Cryptodamages: Monetary value estimates of the air pollution and human health impacts of cryptocurrency mining

9 Pages·2020·1.759 MB·English

by  Andrew L. Goodkind, Benjamin A. Jones, Robert P. Berrens

Energy Research & Social Science 59 (2020) 101281
ContentslistsavailableatScienceDirect
Energy Research & Social Science
journal homepage: www.elsevier.com/locate/erss
Cryptodamages: Monetary value estimates of the air pollution and human
T
health impacts of cryptocurrency mining
⁎
Andrew L. Goodkind, Benjamin A. Jones , Robert P. Berrens
UniversityofNewMexico,Albuquerque,NM,UnitedStates
ARTICLE INFO ABSTRACT
Keywords: Cryptocurrencyminingusessignificantamountsofenergyaspartoftheproof-of-worktime-stampingschemeto
Cryptocurrencies addnewblockstothechain.Expandinguponpreviouslycalculatedenergyusepatternsforminingfourpro-
Blockchain minentcryptocurrencies(Bitcoin,Ethereum,Litecoin,andMonero),weestimatethepercoineconomicdamages
Energyuse ofairpollutionemissionsandassociatedhumanmortalityandclimateimpactsofminingthesecryptocurrencies
Airpollution
intheUSandChina.Resultsindicatethatin2018,each$1ofBitcoinvaluecreatedwasresponsiblefor$0.49in
Humanhealth
healthandclimatedamagesintheUSand$0.37inChina.ThesimilarvalueinChinarelativetotheUSoccurs
Bitcoin
despitetheextremelylargedisparitybetweenthevalueofastatisticallifeestimatefortheUSrelativetothatof
China.Further,witheachcryptocurrency,therisingelectricityrequirementstoproduceasinglecoincanleadto
an almost inevitable cliff of negative net social benefits, absent perpetual price increases. For example, in
December2018,ourresultsillustrateacase(forBitcoin)wherethehealthandclimatechange“cryptodamages”
roughlymatcheach$1ofcoinvaluecreated.Weclosewithdiscussionofpolicyimplications.
1. Introductionandbackground BeginningwithBitcoinin2009(attributedtotheelusiveNakamoto
[7]),therearenowmorethan2500cryptocurrencies,whichgenerally
As an anonymized and decentralized production process, digital share a reliance on the blockchain technology. In brief, blockchain is
currencies using cryptographic blockchain verification for securing commonly referred to as a decentralized, public ledger that uniquely
transactionsareseenasatransformativetechnologythatcircumvents identifies all transactions using any particular cryptocurrency; the
traditionalbankingandgovernmentalregulatorymechanismsofmore blockchain contains the blocks of digital information that record bat-
centralizedfiatcurrencies([1];andseereviewin[2]).Forproponents, ches of transactions (with dates, times, amounts, participants [with
cryptocurrenciesgenerateimportantsocialbenefitsforsecurelytrans- uniquedigitalsignatures,ratherthannamesoridentifyinginformation]
ferringmoneyandinformation,aswellasservingaspotentialstoresof andauniquetransactionidentifierorhash),1inadistributedpeer-to-
wealth.Thelibertarian,anti-establishmentappealisrepresentedinthe peer network of computers. Independent individuals or groups of mi-
independent production process (“mining”), which is independent of nerscompeteusingbruteforcecomputingpowertobethefirsttosolve
any government, regulatory entity, or particular geographic locality. complexalgorithmsandprovidethehashidentifierforablock,andif
Originally,allthatwasnecessarytominewasaccesstoelectricity(the successful, then are rewarded with units of the cryptocurrency. Data
cheaperthebetter),computerhardware(withsomespareCPUcycles), fromnew,verifiedtransactionsarethenaddedtotheblockchain.While
an internet/network connection (the faster the better), and a heavy alternativesexist,cryptocurrencyapplicationsusingblockchainremain
dose of entrepreneurial spirit [1]. But, for many prominent crypto- dominated by the proof-of-work (POW) process used in the original
currencies,opportunitiesforprofitableminingcurrentlyappeartore- Bitcoin,wheretheprobabilityofsuccessfulminingisincreasedbythe
quirehighly dedicatedorspecializedhardware concentratedinlarge- amountofcomputingworkexpended.
scaleminingfarmsorpools[1,3,4,5,6]. Thus,mininggeneratesfinancialvalue,butconsumeselectricityin
⁎Correspondingauthor.
E-mailaddress:[email protected](B.A.Jones).
1Ahashfunction,ascommonlyusedincryptography,generallyreferstousingmathematicalalgorithmstomaporconvertsomearbitrarilylongstringofnumbers
andlettersintosomeuniqueoutput(orhash)offixedsize.
https://doi.org/10.1016/j.erss.2019.101281
Received20March2019;Receivedinrevisedform23August2019;Accepted1September2019
Available online 24 September 2019
2214-6296/ © 2019 Elsevier Ltd. All rights reserved.
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
doing so. The complication is that the supply of any cryptocurrency number of coins mined dropped to 700,000, but electricity consump-
coinistypicallyfiniteandmadeavailableaccordingtoprescribedrules tion increased to 47.9 billion kWh. This usagecreates negative social
thatasymptoticallyapproachsomefixedamountataspecifiedpointin externalities,mostsignificantlybycontributingtoclimatechangeand
time.Specifically,asthesupplyofnewcoinsslows,theimplicationofa impactinghumanhealthfromtheburningoffossilfuels.Itwasrecently
POWprocessisthatthecompetingcomputingefforttominecoinsmust arguedthatCO2emissionsfromBitcoinminingalonecouldpushglobal
necessarily increase, thus requiring ever increasing amounts of elec- warming above the 2°C threshold of concern [13]. Economists use a
tricity.Thisprocessamountstosequentialroundsofawinner-take-all batteryoftechniquestoestimatethemonetarydamagesconnectedto
tournament (akin theoretically to R&D tournaments [8]), where the these non-market negative externalities, which by definition are not
costsassociatedwithwinningeachroundprogressivelyincrease.Thus, accounted for in the market production or consumption of a good or
the “strange math” [6] of cryptocurrency provision based on a POW service (see [14]). In this case, the health and climate impacts of
process generates intense electricity resource use [5,8,9], potentially cryptocurrencymining,whichwerefertohereas“cryptodamages.”
