You are on page 1of 37

NBER WORKING PAPER SERIES

DILEMMA NOT TRILEMMA:


THE GLOBAL FINANCIAL CYCLE AND MONETARY POLICY INDEPENDENCE
Hlne Rey
Working Paper 21162
http://www.nber.org/papers/w21162

NATIONAL BUREAU OF ECONOMIC RESEARCH


1050 Massachusetts Avenue
Cambridge, MA 02138
May 2015

I am grateful to Silvia Miranda-Agrippino and to Evgenia Passari for excellent research assistance
and to Richard Portes for discussions. I gratefully acknowledge the ERC (starting grant 210584) for
funding. This paper was presented in the Jackson Hole Symposium. The views expressed herein are
those of the author and do not necessarily reflect the views of the National Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peerreviewed or been subject to the review by the NBER Board of Directors that accompanies official
NBER publications.
2015 by Hlne Rey. All rights reserved. Short sections of text, not to exceed two paragraphs, may
be quoted without explicit permission provided that full credit, including notice, is given to the source.

Dilemma not Trilemma: The global Financial Cycle and Monetary Policy Independence
Hlne Rey
NBER Working Paper No. 21162
May 2015
JEL No. E5,F02,F33,G15
ABSTRACT
There is a global financial cycle in capital flows, asset prices and in credit growth. This cycle comoves
with the VIX, a measure of uncertainty and risk aversion of the markets. Asset markets in countries
with more credit inflows are more sensitive to the global cycle. The global financial cycle is not aligned
with countries specific macroeconomic conditions. Symptoms can go from benign to large asset price
bubbles and excess credit creation, which are among the best predictors of financial crises. A VAR
analysis suggests that one of the determinants of the global financial cycle is monetary policy in the
centre country, which affects leverage of global banks, capital flows and credit growth in the international
financial system. Whenever capital is freely mobile, the global financial cycle constrains national monetary
policies regardless of the exchange rate regime.
For the past few decades, international macroeconomics has postulated the trilemma: with free capital
mobility, independent monetary policies are feasible if and only if exchange rates are floating. The
global financial cycle transforms the trilemma into a dilemma or an irreconcilable duo: independent
monetary policies are possible if and only if the capital account is managed.
So should policy restrict capital mobility? Gains to international capital flows have proved elusive
whether in calibrated models or in the data. Large gross flows disrupt asset markets and financial
intermediation, so the costs may be very large. To deal with the global financial cycle and the dilemma,
we have the following policy options: ( a) targeted capital controls; (b) acting on one of the sources
of the financial cycle itself, the monetary policy of the Fed and other main central banks; (c) acting
on the transmission channel cyclically by limiting credit growth and leverage during the upturn of
the cycle, using national macroprudential policies; (d) acting on the transmission channel structurally
by imposing stricter limits on leverage for all financial intermediaries.

Hlne Rey
London Business School
Regents Park
London NW1 4SA
UNITED KINGDOM
and NBER
hrey@london.edu

Introduction

Ifonelooksattheevolutionoffinancialintegrationoverthepasthalfcenturyintheworldeconomy,
onemightconcludethatfinancialopennessisanirresistiblelongruntrend,hailedbypolicymakers
andacademiceconomistsalike.Bothemergingmarketsandadvancedeconomieshaveincreasingly
openedtheirborderstofinancialflows.Thescopeforinternationalcapitalflowstoprovidewelfare
gainsortodoharmhaswidenedconsiderablysincethe1990s.
Ininternationalmacroeconomicsandfinanceweoftenthinkwithintheframeworkofthe
trilemma:inafinanciallyintegratedworld,fixedexchangeratesexportthemonetarypolicyofthe
centrecountrytotheperiphery.Thecorollaryisthatiftherearefreecapitalflows,itispossibleto
haveindependentmonetarypoliciesonlybyhavingtheexchangeratefloat;andconversely,that
floatingexchangeratesenablemonetarypolicyindependence(seee.g.ObstfeldandTaylor(2004)).
Butdoesthescaleoffinancialglobalizationandinparticulartheroleofglobalbanksputeventhis
intoquestion?Arethefinancingconditionssetinthemainworldfinancingcentressettingthetone
fortherestoftheworld,regardlessoftheexchangerateregime?Isthereaglobalfinancialcycleand
ifyes,whatareitsdeterminants?
Riskyassetpricesaroundtheglobe,fromstockstocorporatebonds,haveastrongcommon
component.Sodocapitalflows.Creditflowsareparticularlyprocyclicalandvolatile.Ascredit
cyclesandcapitalflowsobeyglobalfactors,theymaybeinappropriateforthecyclicalconditionsof
manyeconomies.Forsomecountries,theglobalcyclecanleadtoexcessivecreditgrowthinboom
timesandexcessiveretrenchmentinbadtimes.Astherecentliteraturehasconfirmed,excessive
creditgrowthisoneofthebestpredictorsofcrisis(GourinchasandObstfeld(2012),Schularickand
Taylor(2012)).Globalfinancialcyclesareassociatedwithsurgesandretrenchmentsincapitalflows,
boomsandbustsinassetpricesandcrises.Thepictureemergingisthatofaworldwithpowerful
globalfinancialcyclescharacterisedbylargecommonmovementsinassetprices,grossflowsand
leverage.Itisalsoaworldwithmassivedeviationsfromuncoveredinterestparity.Thereare
interrelationswiththemonetaryconditionsofthecentrecountry(theUS),capitalflowsandthe
leverageofthefinancialsectorinmanypartsoftheinternationalfinancialsystem.Theglobal
financialcyclecanberelatedtomonetaryconditionsinthecentrecountryandtochangesinrisk
aversionanduncertainty(Bekaertetal.(2012),MirandaAgrippinoandRey(2012),BrunoandShin
(2013b)).

Butevenifcapitalflows,especiallycreditflows,arelargelydrivenbyaglobalfactor,theymightstill
bringimportantbenefitstotheworldeconomy.Abriefreviewoftheempiricalevidenceandthe
quantificationofstandardgrowthmodels,however,showshowelusivewelfaregainstocapital
flowsappeartobe,thoughitcouldjustbethattheyarehardtomeasure.
InpartI,Idescribethecharacteristicsofcapitalflows(grossandnet),showimpressiveco
movementingrossflowsanddiscusshowtheyrelatetoglobalfactors,asproxiedinparticularby
theVIX.InpartII,Ishowtheexistenceofanimportantcommonfactorininternationalassetprices,
whichisalsocloselyrelatedtotheVIX.Iconcludethatthereisapotentglobalfinancialcyclein
grosscapitalflows,creditcreationandassetprices,whichhastightconnectionswithfluctuationsin
uncertaintyandriskaversion.PartIIIanalysestheassociationofdifferenttypesofcapitalflowswith
theglobalfinancialcycleandreinforcestheconclusionthatcreditflowsareparticularlyconnectedto
theglobalfinancialcycle.InpartIV,Ihuntforthedeterminantsoftheglobalfinancialcycleitself
anditstransmissionmechanism,focusinginparticularontheroleofmonetarypolicyinthecentre
country,ontheleverageoffinancialintermediaries,creditcreationandcreditflows.PartVargues
thatourfindingsinvalidatethetrilemmaandleadtoadilemma,anirreconcilableduo:
independentmonetarypoliciesarepossibleifandonlyifthecapitalaccountismanaged,directlyor
indirectlyviamacroprudentialpolicies.PartVIdiscussesbrieflythefindingsoftheliteratureonthe
gainstocapitalmobility.

I ) The global financial cycle and international capital flows


A) Characteristics of international capital flows

Figure1apresentsacomprehensiveheatmapofcapitalinflowsbyassetclasses(FDI,portfolio
equity,portfoliodebtandcredit1)intodifferentgeographicalregions(NorthAmerica,Western
Europe,CentralandEasternEurope,LatinAmerica,Asia,EmergingAsia,Africa2).Thedataare
quarterly1990Q12012Q4andcomefromtheIMFInternationalFinancialStatistics.Theheatmap
colourscorrespondtothesignsofthecorrelationsofcapitalflowsacrossregionsandtypesofflows
(greenwhenthecorrelationispositiveandredotherwise).Asevidencedbytheveryclear
preponderanceofthegreencolourintheheatmap,mosttypesofcapitalinflowsarepositively
correlatedwithoneanotherandacrossregions.Thereisaverystrongcommonalityinliabilityflows

1
2

Technicallyweuseotherinvestmentwhichcontainsbankloansandtradecredit.
Forapreciselistofthecountriesincluded,seeAppendixA.

acrosstheworld.TheonlyexceptiontendstobeFDIinflowsinallregionsoftheworldwithportfolio
equityflowsintoAsiaandsomecreditflowsintoAfricaandintoAsia.Thereareinparticularstrong
positivecorrelationsbetweenallthemajorflowsintoNorthAmericaandWesternEurope.
Theheatmapofcapitaloutflowsbyassetclasses(Figure1b)intothesamegeographicalregions
showsanalmostequallystrongpatternofpositivecorrelations.Theonlyareaforwhichcapital
outflowstendtobeoutofsyncisAfrica,andthisistrueacrossfinancialassets.Further,someFDI
outflowsoutofAsiatendalsotocorrelatenegativelywithotherflows.Otherwise,thecomovement
offlowsisalsoverymarked,inparticularoutofthemainfinancialcentres(NorthAmericaand
WesternEurope)forcredit,debtandportfolioequity.
Ontheotherhandtherearenosystematicpatternsintheheatmapofthecorrelationsofnetflows
(Figure1c).Thecommonalityinflowsisthereforeacommonalityingrossinflowsandoutflowsand
isparticularlymarkedforEurope,theUSandalsoLatinAmerica,EmergingAsiaandCentraland
EasternEuropeandsomewhatlessprevalentelsewhereinAsiaandinAfrica.Intermsoftypesof
assets,FDIdoesnotseemhighlycorrelatedwithothertypesofflows.Afewquestionsspringto
mind:doesitmatterifgrossinflowsandoutflowsfollowacommonpatternworldwideifnetflows
donot?Whatarethecharacteristicsofthisglobalcycle?Doweseeevidenceofacycleinasset
pricesandcreditgrowth?
[Figures1a,b,chere]

B) Comovements with global factors


Whatisbehindthosecomovementsingrossflowsandaretheyassociatedwithglobalcreditgrowth
andassetpricefluctuations?Ithaslongbeennotedthatglobalfactorsareamajordeterminantof
internationalcapitalflows.AsobservedbyCalvoetal.(1996)globalfactorsaffectingforeign
investmenttendtohaveanimportantcyclicalcomponent,whichhasgivenrisetorepeatedbooms
andbustsincapitalinflows.Theliteraturehasidentifiedcyclesintherealrateofinterestandinthe
growthrateofadvancedeconomiesasimportantpushfactorsforcapitalflows.Morerecently,
severalstudieshavefoundthatmovementsintheVIX3arestronglyassociatedwithcapitalflows.
TheVIXiswidelyseenasamarketproxyforriskaversionanduncertainty.Thecarrytradeliterature
suggeststhatcarrytradeflowstendtoincreasewhentheVIXislowandtocollapsewhentheVIX
spikes.Morerecently,ForbesandWarnock(2012)andBrunoandShin(2013a)emphasizethesurge
incapitalflowsassociatedwiththeloweringoftheVIX.

