版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
TheCryptoCycleand
USMonetaryPolicy
NatashaChe,AlexanderCopestake,DavideFurceri,and
TammaroTerracciano
WP/23/163
IMFWorkingPapersdescriberesearchin
progressbytheauthor(s)andarepublishedto
elicitcommentsandtoencouragedebate.
TheviewsexpressedinIMFWorkingPapersare
thoseoftheauthor(s)anddonotnecessarily
representtheviewsoftheIMF,itsExecutiveBoard,
orIMFmanagement.
2023
AUG
©2023InternationalMonetaryFund
WP/23/163
IMFWorkingPaper
AsiaandPacificDepartment
TheCryptoCycleandUSMonetaryPolicy
PreparedbyNatashaChe,AlexanderCopestake,DavideFurceri,andTammaroTerracciano*
AuthorizedfordistributionbyJayPeiris
IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.
ABSTRACT:WeexaminefluctuationsincryptomarketsandtheirrelationshipstoglobalequitymarketsandUSmonetarypolicy.Weidentifyasinglepricecomponent—whichwelabelthe“cryptofactor”—thatexplains80%ofvariationincryptoprices,andshowthatitsincreasingcorrelationwithequitymarketscoincidedwiththeentryofinstitutionalinvestorsintocryptomarkets.Wealsodocumentthat,asforequities,USFedtighteningreducesthecryptofactorthroughtherisk-takingchannel—incontrasttoclaimsthatcryptoassetsprovideahedgeagainstmarketrisk.Finally,weshowthatastylizedheterogeneous-agentmodelwithtime-varyingaggregateriskaversioncanexplainourempiricalfindings,andhighlightspossiblespilloversfromcryptotoequitymarketsiftheparticipationofinstitutionalinvestorseverbecamelarge.
JELClassificationNumbers:E44;E52;F33.
Keywords:USMonetaryPolicy;Cryptoassets;StockMarkets.
DFurceri@;TTerracciano@
Author’sE-MailAddress:NChe@;ACopestake@;
*Firstdraft:August2022.Che(NChe@),Copestake(ACopestake@)andFurceri(DFurceri@)areattheInternationalMonetaryFund.Terracciano(TTerracciano@)isattheIESEBusinessSchool(Barcelona).WearegratefultoItaiAgur,JohnCampbell,GabrielaCondeVitureira,ShaenCorbet,HaradHau,CharlesLarkin,JayPeiris,MichaelRockinger,KennethRogoff,JeremyStein,BrandonTan,SaskiaterEllen,FabioTrojaniandseminarparticipantsattheIMFandtheUniversityofGenevafortheirconstructivecomments.Remainingmistakesareourown.TheviewsexpressedinthispaperarethoseoftheauthorsandshouldnotbeattributedtotheIMF,itsExecutiveBoard,orIMFmanagement.
WORKINGPAPERS
TheCryptoCycleand
USMonetaryPolicy
PreparedbyNatashaChe,AlexanderCopestake,DavideFurceri,and
TammaroTerracciano
2
1Introduction
Cryptoassetsvarysubstantiallyintheirdesignandvaluepropositions,yettheirpriceshavemovedincommoncycles.
1
TotalcryptomarketcapitalizationboomedfromUS$20billionin2016toalmostUS$3trillioninNovember2021,beforecollapsingtobelowUS$1trillioninthelatestcrypto‘winter’.
2
Periodsofexponentialreturnshaveattractedretailandinstitu-tionalinvestorsalike(
BenettonandCompiani
,
2022
;
AuerandTercero-Lucas
,
2021
;
Auer,
Farag,Lewrick,Orazem,andZoss
,
2022
),whilesubsequentcrasheshavedrawnincreasingattentionfrompoliticiansandregulators.Thesefluctuationsincryptomarketsmayalsobeincreasinglysynchronizedwithotherassetclasses:priorto2020,Bitcoinprovidedapartialhedgeagainstmarketrisk,yetithassincebecomeincreasinglycorrelatedwiththeS&P500(
Adrian,Iyer,andQureshi
,
2022
).
