finallaw thatdistinguishunitsby baselineemissionsrate,fuel type, and age-what we referredto in Section VI as the first categoryof allocation rules.51Ouroriginalidea was thatdifferencesbetweenthe SR and BC allo49 See Tietenberg, supra note 1, ch. 5.

50See, for instance,Crandall,supranote 3, at 27. 51Base Case allowanceswere allocatedas follows before ratchetingdown, dividing the resultsof these formulasby 2,000 to convertto tons. (1) All units that began operationin 1985 or earlierandhad ER > 1.2 receivedbaselinefuel use (in Btus) X 1.2, following Section 405(b)(1). (2) All units thatbeganoperationin 1985 or earlierandhad 0.6 ' ER < 1.2 received baseline fuel use X min [actual 1985 ER, maximumallowable 1985 ER] X 1.2, followingSection405(d)(2).(3) All otherunits(withER < 0.6) thatbeganoperationin 1985 or earlier,except units that derived more than 90 percentof their total fuel consumption (on a Btu basis) from gas during1980-89 (the ">90 percentgas" units),receivedbaseline


61

cations would have primarilytechnicalrationales,with political influences affecting primarilythe difference between BC and the Final Allocation (FA) actuallyemployed.As we noted above and will demonstratebelow, realitywas not so tidy. The high pairwisecorrelationsbetweenFA andeach of PR, SR, and BC shown in Table 4 reflectthe huge interstatedifferences in emissions levels. Finally, we used preenactment,state-level compliance cost estimates from a widely circulatedreportpreparedto informthe legislative process to producean estimateof the allowanceallocationthat would have miniOther mized total compliancecosts in the absence of interstatetrading.52 cost analyseswere also developedfor and consideredin the legislativeprocess and would, of course, imply differentcost-minimizingallocations,so the Cost Minimization(CM) allocationconsideredhere is more illustrative thandefinitive.This allocationis of interestbothbecauseof actualandperceived market imperfections,53 and because autarchywas implicitly assumedin muchof the actualdebateabout"fair" allowanceallocations.Table 4 shows that the CM allocationis also highly correlatedwith the FA allocation,againreflectingthe importanceof baselineinterstatedifferences. Many in Congressseemed to believe thattherewould be significantobstacles to interstateallowancetrading.It is thus of some interestto use our estimatedmarginalcost functions,along with consistentestimatesof unfuel use X min [0.6, maximumallowable 1985 ER] X 1.2, following Section405(d)(1). (4) All >90 percent gas units received baseline fuel use X 1985 ER, following Section 405(h)(1). (5) Units that began operationbetween 1986 and 1990 received estimatedfuel consumptionat a 65 percentoperatingfactor X the unit's maximumallowable 1985 ER, following Section 405(g)(1). Finally, (6) all coveredunits underconstructionand expected to begin operationafter 1990 receivedestimatedfuel consumptionat a 65 percentoperating factor X min [0.3, the unit's maximumallowableER], following Sections 405(g)(3) and 405(g)(4). 52TableA-16 in ICF,supranote 36, containsstate-by-stateestimatesof emissionsin 2005 (a) with no controlsand (b) with a common marginalcost of control ($572/ton) that was projectedto reducetotal emissionsto nearthe actualPhaseII cap. These dataimply a point on each state's estimatedmarginalcost of abatementschedulefor 2005. To determinethose schedulesfully, we assumelinearityand a commonintercept.TableA-16 in ICF, supranote 36, gives the total cost of controlfor the case analyzed,includingthe cost of reducingutility NOxemissions by 10.556 million tons. Comparingthe total cost of SO2controlimpliedby an assumedinterceptvalue with the ICF total cost gives an implied cost per ton of NOx reductions.An interceptvalue of $115 gives a cost per ton in the centerof the range discussedby ICF,supranote 36, at C-12. (Interceptvaluesof $80 and$150 yieldedvery similar results;see Joskow & Schmalensee,supra note 16, table 5.) The CM allocationwas then computedby equatingestimatedmarginalcosts acrossstatesandsettingtotalemissionsequal to the Phase II cap. ICF, supra note 36, table A-16, projectedCaliforniaand Vermontto have zero SO2emissionsin 2005 even with no controls;they receivedzero allowancesunder CM. At the other extreme,Oregon and the Districtof Columbiawere projectedto find it uneconomicto reduce emissions at all, even at an allowanceprice of $572/ton; their CM allocationsequal baseline2005 emissions. 53 Stavins, supra note 2.

62

THEJOURNAL OF LAW AND ECONOMICS

controlledemissions,54to estimate the total expected allocation-specific costs in 2005 in the absenceof interstatetrading.These calculacompliance tionsimply thatPR would impose costs about30 percentabove theirminimumvalue, while SR, BC, and FA are estimatedto involve total cost between5 and 10 percentabove the minimum.These latterdifferencesseem unlikelyto be much above the noise in this exercise. These resultssuggest thatin the presenceof transactionscosts therewas at least a plausibleefficiencycase for rejecting PR in favor of any of the other allocations.55 B. Gainersand Losers by Typeof GeneratingUnit Sinceutility service areas do not map easily into House districts,and sincethe Senatehad somewhatmore influenceon the final allowanceallocationsthan the House, states are the naturalunits of political economic analysis.However,most of the Phase II allocationprovisionsdo not relate directlyto states but, rather,to generatingunits with differentattributes. Thedistributionof differenttypes of generatingunits among the states is thusthe maindeterminant,as a matterof arithmeticratherthanof causality, of the effects of differentallocationrules on individualstates. To understandthe lattereffects, we begin with an analysisof how those rules treat generatingunits with differentcharacteristics. Table 5 summarizesbaselinesemissions by and allowancesallocatedto unitsof varioustypes in PhaseII underSR, BC, andFA.56Under generating SR,"dirty" units are allocatedallowancesequal to only about40 percent of baseline emissions, while both "moderate"and "clean" units receive allocationsabove their baselines.If the Phase II allowanceallocationprocesshad been used partiallyto "buy off" the stateswith manydirtygeneratingunits, which were the main targetsof the whole acid rain program, we would have expected to see allowances allocatedto dirty units to be increasedas we move from SR to BC and from BC to FA. Table 5 shows exactlythe opposite:bothmoves decreasethe aggregateallowancesof dirty units,particularlylarge, very dirty units. (Table A units fall in this cate54From ICF, supra note 36, table A-16.