TheVIXistheChicagoBoardOptionsExchangeMarketVolatilityIndex.Itisameasureoftheimpliedvolatility
ofS&P500indexoptions.

Figure2plotscapitalinflowsdisaggregatedbyassettypes(FDI,portfolioequity,portfoliodebtand
credit)asaproportionoftheworldGDPfortheperiod1990Q12012Q4anditreportstheVIX
(invertedscale)onthesamegraph.ParticularlystrikingistheprolongedloweringoftheVIXduring
theperiod20022007,duringwhichcapitalinflowssurged.Flowstendtobehighlycorrelatedwith
oneanotherandnegativelycorrelatedwiththeVIX(exceptFDI).Creditinflowsandportfoliodebt
inflowsshowahighdegreeofcomovementovertime(correlationof0.52).Creditflowsarethe
morevolatileandprocyclicalcomponentofallflowswithaparticularlydramaticsurgeintherunup
tothecrisisandanequallydramaticcollapseduringthecrisis.TheircorrelationwiththeVIX
(invertedscale)is0.24onthewhole1990Q12012Q4sample(quarterlydata).
[Figure2here]

InTable1(a),IpresentthecorrelationsbyregionsofeachtypeofinflowswiththeVIX.Capital
inflowsarenegativelycorrelatedwiththeVIX,evenatageographicallydisaggregatedlevel.
Overwhelmingly,duringtranquilperiodscharacterisedbylowVIX,whenuncertaintyandrisk
aversionarelow,capitalinflowsarelarger.Inlinewithaggregatedata,theonlyconsistent
exceptionsareFDIinflowsforwhichthecorrelationwiththeVIXispositiveinallgeographicalareas.
CreditflowsintodevelopedeconomiesinAsiaarealsopositivelycorrelatedwiththeVIX.

Table1(a):UnconditionalcorrelationsofliabilityflowswiththeVIX,quarterly,1990Q12012Q4.
Correlations
inflows/VIX

Equity
FDI
Debt
Credit

North
Latin
America America

Central
Eastern
Europe

Western
Europe

Emerging
Asia
Asia

Africa

0.03

0.29

0.34

0.36

0.11

0.34

0.23

0.09

0.23

0.10

0.09

0.08

0.17

0.06

0.23

0.17

0.28

0.16

0.29

0.08

0.23

0.22

0.10

0.14

0.21

0.24

0.06

0.13

IfIconditiononotherpushfactors(worldshorttermrealinterestrateandworldgrowthrate),a
similarpatternemerges(seeTable1(b)).TheVIXissignificantlynegativelyassociatedwith
fluctuationsincapitalinflows,exceptforFDIinflows.Theresultsaresimilarwithoutflows,bothfor
theunconditionalandfortheconditionalcorrelationsfortheUSandWesternEurope;theyare
weakerfortheothergeographicalareas.Incontrast,andinagreementwithourpreviousresults,
5

thesamepatternofcorrelationsdoesnotholdfornetflows.Idonotreporttheseresultsdueto
spaceconstraints.
Table1(b):ConditionalcorrelationsofliabilityflowswiththeVIX,quarterly,1990Q12012Q4..
Correlations
inflows/VIX

Equity
FDI
Debt
Credit

North
Latin
America America

Central
Eastern
Europe

Western
Europe

Emerging
Asia
Asia

Africa

0.06

0.31

0.32

0.38

0.08

0.34

0.25

0.10

0.35

0.07

0.06

0.08

0.16

0.07

0.30

0.15

0.36

0.23

0.28

0.06

0.22

0.29

0.15

0.16

0.24

0.26

0.09

0.14

InTable1(c),IinvestigatewhetherfluctuationsintheVIXarealsoassociatedwithchangesincredit
creationandleverageusingvariousmeasures.Wereporttheconditionalcorrelationscontrolling
againfortheclassicpushfactors(worldgrowthrateandshorttermrealrate).FollowingForbes
(2012),Imeasureleverageastheratioofprivatecreditbydepositmoneybanksandotherfinancial
institutionstobankdeposits,includingdemand,timeandsavingdepositsinnonbanks.Theprecise
definitionsofleverageanddomesticcreditcanbefoundinAppendixB.
Table1(c)offersthisstrikingfinding:inallareasoftheworld,creditgrowthisnegativelylinkedto
theVIX.CorrelationstendtobethestrongestinNorthAmericaandWesternEurope.Leverageand
leveragegrowtharealsonegativelyrelatedtotheVIXinallthemainfinancialcentres(North
America,WesternEuropeandAsia),whicharethehomesoftheglobalbanks.Butthecorrelationis
incontrastpositiveforleverageandleveragegrowthinLatinAmerica,CEEandAfrica.

Table1(c):ConditionalcorrelationsofcreditandleveragemeasureswiththeVIX,quarterlydata,
19902012.Theconditioningvariablesaretheworldrealshortrateandtheworldgrowthrate.
Correlations
credit/VIX

North
America

Latin
America

Central
Eastern
Europe

Western
Europe

Emerging
Asia

Asia

Africa

Domesticcredit
growth
Leverage

0.26

0.14

0.14

0.11

0.01

0.30

0.01

0.17

0.05

0.30

0.09

0.12

0.25

0.03

0.32

0.06

0.07

0.21

0.06

0.31

0.01

Leverage
growth

Tosumup,thedatashow(i)commonalityincapitalinflowsandoutflowsacrossregions
andtypesofassets(exceptforFDIflowsandasubsetofAsianandAfricanflows).The
commonalityisparticularlystrongforcreditandportfoliodebtinflows(seeFigures1a,b)but
isabsentfornetcapitalflows(Figure1c);(ii)surgesingrosscapitalflowsinperiodoflow
volatilityanddeclineinflowswhentheVIXgoesup(withtheexceptionofFDIflows);alarge
volatilityandprocyclicalityofcreditflows(seeFigure2andTables1a,b);(iii)increasesin
creditgrowtharoundtheworldinparallelwithfallsoftheVIX(seeTable1c);(iv)increasesin
leverageandleveragegrowthinallthemainfinancialcentreswhentheVIXislow(seeTable
1c).

AsnotedinBrunnermeieretal.(2012)andShin(2012),creditflowsgrewataveryfastratein
the20032007precrisisperiodandcollapsedduringthefinancialcrisis.Thepatternofcapital
inflowsandoutflowsfollowsaglobalfinancialcyclewhichissynchronizedwithfluctuationsin
worldmarketriskaversionanduncertaintyasproxiedbytheVIX.Furthermore,itappearsthat
creditcreationinthebankingsectorandleveragearedancingtothesametune.

II) The global financial cycle: The common component in risky asset
prices.

Havingestablishedtheexistenceofaglobalfinancialcycleforcapitalinflowsandoutflows,credit
growthandleverage,itisnaturaltostudyfluctuationsinassetpricesandtoseewhethertheyalso
followtheglobalfinancialcycle.Onemightthinkthatpricesofequitiesaroundtheworld,pricesof
corporatebondsandofcommoditiesreflecttoalargeextentcontinentspecific,sectorspecific,
countryspecificandcompanyspecificfactors.But,asshownbyMirandaAgrippinoandRey(2012)
usingalargecrosssectionof858riskyassetpricesdistributedonthefivecontinents,animportant

partofthevarianceofriskyreturns(25%)isexplainedbyonesingleglobalfactor4.Thisresultis
remarkablegiventhesizeandtheheterogeneityoftheset.Irrespectiveofthegeographicallocation
ofthemarketinwhichtheassetsaretradedorthespecificassetclasstheybelongto,riskyreturns
loadtoalargeextentonthisglobalfactor.
AsapparentfromFigure3,takenfromMirandaAgrippinoandRey(2012),thefactorisconsistent
withthetimingofmajoreventssuchastheGulfWarstartingfromthesecondhalfof1990,9/11and
thefirstquarterof2009whenthemostrecentfinancialcrisisreacheditsclimax.Overall,theindex
goesupfromtheearly1990suntilmid1998whentheRussiancrisiseruptsfollowedbytheLTCM
bankruptcy,andeventuallytheburstingofthedotcombubble.Fromthebeginningof2003,the
indexincreasesrapidlyuntilthebeginningofthethirdquarterof2007.Thisisshortlyafterthe
collapseofthesubprimemarketandcoincideswiththefirstsignalsofincreasedvulnerabilityofthe
financialmarkets.ThehighdegreeofcorrelationoftheglobalfactorwiththeVIXisstriking.
BuildingontheanalysesofAdrianandShin(2008)andDanielsson,ShinandZygrand(2012),
MirandaAgrippinoandRey(2012)proposeastructuralinterpretationofthefactor.Itcanbe
understoodasreflectingthejointevolutionoftheeffectiveriskappetiteofthemarketaswellas
realizedmarketvolatility.Inturntheeffectiveriskappetiteofthemarketcanbeempiricallyrelated
totheleverageofasubsetoffinancialmarketintermediarieswhoseinvestmentstrategyiswell
approximatedbyaVaRconstraint(brokerdealerintheUS,largeEuropeanbankswithsignificant
tradingoperationsand,moregenerally,banksclassifiedinthecapitalmarketcategoryin
Bankscope5).Giventhatstructuralinterpretation,itisnotsurprisingthatthefactorshould
empiricallybeclosely(negatively)correlatedwiththeVIX.AspointedoutinBrunnermeieretal.
(2012)andBorioandDisyatat(2011),thereisapositivefeedbackloopbetweengreatercredit
supply,assetpriceinflation,andacompressionofspreads.Smallerriskpremiumsamplifythecredit
boom.Measuredriskislowandbalancesheetslookhealthierasassetpricesgoup.Byrelaxing
valueatriskconstraints,thiscreatesadditionalspaceforlendingandforcredit,andsoon.This
mechanismisanimportantpositivefeedbackloopbetweencreditcreationandriskspreads.It
contributestotheprocyclicalityofcreditflowsandtheirimportanceinthebuildupoffinancial
fragility.

Figure3:GlobalfactorandVIX.Source:MirandaAgrippinoandRey(2012).

Forasimilarconclusionbasedonadynamicfactoranalysisinthecontextofsovereigncreditrisk,see
Longstaffetal.(2011)
5
SeeMirandaAgrippinoandRey(2012)fordetails.