However,weknowrelativelylittleaboutthecommondriversofcryptoassetpricesorthefactorsa↵ectingthecorrelationbetweencryptoandequitymarkets,includingUSmonetarypolicy.Thispapertriestoshedlightontheseissuesbyansweringthefollowingquestions.Towhatextentisthereacommoncycleacrosscryptoassets?Arecryptomarketsbecomingmoresynchronizedwithglobalequitymarkets?Ifso,why?GiventhatUSmonetarypolicyhasbeenidentifiedasakeydriveroftheglobalfinancialcycle(
Miranda-AgrippinoandRey
,
2020
),doesUSmonetarypolicyinfluencethecryptocycletoasimilarextent?Ifso,throughwhichchannels?
Westartansweringthesequestionsbyusingadynamicfactormodeltoidentifyasingledominanttrendincrypto-assetprices.Usingapanelofdailypricesforseventokenscreatedbefore2018,whichtogetheraccountforapproximately75%oftotalcryptomarketcapital-ization,wedecomposetheirvariationintoasset-specificidiosyncraticdisturbancesandan
1Forinstance,thewhitepapersofprominentcryptoassetsincludeaimstoprovidepeer-to-peerelectroniccash,moreefficienttransactions,censorship-resistantdecentralizedcomputing,andfunctionalitywithinafinancialservicesecosystem(
Nakamoto
,
2008
;
Buterin
,
2014
;
RippleLabsInc.
,
2014
;
Binance
,
2017
;
Sun
,
2018
).Weexcludestablecoinsfromouranalysis,astheyareintendedtomaintainaconstantprice.
2Source:
CoinMarketC
.
3
AR(q)commoncomponent.Wefindthattheresulting“cryptofactor”explainsapproxi-mately80%ofthevarianceinthecryptopricedata.Thisissubstantiallylargerthanthe20%figureforglobalequitiescalculatedby
Miranda-AgrippinoandRey
(
2020
),whichalsoreflectsthegreaterconcentrationofmarketcapitalizationinthelargestcryptoassetsrel-ativetothatinthelargestequities.Thisfigureisrobustforvariouslagordersq,andwefindasimilarlyhighdegreeofcorrelationwhenbroadeningthepaneltoincludemorecryptoassets.
3
Inasecondstep,westudytherelationshipofthiscryptofactortoasetofglobalequityfactors,constructedusingtheequityindicesofthelargestcountriesbyGDP(inthespiritof
Rey
,
2013
;
Miranda-AgrippinoandRey
,
2020
).Wefindapositivecorrelationovertheentiresample,drivenbyaparticularlystrongcorrelationsince2020.Theincreasingco-movementisnotlimitedtoBitcoinvis-a-vistheS&P500,butpertainsmorebroadlytothecryptoandglobalequityfactors.Disaggregatingacrossequitymarkets,wefindthatthecryptofactorcorrelatesmoststronglywiththeglobaltechfactorandthesmall-capfactorsince2020,whileitissurprisinglylesscorrelatedwiththeglobalfinancialfactor.
Theincreasedcorrelationbetweencryptoandequitiescoincideswiththegrowthintheparticipationofinstitutionalinvestorsincryptomarketssince2020.Althoughinstitutions’exposureissmallrelativetotheirbalancesheets,theirabsolutetradingvolumeismuchlargerthanthatofretailtraders.Inparticular,thevolumeoftradingbyinstitutionalinvestorsincryptoexchangesincreasedbymorethan1700%(fromroughly$25billiontomorethan$450billion)from2020Q2to2021Q2(
Aueretal.
,
2022
).Sinceinstitutionalinvestorstradebothstocksandcryptoassets,thishasledtoaprogressiveincreaseinthecorrelationbetweentheriskprofilesofmarginalequityandcryptoinvestors,whichinturnisassociatedwithahighercorrelationbetweentheglobalequityandcryptofactors.Whendecomposingfactormovementsfollowing
Bekaert,Hoerova,andLoDuca
(
2013
),wefindthatcorrelationinthe
3Sincemostcryptoassetshavebeencreatedonlyinthelastcoupleofyears,abroaderpanelofassetsalsoimpliesashortertimedimension—hencewefocusonthesevenmainassetsinourbaselinemeasure.
4
aggregatee↵ectiveriskaversionofcryptoandequitiescanexplainalargeshare(upto65%)ofthecorrelationbetweenthetwofactors.
