Demand-Supply imbalance during the Covid-19 pandemic: The role of fiscal policy

Board of Governors of the Federal Reserve System International Finance Discussion Papers ISSN 1073-2500 (Print) ISSN 2767-4509 (Online) Number 1353 August 2022 Demand-Supply imbalance during the Covid-19 pandemic: The role of fiscal policy Franc¸ois de Soyres, Ana Maria Santacreu, and Henry Young Please cite this paper as: de Soyres, Franc¸ois, Ana Maria Santacreu, and Henry Young (2022). “Demand-Supply imbalance during the Covid-19 pandemic: The role of fiscal policy,” International Finance Discussion Papers 1353. Washington: Board of Governors of the Federal Reserve System, https://doi.org/10.17016/IFDP.2022.1353. NOTE: International Finance Discussion Papers (IFDPs) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the International Finance Discussion Papers Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers. Recent IFDPs are available on the Web at www.federalreserve.gov/pubs/ifdp/. This paper can be downloaded without charge from the Social Science Research Network electronic library at www.ssrn.com.

Demand-Supply imbalance during the COVID-19 pandemic: The role of fiscal policy * FrançoisdeSoyres† AnaMariaSantacreu‡ HenryYoung§ FederalReserveBoard FederalReserveBankofStLouis FederalReserveBoard August1,2022 Abstract To mitigate the health and economic fallout from the COVID-19 pandemic, governments worldwideengagedinmassivefiscalsupportprograms.Weshowthatgenerousfiscalsupportcontributedtoanincreaseinthedemandforconsumptiongoodsduringthepandemic, butindustrialproductiondidnotadjustquicklyenoughtomeetthesharpincreaseindemand. Thisimbalancebetweensupplyanddemandacrosscountriesledtohighinflation. Ourfindingssuggestasizableroleforfiscalpolicyinaffectingpricestability. Keywords: Covid,FiscalPolicy,Inflation,SupplyChains. JELClassification: E2,E6,F4 *TheviewsexpressedinthispaperareourownanddonotrepresenttheviewsoftheFederalReserveBankof SaintLouis,theBoardofGovernorsoftheFederalReserve,oranyotherpersonassociatedwiththeFederalReserve System. WethankShaghilAhmed,MatteoIacoviello,SeungJungLeeandMiklosVariforusefuldiscussionsand comments.Allerrorsareourown. †Email:francois.m.desoyres@frb.gov; ‡Email:Ana.M.Santacreu@stls.frb.org; §Email:henrylyoung8@gmail.com. 1

1 Introduction In 1970, Milton Friedman famously said, "Inflation is always and everywhere a monetary phenomenon in the sense that it is and can be produced only by a more rapid increase in the quantity of moneythaninoutput."Therecentconcurrenceofasurgeininflationinmanycountriesaround the world and large fiscal stimulus provided in the face of the COVID-19 pandemic has renewed interest in analyzing the potential role of large fiscal spending as a driver of price increases. Thereisalonglineofliteratureabouttherelationshipbetweenmonetarypolicyandfiscal policy, and the unpleasant arithmetics of government debt monetization. A distinction that has become central in that literature is between fiscal dominance and monetary dominance. Sims(2011)usedthefiscaltheoryofthepriceleveltoremindusthatwhenrational,forwardlookingagentsbelievethatnewlyissuednominalgovernmentdebtisonlypartiallybackedby futuretaxes,debtissuanceisinflationary. AfteracarefulanalysisoffiscalpolicyintheUnited Statesinthe1970s, heconcludesthat"fiscalpolicycanbeaprimarytransmissionmechanismora primary source for changes in the inflation rate." Given the large fiscal support implemented in the face of the COVID-19 pandemic, and associated large increase in public debt, a possible shiftfrommonetarytofiscaldominanceraisestheriskofmorepersistentinflation(Goodhart andPradhan(2021),Cochrane(2021)). In this paper, we bypass the role of government debt and money creation and focus directlyontheassociationbetweenfiscalspendingandaggregatedemand.1 Westartbyusing cross-countrydataandestimatingtheimpactoffiscalsupportoneconomicfluctuationsduring the pandemic. While the successive waves of the pandemic and associated changes in mobility were the main drivers of economic activity throughout 2020 and 2021, we argue thatfiscalstimuluspoliciesmighthaveshapedtheresponseofconsumptionandproduction to mobility changes. Indeed, we show that countries behaved differently to lockdowns and reopenings: Countries with a larger stimulus experienced a smaller consumption decrease when mobility went down, and a stronger rebound in periods of reopening. However, fiscal stimulus did not have any noticeable impact on industrial production. By stimulating demand without boosting supply, our results suggest that fiscal support contributed to increasedexcessdemandpressuresingoodmarkets. Motivated by this observation, and based on the premise that a fiscally induced imbal- 1WepresentedanearlierversionofouranalysisindeSoyresetal.(2022). 2

ance between demand and supply could lead to price tensions, we then move to examine the association between exposure to fiscal stimulus, both domestic and foreign, and excess inflation, defined as inflation in excess of the country-specific pre-pandemic average. Given the delay in transmission and the continued increase in inflation from early 2021 onward and across several waves of the virus, we focus on a cross-sectional analysis. Using data on tradeinvalueadded,weconstructcountry-specificvaluesofexposuretobothdomesticand foreignfiscalstimulus,wherethelatteriscomposedoftwocomponents: (i)a“vertical”component,definedasatrade-weightedaverageofothercountries’stimulusmeasures,and(ii)a “horizontal” component capturing the exposure of each country’s import partners to a third country’sfiscalstimulus.2 We find that excess inflation is significantly correlated to each country’s own domestic stimulus and to various exposures of foreign stimulus. A back-of-the-envelope calculation suggeststhatU.S.fiscalstimulusduringthepandemiccontributedtoanincreaseininflation ofabout2.6percentagepointsintheU.S.,2.3percentagepointsinCanada,and0.6percentage pointsintheUnitedKingdom. Therestofthepaperisorganizedasfollows. Insection2, wepresentsomestylizedfacts about the COVID-19 pandemic and how it impacted economic activity around the world. Section3investigatestheassociationbetweenthesizeoftotalfiscalstimulusandthepathof consumption and industrial production during the COVID-19 pandemic. We find that fiscal support during the pandemic was associated with a boost in goods consumption demand withoutanynoticeableimpactonthesupplyofgoods,hencecontributingtogrowingimbalances between supply and demand in the goods market. In section 4 we turn to the role of domesticandforeignfactorsonacountry’sinflation. Ourresultssuggestthatexcessinflation issignificantlycorrelatedtoeachcountry’sowndomesticstimulusandtovariousexposures of foreign stimulus. Section 5 takes stock of the previous findings and highlights high-level riskstotheinflationoutlookinthenextfewquarters,andsection6concludes. 2Theanalysisofglobalvaluechainsinthetransmissionofinflationisrelatedtomanypapersinrecentyears, includingAueretal.(2019),deSoyresandFrancoBedoya(2019),BaldwinandFreeman(2021)orSantacreuand LaBelle(2022). 3