55One often-invokedprincipleof equity is equalityof sacrifice.Using the coefficientof variationof states'estimatedper capitacompliancecosts as a measureof inequalityof sacrifice, PR is estimatedto involve substantiallyless inequalitythan the otherfour allocations and SR have the (Joskow& Schmalensee,supranote 16, table 5). Using this measure,CM least inequalityof sacrificeand FA the most. While these estimatesmust clearlybe treated with considerablecaution,the outcomeof the politicalprocesssuggeststhatthe equityconsiderationsthatdrove it are not well summarizedby equalityof sacrifice. 56Since the calculationsleadingto the CM allocationcould only be done at the statelevel, a breakdownof this allocationby unittype is not possible.PR allocationsaredirectlyproportionalto baselineemissions.

63

MARKET-BASEDENVIRONMENTALPOLICY TABLE 5 EMISSIONSAND PHASE II ALLOWANCE ALLOCATIONSBY UNIT TYPE IMPLIED INITIAL ALLOWANCE ALLOCATIONS

UNIT TYPE Dirty: 1.2 ' ER: ER - 2.5, 2 75 MW Other dirty Moderate: .6 ' ER < 1.2 Clean: ER < .6: Coal Oil/gas Gas (>90%) New: Came on line 1986-90 Other: Planned, exempt, etc. Total

BASELINE Simple EMISSIONS Rule (SR)

Base Case (BC)

Final Allocation (FA)

13,004 9,451 3,553 2,793 363 298 61 4 230 305

5,375 2,901 2,465 2,881 394 323 67 5 250 239

4,887 2,645 2,242 3,107 772 475 292 4 238 134

4,745 2,412 2,333 3,186 864 510 303 50 209 135

16,695

9,139

9,139

9,139

NOTE.-Emissionsand allowancesare expressedin thousandsof tons of SO2.ER is the baseline emissionsratein poundsof SO2emittedper million Btu of fuel burned.Baselineemissionsgenerally equal (1985 emissionsrate x 1985-87 averagefuel use) for all units on line in 1985.

gory.) Moreover,all other unit types on line by 1985 receive allowances underFA thatin aggregateexceed theirbaselineemissions.This patternis consistentwith "We're alreadyclean, don't pick on us!" having been a moreeffective equityargumentthanany notionof equal sacrifice.It is also consistentwith a desireof senatorsfromthe "clean" Westernstatesto pay back the Midwesternand Appalachianstates for the mandatoryscrubbing provisionin the 1977 law.57Finally, along with the resultsof Section V, it is also broadlyconsistentwith high-emissionsstatesbeing willing to accept fewer Phase II allowancesin returnfor more Phase I allowances. Cleanunits do muchbetterunderBC thanunderSR.58The higherallocation to clean units mainlyrepresentsa gain by clean oil/gas units. The formula involved was nominally a response to an argumentthat these units had burnedan "unusual" amountof gas in the base period, so that their baseline emissions were "abnormally"-and thus "unfairly"-low. In connectionwith both this provisionand the "clean states" provisiondiscussed below, it is instructiveto considerthe possible role of SenatorBen57 See Margaret E. Kriz, Dunning the Midwest, Nat'l J., April 14, 1990, at 893, on this point. 58 The lower allocation to "Other" under BC than under SR is an artifact; it primarily reflects a legislative decision t6 exempt some cogenerators and other units from the program altogether.

64


nett Johnston (D-La.) in promotingprovisions favorable to gas-burning units.SenatorJohnstonrepresenteda majorgas-producingandgas-consuming stateandchairedthe EnergyandNaturalResourcesCommittee.Though this committeehad broadoversightauthorityfor federaleconomic regulation of electric utilities and could have plausiblyassertedjurisdictionover aspects of the 1990 legislation,it did not do so. Moreover,it would have been naturalfor gas-burningelectricutilitieswithoutmore directinfluence on the relevantcommitteesto turnto SenatorJohnstonfor assistance. DifferencesbetweenBC and FA reflectmore than a score of otherprovisions, some of which appearin Table 3. Their most strikingimplication in Table 5 is the huge increasein allowancesfor units burningmore than 90 percentgas. This resultsmainlyfromthe "clean states" provision,Section 406, discussedabove. This provisionallocateda pool of bonus allowances to units in "clean" states in proportionto generation,not baseline emissions.

Table 5 shows that only the dirtiestlarge units did less well underFA than underBC, even thoughtheir FA endowmentswere increasedby explicit bonusesfor Phase I units. Small dirtyunits ( SR: None |A| > 5% |A| > 10% PR -> CM: None IAI > 5% lAl > 10% SR -- BC: None |A| > 5% IAI- 10% SR -- FA: None lAl > 5% |A|- 10% BC -- FA: None |Al - 5% IAI > 10% CM -> FA: None I|A| - 5% IAI- 10%

DROPPED

VOTES ON PHASE II ALLOWANCE

.lv~.

o

z

rA 't't

ALLOCATION

CH

T,rtT n 11V i^ulr Urt

;

Yea

Nay

Margin

Yea

Nay

Margin

0 3 7

70 68 62

24 20 18

46 48 44

303 302 257

126 102 99

177 200 158

0 1 6

62 60 54

32 32 28

30 28 26

235 228 199

194 194 184

41 34 15

0 15 31

48 40 28

46 24 6

2 16 22

220 178 147

209 91 3

11 87 144

0 15 22

50 40 34

44 24 18

6 16 16

209 170 158

220 113 75

- 11 57 83

0 19 34

42 28 18

52 28 8

-10 0 10

207 138 39

222 106 24

-15 32 15

0 7 13

48 40 38

46 40 30

2 0 8

242 197 193

187 168 111

55 29 82

NOTE.-Foreach change,congressionaldelegationsor electorsof statesthatgain enoughto pass the votingtest ind electorsof statesthatlose enoughare assumedto vote nay. The averageof 1988 and 1992 electoralvotes was used to