SinceUSmonetarypolicya↵ectstheglobalfinancialcycle(
Miranda-AgrippinoandRey
,
2020
),thehighcorrelationbetweenequitiesandcryptosuggestsasimilarimpactoncryptomarkets.WetestthishypothesisusingadailyVARwiththeshadowfederalfundsrate(SFFR)by
WuandXia
(
2016
)toaccountfortheimportantroleofbalancesheetpolicyoveroursampleperiod.OuridentificationoftheimpactofmonetarypolicyshocksisbasedonaCholeskydecompositionwiththefollowingordering:theSFFR;theTreasury10Y2Yspread,reflectingexpectationsoffuturegrowth;thedollarindex,oilandgoldprices,asproxiesforinternationaltrade,creditandcommoditycycles;theVIX,reflectingexpectedfutureuncertainty;andfinallytheequityandcryptofactors.Inthissetup,endogeneityisnotlikelytobeanissueasitisimplausiblethattheFederalReserveadjustsitsmonetarypolicyaccordingtocryptopricemovementsandthatitdoessoatthedailylevel.
4
WefindthatUSmonetarypolicya↵ectsthecryptocycle,asitdoeswiththeglobalequitycycle,contrastingstarklywithclaimsthatcryptoassetsprovideahedgeagainstmarketrisk.AonepercentagepointriseintheSFFRleadstoapersistent0.15standarddeviationdeclineinthecryptofactoroverthesubsequenttwoweeks,relativetoa0.1standarddeviationdeclineintheequityfactor.
5
Interestingly,aswiththeglobalfinancialcycle(
Rey
,
2013
),wefindthatonlytheUSFed’smonetarypolicymatters,andnotthatofothermajorcentralbanks—likelyreflectingthatcryptomarketsarehighlydollarized.
6
Wefindevidencethattherisk-takingchannelofmonetarypolicyisanimportantchannel
4Notethatourresultsarealsorobusttorelaxingtheaforementionedvariableordering.Whenweinverttheorderofthevariablestoallowthepolicyratetobethemostendogenousone,wefindsimilarresults.Asexpected,wealsofindthatthepolicyratedoesnotrespondtochangesinthecryptofactor.Thus,ourfindingsdonotdependonanarbitraryorderingofthevariables.Inaddition,usingtheavailable
Bu,Rogers,
andWu
(
2021
)monetarypolicyshocks;westillfindasignificantnegativee↵ectofUSmonetarypolicyonthecryptocycleatthemonthlylevel.
5Thisreferstostandarddeviationsofvariationincryptoorequitiesover2018-2023,theperiodforwhichwecanconstructthecryptofactor.
6Forinstance,thetwolargeststablecoinsTetherandUSDCoinarepeggedtothedollar,whileCoinbase,thelargestcentralizedcryptoexchange,islistedontheNewYorkStockExchange.
5
drivingtheseresults,parallelingthefindingsof
Miranda-AgrippinoandRey
(
2020
)forglobalequitymarkets.Inparticular,wefindthatamonetarycontractionleadstoareductionofthecryptofactorthatisaccompaniedbyasurgeinaproxyfortheaggregatee↵ectiveriskaversionincryptomarkets.Putdi↵erently,restrictivepoliciesrendertheriskpositionsofin-vestorslesssustainable,andthustheyreducetheirexposuretocryptoassets.Whensplittingthesamplein2020,wefindthattheimpactonriskaversionincryptomarketsissignificantonlyforthepost-2020period,consistentwiththeentryofinstitutionalinvestorsreinforcingthetransmissionofmonetarypolicytothecryptomarket.Moreformally,wefindthesameresultwhentestingthishypothesisusingasmoothtransitionVARfollowing
Auerbachand
Gorodnichenko
(
2012
),wherethetransitionvariableistheshareofinstitutionalinvestors.
Next,werationalizeourresultsinamodelwithtwoheterogeneousagents,namelycryptoandinstitutionalinvestors.Theformerareretailinvestorswhoonlyinvestincryptoassets,whilethelattercaninvestinbothstocksandcryptoassets.Crucially,cryptoinvestorsareriskaverse,whileinstitutionalinvestorsarerisk-neutralbutfaceavalue-at-riskconstraint.Wecanrewritetheequilibriumreturnsonthecryptoassetsasalinearcombinationoftheirvarianceandtheircovariancewithstocks’returns,scaledbytheaggregatee↵ectiveriskaversion.Thelattercanbeinterpretedastheaverageriskaversionoftheagents,weightedbytheirwealth.Thisimpliesthatthehighertherelativewealthofinstitutionalinvestors,themoresimilarthecryptoaggregatee↵ectiveriskaversionbecomestotheirriskappetiteandthemorecorrelatedarecryptoandequitymarkets.Sincethepresenceofinstitutionalinvestorsincryptomarketsdecreasestheaggregatee↵ectiveriskaversion,weinterprettheincreasingreactionofcryptopricestomonetarycontractionasreflectingthatmoreleveredinvestorsaremoresensitivetotheeconomiccycle(
Coimbra,Kim,andRey
,
2022
;
Adrian
andShin
,
2014
).Finally,wenotethatspilloversfromcryptotoequitiescanariseeveninoursimpleframework:ifinstitutions’cryptoholdingsbecomelarge,acrashincryptopricesreducesequilibriumreturnsinequities.