2 The Covid Crisis and fiscal policy responses around the world The COVID-19 pandemic gave rise to unprecedented global economic conditions. Due to a mix of government-imposed restrictions and voluntary personal decisions, mobility levels, as measured by Google’s geo-location tracking data from smartphones, collapsed in March 2020. Since then, mobility has improved, albeit with some volatility that tracked closely the successivewavesofthepandemic(seeFigure1). Figure1. EvolutionofMobilityduringthePandemic Note:MobilityisasimpleaverageofGoogle’sgrocery,workplace,retail,recreation,andtransportationmobility. Seriesaresmoothedusingaseven-daymovingaverage,andaggregatesareweightedbypopulation.Aggregates areconstructedusingFederalReserveBoardcountryclassifications.AFEcompriseCanada,France,Germany, Italy,Japan,Spain,andtheUnitedKingdom.EMEcompriseArgentina,Brazil,Chile,Colombia,HongKong, Korea,Indonesia,Malaysia,Philippines,Singapore,TaiwanandThailand.SeriesendonMay5,2022. Source:GoogleCommunityMobilityReports. Thesechangesinmobilityaffectedboththesupplyanddemandsideoftheeconomy,hamperingfirms’abilitytoproduce,aswellasconsumers’abilitytoconsume. Onthesupplyside, government-imposed mobility restrictions and personal decisions from workers resulted in adramaticdecreaseinthevolumeofproduction. Onthedemandside, publichealthrestrictions and high uncertainty from both economic and health conditions contributed to a large decreaseintotalrealconsumptionintheearlypartofthepandemic. Consumptionofgoods andservicesbehavedverydifferentlythaninpreviousrecessions(seeFigure2). Inadvanced economies,wherethedataallowustoanalyzerealconsumptionexpendituresbetweengoods andservicesseparately,consumptionofservicesfelldramaticallyandthenstartedrecovering slowly as containment policies eased and vaccines were made widely available. In contrast, 4

goods consumption fell by less during the beginning of the pandemic and experienced a strong recovery thereafter.3 Industrial production, however, was slow to adjust, creating a discrepancy between supply and demand in goods’ markets that likely played a role in the depletionofinventoriesandultimatelyinrecentpricetensions.4 Figure2. RealConsumptionandIndustrialProductionduringthePandemic Note:UnitedStatesandChinaconsumptionseriesendin2022Q1,whereasAdvancedForeignEconomiesand EmergingMarketEconomiesex.Chinaseriesendinin2021Q4.Allindustrialproductionseriesendin2022Q1 andareaggregatedtothequarterlyfrequencybytakingtherespectiveaverageofthemonthlyvalues. AggregatesareconstructedusingrealGDPweights.Chinaconsumptiondataareexpressedinpercapitaterms. Totalindustrialproductionexcludingconstructionseriesareusedwhenpossible,butsomeseriesuse manufacturingindustrialproductioninsteadduetodatalimitations.Goodsconsumptiondataareavailableonly fortheUnitedStatesandAdvancedForeignEconomies. Source:BureauofEconomicAnalysis;OECDNationalQuarterlyAccounts;HaverAnalytics. Inthispaper,wemostlyfocusonthegoodsmarkettoinvestigatetheimbalancebetween 3SeeSantacreuandLaBelle(2022). 4Severalaspectsofthislineofreasoning,includingtheroleofsemiconductorsandtheimportanceofsupplychainbottlenecks,havebeendocumentedinrecentcontributions,suchasAmitietal.(2021),LeiboviciandDunn (2021),SantacreuandLaBelle(2021a),SantacreuandLaBelle(2021b),SantacreuandLaBelle(2022) 5

consumptionandproduction. Intheservicesector,anecdotalevidencesuggeststhattheproductionofmanyserviceswasheavilyhamperedthroughout2020andupuntiltoday–inpart due to the difficulty of finding workers in an environment of health uncertainty and limited daycare options. However, the absence of production data in the service sector precludes us fromquantifyingtheaggregatemismatchbetweensupplyanddemand.5 Tomitigatethehealthandeconomicfalloutofthepandemic,manygovernmentsengaged in massive fiscal support programs. Using the IMF’s World Economic Outlook data for 52 advanced and emerging market economies, we define each country’s fiscal stimulus during the pandemic as the percentage deviation between government spending and the countryspecific pre-pandemic trend. This measure can be constructed for both 2020 and 2021 separately. AsillustratedinFigure3(leftpanel),thecross-sectionalcorrelationbetween2020and 2021 fiscal stimulus is high, which means that generous fiscal support in 2020 is also a good predictoroffiscalstimulusin2021. The right panel of Figure 3 uses the average of 2020 and 2021 values to illustrate the heterogeneityoffiscalsupportacrosscountries.6 Chile, theUnitedKingdom, theUnitedStates, Canada,andJapanareamongthecountriesthatdisplayedthemostgenerousfiscalsupport. Moreprecisely,inoursample,theaverageof2020and2021governmentspendingwas9.45% above each country’s pre-pandemic trend in advanced economies, while it was only 4.95% abovetrendinemergingmarketeconomies. 3 Fiscal Support, Consumption, and Production During Recovery Periods of the Pandemic While the pandemic and associated restrictions were the main drivers of economic fluctuationsoverthepasttwoyears,fiscalstimuluspoliciesmighthaveshapedtheresponseofconsumption and production to lockdowns and consumer preferences, as reflected by changes inmobility. Here, weinvestigatetheassociationbetweenthesizeoftotalfiscalstimulusand thepathofconsumptionandIndustrialProductionduringtheCOVID-19pandemicbyevaluating(i)theelasticityofdemandandsupplytolockdownsandreopenings,asmeasuredby 5Notethat,accordingtotheBureauofLaborStatistics,servicesectorsareonaveragemorelaborintensivethan goodsectors,implyingthatlaborshortagesobservedinseveralcountriescouldhavebeenexacerbatedifdemand forserviceconsumptionhadgrownatafasterrate. 6Tobemoreprecise: theWEOdatahavemuchlargercountrycoverage,andwehaveestimatesofDomestic Fiscal Stimulus for more than 100 countries. However, other data used below restrict our sample size, with coveragerangingfrom23to52countriesdependingontheanalysis. 6

Figure3. PandemicFiscalSupport Note:DeviationfromProjectedSpendingisconstructedbycalculatingthepercentchangebetweeneach government’sfiscalspendingin2020and2021againstaprojectedvalue,respectively.Theprojectedvalueis calculatedbytakingtheaveragefiscalspendinggrowthratefor2015–2019andforecastingoutayearortwo years,dependingonthebaseyear.FederalReserveBoardcountryclassificationsareusedtogroupcountriesinto AdvancedEconomyandEmergingMarketEconomycategories. Source:IMFWorldEconomicOutlookJanuary2022;Authors’calculations. mobilitymovements,and(ii)theimpactoffiscalstimulusonthiselasticity. 3.1 EmpiricalSetupandIdentificationStrategy We use cross-country data to investigate the effect of fiscal stimulus on consumption and on IndustrialProductionduringtheCOVID-19pandemic. Ourconjectureisthatfiscalstimulus supported the increase in consumption during periods of increased mobility but had only a limited impact on countries’ supply as measured by Industrial Production. Our empirical strategyconsistsofprojectingquarterlyrealconsumptionandproductiongrowthonchanges in mobility in the same quarter, as well as on the interaction of these changes with countryspecific fiscal stimulus. Our objective is to analyze how fiscal support affected country-level consumptionandproductioninresponsetomobilityfluctuations. OurfiscalstimulusvariableisconstructedasdiscussedinFigure3andvariesacrosscountries and time. For all quarters of 2020, it takes into account government spending in excess of each country’s pre-pandemic trend. For all quarters of 2021, the variable takes into account fiscal support provided in both 2020 and 2021. This choice allows us to consider, for 7

