AssessingEnergyPolicy: ShouldReboundCount?

BlakeAlcott HughesHall

DissertationsubmittedforthedegreeofMasterin PhilosophyinLandEconomy

Acknowledgments
For their encouragement and their interest in thisandrelatedtopicsIwould liketothankmyclassmatesintheMPhilinEnvironmentalPolicyprogramme Zubaida, Yifan, Tanja, Steve, Sophia, Michelle, Mae, Laura, Kanittha, Ashleigh,Annela,andAngelaswellasBarbaraBrndli,LenBrookes,John Dimitropoulos,TimFoxon,BasGirod,LindyandPhilipGorton,TimJackson, Tim Jervis, John Lintott, Reinhard Madlener, Kozo Mayumi, Unai Pascual, Cecilia Roa, Steve Sorrell, Steve Stretton, Wendy Thurley,PengWangand zlemYazlk.

Thisdissertationistheresultofmyownworkandincludesnothingwhichis theoutcomeofworkdoneincollaborationexceptwherespecificallyindicated inthetext.

..

Abstract
Evaluationsofenvironmentalpoliciestoincreasetechnologicalenergy efficiencyareinvestigatedfortheirtreatment,orlackthereof,ofrebound,the theoryofwhichispresented.Asanapproximation,untilthebackfirequestion issettled,thereisenoughevidencetojustifyevaluationsdividingengineering savingsbytwo.Itisshownhowtheenvironmentalenergyefficiencystrategy relatestootherenvironmentalstrategieswithinI=f(P,A,T).Definitionsand taxonomiesofreboundandofitsmeasurementareanalysed.Whileabroad historicalcorrelationbetweenefficiencyincreaseandconsumptionincreaseis attested,establishingcausalitystatisticallywouldrequirebetter,perhaps aggregate,metricsformeasuringmacroeconomicefficiency.Thereare reasonstopreferphysicalmetrics,especiallysincetheenvironmentalgoalis toreducenegativephysicalimpact. Whilebackfiretheorypositscausalityfrom efficiencytogreaterinputconsumptionandpredictsthecorrelation,efficiency strategytheoryacceptstherealworldcorrelationbutpositscausalityfrom efficiencytolessinputconsumption.Ideasoftheclassicaleconomiststhatare relevanttoreboundarebrieflyanalysed.Theoreticalandempiricalevidenceis notyetsufficienttodecidewhetherreboundisgreaterorlessthanunity.If greater,however,thesevereconsequenceistheabandonmentofthe efficiencystrategy,perhapsinfavourofquotas.

Glossary
AEEI autonomous energy efficiency improvements, i.e. businessasusual
costcuttingonesratherthanpolicyinducedones

Backfirereboundgreaterthan100%ofengineeringsavings Consumptionefficiency aninputoutputratioasinfluencednotbythe

technologyofaprocessbutbyconsumerbehaviour,e.g.boilingonlythe amountofwaterneededorusinglesslumensbyfocusinglight(e.g.onthe pagebeingread)

Efficiency fromtheLatintodoheretheratioofoutputtoenergyinput,as
opposed to efficacy or effectiveness, which disregard input the inverse of

intensity which is the ratio of energy input to output greater efficiency is a

lower inputoutput ratio shorthand for energy efficiency and technological efficiency

Energy fromtheGreekworkdefinedostensivelyexpressedinjoules
loosely,alsothematerialwithpotentialtodoworkatacertaintemperature subjecttoFirstLawofthermodynamics

Engineering savings the theoretical input savings when processes

becomemoreefficientanddemandforoutputisheldconstantthequantityof

whichreboundisapercentage EROI energy return on (energy) investment, i.e. the ratio of the energy
expended to the energy mined or made available, e.g. for oil about 8093% (Smil2003,76)

Exergy energy in available or useful form to do work not subject to First

Lawofthermodynamics,i.e.itsquantitycanchange

Input presupposes time, process and output quantifiable in joules or

workhours,forinstance

Inputoutputratio :forefficiencyandintensityinputisinthenominatorfor
productivityinthedenominator

Jevonstheory shorthandforbackfire Organisationalefficiency theratioofhumanefforttooutputeconomiesof

scale,Tayloritefactoryfloormeasures,ruleoflaw,honesty,slimmanagement hierarchy, trade, specialisation or division of labor, information and populationdense markets, trade, industrial norms, communication, property security,improvementsinhumancapital,movingclosertowork,lesscrime, andwalkinginastraightlinefromAtoB

Output presupposestime,processandinputquantifiablewithdifficulty
aproductor,macroeconomically,allproductsloosely,economicsize,scale, andconsumptionheremeantphysicallyratherthaneconomicallyorinterms ofutility

Rebound consumption or demand for inputs caused by income or price

effects of efficiency increases expressed as a percentage of engineering savings

Technological efficiency : the ratio of physical input to physical output (or

services: seeSection6.4), implyingprocess here usuallyenergyefficiency: ratioofconversionoffueltousefulheatormechanicalenergy,ortoproduce thingshereoftenshortenedtoefficiency

TABLEOFCONTENTS
Acknowledgments............................................................................................2 Abstract............................................................................................................3 Glossary...........................................................................................................4 TABLEOFCONTENTS...................................................................................6 Chapter1AParadox ......................................................................................1 Chapter2PolicyEvaluation...........................................................................4 2.1Introduction............................................................................................4 2.2Evaluationmethodology ........................................................................4 2.2.1EfficiencyChanges .........................................................................4 2.2.2RealInputSavings..........................................................................6 2.2.3FiveMeasurementIssues...............................................................7 2.3MacroeconomicEnergyConsumption...................................................8 2.4TheEconomicEffectsofEfficiency .....................................................12 2.5TheReboundCoefficientandConclusion ...........................................16 Chapter3PolicyContext ..............................................................................19 3.1Introduction..........................................................................................19 3.2TheEfficiencyStrategy........................................................................19 3.3 I=f(P,A,T) ...........................................................................................20 3.4Ratios ..................................................................................................24 3.5SaturationofDemand.........................................................................26 3.6Conclusion...........................................................................................29 Chapter4MicroeconomicMethodology........................................................30 4.1Introduction..........................................................................................30 4.2TheIncomeEffect ...............................................................................31 4.3ReboundTaxonomy ............................................................................33 4.4Products,Sectors,Consumers............................................................36 4.5OutputQualityandNewProducts .......................................................38 4.6CountriesorWorld ..............................................................................40 4.7SubstitutionandTimeEffects ............................................................42 4.8Berkoutetal.2000andGreeningetal.2000 ......................................44

4.9Conclusion...........................................................................................46 Chapter5MethologicalProblems .................................................................47 5.1Introduction..........................................................................................47 5.2Population ...........................................................................................47 5.3EndogenousGrowth............................................................................49 5.4SmallShares .......................................................................................52 5.5Marketfailure.......................................................................................54 5.6Costsofefficiencyincreases ...............................................................56 5.7Conclusion...........................................................................................57 Chapter6TheMacroEmpiricalApproach ....................................................59 6.1Introduction..........................................................................................59 6.2ChangeinEfficiency............................................................................60 6.3TheMonetaryDenominator.................................................................63 6.4TheServiceDenominator....................................................................65 6.5PhysicalDenominators........................................................................69 6.6Conclusion...........................................................................................76 Chapter7 TheClassicalEconomists ............................................................79 7.1Introduction..........................................................................................79 7.2FromEfficiencytotheConsumer.......................................................80 7.3TheLabourAnalogyandFrozenTechnology......................................82 7.4Conclusion...........................................................................................86 Chapter8Discussion....................................................................................87 8.1Concepts .............................................................................................87 8.2PolicyAlternatives ...............................................................................88 8.3FrozenTechnology..............................................................................89 Chapter9Conclusions..................................................................................92 9.1ByChapter ..........................................................................................92 9.2General................................................................................................94 Bibliography ...................................................................................................96

Chapter1AParadox
Heightenedenergyefficiencyisapolicytoolofgovernmentstoachievethe environmental goals of conserving fuel and lowering combustion emissions. Article 2,1 (a) (i) of the Kyoto Protocol to the United Nations Framework Convention on Climate Change, for instance, commits Parties to implement policies and measures in achieving its quantified emission limitationandreductioncommitments[and]inordertopromotesustainable developmentsuchasenhancementofenergyefficiencyinrelevantsectors of the national economy. (UNFCCC 1997) Bodansky reports that from the beginning of the Convention negotiations the envisioned legal commitments included not only taxes, cost internalization, sustainable forest management and best available technology to limit greenhouse gas emissions, [but also] harmonized efficiency standards [and] promotion of energyefficiencyandconservation.(1993,508)

ArticleIII2561.(c)oftheDraftTreatyestablishingaConstitutionforEurope states that with regard for the need to preserve and improve the environment,Unionpolicyonenergyshallaimtopromoteenergyefficiency andenergysaving.(CONV850/03)TheUKEnergyEfficiencyCommitment (Defra2002)andChapter3oftheUKEnergyWhitePaperforeseeefficiency measures in space heating, boilers, appliances, lighting, electric motors, powersupplies,etc.,tobepursuedinteraliabylegislation,theCarbonTrust, andtheEnergySavingTrust(DTI2003),whiletheUKSustainableEnergyAct 2003 prescribes efficiency improvement in Sections 24. (OPSI 2003 also DoE2006[US],BfE/ES2005[Switzerland]andPrognos2004,4[Germany])

Thevariouspoliciesmandate,subsidise,ormerelyencouragetechnological efficiencygains(Glossary)andarefamiliar:lightbulbs,householdappliances, vehicles, whole transport systems, electric power installations, buildings, manufacturing, etc., should use less energy input for a given physically measuredoutput.Energyefficiencyisuniversallyendorsedbypoliticalparties right and left, green as well as industry NGOs, and the press. (Reijnders

1998)Academicecologistsandeconomistsaredivided.Whilesomebelieve that demand for a scarce resource can be reduced by its more efficient use (Jacobs 1991, 91 105), criticism of higher efficiencys usefulness in lowering pollution and conserving resources has become more widespread since first voiced by Jevons (1865 Glossary), Brookes (1978, 1979), Khazzoom (1980) and Saunders (1992) and has led the UK Department of Environment,Food,andRuralAffairs(Defra)tocommissionacademicstudies aimedatsettlingthisurgentquestion(UKERC2005,20064CMR2006Allan etal.2006).

This dissertation asks two questions about this environmental energy efficiencystrategy.First,shouldevaluationsofefficiencymeasurescorrectfor

rebound,i.e.fortheincreasedconsumptioneffectedbythelowerproduction
costs (including household production) caused by policyinduced technological efficiency improvements? The dissertation answers strongly in the affirmative, but the evaluation documents here surveyed do not do this. Second, are these policies environmentally counterproductive, i.e. do they

backfire, causing energy consumption higher than it would have been

otherwise, without the policies? While the answer is tentatively yes, the dissertationmainlytriestodefineconceptsandquestionsandestablishwhat countsasevidence.

Chapter 2 examines the methodology of documents evaluating the effectiveness (and costeffectiveness) of policies,definesthereboundeffect and renders it plausible. Chapter 3 embeds the efficiency strategy in the

I=PAT identity showingenvironmentalImpacttobeafunctionofPopulation,

Affluence,andTechnology,andshowsthatdematerialisationandheightened efficiencyareidentical.Chapter4describesthereboundeffectasanincome

effect and tries to improve on the literatures chaotic and nonparsimonious

taxonomy of rebound, criticising widelycited literature. Chapter 5 criticises parts of the theory implicit in the empirical methodology of rebound studies. Chapter6drawsacorrelationbetweenundisputedlyrisingconsumptionwith risingefficiencybutmainlyanalysesproblemsofmeasuringthevariablesand concludes that since correlation is not causality the debate needs more 2

theoretical work. Chapter7brieflydescribestheargumentsandconclusions of classical economists concerning backfire. Chapter 8 discusses open definitional issues and suggests the alternative strategy of energy quotas. Chapter9drawsconclusions.

Noonemaintainsthatincreasedtechnologicalefficiencyhassucceededin lowering rates of consumption. Empirical work is universally and perhaps necessarily inconclusive, allowing the two contradictory claims that efficiency gainsatleastslowratesofconsumptiondown,orthattheyspeed themupdependingononestheory.ForJevonsthequestionisaparadox (1865,141),forWirlatheoreticalriddle(1997,29),andforSchipperaLoch Ness monster (2000, 351). Since around 1990 rebound literature has burgeoned. No study omits counterfactual assumptions or an otherwise clause about the path consumption would have taken without either the businessasusual or policyinduced intensity decreases. (Khazzoom 1980, 23, 31 Howarth 1997, 3 Brookes 2000, 356 Schipper & Grubb 2000, 370 Moezzi 2000, 525) The objective of the dissertation is to contribute to the clarificationoftermsandissuesanddemonstratetheplausibilityofthethesis thatchangingtechnologicalefficiencyratiosdoesnotnecessarilychangethe absoluteamountofenergyconsumed.

Chapter2PolicyEvaluation
2.1Introduction
Evaluating the environmental effectiveness of energy efficiency policies presupposesquantificationofthechangeinenergyconsumptionorpollutant
1 emissions caused by the policies. Sections 2.2 and 2.3 show how some

Swiss and UK assessments do this on the basis of technical product information enabling computation of the average efficiency of the stock of capitalgoodsbefore,andafter,policyenactment.Actualenergyconsumption over the studied time period can be compared with these potential savings andwithhypotheticaltrendsprojectingquantitiesofenergythatwouldhave

beenconsumedwithoutthepolicies.Section2.4showsthatitisplausiblein
judgingpolicyeffectivenesstocorrectforreboundandcontrolforbackfire(a consumption level even higher than trend) that would render efficiency policies counterproductive. Section 2.5 reviews recent steps to incorporate reboundintoevaluationcomputations.

2.2Evaluationmethodology
2.2.1EfficiencyChanges
Thebottomupmethodofcomputingenergysavingstakesthenumberof newly influenced agents, realised installations, optimisations, and saved personkm, etc. This number is then extrapolated by an average energy savingsorproductionsfactorbasedonengineeringdataabouttheproduct. (ES/INFRAS 2004, 6, 103 also Schipper & Meyers 1992, 65) The extrapolation starts with the average efficiency of preexisting stock and the number of newly more efficient goods and processes, then computes the greateraverageefficiencydeemedtobecausedbypolicies.Thedifferencein theseaverageratiosofjouleinputtooutputmultipliedbythenumberofoutput

units (light bulbs, houses, cars, etc.) in use is called savings or the rate of
changeofconsumptionoftheinput.

Thisdissertationanalysesconsumption,andemissionsonlyimplicitly.

Assume the policy measure mandates kettle efficiency jouleinput per changeoftemperatureofalitreofwaterasevaluatedbyDefra(2006).All kettlesnewlytakenintousethushavewaterboilingefficiencyX,greaterthan

Y, the average efficiency of the existing stock of the capital goodkettlethe

difference in these ratios gives a coefficient of efficiency change. Policy induced change in jouleinput for all kettles is the number of kettles in use timesthedifferencebetweenthebeforeandafterefficiencycoefficientsand the number of litres boiled. Assumptions are necessary about the mix of efficiencies among the stock at the two measurement times. (Fouquet & Pearson 2006, 146) Defra gives figures on standard and eco kettle efficiency,numbersofeachinuse,numberofusesperyear,andnumberof litresboiledperuse.Thisdataontheproductoroutput(hotterwater)yields data on input: energy use per kettle per year times kettle numbers which, takinganefficiencydifferenceof1.42andassumingthewholestockisnow eco kettles, yields energy saved per year of 1.27 TWh. (ibid., 4.1) Similar changesforcars,heatedrooms,powerstations,etc.areroutinelytallied. Sincethepolicygoalisreductionoftheabsolutenumberofinputjoulesand not a mere change in ratios, the crucial and contentious assumption here concerns the number of kettles/kettle uses by which the efficiencychange coefficient is multiplied. Simply counting kettlejoule consumption at two differenttimesyieldsnoinformationabouttheinfluenceofthepoliciesonany measuredchange,notonlybecausesomekettlereplacementwouldhappen anyway(Section2.2.3)butalsobecauseothercausalfactorscouldinfluence totalkettlenumber.Technicalefficiencyinformationtheengineeringpartof evaluationgivesonlythedimensionlessratioofinputperunitofoutput.To try to determine from these ratios an absolute number one computes

engineering savings by holding outputconsumption, e.g. boiled litres,

constant.Ifpoliciesareresponsiblefordoublingtheenergyefficiencyofkettle
2 use, or vehicles, the inputoutput ratio over some time period halves. To

Thisunrealisticallyassumesthehomogeneityovertimeoftheoutputse.g.inkettle

accessoriesorcarsizeandweight.(Section4.7)

derive total consumption ofinput units from this, one must then insert some number of units of consumed outputinthedenominatorandmultiplybyhalf theoriginalbaselinefractionof1/1,namely1/2.Holdingthisnumber(e.g.of kettles/kettle uses) constant gives a theoretical reduction in inputjoules or savingspotentialwithconstantconsumption.3 Thisamountisroutinelyadded to actual (measured) consumption to arrive at trend or input consumption withoutpolicies.(Section2.3)

2.2.2RealInputSavings
Taking engineering savings as real savings assumes that although higher efficiency means lower outlays for energy and freed purchasing power, consumersdonotconsumemore(KonstantVerbrauch(CEPE2003,39)).A typicaldocumentconcerningaprogramforhouseholdheatingandappliance use takes number of households and previous demand for energy services
4 times the fall in the energy ratio. (Defra 2004b, 29) All investigated

regulatoryimpactassessments(RIA)thusmerelysubstitutelowerunitenergy consumption(UEC)withoutalteringunitamounts.Prognossimilarlyassesses Germanys prospects of achieving its emissionsreduction goals, quantifying the variables policies and technology (including efficiency), general economic growth, weather, and energy prices (2004, 3) consumption reducingefficiencyimprovementsforcartypes,forinstance,aremultipliedby the number of cars in the before scenario, yielding an alleged savings of
5 14.7PJ. (22INFRAS2003,34)

Wereefficiencydecreased,thetheoreticalincreaseinconsumptionwouldbeengineering

wastethediscourseofwastingenergyisthusdependentontheconceptofefficiency.Not everysectorbecomescontinuouslymoreefficient:restauranteatingrepresentsperhapsa declineintheefficiencyofeating,andaluminiumdrinkcontainersareperhapslessefficient thanglassones.Efficiencyinducedcostreductionsingasolineleadtomorekilometersand lessefficientvehicles.(Wirl1997,p.31)

4

Thesampleofdocumentsalltobefoundinthebibliographycoversalmostall

assessmentscarriedoutinSwitzerland,butonlyasmallnumberfromtheUKnoconclusion aboutthepopulationofallassessmentdocumentsispossible.
5

Prognosalsocomputestaxeffectsonfinalconsumption(2004,10),butignoretheecotax

rebound,i.e.ensuingefficiencyincreases.

Thus only engineering criteria are used, to the exclusion of the economic possibilityofchangesininputaswellasoutputconsumptionduetodemand stemming from lower perunit energy expenditures and/or lower prices (Section 2.4). A further example is change in petrol consumption per km specifictocarmodelsanefficiencyimprovementEfactortimese.g.775 moreefficientcarssoldtoderivefuellitressaved.(ES/Prognos2004,1011, 22 ES/INFRAS 2004, 6, 76) Using this bottomup methodology DTI estimates 0.6 MtC are saved through boiler efficiency improvements of 5 million boilers (2003, 3338), while INFRAS takes Delta rvalue for wall or
2 window insulation times newly improved m to assess heating oil savings.

(2000, 3, 16) Paradigmatically, the empirical basis of the criteria for evaluating the programmes is only the amount (Mengen)ofmoreefficient itemssoldandtheenergyspecificeffect(spezifischeWirkung)injoulesper

unit (INFRAS 2003, 23). Therefore absolute PJ savings are exactly proportional to the changes in relative or specific unit consumption, and
when actual energy consumption rises rather than falling by the amount predicted by engineering savings, this is attributed to other growth factors (Section2.3).

2.2.3FiveMeasurementIssues
1) Correct calculation of engineering savings includes the lifetime efficiencyofthenewcapitalstockcomparedtothatoftheold,entailing comparisonnotonlyofoperatingenergyinputsbetweenlessandmore efficientequipment,butalsoofdifferencesinembodiedenergy,andalso the costs of capital junking in energy (not monetary) units i.e. of replacing equipment before wear and tear alone render it depreciated.
6 (Robinson 1954, 91 Section 5.6) None of the investigated policy

evaluationscorrectforthese. 2) More efficient capital that would have been brought into use anyway cannot be attributed to policies in evaluating their effectiveness. This

AddingR&D,recyclingandwastetreatmentlikewiselowerslifetimeefficiency.

Mitnahmeeffekt (freeriding) is routinely corrected for with some

coefficient.(ES/INFRAS2004,5053,109113) 3) New more efficient energyproducing installations such as combined heatandpoweronesarecreditedwithenergysavings,newgeneration beingsubtractedfromtheoveralldemandforelectricitytobemetbythe generating stations attached to the grid. The use of the electricity from the CHP [combined heat and power] plants shows up as increased energy efficiency (Barker et al. 2005, 25) ES/INFRAS speaks of electricity saved, i.e. substituted. (2004, 4) The assumption that an equivalent number produced by fossil fuels is taken from the grid is
7 howevergratuitous. 8 4) Nontechnological consumption efficiency (Glossary) , e.g. a policy

educatingdriverstodrivemorefuelefficiently,isevaluatedbymultiplying the number of pupils by an energy savings specific to the course, saving e.g. 150 TJs annually. (ES/INFRAS 2004, 10, 76) The income effectorresultingmoneysavingsraisesdemand,however(Section2.4), andthesejoules,aswell,shouldnotbecountedonetooneassaved. 5) Evaluations universally ignore the possibility that higher operating efficiencycancontributetotheemergenceofnewproducts(Section4.7).

2.3MacroeconomicEnergyConsumption
The engineering computations predict, ceteris paribus, total energy consumption lower than a baseline or trend consumption without policies, based on history and adjusted for assumptions about growth factors like population and endproduct consumption as well as for tax policies, prices, and general growth effects. ES/PROGNOSfor instance, after adjusting for policy effects themselves, observes actual energy consumption increases, e.g.of19.7PJsfor20022003,96%ofwhichareexplainedbythemodel supported analysis. (2004, 12, 1011, 2022) That is, in the face of
7

Similarly,newwindenergyinstallationsareerroneouslysaidtosaveabout2,250tonnesof SeeHannon(1975,96)Etzioni(1998,630)Prettenthaler&Steiniger(1999)Norgard

CO2 permegawatt.
8

(2006).

divergence between actual measured levels of demand and engineering computations, not all engineering savings could be realised due to countervailinggrowthforces:Humanimpactsontheenvironmentarearace between economic (and population) growth and efficiency improvements. (Levett2004,1015Binswanger2001,120)Thetheoreticalparadigmsetsthe trend or reference path based on exogenous growth variables like population (Schipper et al. 1996, 188) or a rising background level of aggregate activity, counteracted by falling energy intensity (Schipper & Meyers 1992, 5860 Chapter 5). For Besiot & Noorman, although appliances become more and more energy efficient, volume growth [e.g.] population growth, number of households, household size still outweighs efficiencygains(1999,375377)

For instance, SwissEnergys assessment of its own strategy of energy efficiency and renewable forms of energy first presents the energysaving effects of each policy. (2004, 3, 4, Fig. 10) Then after summing these engineering savings it laments the trend of increasing traffic volumes and heatedsurfaceareasinbuildings,andtheappearanceofevermoreandever larger appliances, equipment and vehicles on the market. (27) Based on thesetwoindependentlycomputedmagnitudes,thedocumentconcludesthat Without Energy 2000 and SwissEnergy energy consumption in Switzerland would probably be around 6% higher than actual consumption. (22, 29, Figs. 12 & 13 ES/Prognos 2004, 25, 2728) Duchin & Lange use this
9 templatetocalculateworldenergyconsumption.(1994,48,9296,159164

Schipper & Grubb 2000, 368 CEPE 2003, 6, 24, 36, 46) This method of estimating real policyinduced savings is no longer bottomup but rather topdown:Onthebasisofofficialstatisticsofactualenergyconsumptionthe

Theirstudyforthetransportationsectorassume[s]thatthefuelmixwillremain

unchangedbutthatanimprovementintheaveragefuelefficiencyofthefleetof50%is achievedby2020Weassumethatthenumberofmotorvehiclespercapitaandthenumber ofmilesdrivenpervehicleinmostdevelopedcountriesremainfairlystable,withanexpansion between16%and450%..invehiclespercapitaindevelopingcountries.(55)

influence of policies is determined, after deducting a roughly estimated

10 referencescenario.(ES/INFRAS2004,6)

Figure1showstheeffectofefficiencypoliciesontotalenergyconsumption assumingrebound=0.
Figure1:Consumptionwithandwithoutpolicies

Energyconsumptioninjoulesasafunctionoftime(with BAUtechnologicalefficiencyincreases)
Consumptioninjoules
Q1 Projectedconsumption withoutpolicies Realconsumption(withBAUandpolicy inducedefficiencyincreases) Projectedconsumptionwith policies(engineeringsavings)

Q2 Q0 Q3

Time

Holdinggrowthdriversunrelatedtoefficiencyconstant,Q3 resultsifrebound=0. Q2 isactualmeasuredconsumption.WithvariousgrowthdriversQ1 results,ahistorical trendbasedongrowthfactorsopposingtheallegedshrinkageeffectsofefficiency.

Only the middle line (the explicandum) and engineering savings are real measurable quantities. The top line is the hypothetical or predicted consumptionchangewithoutgovernmentinterventionseenafterthefactitis thecounterfactual,whatconsumptionwouldhavebeenwithoutpolicies.The bottomlineisconsumptionchangewithefficiencypoliciesbutwithoutgrowth drivers, i.e. with no change in number of products and services demanded. The difference between the bottom and middle lines is explained either by growth forces stronger than efficiencyinduced savings or by the opposite thesis, namely that efficiency gains lead to higher input consumption. If

10

Thedeductionisofthegrowtheffectsfromwhatwouldhavebeenrealisedwithfull

engineeringsavings.