Overall,ourresultshighlightthatthecryptocycleisremarkablysynchronizedwithglobal
6
equitymarketsandreactssimilarlytomonetarypolicyshocks.Despitetherangeofexpla-nationsforcryptoassetvalues—e.g.,asaninflationhedgeorasaproviderofmoreefficientpayments,censorship-resistantcomputingorpropertyrights—mostvariationincryptomar-ketsishighlycorrelatedwithequitypricesandhighlyinfluencedbyFedpolicies.Thisalsosuggestsemergingcryptoventuresthatbenefitedfromhighcryptoreturnswerecon-comitantlysupportedbythelowinterest-rateenvironment.Finally,wefindthatgrowthininstitutionalparticipationhasstrengthenedtheseconclusionsandincreasedtheriskofspilloversfromcryptomarketstothebroadereconomy.
Literature:Thispapercontributestotheburgeoningliteratureoncryptoassetsbycon-nectingworkonspecificcryptopricesandthecompositionofcryptoinvestorstotheestab-lishedliteratureontheglobalfinancialcycle.
First,ourpaperbuildsonworkassessingthedriversofthepricesofspecificcryptoassets.Theprimaryaimofearlydevelopers,ledby
Nakamoto
(
2008
),wastoprovideanewformofdecentralizedelectroniccashthatpeoplecouldfreelyaccess.Severalscholarshavestudiedthematterthroughtheselenses(
Biais,Bisi`ere,Bouvard,andCasamatta
,
2018
;
Schillingand
Uhlig
,
2019
;
Brunnermeier,James,andLandau
,
2019
;
Cong,He,Li,andJiang
,
2021
;
Auer,
Monnet,andShin
,
2021
;
Pagnotta
,
2022
),yetthehighpricevolatilityandtherelativelylowscalabilityofexistingdistributedledgertechnologyhaveledresearcherstothinkofmostcryptotokensasassetsratherthancurrencies(seeforinstance
Liu,Tsyvinski,andWu
,
2022
;
MakarovandSchoar
,
2020
;
Scaillet,Treccani,andTrevisan
,
2020
).
7
Indeed,manycryptoassetslackexplicitfundamentalvalueorcash-flows(
MakarovandSchoar
,
2020
),andaresubjecttofragmentation,arbitrageopportunitiesandmarketmanipulation(
Griffinand
Shams
,
2020
;
Gandal,Hamrick,Moore,andOberman
,
2018
;
Foley,Karlsen,andPutnins
,
2019
).Inthispaper,weabstractfromcrypto-asset-specificpricingconsiderations,and
7Indeed,toaddresssuchhighvolatility,theindustrydevelopedstablecoins,suchasTetherorUSDCoin,whicharepeggedtoanothercurrency(mostcommonlytheUSdollar).
7
insteadconsiderthecommonmovementintheentireassetclass.Indoingso,webuildon
Iyer
(
2022
),whoprovidesevidenceofthepositivecorrelationbetweenUSequitymarketsandBitcoinandEtherprices,and
Corbet,Larkin,Lucey,Meegan,andYarovaya
(
2020
)whoassesstheimpactofmacroeconomicnewsonBitcoinreturns.
Second,wedrawonanemergingempiricalliteratureexaminingthecompositionandmotivationsofcryptoinvestors,includingtheincreasedparticipationofinstitutions.
Auer
andTercero-Lucas
(
2021
)studytheprofileofUScryptoinvestorsandhighlightthattheyareingenerallessmotivatedbydistrustinthetraditionalfinancialsystemthanbytheprospectsforhighreturns.