creatingnegative—andgrowing—environmentalandhealthcoststhat Rather than attempt a global damage estimate from overall pro-
maybehighandarenotbornebytheminers[10].Ourfocushereison ductionofcryptocurrenciesunderthebroadestofassumptionsforwhat
beginningtomonetizetheseelectricity-relatedsocialcosts. areextremelydynamicandvolatilemarkets,wemakeaninitialattempt
As with any emergent technology, there needs to be careful con- atamorerefinedapproach,usingfourprominentPOW-processbased
sideration of its environmental and health impacts on society. In the coinsandtwocountries.Specifically,givenavailableinformationcon-
emergingliteratureconsideringtheseimpacts,KrauseandTolaymat[1] straints, we examine these externalities on a per-coin generated basis,
pushsuchassessmentssignificantlyforwardbyquantifyingtheenergy where we have more detailed information about the spatial and tem-
and carbon emissions for mining four prominent cryptocurrencies poral characteristics of the required electricity generation. Spatially,
(Bitcoin(BTC),Ethereum(ETH),Litecoin(LTE),andMonero(XMR),all emission rates for CO2 and regional air pollutants from electricity
identified as using a POW process). They pursue the following ques- generationvarysubstantially,basedonthefuelsusedandthegenera-
tions: (i)dothesecryptocurrenciesrequireasimilarenergysupplyto tion methods. Human health effects from regional air pollutants also
function?;(ii)whatconventionalprocessesorserviceswould“crypto- havelargespatialvariabilitydependingontheproximityofpopulations
mining”compareto(e.g.,goldmining),intermsofenergyinvestedand toemissionsources.Thus,toaccountforthespatialvariabilityofcoin
value extracted?; and (iii) what carbon impacts might this energy generation,wecalculateexternalitiesonaper-coin-basisforelectricity
consumptiongenerate?TheyfindthatcryptominingofBTC,ETH,LTC, generationintheUSandChina,whicharetwodominantplayersinthe
and XMR tends to consume more energy than traditional mineral cryptocurrency network, and which have very different electricity
mining such as copper, gold, platinum metals, and rare earth metals productionandairpollutionemissionprofiles.3
(with the exception of aluminum, which has high electricity con-
sumption) in producing an equivalent market value. Their results ad-
ditionallyindicatethatenergyconsumptionrequirementsaregenerally 2. Methods
expected to increase, for reasons previously discussed, and that BTC
consumedmoreelectricitythanIreland(26TWhyr−1)orHongKong WestartbycollectingdataonemissionratesperkWhofelectricity
(44 TWh yr−1) in 2017. Finally, for the 2.5-year period (January 1, generation by country (US and China) for four pollutants commonly
2016 to June 30, 2018), they estimate that the four prominent cryp- createdbyburningfossilfuelstoproduceenergy:carbondioxide(CO2),
etomciusrsrieonncsi.eMsowreerreecreenstplyo,nSsitbolleleftoarl.3[1–1]1,5usimngillIiPo-nadtdornensesess(tto)aotfteCmOp2t ofixniedepa(SrtOic2u)la[1te6,m17a,t1te8r].4(PNMe2x.5t,),wneitcroogmebninoexitdheesse(NemOxis)siaonndrsautelfsurwidtih-
tolocatemining,estimatesubstantiallylargerCO2emissionsassociated thekWhofelectricityusagepercoincreated[1].5Together,thispro-
withBTC,inexcessof20milliontperyear. vides the average emissions released (from all energy sources; fossil
Inthisresearch,weextendthediscussioninKrauseandTolaymat fuelsandrenewables)togenerateonecoin.CO2emissionsareusedto
[1]toaskthenextlogicalquestion:“Whataretheeconomicdamages calculatetheclimatedamagespercoincreated.Exposuretopollution
connected to air pollution, health and climate impacts of mining associated with the other three pollutants (PM2.5, NOx and SO2) in-
cryptocurrencies?”TakingtheKrauseandTolaymat[1]resultsforthe creases the risk of premature mortality [20,21]. Hence, we use these
US and China, we use commonly available simulation models, ex- emissionstocalculatethemortalityimpactofcreatingonecoinineach
posure-responsefunctions,andavailableestimatesofthesocialcostsof country,andthemonetarydamagesoftheprematuredeaths.6
carbon, and value of statistical life (VSL) measures to monetize the Electricity requirements to produce one coin for each day are de-
negativeenvironmentalexternalitiesofairpollutionandhumanhealth rivedfromthenetworkhashrateoftheblockchain,apubliclyavailable
damages.Wefocusonthehealthandclimatedamagesassociatedwith figure [1,22]. The hashrate is a measuring unit, which identifies the
mining four coins over the period 2016 – 2018; the same four coins amountofpowerconsumedbyalltheoperationsinacryptocurrency
considered inKrause andTolaymat[1].Thesefourcoins—BTC, ETH,
LTCandXMR—accountedfor57.2percentofthe$369billionintotal 3Forcontext,asofJune2018,over80%ofBitcoinminingwasperformedby
marketcapitalizationfor1658listedcryptocurrenciesasofMarch15,
sixminingpools,andfiveofthosesixpoolsaremanaged(butnotnecessarily
2018;BTCaloneaccountedfor37.8%[12].2 physicallylocated)inChina[15].IntheUS,cryptocurrencyminingisasig-
Theelectricityconsumptionofminingcryptocurrenciesislargeand nificantsourceofcontroversy,partiallyrelatedtoitsoffsettingeffectsonfossil
growing rapidly. For example, in January of 2016, each BTC mined fueluseandassociatedclimatechangeimpacts(e.g.,[6]).
required 1005kWh ofelectricity; but by June 2018, each coin mined 4Thereareseveralotherfossilfuelpowerplantpollutantsthatarenotcon-
required60,461kWh[1].In2016therewere∼1millionBTCmined, sideredhere,butwhichmayaffecthumanhealth,includingmercury,cadmium,
which consumed 2.5 billion kWh of electricity; in 2018 the total chromium,andnickel.Datalimitationsprecludetheirinclusionhere.Thus,our
estimatesarelikelyalowerboundonactualdamages.
5ThedatainKrauseandTolaymat[1]stopatJune30,2018,sowecollected
originaldataonelectricityusageandcoinsgeneratedtoextendthedatasetfrom
2Forperspective,asofapproximatelyoneyearlater(March2,2019),total July1,2018toDecember31,2018.Theseextendeddataareavailablefrom
market capitalization had fallen to $130.7 billion, but the number of listed https://bitinfocharts.com/[19]forthenetworkhashrate,blocktime,andcoin
cryptocurrencieshasgrownto2526,withBTC,ETH,LTCandXMRaccounting price.
for65.8percentoftotalmarketcapitalization;BTCaloneaccountedfor50.4 6Thereareotherexternalitiesassociatedwiththeseemissionsthatarenot
percent (for current estimates see Investing.com at: https://www.investing. includedinourdamagesestimates.Forinstance,SO2emissionscontributesto
com/crypto/currencies). acidrainandNOxemissionscontributetoground-levelozone.
