P aper No. 907 Banks, shadow banks, and business c

4TheMarketSentimentShock

Thissectionpresentsourmainfindings.Weexaminevariousindicatorswhichreveal theleadingroleofsentimentshocksonEuropeanbusinesscycles.Wethendescribethe mechanismsatplayinourmodelthatexplainthisresult.

4.1ThePredominantMarketSentimentShock

ConsiderFigure2.Toisolatetheeffectsofthemarketsentimentshockontheeconomy, wefeedittoourmodelwhileshuttingoffothershocks.Wethencomparetheoutcome

7ThecountriesareAustria,Belgium,Finland,France,Germany,Greece,Ireland,Italy,Luxembourg, Netherlands,Portugal,andSpain.Asof2019Q4,theyaccountfor97.6percentoftheentire19-country euroarea’sGDP.

8Creditspreadsareonlyavailablefrom2003Q1onward.

12

(solidline)totheactualdata(dashedline).Thekeytakeawayisthatthemodelhit withjustoneimpulsedoesagreatjobatmatchingthehistoricalbehaviorofthemain macroeconomicandfinancialvariables.Thisisallthemorestrikinggiventhatwe estimatetwelveshocksintotal,andonlywhenallofthemareactivedoesthemodel replicatethedataexactly.Thematchisfarfromperfect,though,andleavesampleroom fortheotherforces.Forexample,themodelovershootstheboomineconomicactivity precedingthe2008financialcrisis.Wefindthatcontractionarymonetarypolicyshocks werepullingtheeconomydownduringthatperiod.Themodelalsomissespartof themovementsinhouseholdcredit.There,thehousingshockaccountsformostof thedifference.Notwithstanding,webelieveitisremarkablethatasingledisturbance generatesfactualdynamicsforasmanyrealandfinancialvariables,spanningdifferent sectors,includingpricesandquantities.Thefitforthetwocreditspreads,inparticular, isimpressive.

AnotherindicatoroftheimportanceofmarketsentimentshocksappearsinTable 1.Thetablereportsthecontributionofthedifferentshockstothevarianceinour elevenobservablevariablesatbusiness-cyclefrequency(6-32quarters).According tothiscriterion,themarketsentimentshockisthemostimportantforcedrivingthe businessandfinancialcycle.Itaccountsforthebulkofthevarianceinoutput(55 percent),consumption(49),investment(49),hours(50),thenominalinterestrate(74), thehouseholdspread(89),andthefirmspread(83).Itisalsoasizableimpulse behindhouseprices(31),householdcredit(23),andfirmcredit(32).Theselasttwo variablesaresteeredbydisturbancesspecifictotheirsector,namelythehousingshock forhouseholdcreditandtheentrepreneurialequityshockforbusinesscredit.Also, inflationismostlydrivenbymarkupshocks.Wewillseethatthisvariablebarely respondstoamarketsentimentshockrealization.Thisisconsistentwithevidencein Angeletos,Collard,andDellas(2020)onUSdata,whichwerepeatinthenextsection forEuropeandata,thatprominentbusiness-cycleshocks,whatevertheirnature,donot appeartobeinflationary.

Takentogether,ourresultssuggestthatshockstotherelativereturnbetweenrisky andsafeassetsareamajorsourceofeconomicfluctuationsinEurope.Tothebestof ourknowledge,wearethefirsttoattributesuchaconsistentlyhighshareofvariationin themainmacroeconomicandfinancialaggregatestoasingleforce.

4.2WhyistheMarketSentimentShockSoImportant?

Toanswerthisquestionweturntoimpulseresponses.Figure3plotsthereactionof themodeleconomytoanegativerealizationofthemarketsentimentshock.Thinkof arisk-offmomentinthefinancialsystem.Investorsflytosafetyanddemandahigher yieldontheirriskierinvestments,whichhappentobetheasset-backedsecuritiesissued

13

Table1:VarianceDecomposition,6–32Quarters

Sentiment

ν_t


Technology ε_t,µ_z^∗_,t

Investment ζ_I,t,µ_Υ,t

Household ζ_c,t,ζ_h,t

Firm
λ_pt,γ_t^e,σ_t^e
Policy
g_t,ε_t^p

Output 55 8 16 4 10 7

Consumption 49 13 4 28 5 1
Investment 49 6 29 1 15 0
Hoursworked 50 17 14 4 9 6
Inflation 30 17 1 1 51 1
Nominalrate 74 5 6 2 10 4
Householdcredit 23 20 0 53 3 0
Firmcredit 32 21 8 0 38 0
Householdspread 89 5 0 3 2 0
Firmspread 83 0 3 0 13 0
Houseprice 31 45 6 10 6 1

Note:Thevariancedecompositioniscomputedonbandpass-filtereddatageneratedbythemodelevaluatedatthemodeof theposteriordistribution.

byshadowbanks.Theseinstitutionsareforcedtoscaledowntheirbalancesheet.The priceofsecuritizedloansplungestoclearthemarket.Thistranslatesintolossesfor banks,whichcompensatebyincreasingtheinterestratetheychargetotheirborrowers, householdsandfirmsalike.Theshockthenpropagatestothesesectorsoftheeconomy.

Ontheconsumerside,creditdropsasthespreadwidens.Impatienthouseholdsare forcedtocutonconsumptionandhousingpurchases.Lowerdemandforrealestate reducesthepriceofhousing.Thishurtsthenetworthofborrowers,whobecome morepronetodefault.Banksrespondbyfurtherincreasinglendingrates,exacerbating thesituation.Onthebusinessside,asimilarstorytakesplace.Entrepreneursare forcedtocutoncapitalpurchases.Investmentdrops.Laborbecomessuperfluousand hoursfall.Alowerdemandforcapitalreducesitsprice.Thishurtsthenetworthof entrepreneurs,whoarechargedhigherratesontheirdebt,andmustthereforereduce investmentfurther.Lessconsumptionandinvestmentleadstoafallinoutput.Inflation alsofalls,butquantitativelytheresponseissubdued.Thecentralbankrespondsby loweringthepolicyrate.

Inshort,amarketsentimentshockinourmodelinducesprocyclicalconsumption, investment,employment,inflation,nominalinterestrate,householdcredit,business credit,assetprices;countercyclicalhouseholdandbusinesscreditspreads.These dynamicsarepreciselytheonesweobserveintheEuropeandata.Thatexplainswhy ourempiricalanalysisassignssuchalargeroletomarketsentimentshocksinbusiness andfinancialcycles.
14

Figure3:ResponsetoaNegativeMarketSentimentShock