10

engineeringsavingsequalrealsavings,Q2 minusQ3 shouldequalQ1 minus Q2.Themiddlelinecould,theoretically,slopedownward(Q2<Q0). In terms of kettles, Q3 results if we hold outputconsumption (number of kettles/kettleuses) constant after the change to more efficient kettles. Q1 results from more kettles and/or more kettle use if this growth is large enough, even at the new lower inputoutput ratio inputjoule consumption

rises(Q2).Thetoplineineffectholdsefficiencyconstantwhilethenumberof output units increases, giving Q1. The top line, based as it is on historical
experience, slopes upward and includes the businessasusual (BAU) efficiencyincreasesoccurringdailyasinputcostcuttingoffirms,governments and individuals (Brookes 1978, 94), but here again this is interpreted by efficiencystrategytheoryasbeinginconflictwithgrowthforcesthetopline wouldbeevensteeperwithoutBAUefficiencygains.Inanycase,realsavings is allegedly Q1 minus Q2 (SwissEnergy 2004, 2224), while Jevons theory holdsQ2 >Q1.IncludingBAUefficiencyincreasesincreasesthegapbetween Q2 andQ3. The efficiency strategys macroeconomic case is strong if actual consumption falls over time in comparison to the baseline trend: policy induced(butnotBAU)efficiencyincreasescouldbeanexplanatoryvariable alongsideconsumersatiationandstructuralchange(theshifttolessenergy
11 intensive sectors (Section 3.4)). For instance, SwissEnergy can argue that

the consumption of combustibles was more or less constant despite an increaseintheoverallenergyrelevantheatedsurfaceareaby1.2%,andthis points to additional efficiency gains. (2004, 23) If on the other hand real, measuredlevelsofconsumptionincrease,themoresoatanincreasingrate, theefficiencystrategyhasamoredifficulttask:thepostulatedgrowthforces mustbeallthestrongerthemoreclearlytheywintherace.Iftheydecrease or increase at a decreasing rate, however, Jevons theory must counter the suggestionthatthisisduetogreatertechnologicalefficiency.

11

Attheworldscaleofinvestigationthestructuralargumentreducestothesaturationone.

11

Policy evaluation methodology is based on theory. All relevant literature acknowledgesthisandspeculatesonwhatwouldhavebeenotherwise withoutthem.(Alcott2005,11,16)Howwouldconsumptionhavedeveloped in all energyusing sectors without this improvement in the joule/GDP ratio? Evaluations universally assume that efficiency increases of any provenance

ceteris paribus lower energy input consumption, i.e. register negatively.

Jevons theory observes that Q2 is higher than Q0 and takes efficiency increases as one of the causes of this, on the theory thatinputcostsarein aggregatelower,enablingnewdemandforoutputsrequiringenergyinputs.

2.4TheEconomicEffectsofEfficiency
Doesactualconsumptionthenriseinspiteofefficiencyincreasesor(partly)

becauseofthem?TheInternationalEnergyAgencywrites,Withoutefficiency
gains energy demand would today be half again as high as some earlier time. (NZZ 2005, 21) A seminal theoretical model including BAU efficiency increases (which are arguably much larger than policyinduced ones) first assumes a constrained case, scenario 1, with zero percent autonomous energyefficiencyimprovement(AEEI)andthenahighefficiencycasean AEEI of 1.0% per year. By comparison with scenario 1, energy demands in 2050 would be nearly halved. By definition, no consumption losses are imputed to AEEIs. (Manne & Richels 1990, 72) This dissertation examines Jevons alternative template, namely that technological efficiency increases, because they cheapen energy inputs, are a necessary condition for consumption growth if this is true, then the methodology of evaluations is basedonfalseassumptions.IsJevonsmodelplausible?

If we cease holding the number of consumed products (or their use) constant,inputengineeringsavingsandinputrealsavingscannolongerbe equal.Basednotonthemacroeconomiccorrelationbetweenrisingefficiency and rising input consumption (Chapter 6), but rather on the price consequencesofanefficiencyinducedfallindemand,oneofthetwoinitiators oftodaysdebatewrote: Unfortunately, the estimates of energy savings predicted to result from these mandated standards are derived mechanically[with one 12

exception].Whenmandatedstandardsraisetheapplianceefficiencyby1 percent, demand is expected to drop by 1 percent Examples of such resultsarefoundinreportsbythe[US]DepartmentofEnergyandbythe Staff of the California Energy Commission Thesecalculationsoverlook the fact that changes in appliance efficiency have a price content. (Khazzoom1980,2122) Thetwoquestionsofthisdissertationcannowberestated: Dorealsavingsequalengineeringsavings?Ifnot,evaluationsandthe theoryunderpinningthemoverestimatestheeffectivenessofefficiency policies. If the technological changes,ceterisparibus,causenewconsumption of the goods affected by the higher input efficiency (and/or other goods), what is the size of this newly released consumption? Could higher consumption of the newly more efficiently usedinput therefore beevenhigherthanbeforetheefficiencychangeapolicybackfire? (Khazzoom1980,23) Whilerebound,inaccordancewiththeliterature,isanyconsumptioncaused by efficiency increases, expressed as a percentage of engineering savings with100%sometimescalledunity,reboundgreaterthanunityisbackfire. The rigorous argument is presented by Khazzoom (1980, 2223), Wirl (1997, 31), Birol & Keppler (2000, 460463), Schipper & Grubb (2000, 369 370),and,below,Binswanger: If technological progress makes equipment more energy efficient, less energy is needed to produce the same amount of product or service. However,theamountofproductorserviceusuallydoesnotstaythesame. Because the equipment becomes more energyefficient,thecostperunit of product or service that is produced with this equipment falls which, in turn, increases the demand for the product or the service. If the energy efficiencyofacarisincreasedbytechnologicalinnovations,100kmcan bedrivenwithlessfueland,therefore,atlowercost.Consequently,people may drive more and longer distances because mobility has become cheaper.(2001,120) Thischaracterisationdescribesonlydirectreboundthegreaterdemandfor the product now produced more energyefficiently (Section 4.3) to the exclusion of further rebound effects which by the same causal mechanism

13

raisedemandforrelatedproductsandbyotherconsumersinWirlswordswe not only upgrade quality but the associated increase in the real income allows[us]toraiseallkindsofdemandsincludingthedemandfortheservice in question. (1997, pp. 41 20, 2627, 31) However, computation of total rebound must take into account both new products and worldwide effects.
12 Technically,itmustestimatetheefficiencyelasticitiesofoutputprice andof

energyinputprice,andthenthepriceelasticitiesofdemandforoutputsand
13 energy inputs , arriving in the end at the efficiency elasticity of energy

demand(UKERC2006,67),formallyinput/input/output.
Figure 2 shows Khazzooms depiction of how lower production costs alter the willingness to utilize appliances or cars, even at an unchanged energy inputprice.
Figure2:Lowercostsenableashiftinthesupplycurve

S(E1)=
supplycurveat startingefficiency

Q
D

S(E2) S(E1)

Q2 Q1

S(E2)=
supplycurveat higherefficiency

P
Sameprice,moresupplied
Source:Khazzoom(1980,24)

12

E.g.steelpricesfell8090%partlyduetoBessemerandSiemensMartinprocessesandto I.e.bothsupplierandconsumerdemandfunctionsareneeded.

mechanisedlabourandfactoryfloorefficiency.(Landes1969,259)
13

14

Q here unconventionally on the Y axis is the utilization rate of the refrigeratororcartheunitisachangeintemperatureofagivenvolumeof spacerequiringembodiedandutilisationenergy. Any computation of purported efficiencyinduced savings, whether autonomous or policyinduced, must assume reference quantities under

frozentechnology:Ifsteamengineswerestillattheirefficiencyratiosof1820,
how many steam engines would be made, and how often used, per year today?Anytrendperiodcouldsomedayalsoappearrelativelyinefficient,as the period 18201825 does to us now, yet total consumption of goods then alreadyexisting,plusnewthings,hassincethenincreasedenormously.

The possibility thus emerges that engineeringsavingsmustbediminished bysomefactororcoefficientifwearetotakeaccountofincreased,rebound demand. A correction formally the same as this is now routinely made in deductingfromtheestimateoftheeffectivenessofmeasuresthoseefficiency improvements of capital which would privately have been made anyway (CEPE2003,43),showingthatengineeringtalliesareinprinciplecorrectable, inourcaseforthedemandgeneratedbyeitherleftoverpurchasingpoweror lowerproducercost(inputprices)withtheensuingshiftofthesupplyfunction andpricedecrease. INFRAS already does this inadvertently in their evaluation of the

employment effects of efficiency programmes: Positive effects on

employment result indirectly from the newly freed spending power due to the energy savings Important assumptions of our model are used to estimate theemploymentintensityoftheinvestmentsthatarebothtriggered andhinderedbytheprogrammes.(2000,2,3,emphasisadded)Thisincome effect(dieinfolgederEnergieeinsparungenfreigewordenenMittelChapter4) or higher material welfare caused by the efficiency increases is said to increase employment, but by the same logic it increases demand generally. (Schumpeter 1911, 283, 301) Even if higher demand is only for labour, the newwageearnerswoulddemandenergy.Withanelasticitytobedetermined, and which in turn will determine whether rebound is greater or smaller than unity,consumptiongoodsembodyingandusingenergyarenewlydemanded. 15

2.5TheReboundCoefficientandConclusion
Twenty years after Khazzooms complaint this methodological weakness gainedsomeacknowledgment.Schipper&Grubbwrote: Thepotentialfeedbackbetweenenergyefficiencychangesandthelevelof energyusing activitieshasbeenproposedtoresultinageneralrebound effect. This postulates that since improving the energy efficiency of end use technology (e.g. lighting) will lower the cost of the energy service it provides, consumers will tend to increase their demand for that service, offsetting the apparent savings that an engineering calculation would suggest.(2000,369) Empirical studies of the 1980s and 1990s established a 20 50% direct reboundeffect,i.e.forthesamegoodorservice.(Greeningetal.2000)

One correction for rebound is in the evaluation of UK homeheating efficiency measures, where lower fuel bills allow heating more, a rebound
14 called comfort taking in departure from the scientific literature and

moreoverinaccuratebecauseitappliestohomeheatingonly: the uptake of such measures maynotleadtotheequivalentreductionin consumptionashouseholdersmayreapthebenefitsasincreasedcomfort. This is accepted and allowed for in the Defra calculations, which allow around 50% rebound in all low income households and 15% elsewhere. Preliminary evidence indicates that rebound effects may be small in homeswhichwereadequatelyheatedbeforehand,irrespectiveofincome. (Defra2002,4) 4CMR likewise modifies engineering (expected) savings by an exogenously incorporated direct (Section 4.3) rebound coefficient ranging from 0 to 75%. (2006, 5, 12, 21, 23, 35, 7275) Finally, one assessment indirectly acknowledges the principle by remarking, calculations [of the ExpertGroup]donotincludethereboundeffect.(EEB2000,32)

Similarly, the NRC (2002) in assessing the effectiveness of Corporate AverageFuelEconomystandardswrites: Allowanceismadeforthereboundeffectthatis,thereisasmallincrease inmilesdrivenasfueleconomyincreases.[TheCAFE]policyinstruments leadtoincreasesinfueleconomyofnewvehiclesandthusdecreasethe

14

Thisquainttermdepartsfromthescientifictermdirectreboundandmoreoverappliesonly

tohomeheating.(also4CMR2006)

16

variable cost per mile of driving. This, in turn, encourages consumers to drivemore.[A]10percentincreaseinfleeteconomycanbeexpectedto lead to a1percentto2percentincreaseinVehicleMilesTravelled.The net impact would be an 8percent to 9percent reduction in fuel use for every10percentincreaseinfueleconomy.(sections41and524 25) Thiscorrectsengineeringsavingsby10to20%.

In contrast, an assessment of an evaluation [INFRAS 2002] of Swiss efficiencypoliciestakesengineeringsavingsthroughoutatfacevalue.(CEPE 2003, 6) Two of its assessment criteria are of interest, namely whether the definition of the reference assumptionsisadequate,andwhetheraccountis taken of suppliers hiking up prices to take advantage of subsidies for consumers (Preismitnahmeeffkete). (32, 35) As to the first, it accepts as exogenous the usual parameters of space heated, number of new vehicles and appliances, industrial production (44) or business cycle and structural effects (56). As to the second, they implicitly acknowledge that while increased purchasing power has resulted, it is appropriated by suppliers. Given that by the year 2000 a consensus on the presence of at least some significant direct rebound had been reached, this 2003 study missed an opportunity to note that the reference scenario wasnotnecessarilytoohigh andthatsomegrowthstemsfromefficiencygainitself. Statementsquantifyingenergysavingsshouldbetakenwithagrainofsalt. Ifassumptionsarechallenged,itmustbedemonstratedthatanysavingsare achieved at all. At the very least, quantifications of the size of purported savings are overestimates taking rebound = 0.5 would for instance halve them, with consequences both for judging effectiveness (or distance from policy target) and costeffectiveness. Whereas Defra in 20052006 commissioned studies of the rebound effect, the UK energy departments Energy Review ignores it, presentingsavingsmeasuresasiftechnological changesintheaveragestockofhouses,appliances,vehicles,powerplants, etc.hasnoinfluenceonenergypricesorconsumerdemand.(DTI2006,36 60,149)

17

Thedebatehasledtoaconsensusthatreboundissignificant.OnenewUK study (Allan et al. 2006) concludes that in contrast to the conventional wisdomwhichequatesengineeringsavingsandrealsavings, wheresubstantialacrosstheboardenergyefficiencyimprovementsare being pursued, future evaluation work [should adopt] an approach which allowsthesystemwideeffectsofefficiencyimprovementstobeassessed (7,53)Ourownempiricalanalysissuggeststhelikelihoodofsignificant rebound effects in responsetosystemwidechangesinenergyefficiency (oftheorderof40%)fortheUKasawhole.Ourworkshowsthatenergy efficiencymeasureswouldgenerallybeexpectedtogeneratealessthan proportionate fall in energy use (rebound), andmayactuallystimulateits use(backfire).(3,51) It is too early to judge whether the policies have an (albeit reduced) effectiveness, or are counterproductive, but a coefficient to adjust engineering savings perhaps 0.5 would be far more accurate than not adjustingatall.

18

Chapter3PolicyContext
3.1Introduction
This chapter relates the technological efficiency strategy to other environmental policies, using a taxonomy suggested by the identity I=PAT: environmental Impact equals Population times Affluence times Technology. After defining these terms it will be shown that the rightside factors are interdependent,i.e.whetheranygivenrightsidepolicysucceedsinlowering environmental impact is contingent upon its not influencing other rightside factors inwaysthatraiseimpact.ThustheformI=f(P,A,T)ismoreaccurate. The question of the relevance of ratios to absolute impact leads into a discussionoftheenvironmentalKuznetscurve(EKC).Finally,theassumption ofconsumersaturationisstatedandevaluated.

3.2TheEfficiencyStrategy
The efficiency policies here examined are just one type of energy policy, namely that serving environmental goals of less pollution and less resource
15 depletion by enabling the same economic benefit with less input. Other

environmental measures seek to substitute away from particular pollutants such as carbon dioxide or fine particles, while other strategies employ consumption caps, quotas, ambient standards, subsidies, and taxes to cut consumptionorinternaliseexternalities.Nonenvironmentalenergymeasures canservesupplysecurityorsupplysufficiencyatpoliticallyacceptableprices. Energyefficiencypoliciesseektolowertwotypesofsocietalcostsofenergy consumption:sourcecostsofdepletionofnonrenewables,overharvestingof renewablesandusurpationofspace,andsinkcostsofnegativeemissionsto

15

Thestipulationthatthesepoliciesnotreduceaffluenceisexplicitintheparadigmaticfactor

fourprogramtodoubleeconomicwelfarewhilehalvingresourceuse.(vonWeizsckeretal. 1997)Indeed,efficiencyisarguablyatooltorescuemaximumaffluencewithinsome ecologicallimitsratherthanatooltoremainwithinthoselimits.

19

16 air, water, soil, and other species. If societyswelfare function is such that

theseoutweighthebenefitsofconsumption,atagivenlevelofcosts,thegoal is consumption/emission reduction as opposed to correction of negative

externalities or increasing affluence the welfareeconomics concept of

externalities is unnecessary to justify the strategy because even if all costs wereefficientlyandjustlyallocatedintragenerationally,theycouldstillbetoo high in terms of a cost/benefit analysis constrained by strong sustainability. These costs are here denoted negative environmental impact. A strategy subsumed under I would directly limit consumption or emissions through source quotas or rationing, while other strategiesindirectlyaimatlessharm throughenergytaxes,substitutionofmoreplentifulorlessharmfulsources,or (ourtopic)greaterutilisationandembeddedenergyefficiency.

3.3 I=f(P,A,T)
17 Microeconomically,loweringtheratioofenergyinputperunitofoutput is

costcutting,butmacroeconomicallyitisachangeintheaggregatefactorTin

I=PAT (Ehrlich & Holdren1971Ekins1991,250),anearlyversionofwhich

stated that resource consumption equals population timesconsumptionof goods (as distinguished from resources) times the technological factor environmentalimpactperquantityofgoodsconsumed.(Ehrlichetal.1973, 206)Theauthorsemphasizedthemultiplicativerelationsontherightside.In theirexampleofautomotiveleadbetween1946and1968,eachfactorwas1 at the beginning a population increase of 42% (1.42) multiplied by a per capitaconsumptionincreaseof100%(2.00),timesatechnologicalworsening (use of lead per kilometer driven) of 81%(1.81)yieldsanincreaseof414% (5.14minus1)inImpact.(214)

16

Parsimonyadvisessubsumingsinkproblemsunderthefurthersourceproblemsofthe

(over)consumptionofcleanairandwater,goodsoil,etc.Exceedingsinkorabsorbtion capacityisthenequaltooverconsumptionofsources.
17

Unlikemostanalyses(e.g.Wirl1997Binswanger2001UKERC2006),outputisdefined

physically,e.g.astheartificiallightinaroomratherthaneithertheserviceoflightingorits monetaryvalueinGDP(Sections6.3&6.4).

20

This taxonomy embeds the efficiency strategy within T, the lowaffluence

18 sufficiencystrategy withinA,thepopulationstrategywithinP,andthedirect

leftside strategies of taxes, quotas, and rationing within I, the former two being ratios and the latter two absolute (or ambient) numbers. T also subsumes economic efficiencies (Smith 1776, III.ii.20, IV.i.3745, IV.ix.17), here termed organisational efficiency since its locus is human groups, includingeconomiesofscale,trade,education,legalsecurity,propertyrights, low transaction costs, Taylorite factoryfloor measures, management hierarchies,etc.(Alcott2005,10Solow1970,3334Greenhalgh1990,293 Sanne 2000, 494) Rebound afflicts all types of efficiency as well as sufficiency:byincrementallyloweringdemandanddepressingprice,marginal
19 consumersareenabledtotakeuptheslack .(Inhaber1997,xii)

Impact is the bad environmental sideeffect of consumption, defined ostensively by Spangenberg & Lorek as climate change, ozone depletion, acidification, eutrophication, biodiversity loss, soil erosion, inland water degradation,wasteproblems,healthrisks,anddepletionofnaturalresources. (2002, 133 Turvey 1966, 4849 Ekins 1991, 244) Energyuse impact is described by Smil. (2003, 105116, 339349) Elaborations on IPAT include the STIRPAT or stochastic impacts by regression model measuring ecological elasticity which, however, does not discuss the technology elasticityofimpactbecausethereisnosingleoperationalmeasureofTthatis free of controversy. (York et al. 2003, 355 Section 6.5) This and other models are consistent with the model of rightside interdependencies
20 presentedhere.(Swaney1991,504Durham1992)

18

Livingsufficientlyinsteadofluxuriouslyisdifferentfromconsumerefficiency(Section

2.2.3)insteadofboilingonlytheamountneededinthekettle,onegoeswithoutthecoffee.
19

Slackresultsasthedemandfunctionshiftstotheleft,intersectingthesupplyfunctionata TheKayaIdentityspecificallytakesImpactascarbonemissionsandTechnologyas

lowerpriceandthuscausingnewdemandbyotherconsumers.
20

primaryenergyinputperGDPunittimescarboncontentperenergyunit.(Nakienovi1997, 272273).

21

Consumption(C)equalsPxA,thenumberofconsumerstimestheamount ofconsumptionperconsumer,yieldingI=CxT,i.e.consumptionmodifiedby technology either in terms of what gets consumed (e.g. more benign hydrochlorofluorocarbons instead of chlorofluorocarbons (Gardner & Stern 1996,1214,19)),orhowmuchgetsconsumed(pergoodorprocess)ofan environmentally relevant substance (like fuel) this dissertations topic.
21 Consumptionisphysicalratherthanmonetaryorutilitarian(Chapter6) yet

normatively ambiguous: for example energy is a good giving life and affluence,yetthesameenergyisthebadofimpact(depletionandpollution). Asimpact,itistheprocessofusingupratherthantheambiguoususe in accordance with Princen (1999, 355) and in contrast to Ayres (1978, 57) and in agreement with its Latin sense of taking with or destruction, being
22 somethingtobeminimized(Boulding1949).

Sincethisdissertationsexplicandumisamountsofconsumption,theterms

sizeorscaleoftenoftheeconomyorofeconomicactivityorjustGDP
willbeusedcrucialfromanenvironmentalviewpointisthisscalerelativeto the ecosystem (Boulding 1966). In line with Dalys (1992) differentiation between economic issues of allocation, distribution, and size, concern with sizereplacestheclassical,welfareeconomicsandenvironmentaleconomics concern with distribution of costs and benefits among economic actors. (Ricardo 1817, 1 & 2 Pigou 1920) In terms of the ecological footprint metaphor, the motivation for studying the rebound question is that the ecological footprint (Wackernagel & Rees 1996) of the world economy is largerthantheworld(Aubauer2006). Figure 3 shows in time series how quantities can be inserted into I=PAT. The T column shows that the technology of this economy determines the

efficiency or ratio of tonnes of coal needed to produce an amount of a

21 22

ThisfollowsSmith(1776,I.viii.21,I.xi.l,II.i.28),Say(1803,38,62),andMalthus(1820,28). Sincehowevermatter/energycannotbedestroyed,thedefinitionofconsumptionrequires

anthropomorphicqualificationintermsofavailability,usefulnessandwaste.(Ayres1978,43 44,119120Alcott2004,612)

22

composite good including cars, driven kilometres, airconditioned driven kilometres,etc.(alsoSchurr&Netschert1960,64)

Figure 3: Total impact codetermined by the technology determining input by weightpergood

I=PATfortonsofcoal
(year)

tons 105 181 192 249 331 353

= population x

(tons/ goods/ person) = per son x tons/goods

1871 1900 1925 1951 1975 1987

31.5 40.5 45.1 50.2 55.9 56.9

=

(3.3)5.3 (4.5)7.6 (4.3)8.7 (4.9)11.3 (5.9)13.1 (6.2)18.7

x
(inputs/ person)

.62 .59 .49 .43 .45 .33

x

FromClapp1994,p.173

In reduced form, I=PAT is false in implying that rightside changes necessarily change the left side. While this is true ceteris paribus as an aggregate, static description, the form I=f(P,A,T) describes the dynamic interdependencies resulting from economic behaviour, rightside changes impactingonotherrighthandfactors,renderingchangeinimpactcontingent:
[b8]

anadditionalinhabitantlowersaverageconsumption:A=f(P) higheraffluenceinfluencesdesiredandpossiblefamilysize:P=f(A) aspopulationrises,diminishingreturnstolandleadtomoreefficient technologies:T=f(P)

as affluence rises, more investment is made in research and development:T=f(A)

technologyinagriculture,sanitation,medicineandenergyefficiency allregardedasexogenousenableahigherpopulation:P=f(T)

technologiesusingelectricityenablehigheraffluence:A=f(T)

23

E.g. if farm efficiency (T) causes population (P) to rise, affluence (A) and impact (I) may well stay the same. The term enable indicates that in all examples something hitherto impossible becomes possible, and is used ratherthancausebecausethebivariatechangesarealsocontingent. This dissertation treats the last two interdependencies: How do technologicalefficiencyincreasesaffectaffluenceandpopulation?(Reijnders 1998, 17) What is the magnitude in physcial, i.e. environmentally relevant terms, of the rise in consumption (AxP) enabledbyfallingenergyintensity andtheconcurrentfallinenergyprice,resp.increaseinpurchasingpower?In the words of Pigou, forwhomnationaldividendsignifiedtotalconsumption or economic scale: All developments of this [efficiencyraising] kind, since they enable something to be produced which was not being produced at all before,orenablesomethingwhichwasbeingproducedbeforetobeproduced moreeasily,mustincreasethenationaldividend.(1920,671)

3.4Ratios
Thetargetsoftheefficiencystrategy,suchaslesselectricityperlightbulb
23 or lumen, are ratios symbolised by TE. Numerators or inputs are to be

lowered relative to denominators or output, whose size is free to grow or shrink.Butbecauseefficiencyisaratio,itdoesntsaveenergyexceptunder conditions that also limit increases in production. (Moezzi 2000, 525)Total quantitiesarethusuntouchedbythepolicies,whichalteronlyeitherlessinput (herefuel)forthesameoutput,thesameinputformoreoutput,ormoreinput butevenmoreoutput.WhilelowernumeratorsdiminishTE (asthesufficiency strategy diminishes A in aggregate average terms), so also do higher denominators in the caseofTmoreproducedgoodsandinthecaseofA more consumers. There is therefore no necessary link between improved
23

Ihavechosenthenumeratorforinputinordertoremainconsistentwith1)thephrase

input/output,whichisintensity,theinverseofefficiency(seeChapter6)and2)thetermTinI

=f(P,A,T)whereinraisingefficiencyishopedtobealoweringofT.

24

efficiency or lower affluence to lower consumption or impact. (Dahmus & Gutowski 2005, 23 Lantz & Feng 2006) Just as falling affluence leaves
24 impact unchanged if population accordingly rises (Malthus 1798, 79, 119) ,

falling energy intensity leaves impact unchanged if consumption proportionatelyrises.DividingTby2doesnotipsofactohalveI.Analogously, policiesrequiringelectricityproducerstouserenewablesinsomepercentof totaloutputsaynothingaboutthenumberofjoulesgeneratedbyfossilfuels.
25 (Section2.2.3)

Khazzooms analysis assumed any positive price elasticity of demand, however small. (1980, 22) The slopes of these functions determine whether rebound is less than or greater than one, but we cannotderive slopes from otherratios.Whilesomereboundisatheoreticalcertainty,itssizeisadifficult empiricalquestion.Anefficiencyshockchangesphysicallyexpressedinput output ratios (Allan et al. 2006, 5), and two other ratios stand between this and final inputconsumption change: that of efficiency change to price changes (of either the products comprising output or of energy inputs themselves),andthatofpricechangestodemandchange.Thatis,weseek theefficiencyelasticityofdemand.(UKERC2006,5)Howdotheseintensive, dimensionless quantities transform into the extensive quantities of
26 environmental interest? The next section examines a similar relationship

betweentheratioofincomepercapitaand(absolute)environmentalimpact, namely the claim that while engineering savings can be eradicated by

24

CorrelationsbetweenPandthelefthandfactorhavebeencomputed,e.g.Shis(2003)

findingthata1%riseinPcausesa1.42%riseinglobalcarbondioxideemissions.(also Raskin1995)
25

Classicaleconomistssimilarlyworkedongettingfromthewagerate(ratio)toabsolute

quantitiesofhoursworked,populationsize,andconsumption.(McKinley1960,9296)Alater challengeistherelationshipbetweenlabourproductivityandthesizeoftheeconomy. (Rosenberg1982,23,55)

26

MalthuscastigatedRicardo:But,toestimaterentandwagesbytheproportionwhichthey

beartothewholeproduce,must,inaninquiryintothenatureandcausesofthewealthof nations,leadtoperpetualconfusionanderror.(1820,164170,175,211)

25

greater consumption, the extent of their eradication depends on income levels.