2
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
networktomineblocksandreceivethecurrencyreward.Statedsimply, electricitygenerationsectordividedbythenetelectricitygeneration
“hashrates are thenumberof calculations(hash functions)performed [29]. For both countries, we did not want to exclude non-emitting
onthenetworkinseconds”([1],p.711).Hashratesincreasewiththe electricitygeneration,ratherwewanttheemissionprofileofaunitof
intensityofthecomputationalcompetition. electricity from all sources in each country. The choice of pollutant
The great unanswered question faced when exploring cryptoda- species is based on their high rates of emissions from electricity
mages is that while we can identify select geographic hotspots of generation,andtheirknownimpacttoprematuremortality.Exposure
production we currently do not know in the aggregate where the to PM2.5 is associated with an increased risk of adult-premature
electricityusedincryptocurrencyminingisphysicallyproduced.This mortality[20,21].PM2.5airpollutionconsistsofprimaryPM2.5and
is because we, like Krause and Tolaymat [1], do not know the phy- secondary species: SO2 and NOx are PM2.5 precursor emissions that
sicallocationsofcryptocurrencyminers,whetherindividuals,groups formintosecondaryPM2.5intheatmosphere.Weestimatethemor-
oraggregatesin,say,aregionorcountry.Thereisconsiderableevi- tality impacts of SO2 and NOx from their contributions to PM2.5
dence of concentration of mining operations in particular locations, concentrations.
typicallywherereliableelectricityischeaplyavailable,though,pre- Calculating human exposure to PM2.5 from electricity emissions
cisedataarelacking(e.g.,[6,15]).IntheUS,perhapsthemostwell- (step(ii)above)iscomplicatedandthedataarenotreadilyavailable
known concentration is the Mid-Columbia Basin area in central and for China. We do, however, have detailed estimates of these re-
eastern Washington State, where cheap electricity is produced by lationships for US emissions. We use this US data to estimate the
hydropower along the Columbia River [4,6], however, mining in relationship for China, which we explain in detail below. First, we
other US locations also occurs (see [11] and news references in describetherelationshipbetweenelectricityemissionsandexposures
[23,24], and [25]). Thereis also evidence of large mining campsin intheUS.TransportofemissionsofeachpollutantfromeveryEGUin
China [15]. With time and emergent research (e.g., see [11]) there the contiguous US is modeled using the Intervention Model for Air
may be improved information about the amounts of electricity de- Pollution (InMAP) [30,31]. InMAP estimates the change in PM2.5
votedtominingcryptocurrenciesforparticularlocationsorregions, concentrations from a unit of emissions. Changes in PM2.5 con-
butitiscurrentlynotavailable. centrations are translated into changes in the risk of premature
Therefore,weaimtocharacterizetheexternalitiesassociatedwith mortalityusingacommonlyadoptedconcentration-responsefunction
the production of a single coin in both the US and China, where [20]. Combining these results with emissions per unit of electricity
emissionsratesperkWhwilldiffersubstantiallygivenunderlyingdif- forthethreepollutants,weestimatethemortalityimpactperkWhof
ferences in how power is produced—i.e., China relies extensively on electricity generation at each EGU. The sum of the impact for the
coal power (>60% of electricity generation) while the US is more three pollutants is the total mortality impact per kWh of electricity
balanced(32%naturalgas,30%coal,20%nuclear,etc.).Emissionrates generated at every EGU in the US. We then allocate the emission
ofCO2perkWhofelectricityproducedin2016wereobtained7[17,18] profileofelectricityatEGUstolocations(orgridcellsinthemodel)
andcombinedwithelectricityusagepercointoproduceCO2emissions wherepeopleconsumeelectricity.9Theemissionprofileinagridcell
per coin created. Then, we convert CO2 emissions into estimated cli- isbasedonaweightedaverageofallEGUswithin250kmofthecell
matedamagesusingtheUSFederalGovernment'ssocialcostofcarbon centroid, weighted by the inverse distance to the EGU and the
(SCC)for2020 emissionsandassuming a3% discountrate[26].The quantity of electricity generated (see Fig. 1). Combining the health
SCCisestimatedat$51t−1in2018USDollars.8 damageswithclimatedamagesperkWh,weestimatethatone-third
Estimating mortality impacts of emissions from cryptocurrency of electricity in the US is produced with less than 1 cent kWh−1 of
generation is more complicated. The general steps to convert emis- damages, and one-tenth of electricity is produced with greater than
sionsinto mortality impacts,and ultimatelyinto monetarydamages 10centskWh−1ofdamages.
areasfollows:(i)datafromemissionratesfromelectricitygeneration We use the mortality impact per unit of emissions in the US to
are collected; (ii) human exposures to the pollutants are estimated; estimate the mortality impacts in China. Crucially, the impact of
(iii)exposureisconvertedintomortalityimpactsusingexposure-re- emissionsisstronglyrelatedtothesizeofthepopulationinthearea
sponse functions, and; (iv) premature mortality is converted into surrounding the emission source [32,33]. We relate the population
monetarydamagesusingthevalueofastatisticallife(VSL).For(i), densityina50kmradiusaroundeachemissionsourceintheUSwith
wecollectdatafortheUS[16,17]andChina[18].FortheUS,these themortalityimpactofthoseemissions(seeFig.2).Weapplythese
dataareemissionsofPM2.5,NOx,andSO2andelectricitygeneration relationships to China based on its unique population density to es-
foreveryelectricitygeneratingunit(EGU)inthecountry.ForChina, timatetheaveragemortalityimpactperunitofemissions.Whilewe
we collect aggregate data of emissions of these pollutants from the know the population density of both countries, this is not the ap-
propriate value to apply to the estimated functions. The population
density around EGU emission sources in the US is substantially
greater, onaverage,thanthepopulationdensityofthecountryasa
me7tUhSaneeleacntdricnitiytroeumsisosxiiodnes. aCrOe2inemCisOsi2onesquciovnasletintutste(C9O9.24e%), owfhCicOh2einfcolrudUeSs wsiohnoslei.sF8o8r pexearsmopnlse,kmth−e2p,ocpoumlaptairoenddwenitshity37arpoeurnsodnEsGkUm−SO22foermthise-
electricitygeneration.