Tosumup,wehavenowestablishedinflowdata(acrossmosttypesofflowsandregions,butwith
someexceptions)andinpricedata(acrossasectorallyandgeographicallywidecrosssectionofrisky
assetprices)theexistenceofaglobalfinancialcycle.Interestingly,theVIXisapowerfulindexofthe
globalfinancialcycle,whetherforflowsorforreturns.Ouranalysissofaremphasizesstriking
correlationsandpatterns,butcannotaddresscausalityissues.LowvalueoftheVIX,inparticularfor
longperiodsoftime,areassociatedwithabuildupoftheglobalfinancialcycle:morecapitalinflows
andoutflows,morecreditcreation,moreleverageandhigherassetpriceinflation.

III) Capital flows and market sensitivities to the global financial cycle

InthispartIattempttogaugefurthertheimportanceoftheglobalfinancialcyclefordifferentasset
markets(stockprices,houseprices)aswellasfortheleverageoffinancialintermediaries.Having
reportedtheimportanceoftheglobalcycleforthefluctuationsofthesevariablesinthetimeseries
dimension,Istudyinmoredetailsthefactorsaffectingthecrosssectionalsensitivitiesofthese
variablestotheglobalfinancialcycles.Moreprecisely,Ifocushereonthepossibilitythatlarger
volumesanddifferenttypesofcapitalflowsmatterforthesensitivityofnationalmarketstothe
globalfactor.

Iinvestigatewhethercrosssectionally,thesensitivitiesofcountryspecificvariablestotheglobal
(logged)canberelatedtodifferenttypesandintensitiesofcapitalflowsintoeach

factor

market.Thecountryspecificvariables

arestockmarketreturns

,andhousepriceinflation

.Irunthefollowingsetofregressions:

growth

where

,bankingsectorleverage

denotesflowsintocountryi(inflows,outflows,differenttypesofflows)normalizedby

theGDPofcountryi,

isavectorofcontrolvariables(laggedGDPgrowthofcountryiandlagged

nominaleffectiveexchangerateofcountryi).Ialsoinclude
Theinteractionterms

and

,thechangeintheglobalfactor.

aremeanttocapturethepossible

heterogeneoussensitivityofagivenmarkettotheglobalfinancialcycledependingontheintensity
andtypesofcapitalflowsitreceivesorexports.Irunfixedeffectsestimatorswithclustered
standarderrorsbycountryandincludealineartimetrend.Wecheckedthestationarityofvariables
usingaPesarantest.Wehavealargenumberofobservations(between2770and3462depending
onthespecification).Table2reportstheresultsofselectedspecifications.Panel(a)reportsour
resultsforstockmarketreturns(logdifferenceoflocalstockmarketindices),AppendixCpresents
resultsfor2(b)bankingsectorleveragegrowth(differenceofleverageratio)andfor2(c)forhouse
priceinflation(logdifferenceofpropertypriceindices).

Table2(a)Stockmarketreturnssi,tisthedependentvariable(19902013)
flit

CreditL

C.nonbankL

CreditA

DebtL

DebtA

EquityL

flit
*VIXt

0.0952***
(12.64)
0.1743***
(15.14)
0.00
(0.02)

0.0914***
(12.78)
0.1669***
(16.61)
0.0042
(1.17)

0.0951*** 0.0952*** 0.0962*** 0.0959***


(12.25)
(12.66)
(12.34)
(11.96)
0.1759*** 0.1737*** 0.1751*** 0.1758***
(14.71)
(14.81)
(14.54)
(15.51)
0.0002
0.0025*
0.0006*** 0.0010***
(0.99)
(1.98)
(3.54)
(6.63)

flit1
*VIXt1

0.0007*
(1.88)

0.0012**
(2.13)

0.0004***
(5.61)

VIXt

Adj.
R2
N

0.0005*
(1.77)

0.0006**
(2.4)

0
(0.55)

EquityA
0.0995***
(12.94)
0.1744***
(13.39)
0.0016***
(3.2)
0.0001
(0.26)

0.24
0.222
0.234
0.239
0.245
0.254
0.255
3042
3267
3073
2924
2971
2631
2770
Fixedeffectestimator,standarderrorsadjustedforclusteringoncountry,tstatinparentheses.All
specificationsincludethecontrolvariablesandalineartimetrend.Eachcolumncorrespondstoa
differentspecificationoftheflowintheinteractedterm.

10

Panel(a)showsthatstockpricesaresignificantlynegativelyrelatedtotheglobalfactor(theVIX)and
toitsgrowthrate.Creditflowsintoandoutofcountryitendtobeassociatedwithahigher
sensitivityofthestockmarketofcountryitotheglobalfinancialcycle(theinteractiontermis
significantlynegative).But,interestingly,debtoutflowsandespeciallyequityinflowsandoutflows
tendtobeassociatedwithalessersensitivitytotheglobalcycle(interactiontermpositive).Socross
sectionally,justlikeinthetimeseries,creditflowsseemmorestronglyrelatedtotheglobalcycle
thanotherflowsandinparticularthanequityflows.
AscanbeseeninAppendixC,theresultsforbanksleverage(b)andhouseprices(c)aresimilarin
somerespect.Thereisanegativecorrelationofbanksleverageandhousepriceinflationwiththe
VIXandapositivecorrelationwiththegrowthrateoftheVIX.Thereishoweverinthisspecification
nosignofanyflowsassociatedwithahighersensitivityofleverageofbanks(orofhouseprices)to
theglobalfinancialcycle(theinteractiontermisneversignificant).
Onceagain,itisworthemphasizingthattheseregressionsindicatecorrelationsandnotcausality.In
thetimeseriescreditflowsareveryprocyclical.Inthecrosssection,creditflowstendtobe
associatedwithastrongercorrelationofstockmarketreturnswiththeglobalcycle,whileequity
flowstendtobeassociatedwithaweakercorrelation6.

IV) Monetary policy, capital flows and the global financial cycle

A) Recursive VAR analysis

Theglobalfinancialcycleappearsincomovementsofgrossflows,assetprices,leverageandcredit
creation,whichareallcloselylinkedtofluctuationsintheVIX.Butwhatareitsdrivers?
Giventhestrongprocyclicalityofcreditflowsandthewayglobalbanksoperate(e.g.Shin(2012)
andBrunoandShin(2013a))itisnaturaltoinvestigatetheeffectontheglobalfinancialcycleof
refinancingcostsindollars,i.e.FederalReservemonetarypolicy(seeRajan(2006);BorioandZhu
(2008)).Shin(2012)describeshowEuropeanglobalbanksinparticularweremajoractorsin
channellingUSdollarliquidityworldwidebeforethecrisis.ForeignbankbranchesintheUSwere
raisinglargequantitiesoffundsindollarsandtransferringthemtooverseasmarkets.European

Itwouldofcoursebeinterestingtoestablishacausallinkbetweencrossbordercreditflowsandsensitivitytothecycle.
Butforthiswewouldneedinstrumentalvariables.

11

globalbankswerenotonlyintermediatingsavingsbackintheUSmarketbutwerealsoserving
Asian,LatinAmerican,AfricanandMiddleEasternmarkets.Iwillthereforetreattheleverageof
Europeanbanksasakeyvariableoftheanalysis.Thedollaristhemaincurrencyofglobalbanking.
Sincesurgesincapitalflowsespeciallycreditflowsareassociatedwithincreasesinleverage
worldwide,anaturalinterpretationisthatmonetaryconditionsinthecentrecountryare
transmittedworldwidethroughthesecrossbordergrosscreditflows.Itisthereforethosegross
flowsthatshouldbetrackedinordertoassessfinancialfragilityandoverallcreditconditions,as
emphasizedbyBorioandDisyatat(2011),Gourinchas,TruemplerandRey(2012)andObstfeld
(2012).Itisalsoonlybylookingatgrossflowsthatonecankeeptrackofcurrencyandmaturity
mismatchonbalancesheetsoffinancialintermediariesandhouseholds.Bothofthesemismatches
arewellknowncontributorstofinancialinstability.
Thisisofcoursenottosaythatnetflowsareirrelevant:currentaccountimbalancesarekeyforthe
longrunsustainabilityofthenetexternalassetposition,asalongliteratureshows(recently
surveyedinGourinchasandRey(2013)).

Toanalyzethedynamicinteractionbetweenmonetarypolicy,riskaversionanduncertainty,
leverageandcreditflows,IperformarecursiveVARanalysis7.IbuildonthestudyofBekaert,
HoerovaandLoDuca(2012).Theyshowthatmovementsinthefederalfundsratehaveaneffecton
uncertainty(expectedstockmarketvolatility)andriskaversion,twocomponentstheyextractfrom
theVIX.Likethem,IfocusonthedynamiclinksbetweenthefederalfundsrateandtheVIXbutI
alsostudytheirdynamicinterrelationswithcreditcreation,leverageandcreditflows.Iuse
quarterlydatafortheperiod19902012.Iimposecontemporaneousrestrictions(Cholesky)onthe
responsesofthevariables,basedoninstitutionalknowledge.Iorderthevariablessuchthatthefirst
variablecannotrespondtocontemporaneousshocks(withinthequarter)ofanyothervariables,the
secondonecanrespondtocontemporaneousshocksaffectingvariable1butnotanyothersetc
IassumethatGDPandpricesrespondwithalagastheyareslowmoving,whiletheglobalfactor
(VIX)canrespondcontemporaneouslytoanyvariable(andisthereforeorderedlast).Theeffective
FedFundsrate(FFR)isourpenultimatevariable:itcanrespondtoanyvariablewithinthequarter
excepttotheVIX.Financialvariablessuchascredit,flowsandleverageareinbetween:leverageis
orderedimmediatelybeforetheFFR.Iincludethefollowing7variables(inthisorder):USGDP,US
GDPdeflator(GDPDEF),globalcredit(logged)(CREDIT),globalcreditinflows(INFLOWS),European

TheanalysisborrowsfromMirandaAgrippinoandRey(2012)whoprovidesamoredetailedexercise
disentanglingeffectsonmarketeffectiveriskaversionandvolatility.BrunoandShin(2013b)presentasimilarly
inspiredandindependentlydevelopedanalysisbuttheyfocusonthedynamicrelationwiththedollar
exchangerateandtheovershootingpuzzle.

12

banksleverage(definedasthemedianofEUbankleverage)(EULEV),FedFundstargetrate(FFR)and
VIX(logged).8
Inotethatsincewearefirstandforemostinterestedintheimpactoftheshocksinthelastthree
variablesintheVAR(VIX,FFRandleverage),howtheothervariablesareorderedmakesno
differenceforthosethreeshocks(allIneedispartialidentification).Forexample,givenaFFRshock
bothcreditandflowswillstayputinthefirstquarterandthenarefreetoreact,sotherelativeorder
ofthosetwodoesnotmattertoFFR.Thatorderonlymattersinbetweenthetwo,becauseIam
assumingthatiftherewereashockoncrossborderflows,globalcreditwouldtakeaquarterto
reacttothat,butIamnotfocusingonthis.
IuseatwolagVAR,usingtheusualcriteria(BICandLR).Bootstrappedconfidenceintervalsare
computedusing1000replications;lightanddarkgreyshadedareascorrespondto95and86%
confidenceintervalsrespectively.Ireportasubsetofkeyimpulseresponsesinthetext(Figure4a
and4b).ThecompletesetofimpulseresponsefunctionsarereportedinFigure5intheAppendix.