8
Auer,Farag,Lewrick,Orazem,andZoss
(
2022
)arethefirsttofocusontheroleofinstitutionalinvestorsincryptomarkets,andshowthattraditionalfinancialinstitutions,especiallylightlyregulatedbanks,arestartingtoholdcryptoassets.
9
Weusethisliteraturetohelpexplaintheco-movementbetweencryptoandequities,andtoconstructastylizedframeworkforinvestigatingpotentialspilloversbetweenthetwo.
Third,wecontributetotheliteratureontheglobalfinancialcycle.
10
Inherseminalcontribution,
Rey
(
2013
)showstheexistenceofasinglefactorthatexplains20%ofthevariationinglobalassetprices.Inmorerecentworks,
Miranda-AgrippinoandRey
(
2020
)and
Miranda-AgrippinoandRey
(
2021
)highlighthowUSmonetarypolicya↵ectsthisglobalfinancialcyclethroughtherisk-takingchannel.Achangeininterestratesforcesfinancialintermediariestochangetheirleverageandthusthee↵ectiveriskappetiteofthemarginalinvestor.AUSmonetarycontractionthusnegativelya↵ectsglobalequityprices,erodingtheindependenceofothercentralbanksandreinforcingthedominantroleoftheUSdollar(
PassariandRey
,
2015
;
FarhiandMaggiori
,
2018
).Withinthisliterature,weparticularlyfocusontherisk-takingchannelofmonetarypolicy.
Coimbraetal.
(
2022
)developacom-
8Similarly
Hackethal,Hanspal,Lammer,andRink
(
2021
)and
DidisheimandSomoza
(
2022
)documentthebehaviourofcryptoretailinvestorsandtheirportfolioallocationbetweenequityandcryptoassets.
9Nonetheless,banks’exposureremainslimitedwithrespecttothesizeoftheirbalancesheets.Inaddition,
Cornelli,Doerr,Frost,andGambacorta
(
2023
)documentdi↵erencesintradingbehaviourbetweensmallandlargeinvestorsduringcrisisepisodes.
10Forearlydiscussions,see:
Diaz-Alejandro
(
1985
);
Calvo,Leiderman,andReinhart
(
1996
).
8
prehensivedynamicmodelwithheterogeneousintermediaries,whichfeaturestime-varyingendogenousmacroeconomicrisk.Intheirframework,thevariationinrisk-aversionacrossagentsdeterminestheaggregateriskoftheeconomy.Relatedly,
AdrianandShin
(
2014
)high-lighthowthecyclicalityofleveragedependsontheconstraintsoffinancialintermediaries.
FostelandGeanakoplos
(
2008
)showhowleveragecyclescanbeexplainedbydi↵erencesinagents’beliefs,whereas
KekreandLenel
(
2018
)and
Gourinchas,Rey,andGovillot
(
2010
)focusonheterogeneityinriskaversion.Wecontributetothisliteraturebyincorporatinganalysisofthecryptocycle.
Therestofthispaperproceedsasfollows.Section
2
derivesthecryptofactor,thenSection
3
investigatesitsrelationshiptoequitypricesandtheglobalfinancialcycle.Section
4
examinestheimpactofUSmonetarypolicyonthecryptofactor,andSection
5
rationalizesourfindingsinaheterogeneous-agentmodel.Section
6
concludes.
2TheCryptoFactor
Thepricesofcryptoassetsarehighlycorrelated.Table
1
reportsthecross-correlationsamongthecryptoassetswiththelargestmarketcapitalization.Theseareremarkablyhigh,andmuchlargerthanthecorrelationsdocumentedacrossequitymarkets(see,forinstance,
Rey
,
2013
).Forexample,Bitcoinhasa52%averagecorrelationwithothercryptoassets.Wethusconjecturetheexistenceofacommoncryptofactorthatco-moveswithcryptoprices,inthesamespiritastheglobalequityfactorpioneeredby
Rey
(
2013
).