8Thesocialcostofcarbon(SCC)isapresent-valueddollarmeasureofthe contiguous US. We use the ratio of each pollutant to population
long-term damages caused per t by carbon dioxide (CO2) emissions into the density tothe contiguousUS populationdensity toadjust thepopu-
atmosphere.Atopicofconsiderablepolicydebate,SCCestimatesareusedby lation density of emission sources in China. The SO2 population
theUSEPAandotherUSgovernmentbodiesincost-benefitanalysesofreg- density ratio is 2.9, which when applied to China, with an overall
ulatoryactions(e.g.,forclimatechangeimpacts)[27].Forthisanalysis,weuse population density of 145 persons km−2, yields an estimated “SO2”
themostrecentUSInteragencyWorkingGroup(2016)SCCestimateproduced population density for China of 346 persons km−2. We apply this
duringtheObamaAdministration,andconvertedinto2018values.TheInter- population-density value to the mortality-impact relationship (best-
agencyWorkingGroupalsoproducedestimatesusinghigher(5%)andlower
fitlinesinFig.2).Thisprocessproducesanestimateofthemortality
(2.5%)discountrateswhichsubstantiallyimpactthemagnitudeoftheSCC.It
pertonneofemissionsforeachpollutantfromelectricitygeneration
should be noted that in 2017 the Trump Administration altered the metho-
dologyusedincalculatingSCCtoproduceestimatesinthe$1to$6ranget−1of in China. Our estimate of the ratio of mortality impacts per unit of
CO2emissions,whichisdrasticallylowerthanestimatesusedhere.However,
othersourceshaverecentlyarguedthatSCCestimatesshouldbesignificantly
higher than those used here (e.g., see [27]), due to significantly under- 9TheInMAPmodelhasvariably-sizedgridcellsbasedonpopulationdensity.
estimatingtheeffectsonagriculture[28]. Thegridcellsrangefrom48km2inruralareasto1km2indenseurbanareas.
3
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
Climate damages Health damages
Damages ($/kWh)
0 0.02 0.04 0.06 0.08 0.10
Fig.1.ClimateandhealthdamagesperkWhofelectricityusedineachlocation.Theemissionprofileofelectricityuseineachlocationisaweightedaverageofthe
EGUswithin250km,weightedbytheinversedistancefromtheEGUandthequantityofelectricitygenerated.
Fig.2.RelationshipbetweenUSSO2,NOx,andPM2.5electricityemissionmortalityimpactst−1andpopulationdensityin50kmradiusaroundemissionsource.
Dashedlinesarebest-fitestimatesofthedata.
PM2.5fromEGUsbetweenChinaandtheUS(ratio:1.8)issimilarto risks of mortality and compensation.10 Multiplying total climate and
theratioestimatedbyApteetal.[32](ratio:2.1)forurbanemissions. health damages per kWh of electricity generation by the electricity
WecombinetheseresultswiththequantityofemissionsperkWhof requirementstoproduceonecryptocurrencycoin,producesanestimate
electricitytogetmortalityimpactsperunitofelectricity. ofthesocialdamagesorcryptodamagespercoingenerated.
Next, we convert mortality and climate impacts per unit of elec- Usingthesedamageestimates,fortheUS,wecalculatethenetsocial
tricity into monetary damages. Emissions of CO2 are converted into value of producing each cryptocurrency, for each day in our dataset
climate damages by multiplying emissions by the SCC ($51 t−1). (January1,2016toDecember31,2018),andforanylocationwhere
Mortality impacts in the US are converted into health damages by
multiplying estimated mortality by the EPA's recommended VSL [34]
(adjustingforinflationandaverageincometo2018,theVSLis$11.53 10ForacomprehensivereviewoftheVSLliterature,andthehistoryofusein
million).FormortalityimpactsinChina,wecalculatehealthdamages regulatoryarenasintheUSandelsewhere,seeViscusi[36].Viscusi[36]argues
thatmorewidespreaduseofVSLscouldcreateasaferandmoreequitableso-
usingaChina-specificestimateoftheVSL[35](adjustingforinflation
ciety,whilealsorecognizingthatrelativetotheUSothercountriestypically
to 2018in USDollars, theVSL is$1.12 million). With the use ofdif-
undervalue the risks to lives, even after accounting for income differences.
ferentVSLsineachcountry,wearemakingnovaluejudgementabout
Viscusi[37]recentlyarguesthatinternationalestimatesoftheincomeelasti-
therelativeworthofahumanlifebetweentheUSandChina.Rather,to
citiesofVSLscouldbeusedforimprovedcalibrationandtransferofUSesti-
produceamonetaryestimateofthedamagesweusecurrentlyavailable matestoothercountries,afteraccountingforincomedifferences.Forexample,
peer-reviewed studies of the observed and stated tradeoffs between doingsoforAustraliawouldmorethandoublethecurrentbest-practicesVSL
estimate[37].
4
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
electricity is consumed. We define the net social value as the private 3. Results
benefitsandcostsofminingacryptocurrencylessthesocialdamagesof
consuming the electricity. Specifically, the net social value of mining With the above information, we calculate the daily mortality and
onecoinisthepriceofthecoinattimet,minusthecostofelectricityin climate impacts and associated monetary damages of producing one
locationsattimetminusthehealthandclimateexternalitiesofelec- coinforeachofthefourcryptocurrencies(BTC,ETH,LTC,andXMR)in
tricityinlocationsattimet: theUSandChina(Table1).Then,dividingbythemarketpriceofeach
cryptocurrency,wecalculatethehumanhealth,climate,andtotalda-
Net socialvalues,t=coinpricet (electricityprices,t×electricityper coint) externalityper coins,t (1)
Table1
Mortalityimpacts,climatedamages,andhealthdamagesofcoinminingcreatedbycountry,year,andcryptocurrency.
Mortalitypermillioncoins Climatedamages($percoin) Healthdamages($percoin) Damages(%ofcoinvalue) Globalcoinsmined(millions)
USA China USA China USAa Chinab USA China
BTC 2016 4.6 9.6 74 86 53 11 21% 16% 1.0
2017 19 40 311 359 222 45 19% 14% 0.70
2018 115 239 1849 2135 1321 268 49% 37% 0.68
ETH 2016 0.03 0.06 0.49 0.57 0.35 0.07 9% 7% 10.8
2017 0.59 1.2 9.5 11 6.8 1.4 8% 6% 8.7
2018 2.5 5.2 40 46 29 5.8 21% 16% 6.6
LTC 2016 0.01 0.02 0.15 0.17 0.10 0.02 7% 5% 5.3
2017 0.08 0.16 1.22 1.4 0.87 0.18 5% 4% 5.4
2018 0.94 2.0 15.2 18 11 2.2 37% 28% 5.3
XMR 2016 0.02 0.05 0.35 0.41 0.25 0.05 27% 21% 3.1
2017 0.19 0.39 3.0 3.5 2.2 0.44 8% 6% 1.9
2018 1.0 2.1 17 19 12 2.4 22% 17% 1.1
Notes:BTC=Bitcoin,ETH=Ethereum,LTC=Litecoin,XMR=Monero.MortalityimpactsassociatedwithpowerplantemissionsofPM2.5,SO2,andNOxdueto
countryandyear-specificcryptocurrencyminingactivity.Alldamagesarein2018USDollars.
a HealthdamagesinUScalculatedwith$11.53millionVSL.
b HealthdamagesinChinacalculatedwith$1.12millionVSL.