Ourkeyfindingsarethefollowing:
(i)

Anincreaseintheeffectivefederalfundsrate(FFR)leadstoanincreaseintheVIXafter
about5quartersanduntil11quarters.(Figure4a)

(ii)

AnincreaseintheVIXleadstoafallinEuropeanbanksleverage.(Figure4b)

(iii)

AfallintheVIXleadstoanincreaseincrossbordercreditflowsupto6quarters.(Figure
4b)

(iv)

AnincreaseintheVIXleadstoafallinglobaldomesticcreditfrom4quartersonwards.
(Figure4b)

(v)

AnincreaseintheVIXleadstodeclineintheFFR.(Figure4b)

(vi)

AnincreaseintheFFRleadstoafallinEUbankleverageafter15quarters.(Figure5)

(vii)

AnincreaseintheFFRleadstoafallingrosscreditflowsafter12quarters.(Figure5)

(viii)

AnincreaseinEUbanksleverageisassociatedwithanincreaseindomesticcreditfroma
1quarterhorizon.(Figure5)

(ix)

AnincreaseinEUbanksleverageisassociatedwithafallintheVIXafterabout8
quarters.(Figure5)

SeeAppendixBfortheprecisedefinitionsofthevariables.

13

TheVARresultsarethereforeconsistentwiththefollowinginterpretation.WhentheFederalFunds
rategoesdown,theVIXfalls(afterabout5quarters),Europeanbanksleveragerises,asdogross
creditflows(after12quarters).AfallintheVIXleadstoanincreaseinglobaldomesticcreditafter4
quarters.
FurthermoreIfindthatincreasedbankleverageandcapitalflows,aswellascreditexpansion
(thoughforcredititisonlymarginallysignificant)areassociatedwithasubsequentfallintheVIX
index.Thisisconsistentwiththefollowingmechanism:ascreditandcapitalflowsgoup,spreadsfall:
asnotedinparticularbyAdrianandShin(2010),thequasiconstancyofriskweightedassetsinthe
balancesheetofglobalbanks(mostlythebankshavinglargecapitalmarketdivisions)attimeswhen
theunweightedvolumeofassetsrisessubstantiallysuggestsafallinmeasuredriskduringexpansion
times.Whenleverageishighandcreditisabundant,spreadsarecompressedandmeasuredriskis
low.ThistranslatesintoadeclineintheVIX.Thereisthereforeapositivefeedbackloopbetween
loosemonetarypolicy,fallintheVIX,riseincredit,capitalflowsandleverageandfurtherfallinthe
VIX9.
FromFigure5,IalsonotethatanincreaseintheVIXhasasignificantnegativeeffectonGDP(asin
Bloom(2009))andontheGDPdeflator.Asexpected,anincreaseintheFFRratehasadampening
effectonprices.IalsonotethatmonetarypolicyloosenswhentheVIXgoesup(Figure4b).

Figure4a:25bpincreasetotheeffectivefederalfundsrate.

ThisinterpretationaccordswellwiththemicrostudiesofJimenez,Ongena,PeydroandSaurina(2012)onEuropeandata

whofindsthatbanksgrantsmoreloanstoriskierfirmsinalowinterestrateenvironmentandofDellAriccia,Laevenand
Suarez(2013)whohavesimilarfindingsusingUSdata.

14


Figure4b:Responsestoa1%increaseintheVIX.

B) Robustness

Icheckrobustnessontheprecrisissample19902007.Ialsocheckrobustnessbydroppingsomeof
ourvariables(Idropsuccessivelycredit,leverage,flowsonebyone)andbydroppingalagtomake
sureoverfittingisnotanissue.Importantly,twostudieswithadifferentfocusbutsomerelated
results(Bekaertetal.(2012)andBrunoandShin(2013b),allowustoassessfurthertherobustness
ofsomeofthefindings.
Bekaertetal.(2012)decomposetheVIXindexintoacomponentreflectingexpectedstockmarket
volatilityandintoavariancepremiumreflectingriskaversion.Theyrunastructuralfourvariable
VARwithabusinesscycleindicator,thetwocomponentsoftheVIXandtheUSshorttermrealrate
(definedastheFedFundsendofmonthtargetrateminustheCPIannualinflationrate)astheir
benchmark.Theyfindthataloosemonetarypolicyreducesriskaversionanduncertainty;andthat
periodsofhighVIXarefollowedbyloosermonetarypolicy.Theyprovidenumerousrobustness
checks,inparticularwithrespecttothemeasurementofmonetarypolicyshocksandbycomparing
resultsontheprecrisissampleandthewholesample.Theirfindingsarecompatiblewithmyresults
showinganincreaseintheVIXfollowingatighteningintheFFRandalooseningofmonetarypolicy
afteraVIXincrease(seeFigure4a,b).
15

BrunoandShin(2013b)runsa4variablerecursiveVARwithFFR,logVIX,leverageandthereal
effectivedollarexchangerateonquarterlydatafortheperiod19952007.Formonetarypolicy
measurestheyusetherealFFRandtherealeffectiveFFR,growthofUSM1,andtheresidualofa
Taylorrule.TheyuseUSbrokerdealerleverageinsteadofthebroadermeasureofEUleverage.
TheyfindthatapositivemonetarypolicyshockleadstoanincreaseintheVIXafterquarter4;toa
declineinUSbrokerdealerleverageafterabout10quarters;anincreaseintheVIXleadstoadecline
inUSbrokerdealerleverageafterquarter10.Theseresultsarecompatiblewithmyresults(i),(vi),
(ii)withsomedifferencesintiming.Theauthorsalsofind,afteraugmentingtheirVAR,thatan
increaseintheFFRreducescreditflows(intheircasedefinedasthefirstdifferenceofUSdollar
liabilitiesofbankslocatedoutsidetheUS)afterabout7quarters;thatanincreaseintheVIXreduces
flows.Theseadditionalresultsarealsocompatiblewithmine(see(vii),(iii))withsomedifferences
intiming.BrunoandShin(2013b)presentinadditionveryinterestingevidenceonthedollarreal
effectiveexchangeratedynamicsandthedelayedovershootingpuzzle.

C) Economic significance of the results

Aretheshockstothefederalfundsrateanimportantsourceofvariationforthedynamicsofthe
globalfinancialcycle,indexedbytheVIX?
Intheir4variableVAR,BrunoandShin(2013b)findthatshockstotheFFRexplainalmost30%ofthe
varianceoftheVIXathorizonslongerthan10quarters.
Similarlyintheir4variablestructuralVARmodel,Bekaertetal(2012)findthatmonetarypolicy
shocksaccountforover20%ofthevarianceofriskaversionathorizonslongerthan7quarters.They
alsoaccountforacomparablepartofthevarianceofuncertainty.IntheirsixvariableVAR,the
monetarypolicyshockaccountsforabout12%ofthevarianceofriskaversionathorizonslonger
than10quarters.
DependingontheexactspecificationoftheVARanalysis,IfindthatshockstotheFFRexplainfrom
about4%ofthevarianceoftheVIX(inthe7variableVARonthewhole19902012sample)toabout
10%(ina4variableVARonthe19902007sample).Thatnumbergoesupto17%ifIuse,likeBruno
andShin(2013b)USbrokerdealerleverageinsteadoftheEUbankleveragevariable.
Althoughthereissomevarianceintheestimatesdependingonthenumberofvariablesandthe
exactspecificationoftheVAR,theseareeconomicallysignificant,possiblylargeeffects.

16

V) Taking stock: monetary conditions, capital flows and the global


financial cycle

Thereisaglobalfinancialcycleincapitalflows,assetpricesandincreditgrowth.Thiscyclecomoves
withtheVIX,ameasureofuncertaintyandriskaversionofthemarkets.Assetmarketswithmore
creditinflowstendtobemoresensitivetotheglobalcycle.Theglobalfinancialcycleisnotaligned
withcountriesspecificmacroeconomicconditions.Inanumberofcountries,thiscanleadtoexcess
creditgrowth(oralternativelytomonetaryconditionswhicharetootight).Excesscreditgrowthis
oneofthebestpredictorsofcrisis.GourinchasandObstfeld(2012)showthatacrossalltypesof
crisis,threevariablesplayastatisticallyandeconomicallysignificantrole:theratioofdomestic
credittooutput,therealexchangerate,andtheratioofofficialreservestooutput.Schularickand
Taylor(2012)demonstratethatcreditgrowthisapowerfulpredictoroffinancialcrises,suggesting
thatsuchcrisesarecreditboomsgonewrongandthatpolicymakersignorecreditattheirperil.
SimilarfindingsareechoedinLundJensen(2012),whofindsthathighassetpriceinflationis
associatedwithsystemicbankingcrises.Ouranalysisclearlyimpliesthatgrossflows(particularly
creditanddebt)shouldbemonitoredclosely(inparallelwithnetflowswhicharekeyfor
sustainabilityissues)inordertoassessfinancialfragilityandoverallcreditconditions.Itisalsoonly
bylookingatgrossflowsandgrosscrossborderpositions(theentirebalancesheetofcountries)
thatonecankeeptrackofcurrencyandmaturitymismatch.Bothofthesemismatcheshaveproved
tocontributetofinancialinstability10.

Theimportanceoftheglobalfinancialcycleincreatingboomandbustcyclesinemergingmarkets
andadvancedeconomiesalikewithcapitalinflowssurgesgoesbackalongwayandhasbeen
mentionedindifferentcontextsbyDiazAlejandro(1983),Calvoetal.(1996),(identificationofpush
factorsforcapitalflows),EichengreenandPortes(1987),ReinhartandReinhart(2008)(capital
flowsbonanzas),LaneandMcQuade(2012)andmanyothers.Theroleofcrossborderflowsin
disruptingfinancialintermediationintheperiodleadingtothe2008crisisisstressedbyPortes
(2009)andObstfeldandRogoff(2010);ReinhartandRogoff(2009)drawsimilarconclusionsfrom
thehistoricalrecord.

OurVARanalysissuggeststhatoneimportantdeterminantoftheglobalfinancialcycleismonetary
policyinthecentrecountry,whichaffectsleverageofglobalbanks,creditflowsandcreditgrowthin
theinternationalfinancialsystem.Thischannelinvalidatesthetrilemma,whichpostulatesthatin

10

SeeforexampleKalemniOzcanetal.(2012).

17

aworldoffreecapitalmobility,independentmonetarypoliciesarefeasibleifandonlyifexchange
ratesarefloating.Instead,whileitiscertainlytruethatcountrieswithfixedexchangeratescannot
haveindependentmonetarypoliciesinaworldoffreecapitalmobility,myanalysissuggeststhat
crossborderflowsandleverageofglobalinstitutionstransmitmonetaryconditionsglobally,even
underfloatingexchangerateregimes.
Soshouldpolicyrestrictcapitalmobility?