69

not much affected,67legislatorsmay be free to indulge their own preferences-which may dependon ideology, logrolling,PAC contributions,or a host of otherfactors.We have assumedthreedifferentthresholdsof concern: any change at all, any change above 5 percentin absolutevalue, and any change above 10 percentin absolute value. Those legislators whose states' allowancechangesdo not pass the relevantthresholdare assumedto divide their votes evenly; for the sake of clarity they are simply omitted from the vote countsin Table 7. For the sake of completenesswe have applied the same calculationsto electoralvotes (includingthose of the District of Columbia). Table7 makesclearthatPR is a politicalnonstarteras well as potentially expensive (SectionVIIAabove):a changefrom PR to SR or to CM passes overwhelminglyin both Houses under any of our thresholdsof concern. There arejust too many relativelyclean states that would suffer underPR for it to gathera majorityagainstany alternativethatconcentratesthe pain in a smallernumberof dirtystates.This is consistentwith SR being at the core of most proposalsmade duringthe 1980s and with those proposals having been blockedfrom passageby powerfullegislatorsfrom states that this change makes worse off, as discussed above. Once these legislators could no longer simply block acid rain legislation,majoritarian politics increasedtheirpainby reducingthe allowancesbelow those they would have receivedunderproportionalreduction.This is also broadlyconsistentwith the ultimate rejection of efforts to fashion a cost-sharingprogrambuilt arounda nationaltax on electricity,a possibility that was seriously discussed duringthe 1980s. Such a tax would, of course, have benefitedprecisely those states that lose from a shift from PR to SR, BC, or CM. A change from SR to BC also passes both Houses, as well as the electoral college. Note that its margin increases uniformlyas we impose a strictervoting test. The actualallocationof allowances(FA) defeatsCM in the House and electoralcollege and generallywins in the Senate as well. On the other hand, if we assume that every loss of allowances,no matter how relativelyor absolutelysmall, leads to a "Nay" vote, FA fails in the Senate againstBC and in the House againstboth BC and SR. When even a 5 percentthresholdof significanceis imposed, however, FA beats both alternativeseasily in the House, easily beats SR in the Senate, and needs only a nudge to beat BC in the Senate. On the whole, Table 7 supportsthe notion that the Phase II allowance allocationprovisionswere craftedwith sufficient(implicitor explicit)concern for theirviabilityon the floorsof both chambersto makethem no less 67 See, for instance, Kalt & Zupan, supra note 6.

70


attractivethan at least some obvious alternatives.If this had not been the case, one would expect to have seen votes involving alternativeallocation provisions. IX.

ESTIMATING POLITICAL DETERMINANTS OF ALLOWANCE ALLOCATIONS

Ouranalysisthus far does not suggestthatthe PhaseII allowanceallocations can be easily explainedby a small numberof "standard"political economy variables.We appearto be dealingwith a processof majoritarian politics (once the dam holdingback acid rainlegislationwas broken)combinedwith a numberof specialinterestprovisionsto satisfynarrowconstituencies. Committeesof jurisdictionwere not unimportantin the legislative process, but, particularlyin the Senate and in conference,issue-specific groupsof legislatorsplayed criticalroles.68 Because an abundanceof quantitativeinformationis availablehere, we can use regressionanalysisto examinewhetherand how variablesmeasuring the importanceof variousinterestgroups,the presenceof senatorsand congressmenin leadershippositions, and competitive races for Senate, governor,and/orpresidentin particularstateshelp explainthe observedallowance allocationin ways consistentwith varioustheoriesof distributive politics.This analysisis similarin spirit(andresults)to the extensiveliterature that relates congressionalappropriations to variouspolitical variables (and that fails to find strongsupportfor any simple theoriesof distributive politics).69

As above, our analysisconcentrateson the Phase II allocationfor years 2000-2009, both because it is more complex and important(in expected dollar terms) than the Phase I allocationand because it involves a larger samplesize. Becauseof the importanceof complexinterstatedifferencesin initial conditions,we focus on explainingdifferencesbetween the states' actual allocations(FA) and the averageof the allocationsimplied by our threebenchmarks:SR, BC, and CM. (See Table 6 above.) This variableis definedas APHASEIIin Table 8.70 We focus on differencesin numbersof allowancesbecause allowances are homogeneouspropertyrights that should have the same marketvalue no matterto whom they are given. Therefore,the political cost of getting 68 Cohen, supra note 16, stresses that this bill was not atypical of recent experience in this last regard. 69 See, for instance, the references cited in note 5, supra. 70 Joskow & Schmalensee, supra note 16, describe the generally minor differences between the results for this average variable and those obtained for each of the three differences involving individual benchmark allocations.


71

an incrementalallowance for one's own constituentsshould not depend heavily on the state in which they happento reside. Nonetheless,we performeda numberof experimentsinvolvingpercentageandper capitadifferences, withoutobtainingresults qualitativelydifferentfrom those reported below. As Table 8 describes,we employedseveralexogenousvariablesintended to captureinterstatevariationsin the importanceof interestgroupsinvolved in debates about acid rain legislation. These include a variablethat measuresprojectedjob loss in the coal-miningindustryas a resultof the legislation (HSMINERS),71 variables that distinguishbetween clean and dirty states with differentlevels of SO2emissions (EMISSIONS)and different emissions rates (EMRATE),72 and a variabledesigned to measureinterest in relying on scrubbersby applicationsfor Phase I extensionallowancesto supportscrubberinvestments(PH1EXT). One might expect that states for which HSMINERSis large would be very interestedin obtainingallowancesas compensationfor losses of mining jobs. On the other hand, allowancesare given to electric utilities, not miners.It is thus at least equallyplausible,particularlyin light of some of the results of Section VII, that representativesof these states would have neglected the pursuitof allowances in favor of seeking aid for displaced miners and/or attemptingto strengthenincentives to scrub. Thus while states with high values of HSMINERScared more than others about the acid rainprogram,it is unclearwhetherthatconcernshouldbe expectedto producemore or fewer Phase II allowances. We would expect EMISSIONSand/orEMRATEto have positive coefficientsif the "dirty" stateswere able to use the PhaseII allocationprocess to makeup for some of whatthey lost throughpassageof acid rainlegislation aimedat existing dirtyplants.Negativecoefficients,on the otherhand, would be consistentwith clean stateshavingbeen able to use the allocation process to their advantage-the patternsuggestedby Section VII. Finally, we would expect PH1EXTto be negativeif the states interestedin scrub71 Two other conceptuallyweakermining-relatedvariableswere computed.(1) Exceptfor Kentuckyand West Virginia,which are divided into two regionseach, ICF, supra note 36, projectionsof miningjob losses arebasedon state-levelnet employmentchanges,ratherthan gross flows out of high-sulfurmining.(2) Aid actuallyreceivedby May 1995 underthe displacedworkerprovision(Title XI) thatwas pushedhardby mining-staterepresentatives(see Section4, above)amountedto less than$29 millionandcould not have been well anticipated in 1990. Both these variableswere highly correlatedwith HSMINERS,and neitheroutperformedit significantlyin regressions. 72 We also consideredusing emissionsfrom or allowances given to PhaseI units as independentvariables,but both were almost perfectlycorrelatedwith EMISSIONS(p = 0.96). The shareof stateemissionsaccountedfor by PhaseI plantsdid not sufferfromthis infirmity, but its coefficientnever approachedstatisticalsignificancein any experiment.