Fig.3.HealthandclimatedamagesfromproducingonecoinintheUSasapercentageofthevalueofminingonecoinfromJanuary1,2016toDecember31,2018.
Notes:BTC=Bitcoin,ETH=Ethereum,LTC=Litecoin,XMR=Monero.
5
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
Fig.4.NetsocialvalueofproducingoneBitcoin(BTC)acrosstime(betweenJanuary1,2017andDecember31,2018)andspace.Panelsa-dshowthenetvalueof
usingelectricityinanylocation(gridcell)intheUStoproduceoneBTCforthefollowingdates,respectively:(a)March1,2017,(b)November15,2017,(c)March
15,2018,and(d)July1,2018.
magesperdollarvalueofacoinmined.Fig.3illustratesthechanging generation. Thus, Fig. 4 examines the net social value of mining a
profile of total damages from cryptocurrency mining relative to the cryptocurrencyinanylocationintheUS,accountingforthebenefitand
price of a coin over time in the US. The volatility of this measure is costofgeneratingacoinandbuyingelectricityfortheminer,respec-
related to two factors: the price of coins and the electricity require- tively, and the externality costs of the electricity generation in that
mentstogeneratecoins.Atthebeginningofourtimeframe,bothprice locationbornebysociety(seeEq.(1)).Fig.4focusesjustonBTC,for
andelectricityrequirementswerelow,withdamagesaveraging21%of simplicity.Itshouldbenotedthisisthemaximumnetsocialvalue,aswe
the value of a BTC in 2016. Then the price surged (topping out at areexcludingall otherpotentialcostsofcryptocurrencymining (e.g.,
$19,345 for BTC on December 16, 2017), reducing the damages to equipment, cooling, or opportunity costs). Fig. 4 shows the price and
under 10% of the coin value. Since this date, the price of these four electricity requirements of BTC across time, and net social value of
prominent cryptocurrencies crashed but the electricity requirements generating a coin spatially on four specific dates: (a) March 1, 2017,
rapidlyincreased.AsofDecember31,2018,thedamagesfromaBTC when price and electricity consumption were low, and there were
minedintheUSwereestimatedtobe95%ofthevalueofthecoin[BTC modestnetbenefitsacrossseveralregionsoftheUS;(b)November15,
priceonDecember31,2018:$3747;damagespercoinmined:$3551 2017,whenpricebegantosurgeandelectricityrequirementsremained
($1480frommortalityimpacts,$2071fromclimateimpacts)].11While relatively low, leading to large net benefits to mining in almost any
the damages as a share of coin value were highest for BTC in 2018, region; (c) March 15, 2018, when the price started to drop and elec-
producingoneoftheotherthreecoins(ETH,LTC,XMR)createdsub- tricityconsumptionincreased,wherethereexistedalargespatialdis-
stantial damages, between 21 and 37% of the coin's value. For each tribution of net benefits, with positive regions in the Northwest and
cryptocurrency,thisdamageratioincreasedbetween2017and2018. Southeast,andnegativeregionsintheMidwest;and(d)July1,2018,
Fig. 3 shows the damages of mining a coin in the US using the whenthepricedroppedandelectricityconsumptionincreasedfurther,
average electricity mix; however, the price and damages from elec- withsubstantialnegativenetbenefitsinalmostallregionsoftheUS.
tricitygenerationwillvarysubstantiallybylocationandfuelsourcefor Giventhemagnitudeofthecryptodamagesandthefactthatcryp-
tocurrency miners do not bear the full cost of the negative environ-
mental externalities from their activities (just the price of electricity)
11During December 2018, there were several days when the damages ex- they do not receive the correct market signals regarding when and
ceededthevalueofthecoinsgenerated,asseeninFigure3.
6
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
where mining can be pursued at the minimum cost to society. cryptodamages—thehumanhealthandclimateimpacts—willeventually
Presumably,minersseeklocationswiththelowestcostelectricity.We exceedeach$1valuecreated.Theaboveestimates($0.49intheUSand
estimatethatthecorrelationbetweenelectricitypriceanddamagesper $0.37inChina)areaveragedovertheyear(2018);but,notablyaver-
kWh of electricity generated (weighted by net generation) is 0.18, agedacrossDecember2018weobservedeach$1ofBTCvaluecreated,
suggesting that the lowest priced electricity locations will generate generated $0.95 of cryptodamages in the US. For any cryptocurrency
slightlyhigherthanaveragedamages.Ifminersaremorelikelytolo- tied to the POW process (or something similar), the rising electricity
cate in above-average damage per kWh locations, then our estimates requirements to produce a single coin leads to a situation where the
using the average electricity mix (in Fig. 3) will be slight under- price must continue rising, faster than the social costs, to maintain
estimatesofthecryptodamages. positivenetbenefitsforsociety.Withoutperpetualpriceincreases,coin
Available evidence indicates that a substantial share of crypto- mining may follow an almost inevitable cliff of negative net social
currencyminingoccursinChina[15].Weestimatethatthedamagesof benefitsastheenergyuserequiredforminingincreasesbygreaterand
miningcoinsinChinawillbe,onaverage,slightlysmallerthaninthe greateramountsduetothePOWprocess.