VI) Benefits of international capital flows

Ifrestrictingthemovementofcapitalacrossborderistobeapolicyoption,itspotentialbenefit
shouldbeassessedagainstitscosts.Sowhatdoweknowaboutthegainstointernationalcapital
mobility?

Theliteraturehasattemptedtomeasuregainstofreecapitalmobilitymostlyintwoways:by
calibratingstandardinternationalmacroeconomicmodelsandevaluatingwelfaregainswhengoing
fromautarkytofinanciallyintegratedmarkets;bytestingforgrowtheffectsandbetterrisksharing
(lowervolatility)followingfinancialintegration,usingeitherpaneldataoreventstudies.

A) Calibration of standard models

Theneoclassicalgrowthmodelisbehindmanyofoureconomicintuitionsregardingwhythe
freeflowofcapitalcouldbebeneficial.Withinthismodel,financialintegrationbrings
improvementsinallocativeefficiency(capitalflowstoplaceswiththehighestmarginalproduct)
andbetterrisksharing.Interestingly,evenwithinthatparadigm,realisticcalibrationsindicate
thatgainstendtobesmall.GourinchasandJeanne(2006)show,inthecontextofsmallopen
economiesandinadeterministicsetting,thatgainsaresecondorder.Allthatinternational
financialintegrationdoesistospeeduptransitiontowardthesteadystateoftheeconomy.
Coeurdacieretal.(2013)allowforuncertaintyandestimatewelfaregainsfromallocative
efficiencyandrisksharingtogether,inthecontextofageneralequilibriumneoclassicalgrowth
model.Thewelfaregainsaresmall,eveninsuchaworldwheretheinteractionbetweenthe
precautionarysavingsmotivesandallocativeefficiencyeffectsismodeledexplicitlysothe
twomainchannelsofgainsfromintegrationcanexpressthemselves.Wefindtheyareonthe
18

orderofafewtenthsofapercentofpermanentconsumptionforrealisticcalibrations.

B) Empirical evidence from panel data and event studies

Crossborderinvestmentpositionshaverisenforadvancedeconomiesfrom68%ofGDPin1980to
438%ofGDPin2007;foremergingmarketstheyhavegonefrom35%to73%ofGDPduringthe
sameperiod(LaneandMilesiFerretti2007andLane(2012)).Ifcapitalflowsbringgains,weshould
beobservinglargeeffectsinthedata,duetothesheerscaleoffinancialglobalisationsincethe
1990s.Therearenumerousstudiesthattrytotestforeffectsofinternationalcapitalflowson
growthoronconsumptionvolatility.Surprisingly,theseeffectsarehardtofindinmacroeconomic
data.Asattestedbythemostrecentsurveysreviewingalonglistofempiricalpapers,itishardto
findrobustevidenceofanimpactoffinancialopennessongrowthoronimprovedrisksharing(see
Eichengreen2002;Jeanneetal.2012;Koseetal.2006;Obstfeld2009).Somepaperspointtoward
theexistenceofthresholdeffects:capitalflowsarebeneficialonlyafteracountryhasreacheda
certainamountofinstitutionalorfinancialsectordevelopment(seeBekaertetal.(2005)).Thereare
alsosomedifferencesifonelooksacrossdifferenttypesofcapitalflows:FDIflowsseembetterat
deliveringgrowthandrisksharingbenefitsthanothers.Butthisevidenceisnotveryconclusive
becausethesampleusedoftenmakesadifference(seeJeanneetal.(2012)).Theliteraturebased
oneventstudiesisoftenmorepositive(seeHenry(2007))andpointstowardsafallinthecostof
capitalandincreasedinvestmentatthetimeoffinancialintegration.Butthesimultaneityofother
economicreformsorpoliciesputinplaceatthetimeoffinancialopeningishoweveroftena
concern.Further,fromatheoreticalpointofview,evaluatingwelfaregainsrequirestrackingthe
pathofintegratingeconomiesfromthepointofcapitalaccountintegrationtotheirsteadystate.
Alongsuchpaths,onecanobserveinitialinvestmentincreasesandcurrentaccountdeficitswhich
thenreverselateronascountrieshavetorepaytheirexternaldebt.Welfaregainsalongthesepaths
arefoundtobesmall(seeCoeurdacieretal.(2013)).

Sobothontheempiricalsideandonthecalibrationside,itissofarhardtofindrobustsupport
forlargequantifiablebenefitsofinternationalfinancialintegration.Idonotclaimthatthere
arenobenefitstointernationalfinancialintegration,onlythattheyhavebeenremarkably
elusivesofargiventhescaleoffinancialglobalizationtheworldhasundergone.Inthatlight,it
wouldbeusefultoidentifymorepreciselythechannelsforwhichcapitalflowsmaybe
beneficial.Onepossibilityistolookmorecloselyatpotentialeffectsontotalfactorproductivity

19

ofcertaintypesofflows.Theexistingliteratureonthistopichastodealwithhardidentification
issuesandisalsonotveryconclusive(foradiscussionseeObstfeld2009,p.89).Another
possibilityisthatfinancialFDIfavoursfinancialmarketdeepeningandtherebyimprovesgrowth
prospects11.Yetanotherpossibilityistoinvestigatemorecloselytherisksharingpropertiesof
theexternalbalancesheetofcountriesduringcatastrophiceventssuchasthe20072008global
financialcrisis.Gourinchasetal.(2012)showthatthereweremassivewealthtransfersbetween
theUSandtherestoftheworldwhentheglobalfinancialcrisishit(about2trilliondollars
valuationlossesontheUSnetexternalassetposition,whichisequivalenttoawealthtransfer
totherestoftheworld).TheUS,centreoftheinternationalmonetarysystem,actedasaglobal
insurer.Itiseasytoseehowthisinsurancetransferisimplemented:sinceemergingmarkets
tendtobelonginUSgovernmentdebt(thereserveasset)andshortequityandFDI(andvice
versafortheUnitedStates)12,intimesofcrisisthevalueofalargepartoftheirassets(US
governmentbonds)isstableorevengoesupwhilethevalueoftheirliabilities,consistingof
riskyassets,collapses.Thus,whilelargeexternalbalancesheetscanhelppropagatefinancial
crisis,theycanalsocontributetorisksharingdependingontheirexactstructure.Thisisa
furtherreasonwhymonitoringgrossflowsandgrosspositions(asopposedtoonlynetflowsor
currentaccounts)isessential.

Tosumup,gainstointernationalcapitalflowshaveprovedelusivewhetherincalibratedmodelsor
inthedata,thoughperhapsthisisjustbecausethosegainsarehardtomeasure.Forexample,they
mightoccurthroughimprovementsinTFP,whichwehavenotbeenabletomeasureprecisely(but
thenwhydontweseethemingrowthrates?)ortheymightmanifestthemselvesmainlywhen
largeshockshit.Onethingisclearatthisstage:wecannottakethemforgranted.

VII) Policy options: dealing with the dilemma

Grosscapitalinflows,leverage,creditgrowthandassetpricesdancelargelytothesametune.They
comovewiththeVIX.Thereisaglobalfinancialcycle,whichmaynotbeappropriateforindividual
countries.Symptomscangofrombenigntolargeassetpricebubblesandexcesscreditcreation,a
conditionwhichhasbeenidentifiedrepeatedlyasoneofthebestpredictorsoffinancialcrises.VAR

11

FordetailedworkonoperationsofinternationalfinancialinstitutionsseeinparticularCetorelliandGoldberg
(2012).

12

Gourinchasetal.(2010)showhowtoendogenizesuchasymmetricportfolioswhenthecentrecountryofthe
internationalmonetarysystemismoreriskneutralthantherestoftheworld.

20

analysessuggestmonetaryconditionsaretransmittedfromthemainfinancialcentretotherestof
the world through gross credit flows and leverage, irrespective of the exchange rate regime. This
putsthetraditionaltrilemmaviewoftheopeneconomyintoquestion.Fluctuatingexchangerates
cannotinsulateeconomiesfromtheglobalfinancialcycle,whencapitalismobile.Thetrilemma13
morphs into a dilemma independent monetary policies are possible if and only if the capital
accountismanaged,directlyorindirectly,regardlessoftheexchangerateregime.
Thisimpliesthatgrossflows,particularlycreditflows,areofgreatimportanceforfinancialstability
andhavetobemonitoredcarefully.Itisalsoonlybylookingatgrossflowsandgrosscrossborder
positions(theentirebalancesheetofcountries)thatonecankeeptrackofcurrencyandmaturity
mismatch.Bothofthesemismatcheshaveprovedtocontributetofinancialinstability,timeand
timeagain(seeforexampleFarhietal.(2012)).Oncemore,thisisnottosaynetflowsdonot
matter,astheyareimportantforsustainabilityissues14.
Aswelfaregainsfromcapitalflowscannotbetakenforgranted(thoughthejuryisstillout),we
shouldconsiderthefollowingrangeofoptionstoweakenthepotencyoftheglobalfinancialcycle
andtherebyincreasefinancialstability.Onecould:a)imposetargetedcapitalcontrols;b)actonone
ofthesourcesofthefinancialcycleitself:themonetarypolicyoftheFedandothermaincentral
banks;c)actonthetransmissionchannelcyclicallybylimitingcreditgrowthandleverageduringthe
upturnofthecycleusingnationalpolicies(andpossiblydoingthereverseduringdownturns)i.e.
puttinginplacemacroprudentialpolicies;d)actonthetransmissionchannelstructurallyby
imposingstricterlimitsonleverageforallfinancialintermediaries.

a) Capitalcontrols
Onecouldconsidercapitalcontrolseithercyclicalorpermanenttoinsulatetheeconomyfromthe
globalfinancialcycle.Permanentcapitalcontrolscanbeappliedonsubsetofassetseitheronthe
inflowsideortheoutflowside.Itis,atthisstagehardtoassessrigorouslytheeffectofsuchpolicy
onfinancialstabilityanditssideeffects,as,intherecentperiod,permanentcontrolshavebeen
implementedexclusivelyinasubsetoflowincomecountries,whichhaveveryspecificcharacteristics
(seeKlein(2012)).Overcomingthisselectionissueisamajorchallenge.

13

Alternativelytheimpossibletrinitybecomesanirreconcilableduo.
GourinchasandRey(2007)estimatedthataboutathirdoftheadjustmentoftheUnitedStatestowardsits
longrunbudgetconstraintcamefromvaluationeffectswhiletherestcamefromnetexports(i.e.current
accountandnetflows).