TABLE 8 ANALYSIS VARIABLES EMPLOYED IN PHASEII REGRESSION Variable APHASEII

to

HSMINERS

EMISSIONS EMRATE PHIEXT SEN

Mean

Max

.00

61.7

1.18

21.6

Min

-93.7

SD

28.4 3.64

0

.16 .01

473 1.03

348 1.49

2,303 4.20

16.6

190

0

.27

1

0

.45

6.14

50.5

0

9.95

42.1

Desc

Differencebetweenactual(FA) Ph allocationsunderSR, BC, and C 2000 to 2009 Estimatednumberof minersof hig (fractionof 1992 demonstratedr million Btu [fromU.S. EnergyI Reserves:An Updateby Heat an 1992), table C-1]) and (average 1990 [fromNationalCoal Assn. Baseline SO2emissions,thousand State averageSO2emissionratefr units, poundsper million Btu of Phase I extensionallowancesrequ averageper year from 1995 to 1 CompetitiveSenateelectiondumm competitiveSenaterace in 1990 competitivein Nat'l J., March1 erwise

GOVEV

Competitiveand importantgovern variablefor competitivegovern (the averageof the state's 1988

SWINGEV

10.0

.65

48.2

8.85

HLEAD

.10

1

0

.31

HCR

.12

2

0

.39

1.35

8

0

1.84

SLEAD SCR

.08 .06

1 1

0 0

.28 .24

SSUB

.29

1

0

.46

APHASEI

.00

58.2

HCOMM

-26.1

13.7

Importantswing state:productof RPCTis the percentageof the s the 1988 Presidentialelection, a RPCT} and {the averageof the votes} Numberof House leadershipslots tion Numberof House committee(Ene tee (Energyand Power,Healtha rankingmemberslots (6) filled b Numberof House committee(Ene numberof subcommittee(Ener state'sdelegation Numberof Senateleadershipslots Numberof Senatecommittee(Env committee(Environmental Prot memberslots (4) filled by the st Numberof Senatesubcommittee( filled by the state's delegation ActualPhaseI allowancesminusa (fromTable 2), thousandsof ton

NOTE.-Except as noted,dataarefromEPA (principallythe NADB) andstandardreferenceson U.S. politics.Sam excludedfromthe sample,as fromthe acid rainprogram,and the Districtof Columbiais included.

74


bing (eitherbecauseit was the least-costcontroloptionor becauseof pressuresto save high-sulfurcoal miners'jobs) gave up Phase II allowancesin exchangefor greaterscrubberincentivesduringPhase I. We also computedtwo sets of more narrowlydefined "political" variables. In the spirit of models of partisandistributivepolitics, variablesin the first set are designedto measurestates' electoralimportancewhen the 1990 legislation was being considered.These variablesinclude a dummy variableindicatingwhethertherewas a competitiveelection for the Senate expectedin 1990 (SEN), the nationalimportanceof an upcomingcompetitive governor'srace (GOVEV),anda variablethatmeasuresthe importance of a stateas a "swing state" in the 1988 presidentialelection (SWINGEV). Since 1990 was an election year, it seems plausiblethat states would have had more clout in the zero-sumallowance allocationgame if they had a competitivesenatorialrace (SEN) or if they were an importantstatewith a competitivegubernatorialrace (GOVEV).It also seems plausiblethat important states that were swing states in the 1988 presidential race If these electoralim(SWINGEV)would have extrabargainingstrength.73 portancevariablesinfluencedallocations,they should have positive signs. Since the issues in the acid rain program,and in the allowanceallocation process in particular,did not reflect a clear split between Democratsand Republicans,we have not includedvariablesmeasuringpartyaffiliationsor ideologicalratingsof each state's legislators. The second set of political variablesreflectsthe nonpartisandistributive politics literature,which implies that the ability of an individuallegislator or a groupof legislatorswith similarintereststo affect acid rainlegislation depends,in part,on whetherthey occupy positions on key committeesor subcommitteesor hold leadershippositions that provide special influence over the provisionsof the bill reportedto the Senate or the House floor.74 The variablesin this second set include the numberof House and Senate leadershipposts (HLEAD and SLEAD), the numberof House and Senate committee and subcommitteechairmanshipsand ranking member slots filled by a state'srepresentatives(HCRand SCR), andthe numberof committee and/or subcommitteeslots filled by a state's representativesin the 73 One mightalso suspectthatrich stateswouldhave moreclout, all else being equal(perbut incomepercapitahadessentially hapsbecauseof the presenceof campaigncontributors), no explanatorypower in any equation. 74 See BarryWeingast& M. Moran,Bureaucratic Discretionor CongressionalControl,91 J. Pol. Econ. 765 (1983); BarryWeingast& W. J. Marshall,The IndustrialOrganizationof Congress;Or,Why Legislatures,LikeFirms,Are Not Organizedas Markets,96 J. Pol. Econ. 132 (1988); KennethA. Shepsle& BarryR. Weingast,WhenDo Rules of ProcedureMatter? 46 J. Pol. 206 (1984), and The InstitutionalFoundationsof CommitteePower,81 Am. Pol. Sci. Rev. 85 (1987).