3.5SaturationofDemand
The environmental Kuznets curve (EKC: an inverted Ushaped curve with GDPpernationorpercapitaontheXaxisandenvironmentalimpactontheY axis)isrelevantbecause1)theproblemofderivingabsolutequantitiesfrom ratio changes is the same in both discussions and 2) one condition for low rebound is consumer saturation, and the EKC hypothesis posits either consumer saturation (declining marginal utility) or structural change to less materialintensityasafunctionofincomeabovealevelthatmayormaynot besustainable.Ifthemarginalconsumptiontriggeredbythepriceorincome effects of efficiency (Section 4.3) is somehow less energy intensive, the efficiency elasticity of input demand is lowered. The results of EKC studies
27 are here ignored in favour of this conceptual question and the

methodologicalproblemsoftheaxesmetrics.

Again, if the independent variable is perperson or countryaverage purchasing power and if the dependent variable is an amount such as total tonnesofcarbondioxideintheatmosphere,causalinfluenceiscontingenton the size of the denominator, in this case number of persons. The further problemisthatoftenthedependentvariableisnotcalibratedasanabsolute number using a measure such asvolumeorweight,oranambientmeasure
3 suchasweightorpartsperm butratheraspollutantorimpactpercapitaor
28 per unit of output (often GDP) some studies use both. When the former ,

some indicators of environmental degradation show the invertedU shape (usually in rich countries rather than the world) whereas others rise
29 monotonically. When the latter , practically all show this shape. This

27 28

StudiesofUKenergyintensityroutinelyignoreairtravel!(Fawcett2004,1080) AbsolutemetricsareinteraliainpartsofShafik&Bandyopadhyay(1992),Selden&Song RelativemetricsareinteraliainpartsofHettige(1992),Panayotou(1993),Selden&Song

(1994),Shafik(1994),andGrossman(1995).
29

(1994,148,151,157),Shafik(1994),Grossman(1995),andHoltzEakin&Selden(1995).

26

ambiguityisfataltocleardiscussionofaffluenceseffectontotalimpactand, sinceefficiencyinfluencesaffluence,toourquestionofefficiencyseffecton totalconsumption.

Moreover,imprecisemetricsabound,suchasenvironmentalimpact,some measures of environmental quality, some environmental indicators, or pollution levels. (e.g. Birol & Keppler 2000, 458459) Efforts at clarity are Jnicke et al.s differentiation between absolute [and] relative structural
30 improvements (1989, 30, 32) and Opschoors (1995) and de Bruyn &

Opschoors(1997)distinctionbetweenweakandstrongdematerialisation, the former denoting merely per unit decrease as of a certain empirically determined level of purchasing power. Honouring this distinction definitely weakens the broad hypothesis that at high average incomes absolute or ambientimpactfalls.(266)Inanycase,studieswithambiguousorrelativeY axesarethusmisleadingbecausemanyindicatorsofpollutionanddepletion rise absolutely while falling relatively (Herman et al. 1989, 50), a further instance of the lack of logical connection between ratios and extensive quantities. Radetzki & Tilton, whose intensity of use hypothesis that metal consumptioncanbepredictedfrompercapitaincome(1990,2629)wasthe 1970sforerunneroftheEKC,thusdoubttheEKChypothesis: Material substitution, new technology, and other factors, may also influenceintensityofuse.[and]arediscreteeventsthatoccuratregular intervals. Their number, speed of adoption,andultimateimpactonmetal use [consumption] in specific applications vary greatly in response to a variety of factors, including the opportunities they provide for reducing costs and expanding markets. The development of the aluminum beveragecan,forexample,increasedtheintensityofaluminumuseinthe UnitedStates.(28,emphasisadded) Also applied to energy, such structural change isnt necessarily dematerialisation. All production requires matter and energy, including the

30

AlsoinGrossman(1995,34),HoltzEakin&Selden(1995,98),Torras&Boyce(1998,149,

167),Johnson(2001,742).

27

productionoflabourandcapital.(Young1991,172)Yetthisisassumedfor instance in the seminal contribution to the rebound literature of Schipper & Meyers, who held the structure of economic activity sectors (agriculture, industry, service) to be one of three major determinants of energy consumption.(1992,5859)Theconceptofstructureherehoweverreducesto the concept of energy intensity: a sectors energy intensity a dimensionlessnumberissaidtodeterminetotalconsumption.However,this chapterhastriedtoshowthatforthischainofreasoningtoholdformarginal consumptiontonecessarilybelessenergyintensiveinaggregatesaturation (i.e.oneexplanationoftheEKChypothesis)musthold,underthedebatable premise that, like Adam Smiths stomach (1776, I.xi.c.7), a human life can
31 handle only so much energy throughput. Structural change is irrelevant,

especially if e.g. energy consumption remains constant or even grows a bit whilefurtherincomeandexpenditureraisesutilitytohigherandhigherlevels: the issue is saturation. Efficiency improvements thus lower consumption marginallyonlyifmaterialsaturationobtains,andthiscanbeestablished,ifat all,onlyempirically. The relatively strong case of the EKC hypothesis against high rebound/backfire thus identifies two kinds of consumer saturation: 1) The nature of utility functions could be such that marginal expenditure is less energy intensive marginal consumption desires are satisfied with lower energy inputs. 2) After consuming more efficiently, we can do absolutely
32 nothing,i.e.chooseleisureornonconsumption.(Section4.3) However,the

EKC concept of (the ratio of) dematerialisation is ambiguous: while it is, in one sense, identical to resource efficiency increase as understood in the debateovertechnologicalefficiencygains,itisherethedependentvariable. Inthestrictlytechnologicalsense,though,itistheindependentvariable.Thus the claim that rebound is low because of structural change to greater efficiency (Schipper & Meyers 1992, 5960 Berkhoutetal.2000,425)begs

31 32

Alargeliteraturedoubtsthis.(Alcott2004) Worldwidesaturationseemsbestignoredforthenextcenturyorsogiventhehuman

tendencyfordisplayconsumptionandthepovertyofover3billionhumans.

28

our question. Finally, the argumentthatefficiencyinducedstructuralchange lowers consumption is reducible to a claim about consumer saturation (Section4.3).

3.6Conclusion
Comparing I=PAT and I=f(P,A,T) sheds light on why this issue is called Jevons paradox. As a static description the former holds: a system with lower T (greater efficiency of any sort) also has lower I. If we respect the consumer,i.e.Jevonsnaturallawswhichgovernconsumption(1865,25), we conclude that production functions alone settle nothing. Including consumer behaviour brings in dynamics and psychology the economic version contradicts the engineering computation, which is therefore insufficienttoguidepolicy.Abroaderformallessonfortherebounddiscussion is that dimensionless numbers say nothing about absolute quantities of environmental interest, because ratio reductions enable both less or more consumption.TheEKChypothesisstatementthateconomywide,asincome perpersongrows,resourceconsumptionfallseitherabsoluteorperperson seemstonotonlylackinferrentialpowerbuttobegquestionsofconsumer demandasafunctionofsmallscaleefficiencychanges.

29

Chapter4MicroeconomicMethodology
4.1Introduction
Empiricalreboundstudiesstartwiththeeventofatechnologicalefficiency increase, then investigate change in demand for the newly more efficiently providedgoodorservice.Then,withefficiencycoefficientsforanyinput,input demandisderivedforthisgoodorserviceonly,yieldingdirectrebound.This chapter attempts clearer definitions and taxonomies of rebound effects, relying on Khazzooms description of a shift in the supply function for the utilisation of newly more efficient household appliances. It identifies the systemboundaryproblemofinvestigatingonlycertainsectorsandcountries totheneglectofotherextantsectors,newproductsorsectors,andtheglobal economy, ending with a criticism of Greening et al.s influential survey of empiricalstudies.

The backfire debate to date lacks clear terminology and taxonomy of the effectsofefficiencyincreases.Forthemicroeconomicconceptofengineering savings, for instance, although precisely defined, other terms abound like technical energy conservation potential, benchmarking, energy efficiency improvement potential, calculated savings, changes from an engineering point of view, gross saving, technical computations, initial benefit, potential dematerialisation, and even begging the question energy conservation.

For the quite clear term rebound one encounters comfort taking, takeback,andfeedback,andforreboundgreaterthan100%ofengineering savings not only backfire but also boomerang, very large rebound and rebound effect writ large (Ayres & van den Bergh 2005, 100). Rebound taxonomies include the categories direct, indirect, own, primary, conventional, secondary, respending, narrowly defined, producer side, substitution, structural, transformational, income, direct price, price, composition, output, sectoral, firstorder, firstcost, Jevons, general equilibrium, overall, longterm, shortterm, economywide, aggregate

30

andactivityeffects.Approachesaretermedmicro,macro,bottomup,and topdown.

4.2TheIncomeEffect
Khazzooms analysis (1980 Section 2.4) showed that lower fuel input requirementsforaunitoffuelrequiringoutput,holdingpricesconstant,mean cheaperunitoutputprices:withamoreefficientcar,aconsumercandrivethe samenumberofkilometresforlessoutlay.Previousbudgetsarenotusedup, i.e.purchasingpowerisnewlycreated.Ineffect,thewholeeconomyhascut energy costs without raising others, an economywideincome effect. To the extent that this purchasing power is availed of, new demand appears for furtheroutputwhichitselfrequiressomefuelorotherenergy.Thenewinput demand is rebound and is some percentage of the demand reduction that would obtain were none of the new purchasing power availed of (i.e. engineeringsavings).Economywide,outputintermsofgoods(orservices, butnotnecessarilymoneyorGDPSections6.36.5)hasrisen.Consumption some combination of population and affluenceishigher.Butsinceinput perunitofoutputfell,consumptionoffuelorenergyinputcouldenduplower (rebound<1),orhigher(backfire).

Itisequallytruethatifsmeltingbecomesmorecoalefficientandtheamount of metal produced remains constant, demand for coal and its price drop. However,withthehigherprofits,andassumingcompetitionamongsmelters, metal prices drop through underselling (Mill 1848, 133134), stimulating demandformetalandindirectlyforcoal.Thegainsofincreasedproductivity have been distributed to consumers by means of price changes. (Salter 1966, 10 Malthus 1820, 52, 55 Jevons 1865, 8, 141, 156 Berkhout et al. 2000, 426) But even if the new demand raises coal prices to a new equilibriumlevelequalingtheold(Saunders1992,135),metalconsumption andperhapscoalinputconsumptionishigherthanattheabstractpointin timejustafterthetechnologicalefficiencyincrease,againimplyingsomecoal consumptiongreaterthanengineeringsavings.Thatis,onceagain,societyis richerintermsofoutput.Withthesameinputslabourandmatter/energyit

31

hasmoreoutput.(Birol&Keppler2000,461)Purchasingpowerisonceagain newlycreated.

FromBrookesperspective: Themarketformoreproductivefuelisgreaterthanforlessproductivefuel, oralternativelyforaresourcetofinditselfinaworldofmoreefficientuse is for it to enjoy a reduction in its implicit price with the obvious [sic.] implicationsfordemand.(2000,355) Inthetermsofclassicaleconomics,theexchangevalueofaunitfallsbutthat of the whole massrises.(Malthus1820,134135,288)(Schumpeter1911, 283, 304 Rosenberg 1982, 106) Because efficiency is a ratio a subtlety arises: Jevons writes, Economy multiplies the value and efficiency of our chiefmaterial.(1865,156)Itakevalueasusevalueorthewholemassof the coal input, if expressed monetarily not necessarily per unit but as total unitstimespriceperuniteconomyreferstoloweringinputperunitofoutput, butefficiencyisclosertoefficacyasintheLatinefficare,tomakeortodo: thelowerratiomeansthatmorecanbeproduced.
33 One conclusion is that efficiency increases always increase total wealth ,

i.e.entailnotonlynonetcosttosocietybutaprofit.(Say1803,71,101,295) Whether they also reduce input consumption is still an open question. A second conclusion is that while real price reductions of energy inputs could occur,bothwithorwithoutthempurchasingpowerincreasesandanincome effect ensues, raising demand. Similarly, in the aftermath of labourinput efficiencyandlowerproductioncostsintextilesMalthusobservedthatthe whole value of the cottons produced in this country has been prodigiously increased,notwithstandingthegreatfallinprice.(1820,31449)Forlabour as well as fuel inputs then, some amount of rebound ensues, which in Malthusviewisindeedgreaterthanthelaboursavingsduetoefficiency: Whenamachineisinvented,which,bysavinglabour,willbringgoodsinto the market at a much cheaper rate than before, the most usual effect is suchanextensionofthedemandforthecommodity,byitsbeingbrought withinthepowerofamuchgreaternumberofpurchasers,thatthevalueof the whole mass of goods made by the new machinery greatly exceeds

33

I.e.somecombinationofpopulationsizeandaffluence.

32

theirformervalueand,notwithstandingthesavingoflabour,morehands, 34 insteadoffewer,arerequiredinmanufacture.(281192,287) The question now arises whether all efficiency rebounds are subsumable undertheterm incomeeffect,enablingamoreparsimonioustaxonomy.

4.3ReboundTaxonomy
The following taxonomy contains five different ways this freed or created purchasingpower(incomeeffect)canbeused: bythesameconsumerforthesameproduct bythesameconsumerforadifferentproduct byadifferentconsumerforthesameproduct byadifferentconsumerforadifferentproduct for leisure, i.e. all consumers reduce their purchasing power to a degreeproportionaltoengineeringsavings. The first and third categories are the wellinvestigated own rebound of the product or service, usually called direct rebound but often confused with rebound, i.e. the generic term. Perhaps deserving of a sixth category is a substitution effect, when energy as the now cheaper factor of production (also in homeproduction functions) is purchased 1) instead of other factors and 2) when the other factors are by some definition less energy intensive thanthatnewlypurchased.(Section4.7)

Where the same consumer decides either to lower his purchasing power by earning less, thus demanding no more goods than previously, or purchasesagoodcontainingorimplyinglessembodiedoroperatingenergy, price falls enable different or marginal consumers to demand more, at no cost to themselves. Rebound study must also includedifferentproductsfor cases where demand for the product or process now more efficiently and cheaply available is relatively saturated or price inelastic. Even in the case where engineering savings are fully realised, i.e.evenifatlowerfuelorcar prices nobody drives more, untapped purchasing power (greater income)
34

SeeChapter7fortheclassicaleconomistsconclusionthatlabourinputefficiencyclearly

resultedinbackfire,i.e.greateremployment.

33

remainsforotherextantornewlyinventedproducts.Evenwithdirectrebound
35 of 100% , if the efficiency elasticity of price is sufficiently large consumers

mayretainuntappedpurchasingpower.Butcertainlywithdirectreboundless than100%,potentialdemandforothergoods,orthesameorothergoodsby otherconsumers,becomeseffectivedemand. Measuring direct rebound alone is of little value. Yet Schipper & Meyers seemtoadvocatejustthis: ...oneshouldexaminetheenergyefficiencyofwelldefinedsectorsorend uses, For refrigerators and freezers there is little, if any, scope for increasedutilization.Forclotheswashersanddryersitseemsdoubtfulthat households would use a more efficient device more frequently than an averageone.(1992,55) Thesectorissowelldefinedthatitignoresevenlargerormorerefrigerators and freezers Greening et al. thus enter for white goods a potential rebound of zero! (2000, 398) Another aspect is that a nonconsumer can becomeaconsumer:Ifduetoefficiencygainsrazorbladesbecomecheaper
th and/orbetter,ashappenedinthemid19 century,peopleshavewhodidnot

before.Withevencheaperblades,oneusesthrowawaydoublebladerazors. And if electricity becomes more efficientlyprovidedandthereforeevermore cheaply available, people switch to less energyefficient electric shavers. Finally,asenergybecomesgenerallycheapereitherperunit(priceeffect) or using up less of ones budget (income effect) other things besides shavingcanbedone. Taxonomy faces the difficult task of incorporating two criteria. Who consumes what? One influential study classifies only categories 2 and 4 as respending in general or income effect (which they deem small), i.e. excluding what I am calling direct rebound. (Schipper & Grubb 2000, 384) Grubb calls this dimension the joker of new demands, rather than the efficiency of meeting existing ones and sees increasingly profligate and wasteful attitudes to energy as consumers become richer. (1990a, 239)

35

Glomsrd&Taoyuan(2005)andDahmus&Gutowski(2005)founddirectreboundbackfire

insomesectors,ashadMalthusforthecottonindustry.(1820,314)

34

Categories 2 and 4 are here called indirect. One similar taxonomy distinguishes sufficiently between direct and indirect rebound effects, but somewhat vaguely between indirect and economywide effects (which together comprise macroeconomic rebound): While the former are said to
36 consistofincomeandpriceeffectsforothergoodsandservices ,thelatter

are, less precisely, cumulative impact of numerous energy efficiency improvementsthroughouttheeconomy.(4CMR2006,4,9,51)

Category5isalimitingcasewhereourwishes,inaggregate,aresatisfied, and we react to lower prices/greater income by working and earning less, taking the efficiency dividend as nonconsuming leisure. In the search for environmentalsolutionsthisemotionallypowerfulvisionhasspawnedalarge literature. (Schor 1992, 1999 Princen 1999, 354 Rpke 1999 Alcott 2004, 776778) Were this case(zeropriceelasticityofdemand)reallytoobtain,it wouldeliminatetheinterdependencyinI=f(P,A,T)betweenTandA:LowerT leads to higher A only under the ceteris paribus assumption of latent consumerdemand.(Jevons1865,25)

Whilecategory5excludesmarginalconsumersbydefinition,somestudies excludethemandtherebycategories3and4formethodologicalreasons that do not imply an assumption of full saturation. Wirl for instance excluded marginal consumers [although] many important energy services are at present consumed by a minority, or a fraction of the households. However,astheefficiencyofanapplianceimproves,moreconsumerscan now afford the now cheaper service and thus acquires the corresponding
37 appliance. (1997, 32) Schumpeter captured this process with the doctrine

that entrepreneursprofitsfromnewcombinationsofproductivefactorsare over time transferred to all members of society. (1911, 297, 301302, 306)

36

ItismoreoeverclaimedwithoutargumentthatTheindirecteffectsfromindividualenergy Wirlremarksthatthisgetsaroundtheconservation[or]energyparadox,i.e.leadsto

efficiencyimprovementsmayberelativelysmall.(4CMR2006,9)
37

seriousunderestimationofrebound.(1997,1932,36,112)

35

Thus, the more so because the energy market is global, measuring the first twocategoriesaloneislikewiseoflittlevalue.

4.4Products,Sectors,Consumers
This dissertation suggests the term income effect as a replacement for many of the literatures ptolemaic array of rebound effects including often vaguemacroeconomicorgeneralequilibriumeffects(butseeSection4.7). It denotes an economywide enlargement of consumer surplus. (UKERC 2006,13)Subcategoriesaredirect,indirect(includingnewproducts,Section

4.5),andleisureeffects(category5).Figure4showshowtheliteraturealmost
universally limits study scope, or system boundaries, thus neglecting openness (Allan et al. 2006, 18), and underestimating rebound. Research beginsatthebottomwithefficiencychangeinaparticularprocess,e.g.kettles boilingwatermorecheaply.

Figure4:Thetaskofmeasuringall(notonlydirect)reboundeffects

AfterEngineeringSavings
r ealboneof contention ignored

BACKFIRE,PHYSICAL GROWTH
INDIRECTREBOUNDwith newpr oducts&mor epopulation INDIRECTREBOUNDwith

100%

underestimated other pr oducts&consumer s undisputed

DIRECTREBOUND withother consumer s

(per cent of eng savings)

0%
TheanalysisofOwenforinstancedetailssavingsduetobetterinsulation andheatingefficiency1995whichhelpedcounteracttheexogenousgrowth of comfort standards and number of households, but does not stray fromthissector.(2000,561)ThatofdeHaanetal.(2006)examinesrebounds 36

of carkilometres and secondcar ownership among new owners of fuel efficient cars, but not other parts of demand stemming from the implicitly increased income of those who can now spend less on fuel, for the same amount of mobility, but perhaps more heating their swimming pools. Binswanger therefore conservatively claims that singleservice model[s] can be misleading. (2001, 121) The consequences of efficiency show no respect for sectoral boundaries such as those of the International Standard IndustrialClassication(ISIC).

Many like Roy attest the difficulty of measuring indirect effects, i.e. a whole range of behavioral responses of the endusers that follow any technical efficiency improvement all of which may, however, not be traced empirically. (2000, 433) Grubb sees after literally hundreds of scenario studies the hopeless complexity even of comparing enduse efficiency of
38 hundreds of different processes (1990a, 235, 195 Chapter 6) For

Khazzoom the problem is that price elasticity estimates are not detailed enough to allow us to pinpoint for all enduses where an improvement in applianceefficiencywillresultinanetincreaseinenergydemandandwhere it will result in energy saving. (1980, 32) But unlessallconsumerreactions are investigated, we cannot answer our question and DTI can conclude counterfactuallythatOverthelast30yearstheeconomysenergyintensity has improved by around 1.8% each year. Without this, home heating, for example,wouldusemorethantwicetheenergyitusestoday.(2003,32)Did forinstanceenergyintensityhavenoinfluenceonthenumberofhomesbuilt duringthose30years?

Other income effects result from organisational efficiency gains (Section 3.3), usually subsumed under Paretooptimal improvements, or from
increases in human skill, sometimes regarded as greater human capital

38

HowarthcallsthesortingoutoftheempiricaldimensionsoftheKhazzoomBrookes

hypothesis[an]ambitioustask.(1997,4)Greeningetal.writethatonlyageneralequilibrium analysiscanpredicttheultimateresultofthesechanges.(2000,396Brookes2000,360, 362Binswanger2001,122126)

37

efficiency. Similar in effect to discovering higherquality or more available natural resources, these are of net benefit to society economically and universallyseentobegrowthdriversinphysicalaswellasutilityormonetary units as opposed to technological efficiency gains, which are sometimes saidtoreducegrowthinthephysical,materialsense.

4.5OutputQualityandNewProducts
Technological efficiency improvements are easily measured if the output stays the same in all respects, aside from input requirements e.g. an

automobile kilometre holding weight and other qualities of the car constant.
39 However, car weights, sizes, etc. do change. When cuttingtools change

fromsteeltoceramicstocarbide(diamonds),arethesenewproductsorjust more efficient cutting tools? (Rosenberg 1982, 65) Surely completely new products emerge: the first kettle was perhaps a more efficient version of an open pan, but the first pan, and the first electric current generator, were simplynew.

This affects rebound bookkeeping and taxonomy when 1) new products representnew,additionaldemandforsomecombinationofoutputandenergy input and 2) efficiency improvement was a necessary condition for the new product.Thisisdifficultbecause aninnovationfromoutsidewill[often]notmerelyreducethepriceofthe product in the receiving industry but will make possible wholly new or drastically improved products or processes. In such circumstances it becomes extremely difficult even to suggest reasonable measures of the payoffs [rebound, growth] to the triggering innovation, because such innovations, in effect, open the door for entirely new economic opportunities and become the basis for extensive industrial expansion elsewhere.(Rosenberg1982,75) Therefore attempts to measure more than direct rebound must judge efficiencygainscontributiontotheinventionofnewthingslikesyntheticfibres or lasers. Early on, Raeobservedthatinventionandwealthreinforcedeach other: improvement in steam pumps suggested new areas of application,

39

Sometimesatthecostofgreaterfuelinput,aswithheavierorairconditionedcars,orother

materialinputaswhengreaterboilerefficiencyitselfrequiresthickerwalls.

38

namely transportation and milling, while printing efficiency enabled applicationsfornotonlyreadingbutforplayingcardsandcoinmintingaswell ashugenumbersofBibles.(1834,245248)ButtomyknowledgeJevonsis theonlyauthortoincorporatesuchefficiencyeffectsintoatheoryofrebound (Alcott2005,1315),andthisareaofindirectreboundseemstobeoutsidethe remitofcurrentUKstudies(Chapter1).

Can examples make this category plausible? Rosenberg gives several: Neilsons hot blast smelting process enabled the use of anthracite coal, turning this fossil fuel into a resource in the first place and causing its consumption. (1982, 23) Henry Maudsleys slide rest for turning resulted in better locks at lower cost, expanding demand for locks and their material energy inputs. (47) The compound steam engine had toawaitcheap,high quality steel, i.e. smelting efficiency enabled pumps, steamships, and railroads and thus higher demand for steel [and coal]. (61, 157, 187 1994, 166)Moreefficientproductionoffertilisersandsubsequentpricedeclineled toanincreaseoffertiliserandlanduseaswellasfood.(75)Balzhiserattests acorrelationbetweendecliningelectricityintensityandacontinuingstream of new industrial applications [and] new uses for electricity and a continued growthincommercialsectors(Balzhiser1989,99Price1998,7879)

Thus, some of the knowledge that makes it possible to produce (1) a greater volume of output or (2) a qualitatively superior output from a given amount of resources is knowledge of more efficient processes. (Rosenberg 1982,34Schumpeter1911,287288)Ontheotherhand,whilethepuddling steelmaking technique turned previously unused impure or scrap iron into (useful)resources,enlargingtheresourcebase(87)andcausingevenmore fueluse,ispuddlingmorefuelefficientthanearliertechniques?Isthegreater fuel efficiency of aircraft engines not offset by their greater thrust, lowering efficiencyscontributiontoincreasedfuelconsumptionentailedbyincreased airtravel?(127128)Efficiencygainsurelyloweredinputcosts,butnotallof the new demand for air travel is attributable to such a rebound effect,

39

40 stemming perhaps instead from changed tastes. Finally, new products are

also crucial to any rebound beyond direct rebound because they can shift preferences away from leisure and nonconsumption. In sum, while the environmental energy policies here examined define themselves in terms of

existing products and/or processes, in the longer term new products and
activitiesappear.Whatnewusesaredevelopingatthismomentastheresult ofpriorefficiencygains?