each country, the whole stimulus implemented throughout the first and second year of the pandemic, which we then relate to the way countries reacted to the different waves of the pandemic for both 2020 and 2021. As widely noted, the transmission from fiscal stimulus to householdconsumptioncantakesometime(seeRamey(2011)orRameyandZubairy(2018) for an extensive discussion of the timing of transmission). As a result, 2020 fiscal support is likelytoimpactconsumptiondecisionsforboth2020and2021,whichimpliesthatoneneeds to take into consideration fiscal stimulus disbursed in 2020 for data in 2021.7 We also note that this construction accounts for the slow unwinding of savings, which may be attributed toconsumersspendingsignificantlymoreoncerestrictionswererelaxedin2021withtheproliferationofvaccines. To account for possible non-symmetric effects during periods of tightening or relaxation ofpublichealthrestrictions,wesplitquarter-on-quartermobilitychangesintotwovariables, separatingpositiveandnegativemovements. Aswillbeclearfromourresultsbelow,suchan asymmetricempiricalspecificationisinlinewiththepatternobservedinthedata.8 Alltold, ourempiricalmodelcanbewrittenas: Cons_Growth = β Mob_Increase +β Mob_Decrease (1) ct 1 ct 2 ct +β Mob_Increase ×Fiscal_Stim. +β Mob_Decrease ×Fiscal_Stim. 3 ct ct 4 ct ct +FE +ε c ct Intheabovespecification, β and β capturetheimpactofmobilitychangesonconsump- 1 2 tion growth for countries without fiscal stimulus. The interaction terms, whose effects are measured by β and β , capture how fiscal support changed the way consumption reacted 3 4 to mobility movements. Recall that our fiscal stimulus variable is country-specific and timeinvariant. Hence,ouridentificationstrategyreliesoncomparingtheelasticityofconsumption to mobility changes for countries with different levels of fiscal support: β will be different 3 fromzeroiffiscalstimulusmodifiesthiselasticityduringmobilitydecrease(i.e.,lockdowns), and β will be different from zero if fiscal stimulus modifies this elasticity during periods of 4 mobilityincrease(i.e.,reopenings). Inaseparatesetofregressions,weperformthesameanalysisusingIndustrialProduction 7Note, however, thatanyfiscalstimulusimplementedin2021cannotaffecttheelasticityofconsumptionto lockdownsin2020. Hence,itwouldnotmakesensetouse2021fiscalsupportforquartersin2020inourpanel dataset. 8WediscussthispointmorepreciselyandpresenttheresultsofthesymmetriccaseinAppendixB.1Table13. 8

growth as the dependent variable, which allows us to assess the effect of fiscal support on thecountry-levelsupplyofgoods. Inourbaselinespecifications,weusecountryfixedeffects to account for observable factors such as heterogeneous trend growth across countries. Our resultsarequalitativelysimilarwiththeadditionofseveralfixedeffects. 3.2 BaselineConsumptionandIndustrialProductionResults OurmainresultsarepresentedinTable1. Wefindthatgovernmentsthatprovidedgenerous fiscalsupportmitigatedthedropingoodsconsumptioninperiodsofmobilitydecrease,while they boosted consumption in periods of increased mobility. The effect of fiscal stimulus on servicesconsumption,however,isinsignificant. Finally,ourresultsrevealthatgenerousfiscal spending did not significantly contribute to supply expansion: Countries with larger fiscal support did not have a significantly different association between mobility and Industrial Production. Inotherwords,supplydidnotadjustquicklyenoughtomeetthesharpincrease indemandforgoods. Looking at the first column in Table 1 reveals that a one standard deviation increase in fiscal stimulus, which is an increase of government spending of about 6.7% compared with the pre-pandemic trend, raises the responsiveness of total consumption to positive mobility movement by about 16 percent. More precisely, for countries without fiscal stimulus, consumption growth increased by 0.25 percentage points in response to a 1 percentage point increase in mobility. For countries with government spending 6.7% above their trend, this elasticityincreasedto0.25+0.04=0.29percentagepoints. Separatinggoodsandservicesconsumptionincolumns2and3,respectively,wenotethat the effect of fiscal stimulus on total consumption is entirely driven by goods consumption, whereaonestandarddeviationincreaseinfiscalsupportisassociatedwitha26percentsurge inthelinkbetweenconsumptiongrowthandmobilityrebound(from0.19to0.19+0.05=0.24), while it reduced the drop in goods consumption during periods of mobility decline by 13 percent(from-0.31to-0.31+0.04=-0.27). In a robustness exercise presented in appendix B, we use an alternative construction of ourfiscalstimulusvariable,whichfocuseson2020governmentspendingonly. Therationale forthisexerciseisthatfiscalsupportin2021mighthavebeendisbursedlateinthatyearand hence would not be relevant for understanding consumption behavior in most of the year. Suchatime-invariantversionofourfiscalstimulusvariableforeachcountrymeansthatthe identificationofinteractiontermsreliesoncomparingcountrieswitheachother, assessingif 9

Table.1. ConsumptionandIP’srelationshipwithmobilitymovements,andcountrylevel’s FiscalStimulus (1) (2) (3) (4) Total Goods Service Industrial Consumption Consumption Consumption Production Mob. Increase 0.25*** 0.19*** 0.32*** 0.20* (0.04) (0.03) (0.06) (0.12) Mob. Decrease -0.38*** -0.31*** -0.43*** -0.64** (0.04) (0.05) (0.06) (0.26) Mob. Increase× 0.04* 0.05*** 0.02 0.00 DomesticFiscalStimulus (0.02) (0.01) (0.03) (0.06) Mob. Decrease× -0.01 0.04*** -0.05* 0.20 DomesticFiscalStimulus (0.02) (0.01) (0.03) (0.15) ✓ ✓ ✓ ✓ CountryFE R2 0.74 0.51 0.67 0.21 Observations 184 184 184 352 NumberofCountries 23 23 23 44 Note: Left-hand-side variables and mobility variables are quarterly growth rates. Domestic Fiscal Stimulus is constructed using a similar definition of deviation from projected spending shown in Figure 3, but for a given country in 2020, the value solely takes the 2020 deviation from projected fiscalspending. However,in2021thevalueisconstructedbyaveragingthe2020and2021valuesof deviation from projected spending. The variable is then standardized by dividing by the variable’s standarddeviation.TheconstantandstandaloneDomesticFiscalStimulusvariablesareomittedfrom thetableforbrevity. Dataextendfrom2020Q1-2021Q4. Standarderrorsareclusteredatthecountry levelandshowninparentheses. **p<0.01,**p<0.05,*p<0.1 countries with large fiscal stimulus in 2020 were associated with different elasticity of consumption to mobility. Results show that, with this specification, all of our results are both qualitativelyandquantitativelysimilartothoseinTable1.9 Insummary,ourresultspointtoanasymmetriceffectoffiscalspendingonhowconsumption reacted to mobility changes. In periods of economic reopening and mobility rebound, fiscalsupportamplifiestheincreaseinconsumption. Inperiodsofmobilitydecline,however, fiscalsupporthelpshouseholds“softentheblow”ofreducedactivity,implyingasmallerdecline in consumption in countries with large fiscal stimulus. Hence, in these periods, fiscal supportisexpectedtocounterbalancetheeffectofmobility, andtheinteractiontermhasthe 9Thisconsistencyisnotsurprisingbecausethecross-sectionalcorrelationbetween2020and2021fiscalstimulusis0.63,whichmeansthatgenerousfiscalsupportin2020isalsoagoodpredictoroffiscalstimulusin2021.On averageinoursampleof52countries,governmentspendingwas6.19%abovepre-pandemictrendin2020,while itwas7.23%aboveofthetrendin2021. 10

oppositesignofthestandalonemobilityvariable. Using our point estimates and country-specific values of fiscal support, Figure 4 quantifiestheroleoffiscalsupportinshapingtheresponseofgoodsconsumptiontofluctuationsin mobilityinseveralgroupsofcountries. Asexpected,theUnitedStatesisoneofthecountries whereconsumptionbehaviorwasthemostimpactedbyfiscalstimulusduringthepandemic, withconsumptiondecreasingsignificantlylesswhenmobilitydropsandreboundingsignificantlymorewhenmobilityincreases. Figure4. ChangeintheGrowthRateofGoodsConsumptioninresponsetomobilitychanges Note:TheAFEaggregatecomprise19countriesandtheEMEaggregatecomprise32countries.Countriesare equallyweightedwithintheaggregates.CountriesareclassifiedusingFederalReserveBoardcountry classifications. Source:Authors’calculations. 3.3 ALookatEmploymentRecovery In order to further explore the finding that fiscal support did not increase the rebound in industrial production during periods of mobility increase, we extend our analysis to study the impact of fiscal stimulus on the labor market, using data on both employment and labor forceparticipation. Wefirstuseemploymentdatafor29countries,separatedbetweengoods and services employment, and assess the effect of fiscal spending on employment growth 11