14

21

Temporarycontrols,especiallyoncreditflowsandportfoliodebtwhenthecycleisinaboomphase
couldbeused.Thisoptionhasbeentestedinvariouscontexts:theChileanencaje(19911998);the
2010and2011BraziliantaxesonequityinflowsetcOftenthough,controlshavebeenusedwith
theprimaryaimofpreventingexcessiveappreciationoftheexchangerate.Whencapitalflowsin,an
excessiveexchangerateappreciationmayhurttheexportsector.Asaresult,CentralBankersmay
wishtointerveneontheforeignexchangemarkettokeepthecurrencydown,accumulating
reserves.Theyfacethetradeoffofhigherinflationorincreasedsterilizationcostswithalikelyside
effectofanincreasedinterestrateleadingtofurtherinflows(alsoreinforcedbyexpectationsof
furtherappreciationoftheexchangerate).Taxinginflows,ifeffectivelyimplementedcanactasa
circuitbreakerinsuchasituation.Thereisalivelydebateastherearealsodifficultmeasurement
issueslinkedtoselectionandendogeneityontheeffectivenessandsideeffectsoftemporarycapital
controlsinthiscontext(seeforexampleForbesetal.(2012),ChamonandGarcia(2013),Klein
(2012),Werning(2012)).
Ultimately,sinceinourcontext,itisreallyexcessivecreditgrowththatisthemainissueofconcerns,
capitalcontrolsshouldbeviewedmoreaspartialsubstituteswithmacroprudentialtools.Thelatter
tendtobemoretargeted.Butcapitalcontrolsmaybeappropriateifthereisalotofdirectcross
borderlendingandthebankingsystemcanbecircumvented(seeOstryetal(2011)).Itisimportant
tonotethatmacroprudentialpoliciescanweakenthelinkbetweendomesticmonetarypolicyand
capitalinflows,withouttheimpositionofcapitalcontrols.Forinstance,bypreventingexcessive
creditgrowthinboomtimes,theCentralBankmayreducetheincentiveforbankstoborrow
externallywhendomesticmonetarypolicytightens.

b) Internalisationoftheglobalspilloversofthecentresmonetarypolicy
Onecouldconsideractingononeofthesourcesoftheglobalcycleitself,themonetarypolicystance
inthemainfinancialcentres.MonetaryconditionsinlargefinancialcentressuchastheUSshape
theglobalfinancialcycleviatheendogenousresponseofleverageandtheprocyclicalityofcross
bordercreditflows.Thistransmissionmechanism,unhinderedbytheflexibilityofexchangerate
transformstheimpossibletrinityofafixedexchangerate,independentmonetarypolicyandfree
capitalmobilityintotheirreconcilableduo(adilemma)ofindependentmonetarypolicyandfree
capitalmobility.Thespillovereffectsoflargecountriescentralbankspoliciesontoothercountries
areatpresentnotinternalized.Centralbankersofsystemicallyimportantcountriesshouldpaymore
attentiontotheircollectivepolicystanceanditsimplicationsfortherestoftheworld.Onepractical
wayofimplementingthis,proposedinEichengreenetal.(2012)wouldbeforasmallgroupof
systemicallysignificantcentralbankstomeetregularlyundertheauspicesoftheCommitteeonthe
22

GlobalFinancialSystemoftheBIS.Thisgroupwoulddiscussandassesstheimplicationsoftheir
policiesforgloballiquidity,leverage,andexposures,andtheappropriatenessoftheirjointmoney
andcreditpoliciesfromthepointofviewofglobalprice,output,andfinancialstability.Itcould
issueashortreportdiscussingpolicytradeoffsandinternationalinconsistencies.Withtime,this
shouldatleasthelptounderstandbetterthesecomplexissuesalsobystimulatingmoreresearchin
theseareasandmightencourageCentralBankerstointernalizesomeoftheexternalspilloversof
thepolicies.Thedifficultiesofsuchapolicyoptionareobvious:internationalcooperationon
monetaryspilloversmayconflictwiththedomesticmandatesofcentralbanks.Forexample,
internationalfinancialstabilityanddomesticactivityandinflationtargetsmaybeatoddsatleastin
theshorttomediumrun.Furthermorethemanagementofaggregatedemandinsystemically
importanteconomiesalsohasimportantconsequencesforeconomicactivityintherestofthe
world.Itiseasytoseethatthetradeoffsareextraordinarilycomplex.

c) Mutingthetransmissionchanneloftheglobalcyclebytakingcyclicalmeasures(macro
prudentialmeasures)tolimitexcessivecreditgrowth.
Since,foracountry,themostdangerousoutcomeofinappropriatelylooseglobalfinancial
conditionsisexcessivecreditgrowth,asensiblepolicyoptionistomonitordirectlycreditgrowth
andleverageineachmarket.Recently,muchefforthasgoneintoputtinginplacemacroprudential
measureshavingjustthisgoal.Thearsenalhasseverallayers.Basel3hasacountercyclicalcapital
cushionthatcanbeactivatedinboomtimes.Loantovalueratiosanddebttoincomeratioscanbe
usedinordertorestrictlendingandkeeprealestatepricesincheck.Oneshouldalsomonitorclosely
lendingstandardsandtradingstrategiesduringperiodsofhighcreditgrowth.Thereisawealthof
experiencebeinggatheredaroundtheworldrecentlyonthepracticalimplementationofmacro
prudentialtools(seeforexampletheReserveBankofIndiaortheBankofKoreawhere

macroprudentialmeasureswereimposedincludingleveragecapsonFXderivativespositionand
amacroprudentialstabilitylevyonnoncoreFXliabilitiesofbanks(BankofKoreaReport2013)).
Itisobviousthatcountryspecificinstitutionaldetailsandmarketorganizationmatteralot.A
centralizedrepositoryoftheknowledgeandexperiencegatheredsofarbysupervisorsandcentral
bankerswouldbehighlyvaluable.
Beyondthetools,oneofthebigpracticalissuesistodeterminethetimingofintervention.When
shouldoneactivatecircuitbreakerstocutthepositivefeedbackloopsdescribedinPartIV?
Itisimportant,nottowaittoolong;nottowait,forexample,forthequasicertaintythatthereisa
bubbleinassetpricesorrealestatetointervene.
23

OneoptionistodeviseautomaticrulesbasedonthecredittoGDPratioandactassoonasacertain
thresholdiscrossed(Borioetal.(2011)).Thishastheadvantageofbeingrobusttolobbyingof
interestedparties.Italsoovercomesthewellknownbiastowardsinactionwhengoodtimesare
unfoldingandeveryoneishappilysharingthedividendsofincreasingassetreturns,forgettingabout
theriskbuildingup.
Anotheroptionistostresstestthebalancesheetofthefinancialsector(banksandshadow
banks)veryfrequently,eitherinatargetedwayorbroadly,andjudgewhetherlargebut
realisticchangesinassetpricescouldjeopardizefinancialstability.Stresstestingisadifficult
exerciseingeneralandestimatingsecondroundeffectsisparticularlychallenging.
Furthermore,thisisnotapopularundertakingwithmarketparticipants,asitrequiresregular
inputsontopofmandatoryreportings.Italsorequirescarefulthinkingaboutcommunication
policy(and/orabsoluteconfidentialityasthecasemaybe).Moreover,fiscalbackstopstrategies
areneededtoguaranteethecredibilityofthestresstesting.Noneofthisiseasy.Butdoing
stresstestsregularlyandoften,evenifthisisanimperfectprocess,isanecessarymonitoring
tool.Itimprovestheknowledgeofsupervisorsandinsurestheyareuptodatewiththerecent
marketdevelopments;importantlyitmayalsogiveconstructivechallengestotheinternalrisk
monitoringofinstitutions.Itmayrevealfailuresincorporategovernanceinorganizationswhere
incentivesarenotnecessarilyalignedtokeepriskincheckorwhereinformationisnotavailable
orcentralizedadequately.Itmayevenrevealblindspotsofrisktakingactivitiesoccurring
belowtheradarscreenoftheChiefRiskOfficer.Anaggressivepolicyoffrequentstresstests,
sometargeted,somebroad,canprovideaflexiblewaytotackletheissueofexcessivecredit
growthandleverage.Testsareparticularlyinformativewhentheyindicatewhetherif

currentfinancingconditions,includingtheshadowbankingsector,weretobe
disrupted,financialstabilitycouldbeendangered.15

d) Mutingthetransmissionchannelstructurallybydampeningtheamplificationcapacityof
financialintermediaries:tougherlimitsonleverage.

Weshouldnotforgetthat,usually,thereareanumberofimportantdomesticdistortionsthatinteractwithcapitalflows

15

andcreditgrowth.Inpractice,forpoliticalreasons,weseemanysubsidiestoinvestmentinrealestateandtodebt.These
subsidiesareinstrumentaltocreatingtheinitialbubbleorthebeginningofabubbleinrealestatepricesandinvestment.
Byallmeans,thefirstthingtodoistoremovethesedistortions.Itisalsoimportanttorememberthatexcessiveborrowing
byacountrymeansthatsomeoneelseislendingexcessively:macroprudentialpoliciesapplytolendersjustaswellasthey
applytoborrowers.

24

Attheheartofthetransmissionmechanismdescribedinthispaperistheabilityoffinancial
intermediaries,whetherbanksorshadowbankstoleverageupquicklytoveryhighlevelswhen
financingconditionsarefavourable.Creditisexcessivelysensitivetothefinancingcosts.Istart
again,asinc)withtheusefulobservationthatthemostdangerousoutcomeofinappropriatelyloose
globalfinancialconditionsisexcessivecreditgrowth.Itispossibleinadditiontoorinsteadof
monitoringthecyclicalpropertiesofcreditgrowthtocutstructurallytheabilityoffinancial
intermediariestobeexcessivelyprocyclical.Onepolicyleverseemsparticularlyappropriatefor
doingthis:theleverageratio.Byputtingatougherlimitonleverage,theabilityofthefinancial
systemtoengageinthefeedbackloopsdiscussedinsectionsVandVIwillbecurtailed.Complex
macroprudentialpoliciesdescribedabovearenotnecessarilyrobust.Errorsofjudgementsby
supervisors,ChiefRiskOfficers,CEOsandboardsarepossibleandevenlikelyinourexcessively
complexfinancialandregulatoryenvironment.Tougherleverageratiosmaybeinprincipleawayto
decreasethe(verifiablyhuge)costoftheseerrors,withoutimposinganylargecosts,ifatall,onthe
realeconomy(seeHaldane(2012),Jenkins(2012),AdmatiandHellwig(2013))16.