75

House and Senate (HCOM and SSUB). The leadership in the House (HLEAD)and Senate(SLEAD)generallyhas seats at any negotiatingtable aboutwhich they care. Committeemembershipand, especially, chairmanship or service as the rankingminoritymembercan convey issue-specific influencevia agendacontrol. As we discussedin Section IV, on the House side both the Energy and CommerceCommittee(chairedby CongressmanDingell) and two of its subcommitteesplayedimportantroles in the CleanAir process,thoughonly one of the subcommittees(chairedby CongressmanSharp)dealt with the acid rain programexplicitly. Thus HCR counts all chairmenand ranking members involved in Clean Air, while HCOMM gives extra credit for membershipon the subcommitteethatdealt with the acid rainprogram.On the Senate side the process was very different.The Senatebill was essentially writtenin negotiationswith the administrationin early 1990. While SenatorBaucusgenerallychairedthe negotiationsessions, SenatorMitchell assumedthe chairat key momentsandwas heavily involvedthroughoutthe process. Similarly,while most senatorsin the room at any one time were likely to be membersof SenatorBaucus's Subcommitteeon Environmental Protection,the sessions were open to all senators,and many nonmembers participatedpersonallyon issues with which they were particularlyconcerned and had staff in regularattendance.The variablesSCR and SSUB attemptto reflectthe essentialelementsof this process.75We would expect all the congressionalcontrol variablesin this second set to have positive coefficients. Some of our regressionsalso includedAPHASEI,a variable,takenfrom Table 2, thatmeasureshow well or poorly a state did in the PhaseI allocation process relative to the ER = 2.5 benchmark.76 Our idea here is that states that did relatively well in Phase I for reasons not reflectedin our Phase II independentvariablesmight also have done well in Phase II, reflectingthe sameunobservedpoliticalforces.This variableis clearlyendogenous, and its coefficientcannotbe given an unambiguousstructuralinterpretation.77 75 Idaho,with one subcommitteemember,Steve Symms(R), was excludedfromour sample becauseit had no fossil-fueledgeneratingunits and was thus not includedin the allowance allocationprocess. 76 This variabledoes not reflect actualor anticipatedextension (scrubberbonus) allowances;it is from the second-lastcolumnin Table 2. 77 Several additionalvariableswere employedin a varietyof unsuccessful experiments. One mightexpectthatrepresentatives of stateswith high electricityratesor expectingto need large numbersof allowancesto accommodategrowthwould both be particularlyinterested in obtainingincrementalallowancesandparticularlyable to argueeffectivelyfor them,especially in light of Florida's ability to obtain Section 405(I)(1), but a range of experiments failed to supporteitherhypothesis.(We used the productof the 1980-90 populationgrowth

76


Table9 presentsillustrativeestimationresultsfor a series of equationsin whichAPHASEIIis the dependentvariableand alternativecombinationsof the threesets of variablesdiscussedabove are the independentvariables.In the equationdescribedin the first columnof Table 9, as in all otherequations estimated with a large numberof plausible independentvariables, most coefficientsare not significantlydifferentfrom zero. Several "political" variablesin thatequationneverhad significantcoefficientsin any specification,and we dropthem from furtherconsideration. One of these was SEN, even thoughall of SEN's correlationswith the other independentvariableswereless than0.25. In addition,neitherHCOMMnor SSUB ever had significantcoefficients,perhapsreflectingthe general decline in the importanceof committeesand the concomitantrise in the importanceof otherissue-specificgroupsstressedby Cohen.78The coefficient of SCR was never significant,even thoughthat of HCR was positive and significantin all specifications. Finally,the coefficientof HLEADwas generallynegativeand sometimes significant.Fourof the five House leadershipslots were filled by representatives from Georgia,Illinois, Missouri,and Pennsylvania-all high-emissions states. (The correlationof HLEADwith EMRATEis 0.44.) It seems most likely thatthereis no real "leadershipeffect," since a negativeeffect is implausibleand SLEADnever had a positive and significantcoefficient. The negative HLEAD coefficient simply tells us that "dirty" states did poorly in Phase II allocationsdespite being well representedin the House leadership.Accordingly,we drop both HLEAD and SLEAD from further consideration. Droppingthe variablesjust discussedleads us to the second equationin Table 9. Thatequationhas two groupsof independentvariables,with high intragroupcorrelationsand low intergroupcorrelations.The first group consists of four variables that we think of as measuring "dirtiness": HSMINERS,EMISSIONS,EMRATE,and PHIEXT.The lowest of the six pairwisecorrelationsamongthese variablesis 0.38, and the second-lowest is 0.49. The second group consists of three variablesthat we think of as measuringpolitical/electoral"clout": GOVEV,SWINGEV,andHCR.The rates and baseline emissions as a measure of growth-related allowance "needs.") Optimistic economists might expect that states with high baseline average or marginal costs would be both eager and able to obtain incremental allowances, all else being equal, but coefficients of such variables (based on ICF, supra note 36, as above) never approached statistical significance. Finally, we attempted to measure the importance of two other Clean Air issues, ozone nonattainment and alternative fuels, that might have been involved in cross-Title deals. Specifically, variables measuring the percentage of each state's population in severe or extreme ozone nonattainment areas and each state's production of corn and natural gas (inputs into alternative fuels) never had significant and sensible coefficients in any specification. 78 Note 16 supra.