usingasimilarframework: Emp_Growth = β Mob_Increase +β Mob_Decrease (2) ct 1 ct 2 ct +β Mob_Increase ×Fiscal_Stim. +β Mob_Decrease ×Fiscal_Stim. 3 ct ct 4 ct ct +FE +ε c ct The results are presented in the first three columns in Table 2. We find that employment was less sensitive to mobility than either consumption or Industrial Production. We also noteastrongasymmetryinthesensitivitytomobilitychanges,withemploymentcontracting strongly in periods of restrictions while only picking up modestly in periods of reopening. Related to our previous point, country-specific fiscal stimulus does not appear to have any significant impact on the relationship between employment and observed mobility, which is consistent with our previous observation that while fiscal stimulus boosted consumption, it hadalimitedimpactonproduction. In column 4, we also investigate how labor force participation (LFP) changed during the subsequentwavesofthepandemic,andhowthiselasticitychangedforcountrieswithlarger fiscalsupport. Consistentwithpreviousobservations,LFPdecreasedstronglyduringperiods of lockdowns and recovered only modestly in periods of reopening. However, it is interesting to note that countries with larger fiscal stimulus experienced a smaller decline in LFP in periods of mobility decline, as revealed by the positive and significant coefficient for the interaction between mobility decrease and fiscal support. This result is especially interesting when compared to the insignificant effect of fiscal stimulus on the way employment reacted to mobility changes, as seen in columns 1 to 3. This contrast highlights that while governmentsupportdidnotsignificantlymodifypeople’swillingnessorabilitytoworkduringthe pandemic,itseemstohavehelpedmaintainworkersinthelaborforce. Overall, the sluggish employment recovery suggests lingering labor market constraints. Manypossibleexplanationshavebeendiscussedinrecentmonths,includingchildcareissues, healthconcerns,earlyretirement,greatresignation,lowermigration,etc. 3.4 FromSupplyandDemandImbalancetoBottlenecksandInflation Theprevioussectionhighlightedthatfiscalsupportduringthepandemicboostedgoodsconsumptiondemandwithoutanynoticeableimpactonthesupplyofgoods. Alltold, thelarge 12

Table.2. Employment’srelationshipwithmobilitymovements,andcountrylevel’sFiscal Stimulus (1) (2) (3) (4) Total Goods Service LaborForce Employment Employment Employment Participation Mob. Increase 0.01 0.06 -0.02 0.03 (0.02) (0.06) (0.03) (0.04) Mob. Decrease -0.22*** -0.13*** -0.25*** -0.24*** (0.06) (0.06) (0.08) (0.06) Mob. Increase× 0.01 -0.04 0.03 0.00 DomesticFiscalStimulus (0.02) (0.04) (0.02) (0.02) Mob. Decrease× 0.06 0.00 0.08 0.06** DomesticFiscalStimulus (0.05) (0.04) (0.06) (0.03) ✓ ✓ ✓ ✓ CountryFE R2 0.32 0.19 0.31 0.31 Observations 232 232 232 245 NumberofCountries 29 29 29 31 Note: Left-hand-side variables and mobility variables are quarterly growth rates. Domestic Fiscal Stimulus is constructed using a similar definition of deviation from projected spending shown in Figure 3, but for a given country in 2020, the value solely takes the 2020 deviation from projected fiscalspending. However,in2021,thevalueisconstructedbyaveragingthe2020and2021valuesof deviation from projected spending. The variable is then standardized by dividing by the variable’s standarddeviation.TheconstantandstandaloneDomesticFiscalStimulusvariablesareomittedfrom thetableforbrevity. Dataextendfrom2020Q1-2021Q4. Standarderrorsareclusteredatthecountry levelandshowninparentheses. **p<0.01,**p<0.05,*p<0.1 increase in demand triggered by the fiscal stimulus policy, together with the slow pace of adjustmentinproduction,likelycontributedtothecurrentimbalanceinthegoodsmarket. Thetimingoftransmissionfromfiscalsupporttoinflationis, however, uncertain. Asaggregatedemandwassupportedbygovernmenttransfers,goods-producingfirmsfirststarted todigintoinventoriesandincreaseorderstosuppliers,bothdomesticandforeign. Suchanincreaseindemandcoupledwithlimitedproductionandshippingcapacityfirstledto“supply chainbottlenecks."Asanillustration,theSupplierDeliveryTimecomponentofManufacturing PMIs across 30 countries, plotted in Figure 5, shows the presence of supply chain delays acrosstheworld,especiallyintheUnitedStatesandotheradvancedeconomies.10 10PurchasingManagers’Indices(PMIs)areindicesoftheprevailingdirectionofeconomictrends. Theysummarizewhethermarketconditions,asviewedbypurchasingmanagers,areexpanding,stayingthesame,orcontracting.Inouranalysis,wefocusonthecomponentrelatedtosupplierdeliverytimes. 13

Figure5. SupplierDeliveryTimes’componentofPMIs Note:Valueslargerthan50denotelongersupplierdeliverytimes.AFEandEMEex.Chinaaggregatesare constructedusingbilateralU.S.merchandiseexportweights.DataendinApril2022. Source:S&PGlobal;HaverAnalytics. Ultimately,thesurgeinaggregatedemandcontributedtothesurgeininflation,whichwe discuss in the next section. However, given the delay in transmission and the continued increaseininflationfromearly2021onwardandacrossseveralwavesofthevirus,wefocuson a cross-sectional analysis, instead of a within-country time variation. Indeed, while this sectionwasdevotedtoquarter-on-quarterchangesinmobilityandhowtheyimpacteddemand and supply in countries with different levels of fiscal support, our next section takes stock of the imbalance and investigates the “end result” of this process in terms of inflation, using dataupuntilFebruary2022(which,forreference,ispriortotheinvasionofUkraine). 4 Fiscal Support and Inflation As the pandemic disrupted the economy for longer than many expected, inflation started to displayastrongupwardtrajectory. Figure6plotstheevolutionofbothheadlineandcoreinflationduringthepandemic. AfteraninitialdecreaseinthemidstofthefirstCovid-19wave, bothheadlineandcoreinflationincreasedsteadilyfromtheendof2020andthroughout2021. Thefirstfewmonthsofdatafor2022revealacontinuedsurge,especiallycoreinflationinadvanced economies, which suggests a persistent imbalance between high aggregate demand and constrained aggregate supply. In our subsequent analysis, we focus on inflation data 14

up until the invasion of Ukraine (i.e., until February 2022), in order to avoid taking into account large movements in commodity prices that arose from the war and could have been a confoundingfactorinouranalysis. Figure6. EvolutionofInflationduringtheCovid-19pandemic Note:TheAFEandEMEaggregatesareweightedbyU.S.bilateralimportsharesandcomprisecountries forecastedbytheFederalReserveBoard.TheU.S.seriesendinApril,whereasAFEandEMEaggregatesendin March2022. Source:HaverAnalytics. As we have previously shown, countries with large fiscal support experienced substantial increases in consumption of goods. Moreover, the steep surge in goods consumption in those countries may have also created extra demand in other countries through an increase in demand for imports. This demand surge was met by limited supply capacity and bottlenecks. Indeed,whilebothproduction,transportation,andshippingcapacityhaveadaptedto increasing global value chain participation over the past few decades, the necessary infrastructureappearedtobequiteinelasticintheshortrun. Weinvestigatetheroleofdomesticandforeignfactorsonacountry’sinflationbyconducting a cross-country regression analysis. For each country, we compute a measure of “excess inflation” by taking the February 2022 12-month inflation rate and subtracting the average rate of inflation each country experienced during 2015-2019. We then construct several measuresofexposuretodomesticandforeignfiscalstimulusandprojectexcessinflationonsuch measures. 15