9
1.00
0.820.640.62
1.000.640.67
1.000.52
0.750.570.310.700.610.470.450.590.630.820.000.660.730.56
0.69
0.470.340.640.590.490.380.550.530.800.030.590.750.55
0.560.510.240.580.470.460.320.550.470.630.030.510.590.48
1.000.480.300.630.590.420.380.640.540.72-0.010.580.660.53
1.000.160.520.370.340.430.540.470.490.520.530.430.43
1.000.230.250.510.150.210.140.330.000.270.300.22
1.000.560.430.540.590.600.660.520.700.550.60
1.00
0.34
0.27
0.44
0.46
0.58
0.00
0.52
0.55
0.48
1.000.320.340.430.450.310.420.390.40
1.000.510.540.38-0.010.330.340.27
1.000.510.530.480.590.460.50
1.000.560.450.590.440.49
1.00-0.010.600.720.55
1.00
-0.01
0.04
-0.01
1.000.540.48
1.000.53
1.00
Table1:Correlationsamongcryptoassets
Bitcoin
EthereumBinanceCoin Ripple Cardano Solana DogecoinPolkadot
Tron ShibaInuMakerDaoAvalanche
Uniswap
Litecoin FTXChainlink MoneroTHETA
1.00
0.65
0.42
0.26
0.49
0.58
0.41
0.33
0.48
0.44
0.67
0.00
0.53
0.59
0.46
BitcoinEthereumBinanceCoinRippleCardanoSolanaDogecoinPolkadotTronShibaInuMakerDaoAvalancheUniswapLitecoinFTXChainlinkMoneroTHETA
Notes:Thistableshowspairwisecorrelationsbetweenselectedcrypto-assetreturns.DataisfromJanuary2018toMarch2023.
Tosummarizethefluctuationsincryptomarketsintoonevariable,weusedynamicfactormodelling,adimensionalityreductiontechnique.
11
Thisallowsustodecomposeasetofpricesintotheiridiosyncraticcomponentsandacommontrend.Specifically,westartwiththedailypricesofthelargestcryptoassetsthatwerecreatedbeforeJanuary2018(excludingstablecoins).Thisleavesuswithsevencryptoassets,accountingfor75%oftotalmarketcapitalizationinJune2022.
12
WethenwritethispanelofcryptopricespitasalinearcombinationofanAR(q)commonfactorftandanasset-specificidiosyncraticdisturbance✏it(whichinturnfollowsanAR(1)process):
pit=λi(L)ft+✏it
ft=A1ft−1+...+Aqft−q+⌘t
✏it=pi✏it−1+eit
(1)
⌘t⇠N(0,⌃)
eit⇠N(0,σ)
whereListhelagoperatorandλi(L)isaq-ordervectoroffactorloadingsforasseti.Estimatingthissystemusingmaximumlikelihood,selectingqusinginformationcriteria,
11Fortheevolutionofthemethod,see,amongothers:
Geweke
(
1977
);
SargentandSims
(
1977
);
Forni,
Hallin,Lippi,andReichlin
(
2000
);
BaiandNg
(
2002
);
StockandWatson
(
2002
);
Miranda-Agrippinoand
Rey
(
2020
).
12Theseare:Bitcoin,Ethereum,BinanceCoin,Ripple,Cardano,DogeCoin,andTron.
10
producesourcommonfactorft.
13
Itisalsopossibletospecifymultiplefactorsthata↵ectpricesdi↵erently,andweusethislatterspecificationwhenweconsidermultipledistinctsub-classesofcryptoassets.
Figure
1
showsthecryptofactorandtheunderlyingpriceseriesfromwhichweextractit.Thecryptofactore↵ectivelycapturesthesalientphasesthatcharacterizedcryptomarkets—suchasthedeclineatthebeginningof2018,thesubsequent‘cryptowinter’,thelatestboomwiththepeaksinBitcoinandDogecoin,andfinallytheslumpofTerraandFTXof2022—withoutbeingoverlyinfluencedbyisolatedspikeslikethoseofRippleandTron.
Figure1:Thecryptofactor
Notes:Thisfigureshowsthecryptofactor(blue)andthestandardizedcryptopricesfromwhichitisconstructed(grey)usingadynamicfactormodel.
Togaugetheimportanceofthisfactormoresystematically,weregresseachpriceseriesinturnonthecryptofactor.Onaverage,80%ofvariationintheunderlyingseriesisexplainedbyourcryptofactor.
14
Thisfigureisabove68%forallsevenassets,underscoring
13WeusethePythonpackagestatsmodels/DynamicFactor.Forfurtherinformationaboutthemodelandalgorithm,see
/dev/examples/notebooks/generated/statespace_
dfm_coincident.html
.
14SeeAppendixFigure
A.1
forthebreakdownacrossindividualcryptoassets.