Taking energy consumption as the dependent variable, it is difficult to isolate the role of input efficiency, ceteris paribus. If we expand the system boundary backwards to mining, we see that lower costs of obtaining usable energy constitute an efficiency increase with effects in every economic
41 sector. Andweretherenoqualitativelychangedornewproductswhichhave

efficiencygainasaprecondition,consumersaturationwouldbeconsiderably nearer. Moreover, efficiency and invention reinforce each other. The conclusion of Rosenbergbecomesahypothesisforreboundstudy,viz.,that new technologies that improved energy efficiency have often led to a significant increase, not to a reduction, in fuel consumption improved energyefficiencymayitselfstimulatenewusesofenergythataredifficultto anticipate.(1994,165,166)

4.6CountriesorWorld
Investigationsofonlycertaincountriesexhibittwoproblems: Marginalordifferentconsumersinothercountriesareignored.Butif efficiency lowers fuel prices internationally, and fuel is traded, the income effect can be enjoyed by anyone. Moreover, not only does povertyassurelittlesaturation,butsomesuggestthatmarginalenergy intensitiescouldbehigherindevelopingcountries.(Binswanger2001, 126 Shi 2003, 32, 3839) Therefore Schipper &Grubbcorrectlynote thattheirfindingscannotbeextendedreadilytodevelopingcountries
40

Capitaljunkingoreconomicobsolescenceinconnectionwithnewproductsraises ButEROIalsofalls(diminishingreturns)asmoreavailablesourcesareexhausted.

consumptionfurther.(Babbage1832,231233Rosenberg1994,34,51)
41

40

(2000,387),buthowusefulistheirconclusionthatfeedback[rebound] effects are small in mature sectors of mature economies? (368 Ayres2000Opschoor2000) Duetoenergythatisembodiedintradedgoods,studiesofcountriesor groups thereof should always correct for trade. (Cleveland & Ruth 1998, 4445) While data especially on embodied energy imports are hard to compile, there is evidence that goods imported to the often investigatedOECDcountriesarefromcountrieswithrelativelyenergy intensive technology. (Greenhalgh 1990, 298 Hinchliffe 1995, 94 Saunders 2000a, 439) The methodology for computing relative ecological footprints, though, using material flows and inputoutput analysis,iswellknown.(Wackernagel&Rees1996vandenBergh& Verbruggen 1999) Furthermore, evidence that declining rates of increase in total OECD consumption is at least compensated for by developing countryconsumptionisempiricallycollectedonaregular basis.(e.g.Brown1998,114)ThishardensthethesisthatbothEKCs and the gap between trend projections and actual consumption in evaluationanalyses(Section2.3)tosomeextentresultfrommarginal, nonOECDconsumersandimportedembodiedenergy.

Nevertheless most rebound studies do not systematically integrate these crossboundary movements into calculations either of consumption or of energy intensity of production. (Sebold & Fox 1985 SchmidtBleeck 1999, 24 Greening et al. 2000) But surely structural change from agriculture to industry to services in rich economies (Jnicke et al. 1989, 2426) depends partly on imports of agricultural and industrial goods. One recent study acknowledgessuchimports,inconsequenceoftheincomeeffect,notingthat the rebound effect defined for UK consumption is thereby reduced but it seemstoignorethemduetomeasurementdifficulties.(4CMR2006,24,52 53) Outside the rebound literature, however, Dahlstrm & Ekins (2006) for instanceintheirstudyofthesteelindustryintegrateflowsofenergyintoand outoftheUK.AndWeiszetal.explicitlyconceptualisetheproblemthatsince thenationstateasaunitofanalysisformaterialfinalconsumptionand the nation state as a territory where certain materials are extracted and 41

transformedintoproductsaredriftingmoreandmoreapart,crossboundary
42 analysisisessential.(2006,694) Tomeasurethis,onecomparisonoftrade

between Japan (an Annex I country) and South Koreausesthebilateral balanceofemissionsembodiedintrade(BEET).(Rhee&Chung2006)

For our question, country studies seem of little value in the absence of a methodologytoincludeinternationaleffects.Theassumptionofonestudythat UKefficiencypoliciesconcededtolowercostsandpricesnationallyhave no effect on international prices seems moreover unlikely. (4CMR 2006, 9) For a wide range of environmental issues, the concept of the country or economyindeedcouldbelargelyobsolete.(SaintPaul1995)

4.7SubstitutionandTimeEffects
Conceptual difficulties in rebound taxonomy remain. Schipper & Grubbs categoryofmicroreboundincludesasubstitutionrebound,anothervariety ofthemicrorebound[which]couldoccurifhigherefficiencythroughlowering thecostofusingenergyleadstosubstitutionofmoreenergyforotherfactors ofproduction.(2000,368)Burniauxetal.write: There is a link between technical progress, output prices and real income[T]heriseinenergyproductivitytendstolowertherelativeprice of energy, thereby generating a substitution effect from nonenergy towards energy goods. In the aggregate the increase in autonomous energy efficiency also generates a real income gain that leads to higher consumption of both energy and nonenergy goods. The netresultisthat emissions do not decrease in the same proportion as the AEE increase because the energy conservation effect is partly compensated by the 43 relativepriceandincomeeffects.(1995,246)

Jevons explanation of greater coal consumption similarly included the fact thatcoalcostshadremainednearlystableinseventyyears,whiletheprice oflabourandallmaterialswasdoubled.(1865,118)

42

Thedauntingtaskofcomputingthisforallpairsofcountriesisprescribedinthemonitoring Hereagainisaracebetweenefficiencyandemissionsmoreover,itisnotclearhowthe

proceduresforeseenbytheKyotoProtocol.(UNFCCC,Art.5.2)
43

authorsmeasureAEEchange(Chapter6).

42

PerhapsthesubstitutioneffectstrengthenseachcategoryinSection4.3.On the other hand, does it matter to rebounds size whether the newly cheaper energy is bought instead of something else, e.g. labour, orin additionto it? Consideratwoproductmodel.Iftheyareperfectsubstitutes,consumptionof the newly cheaper one, X, entails a onetoone decrease in consumption of theother,Y.ExpenditureonXrises,butthisismorethancompensatedwithin hisbudgetbythedecreaseinhisexpenditureonYbecauseYis(remains) more expensive. The result is that the consumer has the same amount of product (or utility) with purchasing power left over (income effect). On the otherhand,ifapotentialincreaseinconsumptionofXcannotsubstituteforY and consumption of units of X do not increase, then because X is cheaper purchasingpoweragainremains,potentiallyforZ.

Often substitution effects are taxed as price effects because the more efficiently used input becomes cheaper. However, price effect often refers instead to the price elasticity of the relevant output. When this is low, then

directreboundislow:WhenLovinsassumesthatthepriceeffectisgreater
than the income effect, he can only be measuring direct or own rebound. (1988,158)OnecaninanycaseconcurwithCleveland&Ruthsdescription: Energy efficiency gains look to the consumer a lot like price reductions, spurring increased demand for energy either directly through price elasticityeffects(e.g.peoplebuyingmoregasolinewhenitspricedrops)or indirectly through released purchasing power redirected to energyusing goodsandservices.(1998,44) Thealternativepurchasesareonaveragemoreenergyintensive.Ofcourse, notonlyisincomeredirectednewincomeiscreated. AnothertypeofeffectthatisdifficulttocategorizeisidentifiedbyCogoy:If technicalprogressimprovesthetimeefficiencyofenjoymentprocesses,room for enlarging existing types of consumption activities or for introducing new ones is opened. (1999, 388) This has been called a rebound effect with respect to time. (Binswanger 2001, 127ff) Like EROI increases, this would seem to pervade all sectors and marginal consumers. The relationship

43

between material and time efficiency deserves much more attention: freed
44 timeisalsothepreconditionfortakingefficiencygainsasleisure.

4.8Berkoutetal.2000andGreeningetal.2000
This section criticises twoofthemostwidelycitedpapersintheliterature, illustratingproblemsofthemicroeconomicapproach.First,bothbetweenthe
45 papers and within each, definitions are either lacking or contradictory.

Berkhout et al.s (2000) taxonomy of effects includes at least 13 classes, inconsistently used. When results are shown of rebound between, say, 0 and 60%, one cannot tell which definition is operative: direct rebound and the overall, broader rebound effect, RE are not clearly distinguished. Therefore their mainconclusionThereisagreementthattheREissmall (431) is misleading. While the evidence they offer shows that direct or lowerorder rebound is well less than unity, they switch to the general categoryREfortheconclusion.Asinmuchliterature,thetermreboundis ineffectcooptedtomeandirectrebound.

Second,whiletheauthorsacknowledgethatnondirectreboundisdifficult
46 to assess , and note that the reduced energy bill enlarges purchasing

power, they nevertheless cite and endorse the results of a longterm macroeconomic study concluding that [p]robably, the secondorder effects willbesmallerthanthefirstordereffectslikeastoneinthepond(425427). Finally,onecouldarguewiththeiruseofpriceelasticityofenergydemandas theproxyforthe(rebounddetermining)efficiencyelasticityofdemand.(428 430)

44

Jevonsremainedtruetoforminnotingthatalthoughcoalwassubstitutedforwood,woods Adetaileddocumentprovingthisisavailablefromtheauthor. deHaanetal.maintainthatevensomedirectreboundsaretoodifficulttomeasure,namely

usealsorose.(1865,249)Wascoalsubstitutedforwood?
45 46

thenumberofreboundvehiclekilometresconsumedbypurchasersoffuelefficientPrius carstheythereforerestrictthemselvestocarweight,andnumberofcarsperhousehold. (2006,596,603)

44

Greeningetal.(2000)likewiseontheonehandreturntoanarroweruseof thetermreboundortakebackandemployothertermsfortheeconomywide phenomenon, yet on the other hand extend [their] typology beyond its origins in the microeconomics literature to include secondary and economy
47 wide sources. (390) They likewise shift without explanation from direct to

total rebound. First they give secondary evidence from 74 studies that a

givenfuelusingactivityrevealslowreboundandattestsatiationforagiven
service. (390, emphasis added) 48 They then state the nonsequitur that increasesinefficiency,althoughpartiallyoffsetbyincreasesinconsumption, willresultinanoverallreductioninenergyconsumption.(392)Theycallthe lowtomoderateestimatesofdirectreboundestimatesofreboundassuch. (399) Moreover, no proof is offered that secondary effects are probably insignificantorrelativelyminor.(391)Insum,theirevidenceisaboutdirect rebound,yettheirconclusionsareabouttotalfuelconsumption(392)

Greeningetal.identify of market response to changes in fuel efficiency: (1) direct rebound effects, (2) secondary fuel use effects, (3) marketclearing price and quantity adjustments (especially in fuel four categories markets) or economywideeffects,and(4)transformationaleffects(390) They however regard the last two effects too difficult to investigate, and the secondasinsignificant.Thus,exactlytwentyyearsafterKhazzoomsoriginal contribution, we are left with direct rebound only but direct rebound cannot speakforthewholereboundeffectasthebasisforpolicyrecommendations. The epistemology of these studies is inadequate for the badly needed clarificationofthedebate.

47

Including(exogenous)fuelmarketsextendsthetypologytoofar,becauserebound Forspacecoolingninestudiesreportdirectreboundfrom050%theresultforlongrun

measuresonlyconsumerresponsetoefficiencychange.
48

aggregateimpactsisgivenas<1000%forappliances,althoughthestudieswere inconclusiveinresultsandindirecteffectsintermsofthepurchaseoflargerunitswithmore featureswerereported,theirfigureforpotentialsizeofthereboundisastraightforward0%.

45

4.9Conclusion
Estimating rebound presupposes definitions, taxonomy, and transparency about exactlywhatpartofnewconsumptionintheeconomyismeantwhen theseareuncleardescriptionsofbackfiresuchasthefollowingaredifficult: Ifsavingenergyistoleadtogreater,notlessenergyusethanotherwise, theneitherenergyintensitiesafterenergysavingmustsomehowwindup greaterthanbeforeenergysaving,ortheactivitiesandoutputforwhichthe savings were made must increase by more than the savings decreased energy use, or the overall mix of output must evolve in a way towards greater, not lower energy uses [sic.] than otherwise. (Schipper & Grubb 49 2000,383) Heightened demand for the activities for which the savings were made correspondstomycategories1and3,whiletheoverallmixcoverseither2 and4orallfour(Section4.3),theevolutionofwhichthewritingsofbackfire theorists from Jevons through Saunders have tried to describe. It is hoped that the taxonomy proposed in this chapter, and the identification of misleadingsectoralandcountryboundariesinreboundstudies,contributeto bringingsomelightintotheblackboxofthemacroeconomiceffects.

Schipper&Grubbwrite: Overall,ouranalysisofdisaggregatedsectoralandsubsectoralenergyuse and activity trends in a variety of IEA economies suggests that any feedback effect is small compared to both the effects on energy use of changesinenergyintensitiesandoveralleconomicgrowth.(2000,367) We have shown the severe limitedness of sectoral and country studies.

Section 5.3 questions treating overall economic growth only as an

independentvariable.However,themostdoubtfulelementofthisconclusion regards changes in energy intensities: Since energy intensities are technologicalefficiency,thefeedbackeffectispartofthechangeinenergy intensityeffectandcannotbecomparedwithit.Theseintensitychangesare theindependentvariablecausingthereboundeffect.Empiricalworkseemsto needabettertoolkitofdefinitions.Onceilldefined,thetaxonomyofrebound isasaruleunparsimoniousandcontradictory.

49

Thephrasessavingenergyandenergysavingbegourquestion.

46

Chapter5MethologicalProblems
5.1Introduction
Thecontentiousbackfirequestioninvolvesacerbicexchanges(Lovins1988 Khazzoom1989Grubb1990bBrookes1992Grubb1992)Perhapsbecause of ambiguous definitions and nonrigorous taxonomies and because study scope is limited to sectors and countries, estimates of rebound (direct rebound?), range from insignificant (Lovins 1988, 156157) or small (Schipper & Grubb 2000, 367368, 384386) to greater than 100% (Jevons 1865 Brookes passim Saunders passim). This chapter continues to try to clarify a number of definitional and methodological issues that are inadequatelydiscussedintheliterature(UKERC2006,3),bothtoexplainthis variationandtoquestionsomeassumptionsbehindreboundmeasurement.It critically examines five concepts for their relevance to the measurement of rebound and their employment in arguments for low rebound: population,

economicgrowth,energysproportionofGDP,marketfailure,andthecostsof efficiency.
Growth of efficiencyinduced energy input consumption is our dependent variable.Wedistinguishbetweenthisandthevaguetermeconomicgrowth, which can use monetary, utility or service or physical metrics. When for instance Birol & Keppler write that It is the rebound effect that translates technological efficiency improvements into economic growth (2000, 462), they are not endorsing backfire because their metric is monetary and thus inclusiveofnonphysicalservicemeasures.Twopremisesofthechapterare thatenvironmentalenergypolicymeasuresonlythephysicalscaleofenergy consumption, and that rebound is calibrated onlyasapercentoftheoretical inputsavingsduetotechnologicalefficiencychanges.

5.2Population
Population increase isoftensaidtocausealargersizeoftheeconomya populationeffectexpandingconsumption(Schipperetal.1996174)isseen 47

torace(Section2.2.3)againstefficiencygainswhichareshrinkingit(netof somerebound).Lantz&FengforinstancecompareelasticitiesbetweenCO2 emissionsandthethreedriverspopulation,GDPpercapitaandtechnology, the former two increasing, the latter decreasing, emissions however, if population is not wholly exogenous, but itself influenced by more efficient technology, its coefficient (0.8 compared to 0.1 for technology) is probably overstated. (2006, 235) Howarths model of energy consumption similarly containstheexogenousfactorconstantrateofpopulationgrowth.(1997,4)

But what causes population increase? The claim here is that while population times affluence is consumption, neither is purely exogenous but rather determined partly by institutions and organisationalandtechnological efficiencies, the latter our independent variable. Greater production per hectare,perworkhour,permachineandperunitofmaterialinputsupportsa greaterpopulation,apositionofclassicaleconomistswhothusplayeddown the explanatory importance of merely settling new land or working harder. (Malthus1798,7475,206Jones1831,196)MorerecentlyCipolla(1962,49
50 53, 9495, 105), Clapp (1994),Johnson(2000)andothers ,usinghistorical

methodology, argue that without efficiency increases in producing food, warmth, clothing, transportation, metals and fuel, neither affluence nor population could be of todays magnitude. This efficiency/population connection is put by Giampietro intermsofMalthusianinstability,Jevons Paradox and the desire for greater affluence a ratchet effect showing that, like land availability, productivity enables more people to live healthy livesthetendenciesofhumanreproductivebehaviour(Malthus1798,52,70) add to this necessary condition to yield sufficient conditions for population growth. (1994, 680681) Hannons argument is formally similar, that vegetarianism doesnt necessarily lower impact because of a reaction that peopleseemtohavetoeatingloweronthefoodchaininthefaceofland scarcity:producemorepeopleperfamily.Thesegainsinefficiencyallowfor largerpopulations(1998,215)

50

AlsoPimentel(1994),Bartlett(1994),Giampietro&Mayumi(2000).

48

What is the relative size of an endogenous, as opposed to an exogenous component of population? And of the endogenous part, how much is explained by technological efficiency gain? Whatever the proportions, this contribution of efficiency to energy consumption, through enabling larger population, must be booked under rebound, unlike the part stemming from biology and sexual or parental emotions. Failing to do this leads to underestimatingrebound,aswhenSchipper&Grubb,forinstance,although they normalise observations of absolutequantitiestoeitherpopulationor GDP, see none of this significant population growth as stimulated by the increasesinenergyefficiency.(2000,368)

5.3EndogenousGrowth
Thegreatestpartofreboundliteraturethusexplainsgrowthinthenumberof joules consumed with three variables: economic growth, population growth and pulling in the other direction and thus begging our question energy efficiency increases. When considering a trend holding technological efficiencyconstant,forinstance,Schipper&Grubbwrite: Of course, the scale of the system keeps increasing with population, household formation, and the climb of incomes and sectoral output [O]ver the time it takes to implement techniques that save energy, the entiresystemisgrowing.(2000,368,370) Alternatively, population can be subsumed under growth and an additional factoradded,aswhenSchipper&Meyersusethethreeimpactfactorslevel ofaggregateactivity,thestructureofactivities,andtheenergyintensities of specific types of activities (1992, 58), or when Manne & Richels three exogenous, mutually independent parameters are potential GNP growth,
51 elasticity of price induced substitution between capitallabour and energy ,

andtherateofautonomousenergyefficiencyimprovements(1990,51)

Thisparadigmoftheracethusenablesthestatementthat[e]conomywide fuel demand can increase in the face of fuel conservation simply because economiesgrow.(Saunders2000a,442Fouquet&Pearson2006,172)But
51

Themoresosinceefficiencyincreaseslowerprices,treatingpricesassolelyexogenous

likewisebegsmanyquestions.

49

why do incomes, output and populationincrease? Schipper & Grubb further

speculate that [w]e still may find that energy use has increased even if energy efficiency has improved, because population or GDP grew significantly, growth not significantly stimulated by the increases in energy efficiency. (2000, 368) This seems to concede, in agreement with the analysishere,thatefficiencystimulatesGDP,butthatthisisnotsignificant. But instead of merely lending economic growth itself autonomous causal power,weshouldaskwhatdoesthencausethisgrowth. ItseemscorrectforthereboundandEKCliteraturetopartlyexplainthesize oftheeconomyendogenouslyintermsofeconomiesofscaleorofpopulation andinfrastructuredensity(Petty1675,255)or(beggingourquestion)interms of more research and development for efficiency increases. Difficulties of circularitynotwithstanding,growthitselfisinthissenseexplicans.Buttothe extentthatproductivityleadstogreaterconsumptionifthisisconceded also explicandum. Schmookler, for instance, in conceptualising technology and economic growth as both independent and dependent variables, emphasies that the amount of invention is governed by the extent of the market. (1966, 104106 Rosenberg 1982, 141) But he ignores the reverse causality attested by the classical economists and Solow, namely that the extensionofthemarketisitselfafunctionoftechnology.(Say1803,8687, 295301Malthus1820,5356Mill1848,8789Solow19571970)

If we are to avoid begging our question, then in addition to undisputed forces like organisational efficiency or population (to its exogenous extent), engineering innovations should no longer be assumedatoneandthesame time to raise economic growth but lower growth in energy consumption. When,however,inexplainingenergyusethescaleoftheeconomyistaken as wholly exogenous, efficiency of energy use is ruled out as a cause of growth of energy use. If economic growth causes greater energy consumption, as in the trend scenarios of evaluation studies, yet efficiency

cannot cause economic growth, then rebound is by assumption

underestimated.Ifontheotherhandgreaterefficiencycancauseeconomic growth,butthisisdefinedintermsofGDPorutility,thedifficultquestionsof 50

the decoupling of growth from physical matter andenergyonceagainarise. (Section3.5)

Inotherwords,Jevonsparadoxoncemorearises,duetotheallimportant implicitdistinctionbetweengrowthofoutputandgrowthofinputslikeenergy. Efficiencyinduced income effects an economywide increase in absolute purchasingpowerareontheonehanduniversallyconcededtocausesome increaseindemandforinputscomparedwithengineeringsavingsotherwise rebound equals zero. On the other hand they are said themselves to lower absolute consumption levels of energy inputs, and actual measured rises in energy consumption are then attributed to the exogenous drivers like populationandGDP.Onlytheambiguityinthetermgrowthcanexplainthis. Onlyifgrowthofoutputoreconomicactivityisconceptuallydecoupledfrom
52 growth of physical throughput can efficiency increase raise one while

loweringtheother.Whereasifahigheroutput/inputratioisenergyefficiency increase,ourquestionisbegged.

Inanycase,onecontributiontotheliteraturewithinthisparadigmthus(self contradictorily)concludes: [I]ncreased growth should, all else equal, lead to increased energy demand. If this growth effect is large enough, it might counter the direct effectsofreductionsinenergyoutputcoefficientssothatimprovedenergy efficiencyactuallygivesrisetoincreasedenergyuse.(Howarth1997,2) Again,notonlyisthegrowtheffectfullyexogenous,butitisunclearwhatthe roleofimprovedenergyefficiencyis.Howarthcallstheratioofenergyuse togrossnationalproductacrudebutfrequentlyemployedindicatorofenergy
53 efficiency(2) ,butenergyefficiencydenotesexactlythisratioinadditionto,

ambiguously,morenarrowtechnological,physicalprocesschanges.

Pertaining directly to rebound measurement, Howarth then calculates that for the years 19291970, [t]o accept the KhazzoomBrookes [backfire]

52

Throughputisinputofmatter/energyintotheeconomicsystemplusoutputofgood SeeSection6.3.

productsandwaste.
53

51

hypothesisonemustassertthatimprovementsinenergyefficiencywere responsible for a full 29% of the increase in gross national product that occurred during this period he opposes this to Denisons finding that productivity improvements and capital investment account for some 54% of theincreaseinnationalincomebetween1929and1982.(4)First,again,this leavesoutofconsiderationthatlabourortotalfactorproductivitycouldresult partly from energy productivity. (Schurr 1982) Second, Howarth then dismissesthe29%figureabove:Claimsofthissort,however,seempalpably implausible. (4) They are empiricallyimplausible(7),butplausibilityisnot defined. In the absence of clear terminology he appeals only to casual observations(3)andtheintuitionsoftechnologyanalysts(2)inarguingfor lowrebound.Schipper&Grubbsimilarlyassert: It is fanciful to suppose that the 20% energy savings, together with the restrainingeffectofstructuralchanges,wouldboostGDPby33%.(2000, 384,emphasisadded) Suchargumentation,however,isunscientific.

Howarthfurtherstatesthat[g]rowthaccountantshavenotidentifiedenergy efficiencyasanessentialaspectoftheircalculations(4)Thisadhominem argument seems to ignore, however, a large body of literature including Solowscalculationthattechnologicalprogressisstrongerthanorganisational progress (1970, 3335, 38), and that together they account for 87.5% of growth(1957,320).Thus,aproblemfortheparadigmthatapriorideniesthe growtheffectsofefficiencyisthatthefactorswhichthendocausethegrowth mustbePromethean.Theymustnotonlycompensatefor,butovercome,the consumptionshrinkingeffectsofefficiency.

5.4SmallShares
A common argument for small rebound is that energys share of GDP is small. In his attempted rebuttal of Brookes (1990), for instance, Grubb argues against the significance of a respending effect [where] purchasing power is released for other energycontaining services he asserts that because total energy costs are generally a few percent of GDP this effect will, even if one assumesalinearenergyGDPrelationship,beofthis 52

order. (1990, 784) Greening et al. similarly claim that Since energy is a relatively minor share of an individual consumers total expenditures, the secondaryeffectsareprobablyinsignificant(2000,391),whileSorrellholds thatIndirectreboundeffectsmaybesmallbecauseenergytypicallymakes upaminorshareoftotalconsumerexpenditure.(2005,2.2Schipper2000, 353) HowarthsubstitutesservicesforGDPandseestwosmallratios: Since energy costs are typically a small component of the total cost of owning and operating energyusing equipment, large improvements in energyefficiencymaygeneratecomparativelysmallreductionsinthecost ofenergyservices[and]energycostsareasmallcomponentinthecost of energy services, so that large changes in energy intensities are accompaniedbymodestchangesindecisionmakersincentives.(1997,3, 2) If energy/energy service and energy service/GDP are both small, rebound melts. Even some researchers who measure more than just direct rebound, and thus deem it relatively large, seem to measure rebound not against engineering savings but GDP: It is true that the significance of [the broad incomeeffect]ismoderatedbyenergysshareintherelevantscalevariable (e.g. GDP in a KLE valueadded production function), which is typically of modestscale.(Allanetal.2006,1819)

But whether small is 0.5% or 6% (Howarth 1997, 7), the argument misunderstands the concept of rebound. Rebound does not depend on the percentage of GDP of engineering savings, but of the percentage of engineering savings of new consumption enabled by lower production costs andmediatedbytheefficiencyelasticityofdemand.Thisconsumption,resp. the potential energy savings, could be a tiny part of GDP or total energy consumption and still be greater than engineering savings. This issue is purely conceptual: rebound is not a claim about everything elsegoingonin theeconomywhichmightincreaseenergyconsumption.

53

5.5Marketfailure
Anothercommonargumentclaimsthatreboundislowerthanusualinareas oftheeconomydominatedbymarketfailure: The sharpening energy efficiency debate has revived claims that the apparentsavingsfromusingmoreefficienttechnologieswouldbelargely offsetbythemacroeconomicresponsethetendencytousemoreenergy services because they are made cheaper. When energy price or availability constrains demand this is correct. In more normal circumstances,itmaybetrueforefficiencychangeswhichoccuraspartof general economic trends, but not for policydriven measures aimed at bringingmoreefficienttechnologiesintoimperfectmarkets.Insuchcases there will usually be real net energy savings only slightly smaller than those suggested by a simple engineering analysis, and clear economic benefits. Conservation policies can choose to focus upon areas, includingthosedominatedbymarketfailures,wheretheimplicitpricefalls from increased efficiency have little impact on activity levels. (Grubb 1990b,783) First, when is energy price or availability not a constraint on demand? Second, while conceding that higher BAU efficiency causes enough consumption growth to largely offset the lowering effect of more efficient technologies, Grubb claims the opposite for nonBAU changes, i.e. those forcedbypolicy.Butwhy,intheseareas,doestheincomeeffectnotobtain? He suggests seven states of affairs of market failure including lack of knowledge and skills, separation of expenditure and benefit, and rapid paybackrequirements,butnodetailedargument.(1990b,784785)

Perhaps the technical change is not carried out in the first place. Or low priceelasticityofdemandbrakestheeffectsofpricefalls,orpricesdonotfall due to lack of competition. Concerning the first (that market failure leaves actualefficiencybelowtechnologicallypossibleefficiency),thisisnotrelevant to measurement of rebound, but only to engineering savings, against which rebound would be measured, were there any. Indeed, the issue of market failure seems relevant only if one assumes efficiency to be environmentally efficate: Grubb and Lovins argue that energy efficiency policies primarily addressmarketfailures/barrierswhichpreventcurrentlymostenergyefficient

54

54 technologyorsystemsbeingused.(4CMR2006,5,14) Thisargumentfrom

marketfailurebegsourquestion.