77

MARKET-BASEDENVIRONMENTALPOLICY TABLE 9 RESULTS PHASEII REGRESSION Independent Variable Constant HSMINERS EMISSIONS EMRATE PHIEXT SEN GOVEV SWINGEV HLEAD HCR HCOMM SLEAD SCR SSUB APHASEI R2 SE

Dependent Variable = APHASEII -4.032 (10.55) -.043 (1.194) -.0311 (.0220) 1.400 (6.534) -.604* (.198) -4.594 (9.026) .393 (.593) -1.012 (1.030) -22.79 (15.02) 30.67* (14.24) 3.923 (3.894) .788 (13.84) 9.549 (13.85) 9.481 (8.183) ? .538 22.67

5.142 (8.770) -.814 (1.088) .023 (.020) -4.642 (5.849) -.421* (.158) .555 (.530) -.522 (.731) 27.82* (12.68)

1.746 (3.618)

4.094 (7.810) * ** ...

*.. -.334* (.078) ...

-9.054* (3.708) .* ...

*

.942* (.433)

-2.646 (3.983) ** ...

-.466* (.065) ...

1.039* (.299)

-5.412 (5.386) -2.660* (.976) ...

...

.854* (.404)

30.47* (8.328) ...

*?--

*

.461 22.58

.403 22.40

?

**

.184 26.20

1.140* (.200) .630 17.83

...

.745* (.260) .379 23.11

NOTE.-Standarderrorsare in parentheses.Samplesize = 48. * Significantat 5%.

lowest of the three pairwise correlations among these variables is 0.58. Within each of these groups, the different variables are conceptually quite distinct. If their performance in regression experiments could also be clearly distinguished, it might be possible to base a structural story on the results. These data, however, are not so kind. Note first that in the second equation in Table 9, only one variable from each of these two groups has a coefficient that is significant at the 5 percent level. Similarly, in the 18 equations (not shown) with two variables from each group, at most one from each group is significant. In the 12 regres-

78


sions with only one variablefromeach of these groups,however,all "dirtiness" coefficientsare negative,all "clout" coefficientsare positive, all 24 slope coefficients are significantat 5 percent,and 16 are significantat 1 percent.The thirdand fourthcolumns in Table 9 show the specifications within this set with the highest and lowest values of R2,respectively. These results provide strong evidence that "dirty" states tended to do poorly relativeto our benchmarksin Phase II, while states with "clout" tendedto do well. Unfortunately,high correlationswithin our two groups of independentvariablesmake it impossibleto use these datato determine with any confidencewhat elementsor aspects of "dirtiness"and "clout" were most important.We are thus unable to discriminateamong a large numberof plausiblestructuralhypotheses. We ran the same set of 12 regressionsjust discussedusing APHASEIas the dependentvariableand restrictingthe sample to the 21 Phase I states. All coefficientsof both "dirtiness" and "clout" variableswere positive, thoughonly one of each was significantat 5 percent.These resultsat least suggest thatthe dirtieststatesconcentratedon PhaseI, wherethey did relatively well on average,at the expense of Phase II, where they fared less well. These resultsalso suggestthatthe "clout" variablesare at least correlated with the abilityto affect the legislativeprocesspositively-at least in the context of acid rain in 1990. Finally, we re-ranthe 12 regressionswith one "dirtiness"variableand one "clout" variableon the right,addingAPHASEIas a thirdindependent variable.All coefficients of "dirtiness" variableswere negative and significantat 5 percent;all coefficientsof "clout" variableswere positive and significant;and all coefficientsof APHASEIwere positive and significant. The last two columns of Table 9 show the specificationsamong these 12 with the highest and lowest R2values, respectively.These results suggest that states that managedto do well in Phase I for reasons not correlated with our "dirtiness"and "clout" variablesalso did well in Phase II. Unfortunately,thereseems to be no way to use these datato tell what sortsof forces this effect mightreflect,and the complex legislativehistorysummarized above providesno obvious candidates. This analysissuggestsfourtentativeconclusions.First,andperhapsmost important,theredoes not appearto be any simple, structuraltheoryof distributivepolitics thatis well supportedby the data.In particular,the failure of most congressionalleadership and committee membershipvariables seems inconsistentwith theories in which power over most legislation is concentratedin the handsof a few people who happento occupy key positions. This resultdoes not in any sense refutethe literaturethatemphasizes the role of committees,subcommittees,andleadershippositionsin congressionalbehavior,however.After all, CongressmanDingell and SenatorByrd


79

managedto block Clean Air legislation for a decade, with the help of a Republicanpresidentopposedto new environmentallegislation.But, once acidrainlegislationgot throughthe gate, the distributionof influentialcommitteeassignmentand leadershippositionsdid not help muchin predicting allowanceallocations.As our discussionof Table 3 indicates,some legislatorswith key committeeposts clearlyused themto benefittheirconstituents throughthe allocationprocess,but othersdid not, and severalstateswithout obviousinfluenceon the relevantcommitteesor in the leadershipdid quite well. Second,there is good evidence that "dirty" states-those on average with many high-sulfurcoal miners, high total emissions and emissions rates,and much interestin using scrubbingto comply with Phase I emissions limits-did relatively poorly in the Phase II allowance allocation game,all else equal.Thereis weak evidence suggestingthatthe very dirtiest states did relativelywell in Phase I, suggestingin turna willingnessto give up Phase IIallowancesto obtainPhase I allowancesfrom states less concernedwith Phase I compliance.Third, there is strong evidence that stateswith political "clout"-because they were large states that were swingstates in the 1988 presidentialelection, or because they were large statesthathappenedto have competitivegubernatorialcampaignsin 1990, orbecause they had representativesin the House Energy and Commerce leadership-tendedto do well in PhaseII, andweak evidencethatthey also didwell in Phase I, all else equal.79 Finally,there is strongevidence that states that did well relativeto our PhaseI benchmark,holding "dirtiness" and "clout" constant, also did wellin Phase II. In a way, this just reaffirmsour firsttentativeconclusion: somethingnot capturedby any of our "dirtiness" and "clout" variables produced positive resultsin both phases.We do not know whetherthis factorprimarilyreflectsdifferencesin legislators'effectiveness,logrollingon issuesoutsidethe acid rainTitle (or even completelyoutsidethe CleanAir 79At the suggestionof an editor,we rana numberof regressionsusing as dependentvari+ APHASEIand DIF = ablesSUM = D(APHASEII) D(APHASEII)- APHASEI, where D= (1.05)-5 is a discountfactorreflectingthe 5 yearsbetweenthe startsof PhasesI and II. Asbefore,at most one "dirtiness"andone "clout" variablewas significantat the 5 percent levelin any one regression.In both SUM and DIF regressionsinvolvingone variable from eachgroup(12 regressionseach), all "dirtiness" coefficientswere negativeand all "clout" coefficients were positive. In the SUM regressions,none of the "dirtiness"coefficientsand eightof the "clout" coefficientswere significant.In the DIF regressions,all of the "dirtiness"coefficientswere significant,along with six of the "clout" coefficients.These results are consistent,at least, with the notionsthatthe "dirtiest"statesgave up Phase IIallowances inexchangefor PhaseI allowancesand that "clout" was valuable.Mechanically, however, these resultsreflectthe high correlationsbetweenAPHASEIIandboth SUM = 0.91) and (p DIF(p = 0.80).