11
thehighdegreeofco-movementoveroursampleperiod.Forcomparison,theglobalequityfactorcalculatedby
Miranda-AgrippinoandRey
(
2020
)explainsonly20%ofglobalequityprices,highlightingthegreaterco-movementandconcentrationofmarketcapitalizationinthecryptomarket.Ourfindingsthusstronglycorroboratethehypothesizedexistenceofasinglecryptofactorthatdrivesthepricesoftheentirecryptomarket.
Giventhelimitedrangeofassetsusedtocalculateourfactor,wealsoconfirmthatourcryptofactorreflectsmorerecenttrendsinnewerassets.
15
Todoso,weexamineabroadersampleofassets,groupedintofivecategories:FirstGenerationtokens(Bitcoin,RippleandDogecoin),SmartContractsplatformtokens(Ethereum,BinanceCoin,Cardano,SolanaandPolkadot),DeFitokens(Chainlink,Uniswap,MakerandAave),Metaversetokens(Flow,ApeCoin,theSandbox,DecentralandandThetaNetwork)andInternetofThingstokens(Helium,Iota,IoTexandMXC).Wethenestimateanewmodelwithfivedi↵erentfactors,whereeachfactorcanonlya↵ectoneclass.TheresultsareshowninFigure
2
,alongwiththegeneralcryptofactorestimatedabove.
16
Allclassesarehighlycorrelatedwiththegeneralcryptocycle,validatingourfocusonthecommontrend.
17
15Wedonotincludethesenewerassetsinthecalculationofthemainfactor,astheywouldfurtherlimitthetimespanofoursample.
16Notethatthetimespanforeachofthenewfactorsissubstantiallyshorter,giventhatmanywerecreatedonlyin2021.
17ThemainexceptionisthejumpintheMetaversefactorinlate2021,whenFacebookre-brandedtoMeta.Outsideofthisidiosyncraticshock,movementsintheMetaversefactoralsofollowthegeneraltrend.
12
Figure2:Cryptosub-factors
Notes:Thisgraphshowstheoverallcryptofactorandfivecryptosub-factors,standardizedandsmoothed.Thesub-factorsareconstructedfromthefollowingassets:FirstGenerationtokens—Bitcoin,RippleandDogecoin;SmartContractplatformtokens—Ethereum,BinanceCoin,Cardano,SolanaandPolkadot;DeFitokens—Chainlink,Uniswap,MakerandAave;Metaversetokens—Flow,ApeCoin,theSandbox,Decentra-landandThetaNetwork;andInternetofThingstokens—Helium,Iota,IoTexandMXC.
Finally,andconsistentwithanecdotalevidence,thecryptofactorcorrelateswithaproxyforleverageincryptomarkets.Figure
3
plotsthecryptofactoragainstcryptoleverage,definedusingthetotalvaluelocked(TVL)indecentralizedfinance(“DeFi”)contractsnor-malizedbytotalcryptomarketcapitalization.
18
Thisshowsthatthesystemwasrelativelyunlevereduntiltheendofthe2018-2019cryptowinter,afterwhichleverageincreasedsub
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 学校(园)食品安全专项督查表
- 智杰教育:急重症护理
- 经皮经肝胆道引流术患者的术后引流护理
- 生活护理课件资源体系
- (编制说明)《智慧农业设施作物数据采集规范》(征求意见稿)
- 家国情怀与统一多民族国家的演进说课稿-2025-2026学年高中历史必修 中外历史纲要(上)统编版(部编版)
- 2026年假日野花开测试题及答案
- 2026年逆商情商测试题及答案
- 2026年招聘专员的测试题及答案
- 2026年宝鸡英创入学测试题及答案
- 2026年水利安全生产考核b证题库附参考答案详解【培优】
- 2026四川泸州北方化学工业有限公司社会招聘保卫人员8人笔试备考题库及答案详解
- 2026年安徽合肥市高三二模语文试卷试题打印版
- 安全骑行 平安五一2026年北京电动车新规全解析
- 盒马鲜生活动方案
- 施工现场实名制奖惩制度
- 4.1《权利与义务相统一》课件 2025-2026 学年统编版道德与法治 八年级下册
- 功与功率课件2025-2026学年高一下学期物理人教版必修第二册
- (完整版)施工现场质量、安全生产管理体系
- 2025年中职学前教育笔试题目及答案
- 特种设备(每周)安全排查治理报告
评论
0/150
提交评论