To my knowledge no one denies thatonceefficiencyincreasesaremade, an income or price effect ensues. Thus, if the argument is that in these areaspriceelasticitiesofdemandarelow,thereisstillindirectrebound,i.e. effective demand forotherconsumption.Malthusaddressedthis,notingthat the increase of wealth was perhaps less when the commodity to which machineryisappliedisnotofsuchanaturethatitsconsumptioncanextend withitscheapnessbuthesawthatthusaportionofthenowidlecirculating capital is set free for the purchase of fresh commodities creating new demandandnewemployment.(1820,282)Grubbslistofmarketfailuresthus has relevance for direct rebound, but not for the respending or macroeconomicreboundhereunderdebate.

Alternatively, in some (monopoly) areas the entire increased purchasing powercouldaccruetoproducersasprofits.Buteventhenlowerdemandby suppliers can cause a fall in energy prices, opening the market to marginal consumers. Furthermore, the higher profits would generate demand through investment spending, distributed dividends, or political clamour for privatisationoftheprofitablesector.(Malthus,ibid.Schumpeter1911Grubb 1992, 393) Finally, it is not clear why rebounds size should depend on whether the efficiency gain is BAU or policyinduced. That BAU ones are cheaperisconsideredinSection5.6,butatfirstglanceengineeringsavings depend only on physical changes, not their motivation. Using more efficient lightbulbsleavespurchasingpowerinanysetting. Theliterature,nottomentiontheevaluationdocumentsreviewinChapter2, often conflates welfare economics concepts like market failure with environmentalones.Brookeschallengetotheefficiencystrategyitselfstated

54

Ifefficiencypoliciesbackfire,however,theyareenvironmentallydamaging.Andtothe

extentthateconomicefficiency(e.g.lowertransactioncostsorfreecompetition)furthers economicgrowth,ittoomitigates,ceterisparibus,againstenvironmentalgoals.

55

thatReductionsinenergyintensitythatarenotdamagingtotheeconomyare associated with increases, not decreases, in energy demand at the macroeconomic level. (1990, 199) Schipper & Meyers argued that Narrowingthe difference between the average energy efficiency in the market and the level of efficiency that would be economically beneficial fromasocietalpointofview,isanimportantroleforpublicpolicy.(1992,305) Grubb similarly writes that market imperfections make efficiency
55 improvements economically attractive. (1992, 393) Jevons argued that

indeed all efficiencies raise affluence, but we are here investigating the

environmentalimpactoflevelsofconsumptioninfluencedbyrebound.

5.6Costsofefficiencyincreases
Engineeringsavingscomputationsincludetheenergynecessarytoproduce themoreefficientequipment(embodiedenergyandR&Dandjunkingcostsin termsofjoules):ifthemoreefficientcapitalcostsmoreorisinstalledearlier, the overallefficiencyincreaseissmaller.Computingreboundthenproceeds as usual. Yet some argue thatcapitalcostsformanimportantpartofthe total cost of providing energy services and the higher cost of energy efficientconversiondeviceswillreducethemagnitudeofthereboundinmany instancescapitalcostsinfluencetheefficiencyelasticityofenergydemand. (UKERC 2006, 1112) The argument seems to be that high capital costs hinder the purchase of more efficient equipment, resulting in lower average technical efficiency. (Besen & Johnson 1982,111)Capitalandmaintenance costsandopportunitycostsofspaceandtimetakentogethercanleadtothe decisionnottochangetoequipmentwithloweroperatingcosts.Butagainin thiscase,thereisnoengineeringsavingsandthusnorebound.Orisitagain beingassumedthatefficiencyincreases,ceterisparibus,lowerconsumption? Efficiencys cost seems relevant to engineering savings and overall energy consumption,butnotspecificallytorebound.

55

Theseideologicallycrucialquestionsconcerningpossibleconflictbetweenwelfareand

environmentalgoalsdeservestudy.(seePigou1920,134,183196)

56

Economists have long known that [t]he efficiency of the instruments produced must... be generated by greater cost... [in labour and material] (Rae,111116,100102Roodetal.2003)Intodaysdebate,however,there is disagreement on the relative capital costs of more, and less, efficient equipment.(Khazzoom1982)Whatseemsclear,though,isthatcapitalcosts relativetolowerfutureoperatinginputsavingscouldtheoreticallybesohigh thatnoincomeeffectremainsi.e.thereisanefficiencyloss.(Brookes1990) But assuming no market failure, all economically beneficial efficiency increases would be made. Additional policyinduced ones would lower lifetime efficiency in the mandated areas. The paradoxical question then again arises: If this also lowers affluence, are environmental and economic goalsinfundamentalconflict?

5.7Conclusion
Radetzki & Tiltons intensity of use (IU) methodology to predict metal demandisthesameasthosediscussedabovetodetermineenergydemand, andraisestwoofthesamedifficulties.Consumption(demand)ofthemetal dependsontheoutputinphysicalunitsoftheithfinalgood,theamountof themetalconsumeddirectlyandindirectlyintheithgood,andthenumber of final goods produced throughout the economy the amount of output is then related to national income [GDP], yielding the familiar result that changes in metal demand over time aretheresultofthreecausalfactors theoveralllevelofnationalincome,thematerialcompositionofproducts, and the product composition of income. (1990, 2526) Again, in taking GDPgrowthasanindependentvariableitneglectstheimportantquestionof what causes growth, and while knowing the share of the economy paid for metalsmighthelpcomputephysicalmetaldemand,itwouldnothelpcompute metalrebound.

Someoftheproblemsperceivedinthischapterareevidentinthetopdown policyevaluation methodology of 4CMR (2006). First, the two exogenous factorsseemtobedirectrebound(takenfromempiricalstudies)andenergy saving projected from energyengineering studies for the policies. (35,

57

52, 23) Second, two counterfactual scenarios for energy consumption are compared,onewithandonewithouttheefficiencypolicies,bothassumingan exogenous GDP growth factor (about 2.65%). (6, 36) Various types of rebound are then computed allowing forthedirectreboundeffect.(23,52) The scenario with policies includes the sectoral effects on energy use by assumption... (32) and predicts less energy use than the scenario without policies(56,passim).

Threequestionsthatariseare:First,evenifthefigurefordirectrebound (15%) were correct, what indirect rebounds were either left out or underestimated? Second, what if backfire, at least in many sectors, is true, and therefore some of the GDP growth must also be attributed to AEE or policyinduced increases? Third, the assumption of the computation of the difference between the two scenarios seems to be that rebound is (well) below100%theveryquestionthatbackfire(andrebound)studiesaretrying to answer. The study concludes that the policies are effective, savingabout 8% of the energy which otherwise would have been used by 2010. (6

Sections2.2&2.3)
Wedoconfrontaparadox.Theargumentstyingrebound,oratleastenergy consumption, to population, exogenous growthingeneral,energysshareof GDP, market failures, and the capital costs of efficiency gains are all searching for an explanation of why, in the face of more or less well documented energy efficiency increases, energy consumption goes up and up.Aparsimoniousandperhapsnaveansweristhattechnologicalandother efficiency increases cause greater physical consumption. While parsimony does not establish truth, it recommends Jevons theory for further investigation.

58

Chapter6TheMacroEmpiricalApproach
6.1Introduction
This chapter no longer starts with the concepts of engineering savings, rebound and backfire but with the scale of the world economy more precisely of energy consumption and changes thereof. Population times affluence in I=PAT, modified by technology either downwards (according to efficiency strategy theory) or upwards (according to Jevons) determines aggregateconsumptionoffossilfuel,thegrowthofwhichoverrecenthistory is assumed without documentation. We need therefore the determinants of populationandaffluence.Isthereacorrelationbetweenlevelofconsumption of natural resources and level of technological efficiency? Determining the latterrequiresametric,perhapsanaggregateone.Whiletheinputnumerator
56 (joules)isunproblematic ,monetary,service,andphysicaldenominatorsare 57 not. Eachisexploredinturn,theconclusionbeingthattheleastproblematic

arephysical:ifleftrelativelyunaggregated,onecanperhapsaveragespecific sectors,eachwiththeirownmetric.

While a consensus attests correlation between resource use and the

58 efficiency of resource use , causation is disputed. The hypothesis is stated

that efficiency increases do raise overall consumption, alongside the conventional factors of organisational efficiency, the efficiency of human capital, appropriation of space, hoursworked,andpopulationandeconomic
59 scale to their exogenous extent. To enhance the plausibility of this

56 57

Thisignoresdistinctionsbetweenprimaryandsecondaryenergy,andexergy. Sometimestechnologyissimplytimeontheindependentvariableaxis.(Lantz&Feng

2006,233)
58

Thisdissertationregardsthetermsresourceefficiencyandresourceproductivityas

virtuallysynonymous,althoughwhilethelattertermusuallymeasuresoutputintermsof moneyorutility,theformerhereinanycaseismeasuredoveragainstphysicaloutcomes only.

59

Itisspeculatedthatquantityandefficiencyofcapitalisreducibletothese.

59

hypothesis, secondary literature from the interface of economics and engineering is consulted. Highly aggregated empirical work, with whatever metric, has to date yielded only weak evidence. If the hypothesis is true, however, the consequence for environmental policy would be the abandonmentofefficiencypoliciesascounterproductive.

6.2ChangeinEfficiency
Measuring macroeconomic efficiency is challenging. That inputs of matter and energy per consumption unit (physical commodities, services, or Gross World Product) have indeed fallen since Adam Smiths day is claimed by many (Jevons 1865, 145146, 388 Cipolla 1962, 49, 99 Schurr 1982, 5 Clevelandetal.1984Kaufmann1992,522004,6364Inhaber&Saunders 1994,21Clapp1995,161171Giampietro&Mayumi1998,2024Vringer& Blok 2000) However, how can we compare efficiency over twentythree decades,ifnotwiththedenominatoraveragehumanlife?Thecapitalgoods andprocessessubjecttoefficiencyvariationhaveafterallchanged(Section

4.5),renderingthedenominatoramovingtarget.However,measuringinput
againstahumanlifeorthetotalofhumanutilitywouldnotbemeasuringthe

technological ratios that are the tools of environmental policy. Therefore we

mustsomehowmeasuregoods,artifacts,transformedmaterial.

Technologicalefficiencyisheredefinedasaratioadheringontologicallyto theproductionconsumptionprocess,preanalyzedintooutputsandinputsand done by manmade capital units. (Rudin 2000, 539 Rood et al. 2003, 492) Theliteraturesimilarlyusestermsliketransformation,transfer,throughput, andthetermproductionitself.ForEnos,process[is]aunitofenquiryforthe studyofimprovements[in]technologyplusitsapplication,i.e.thephysical and chemical relationships governing the transmutation of elements. (1994, 2 Barker et al. 2005, 5, 17) The process of converting energy to lumensviathecapitalgoodlightbulbisanexampleofatargetofefficiency policies. Van den Berghs statement that [t]he production process may be envisioned as a transformation by agents of material inputs into goods and

60

wasteoutputs(1999,555)remindsusnotonlythatwecanefficientlyproduce

waste,butthatthemetricfornegativeimpactisnecessarilyphysical.
Capital, labour, and (sometimes) energy/material (KLEM) are common determinants of production quantity. The last chapter tried to show that populationandthephysicalscaleoftheeconomyarepartlyendogenous,i.e. explicable by efficiency changes either upward or downward. The hoary conceptofcapitalishereusedbroadlyintheclassicalsenseofcirculating aswellasfixedcapital(Smith1776,II.i.4&5Malthus1820,192Mill1848, 751) or Raes instruments (1834, 8689, 170171). This concurs with Bouldingsinventoryofartifacts(19491966,910Daly1974,15)orSolows capitalinnaturalunits(1970,35).FollowingSchumpeter(1911,2021,37,
60 210211) and many classical economists capital is treated endogenously,

i.e.reducibletolandandlabour.

Capitalalsoaugmentsmaterialandlabour.(SmithIV.ix.34Mill60,63,92, 100115 Saunders 1992) In disentangling labour from capitalproductivity used the homely example of an improvement in the design of a typewriter thatgivesonesecretarythestrengthof1.04secretariesafterayearhasgone bythisyields,usingthestrengthmetaphor,theconceptofefficiencyunits to measure the augmentation of both labour and capital. (1970, 3435) Capitalsreasonforbeingistoimproveefficiencyimportantfortherebound question because while backfire is thus not guaranteed (Section 4.3), it is
61 enabled.

Although controlled experiment is impossible, the fortuitous case of two societiesdifferingonlyinleveloftechnologicalefficiencyisnotruledout.Yet statistics seems the most likely method of testing the hypothesis that increased efficiency, ceteris paribus, causes an increase in consumption of

60 61

E.g.Smith(1776,II.iii.25,3334)Say(1803,293)Mill(1848,100,182). Whileremovingcapitalfromtheproductionfunctionisperhapsnotpertinent,itishopedthat

bringingmatter/energybackintoithelpsfocusattentiononthenaturalresourcesatthecentre ofenvironmentalquestions.

61

the more efficiently used input. The problem of metrics is therefore unavoidable. Solows pioneering work in establishing technical progress ratherthanquantitiesoflabourandcapital(andtheirratios)asthestrongest causesofgrowthhadnorigorousmetricbutincludedbothtechnologicaland
62 organisational changes. (Rosenberg 1982, 24) This large residual was

arrived at more by a process of elimination than by explicit testing. The classicaleconomistslikewiseattestedthecontributionofprocessefficiencyto the amount of a nations wealth, more on the basis of theory. Lauderdales examples of the spade, plough and stockingmaking machine (1804, 162 168), further analysed by Schumpeter (1911, 280284), or Says of the flourmill (1820, 133134, 140), show productivity increases to increase aggregate production and sustenance as well as employment. Jevons showed empirical correlations between coal and iron consumption and increasing efficiency of steam engines and smelters. (1865, 113, 119, 155, 262265,387388)

But even if efficiency almost surely contributes to economic growth and

63 affluence, does it contribute to input growth? Solows conceptualisation of

changeinphysicalcapitalisamottoforthischapter: Obviouslymuch,perhapsnearlyall,innovationmustbeembodiedinnew plantandequipmenttoberealizedatall.Onecouldimaginethisprocess taking place without net capital formation as oldfashioned capital goods arereplacedbythelatestmodels,sothatthecapitallaborrationeednot change systematically. But this raises problems of definition and measurement even more formidable than the ones already blithely ignored.(1957,316317)

62

Domarshistoricalplayaboutgrowthmodelsfeaturedawholegroupofactorsalongside

capitalandlabor,includingtechnicalprogress,whichhecallsMeasureofOurIgnorance andhisownhero,theResidual.(1961,709)
63

SaundersquotesSolowthatitshardtobreakthehabit,...factoraugmentingdoesnot

meanfactorsaving.(1992,131)Jevonswarnedofconfusionbetweencoalsavedandcoal spared.(1865,13)

62

6.3TheMonetaryDenominator
In measuring ecoefficiency Reijnders names five denominators: a product (such as the automobile), a service (e.g., transport over a certain distanceataspecifiedspeed),anareaofneed(e.g.,clothing),asectorofthe economy (e.g., energy supply and demand), or the economy as a whole. (1998, 14) The metric for the last is monetary, relying on the System of National Accounts concept of income or GDP. Justifications of this denominatorareitsconvenienceandtheavailabilityofdata(Radetzki&Tilton 1990, 28) plus the idea that it is a good proxy for output in some sense meaningful in studying physical environmental impact. (Moezzi 2000, 528) Whataretheproblemsofthisdenominator?

TypicalstatementsusejoulesperunitofGNPoreconomicactivityorper relevantmeasureofactivityoroutput(Schipper&Grubb2000,369),e.g.: Over the last 30 years the economys energy intensity (total energy consumption divided by total GDP [as] Mtoe/$USbn) has improved by around1.8%eachyear.(DTI/Defra2003,3.3) GDPandtheubiquitousbutamorphouseconomicactivityaresynonymous. Greening et al. note a widespread vagueness aboutwhat enduse activity withinGNPisbeingmeasured(2000,392),bothconceptsbeingfullyabstract. Many doubt GDPs viability. For Smil GDP, if used simply as [an] absolute measureforcategoricalcomparisonsofeconomicandenergeticefficiencies... [is]grosslymisleading.(2003,66,81Manne&Richels1992,40)Specifically forenvironmentalresearchJnickeetal.holdthat[r]esourceconsumptionby thenationaleconomyandtheprocessofenvironmentallysignificantstructural changearenotappropriatelydescribedbytheproductionvaluesusedinthe nationalaccounts.(1989,25)OftenGDPisfirstcogentlycriticised,e.g.as notreallyanaggregatemeasureofallactivitiesthatuseenergy,butrather andattempttomeasuretheexchangevalueofeconomicactivitiesbutthen relieduponforanalysis.(Schipper&Meyers1992,54)

A set of problems in using GDP outside the realm of fiscal bookkeeping includes its not measuring resource depletion, unpaid labour or bartered products(Daly&Cobb1990,401455ElSerafy1991Smil2003,7374),all 63

of which are relevant to the study of environmentally problematic consumption. Moreover, monetary indexes yield lower ratios of energy intensitywhenevervalueaddedrisesforreasonshavingnothingtodowith materialefficiencyorenergysshareinintermediateproducts.Lowermaterial intensity results similarly when physical inputs of goods like air, water, and space, lacking a price, are not deducted as intermediate expenditures in effectloweringthenumerator.GDPthusseemstomeasuresomethingother than the aggregate output consumption which concerns environmental policies. That is, GWP measures prices. But, as shown by the hoary water and diamondsparadox(Smith1776,I.iv.13),thesebearnonecessaryconnection tophysicalattributescontributingtoenvironmentalimpact.Appliedtometals Cleveland & Ruth likewise notethatmonetaryunitsfortheintensityofuse ratioincludeforcesunrelatedtoquality,suchascartels,regulation,andlabor and capital costs, [that] can affect the price or value of metals. (1998, 35) Radetzki&TiltonpointoutthatThepriceofmetalmayriseovertimeeven thoughqualityremainsunchanged(1990,21),whileRobinsonnotesthatthe measuringrodofmoneymisleadsbythemerefactofsalesandadvertising costs(1954,18).

A further case of GDPs autonomous rise is that the fuel/GDP ratio can decrease simply because fuel prices rise. Yet this is not evidence of fuel conservation. (Saunders 2000a, 442) And even if there is backfire, i.e. the elasticity of fuel consumption over against efficiency is greater than unity, if GDPforanyreasongrowsevenfaster,alowerfuel/GDPratiomasksthefact
64 that real consumption has risen. Finally, Kaufmann showed that the

efficiency ratio with a monetary denominator cannot tell the difference betweenaratiodecreaseofenergytounitofGDPcausedby1)lowerinput for a given quality of service and 2) the same input for a lower quality of service. (1992, 54 Cleveland et al. 1984, 893 Wackernagel & Rees 1997, 17).
64

Thiscouldbethenormalcase.(Kolstad2002)

64

Moreover, insofar as efficiency increase lowers the price of energy, and energy prices comprise part of the GDP denominator, they raise energy intensity by lowering the denominator. Yet because usually growth in the monetarydenominatorforwhateverreasonregistersasanincreaseinall productivefactors(notonlyenergy),ittellsuslittleabouttherelativestrength of the factors. Schurrs counterintuitive results were that total factor productivity rises more rapidly than the increase in energy consumption associated with production in absolute terms the paradox was that favorable energy supply conditions led to our getting more from energy. (1982,710)Couldthisbeclarifiedbyadoptingaphysicalinsteadofmonetary metric?

Finally,Smilsdeconstructionofthemonetaryconceptofenergyintensity (2003,7178)emphasisesthedifficultyofcomputingpurchasingpowerparity ininternationalcomparisons,andofaggregatinghissixvariablesdetermining differences in nations energy intensity into an overall index, an index of limited interest to the technological ratios sought here, as shown by his example of the USA and Canada, which have high energy/GDP ratios but also the most technically efficient machinery and equipment sector (77). Thus, it seems gratuitous to assume that GDPs advantages and disadvantages somehow balance out (Reijnders 1998, 14). Bartelmus may be right that physical metrics evidence problems of arbitrary weighting and indicatorselection(1997,334),butthemonetarymetricfaresnobetter.

6.4TheServiceDenominator
Do services used pervasively in the rebound debate serve the denominatorbetter?Reijnderssecondcategory(Section6.3)wasaservice (e.g., transport over a certain distance at a specified speed). Similar definitionsarethermalcomfort,lightingandcooking(Foxon2000,590)and theoutputfromahomeproductionfunctionrequiring(market)commodities and time [such as] mobility (passenger kilometers), lighting (lumen hours), washing (kg clothes), etc. (Wirl 1997, 14). That is, the chosen unit is

65

passenger kilometre instead of the physical tonne kilometre. (UKERC 2006,4Binswanger2001,120121)Thesearetwodifferentperspectiveson oneandthesamephenomenon.

Service is not without ideologicalfreight.Advocatesoflowreboundinsist on the concept, arguing for instance for the significance of the fact that physical energy costs are only a small part of the costs of energy services (Howarth 1997, 3 Section 5.4) Foxon claims that increasing resource efficiency followsnaturally from the perspectiveofconsideringresourceand material flowsneededtosatisfyenduseservicedemands(588,myitalics). Saunders on the other hand claims that his backfire model does not suffer from the distinction (2000b). The material input per service unit (MIPS) concept is also used to advocate the vision that we cankeepordoubleour

utility (affluence) with less resource consumption, i.e. an efficiency strategy

with both welfare and environmental goals. (SchmidtBleeck 1994 Hinterberger et al. 1997 von Weizscker et al. 1997) The service concept forms moreover a bridgetotheEKCdiscussion(Section3.5)andconsumer efficiency:Perhapsifwecarpool,wemoveclosertoabsolutesatiationwithout
65 consumingmorebutwealsofreepurchasingpowertoconsumemore.

The approach here, however, asks more dryly whether the concept of serviceisneededtoanswerourquestion.Whatfactsaboutefficiencychange andensuingconsumptionchangemightrenderituseful?Howeverfruitfulan
66 analysis oftheambiguousphrasegoodsandserviceswouldbe ,themain

problem with services is that technological efficiency would by definition increase when two persons, rather than one, ride in a car. Passenger kilometer times two yields doubled efficiency without, however, any technological or engineering changewhatsoever.Theratioofpaperandink inputstotheserviceoutputofajournalriseswitheachadditionalreader.The

65

Theconsumerefficiencystrategyisdistinctfromboththeefficiencystrategyandthe

sufficiencystrategy(wherebyinordertolowerconsumptionwedowithoutsomeutiltiy).All threereboundduetopriceeffects.
66

Robinsonwroteironicallyofthemassofsolidgoodsandusefulservices.(1954,19)

66

analogywithGDPisclear.Awholeeconomysefficiencycanvarydepending not on technology but solely on behavioural change which, to be sure, likewise does not guarantee lower energy consumption. The difference is captured by Grubbs conceptualisation of the anomaly of efficient productionandinefficientuse(1990,198Cogoy1999,388,397).

Ifefficiencycanbefullydescribedinphysicaltermsalightbulbofagiven strengthandinputwhatisgainedbycallingthisalightingservice?Thenew analyticalparadigmholdsthat[c]onsumersareassumedtoderiveutility(U)

67 fromconsumingthese[energy]services ,ratherthanfromconsumingenergy

commoditiesandothermarketgoodsdirectly.(UKERC2006,3Foxon2000, 589)Butwhileutilityisperhapsderivedfromservice,serviceisnecessarily
68 derived from energy. Shove likewise insists that people do not really

consume energy. Instead, they consume the services heating, lighting, showering, cooking, television watching, computer interaction that infrastructuresofgas,electricityandwatermakepossible.(2004,1054)But thisundefendedredefinitionofconsumption(andinfrastructure)resultsinthe incorrectassertionthatenergyisnotconsumed!Whatnewtermisproposed, then, to describe the human activity that results in ever more depletion and pollution?

Perhaps the claim is that defining efficiency change in terms of services lowers the efficiency elasticity ofpriceordemand.However,evenassuming thatwederivethesameservicefromcarpoolinganddrivingalone,offeringan energy saving, this again represents a violation of the ceteris paribus condition of our question: In looking for change in consumer demand after efficiency gain and price fall, this would not hold the factor consumption
69 efficiencyconstant. Thatis,adoptingservicesasthebasicconcept(UKERC

2006) opens up the danger of analysing two changes at once, and it is a

67 68 69

Consumingaservicejibesbadlywithordinarylanguage. DoesthisparadigmfailthetestofOccamsrazor? Ifbehaviouralchangeisafunctionofefficiencychange,itisprobablethatmoreefficiently

providedproductsareusedlessconsumerefficiently.

67

conflation of physical and service denominators to write that technical efficiency measures the conversion in an economic system of basic
resource inputs notably available energy into agivenfixedsetofgoods andservicesforconsumers.(Ayres&Miller1980,359)

A further problem is that, especially if we correctly include transportation energyandembodiedenergyintheanalysis,allservicesareenergyservices just as they are all labour services. That is, something is said to be an energy service as opposed to what other kind of service? Is the difficult concept of the energy content of an expenditure unit needed? Finally, we ultimately need a ratio whose denominator directly affects environmental quality.Again,andmetaphorically,theenvironmentdoesnotcarehowmany people get how much utility out of consuming it. Only physical amounts matter.(Jacobs1991,25)Theconcepthoweverseemstoshifttheontological locusofaservicefromtheenvironmentallyrelevantgoodsandprocessesto
70 the relation between humans and a serviceyielding stock of artifacts.

Parsimony often seems lacking: Berkhout et al. suggest that the consumer can be regarded as a producer of useful services [sic.] using energy as an input(2000,427,emphasisadded),whileDimitropoulos&Sorrelldevleopa complex production function with demand for energy (E), demand for energyservices(ES),theutilityderivedfromservices(U)andafeature of services called useful work (S) as well as broader attributes (A) of services(UKERC2006,34).