Conclusion

Ofthesefouroptions,ifhistoryisofanyguidance,puttinginaplaceaneffectiveinternational
cooperationamongthemaincentralbankstointernalisethespilloversoftheirmonetarypolicieson
therestoftheworldseemsoutofreach17.Andtherearesomereasonsforthat:international
cooperationonmonetaryspilloversmayconflictwiththedomesticmandatesofcentralbanks.For
example,internationalfinancialstabilityanddomesticactivityandinflationtargetsmaybeatodds,
atleastintheshorttomediumrun.Furthermorethemanagementofaggregatedemandin
systemicallyimportanteconomieshasimportantconsequencesforeconomicactivityintherestof
theworld.Thisisamajorconsideration.Therestoftheworldcannotatthesametimecomplainof
excessivecapitalinflowsduetoloosemonetarypolicyinthecentrecountriesandwishforahigher
levelofeconomicactivityanddemandstimulusinthesamecountries.Tradeoffsareextraordinarily

16

Ofcourseexcessivecomplexityinregulationhasalsothedownsideoflettingawellresourcedindustryfind
loopholesorcreatethemaswellasencouragingriskybetsguidedbyregulatoryarbitrage.Complexityoften
goeswithlackoftransparencyandheterogeneousimplementation.
17
Policycoordinationwasamajorthemeininternationalmacroeconomicsinthe1980s(see,e.g.Buiterand
Marston(1985)andBryantandPortes(1987)).TheG7summitsof1986(Tokyo)and1987(Venice)emphasized
multilateralsurveillance.Tothisday,howeverneithertheeconomicanalysisnorthepolicypronouncements
havehadanyobservableeffectonactualmonetarypolicies.

25

complexandpolicyactionwillmostlikelyremainbiasedtowardsnationalpriorities.Atransparent
foruminwhichthecollectivemonetarypolicystanceofthesystemicallyimportantcentralbanksis
activelydiscussedandinconsistenciesanalysedwouldreducetheriskofvolatilityincapitalflows18.

References

AdmatiAnatandMartinHellwig(2013)Thebankersnewclothes:whatswrongwithbankingand
whattodoaboutit,PrincetonUniversityPress.

Adrian,TobiasandHyunSongShin(2010)LiquidityandLeverage,JournalofFinancial
Intermediation,19,418437.

Adrian,TobiasandHyunSongShin(2012)ProcyclicalLeverageandValueatRiskFederal
ReserveBankofNewYorkStaffReport338,
http://www.newyorkfed.org/research/staffreports/sr338.html

Adrian,TobiasandNinaBoyarchenko,2013,IntermediaryLeverageCyclesandFinancial
Stability,FederalReserveBankofNewYorkStaffReportNo.567.

BankofKoreaReport(2013),MacroprudentialPolicies:Koreasexperiences.

BekaertGeert,CampbellHarveyandChristianLundblad(2005)Doesfinancialliberalizationspur
growth?JournalofFinancialEconomics77(1),355.

Bekaert,Geert,MarieHoerova,andMarcoLoDuca.(2012).Risk,UncertaintyandMonetaryPolicy,
workingpaper,EuropeanCentralBank.

Borio,ClaudioandPitiDisyatat(2011)Globalimbalancesandthefinancialcrisis:Linkornolink?
BISWorkingPapersNo346http://www.bis.org/publ/work346.pdf

Borio,ClaudioandHaibinZhu(2012)Capitalregulation,risktakingandmonetarypolicy:amissing
linkinthetransmissionmechanism?JournalofFinancialStability,8(4),236251

Brunnermeier,M.DeGregorio,J.etal.2012.BanksandCrossBorderCapitalFlows:Policy
ChallengesandRegulatoryResponses.CommitteeonInternationalEconomicPolicyandReform.

18

Wealsonotethatmonetarypolicyisonlyoneofthedriversoftheglobalfinancialcycleandthatmore
researchshouldbedonetouncoverotherimportantdrivers.

26

Bruno,ValentinaandHyunSongShin(2012a)CapitalFlows,CrossBorderBankingandGlobal
Liquidityworkingpaper,PrincetonUniversity.

Bruno,ValentinaandHyunSongShin(2013b)CapitalFlowsandtheRisktakingchannelof
monetarypolicy,workingpaperPrincetonUniversity.

Buiter,Willem,andRichardMarston(1985),eds.,InternationalEconomicPolicyCoordination,
CambridgeUniversityPress.

Bryant,Ralph,andRichardPortes(1987),eds.,GlobalMacroeconomics:PolicyConflictand
Cooperation,Macmillan.

Calvo,GuillermoA.,LeonardoLeiderman,andCarmenReinhart(1993)CapitalInflowsandReal
ExchangeRateAppreciationinLatinAmerica:TheRoleofExternalFactors,IMFStaffPapers,40(1),
108151.

Calvo,GuillermoA.,LeonardoLeiderman,andCarmenReinhart(1996)CapitalFlowstoDeveloping
Countriesinthe1990s:CausesandEffects,JournalofEconomicPerspectives,10,Spring1996,123
139.

Cetorelli,NicolaandLindaS.Goldberg(2012)BankingGlobalizationandMonetaryTransmission
JournalofFinance67(5),18111843.

ChamonMarcosandMarcioGarcia(2013),CapitalcontrolsinBrazil:Effective?,manuscript,IMF.

CoeurdacierNicolas,HlneReyandPabloWinant(2013)FinancialIntegrationandGrowth
inaRiskyWorld,manuscript,LondonBusinessSchoolandSciencesPo.

DellAriccia,Giovanni,LucLaevenandGustavoSuarez(2013)BankLeverageandMonetary
PolicysRiskTakingChannel:EvidencefromtheUnitedStatesworkingpaper,International
MonetaryFund.

DazAlejandro,CarlosF.GoodByeFinancialRepression,HelloFinancialCrash.Journalof
DevelopmentEconomics19(February1985):124.

Drehmann,M,CBorioandKTsatsaronis(2011):Anchoringcountercyclicalcapitalbuffers:
theroleofcreditaggregates,InternationalJournalofCentralBanking,vol7(4),pp189239
EichengreenBarryetal.(2011)RethinkingCentralBanking,CommitteeonInternationalEconomic
PolicyandReform,BrookingsInstitution.

EichengreenBarry2002,CapitalAccountliberalization:whatdothecrosscountrystudiestellus?
WorldBankEconomicReview15,341366.

EichengreenBarryandRichardPortes(1987),TheAnatomyofFinancialCrises,inRichard
PortesandAlexanderSwoboda,eds.,ThreatstoInternationalFinancialStability,Cambridge
UniversityPress.

FarhiEmmanuel,PierreOlivierGourinchasandHlneRey,2011Reformingthe
InternationalMonetarySystem,CEPRebookandConseildAnalyseEconomique.

27

ForbesKristin(2012)"The"BigC":IdentifyingandMitigatingContagion",36thJacksonHole
Symposium.

ForbesKristin,MarcelFratzscher,ThomasKostkaandRolandStraub(2012)"BubbleThyNeighbor:
DirectandSpilloverEffectsofCapitalControls",NBERWorkingPaper#18052.

Forbes,KristinJ.andFrancisE.Warnock(2012)CapitalFlowWaves:Surges,Stops,Flight
andRetrenchmentJournalofInternationalEconomics88(2):235251.

Geanakoplos,John(2010)TheLeverageCycle.InNBERMacroeconomicsAnnual2009,ed.
DaronAcemoglu,KennethRogoff,andMichaelWoodford,165.Chicago:UniversityofChicagoPress.

Gourinchas,PierreOlivierandOlivierJeanne(2006),Theelusivegainsfrominternational
financialintegration,ReviewofEconomicStudies,73,715741.

Gourinchas,PierreOlivierandMauriceObstfeld(2012)StoriesoftheTwentiethCentury
fortheTwentyFirst,AmericanEconomicJournal:Macroeconomics,4(1),22665.

Gourinchas,PierreOlivier,NicolasGovillotandHlneRey(2010)ExorbitantPrivilegeand
ExorbitantDutyManuscript,LondonBusinessSchool.

Gourinchas,PierreOlivierandHlneRey(2007)InternationalFinancialAdjustmentJournalof
PoliticalEconomy,115(4),665703.

Gourinchas,PierreOlivier,HlneRey,andKaiTruempler(2012)"TheFinancialcrisisandthe
geographyofwealthtransfers."JournalofInternationalEconomics,88(2),266283

Haldane,Andrew(2012)Thedogandthefrisbee,givenattheJacksonHole36theconomicpolicy
symposium.

HeZhiguoandArvindKrishnamurthy.IntermediaryAssetPricing.AmericanEconomic
Review,103(2):732770,2013.

HenryPeter(2007),"Capitalaccountliberalization:theory,evidenceandspeculation,Journal
ofeconomicliterature,vol.45(4),887935.

Jeanne,Olivier,ArvindSubramanianandJohnWilliamson(2012),Whoneedstoopenthecapital
account?,PetersonInstituteforInternationalEconomics.

JenkinsRobert(2012),Letsmakeadeal,speech,BankofEngland,FinancialPolicyCommittee.

Jimenez,G.S.Ongena,J.L.PeydroandJ.Saurina(2012)HazardousTimesforMonetary
Policy:WhatDoTwentyThreeMillionBankLoansSayabouttheEffectsofMonetaryPolicy
onCreditRiskTaking?,Econometricaforthcoming.

KalemniOzcan,EliasPapaioannouandFabrizioPerri(2012),Globalbanksandcrisis
transmission,JournalofInternationaleconomics.

Klein,Michael(2012)CapitalControls:GatesversusWalls,BrookingsPaperson
EconomicActivity.
28


Korinek,Anton.TheNewEconomicsofPrudentialCapitalControls.IMFEconomic
Review59(August2011):523561.

KoseMEshwarPrasadandKenRogoff2006Financialglobalization:areappraisal,IMF
StaffPapers,56,862.

Lane,PhilipandGianMariaMilesiFerretti(2007)."TheExternalWealthofNationsMark
II:RevisedandExtendedEstimatesofForeignAssetsandLiabilities,1970.2004"Journalof
InternationalEconomics,73,223250.

Lane,PhilipandPeterMcQuade(2012),Domesticcreditgrowthandinternationalcapital
flows,mimeo,TrinityCollegeDublin.
Longstaff,Francis,JunPan,LassePedersenandKennethSingleton(2011)Howsovereignis
sovereigncreditrisk?,AmericanEconomicJournal:Macroeconomics3:75103.

LundJensen,Kasper,(2012)MonitoringSystemicRiskBasedonDynamicThresholds,IMF
WorkingPaperNo.12/159.

Mendoza,EnriqueG.andMarcoE.Terrones.AnAnatomyofCreditBooms:Evidencefrom
MacroAggregatesandMicroData.WorkingPaper14049,May2008

MirandaAgrippino,Silvia,Rey,Hlne(2012).WorldAssetMarketsandGlobalLiquidity,presented
attheFrankfurtECBBISConference,February2012,mimeo,LondonBusinessSchool.

Obstfeld,Maurice(2009),InternationalFiannceandgrowthindevelopingcountries:whathavewe
learned?,IMFstaffpapersvol56,n1.

Obstfeld,Maurice(2012a)FinancialFlows,FinancialCrises,andGlobalImbalances,
JournalofInternationalMoneyandFinance,31,469480

Obstfeld,Maurice(2012b)DoestheCurrentAccountStillMatter?,AmericanEconomic
Review,102(3),123.

Obstfeld,MauriceandAlanTaylor(2004)Globalcapitalmarkets:integration,crisisand
growth,CambridgeUniversityPress.