US.Table1showsthatfor2018,damagesfromaBTCcoingeneratedin Inexploringthesocialcostsideofcryptocurrencymining,ouranalysis
the US were 49% of the coin value, and the damages from a coin isrestrictedinsomeways.Forexample,theremaybesignificantnegative
generated in China were 37% of the coin value. Emission rates (CO2 communityeffectsconnectedtominingboomtowns(e.g.,see[4])thatare
kWh−115%andSO2kWh−116%higher)andmortalityt−1ofemis- notaccountedforhere.Ouranalysisislimitedtohealthandclimateim-
sions (SO2 mortality t−1 39% higher) are higher in China compared pactsconnectedtotwocountries,andfourprominentcryptocurrencies,so
withtheUS,however,theVSLthatweuseinouranalysisisestimated inthatwayitmaybeasignificantunderestimateoftheemergentunder-
to be 90% lower. Combining these factors, the climate impacts from standing of cryptocurrency externalities. Related, our results are likely a
Chinacryptocurrencyminingconstitutealargershareoftotaldamages lower-boundonactualhealthdamagessinceotherpowerplantpollutants
(89%fromclimate,11%fromhumanhealtheffects),comparedwithUS suchasmercuryandcadmium(producedbycoal-firedpowerplants),are
damages where approximately 60% of damages are from climate im- notincludedinthisanalysisduetodatalimitations.Further,itshouldbe
pacts.ItisimportanttonotethatthisdifferencebetweenChinaandthe clearthatthisinitial,exploratoryinvestigationofdamagesis:(i)restricted
USisdriven,inlargepart,bythedifferenceinthemagnitudeoftheVSL to a per coin basis (we make no inference about total or aggregate da-
foreachcountry.12 mages);and(ii)wedonotknowwherecoinsaremined(butthismight
Despite the relatively similar estimate of total monetary damages changeinthefuture).Ratherweprovideanestimateofthedamagesifa
percoinproducedinthetwocountries,themortalityimpactsofcryp- coinwereminedusingtheaverageemissionsprofileofelectricityintheUS
tocurrencyminingaretwiceashighinChinacomparedwiththeUS.In and China. If “flocks” of independent crypto-miners are endogenously
China, for every 50,000 BTC mined (a quantity that was produced sortingintolocationswithcheapelectricity,thenourestimatesherecould
globally approximately every month in 2018) there are 12 additional beanover-orunder-estimateofthesocialdamages.Asnotedearlier,an
deaths from exposure to particulate matter air pollution. example of such sorting can be seen in the Mid-Columbia Basin area of
Hypothetically, if the entire global production of the four coins we Washington State in the US, with cheap and renewable hydroelectricity
examined were produced in China in 2018, there would be ∼210 [4,6].Alternatively,manylocationswithcheapelectricityhaverelatively
deaths;whereasifallthecoinswereminedintheUS,therewouldbe largeexternalities(e.g.,iftheyarebasedoncoal-firedelectricity).
∼100deaths. Thereareseveralimportantavenuesforfutureresearch:(i)theoretical
modeling of mining sorting behavior; (ii) documentation and empirical
estimatesforgeographicalmininghotspots,andthespatialvariabilityof
4. Discussionandconclusions
localized pollution emissions, and impacted populations; (iii) improved
understandingofthehowthepositive/negativenetsocialbenefitsareex-
Given various uncertainties and weaknesses in traditional financial
pectedtovarytemporallyoverthelifeofacryptocurrency;(iv)explora-
institutions,Nakamoto's[7]introductionofBitcoinarguedthat:“Whatis
tions of improved methods for accurately assessing cryptodamages, and
neededisanelectronicpaymentsystembasedoncryptographicproofin-
decentralized production process in cryptocurrencies, given complicated
stead of trust, allowing any two willing parties to transact directly with
internationalpolicydebatesabouttheheterogeneousVSLestimatesacross
eachotherwithouttheneedforatrustedthirdparty.”Whiletheremaybe
countries (see [36,37]); and (v) theoretical and empirical evaluations of
importantsocialbenefitsconnectedtocryptocurrencies(anissuethatwe
alternatives to the POW computational competition currently pre-
havenotattemptedtofullyevaluatehere),thefocusofthisanalysishas
dominately used in the blockchain technology, which might be sig-
beenoninvestigatingthesocialdamagestheycreatevis-à-visenergyuse
nificantlylesselectricityintensive,andmorecapableofsustainingthenet
and associated air pollution emissions. For the US and China, our main
socialbenefitsofacryptocurrency.Narayananetal.[38]andVranken[39]
findingisthatin2018,each$1ofBitcoinvaluecreatedwasresponsiblefor
provideexcellentreviewsofthepropertiesofthemanyalternativemining
$0.49inhealthandclimatedamagesintheUSand$0.37inChina.Put
puzzlesandblockchaindesignalternatives,includingexpectedenergyuse
differently,thehumanhealthandclimatedamagescausedbyBitcoinre-
(seediscussionsofproof-of-space[5]andproof-of-stake[40],asjustsev-
presented almost half of the financial value of each US dollar of Bitcoin
eralofthevariousalternatives).
created (as represented by market prices). Further, the slightly smaller
Understandingthenegativeenvironmentalexternalitiesassociatedwith
valueinChinarelativetotheUS(foreach$1ofBitcoincreated)occurs
cryptocurrencyminingisemergingbutremainsunder-investigated[39,41]
primarilyduetotheextremelylargedisparitybetweentheVSLestimatefor
and wehopethat thisinitial investigationinto monetizingtheir impacts
theUS($11.53million)relativetothatofChina($1.12million),amore
spursadditionalresearch.Weclosebysharingourbriefthoughtsonpolicy
than10-to-1ratio.
implicationsofrecognizingcryptodamages.
If the energy demand timelines for producing Ethereum, Litecoin,
It is perhaps useful to think about the theoretical “last Bitcoin
andMonerofollowBitcoin,thenwemightanticipatesimilarperdollar
produced,” or any similarly supplied POW-based cryptocurrency de-
healthandclimatedamagesinthenearfuture.Itisalsoclearlypossible
sign. As noted in Truby [10], as the world confronts global climate
in the prescribed supply rules for a cryptocurrency that the
change, the POW process requires ever greater computational power
andenergyconsumption,“withnoconsiderationofitsenvironmental
12Table1showsthathealthdamagespercoininChinaareone-fifthaslarge impact.”Givenfinitesupplyrules,therewilllikelybeapointbeforethe
asintheUS.If,however,weusedthesameVSLinbothcountries,thehealth last possible coin (or whatever the currency reward unit) is supplied
damagesinChinawouldbetwiceaslargeasintheUS(resultsavailableupon
request).
7
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
whereitisnolongerprivatelyprofitabletomine(asmarginalprivate Acknowledgments
costs exceed marginal private benefits) [5]. We argue here that the
social paradox of the last Bitcoin is that in the presence of significant Theairqualitymodelusedinthispublicationwasdevelopedaspartof
cryptodamages the socially efficient end of production may be much the Center for Air, Climate, and Energy Solutions (CACES), which was
earlier (where marginal social costs exceed the marginal social bene- supportedunderAssistanceAgreementNo.R835873awardedbytheU.S.
fits). Thus, it is in the global public interest to prevent this socially Environmental Protection Agency. It has not been formally reviewed by
inefficientproductionandcollectivelymoveusawayfromblockchain EPA.Theviewsexpressedinthisdocumentaresolelythoseofauthorsand
designalternatives(i.e.,POW)thatarehighlyenergyintensive. donotnecessarilyreflectthoseoftheAgency.EPAdoesnotendorseany
Truby[10]providesathoroughandinsightfulreviewofthevarious productsorcommercialservicesmentionedinthispublication.
alternative approaches, and complex regulatory and fiscal challenges
for “decarbonizing” cryptocurrencies.Similar tothe difficulties of ad- References
dressing non-point source pollution problems,13 cryptomining poses
significant implementation challenges for designing a regulatory fra- [1] M.J.Krause,T.Tolaymat,Quantificationofenergyandcarboncostsforminingcrypto-
mework. As an alternative to restricting or prohibiting a production currencies,Nat.Sustainability1(2018)711–718.