Theconceptofservicesdominatestherebounddiscussion,yetIknowofno demonstration that the efficiencyinduced demand for capital goods and the energy to use them is in some way crucially different from the efficiency induceddemandfortheservicesthatthesegoodsandenergyprovide.Inany case, it is difficult to approach our questionphysicallyandatthesametime accommodate the concept. Ayres, whose taxonomy ofproductionprocesses

70

Onedefinitionevenredefinesthetermcommodity:Energyservicesrefertothe commodity,whichisactuallyusedordemanded,i.e.refrigeration,hotwater,andprocess heat.(Greeningetal.2000,389)

68

and resource/commodity definitions are throughout physical, fails in his attempt to distinguish production of goods from production of services. (1978, 5354) Extraction, conversion, and manufacturing are indeed transformations producing goods requiring consumption of material goods andenergybutsincethisisexactlyhisdefinitionofproductionofservices, thereisnodifference.(54)Healsoattemptstomapthesetransformations, firstfromfactorsofproductiontogoodsandservicesandthencetoutilityor

welfare, and then from exhaustible or renewable natural resources to finished materials and forms of energy, and then to material products and structures, then to (abstract) services, and finally to utility. (67) Yet no
differenceemergesbetweenservices,structuresandutility.Heabandonsthis attempttoaccomodatewelfareeconomics,decidingthatProductionofpure services such as transportation is omitted from further consideration for conceptualreasons.(1978,66)

Returningtothenormativeroleoftheconcept,Ayreswrites: [A]technologicalescapehatchtothelimits[andgeneralimpact]bindstill remains open, however. It is that final services (as consumed) may not haveanirreducibleminimummaterialorenergycontent.Inotherwords,it maybetechnologicallypossibletoincreasetheservicecontentofagiven unitofmaterialsorenergyeffectivelywithoutlimit.Sinceservicecontentis nonmaterial, the second law constraints against unlimited creation of negentropydoesnotapplyinthiscase.(1978,50) He then confesses his lack of a conclusive argument against this escape hatch,andindeed,thestudyoftheuseoftheserviceconceptwouldrequire aninterdisciplinaritybeyondthescopeofthisdissertation.

6.5PhysicalDenominators
Recall that the first of Reijnders possibilities for a metric of output is a product (such as the automobile). (1998, 14) Assuming some taxonomy for products, sectoral efficiency changes seem in fact amenable to measurement of engineering savings. (Ausubel 1989 Dahmus & Gutowski 2005) For instance, Ayres materialflow framework covers elements, compounds, energy, transformations and final goods in terms of stocks and flows(1978,5366)andcancalculateefficienciesforallprocesses(53Ayres

69

&vandenBergh2005,102103).Certainly,integratingphysicalprinciplesor physicalchemicalclassificationintothemicrostructureoftheeconomyor commoditybasedclassificationsystemencountersproblemsofaggregation and both taxonomies, for instance, freely aggregate toxic and nontoxic materials. (57) These problems are, however, less severe if only material efficiencyandnottoxicityisconsidered,andfurtherifonesettlesforlessthan totalaggregation.

Worldphysicalenergydepletionandpollution,again,isourrelativelyeasily measured dependent variable to be related to changes in world technical efficiency, average across sectors. This section briefly surveys some candidates for a physical metric, relying on literature by the cited authors Ausubel,Ayres,Cleveland,Enos,Jevons,Landes,Rosenberg,andSieferle. InRosenbergsopinion: Notonlyisitdifficulttosortoutthecontributionoftechnicalprogressto economic growth from other related contributions capital formation, education,resourceallocationbuttherearenounambiguousmeasures of output over time periods long enough to permit large changes in both pricesandtherelativeimportanceofeachcomponentofoutput(theindex numberproblem).(1982,2355) If physical energy consumption rather than economic growth is the

explicandum, and energy efficiency rather than the broader technical

progress the hypothesised explicans, can we find an appropriate physical metric?

Such a metric could be vehicle miles, tonne kilometres, lumen hours, or kWh/yearforagivenrefrigeratorofagivensizeandfeatures).(Schipper& Grubb2000,369).Whileweight,volume,energycontent,andchemical(table of elements) content are units applicable to all process outcomes (fully aggregate),vehiclesandrefrigeratorsarenot.Still,theycompriseReijnders firstcategoryofproducts,asopposedtohisothergroups:services,sectors, areas of need and the economy as a whole. (1998, 14) These products, as objects from which more aggregate units are abstracted, should not be ignored.Asobjects,theseproductsincludeproducerandconsumergoods (i.e. fixed and circulating capital including food), but also products like a

70

lighted space or heated space excluded from capital here are natural resources, which are inputs only, and human beings the L and M in productionfunctions.

An example of a capital object is a poorlyinsulated house containing an open woodburning fireplace. It can be made with more, or less, input or

embodied energy. The second sort of product is neither an object nor a

service, but rather the heated volume of the house, definable in terms of
3 temperature and m , made with more or lessoperating orutilisation energy.

Changesinboththehouse(insulation)andthefiringplace(enclosurewitha ceramicstove)reducetheinputrequirementsfortheresultingheatedspace. These processes and ratios are subject to research independent of the motivation for them, i.e. the heated house constitutes an environmentally relevantstateofaffairswhetherornotanyoneisinit.

The example of rotating axles and rods shows moreover that capital efficiency improvements can change several ratios and several qualities (Section4.5): 1) As hardwood was replaced by metal rings which were replaced by ball bearings and then roller bearings each different capital object broughtmoreoperatingefficiency,i.e.loweredtheratioofenergyinput touseful(thedesired)work. 2) Each type of capital object functioned longer, i.e. loweredtheratioof
71 embodiedenergytooutputspreadoverthegoodslifetime.

3) If the new machine part contains more embodied energy, this lowers theworkefficiencyratio. 4) Aqualitativechangeintheoutputitselfcantakeplace,inthiscasethe qualityoftherotationse.g.fasterorsmoother.Thispointagainraises the problem of efficiency measurement that the characteristics of outputoftenchange.Thismovingtargetproblem(Section6.2)besets

alldenominators,monetaryandserviceonesaswell.
71

Durability,recycling,andembodiedenergyusuallywithinthequestionablemetaphorof

lifecycleanalysisareheresubsumedunderefficiency.(Ayres&vandenBergh2005,101)

71

Ayres&vandenBerghtentativelysuggestamatter/mattermeasureofthe technical efficiency of the production process in terms of mass instead of energy the ratio of mass inputsembodiedinthephysicaloutput(finished products) to the gross mass of material extracted from the environment. (2005,103)Butifwekeepjoulesastheinputmetric,wecanstilltakemass embodiedinoutputasdenominator.Suchphysicalmetricsforanaggregate
72 efficiency index are then in units of weight , differentiated by chemical

composition. Metrics of mass or weight however, have the weakness of not distinguishing between different chemical compositions of inputs. Taking a

badinthedenominator,thejouleratiosofnaturalgastoCO2 massdifferfrom
thoseofpetroleum. Intheirdiscussionofthemethodologyforstudyingconsumptioninthemetal sector, Radetzki & Tilton (1990, 1921) consider solving the problem of incommensurability of outputs by weighting the weights of each of twelve steelproduct types in the denominator by an abstract coefficient. They also attest important product differences that are often not physical, as when copperofmultiplexingqualityenabledthesameamountofwiretocarrymore conversations.Butforreasonsofsuchcomplexitiesanddatalimitations,and because monetary metrics raise even more difficulties, they advocate a simpletonnagemetric.However,amorecompletephysicalmetricthanones using mass would have to include temperature, light, and perhaps even space.

Cleveland & Ruth also see the serious flaw [of] aggregation by weight or volume as ignoring material quality, the marginal amount of economic

output generated per mass unit of material input and therefore similarly
weight mass units with its relative economic usefulness. (1998, 35) The conceptofeconomicusefulnessisclearlynolongerapurelyphysicalmetric, especially if it is relative. Birol & Keppler propose the similar idea of efficiencyunitsoffuel:oneandthesamelitreofpetrolwouldcountasmore efficiency units as it becomes more productive of useful work. (2000, 461)
72

Poundandpesoarebothmonetaryandphysicalunits.

72

However, since usefulness and productivity are themselves ratios, these systems place beneath the joule numerator a joule/work, joule/tonne or joule/unitofGDP denominator. Forinstance,efficiencywouldbetheratioof joulestotheratioofjoules/dollarbutthentodefinetheseaggregatedratios nowinthedenominatorentailstheoriginalproblemsdiscussedabove. Canonesolveaggregationproblemsbyusingjoulesinthedenominatoras

well as numerator? What then distinguishes the inputjoule from the output
one? For instance, Ayres suggests a ratio of exergy to exergy available workofinputstoavailableworkoffinaloutputs(1978,52)thatoffersno adequate distinguishing criterion relying on the Second Law of thermodynamics concept of entropy, this metric includes the quality (availability)ofthejoulesratherthantheirmerequantity.Healsoproposesa firstlawofenergyefficiency(),theratioofenergytransfer(ofadesired kind)tototalenergyinput(44).Underthislaw,outputjoulesaremoredesired oreconomicallymorevaluable.Inthermodynamicterms,thistransfercould also be described as theconversion of energy withpotentialtoformswhich are actually useful. However, since the energy input must also have potentialtodowork,i.e.lowentropy,itwouldbemoreaccurateheretospeak
73 ofexergyinputs.

Enos likewise seeks a metric going beyond comparison of products in physical terms: for instance, when petroleum cracking increases gasolines efficiency at the expense of fuel oils, we dont know which is better [more useful]. (1958, 180, 183) Other attempts to overcome the disadvantages of joules,kilograms,andspace(inecologicalfootprintmetric)aremadebyvan den Bergh, who proposes weighting physical quantities byeconomic value (1999, 551, 559) and Dahlstrm & Ekins, whose value chain analysis technique attaches monetary flows to the traditional material flow analysis (MFA: Ayres 1978, 1998 Duchin & Lange 1994) categories of added

73

Canalloutput(ourdenominator)bereducedtowork?Becausephysicalobjectsaresurely

alsooutput,onemustatleastintegratetheconceptofembeddedenergy[orexergy].

73

chemicalelements,weight,waste,shape,andrecycledtonnage.(2006,509, 515,518,517alsoEnos1994)

Two issues arise from these attempts. First, the changes we seek to measuredonotrelyontheconceptsofdesiredorbetter.Second,having different types of joules in numerator and denominator can reveal important cases where input is higher than output, as when a number of fossil fuel joulesyieldamuchlowernumberofmaizeones.(MartinezAlier1987,Ch.2) Such processes with intensities greater than unity are obviously sometimes economically desirable. EROI is another such a ratio, measuring joule investment over against joule return, the lowering of which, by making
74 energycheaper,isperhapsthecauseofaspecial,economywiderebound. 75 Iftheyarethesame,ajouleislikeaseedwhichproduces100seeds.

JoanRobinson(1954)soughtanonmonetarymetricfortechnicalprogress, choosing the capital/labour ratio with capital physically measured as the value of a stock of goods in terms of commodities or equipment, workin progress[and]materialsandlabourmeasuredintermsoftime.(122,65)She concluded, however, that indexnumber ambiguities are insoluble (6465, 115): Economics isthescientificstudyofwealth[ourdenominator],andyet wecannotmeasurewealth.(24Victor1991,204206)

However,isaggregationnecessary?Canwenotaveragephysicalratiosin differentunitsoverallsectorstoarriveatamacroeconomicindex?Arecent computation of domestic material consumption in the EU15 for instance, rejects metrics of weight, weight adjusted by life cycle analysis (LCA), and exergy,choosinginsteadalessaggregatedMFAusingfossilfuels,industrial minerals and ores, construction minerals and biomass. (Wiesz et al. 2006,
74 75

RisingEROIhascausedfallingrealoilprices.(Smil2003,94,76) Ayres&vandenBerghinsistoncountinghighentropyprocesswaste,thedifference

betweenworkdonebytheeconomicsystem[and]theexergyofallinputs.(2005,103) Exergyisenergyavailableforusefulworkoreconomicactivityandunlikeenergycanbe destroyed.(Ayres1978,52)Thisisconsistentwitholderanalysesoftheconservationand transformationofmattertocreateutility.(Say1803,62,387388Rae1834,29,82,99100)

74

681)WhiletheauthorsconcludethatMFAisbesetbytheusualproblemsof incommensurability,changeovertimeandfindingacommondenominatorfor crosscountrycomparisons(693),lessaggregationsuggestsasolution.

Inspiteoftheproblems,iflimitedtoparticularsectors,manywritersdeem change over time to be physically measurable. Rosenberg traces the following: 1) efficiencyincokeuseforsteel,whichacenturyafterJevonswasstill risingfrom1,900poundspertonofpigironin1949to1,200poundsin 1968,and 2) efficiencyincreaseinlampsrisingbetween1882and1945enoughto enablean80foldreductioninlightingprice.(1982,61,65Saunders 2000a, 445) According to Smil, In 1900 efficiencies of thermal electricity generation were as low as 5%. Todays best thermal plants have conversion efficiencies of just over 40% but cogenerationcanraisethisratetoalmost60%.(2003,2223Schurr 1982,5)TwostudiesoutsidetheOECDusingphysicalmeasurements of average efficiency for certain types of capital are Roy (2000) and Glomsrod&Taoyuan(2005).

Jevons empirical measurements of efficiency increases were made explicitlyinsupportofhisbackfiretheory.(Alcott2005,1718)Further,some recentauthorsfindempiricalevidenceforreboundgreaterthanunity,based on correlations of rising inputenergy consumption and rising efficiency. Ausubelshowsefficiencyincreasesoverroughly200yearsfor: prime movers [steam engines], lamps, and ammonia production measuredasmachinesorprocessesforenergytransformation[fromcoal to work or electricity to lumens] according to the second law of thermodynamics. [I]t is clear that many engineering systems tend to follow steady trajectories over long periods of time toward higher efficiency.(1989,8384) Greenhalgh shows engineering efficiency gains of over 20% for household appliances in Denmark between 1977 and 1986, alongside rising electricity consumption. (1990, 297) Rudin does the same for U.S. energy use in commercial buildings (8% more efficient from 1979 to 1995) and cars (30%

75

from 1967 to 1997). (2000, 542543) Smil likewise covers efficiency change and energy consumption. (2003, 614, 34, 41) Dahmus & Gutowski have undertaken the most systematic attempt to date to regress overall physical consumptionuponefficiencygainsinsevenindustrialsectors:globallyandin joules/kilogramsforpigiron,aluminiumandnitrogenfertilisersandintheUS e.g. for refrigerators and airplane travel (albeit in passenger kilometres). (2005) Adding a third parameter to the efficiencyconsumption equation, namely

prices,revealsthedynamicrelationshipthathighermaterialinputpricesraise
efficiency,butefficiencylowersmaterialinputprices.(Smil2003,8288,149 161) Fouquet & Pearson (2006) show for the UK efficiency increases in lumensperwatt(150)andstrongcorrelationsbetweensuchefficiencygains, lighting prices and lumen consumption (158166). Their datashowsthatthe priceofalumenhourin2000wasonethreethousandthofthatin1800,while the number of consumed lumenhours rose twentyfive thousand times, yielding a theoretical price elasticity of demand. (173174) One cannot, however,thencorrectthisbyafactorjouleintensityofalumenhourbecause the prices are also in terms of lumenhours, not joules. However, especially since prices should not be treated fully exogenously, to interpret such correlations requires causal theory: even if the historical elasticities are preciselyrecordedbetweenefficiencies,prices,andbothinputsandoutputs, onecanalwayssaythatotherwiseinputconsumptionwouldhavebeeneven

higher!(Schipper&Grubb2000,370)

6.6 Conclusion
1) A correlation, much less causation, between energy consumption and energyefficiencycannotbeshownuntilthelatterisclearlydefined,the preconditionforwhichisaclearandappropriatedenominator. 2) Monetarydenominatorsareconvenientbutinappropriate. 3) Methodologyshouldstartwithdemandforenergyithasnotbeenshown that demand for energy services is necessary to study consumer

76

reaction to technological efficiency change therefore service denominatorsarenotappropriate. 4) Relatively unaggregated physical denominators have yet to be developed. 5) Nometricfortechnologicalefficiencyissufficientlyclearandappropriate formacroeconomicempiricalwork. 6) Studiesconflatingthethreemetricsareunfruitful. 7) In the literature, output is used ambiguously to mean both specific products,processes,orsectors,andthewholeeconomy. Ten methodological problems of studying both energy consumption and rebound,conceptualisedinthelastthreechapters,areevidencedbySchipper &Meyers(1992)paradigmaticstudy: Populationisuninfluencedbytechnology. Growthisonlynegativelyinfluencedbytechnology. TheGDPdenominatorisadoptedinspiteofitsweaknesses. Technologicalandconsumer(operational)efficiencyareconflated. Entiresectorsandcountriesareexcludedduetolackofdata. Indirect(includinginternational)reboundislargelyignored. Costeffectivenessandeffectivenessareconflated. Ratios of energy to service, service to sector, and sector to economyareuseduncritically. While it is assumed that an aggregate metric is needed, none is achieved. Thedenominatorchangesfromsectortosectorasamiddleway.

Theevidenceforthischaptersconclusionthatenergyefficiencyhasindeed risen worldwide over decades relies only on a small sample of secondary engineering and economics literature. A physical efficiency metric would, however,aidintestingvariableswithstandardstatisticalmethodscomparing countries (corrected for trade of embodied energy), comparing time periods

77

within countries, and following changes in rates of change in efficiency and

76 energyconsumption.

This chapter attempted to look at the forest in addition to the trees of the microeconomic tracing of rebound through all sectors and consumers. Yet even insofar as the variable technological efficiency is measurable at the worldscale,thiscorrelationestablishesnocausality:becausetechnologyonly

enables more output consumption whatever the consequences for input

consumptionbothJevonstheoryandefficiencystrategytheorymustmake explicit assumptions about consumer preferences for more consumption as opposedtoreapingleisureandpermanentinputcutsfromtechnologygains. That is, disagreement reigns over the counterfactual. While Jevons theory indeed predicts the correlation, the efficiency strategy assumes that without efficiencyincreases(otherwise)consumptionwouldhavebeenevenhigher, i.e. topredictthecorrelation,itmustexplainrisingconsumptionwithfurther, Prometheanfactors.Atthepresent,giventhestateofexistingtheory,further empiricalresearchatthelessthanworldscaleseemsunfruitful,yetamodel includingallsectorsandallconsumersdoesnotyetexist.

76

Perhapstheliteratureonthepriceelasticityoftechnologicalefficiencywhenprices

exogenouslyrisecanilluminateelasticitieswhentheyfall.(Smil2003,8288,149161)

78

Chapter7TheClassicalEconomists
7.1 Introduction
th The simultaneous growth of technical progress and wealth in 19 century

Europe was by all accounts visible and was attested by all classical economists.DavidRicardoforexamplewrote: Bytheinventionofmachinery,byimprovementsinskill,byabetterdivision oflabour,amillionofmenmayproducedouble,ortrebletheamountof riches,ofnecessaries,conveniences,andamusements,inonestateof society,thattheycouldproduceinanother.(1817,273) More specifically, William Stanley Jevons presented empirical data showing that steam engine and smelting efficiency rose alongside coal and iron consumption.(1865,154,261271,387388)JeanBaptisteSaywentfurther, claimingcausalitybetweenproductivityandproduction: Butwhenceisderivedthislargersupplyofwealth,thatnobodypaysfor? Fromtheincreasedcommandacquiredbyhumanintelligenceoverthe productivepowersandagentspresentedgratuitouslybynature.A powerbeforeknownandavailableisdirectedwithsuperiorskilland effect,asinthecaseofeveryimprovementinmechanism,wherebyhuman oranimalpowerisassistedorexpanded.(1803,101Section4.2) ThischapterverybrieflysurveysthemajorwritingsoftheseauthorsandJohn Stuart Mill, John Rae, Thomas Robert Malthus and Adam Smith, in order to discovertheirargumentsandconclusionsconcerningseveralthemesrelevant totherebounddebate: whatefficiency,productivity,orimprovementis theeconomicprocessoflowerproductioncostsandfallingprices thelargesocietalincomeeffect,i.e.increaseintotaloutput the contribution of organisational efficiency and the consumers propensities thepossibilityofcausalityandevenbackfire theevidencethatlaboursavingimprovementssavednolabour thepossibilityofsatiation,leisure,andnonconsumptionand thatgrowthwithouttechnicalprogressisimpossible.

79

7.2FromEfficiencytotheConsumer
Thetermartsdenotedscience,technologyandinvention.Itwasclearthat knowledgeofnatureledtoimprovementoflandandmachinecapital.In Raesviewinventionandimprovement causethesamereturnstobeproducedfromalessexpenditure,or greaterreturns,fromthesameexpenditure.Itistheintentionofthe inventivefaculty,whenitappliesitselftothearts[to]renderthelaborof themembersofthesocietyinwhichitoperatesmoreeffective,andenable themfromthesameoutlaytoproducegreaterreturns,orfromlessoutlay toproducethesamereturns.(1834,131,258259) Theconceptofanimprovedinputoutputorexpenditurereturnratiois similarlyclearinMillsstatementthatanyprogressinthosearts,any improvedapplicationoftheobjectsorpowersofnaturetoindustrialuses, enablesthesamequantityandintensityoflabourtoraiseagreaterproduce. (1848,130)Smithhadalreadydistinguishedproductivityfromincreasein sheerquantitiesofinputs:Toaugmenttherealrevenue,theannual produceofthelandandlabourofanysociety,oneneededeithermore labour,orgreaterimprovementinproductivepowers,dependingonthe abilityoftheworkeranduponthemachinerywithwhichheworks.(1776, IV.ix.3435) Theresultoftheenhancementofproductivepowerswasseentobelower prices.Toillustrateloweredcommodityprices,Ricardoemployedthewell wornexampleofstockings,which,whenefficientlyproduced,would inevitablefallin[exchange]value.(1817,25)Malthuslikewiseobservedthat afterverygreatimprovementshavebeenmadeinthemachineryusedin producingthem,stockingscanbemadeathalfthepricewithoutlosing anyusevalue(1820,241242)[I]mprovedmachinerylowerscostsand[w]e allallowthatwhenthecostofproductiondiminishes,afallofpriceisthe consequence(60233).CharlesBabbagedescribedsuchengineering economicprocessesindetail,forinstanceinthecaseofloweredcostsin makingtanks.(1832,100)ForMillexpensivemachineryproducingupto thefullpowersofthemachineloweredproductioncostssuchthattheowner gainedthepowerofpermanentlyundersellingprevioussuppliers.(1848,

80

133134)Forhimtheproductivenessofindustrycausesdiminishing
77 proportional[marginal]cost.(180182).

Theselowerpricesmeantageneralincomeeffect.Smithintroducedthe conceptoftheagriculturalsurplusresultingfromreductionsintherealprice ofmanufacturedproduce.(1776,I.xi.o.4alsoMill1848,1213,411)Says interpretationwasthat [e]verysavinginthechargesofproduction,thatistosay,everysavingin theproductiveagencyexertedtoraisethesameproduct,isanincreaseof therevenueofthecommunitytoanequalextent.(1803,295) Byproductiveagencyhemeanslandaswellaslabourefficiencyraisesthe totalproductorrevenuewhenanewmachinelowerschargesofits production,therevenueoftheconsumersisbenefited.(87)

InMalthuswords,[A]largemassofmanufacturedarticlescanbeobtained withmuchgreaterfacilitythanbefore[and]thevalueofthewholemassof goodsmadebythenewmachinerygreatlyexceedstheirformervalue.(1820, 49,281)Raesexampleofanimprovementintheartofbakingbread illustratesthepointthatitwouldnotbenefitthebakersexclusively,butwould befeltequallyoverthewholesociety(1834,259118,173)Millidentifieda productionfunctionorlawoftheincreaseofproduction,aconsequenceof bothmoreelementsorrequisitesofproductionandtheirgreater productiveness.(1848,154)

Thesewritersweredescribingthepurchasingpowerthatwasfreedor,more exactly,created: InventionistheonlypoweronearththatcanbesaidtocreateItenters asanessentialelementintotheprocessoftheincreaseofnationalwealth, becausethatprocessisacreation,notanacquisition.(Rae1834,15) Anacquisitionistakenfromsomeoneelse,whilethebenefitsofgreater efficiencyaccruetoallanaggregateincomeeffect.Malthusclaimedthat

77

TheseobservationsconstituteKhazzoomspriceelementintherelationshipbetween

inputefficiencyandinputconsumption.(1980,22)

81

evenifthereweresaturationinsomearea,therewouldbeaportionof
78 revenuesetfreeforthepurchaseoffreshcommodities.(1820,281)

Thenewproductivity,ofcourse,beingaratio,onlyenabledgreater production(andconsumption).Saynotedthattheproductiveagencythus releasedmaybedirectedtotheincreaseofproduction(1803,295,emphasis added),andMalthusaddedthatwecouldchooseleisureorindolence insteadofthemorecheaplyavailablegoods.(1820,170,268)Milllikewise insistedthattheriseinconsumptionwascontingentuponournottaking leisure.(1848,130,133)Infact,afterestablishingthatonepurposeof industrialimprovementisabridginglabour,heironicallyclaimsthat [h]ithertoitisquestionableifallthemechanicalinventionsyetmadehave

79 lightenedthedaystoilofanyhumanbeing.(756757)

Raealsoemphasizedthatallinstrumentspossessacapacityfor supplyingthewants,orsavingthelaborofman.(1834,166,171)However, inthefaceofinventionandincreasingtotalwealthofthecommunityhe asked: Whatdowefindtohavebeenthemostprominentaccompanimentofthis change[inimprovement]?Isitadiminishedexpenditureanincreased parsimonyafrugalitybeforeunknown?Ibelievenot.Anygreat diminutionoftheexpenditureofawholecommunity,itwillbefounddifficult totrace(23) Wemustntconsumemore,but,becauseofefficiencygains,wecan,and usuallydo.Saturationisachangeinpeoplestastes,notanefficiency change.