First,DomesticFiscalStimuluscaptureseachcountry’sfiscalsupport,asdiscussedinSection 2. Second, Total Exposure to Foreign Fiscal Stimulus measures a country’s exposure to foreign stimulus and contains two parts: (i) a “vertical” component, defined as a tradeweighted average of other countries’ stimulus measures, and (ii) a “horizontal” component capturingtheexposureofeachcountry’simportpartnerstoathirdcountry’sfiscalstimulus. Intuitively, the United States can be exposed to fiscal stimulus from Canada, both through a high import share (i.e, imported inflation) and through a high export share (i.e., higher demandfromCanada). Thiscapturesverticalforeignexposure. Moreover,thepriceofCanada’s exportstotheUnitedStatesmaybepushedupbyCanada’sexposuretoMexico’sfiscalstimulus. Thiscaptureshorizontalforeignexposure. Inpractice,weusevalue-addedtradedatafromOECD’sTiVAdatabase,whichallowsus to account for both direct and indirect linkages through global value chains. We are focusing on 2018 values, which is the latest available year. For any country c, the mathematical definitionofourforeignexposurevariablescanbewrittenas T +T ∑ c→j j→c VerticalExp. toForeignStim. = FiscalStim. (3) c GDP j j∈Partners(c) c T +T ∑ ∑ j→k k→j HorizontalExp. toForeignStim. = FiscalStim. (4) c GDP k j∈Partners(c)k∈Partners(j)−c c Where T denotes the value-added trade flow from country c to country j. Partners (c) c→j is the set of all trade partners of country c. In the definition of Horizontal Exposure to Foreign Stimulus,notethatthesecondsummationisdoneoveralltradepartnersexceptthecountryc. IncontrasttothepanelanalysispresentedinSection3,weusecross-countryregressionssince thetimingoftransmissionfromfiscalstimulustosurgeinaggregatedemandandultimately to inflation is uncertain. Indeed, fiscal support in any given quarter likely supported consumers’demandbothcontemporaneouslyandinthefollowingquarters,withthetotaleffect of fiscal support accumulating throughout the period where fiscal spending remains above trend. Indeed, households savings increased sharply during the pandemic and remained aboveitspre-pandemiclevelbymid-2022. Our identification is not without limitations and our results should be viewed as illustrative, highlighting perhaps the higher end of potential price pressures from fiscal stimulus duringthepandemic. Wediscusssomeoftheselimitationsinsections4.3to4.5. 16

4.1 BaselineInflationResults OurmainresultsarepresentedinTable3. Wefindthatexcessinflationissignificantlycorrelatedtoeachcountry’sowndomesticstimulusandtovariousexposurestoforeignstimulus. Bothverticalandhorizontalexposuretoforeignstimulusappeartobesignificantlycorrelated with domestic excess inflation, regardless of whether they are taken separately or each used inconjunctionwithdomesticfiscalstimulus,ascanbeseenincolumns2to4. Whentakingall variables into account, as in column 6, only horizontal exposure remains statistically significant. Weinterpretthisfindingwithcaution, sinceverticalandhorizontalexposurevariables arehighlycorrelated. Finally,column7showsthatexcessinflationisalsostronglyrelatedto ourTotalExposurevariable. Table.3. FiscalStimulusandInflation,theroleofbothdomesticandforeignforces (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Headline Headline Headline Headline Headline Headline Headline Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 1.30*** 1.12*** 1.05*** 1.06*** 1.09*** Stimulus (0.37) (0.33) (0.35) (0.35) (0.33) VerticalExposure 1.49*** 1.34*** 0.11 toForeignStimulus (0.44) (0.43) (0.60) HorizontalExposure 1.69*** 1.52*** 1.42** toForeignStimulus (0.42) (0.37) (0.61) TotalExposure 1.44*** toForeignStimulus (0.41) R2 0.19 0.25 0.32 0.39 0.44 0.45 0.42 Observations 52 52 52 52 52 52 52 Note: Excess Headline Inflation is computed by subtracting February 2022 12-month inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation fromprojectedspendingshowninFigure3,butittakestheaverageof2020and2021deviationsfrom projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is thestandardizedweightedaverageofVerticalFiscalStimulusofacountry’stradingpartners. Total ExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. Theconstantisomittedfromthetableforbrevity. Robuststandarderrorsareinparentheses. **p<0.01,**p<0.05,*p<0.1 17

4.2 Back-of-the-EnvelopeCalculation: QuantifyingtheImpactofFiscalStimulus onInflation Inordertogiveamorepracticalinterpretationofourfindings,weuseourpointestimatesto computecountry-specificvaluesforthecontributionoffiscalstimulustoinflation,asshown inFigure7. Theleftpanel(Figure7.A)presentstheimpactofdomesticandforeignexposure on excess inflation for several regions, based on our regression.11 The impact of domestic fiscalstimulusoninflationishighestintheUnitedStatesandinChile. Canada,acountrywith strong trade links with the U.S., features a high level of excess inflation related to exposure to foreign fiscal stimulus. In large economies with limited openness to international trade, such as the United States, domestic stimulus is a more important driver of excess inflation that foreign stimulus. However, more open countries, or countries that used limited fiscal stimulus during the pandemic, are relatively more impacted by exposure to foreign fiscal stimulus. In our sample, domestic stimulus is associated with 2.6 percentage points (pp) in excessinflationintheUnitedStatesand1.1ppinGermany. Intherightpanel(Figure7.B),wedigdeeperintoforeignexposureandderiveameasure of “international spillover” of U.S. fiscal stimulus. In particular, we isolate the share of U.S. stimulus in foreign exposure for several countries and compute the associated excess inflation in those countries. Our estimation implies that U.S. fiscal stimulus was associated with excess inflation of about 2.3 pp in Canada and 0.6 pp in the United Kingdom. For reference, we present the inflation impact of exposure to domestic and foreign fiscal stimulus for all countriesintheappendix. 4.3 Robustness: Addressingtheomittedvariablesbias The estimation relies on the association between excess inflation and exposure to domestic andforeignfiscalstimulus,butitmightbethecasethatcountriesthatengagedinlargerfiscal supportarealsothosethathavebeentheworsthitbythepandemic. Insuchacase,andifthe severityofthepandemicisitselfpositivelycorrelatedwithexcessinflationoverandbeyond theeffectoffiscalsupport,ourresultswouldsufferfromanomittedvariablebias. Moreover, thebiaswouldbepositivebecauseourfiscalstimulusvariablewouldcaptureboththedirect effectofthepandemicandtheeffectofthefiscalspending. Wenowaddresstheseconcerns. Toaddressthisissue,weconstructseveralmeasuresofthe“severity”ofthepandemicfor 11Forthischartandthetableintheappendix,weusethepointestimatesfromTable4,column7. 18

Figure7. FiscalStimulusandExcessInflation Note:TheEuroareacompriseFrance,Germany,Italy,andSpain.EMEscomprise32countriesusingFederal ReserveBoardcountryclassifications.Countriesareequallyweightedwithintheaggregates. Source:Authors’calculations. each country using different approaches and use them as controls in our main specification. First, we use Industrial Production movements as an indicator of the supply-side impact of thehealthrestrictionsandcompute,foreachcountry,thesumofallnegativegrowthratesas aproxyfortheaveragestrengthofproductionreductionforeachcountry. Sinceourdataare quarterly, we take the sum of all quarters when Industrial Production had negative growth. Moreover, because the “severity” of the pandemic is a multidimensional object, we also experimentwithtotalIndustrialProductiongrowthasameasureofproductionrestrictions. Second, we take a broader view of pandemic-related restrictions and use mobility movementsasanindicatoroftheseverityofhealthrestrictions. Hereagain,weconstructtwopossible measures of severity, first taking the sum of all negative growth rates for each country, andthenconstructingtotalgrowth. Table 4 presents our results, showing all possible combinations of our controls. Comparingthefirstcolumn,whichsimplyrestatesourresultsinTable3,toallotherspecificationsin columns2to7, weseethatourresultsareveryrobustandallpointtowardastrongassociationbetweenexcessinflationandexposuretobothdomesticandforeignfiscalstimulus. 19