80


bill), or other effects. Whateverthis factorreflects, it appearslikely from our earlierwork that Illinois had it and Georgiadid not. We do not believe that these regressionresultsshould be interpretedas implyingthat interestgrouppolitics, congressionalinfluence,or considerations of state and federalelectoralpolitics did not play an importantrole in the allocationof S02 allowances.Ourearlierdiscussionshows thatthere is clearlyevidence of rent-seekingbehaviorand congressionalinfluenceat work. However,these effects are apparentlytoo subtleand too complex to be capturedin any but the crudestway in this kind of summaryregression analysis.This is consistentwith the resultsof relatedwork analyzingcongressionalappropriations. X.

CONCLUSIONS

Environmentalregulationis an excellent exampleof interestgrouppolitics mediatedthroughlegislative and regulatoryprocesses.The history of federalregulationsgoverningpower plant emissions of SO2represents,in many ways, a classic case. Concentratedand well-organizedinterestsin a few statesthatproducedandburnedhigh-sulfurcoal were able to shapethe CleanAir Act Amendmentsof 1970 and,particularly,1977 to protecthighsulfurcoal and impose unnecessarycosts on large portionsof the rest of the country.During most of the 1980s, the Midwesternand Appalachian utility and mining elements of this coalition managedto use their control over key congressionalleadershippositions,combinedwith presidentialopposition to new environmentallegislation,to block new acid rain legislation. However,once it becameclear that acid rain legislationwas likely to be enacted as part of a largerreformof the Clean Air Act, our analysis indicatesthat this coalition was unableto avoid appreciablecontrol costs shareof emissions allowances. by obtaininga disproportionate With regardto Phase I allowances(apartfrom scrubberbonuses),three of the states with the greatestemissions and cleanuprequirements(Ohio, Indiana,andIllinois)did relativelywell comparedto otherstates,while four others(Pennsylvania,West Virginia,Kentucky,andGeorgia)did relatively poorly. Aside from Illinois, the utilities and, indirectly,high-sulfurcoal minersin these states benefitedfrom bonus allowancesallocatedto Phase I units that scrubbed.However,aside from Illinois, the traditionalcoalition of high-sulfurcoal producersand high-sulfurcoal users were not able to claw back a disproportionate shareof PhaseII allowances.Indeed,they lost even more duringthe legislativeallocationprocessthanthey would have if several simple alternativeallocationrules had been utilized. Specifically, the relativelylargernumberof clean stateswith little to gain per capitawere more successful in Phase II than the relatively small numberof "dirty" stateswith much to lose per capita.


81

If anything, the resulting allocation of Phase II allowances appears more to be a majoritarian equilibrium than one heavily weighted toward a narrowly defined set of economic or geographical interests. It is not strongly consistent with the predictions of standardmodels of interest group politics or of congressional control. In some cases, influential senators and congressmen managed to capture special benefits for their constituents. In other cases, particular states did much better (or much worse) in the allocation process than might have been predicted by simple theories of distributive politics. On average relatively "dirty" states did poorly in Phase II (perhaps because they were more concerned with Phase I and benefits for miners), while states with political "clout" did relatively well in both Phases. These results do not have great explanatory power, however, and we can only conclude that the fight to grab allowances, within a range of allocations that could not be easily defeated in the Senate or House, reflects both a more complex and a more idiosyncratic pattern of political forces than one might expect from previous work on the political economy of clean air. Of course, none of this takes away from the fact that Title IV of the 1990 Clean Air Act Amendments put in place a major long-term program to reduce pollution using an innovative tradable emissions permit system. At least in theory, the allowance system gives utilities enormous flexibility in meeting aggregate emissions reductions goals and may thus allow them to meet those goals at much lower cost than under traditional command and control approaches. Demonstrating this theory in the large-scale acid rain program may lead to fundamental changes in environmental policies and significant reductions in their costs. BIBLIOGRAPHY Ackerman,BruceA., and Hassler,WilliamT. Clean Coal and DirtyAir. New Haven, Conn.:Yale UniversityPress, 1981. Baron,David P. "DistributivePolitics and the Persistenceof Amtrak."Journalof Politics 52 (1989): 883-913. Cohen,RichardE. "WhenTitansClashon CleanAir." NationalJournal(April7, 1990): 849-50. Cohen, RichardE. Washingtonat Work:Back Roomsand Clean Air. New York: Macmillan,1992. and the Coal Lobby." In Crandall,RobertW. "Air Pollution,Environmentalists, The Political Economyof Deregulation:InterestGroupsin the RegulatoryProcess, editedby RogerG. Noll andBruceM. Owen.Washington,D.C.: American EnterpriseInstitute,1983. Crandall,Robert W. "An Acid Test for Congress?" Regulation 8 (September/ December1984): 21-28. Davis, JosephA. "Acid Rain to Get Attentionas ReaganChangesCourse." Congressional Quarterly(March22, 1986): 675-76.