7.3 TheLabourAnalogyandFrozenTechnology
RaesandJevonsanalysisalsoincludedefficiencysroleinopeningup entirelynewuses(Section4.5),technologicalinterdependenciesrecountedin
78

NotonlySmith,butallthesewritersattributedwealthsincreasetoorganisationalefficiency

aswellasmechanicalefficiency.(e.g.Rae1834,29,95,165,314andMill1848,130135, 184189,706707).
79

Vebleninhis400pagebookbrokehisruleagainstcitationsexactlyonce,quotingthis

passagefromMill.(1899,111)

82

detailbyBabbage(1832).Butdotheseincomeandinventioneffectsaddup towhatwearecallingbackfire?Nothavingaskedthisquestion(untilJevons), thereisofcoursenostraightforwardanswer.Butcertainlymuchgreater productionensued,implyingmoreuseofresources.Malthussummedupthe chainofeventsbywritingthatifanarticlewereto varyinthedifficultyofitsproduction,thecostofproducing[it]were greatlydiminished,thefallofpricewouldbeoccasionedbyanincreased abundanceofsupply,eitheractualorcontingent.Inalmostallpractical casesitwouldbeanactualandpermanentincreasebecausethe competitionofthesellerswouldlowertheprice,anditveryrarelyhappens thatafallofpricedoesnotoccasionanincreasedconsumption.Itis knownthatfacilitiesofproductionhavethestrongesttendencytoopen markets.(1820,5356,287281) Dependingonwhatdefinitionofcirculatingcapitalistakenperhaps whateverisnotfixedthenMillisperhapsdescribingresourceusewhen attestingthat transformationsofcapitalareofthenatureofimprovementsinproduction, which,insteadofultimatelydiminishingcirculatingcapital,arethe necessaryconditionsofitsincrease[T]hereishardlyanycreationof fixedcapitalwhich,whenitprovessuccessful,doesnotcheapenthe articlesonwhichwagesarehabituallyexpended.[and]almostall improventsinmachinerycheapenthelabourersclothingorlodging (1848,751) Raesspeaksmoredirectlytoourquestionofmaterialinputconsumption. Forinstance: Thevariousagriculturalimprovementswithwhichinventionenrichedthat artinBritain,occasionedagreatamountofmaterialtobewroughtup, whichbeforelaydormant.Sometimesaverysmallimprovementmay putalargerangeofmaterialswithinreachoftheaccumulativeprinciple (1834,261262) Furthermore, Amultiplicationofinstrumentsisofnoavail,unlesssomethingadditional begivenonwhichtheymayoperate.Wheninventionsucceedsin discoveringtheseadditionalriches,themereviewissufficient,inevery wellregulatedcommunity,toinduceitsmemberstoformthenew instruments,necessarytodrawtheserichesforth.(29) Raeseemstobeassertingthatbothoutputs(endproducts)andnatural resourceinputsexperiencegreaterdemand.

83

Butitwasonlyonthetopicofinputsoflabourthattheclassicaleconomists reachedaconsensusthathigherproductivitymeantmore,ratherthanless, employmentthanotherwise.Malthuswasparticularlyadamantonthispoint, whichwassometimesdoubtedbyRicardo(Sraffa1951,inRicardo1817,lvii lx),tracingwiththeexampleofcottonmachineryandmanufacturesthe redundancyofworkersintheshorttermbutoverall,longerterm,theneedfor morehandsconcerningemployment,therewaslittlereasontoapprehend anypermanentevilfromtheincreaseofmachinery.(1820,281,287)

Say(1820)deliveredascathinganddetailedrefutationofSismondionthis score,alwaysarguingthattheincreasedproduceatcheappriceswasmore thanenoughtoreemployworkersnolongerneededinthenewlymore efficientsectors.Forreasonsofspace,thistopicoftheneteffectsonsaving moreefficientlyusedlabourinputscannotbepursuedhowever,the argumentsareofgreatrelevancetotodaysenergyefficiencyrebound debate.ByJevonstime,inanycase,economistshadconcludedthatlabour savingprocessessavednolabourespeciallytakingintoaccountpopulation

80 increase,theyraisedemploymentlevels.(Jevons1865,140)

th SimilartoJevons,whoarguedthatthepopulationandaffluenceofmid19

century Britain is unimaginable at the efficiency level of Saverys steam engine (1865, 143, 118 Alcott 2005, 1516), Say asks us to assume the opposite,namelyanefficiencydecline: Bytheruleofcontraries,asarealadvanceofpricemustalwaysproceed from a deficiency in the product raised by equal productive means, it is attendedbyadiminutioninthegeneralstockofwealth.(302) Weaskaftertheconsequencesoftechnologicalefficiencyincreases,butSay challengesustoassumetheoppositeandseeifthiscouldconceivablylead tohigherinputconsumption.Aretodayslevelsofconsumptionpossibleunder theassumptionoftechnologyfrozenatsomeearlierhistoricallevel?

80

SeealsoSmith,IV.ix.34Say,8690Jones1831,249250Mill,63,66,92,100115

Greenberg1990.

84

An anonymous writer similarly took the argument at its word that labour efficiencygainsledtolesspopulationandaffluence: If the use of machinery is calculated to diminish the fund out of which labourersaresupported,thenbygivinguptheuseoftheploughandthe harrowandreturningtothepastoralstate,orbyscratchingtheearthwith ournails,theproduceofthesoilwouldbeadequatetothemaintenanceof a much greater number of labourers. There are many labourers now in England, and the gradations of ingenuity and skill in machinery are numerousbutasthenumberoflabourersandthefundsfortheirsupport would be gradually increasedinproportionaswefellbackupontheless perfect machinery, so, at last, when we deprived ourselves entirely of its assistance, the produce and hence the population of England would be increased beyond what has ever been exhibited in any country upon the 81 surfaceoftheglobe....(Anon.1826,102) The efficiency strategy theory holds that higher efficiency causes less resource consumption turned around, lower efficiency would then raise consumption. Say,aswell,addsthisobverseargument: Acountryisrichandplentiful,inproportionasthepriceofcommoditiesis low.Forargumentssake,Iwillputthematterinthemostfavourablelight for those who dispute this maxim, and suppose themtourgeanextreme case, namely, that, by successive economical reductions, the charges of productionareatlengthreducedtonothingWhatthen?Everyobject of human want would stand in the same predicament as the air or the water[Smith1776,I.iv.13Lauderdale1804,4345],whichareconsumed without the necessity of being either produced or purchased. In like manneraseveryoneisrichenoughtoprovidehimselfwithair,sowould hebetoprovidehimselfwitheveryotherimaginableproduct.(303304) That is, if efficiency causes price and costofproduction decrease, raising demand,imaginethesituationifproductioncostsfelltozero.Accordingtothis

reductio adabsurdum, to the extent that consumption entailsemissionsand

human appropriation of space, at lower efficiency population and environmental impact would be astronomical. This reasoning must be counteredbytodaysefficiencystrategists.

81

Anon.thenchidesRicardoforentertainingtheantimachineposition.

85

7.4Conclusion
ThefirstandtodatemostexhaustivetreatmentofourquestionisJevons

TheCoalQuestion(1865),whoseargumentstheauthorhaselsewhere
analysed(Alcott2005).Jevonspredecessorscommentedontheeighttopics listedatthischaptersbeginning,andonlyindirectlyontheresource consumptionreboundquestion.However,theirinterpretationsofthepalpable correlationbetweenagrowingeconomyandgreaterproductivityiscloserto thebackfirepositionofJevonsthanthesavingspositionofenergyefficiency strategists.Deeperanalysisofthesethinkersseemsadvisable.Inparticular,
th anexactanalysisofthe19 centurydebateoverreboundandbackfire

concerninglabourinputsmightrevealthattodaysdebateovermaterialand energyinputsisreinventingthewheel.

86

Chapter8Discussion
8.1Concepts
Estimates of total rebound range at least from 26% (4CMR 2006, 6) to significantly over 100% (Jevons 1865 Brookes passim. Saunders passim.) This fact suggests that there are issues of definition, taxonomy, scope of study, and assumptions in baseline scenarios that areunresolved.Untilthis kindofgroundworkisdone,thisvariationwillcontinuetopreventusefulpolicy advice, the more so because the two extremes recommend contradictory
82 policies. The concepts of energy services, useful work, and economic

activity, for instance, are perhaps superfluous, and blur distinctions concerning what consumption is being measured in rebound study. The cacaphonousarrayoftypesofrebound,furthermore,confusediscussion.

Forexample,isthetermenergyforjoules(ameasureforenergyaswellas work),orusefulwork?Isthedistinctionbetweenexergyandenergyuseful? When we purportedly consume services, are these, for example, car kilometres, passenger kilometres, or even the pure utility of the act? The mostdamagingambiguity,however,concernsoutput,production,economic activity and growth: Are these physical, monetary, or welfare units? If the last, then it is true that economic growth and stable or declining resource consumption are compatible. If the first, this is (tautologically) false. The conceptual difficulty about declining energy content of consumption, structuralchangeanddematerialisationordecouplingis:Fromwhatarethe inputs decoupled? If from service or utility or welfare, then it must be shownthattheconceptisrelevantforthestudyofenvironmentalimpactand more specifically rebound. If from physical output, though, then this is technological efficiency increase, and any claim that it weakens rebound effectsbegsthequestion.

82

Theanthropocentricityoftheconceptofusefulworkisillustratedbythecaseofthewaste

heatfromnuclearpowerplantsattheoceansedge:iftechnologyexiststousethisfor desalinationofwater,thesametype(form)andamountofenergyhasbecomeuseful.

87

8.2PolicyAlternatives
Efficiency is positively connotated (Moezzi 2000 Sanne 2002) do we
83 sense that it is no threat to material welfare? The factor four political

economicvisionseeksafteralltocutresourceconsumptioninhalfanddouble

affluence. The rare political unanimity in favour of efficiency policies

encompassingGreensandConservatives,GeorgeBushandGordonBrown is even predicted by backfire theory, as Jevons intimated. (1865, 136, 460) The rival quota strategy caps and permits faces ecological limits more clearlyanddethronesthegoalofmaintainingmaterialaffluenceasonewhich

must be fulfilled. It is not only the principle of the Kyoto Protocol (also
Bodansky 1993, 509), but receives academic attention (Daly 1974, 1920 Hannon 1975, 101 Greenhalgh 1990, 296297 Victor 1991, 209 Jacobs 1991,123Starkey&Anderson2004Simms2005Aubauer2006,644646). Quotas are, with good reason, not positively connotated i.e. if we want to conserve, or lower impact, [i]t would be more straightforward to direct that thereshouldbereductionsinworldeconomicactivity,ofspecificemissions, orseekworldwideagreementtoplacingheavytaxesontheoffendingfuels. (Brookes1990,201)

There is a parallel to more local problems Owens & Cowell observe that becauseitissodifficulttoreduce therateoftrafficgrowthaviewthatpolicyshouldfocusonreducing pollutionandcongestion,ratherthanthevolumeoftrafficperse,has prevailed,convenientlyshiftingattentiontowardsvehicleperformance, trafficmanagementandselectedimprovementsintheroadnetwork. (2002,97) Justaseconomicgrowthweakensconcernfordistributionaljusticeproblems, focussing attention on efficiency, something that is relative, familiar, and positive,takesattentionawayfromthenecessityforabsolutegloballimitsto consumption and pollution. Yet the presence of backfire does imply that environmentalbenefitscannotbeguaranteedbysuchpoliciesalone.(Allan
83

Glomsrd&Taoyuanbringattentiontothepossibilitythat[moreefficiencyresultingfrom]

coalcleaningisacaseforeconomicefficiencymorethanenergysavingandclimatepolicy. (2005,536)

88

etal.2006,51)Ifourmaincriterioniseffectiveness,however,limitsmustbe establishedbyprior[exomarket]law.(Stewart2000,212214)

Decisionsaboutpopulation,affluence,andefficiencywouldthentakecare ofthemselves: Undercarbonrationing,therationingschemeitselfshouldprovideallthe incentive that individuals would need to search out the lower carbon energy supplies, lower energy using products and services and so on. Nolongerwoulditbenecessarytohaveseparategovernmentpoliciesand programmes to promote everything from cycling strategies to efficient refrigerators.(Fawcett2004,1078,1077) Thus,quotasareprobablymorecosteffectiveaswell.

It seems that many if not most economists, starting with Smith, regard efficiencyasacauseoflargereconomicsize,physicallyaswellasintermsof welfare. Efficiency of all sorts is usually seen as a means toward affluence. Thus the view must be contradicted that Periodically, claims [of backfire] surface(Schipper&Grubb2000,367)instead,efficiencysharnessingfor environmental purposes seems to be the exception, happening during the
th coal supply worries of the mid19 century and the depletion and pollution

crisisstartinginthe1960s.

8.3FrozenTechnology
Malthusbelievedthatlowerefficiencycauseslowermaterialwelfareinthe senseoflowerinputconsumption,inthiscaseagriculturalland: Even in our agriculture, if the fixed capital of horses, which from the quantity of produce they consume, is the most disadvantageous descriptionoffixedcapital,weredisused[anefficiencyloss],itisprobably, thatagreatpartofthelandwhichnowbearscornwouldbethrownoutof cultivation. Land of a poor quality would never yield sufficient to pay the labourofcultivatingwiththespade,ofbringingmanuretodistantfieldsin barrows,andofcarryingtheproductsoftheearthtodistantmarketsbythe samesortofconveyance.(1820,192) Efficiencystrategy theory, on the other hand, must maintain the opposite: if more efficiency causes less total input use, then less efficiency must cause more use. This leads us again to the thought experiment of holding technological efficiency frozen. Again, Jevons insight: Saverys engine

89

consumed no coal, because its rate of consumption was too high. (1865,
118,143)EvenatWattshighestefficiencylevels,wouldwehavebeenable in terms of time, population, distances travelled, and machine capital to consume as much as humans do today? This seems implausible we would forinstanceprobablyhavenorailroads,cement,orirrigationpumps. Or, following Say (Section 7.3): Holding technological efficiency change (increase) as it has in fact been since, for instance, 1781, what must the efficiencystrategy (antibackfire) theory predict about todays levels of consumption (population times affluence)? It must predict that these have

fallen. However, this scenario seems not only counterfactual but

unimaginable. The theory has it that we consume more things without their being offered more efficiently than when they are offered more efficiently and without thereby having freed time. Inputs are no longer normal goods.

Asinputsbecomemorecostly,weconsumemoreofthem.
ThisreasoningisbehindWirlssuggestingthepolicypossibilitythathighly uneconomical [light]bulbs are prescribed, which would result in a reversal of...rebound.(1997,39)Hotellinglikewisenoted: The method ordinarily proposed to stop the wholesale devastation of irreplaceable natural resources, or of natural resources replaceable only withdifficultyandlongdelay,istoforbidproductionatcertaintimesandin certain regions or to hamper production by insisting that obsolete and inefficientmethodsbecontinued.(1931,137) The rebound discussion should open itself to this corollary of efficiency strategytheory,justasJevonstheorymustexplainanyrealworlddecreases inconsumptionratesandevendecreasesinratesofincrease.

Finally,perhapsweshouldignoretheconceptsofengineeringsavingsand reboundandusegrowththeorypositivelytoexplaintheconsumedamounts of inputs of energy as well as any other material, land, labour, and time. Baseline trends would no longer have a technology factor lowering input consumption, andthebackfirequestionwouldbeopen.Worldstudiesusing physical denominators for efficiency seem the most promising path upon

90

whichtodothis.Still,wehaveaparadox:perhapsdemandforoutputdoes notimplyproportionaldemandforinput.

91

Chapter9Conclusions
9.1ByChapter
ONE Chapter 1 identified the energyefficiency environmental strategy, its popularity, and Jevons challenge to it. The topic remains aparadox: Given ratio decreases of energy intensity for a given product or process, how can aggregateenergyconsumptioncontinuetoincreasebecauseofthis(orisitin

spiteofthis)?
TWO For rebound study a clear concept of engineering savings is indispensable, of which rebound is a percentage rebound is only that consumptionenabledbyefficiencyincreases.Themechanismenablingmore consumption is at once lower production costs and prices, enhanced purchasing power, and substitution for other production factors. Although rebounds existence is indubitable, practically all evaluations of efficiency policies effectiveness ignore it in computing energy savings and in quantifying consumption trends without policies. Correction of engineering savings(orpotentialsavings)byafactorofperhaps0.5isjustifiable. THREE Like organisational efficiency and consumption efficiency, technological efficiency influences the size of the ratio T in I=PAT. Since howeverthesizeofthisratioinfluencesPandAthereisnonecessaryeffect on I. We thus cannot assume, as does the efficiency strategy, that lower T meanslowerI,andthusI=f(P,A,T)isaccurate:anychangeontherightside canchangetheothertwofactors.DeducinganabsolutequantityforIisinany casenotpossiblefromtheratioT.I.e.dynamiceconomicfactscomplicatethe staticengineeringfactsusedinevaluations.LowerTcausallyreducesIonly tothedegreethatconsumersarematerially,orenergetically,satiated.Testing theEKChypothesisthatthisisafunctionofrisingincomerequiresclarity about whether the quantities in the dependent variable are absolute, or are themselvesratios. FOUR Whilethenewlycreatedorfreedpurchasingpowercanbeavailed of, it does nothave to be,i.e.previouslyimpossibleengineeringsavingsdo 92

becomepotential.Itisavailedofforthesameorothergoodsbythesameor other, marginal consumers, yielding a suggested taxonomy to replace the plethora of terms in the literature. The new relative cheapness and productiveness of energy can mean, through a substitution effect, that the newly demanded goods have a relatively high energy content, perhaps countervailinganysaturationeffectpurportedtoresultfromthehigherincome (EKChypothesis). Policyadvicebasedonlyondirectrebound,ratherthantotalrebound,isnot useful. Academic studies should take only total rebound as their ultimate study object, using direct rebound information only as a starting point and specifyingtheroleofdirectreboundstudyintotalreboundstudy.Moreover, becauseoftradeandtheglobalnatureofmanyimpacts,countrystudiesare misleading.Finally,muchinfluentialliteraturesuffersfromuncleardefinitions andinconsistentapplicationoftaxonomy. FIVE Ifefficiencyincreasesraisethepurchasingpowerofmorepeople,for moregoods,thenpartofrisesinpopulationandproductionmustbecounted as rebound. That is, levels of population and economic activity are not determined totally exogenously as assumed in most models. Rebound is by definitiononlythatconsumptioncausedbygreaterproductionefficiency,not allgrowthforces.Inmakingcounterfactualclaimsaboutrealenergysavings, assumptionsoftherealsavingseffectsofefficiencybegthequestion. Because rebound is nothing more than a percentage of potential savings, energysshareoftotalGDPisnotrelevanttoitsmeasurement.Similarly,the costsofefficiencyincreasesinfluencethesizeofengineeringsavings,butnot ofrebound.Finally,ithasnotbeenshownthatwhensectorscharacterisedby marketfailureexperienceefficiencyimprovements,incomeorpriceeffectsare lower than in situations of greater competition, information, or motivation. Whether barriers to efficiency increases are environmentally good or bad dependsonthebackfireissue,butinanycasereboundscienceshouldcease toconflategoalsoflowerimpactandgreatereconomicefficiency. SIX Definingefficiencyfornarrowlydefinedproductsorsectorsispossible, despiteproblemswithnewproductsand/orchangingcharacteristicsofsimilar products computations based on these definitions, after all, yield the key quantityengineeringsavings.However,ifweinvestigatetheforestinsteadof 93

the trees, defining and calibrating technological efficiency for the world economy is difficult. Do we need fully aggregate numerators and denominators? Do we define denominators in terms of money, services or physical units like space, weight, or chemical elements? The problems of GDParesevere,andservicehasnotbeenshowntoberelevanttorebound study, leading to the conclusion that physical metrics should be sought, perhaps averaging efficiency increases, in percentages, over differently measuredsectors. Nevertheless consensus reigns on a broad correlation over time and national economies between lower energy input per unit of each of the above denominators and rising consumption of energy. If this is mere correlationratherthancausality,othercausesmustexplaintheobservedrise inconsumption.Ifrisingefficiencycauseslowerinputconsumption,theforces that do explain consumptions rise must moreover be strong enough to overcome this shrinkage effect of greater efficiency. Empirical statistical methods to answer the backfire question, presupposing an answer to the vexing question of denominators, must study world, total rebound. Yet the opinion that rebound is only an empirical question is incorrect because designingsuchresearchpresupposessometheory. SEVEN Jevons deemed his paradox easily solved. Perhaps this was

because his predecessors Smith, Say, Ricardo, Malthus, Babbage, Rae, Hearn and Mill had constructed plausible accounts of economic growth in roughagreementwiththelaterresultsofSolowwhichexplainedthesizeof the economy, the amount of wealth, in terms of organisational and technological efficiencies concerning labour and, less specifically, material/energy inputs. These writers attested efficiencyinduced price falls, purchasing power effects, expanding markets, and increasing numbers of nonsatiatedconsumersaswellaswhatwenowcallbackfirewithregardto employmentinputs.

9.2General
Theanswertothedissertationstitlequestionisthatassessmentsofenergy efficiency policy should take rebound into account. Whetherbackfire should

94

betakenintoaccountisstilldebatable.Whilebackfiretheorypredictstherise of some combination of population and affluence alongside greater energy consumption, this is not proof. In particular, any theory must make assumptionsaboutconsumersaturationtheonlyforcethatlowersrebound. Increaseintheefficiencyofanyinputfreesthatinput,butthisonlyenablesus to thereafter use more of it at no cost efficiency gain is only a necessary condition of greater consumption of that input, just as it is a necessary condition of more leisure and nonconsumption. The sufficient condition for
84 higher consumption depends on the world demand function. That is, the

priceandincomeelasticityofdemandisnecessarytodeterminetheefficiency elasticity of demand. Yet policy must make assumptions about consumer behaviour based on less than full empirical knowledge, necessitating some theory of consumer psychology. Efficiency enables backfire, and not just rebounduptounityitexpandstheproductionpossibilityfrontier.Therefore,if worldwide latent demand is judged to be great, there is danger that forcing efficiencyisenvironmentallycounterproductive.

84

Malthuspopulationprinciplewastheidenticalargument:neitherincreasedsubsistence

norourtendencytoreproduceweresufficientforpopulationgrowth.(1798,19,33,7678, 115)

95

Bibliography
Alcott,Blake,2004.JohnRaeandThorsteinVeblen.JournalofEconomic Issues38(3):765786. Alcott,Blake,2005.JevonsParadox.EcologicalEconomics54(1):921. Ahmad,Syed,1996.SmithsdivisionoflabourandRaesinvention:Astudy oftheseconddichotomywithanevaluationofthefirst.HistoryofPolitical Economy28(3):441458. Allan,Grant,NickHanley,PeterG.McGregor,J.KimSwales,andKaren Turner,2006.ThemacroeconomicreboundeffectandtheUKeconomy. FinalReporttoDefra.www.defra.gov.uk/environment/energy/research Anonymous,1826.EffectoftheEmploymentofMachinery&c.uponthe HappinessoftheWorkingClasses.WestminsterReviewV(January1826): 10130.AccordingtoMillbyWilliamEllis. Arrow,Kenneth,BertBolin,RobertCostanza,ParthaDasgupta,CarlFolke, C.S.Holling,BengtOweJansson,SimonLevin,KarlGranMler,Charles Perrings,&DavidPimentel,1995.Economicgrowth,carryingcapacity,and theenvironment.EcologicalEconomics15(2):9195. Aubauer,HansPeter,2006.Ajustandefficientreductionofresource throughputtooptimum.EcologicalEconomics58(3):637649. Ausubel,JesseH.,1989.Regularitiesintechnologicaldevelopment.In:Jesse Ausubel&HedyE.Sladovich(eds.),TechnologyandEnvironment. NationalAcademyPress,Washington,D.C. Ayres,RobertU.,1978.Resources,Environment,andEconomics: ApplicationsoftheMaterials/EnergyBalancePrinciple.JohnWiley&Sons, NewYork. Ayres,RobertU.,2000.Commentaryontheutilityoftheecologicalfootprint concept.EcologicalEconomics32:347349. Ayers,RobertU.,&StevenM.Miller,1980.Theroleoftechnologicalchange. JournalofEnvironmentalEconomicsandManagement7:35371. Ayres,RobertU.,&JeroenC.J.M.vandenBergh,2005.Atheoryof economicgrowthwithmaterial/energyresourcesanddematerialization: Interactionofthreegrowthmechanisms.EcologicalEconomics55(1):96 118. Babbage,Charles,1832.OntheEconomyofMachineryandManufactures. Knight,London.

96

Balzhiser,RichardE.,1989.Meetingtheneartermchallengeforpower plants.In:JesseAusubel&HedySladovich(eds.),Technologyand Environment.NationalAcademy,Washington,D.C. Barker,Terry,PaulEkins,andNeilStrachan,2005.EnergyEconomy EngineeringEnvironmentAnE4representationoftheUKenergysystem. WorkingPaperOctober2006,CambridgeEconometricsandthePolicy StudiesInstitute. Bartelmus,Peter,1997.Whithereconomics?Fromoptimalitytosustainability. EnvironmentandDevelopmentEconomics2:323345. Bartlett,AlbertA.,1994.Reflectionsonsustainability,populationgrowth,and theenvironment.PopulationandEnvironment16(1):535. Berkhout,Peter,JosMuskens,&JanVelthuijsen,2000.Definingtherebound effect.EnergyPolicy28(6/7):425432. Besen,StanleyM.,&LelandL.Johnson,1982.CommentonEconomic implicationsofmandatedefficiencystandardsforhouseholdappliances. EnergyJournal3(1):110116. Besiot,W.,&K.J.Noorman,1999.Energyrequirementsofhousehold consumption.EcologicalEconomics28:367383. BfE(BrofrEnergie/SwissEnergy),2005.AnnualReport200405 www.bfe.admin.ch/energie/oo556/index.html Binswanger,HansChristof,WernerGeissberger&TheoGinsburg,1979. WegeausderWohlstandsfalle.Fischer,Frankfurt. Binswanger,HansChristof,1998.Makingsustainabilitywork.Ecological Economics27(1):311. Binswanger,Mathias,2001.Technologicalprogressandsustainable development:Whataboutthereboundeffect?EcologicalEconomics36 (1):119132. Birol,Fatih,&JanHorstKeppler,2000.Prices,technologicaldevelopment andthereboundeffect.EnergyPolicy28(6/7):457469. Bodansky,Daniel,1993.TheUnitedNationsFrameworkConventionon ClimateChange:Acommentary.YaleJournalofInternationalLaw18:451 558. Boulding,KennethE.,194950.Incomeorwelfare.TheReviewofEconomic Studies,volumeXVII:7786.

97

Boulding,KennethE.,1966.Theeconomicsofthecomingspaceshipearth. In:HenryJarrett(ed.),EnvironmentalQualityinaGrowingEconomy.Johns HopkinsPress,Baltimore. Brewer,Anthony,1991.Economicgrowthandtechnicalchange:JohnRaes critiqueofAdamSmith.HistoryofPoliticalEconomy23(1):111. Brookes,Leonard,1978.Energypolicy,theenergypricefallacyandtherole ofnuclearenergyintheUK.EnergyPolicy6(1):94106. Brookes,Leonard,1990.Thegreenhouseeffect:Thefallaciesintheenergy efficiencysolution.EnergyPolicy18(2):199201. Brookes,Leonard,1992.Energyefficiencyandeconomicfallacies:Areply. EnergyPolicy20(5):390392. Brookes,Leonard,2000.Energyefficiencyfallaciesrevisited.EnergyPolicy 28(6/7):355366. Brown,Lester,etal.,1998.StateoftheWorld.Worldwatch/Earthscan, London. Burniaux,JeanMarc,JohnP.Martin,JoaquimOliveiraMartins,&Dominique vanderMensbrugghe,1995.Carbonabatement,transfersandenergy efficiency.In:IanGoldin&L.AlanWinters(eds.),TheEconomicsof SustainableDevelopment.CambridgeU.Press,Cambridge. CEPE(CentreforEnergyPolicyandEconomics),2003.Begleitende EvaluationderWirkungsanalyse2002vonEnergySchweiz.SwissEnergy, Bern. Christensen,Paul,1991.Drivingforces,increasingreturnsandecological sustainability.In:RobertCostanza(ed.),EcologicalEconomics:The ScienceandManagementofSustainability.ColumbiaU.Press,NewYork.
th Cipolla,CarloM.,1962/1974.TheEconomicHistoryofWorldPopulation,6 ed.,Penguin,Harmondsworth,U.K.