[2] M.A.Hossin,M.S.Hosain,Bitcoin:futuretransactioncurrency?Int.J.Bus.Inf.13(3)
activity,economistsoftenadvocateforpollutiontaxesasonepossible (2018)385–404.
instrument to “internalize” otherwise unpriced and unaccounted-for [3] A.DeVries,Bitcoin'sgrowingenergyproblem,Joule2(5)(2018)801–805.
[4] P.Greenberg,D.Bugden,Energyconsumptionboomtownsintheunitedstates:com-
negative externalities. Theoretically, such taxes could be levied in
munityresponsestoacryptocurrencyboom,EnergyRes.SocialSci.50(2019)162–167.
various ways (e.g., on the sale of prominently-specialized mining [5] S.Park,P.Krzysztof,A.Kwon,J.Alwen,G.Fuschbauer,P.Gazi,Spacemint:acrypto-
hardware [3],theenergyuseitselfinknownconcentratedmining lo- currencybasedonproofsofspace,IACRCryptologyeprintarchive,2015.
[6] P.Roberts,ThisisWhatHappensWhenBitcoinMinersTakeOverYourTown,Politico
cales,orbysomehowidentifyingminingprofits)andviadifferentlevels
Magazine,2018March/April.https://www.politico.com/magazine/story/2018/03/09/
of government, and there is perhaps no single preferred option [10]. bitcoin-mining-energy-prices-smalltown-feature-217230.
Monetaryestimatesoftheenvironmentaldamagesofminingcanhelp [7] S.Nakamoto.Bitcoin:apeer-to-peerelectroniccashsystem.(2009),URL:www.bitcoin.
org/bitcoin.pdf.
incalibratingappropriatetaxrates. [8] J.Ma,J.Gans,R.Tourky,MarketStructureinBitcoinMining,NBER,2018Working
However, the ability to locate, and re-locate, cryptomining almost PaperNo.24242.http://www.nber.org/papers/w24242.
[9] M.A.Buth,A.A.Wieczorek,G.G.Verbong,Thepromiseofpeer-to-peertrading?The
anywhere(e.g.,followingthecheapest,under-regulatedelectricitysource)
potentialimpactofblockchainontheactorconfigurationintheDutchelectricitysystem,
andfundminingcampsfromanywherewithcompleteornearanonymity EnergyRes.SocialSci.53(2019)194–205.
create significant challenges to implementing effective regulation. The [10] J.Truby,DecarbonizingBitcoin:lawandpolicychoicesforreducingtheenergycon-
sumptionofBlockchaintechnologiesanddigitalcurrencies,EnergyRes.SocialSci.44
anonymous production process and footloose nature of the industry is (2018)399–410.
likely to ultimately require international cooperation in regulating the [11] C.Stoll,L.KlaaBen,U.Gallersdorfer,ThecarbonfootprintofBitcoin,Joule3(2019)
decarbonization of cryptocurrencies. Our concern is that delays in over- 1–15.
[12] M.Frankel,HowManyCryptcurrenciesAreThere?MotleyFool,2018March16.https://
coming such regulatory barriers risk further locking in path dependence www.fool.com/investing/2018/03/16/how-many-cryptocurrencies-are-there.aspx.
(see[42])forthePOW-basedblockchaintechnology,whichwillbeatodds [13] C.Mora,R.L.Rollins,K.Taladay,M.B.Kantar,M.K.Chock,M.Shimada,E.C.Franklin,
withtheurgentneedforclimatechangemitigation[10].14 Bitcoinemissionsalonecouldpushglobalwarmingabove2°C,Nat.Clim.Change8(11)
(2018)931.
As an alternative to intervening in emergent cryptocurrency mar- [14] A.M.FreemanIII,J.A.Herriges,C.L.Kling,TheMeasurementofEnvironmentaland
kets through prices (e.g., taxation) or outright restricting their devel- ResourceValues,ResourcesfortheFuture,2014.
[15] B.Kaiser,M.Jurado,A.Ledger,TheLoomingThreatofChina:AnAnalysisofChinese
opment,theremaybearoleforgovernmentinvestmentintoR&Dthat InfluenceOnBitcoin,CornellUniversity,2018ArXivworkingpaperURL:https://arxiv.
focuses on designing mining puzzles that greatly reduce energy con- org/abs/1810.02466.
[16] USEnvironmentalProtectionAgency,,2014NationalEmissionsInventory,Version2,
sumption (and thus production costs) while still allowing for secure
TechnicalSupportDocument,OfficeofAirQualityPlanningandStandards,2018URL:
validation of anonymous transactions. Fully vetting and encouraging https://www.epa.gov/sites/production/files/2018-07/documents/nei2014v2_tsd_
suchcurrencyalternativesmightretainthelibertariansocialbenefitsof 05jul2018.pdf.
[17] USEnvironmentalProtectionAgency,eGRID2016TechnicalSupportDocument,Officeof
the blockchain [10] while making “the last Bitcoin” irrelevant. Al-
AtmosphericPrograms,2018URL:https://www.epa.gov/sites/production/files/2018-
thoughtherewillstillbeprivatebenefitscapturedfromminingunder 02/documents/egrid2016_technicalsupportdocument_0.pdf.
possiblelow-energydesignalternatives,asillustratedheretherewould [18] B.Zheng,D.Tong,M.Li,F.Liu,C.Hong,G.Geng,H.Li,X.Li,L.Peng,J.Qi,L.Yan,
Y.Zhang,H.Zhao,Y.Zheng,K.He,Q.Zhang,TrendsinChina'santhropogenicemissions
besignificantpublicgoodbenefits(reducedcryptodamages)tojustify since2010astheconsequencesofcleanairactions,Atmos.Chem.Phys.18(2018)
suchgovernmentinvestmentsintoR&D.Importantly,suchinvestments 14095–14111.
[19] BitInfoCharts.com,(2019),URL:https://bitinfocharts.com/.
in R&D are not mutually exclusive with taxing to internalize crypto-
[20] D.Krewski,M.Jerrett,R.T.Burnett,R.Ma,E.Hughes,Y.Shi,M.C.Turner,C.A.Pope,
damages, and may also come with the potential to generate spin-off G.Thurston,E.E.Calle,M.J.Thun,ExtendedFollow-UpandSpatialAnalysisofthe
benefitsfromblockchaintechnologymoregenerally. AmericanCancerSocietyStudyLinkingParticulateAirPollutionandMortality,Health
EffectsInstitute,Boston,MA,2009.