Table.4. FiscalStimulusandInflation,controllingfortheseverityofthepandemic (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Headline Headline Headline Headline Headline Headline Headline Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 1.09*** 1.18*** 0.98*** 0.90* 1.22*** 1.01*** 1.07** Stimulus (0.33) (0.39) (0.36) (0.47) (0.39) (0.35) (0.43) TotalExposure 1.44*** 1.28*** 1.38*** 1.12*** 1.29*** 1.49*** 1.29*** toForeignStimulus (0.41) (0.41) (0.43) (0.41) (0.41) (0.47) (0.45) Controls SumofNegativeGrowth ✓ ✓ IP ✓ ✓ Mobility TotalGrowth ✓ ✓ IP ✓ ✓ Mobility R2 0.42 0.42 0.39 0.39 0.42 0.40 0.37 Observations 52 46 49 44 46 49 44 Note: Excess Headline Inflation is computed by subtracting February 2022 12-month inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation fromprojectedspendingshowninFigure3,butittakestheaverageof2020and2021deviationsfrom projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is the standardized weighted average of Vertical Fiscal Stimulus of a country’s trading partners. TotalExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. The constant and control variables are omitted from the table for brevity. Robust standard errors are in parentheses. **p<0.01,**p<0.05,*p<0.1 4.4 Robustness: CoreInflation Anadditionalconcernregardingouranalysiscouldbethatheadlineinflationismoredriven by volatile items such as food and energy, which are less affected by country-specific fiscal stimulus and whose prices are fixed at the world level. In such a case, inflation could be driven everywhere by the total world’s stimulus and would not be particularly related to exposure through international trade linkages. To address this issue, we also perform an analysis using core inflation instead of headline inflation. Our results are presented in Table 5. Moreover, similar to the robustness analysis with headline inflation, we also present the 20

Table.5. FiscalStimulusandCoreInflation,theroleofbothdomesticandforeignforces (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Core Core Core Core Core Core Core Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 0.96*** 0.79*** 0.76** 0.76** 0.77** Stimulus (0.30) (0.29) (0.30) (0.31) (0.30) VerticalExposure 0.97*** 0.84** 0.10 toForeignStimulus (0.33) (0.33) (0.59) HorizontalExposure 1.09*** 0.96*** 0.87 toForeignStimulus (0.32) (0.29) (0.59) TotalExposure 0.90*** toForeignStimulus (0.32) R2 0.17 0.22 0.27 0.33 0.37 0.37 0.35 Observations 42 42 42 42 42 42 42 Note: Excess Core Inflation is computed by subtracting February 2022 12-month core inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation fromprojectedspendingshowninFigure3,butittakestheaverageof2020and2021deviationsfrom projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is thestandardizedweightedaverageofVerticalFiscalStimulusofacountry’stradingpartners. Total ExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. Theconstantisomittedfromthetableforbrevity. Robuststandarderrorsareinparentheses. **p<0.01,**p<0.05,*p<0.1 resultsofourestimationusingvaryingsetofcontrols,aspresentedinTable6. Ourmainmessageremainsunchanged: Inallspecifications,wefindthatcoreinflationisstronglyassociated withexposuretobothdomesticandforeignfiscalstimulus. 4.5 OtherRobustnessTests Finally,wealsoperformrobustnessonthedefinitionofour“FiscalStimulus”variable. Inthe baseline results, a country’s fiscal stimulus is constructed using both 2020 and 2021 government spending in excess of each country-specific trend. Such a choice ensures consistency with our panel regressions in section 3. However, given the high savings rate in many advancedeconomiesthroughout2021,onecouldarguethatexcessinflationasofFebruary2022 would be mostly related to the 2020 fiscal stimulus and not related to government spending in2021. Toaddressthisconcern,weranallourspecificationsusingaFiscalStimulusvariable based on the fiscal stimulus in 2020 only. All results are unchanged, which is not surprising 21

Table.6. FiscalStimulusandCoreInflation,theroleofbothdomesticandforeignforceswith controls (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Core Core Core Core Core Core Core Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 0.77** 0.76** 0.75** 0.73* 0.75** 0.79** 0.74** Stimulus (0.30) (0.31) (0.33) (0.38) (0.32) (0.32) (0.35) TotalExposure 0.90*** 0.84** 0.92*** 0.84** 0.83** 0.95*** 0.87** toForeignStimulus (0.32) (0.32) (0.31) (0.32) (0.32) (0.35) (0.36) Controls SumofNegativeGrowth ✓ ✓ IP ✓ ✓ Mobility TotalGrowth ✓ ✓ IP ✓ ✓ Mobility R2 0.35 0.34 0.32 0.30 0.34 0.36 0.32 Observations 42 39 41 38 39 41 38 Note: Excess Core Inflation is computed by subtracting February 2022 12-month core inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation fromprojectedspendingshowninFigure3,butittakestheaverageof2020and2021deviationsfrom projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is the standardized weighted average of Vertical Fiscal Stimulus of a country’s trading partners. TotalExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. The constant and control variables are omitted from the table for brevity. Robust standard errors are in parentheses. **p<0.01,**p<0.05,*p<0.1 giventhehighcorrelationbetween2020and2021fiscalsupport. We also experimented with alternative constructions for the left-hand-side variable in these sets of regressions. Instead of expressing excess inflation as of February 2022, which is ourpreferreddefinition,asitavoidsconsideringtheeffectoftheRussianinvasionofUkraine, we constructed a country-specific average of 2022 inflation using data up until either March orApril(dependingoneachcountry’savailabledata)fromwhichwesubtractedtheaverage pre-pandemic inflation rate. Using this definition of excess inflation, we find that the results arerobustandqualitativelysimilartothoseinTables3and4. Alltold,despitetherobustnesstestsdescribedabove,ouranalysiscouldsufferfromother 22

biases. Forexample, anpositivebiascouldariseinourestimateifthereisapositivecorrelationbetweenfiscalandmonetarypolicysupportacrossoursamplecountries. Insuchacase, our fiscal stimulus variable would capture the effect of both fiscal and monetary policy on inflation, and hence should be interpreted as an upper bound of the effect of fiscal stimulus. Futureworkonthistopicwouldgainfromaddressingsuchconcerns. 5 The Outlook Ahead: Risks Could Come From Services The large swings in goods demand, coupled with limited production and shipping capacity in many countries, gave rise to bottlenecks and ultimately put pressure on goods inflation. Looking at previous recessions, Figure 8 reveals how different the behavior of consumption during the pandemic was from previous recessions. Given data constraints, we focus on the United States (Figure 8.A) and other advanced economies (Figure 8.B). In both cases, it is very clear that services consumption took a very heavy hit throughout the pandemic, while goodsconsumptiondecreasedonlymodestlyinthemidstofthefirstwavebeforerebounding extremelystronglyandremainingabovetrendsincethen. Alltold,rebalancingofconsumptionwasslowerintheUnitedStatescomparedwithmost other advanced economies. While this movement could ease some tensions in the goods market,italsocomeswithnotablerisks. Theservicesectorismorelabor-intensivethanthegoodssector. Indeed,asshowninFigure9, employmentintensityintheservicesector, asmeasuredbytheemploymenttooutput ratio, is more than twice as in the goods sector. Therefore, rebalancing toward services will likely increase labor demand. Given the current tightness in advanced-economy labor markets, increasing labor demand is likely to put pressure on wages and, at the extreme, could createariskofawage-pricespiral. Thatsaid,wagepressuresarecurrentlymoderate,andtotalhoursworkedarestillbelow their 2019 level, suggesting that labor supply has some room for improvement. Assuming rebalancingisdrivenbyareductioninfearrelatedtothepandemic,itcouldbeaccompanied by further easing of labor supply constraints. Ultimately, the impact of rebalancing on inflationwilldependinpartonhowfastlaborsupplyadjuststomeethigherdemandinthelabor market. Shouldtheadjustmentbetooslow,inflationcouldremainhighforlonger. 23