Clapp,BrianW.,1994.AnEnvironmentalHistoryofBritain.Longman, London. Cleveland,Cutler,RobertCostanza,CharlesA.S.Hall,andRobertKaufmann, 1984.EnergyandtheU.S.economy:Abiophysicalperspective.Science 225:890897. Cleveland,CutlerJ.,&MatthiasRuth,1998.Indicatorsofdematerialization andthematerialsintensityofuse.JournalofIndustrialEcology2(3):1550. Cogoy,Mario,1999.Theconsumerasasocialandenvironmentalactor. EcologicalEconomics28:385398. 98

CONV850/03.TheEuropeanConventionDraftTreatyestablishinga ConstitutionforEurope,18.7.2003. CyclopediaofCommerce,1858. Dahlstrm,Kristina,&PaulEkins,2006.Combiningeconomicand environmentaldimensions:ValuechainanalysisofUKironandsteelflows. EcologicalEconomics58(3):507519. Dahmus,JeffreyB,&TimothyG.Gutowski,2005.Efficiencyandproduction: Historicaltrendsforsevenindustrialsectors.WorkingPaper,USSociety forEcologicalEconomicsConference,July2005,Takoma,Washington. jdahmus@mit.eduandgutowski@mit.edu. Daly,HermanE.,1974.Theeconomicsofthesteadystate.American EconomicReview64(2):1521. Daly,HermanE.,1991.Elementsofenvironmentalmacroeconomics.In: RobertCostanza(Ed.),EcologicalEconomics.ColumbiaU.Press,New York. Daly,HermanE.,1992.Steadystateeconomics:Concepts,questions, policies.GaiaINo.6:333338. Daly,HermanE.,1992.Allocation,distribution,andscale:Towardsan economicsthatisefficient,just,andsustainable.EcologicalEconomics 6:185193. Daly,HermanE.,&CliffordW.Cobb,1989.FortheCommonGood.Beacon, Boston. DeBruyn,Sander,andJohnB.Opschoor,1997.Developmentsinthe throughputincomeRelationship:Theoreticalandempiricalobservations. EcologicalEconomics20(3):255268. deHann,Peter,MichaelG.Mueller,&AnjaPeters,2006.Doesthehybrid ToyotaPriusleadtoReboundeffects?Analysisofsizeandnumberofcars previouslyownedbySwissPriusbuyers.EcologicalEconomics58(3):592 605. Defra(UKDepartmentoftheEnvironment,Food,andRuralAffairs),2002. EnergyEfficiencyCommitment.http://www.defra.gov.uk/environment/eec Defra,2004a.EnergyEfficiencytheGovernmentsPlanforAction,April 2004.

99

Defra,2004b.ReviewoftheUKClimateChangeProgramme,Consultation Paper.www.defra.gov.uk/corporate/consult/ukccpreview/ccpreview three.pdf Defra,2006.www.defra.gov.uk/MarketTransformationProgramme,BNCK06: KettleTrends,9/2/2006. DoE(USDepartmentofEnergy)2006.EnergyInformationAdministration www.eia.doe.gov/emeu/efficiencyandOfficeofEnergyEfficiencyand RenewableEnergywww.eere.energy.gov Domar,EvseyD.,1961.Onthemeasurementoftechnologicalchange.The EconomicJournal71(4):709729. DTI(DepartmentofTradeandIndustry),2003.OurEnergyFutureCreating aLowCarbonEconomy(EnergyWhitePaper)www.dti.gov.uk. DTI,2006.TheEnergyChallenge:EnergyReviewReportJuly2006online. Duchin,Faye,&GlennMarieLange,1994.TheFutureoftheEnvironment. OxfordU.Press,Oxford. Durham,D.F.,1992.Culturalcarryingcapacity:I=PACT.Focus2(3):58. EEB(EuropeanEnvironmentalBureau),2000.Document2000/021,Sarah KeayBright.online Ehrlich,PaulR.,AnneH.Ehrlich,&JohnP.Holdren,1973.HumanEcology: ProblemsandSolutions.W.H.Freeman,SanFrancisco. Ehrlich,PaulR.,&JohnP.Holdren,1971.Impactofpopulationgrowth. Science171:12121217. Ekins,Paul,1991.Thesustainableconsumersociety:Acontradictionin terms?InternationalEnvironmentalAffairs3:243257. ElSerafy,Salah,1991.Theenvironmentascapital.In:RobertCostanza(ed.), EcologicalEconomics.ColumbiaU.Press,NewYork. Enos,JohnL.,1958.Ameasureoftherateoftechnologicalprogressinthe petroleumrefiningIndustry.JournalofIndustrialEconomics6(3):180197. Enos,JohnL.,1994.Evaluationsofimprovementsinpetroleumrefining technology:Incentives,researchanddevelopmentandoutcomes.Oxford InstituteforEnergyStudiesEE18online. ES(EnergieSchweiz)/INFRAS,2004.WirkungsanalyseEnergieSchweiz2003. Bern&Zrich.

100

ES(EnergieSchweiz)/Prognos,July2004.SyntheseberichtEntwicklungund BestimmungsgrndedesEnergieverbrauchs2003imVergleichmit2002 und1990.Bern&Basel. Etzioni,Amitai,1998.Voluntarysimplicity:Characterization,select psychologicalimplications,andsocietalconsequences.Journalof EconomicPsychology19:619643. Fawcett,Tina,2004.Carbonrationingandpersonalenergyuse.Energy& Environment15(6):10671083. Fouquet,Roger,&PeterJ.G.Pearson,2006.Sevencenturiesofenergy services.EnergyJournal27(1):139177. 4CMR(CambridgeCentreforClimateChangeMitigationResearch),2006. ThemacroeconomicreboundeffectandtheUKeconomy.FinalReportto Defra,18May.www.defra.gov.uk/environment/energy/research Foxon,TimothyJ.,2000.Resourceefficiencyandserviceprovision.Energy andEnvironment11(5):587595. Gardner,GeraldT.,&PaulC.Stern,1996.EnvironmentalProblemsand HumanBehavior.Allyn&Bacon,Boston. GeorgescuRoegen,Nicholas,1971.TheEntropyLawandtheEconomic Process.HarvardU.Press,Cambridge,Massachusetts. Giampietro,Mario,1994.Sustainabilityandtechnologicaldevelopmentin agriculture.BioScience44(19):677689. Giampietro,Mario,&KozoMayumi,1998.Anotherviewofdevelopment, ecologicaldegradation,andNorthSouthtrade.ReviewofSocialEconomy 56(1):2036. Giampietro,Mario,&KozoMayumi,2000.Multiplescaleintegrated assessmentsofsocialmetabolism:Integratingbiophysicalandeconomic representationsacrossscales.PopulationandEnvironment22(2):155 210. Glomsrd,S.,&W.Taoyuan,2005.Coalcleaning:Aviablestrategyfor reducedcarbonemissionsandimprovedenvironmentinChina?Energy Policy33(4):525542. Greenberg,Dolores,1990.Energy,power,andperceptionsofsocialchange. AmericanHistoricalReview95(3):693714. Greenhalgh,Geoffrey,1990.Energyconservationpolicies.EnergyPolicy18 (4):293299.

101

Greening,LornaA.,DavidL.Greene,&CarmenDifiglio,2000.Energy efficiencyandconsumptionthereboundeffectasurvey.EnergyPolicy 28(6/7):389401. Grossman,GeneM.,1995.Pollutionandgrowth:Whatdoweknow?In:Ian Goldin&L.AlanWinters(eds.),TheEconomicsofSustainable Development.CambridgeU.Press,Cambridge. Grubb,MichaelJ.,1990a.EnergyPoliciesandtheGreenhouseEffect,Vol.1. DartmouthPublishingCo,AldershotU.K. Grubb,MichaelJ.,1990b.Communication:Energyefficiencyandeconomic fallacies.EnergyPolicy18(8):783785. Grubb,MichaelJ.,1992.ReplytoBrookes.EnergyPolicy20(5):392393. Hannon,Bruce,1975.Energyconservationandtheconsumer.Science189. Hannon,Bruce,1998.LettertotheEditor.EcologicalEconomics27(2):215 216. Hearn,WilliamEdward,1864.Plutology:OrtheTheoryoftheEffortsto SatisfyHumanWants.Macmillan,London,andGeorgeRobertson, Melbourne. Herman,Robert,SiamakA.Ardekani,&JesseH.Ausubel,1989. Dematerialization.In:JesseAusubel&HedySladovich(eds.),Technology andEnvironment.NationalAcademyPress,Washington,D.C. Hettige,H.,R.E.B.Lucas,&D.Wheeler,1992.Thetoxicintensityofindustrial production:Globalpatterns,trends,andtradepolicy.AmericanEconomic Review82(2):478481. Hinchliffe,S.,1995.Missingculture:Energyefficiencyandlostcauses. EnergyPolicy23(1):9395. Hinterberger,Friedrich,FredLuks&FriedrichSchmidtBleeck,1997. Material flowsvs.naturalcapital':Whatmakesaneconomysustainable?Ecological Economics23(1):114. HoltzEakin,Douglas,&ThomasM.Selden,1995.Stokingthefires?CO2 emissionsandeconomicgrowth.JournalofPublicEconomics57:85101. Hotelling,Harold,1931.Theeconomicsofexhaustibleresources.Journalof PoliticalEconomy39(2):137175.In:AdrianC.Darnell(ed.),1990.The CollectedEconomicArticlesofHaroldHotelling.SpringerVerlag,New York. Howarth,RichardB.,1997.Energyefficiencyandeconomicgrowth. ContemporaryEconomicPolicyXV(4):19. 102

INFRAS,2000.WirtschaftlicheAuswirkungendesInvestitionsprograms Energie2000.BundesamtfrEnergie,Bern. INFRAS,2003.BegleitendeEvaluationderWirkungsanalyse2002von EnergieSchweiz.EnergieSchweiz,Bern. INFRAS,2004.EvalutationderWirkungsanalysedesLabelsEnergiestadt. BundesamtfrEnergie,Bern. Inhaber,Herbert,1997.WhyConservationFails.Quorum,Westport, Connecticut. Jacobs,Michael,1991.TheGreenEconomy:Environment,Sustainable DevelopmentandthePoliticsoftheFuture.Pluto,London&Concord, Massachusetts. Jnicke,Martin,HaraldMnch,ThomasRanneberg,&UdoE.Simonis,1989. Structuralchangeandenvironmentalimpact.Intereconomics24(Jan. Feb.):2435. Jevons,WilliamStanley,1865/1965.TheCoalQuestion:AnInquiry ConcerningtheProgressoftheNation,andtheProbableExhaustionof rd ourCoalmines.3 edition1905,Ed.A.W.Flux.AugustusM.Kelley,New York. Johnson,G.Dale,2000.Population,food,andknowledge.American EconomicReview90(1):114. Johnson,G.Dale,2001.Onpopulationandresources:Acomment. PopulationandDevelopmentReview27(4):739747. Jones,Richard,1831.EssayontheDistributionofWealthandontheSources ofTaxation.JohnMurray,London. Kaufmann,Robert,1992.Abiophysicalanalysisoftheenergy/realGDPratio: Implicationsforsubstitutionandtechnicalchange.EcologicalEconomics 6:3556. Khazzoom,J.Daniel,1980.Economicimplicationsofmandatedefficiencyin standardsforhouseholdappliances.EnergyJournal1(4):2140. Khazzoom,J.Daniel,1982.ResponsetoBesenandJohnsonsComment. EnergyJournal3(1):117124. Kolstad,CharlesD.,2002.Climatechangepolicy:AviewfromtheU.S. DepartmentofEconomics,UniversityofCaliforniaatSantaBarbara, WorkingPaper2202online.

103

Kristrm,Bengt,2000.Growth,employmentandtheenvironment.Swedish EconomicPolicyReviewforthcoming,draft14Feb.2000online. Landes,DavidS.,1969.ThePrometheusUnbound:TechnologicalChange andIndustrialDevelopmentinWesternEuropefrom1750tothePresent. CambridgeU.Press,London. Lantz,V.,&QFeng,2006.Assessingincome,population,andtechnology impactsonCO2 emissionsinCanada:WherestheEKC?Ecological Economics57(2):229238.

th Lauderdale,8 Earlof,1804/1819.AnInquiryintotheNatureandOriginof PublicWealthandintotheMeansandCausesofitsIncrease.Arch. Constable&Co.,Edinburgh.

Levett,Roger,2004.Qualityoflifeecoefficiency.Energy&Environment15 (6):10151026. Lovins,Amory,1988.Energysavingresultingfromtheadoptionofmore efficientappliances:Anotherview.EnergyJournal9(2):155162. Malthus,ThomasRobert,1798.AnEssayonthePrincipleofPopulation.Ed. AnthonyFlew.Penguin,Harmondsworth,UK.

nd Malthus,ThomasRobert,[1820]1836.PrinciplesofPoliticalEconomy,2 Ed. Eds.WrigleyandSouden,1986.Pickering,London.

Manne,A.S.,&R.G.Richels,1990.CO2 emissionlimits:Aneconomiccost analysisfortheUSA.EnergyJournal11(2):5174. MartinezAlier,Juan,1987.EcologicalEconomics:Energy,Environmentand Society.Blackwell,Oxford. Mill,JohnStuart,1848/1965.PrinciplesofPoliticalEconomy.Ed.J.M. Robson.U.ofTorontoandRoutledge&KeganPaul. Moezzi,Mithra,2000.Decouplingenergyefficiencyfromenergyconservation. Energy&Environment11(5),521537. Norgard,J.S.,2006.ConsumerefficiencyinconflictwithGDPgrowth. EcologicalEconomics57(1):1529. NRC(NationalResearchCouncil),2002.EffectivenessandImpactof CorporateAverageFuelEconomy(CAFE)Standards.NationalAcademies Press. Opschoor,Hans,1995.Ecospaceandthefallandriseofthroughputintensity. EcologicalEconomics15(2):137140.

104

Opschoor,Hans,2000.Theecologicalfootprint:Measuringrodormetaphor? EcologicalEconomics32:363365. OPSI(OfficeofPublicSectorInformation),2003.SustainableEnergyAct. www.opsi.gov.uk. Owen,Gill,2000.Energyefficiencyandenergyconservation:Policies, programmesandtheirEffectiveness.Energy&Environment11(5):553 564. Owens,Susan,andRichardCowell,2002.LandandLimits:Interpreting SustainabilityinthePlanningProcess.Routledge,London&NewYork. Panayotou,Theodore,1993.Empiricaltestsandpolicyanalysisof environmentaldegradationatdifferentstagesofeconomicgrowth. InternationalLabourOffice,TechnologyandEmploymentProgramme, WorkingPaper238(WEP22(W.P.238),Geneva. Pascual,Unai,2002.Landuseintensificationpotentialinslashandburn farmingthroughimprovementsintechnicalefficiency.Ecological Economics52(4):497511. Petty,William,1675/1899.PoliticalArithmetick.In:CharlesHull(ed.),The EconomicWritingsofSirWilliamPetty.CambridgeU.Press,Cambridge. Pigou,ArthurCecil,1920.TheEconomicsofWelfare.4thed.1952. Macmillan,London. Pimentel,David,RebeccaHarman,MatthewPacenza,JasonPecarsky,& MarciaPimentel,1994.Naturalresourcesandanoptimumhuman population.PopulationandEnvironment15(5),347369. Prettenthaler,FranzE.,andKarlW.Steininger,1999.Fromownershipto serviceuselifestyle:Thepotentialofcarsharing.EcologicalEconomics28: 443453. Price,B.B.,1998.Raestheoryofthehistoryoftechnologicalchange.In:O.F. Hamouda,C.Lee&D.Mair(eds.),TheEconomicsofJohnRae. Routledge,London. Princen,Thomas,1999.Consumptionandenvironment:Someconceptual issues.EcologicalEconomics31(3):34763. Prognos,2004.AnalysederWirksamkeitvonCO2Minderungsmassnahmen imEnergiebereichundihreWeiterentwicklung.Bundesministeriumfr WirtschaftundArbeit,Berlin. Radetzki,Marian,&JohnE.Tilton,1990.Conceptualandmethodological issues.In:JohnE.Tilton(ed.),WorldMetalDemand:Trendsand Prospects.JohnsHopkinsforResourcesfortheFuture,Baltimore. 105

Rae,John,1834.StatementofSomeNewPrinciplesontheSubjectof PoliticalEconomy,ExposingtheFallaciesoftheSystemofFreeTrade,and ofSomeOtherDoctrinesMaintainedintheWealthofNations.Hilliard, Gray,andCo.,Boston.Reprint(facsimile)1964,AugustusM.Kelley,New York. Raskin,PaulD.,1995.Methodsforestimatingthepopulationcontributionto environmentalchange.EcologicalEconomics15:225233. Reijnders,Lucas,1998.ThefactorXdebate:Settingtargetsforeco efficiency.JournalofIndustrialEcology2(1):1322. Rhee,HaeChun,&HyunSikChung,2006.ChangeinCO2 emissionandits transmissionsBetweenKoreaandJapanusinginternationalinputoutput analysis.EcologicalEconomics58(4):788800. Ricardo,David,1817/1820/1951.OnthePrinciplesofPoliticalEconomyand Taxation.Ed.PierroSraffa.CambridgeU.Press,Cambridge.
nd Robinson,Joan,1956/1965.TheAccumulationofCapital.2 edition. Macmillan,London.

Rood,G.A.,J.P.M.Ros,E.Drissen,&K.Vringer,2003.Astructureofmodels forfutureprojectionsofenvironmentalpressureduetoconsumption. JournalofCleanerProduction11(5):491498. Rpke,Inge,1999.Thedynamicsofwillingnesstoconsume.Ecological Economics28(3):399420. Rosenberg,Nathan,1982.InsidetheBlackBox:TechnologyandEconomics. CambridgeU.Press,Cambridge. Rosenberg,Nathan,1994.ExploringtheBlackBox:Technology,Economics, andHistory.CambridgeU.Press,Cambridge. Roy,Joyashree,2000.Thereboundeffect:Someempiricalevidencefrom India.EnergyPolicy28(67):433438) Rudin,Andrew,2000.Letsstopwastingenergyonefficiencyprograms energyconservationasanoblegoal.EnergyandEnvironment11(5):539 551. SaintPaul,Gilles,1995.Discussion.In:IanGolding&L.AlanWinters(eds.), TheEconomicsofSustainableDevelopment.CambridgeU.Press, Cambridge. Salter,WilfredE.G.,1966.ProductivityandTechnicalChange.CambridgeU. Press.

106

Sanne,Christer,2000.Dealingwithenvironmentalsavingsinadynamical economyhowtostopchasingyourtailinthepursuitofsustainability. EnergyPolicy28(6/7):487495. Saunders,HarryD.,1992.TheKhazzoomBrookespostulateand neoclassicalgrowth.EnergyJournal13(4):131148. Saunders,HarryD.,2000a.Aviewfromthemacroside:Rebound,backfire andKhazzoomBrookes.EnergyPolicy28(6/7):439449. Saunders,HarryD.,2000b.Doespredictedrebounddependondistinguishing betweenenergyandenergyservices?EnergyPolicy28(6/7):497500. Say,JeanBaptiste,[1803]1836.ATreatiseonPoliticalEconomy.Introby MunirQuddus&SalimRashid.Transaction,NewBrunswick,NewJersey andLondon. Say,JeanBaptiste,1820.LettresaM.MalthusSurDiffrnsSujets dconomiePolitique.ChezBossange,Paris. Schipper,Lee,2000.Ontherebound:Theinteractionofenergyefficiency, energyuseandeconomicactivity(Editorial).EnergyPolicy28(6/7):351 353. Schipper,Lee,&StephenMeyers,1992.EnergyEfficiencyandHuman Activity.CambridgeU.Press,NewYork. Schipper,Lee,R.Haas,&C.Sheinbaum,1996.Recenttrendsinresidential energyuseinOECDcountriesandtheirimpactoncarbondioxide emissions:Acomparativeanalysisoftheperiod19731992.Mitigationand AdaptationStrategiesforGlobalChange1(2):167196. Schipper,Lee,&MichaelGrubb,2000.Ontherebound?Feedbacksbetween energyintensitiesandenergyusesinIEAcountries.EnergyPolicy28 (6/7):367388. SchmidtBleeck,Friedrich,1994.WievielUmweltbrauchtderMensch?MIPS DasMassfrkologischesWirtschaften.Birkhuser,Berlin. Schmookler,Jacob,1966.InventionandEconomicGrowth.HarvardU.Press, Cambridge,Mass. Schor,Juliet,1992.TheOverworkedAmerican:TheUnexpectedDeclineof Leisure.BasicBooks,N.Y. Schor,Juliet,1999.TheOverspentAmerican:Upscaling,Downshifting,and theNewConsumer.BasicBooks,N.Y. Schumpeter,JosephA.,1911[1912].Theoriederwirtschaftlichen Entwicklung.Duncker&Humblot,Leipzig. 107

Schurr,SamH.,&BruceC.Netschert,1960.EnergyintheAmerican Economy,18501975.JohnsHopkins,Baltimore. Schurr,SamH.,1982.Energyefficiencyandproductiveefficiency:Some thoughtsbasedonAmericanexperience.EnergyJournal3(3),314. Schurr,Sam,1985.Energyconservationandproductivitygrowth:Canwe haveboth?EnergyPolicy13(2),126132. Sebold,FrederickD.,&EricW.Fox,1985.Realizedsavingsfromresidential consumptionactivity.EnergyJournal6(2),7388. Selden,ThomasM.,&DaqingSong,1994.Environmentalqualityand development:IstheraKuznetscurveforairpollutionemissions?Journalof EnvironmentalEconomicsandManagement27:147162. Shafik,Nemat,1994.Economicdevelopmentandtheenvironmentalquality: AneconometricAnalysis.OxfordEconomicPapersNewSeriesVol.46, Oct. Shafik,Nemat,&S.Bandyopadhyay,1997.Economicgrowthand environmentalresources.WorldDevelopmentReport.WorldBank, Washington,D.C. Shi,Anqing,2003.Theimpactofpopulationpressureonglobalcarbon dioxideemissions,19751996:evidencefrompooledcrosscountrydata. EcologicalEconomics44,2942. Sieferle,RolfPeter,1982/2001.TheSubterraneanForest:EnergySystems andtheIndustrialRevolution.WhiteHorsePress,Cambridge. Simms,Andrew,2005.EcologicalDebt.TheHealthofthePlanetandthe WealthofNations.Pluto,London. Smil,Vaclav,2003.EnergyattheCrossroads:GlobalPerspectivesand Uncertainties.MITPress,Cambridge,Massachusetts. Smith,Adam,1776/1976.AnInquiryintotheNatureandCausesofthe WealthofNations.Eds.R.H.Campbell,A.S.Skinner&W.B.Todd. ClarendonPress,Oxford. Solow,Robert,1957.Technologicalchangeandtheaggregateproduction function.(ReviewofEconomicsandStatistics39:31220. Solow,Robert,1970.GrowthTheory:AnExposition.ClarendenPress, Oxford. Starkey,Richard,&KevinAnderson,2004.Domestictradablequotas:A policyinstrumentforreducinggreenhousegasemissionsfromenergyuse. TechnicalReport38,December2005, www.tyndall.ac.uk/research/theme2/final_report/t3_22.pdf

108

Stewart,RichardB.,2000.Economicincentivesforenvironmentalprotection: Opportunitiesandobstacles.In:RLRevesz,PSandsandRBStewart (eds.),EnvironmentalLaw,TheEconomyandSustainableDevelopment. CambridgeU.Press,Cambridge. Swaney,JamesA.,1991.JulianSimonvs.theEhrlichs.JournalofEconomic Issues25:499509. SwissEnergy,September2004.PartnerfortheClimate:AnnualReportof SwissEnergy2003/04.EidgenssischesDepartementfrUmwelt,Verkehr, EnergieundKommunikationUVEKandBundesamtfrEnergie,Bern. Torras,Mariano,&JamesK.Boyce,1998.Income,inequality,andpollution: AreassessmentofTheenvironmentalKuznetscurve.Ecological Economics25(2):147160. UKERC(UKEnergyResearchCentre)2005.ScopingNote:Anassessment ofevidenceforareboundeffectfromimprovementinenergyefficiency, SteveSorrellandJohnDimitropoulos,SPRU,October2005. UKERC2006.TheReboundeffect:Microeconomicdefinitions,extensions andlimitations,JohnDimitropoulos&SteveSorrell,SPRU,WorkingPaper April2006. UNFCCC,1997.UNFrameworkConventiononClimateChange,Kyoto Protocol.InternationalLegalMaterials37(1998):3243. vandenBergh,JeroenC.J.M.,1999.Materials,capital,direct/indirect substitution,andmassbalanceproductionfunctions.LandEconomics75 (4):547561. vandenBergh,J.C.J.M,&H.Verbruggen,1999.Spatialsustainability,trade andindicators:Anevaluationoftheecologicalfootprint.Ecological Economics29(1):6172. Veblen,Thorstein,1899/1998.TheTheoryoftheLeisureClass.Prometheus Books,Amherst,N.Y. Victor,PeterA.,1991.Indicatorsofsustainabledevelopment:Somelessons fromcapitaltheory.EcologicalEconomics4(3):191213. VonWeizscker,Ernst,AmoryB.Lovins,&L.HunterLovins,1997.Factor Four:DoublingWealthHalvingResourceUse.Earthscan,London. Vringer,Kees,&KornelisBlok,2000.Longtermtrendsindirectandindirect householdenergyintensities:Afactorindematerialisation?EnergyPolicy 28(10):713727.

109

Wackernagel,Mathis,&WilliamRees,1996.OurEcologicalFootprint: ReducingHumanImpactontheEarth.NewSociety,GabriolaIsland,D.C. Wackernagel,Mathis,&WilliamRees,1997.Perpetualandstructuralbarriers toinvestinginnaturalcapital:Economicsfromanecologicalfootprint perspective.EcologicalEconomics20(3):324. Weisz,Helga,FridolinKrausmann,ChristofAmann,NinaEisenmenger,Karl HeinzErb,KlausHubacek,&MarinaFischerKowalski,2006.Thephysical economyoftheEuropeanUnion:Crosscountrycomparisonand determinantsofmaterialconsumption.EcologicalEconomics58(4):676 698. Wirl,Franz,1997.TheEconomicsofConservationPrograms.Kluwer Academic,Boston. York,Richard,EugeneA.Rosa,&ThomasDietz,2003.STIRPAT,IPATand ImPACT:Analytictoolsforunpackingthedrivingforcesofenvironmental impacts.EcologicalEconomics46:35165.

110