[21] C.A.PopeIII,R.T.Burnett,G.D.Thurston,M.J.Thun,E.E.Calle,D.Krewski,K.Ito,etal.,
Lungcancer,cardiopulmonarymortality,andlong-termexposuretofineparticulateair
pollution,JAMA287(2002)1132–1141.
DeclarationofCompetingInterest
[22] Blockchain.com,(2019),HashRate,March2017-March2019.URL:https://www.
blockchain.com/charts/hash-rate.
None. [23] T.Lutey,HardinCoal-FiredPowerPlantSellstoPossibleCryptocurrencyCompany,
BillingsGazette,2019May1.https://billingsgazette.com/news/state-and-regional/
hardin-coal-fired-power-plant-sells-to-possible-cryptocurrency-company/article_
d602acfa-e3ac-5e6d-85f0-7b175f031c63.html?utm_content=bufferebe9c&utm_
13Non-pointsourcepollutionproblemsrefertopollutionoveralargearea medium=social&utm_source=twitter.com&utm_campaign=LEEDCC.
[24] K.Pierceall,VirginiaBeachBitcoinMiningCompanyThatWon$500,000CityGrantFiles
coming from a wide variety and mix of sources, rather than concentrations
ForBankruptcy,TheVirginian-Pilot,2019April22.https://www.pilotonline.com/
generatedbysmallnumbersofeasilyidentifiablesources(e.g.,dischargepipes) business/article_dd23979e-6514-11e9-bec0-8f0c2416333e.html.
that would be much easier to control or regulate. While it appears that the [25] K.Torpey,BlockstreamRevealsMassiveBitcoinMiningFacilities,FidelityAnEarly
miningofBitcoinandotherprominentcryptocurrenciesmaybeincreasingly Customer,Forbes,2019August8.https://www.forbes.com/sites/ktorpey/2019/08/08/
blockstream-reveals-massive-bitcoin-mining-facilities-fidelity-an-early-customer/#
concentratedinpoolsorcampsofminerswithspecializedhardware(e.g.,see 6db5bc6e4f3f.
Ma2018[8]),thestillanonymousnatureoftheproductionprocessleavesthis [26] InteragencyWorkingGroupontheSocialCostofGreenhouseGases.TechnicalSupport
difficulttopindown(butperhapspossibletoeventuallygeographicallylocate Document:TechnicalUpdateoftheSocialCostofCarbonforRegulatoryImpactAnalysis–
UnderExecutiveOrder12866(2016).https://19january2017snapshot.epa.gov/sites/
[seetherecentinitialinvestigationsbyStolletal.[11]]).
production/files/2016-12/documents/sc_co2_tsd_august_2016.pdf.
14AsofJuly3,2019,thewebsiteCryptoSlate.com[43]listedallPOW-based [27] C.Harvey,ShouldtheSocialCostofCarbonbeHigher?ScientificAmerican,ClimateWire,
cryptocurrencies (including leaders BTC, ETH, LTC and XMR) as holding a 2017November22.
marketshareor“sectordominance”of80.84%,amongallcryptocurrencies. [28] F.C.Moore,U.Baldos,T.Hertel,D.Diaz,Newscienceofclimatechangeimpactson
8
A.L.Goodkind,etal. Energy Research & Social Science 59 (2020) 101281
agricultureimplieshighersocialcostsofcarbon,Nat.Commun.8(1607)(2017)1–9, [35] Z.Yang,P.Liu,X.Xu,Estimationofsocialvalueofstatisticallifeusingwillingness-to-pay
https://doi.org/10.1038/s41467-017-01792-x. methodinNanjing,China,AccidentAnal.Prevent.95(2016)308–316.
[29] BP,BPStatisticalReviewofWorldEnergy,67thEdition,(2018).https://www.bp.com/ [36] W.K.Viscusi,PricingLives:GuidepostsforaSaferSociety,PrincetonUniversityPress,
content/dam/bp/business-sites/en/global/corporate/pdfs/energy-economics/statistical- 2018.
review/bp-stats-review-2018-full-report.pdf. [37] W.K.Viscusi,Pricinglives:internationalguidepostsforsafety,Econ.Rec.94(S1)(2018)
[30] A.L.Goodkind,C.W.Tessum,J.S.Coggins,J.D.Hill,J.D.Marshall,Fine-scaledamage 1–10.
estimatesofparticulatematterairpollutionrevealopportunitiesforlocation-specific [38] A.Narayanan,J.Bonneau,E.Felton,A.Miller,S.Goldfeber,BitcoinandCryptocurrency
mitigationofemissions,PNAS116(18)(2019)8775–8780. Technologies,PrincetonUniversityPress,2016.
[31] C.W.Tessum,J.D.Hill,J.D.Marshall,InMAP:amodelforairpollutioninterventions, [39] H.Vranken,SustainabilityofBitcoinandBlockchains,Curr.Opin.Environ.Sustainability
PLoSONE12(4)(2017)e0176131. 28(2017)1–9.
[32] J.S.Apte,E.Bombrun,J.D.Marshall,W.W.Nazaroff,Globalintraurbanintakefractions [40] P.Fairley,Ethereumwillcutbackitsabsurdenergyuse,IEEESpectr.56(1)(2019)
forprimaryairpollutantsfromvehiclesandotherdistributedsources,Environ.Sci. 29–32.
Technol.46(2012)3415–3423. [41] J.Li,N.Li,J.Peng,H.Cui,X.Wu,Energyconsumptionofcryptocurrencymining:astudy
[33] D.H.Bennett,T.E.McKone,J.S.Evans,W.W.Nazaroff,M.D.Margni,O.Jolliet, ofelectricityconsumptioninminingcryptocurrencies,Energy168(1)(2019)160–168.
K.R.Smith,Definingintakefraction,Environ.Sci.Technol.36(9)(2002)207A–211A. [42] W.B.Arthur,IncreasingReturns,andPathDependenceintheEconomy,Universityof
[34] USEnvironmentalProtectionAgency,GuidelinesforPreparingEconomicAnalyses, MichiganPress,1994.
NationalCenterforEnvironmentalEconomics,OfficeofPolicy,2010URL:https://www. [43] CryptoSlate.com(2019,)POWCryptocurrencies.URL:https://cryptoslate.com/cryptos/
epa.gov/sites/production/files/2017-09/documents/ee-0568-52.pdf. proof-of-work/.
9