Figure8. RealConsumptionofGoodsandServicesintheUSandotherAFEsinrecentrecessions Note:TheAFEaggregateisconstructedusingGDPweightsandincludestheU.K.,France,Germany,andCanada. Source: BureauofEconomicAnalysis;HaverAnalytics;C.D.HoweInstituteBusinessCycleCouncil;EuroArea BusinessCycleNetwork. 6 Conclusion The COVID-19 pandemic was a very peculiar economic shock, affecting both firms’ ability to produce and consumers’ ability to consume, as has been widely noted. In response to this shock, most governments in advanced economies injected large amounts of money into the economy. This policy was successful at boosting consumption, which, together with a relativelyinelasticsupply,mayhaveledtopricetensions. Thisanalysissuggestsapotentially sizablerolethatfiscalpolicymayhaveplayedincontributingtoupwardpricepressures.12 12Inarecentnewbook,KehoeandNicolini(2021)alsoarguethatmanycountriesinLatinAmericahavebeen experiencingthislinkbetweenfiscalspendingandinflationoverthepastsixdecades. Theirlessonisthatgood 24

Figure9. EmploymentIntensityinGoodsandServices Note:EmploymentIntensityisconstructedbydividing2020employment(thousands)by2020output(billionsof chained2021dollars).Aggregatesareconstructedusingindustryoutputasweights.Thegoodsaggregate comprisemining,construction,manufacturing,andutilities,whiletheservicesaggregatecompriseallother non-agriculturalandnon-governmentsectors. Source:BureauofLaborStatistics. Onecaveattoouranalysisisthat,ifthesizeoffiscalandmonetarystimuluswaspositively correlated across countries, our fiscal stimulus variable could capture some of the effect of monetary policy as well, and hence should be interpreted as an upper bound of the effect of fiscalstimulus. Finally, one should also recognize the positive role played by generous government supportthroughoutthisunprecedentedcrisis. Thelargespendingsupportedastrongeconomic rebound,withbothGDPandemploymentrecoveringataremarkablepace,likelypreventing worseoutcomesdespitethepricepressuresthatmayhaveresultedfromthespending. References Amiti,M.,Heise,S.,andWang,A.(2021). HighImportPricesalongtheGlobalSupplyChain Feed Through to U.S. Domestic Prices. Liberty street economics, Federal Reserve Bank of NewYork. economicpolicyisacombinationofgoodfiscalpolicyandgoodmonetarypolicy. 25

Auer,R.A.,Levchenko,A.A.,andSauré,P.(2019). InternationalInflationSpilloversthrough InputLinkages. TheReviewofEconomicsandStatistics,101(3):507–521. Baldwin,R.andFreeman,R.(2021).Risksandglobalsupplychains: Whatweknowandwhat weneedtoknow. WorkingPaper29444,NationalBureauofEconomicResearch. Cochrane,J.(2021). Theendof"theendofinflation". Thegrumpyeconomistblog. de Soyres, F. and Franco Bedoya, S. (2019). Inflation Dynamics and Global Value Chains. PolicyResearchWorkingPaperSeries9090,TheWorldBank. deSoyres,F.,Santacreu,A.M.,andYoung,H.(2022). Fiscalpolicyandexcessinflationduring Covid-19: a cross-country view. Feds notes, Board of Governors of the Federal Reserve System(U.S.). Goodhart, C. and Pradhan, M. (2021). What may happen when central banks wake up to morepersistentinflation? Voxeucolumn. Kehoe,T.J.andNicolini,J.P.(2021). AMonetaryandFiscalHistoryofLatinAmerica,1960–2017. UniversityofMinnesotaPress. Leibovici, F. and Dunn, J. (2021). Supply Chain Bottlenecks and Inflation: The Role of Semiconductors. EconomicSynopses,(28):1–2. Ramey, V. A. (2011). Identifying Government Spending Shocks: It’s all in the Timing*. The QuarterlyJournalofEconomics,126(1):1–50. Ramey,V.A.andZubairy,S.(2018). GovernmentSpendingMultipliersinGoodTimesandin Bad: EvidencefromUSHistoricalData. JournalofPoliticalEconomy,126(2):850–901. Santacreu,A.M.andLaBelle,J.(2021a). RethinkingGlobalValueChainsDuringCOVID-19: Part1. EconomicSynopses,pages1–2. Santacreu,A.M.andLaBelle,J.(2021b). RethinkingGlobalValueChainsDuringCOVID-19: Part2. EconomicSynopses,pages1–2. Santacreu, A. M. and LaBelle, J. (2022). Supply Chain Disruptions During the COVID-19 Recession. EconomicSynopses,pages1–2. Sims, C. A. (2011). Stepping on a rake: The role of fiscal policy in the inflation of the 1970s. EuropeanEconomicReview,55(1):48–56. 26

A Full Country-Specific Results The table presents our estimate of how fiscal stimulus impacted inflation in all countries. It isbasedonourpointestimatesinTable4,column7,aswellasoncountry-specificvaluesfor theexposuretodomesticandforeignfiscalstimulus. Figure10. Fullcountry-specificresults B Alternative Specifications B.1 Furtherdiscussionofourasymmetricspecification Inlightofourresults,wenowprovidemorediscussionofourmainspecificationin(1),which take an asymmetric view and separate positive and negative mobility movements. To bet- 27

ter understand the value of our asymmetric specification, let us consider what a symmetric/linearmodelwouldimply. Considerthefollowingempiricalmodel: Cons_Growth = β Mobility +β Mobility ×Fiscal_Stim. +FE +ε (5) ct 1 ct 2 ct ct c ct Forthesakeoftheargument: whatwoulditmeanifoneestimates(5)andfindsapositive value for β ? In such a case, countries with large fiscal stimulus would be expected to have 2 larger elasticity of consumption with respect to mobility. In other words: when mobility goes up, consumption is expected to increase more in countries with large fiscal stimulus. Importantly, the symmetric effect would also true: when mobility goes down, consumption is expected to decrease more in countries with large fiscal support – which is a mechanical consequenceofusingalinear/symmetricmodel. Inthiscase,fiscalstimulusalwaysamplifies theeffectofmobility. Overall, in a linear/symmetric regression model, if the "stand-alone" term for mobility (β ) and the interaction between mobility and fiscal stimulus (β ) have the same sign, then 1 2 fiscalstimulusamplifiestheeffectofmobilityregardlessofthedirectionofthemove. Ifβ and 1 β have opposite signs, then fiscal stimulus dampens the effect of mobility on consumption. 2 Ineithercase,theeffectissymmetric,byconstruction. Notethatsuchsymmetryisatoddswithouraboveinvestigation: asweshowinTable1, the stand-alone and interaction terms have the same sign when mobility goes up (implying that consumption goes up with mobility, and it does more so for countries with large fiscal stimulus),whereasthestand-aloneandinteractiontermshaveoppositesignswhenmobility goesdown(implyingthatconsumptiongoesdownwhenmobilitydecreases,butitdoesless soforcountrieswithlargefiscalstimulus).13 Taken together, our results in Table 1 suggest that a linear/symmetric specification as in equation(5)isnotappropriate. Indeed,inoursample,suchalinear/symmetricmodelyields weak and unstable results, as can be seen in Table 7, which is simply the mechanical consequence of the asymmetry we uncover in Table 1: in periods of lockdowns, fiscal stimulus dampened the drop in consumption, while in reopening fiscal stimulus amplified the consumptionrebound. 13Recallthat,inourasymmetricmodel,bothmobilityincreaseandmobilitydecreasevariablesonlytakepositivevalues. Thisenablesustohave"intuitive"coefficientsinthestand-aloneterms: apositivecoefficientassociatedwithmobilityincrease,andanegativecoefficientassociatedwithmobilitydecrease. 28