82

JOURNAL OF LAW AND ECONOMICS THE

Lessonsfor Theory RobertW., andHester,GordonL. "MarketablePermits: Hahn, 361-406. Practice."Ecology Law Review 16 (1989): to and TradeableAir RobertW., and Noll, Roger G. "Barriers Implementing Hahn, on RegJournal Yale Interactions." Permits:ProblemsandRegulatory Pollution 1 (1983): 63-91. ulation Times(DecemRobert."TurningOff Acid Rain at the Source."New York Hanley, ber 11, 1983):A12. in the Marketfor Karl. "The Politics and Economicsof AuctionDesign Hausker, Management11 and Analysis Policy Dioxide Pollution." Journal of Sulfur 553-72. (1992): Impactsof the ResourcesIncorporated(ICF). "Comparisonof the Economic ICF 1630)." (S. 1630) andthe HouseBill (S.WashingRainProvisionsof the SenateBill Acid ProtectionAgency. Draft ReportPreparedfor the U.S. Environmental D.C., July 1990. ton, of MarketPaul L., and Schmalensee,Richard."The Political Economy Joskow, for EnCenter MIT Rain Program." Policy:The U.S. Acid Environmental Based Cambridge, 96-003. Paper Working Research ergyand EnvironmentalPolicy MIT, March1996. Mass.: ElizabethM. "The Marketfor Paul Joskow, L.; Schmalensee,Richard;andBailey, forthcoming. Dioxide Emissions."AmericanEconomicReview, Sulfur in the EconomicTheory JosephP., andZupan,Mark."CaptureandIdeology Kalt, 74 (1984): 243-77. ofPolitics." AmericanEconomicReview Rain ControlPolicy in the 1990 Acid The Markets: Politics of The Kete,Nancy. Unpublisheddissertation,JohnsHopkinsUniversity, Air Act Amendments. Clean 1992. of Tenureon the Flow of Kiel,Lisa J., and McKenzie,RichardB. "The Impact 285-93. Benefitsto SMSAs." Public Choice41 (1983): Federal Journal(May 6, 1989): 1098Kriz,MargaretE. "Politics in the Air." National 1102. Journal (April 14, 1990): Kriz,Margaret E."Dunningthe Midwest." National 893-97. Bill: SenateSays Phil, andHager,GeorgeHager."Showdownon Clean-Air Kuntz, 983-87. 1990): 'No' toByrd." CongressionalQuarterly(March31, DistributivePolitics and Levitt,Steven D., and Poterba,JamesM. "Congressional forthcoming. Choice, Public StateEconomicPerformance." Air Act: Compliance Lock,Reinier,and Harkawik,Dennis P., eds. The New Clean1991. Reports, Utilities andOpportunity.Arlington,Va.: Public Reportto the PresiNationalAcid PrecipitationAssessmentProgram.1989 Annual Office, June Printing Government U.S. D.C.: dentand Congress.Washington, 1990. Issues in the Design of NOx NationalEconomic ResearchAssociates, Inc. Key EPRITR-104245. EmissionTradingProgramsto ReduceGround-LevelOzone. 1994. PaloAlto, Cal.: ElectricPowerResearchInstitute,July Regulation."In HandNoll, Roger G. "EconomicPerspectiveson the Politics of andR. D. WilSchmalensee R. by edited 2, vol. Organization, bookof Industrial 1989. Elsevier, lig. Amsterdam:


83

Pashigian,B. Peter. "The Effects of EnvironmentalRegulationon OptimalPlant Size and FactorShares."Journalof Law and Economics27 (1984): 1-28. Pashigian,B. Peter. "EnvironmentalRegulation:Whose Self-InterestsAre Being Protected?"EconomicInquiry23 (1985): 551-84. Pechan,E. H., & Associates,Comparisonof Acid Rain ControlBills. EPA Contract No. 68-WA-0038, Work Assignments94 and 116, OTA ContractL3-5480.0. Washington,D.C.: E. H. Pechan& Associates,November1989. Peltzman,Sam. "How EfficientIs the Voting Market?"Journalof Law and Economics 33 (1990): 27-63. Pytte,Alyson. "A Decade's AcrimonyLiftedin the Glow of CleanAir." Congressional Quarterly(October27, 1990): 3587-92. Roberts,Leslie. "Learningfrom an Acid Rain Program."Science 251 (March15, 1991): 1302-5. Roberts,Leslie. "Acid Rain Program:Mixed Review." Science 252 (April 19, 1991): 371. Shepsle, KennethA., and Weingast,BarryR. "WhenDo Rules of ProcedureMatter?" Journalof Politics 46 (1984): 206-21. Shepsle, KennethA., and Weingast,BarryR. "The InstitutionalFoundationsof CommitteePower." AmericanPolitical Science Review81 (1987): 85-104. Stavins,RobertN. "TransactionCosts and TradeablePermits."Journalof EnvironmentalEconomicsand Management29 (1995): 133-48. Suro, Roberto."Concernfor the Environment."New YorkTimes (July 2, 1989): A1. Tietenberg,Thomas H. EmissionsTrading:An Exercise in ReformingPollution Policy. Washington,D.C.: Resourcesfor the Future,1985. U.S. Environmental ProtectionAgency (EPA),Acid RainDivision. TechnicalDocumentationfor Phase II AllowanceAllocations.Washington,D.C.: U.S. Government PrintingOffice, March1993. U.S. EnvironmentalProtectionAgency (EPA). National Air Pollutant Emission Trends, 1900-1993. EPA-454/R-94-027.Washington,D.C.: U.S. Government PrintingOffice, October1994. U.S. GeneralAccountingOffice (GAO). An Analysisof Issues concerning "Acid Rain." GAO/RCED-85-13.Washington,D.C.:U.S. GovernmentPrintingOffice, December1984. U.S. GeneralAccountingOffice (GAO).AllowanceTradingOffersan Opportunity to ReduceEmissionsat Less Cost. GAO/RCED-95-30.Washington,D.C.: U.S. GovernmentPrintingOffice, December1994. Weingast,Barry,and Marshall,W. J. "The IndustrialOrganizationof Congress; Or, Why Legislatures,Like Firms,Are Not Organizedas Markets."Journalof Political Economy96 (1988): 132-63. Weingast,Barry,and Moran,M. "BureaucraticDiscretionor CongressionalControl? RegulatoryPolicymakingby the FederalTradeCommission."Journalof Political Economy91 (1983): 765-800. Wilson, JamesQ. "The Politics of Regulation."In The Politics of Regulation,edited by J. Q. Wilson. Cambridge,Mass.:HarvardUniversityPress, 1980.