Obstfeld,MauriceandKennethRogoff.GlobalImbalancesandtheFinancialCrisis:Productsof
CommonCauses.InReuvenGlickandMarkM.Spiegel,eds.,AsiaandtheGlobalFinancialCrisis.San
Francisco:FederalReserveBankofSanFrancisco,2010.

Ostry,JonathanD.,AtishR.Ghosh,KarlHabermeier,LucLaeven,MarcosChamon,MahvashS.
Qureshi,andAnnamariaKokenyne.ManagingCapitalInflows:WhatToolstoUse?IMFStaff
DiscussionNoteSDN/11/06,April2011.

Rey,Hlne(2013),Capitalflows:assessingthecosts,huntingforthegains,presentedatthe
IMFresearchconferenceonRethinkingmacroeconomicpolicy,WashingtonDC.

29

Reinhart,CarmenM.andVincentR.Reinhart.CapitalFlowBonanzas:AnEncompassingViewofthe
PastandPresent.InJeffreyA.FrankelandChristopherPissarides,eds.,InternationalSeminaron
Macroeconomics2008.Chicago:UniversityofChicagoPress,2009.

Schularick,MoritzandAlanM.Taylor(2012)CreditBoomsGoneBust:MonetaryPolicy,
LeverageCycles,andFinancialCrises,18702008.AmericanEconomicReview102,102961.

Shin,HyunSong(2012)GlobalBankingGlutandLoanRiskPremiumMundellFleming
Lecture,IMFEconomicReview60(2),155192.

Sims,ChristopherA.1980.MacroeconomicandRealityEconometrica,48,148.

30


AppendixA

Listofcountriesincluded:

North
America
Canada

Latin
America
Argentina

Central&Eastern
Europe
Belarus

Western
Europe
Austria

Emerging
Asia
China

Asia
Australia

US

Bolivia

Bulgaria

Belgium

Indonesia

Japan

Africa
South
Africa

Brazil

Croatia

Cyprus

Malaysia

Korea

Chile

CzechRepublic

Denmark

Thailand

Colombia

Hungary

Finland

New
Zealand

CostaRica

Latvia

France

Ecuador

Lithuania

Germany

Mexico

Poland

Greece

Romania

Iceland

RussianFederation

Ireland

Serbia

Italy

SlovakRepublic

Luxembourg

Slovenia

Malta

Turkey

Netherlands

Norway

Portugal

Spain

Sweden

Switzerland

UK

31

DataonCapitalflows:

Sourceofflowdata:quarterlygrosscapitalinflowsandoutflowsfromtheInternationalMonetary
FundsInternationalFinancialStatistics(accessedthroughIMFwebsiteinMarch2013)for:
PortfolioEquityInflows,OutflowsandNetFlowsconstructedasOutflowsInflows(AssetsLiabilities)
FDIInflows,OutflowsandNetFlows
PortfolioDebtInflows,OutflowsandNetFlows,and
OtherInvestmentInflows,OutflowsandNetFlows
Datatransformations:FlowsarereportedinmillionsofU.S.dollars
IFSdoesnotdifferentiatebetweentruezerosandnotavailables;mostofthetimeswetreatthese
valuesaserrorsandomissions,unlesstheyevidentlyrepresentzeroflows.

MappingoftheflowsfromBPM5(until2004Q4)toBP6(2005Q1onwards)inaccordancetothe
guidelinesofthe6theditionoftheBalanceofPaymentsandInternationalInvestmentPosition
ManualofIMFReconciliationforquarters2005Q12008Q4forwhichthereisdataoverlap.

ConstructionofNetFlowsonlywhendataonInflowsandOutflowsareavailable

WorldGDPGrowth(Quarterly):InternationalMonetaryFundsInternationalFinancialStatistics

(accessedthroughIMFwebsiteinMarch2013).

32


AppendixB
GlobalFactor:commonfactorextractedfromacollectionof858assetpriceseriesspreadoverAsia
Pacific,Australia,Europe,LatinAmerica,NorthAmerica,CommodityandCorporatesamples.For
detailsonextractionandoriginalassetpricesdatasetcompositionpleaserefertoMirandaAgrippino
andRey(2012).
BankingSectorLeverage:constructedastheratiobetweenClaimsonPrivateSectorand
TransferableplusOtherDepositsincludedinBroadMoneyofDepositoryCorporationsexcluding
CentralBanks.DataareinnationalcurrenciesfromtheOtherDepositoryCorporationsSurvey;
MonetaryStatistics,InternationalFinancialStatisticsdatabase.Classificationofdepositswithinthe
formerDepositMoneyBanksSurveycorrespondstoDemand,Time,SavingsandForeignCurrency
Deposits.
EUBankingSectorLeverage:constructedasthemedianBankingSectorLeverageoftheinitial12
EuroAreaCountries(Austria,Belgium,Finland,France,Germany,Greece,Ireland,Italy,Luxembourg,
theNetherlands,PortugalandSpain)andtheUK.
USBrokersDealersFinancialLeverage:constructedastheratioofSecurityBrokerandDealers
FinancialAssetsandTotalLiabilities;FederalReserveBoard;FinancialAccounts,releaseZ.1
DomesticCredit:constructedasthesumofdomesticclaimsofDepositoryCorporationsexcluding
CentralBanks.DomesticclaimsaredefinedasClaimsonPrivateSector,PublicNonFinancial
Corporations,OtherFinancialCorporationsandNetClaimsonCentralorGeneralGovernment
(ClaimslessDeposits);OtherDepositoryCorporationSurveyandDepositMoneyBanksSurvey;
MonetaryStatistics;IFS.Originaldatainnationalcurrencies.
DirectCrossBorderCredit:measuredasdifferenceinclaimsonallsectorsornonbanksectorofa
givencountryofallBISreportingcountriesinallcurrencies;LocationalStatisticsDatabase;
InternationalBankPositionsbyResidence;BIS;Tables7Aand7B.
GlobalInflows:constructedasthesumofdirectcrossbordercredittononbanksectorinthe53
countriessampledforthepaneldataanalysis;listofcountriessampledattheendofthissection.
NominalGDPDatainUSD:originaldatainnationalcurrenciesfromNationalStatisticalOffices;
HaverAnalyticsconversionusingspotendofperiodFXrates.
VIX:endofperiodreadings;ChicagoBoardOptionExchange(CBOE).
StockMarketIndices:endofperiodclosequotes;HaverAnalyticsandGlobalFinancialData.
HousePriceIndices:OECD,BIS.
ExchangeRates:innationalcurrencyperUSDollar;endofperiod;InternationalFinancialStatistics.
NominalEffectiveExchangeRate:BroadEffectiveExchangeRateIndices,BIS
USGDP:RealGrossDomesticProduct(BillionsofChained2005Dollars);BureauofEconomic
Analysis
33

USINFLATION:GrossDomesticProduct:ImplicitPriceDeflator(Index2005=100);Bureauof
EconomicAnalysis
FFRUS:EffectiveFederalFundRates,EndofPeriod(%p.a.);FederalReserveBoard;Selected
InterestRates,releaseH.15

Countriesinthepanel:Argentina,Australia,Austria,Belarus,Belgium,Bolivia,Brazil,Bulgaria,
Canada,Chile,Colombia,CostaRica,Croatia,Cyprus,CzechRepublic,Denmark, Ecuador,Finland,
France,Germany,Greece,HongKong,Hungary,Iceland,Indonesia,Ireland,Italy,Japan,Korea,
Latvia,Lithuania,Luxembourg,Malaysia,Malta,Mexico,Netherlands,NewZealand,Norway,Poland,
Portugal,Romania,Russia,Serbia,Slovakia,Slovenia,SouthAfrica,Spain,Sweden,Switzerland,
Thailand,Turkey,UnitedKingdom,UnitedStates.

34


AppendixC
Table2
Table2(b)BankingSectorLeverageGrowth
flit

CreditL

VIXt
VIXt
flit*VIXt

0.0125***
(3.84)
0.0164***
(4.06)
0.00
(1.28)

C.nonbankL
0.0122***
(3.82)
0.0167***
(4.4)
0.0014
(1.39)

CreditA

,isthedependantvariable(19902013)

DebtL

DebtA

0.0127*** 0.0121*** 0.0127***


(3.87)
(4.02)
(3.87)
0.0171*** 0.0167*** 0.0174***
(4.44)
(4.26)
(4.39)
0
0.0007
0
(0.67)
(1.09)
(0.89)

EquityL

EquityA

0.0140*** 0.0121***
(3.70)
(3.43)
0.0175*** 0.0190***
(4.49)
(4.81)
0
0.0005
(0.36)
(1.24)

flit

0.0002
0.0014
0
0.0001
0.0001
0
0.0003
(0.61)
(1.43)
(0.56)
(0.34)
(0.67)
(0.29)
(1.18)
Adj.R2
0.01
0.008
0.008
0.009
0.008
0.007
0.011
N
2352
2405
2370
2273
2332
2018
2167
Fixedeffectestimator,standarderrorsadjustedforclusteringoncountry,tstatinparentheses.All
1*VIXt1

specificationsincludethecontrolvariablesandalineartrend.

Table2(c)Housepriceinflation
flit

CreditL

VIXt

0.0174***
(3.66)
0.0073***
(2.91)
0.00
(1.04)

isthedependantvariable(19902013)

C.nonbankL
0.0127***
(3.68)
0.0051**
(2.03)
0.001
(0.99)

CreditA

DebtL

DebtA

EquityL

EquityA

0.0170*** 0.0183*** 0.0179*** 0.0166*** 0.0178***


(3.52)
(3.52)
(3.74)
(3.60)
(3.36)
0.0072*** 0.0076*** 0.0082*** 0.0081*** 0.0074***
(2.88)
(2.87)
(3.5)
(4.33)
(2.89)
0.0001
0.0003
0
0
0.0001
(1.07)
(1.16)
(0.3)
(0.59)
(0.65)

VIXt
flit*VI
Xt
flit
0.0001
0.0018
0
0.0001
0
0
0.0001
1*VIX
(1.00)
(1.11)
(0.09)
(0.56)
(0.08)
(0.02)
(0.59)
t1
Adj.
R2
0.07
0.064
0.068
0.066
0.063
0.058
0.076
N
2335
2490
2358
2263
2303
2044
2166
Fixedeffectestimator,standarderrorsadjustedforclusteringoncountry,tstatinparentheses.All
specificationsincludethecontrolvariablesandalineartrend.

35

AppendixD:

Figures1a,b,c:heatmapsofcorrelationsofgrossinflows,grossoutflowsandnetflows

Figure2:plotscapitalinflowsdisaggregatedbyassettypes(FDI,portfolioequity,portfoliodebt
andcredit)asaproportionoftheworldGDPfortheperiod1990Q12012Q4anditreportstheVIX
(invertedscale)onthesamegraph.

Figure5:CompletesetofImpulseresponsefunctionsoftheVAR.

36