Table.7. Robustnesstest: ConsumptionandIP’srelationshipwithmobilitymovements,and countrylevel’sFiscalStimulus Alternativespecification: noasymmetry. (1) (2) (3) (4) Total Goods Service Industrial Consumption Consumption Consumption Production Mobility 0.31*** 0.25*** 0.36*** 0.39*** (0.02) (0.02) (0.02) (0.07) Mobility×FiscalStimulus 0.02*** 0.01 0.03*** -0.09** (0.01) (0.01) (0.01) (0.04) ✓ ✓ ✓ ✓ CountryFE R2 0.73 0.60 0.66 0.20 Observations 184 184 184 352 NumberofCountries 23 23 23 44 Note: Left-hand-side variables and mobility variables are quarterly growth rates. Domestic Fiscal Stimulus is constructed using a similar definition of deviation from projected spending shown in Figure 3, but for a given country in 2020, the value solely takes the 2020 deviation from projected fiscalspending. However,in2021thevalueisconstructedbyaveragingthe2020and2021valuesof deviation from projected spending. The variable is then standardized by dividing by the variable’s standarddeviation.TheconstantandstandaloneDomesticFiscalStimulusvariablesareomittedfrom thetableforbrevity. Dataextendfrom2020Q1-2021Q4. Standarderrorsareclusteredatthecountry levelandshowninparentheses. **p<0.01,**p<0.05,*p<0.1 B.2 AlternativeFiscalStimulusdefinition In this section, we revisit all estimations presented in the main text but investigate the case whereourfiscalstimulusvariableisconstructedusing2020fiscalstimulusonly. Resultsshow thatallourfindingsholdwhenusingsuchadefinition. 29

Table.8. Robustnesstest: ConsumptionandIP’srelationshipwithmobilitymovements,and countrylevel’sFiscalStimulus AlternativedefinitionofFiscalStimulus: basedon2020governmentspendingonly. (1) (2) (3) (4) Total Goods Service Industrial Consumption Consumption Consumption Production Mob. Increase 0.25*** 0.19*** 0.31** 0.18 (0.03) (0.03) (0.05) (0.11) Mob. Decrease -0.37*** -0.29*** -0.44*** -0.61** (0.03) (0.04) (0.05) (0.26) Mob. Increase×FiscalStimulus 0.04** 0.06** 0.03 0.02 (0.02) (0.01) (0.03) (0.06) Mob. Decrease×FiscalStimulus -0.01 0.04*** -0.05* 0.19 (0.01) (0.01) (0.03) (0.14) ✓ ✓ ✓ ✓ CountryFE R2 0.73 0.59 0.67 0.18 Observations 184 184 184 352 NumberofCountries 23 23 23 44 Note: Left-hand-side variables and mobility variables are quarterly growth rates. Domestic Fiscal Stimulus is constructed using a similar definition of deviation from projected spending shown in Figure 3, but for a given country, the value solely takes the 2020 deviations from projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. The constant is omitted from the table for brevity and the standalone Domestic Fiscal Stimulus variable is absorbed by country fixed-effects since it is time invariant. Data extend from 2020 Q1-2021Q4. Standarderrorsareclusteredatthecountrylevelandshownareinparentheses. **p<0.01,**p<0.05,*p<0.1 30

Table.9. Robustnesstest: Employment’srelationshipwithmobilitymovements,andcountry level’sFiscalStimulus AlternativedefinitionofFiscalStimulus: basedon2020governmentspendingonly. (1) (2) (3) (4) Total Goods Service LaborForce Employment Employment Employment Participation Mob. Increase 0.01 0.06 0.00 0.02 (0.02) (0.05) (0.03) (0.04) Mob. Decrease -0.21*** -0.12*** -0.24*** -0.24** (0.07) (0.06) (0.09) (0.05) Mob. Increase×FiscalStimulus 0.00 -0.04 0.02 0.01 (0.01) (0.04) (0.02) (0.02) Mob. Decrease×FiscalStimulus 0.05 0.00 0.07 0.06** (0.05) (0.03) (0.07) (0.03) ✓ ✓ ✓ ✓ CountryFE R2 0.35 0.17 0.34 0.30 Observations 232 232 232 245 NumberofCountries 29 29 29 31 Note: Left-hand-side variables and mobility variables are quarterly growth rates. Domestic Fiscal StimulusisconstructedusingasimilardefinitionofdeviationfromprojectedspendingshowninFigure3,butforagivencountry,thevaluesolelytakesthe2020deviationfromprojectedfiscalspending. The variable is then standardized by dividing by the variable’s standard deviation. The constant is omitted from the table for brevity and the standalone Domestic Fiscal Stimulus variable is omitted duetousingcountryfixed-effectssincethevariableisnowbasedon2020governmentspendingonly. Dataextendfrom2020Q1-2021Q4. Standarderrorsareclusteredatthecountrylevelandshownin parentheses. **p<0.01,**p<0.05,*p<0.1 31

Table.10. Robustnesstest: FiscalStimulusandInflation AlternativedefinitionofFiscalStimulus: basedon2020governmentspendingonly. (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Headline Headline Headline Headline Headline Headline Headline Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 0.92** 0.84** 0.78** 0.78** 0.81** Stimulus (0.40) (0.38) (0.34) (0.34) (0.36) VerticalExposure 1.11*** 1.05*** -0.46 toForeignStimulus (0.39) (0.38) (0.56) HorizontalExposure 1.61*** 1.54*** 1.90*** toForeignStimulus (0.35) (0.34) (0.56) TotalExposure 1.26*** toForeignStimulus (0.36) R2 0.10 0.14 0.29 0.22 0.36 0.37 0.27 Observations 52 52 52 52 52 52 52 Note: Excess Headline Inflation is computed by subtracting February 2022 12-month inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation from projected spending shown in Figure 3 but, for a given country, takes the 2020 deviation from projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is thestandardizedweightedaverageofVerticalFiscalStimulusofacountry’stradingpartners. Total ExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. Theconstantisomittedfromthetableforbrevity. Robuststandarderrorsareinparentheses. **p<0.01,**p<0.05,*p<0.1 32

Table.11. Robustnesstest: FiscalStimulusandInflation,withcontrols AlternativedefinitionofFiscalStimulus: basedon2020governmentspendingonly. (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Headline Headline Headline Headline Headline Headline Headline Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 0.81** 0.85** 0.72* 0.67* 0.90** 0.82** 0.82* Stimulus (0.36) (0.39) (0.37) (0.39) (0.39) (0.38) (0.41) TotalExposure 1.26*** 1.11*** 1.12*** 0.85** 1.13*** 1.29*** 1.08*** toForeignStimulus (0.36) (0.36) (0.38) (0.37) (0.36) (0.38) (0.38) Controls SumofNegativeGrowth ✓ ✓ IP ✓ ✓ Mobility TotalGrowth ✓ ✓ IP ✓ ✓ Mobility R2 0.27 0.30 0.26 0.31 0.29 0.27 0.25 Observations 52 46 49 44 46 49 44 Note: Excess Headline Inflation is computed by subtracting February 2022 12-month inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation from projected spending shown in Figure 3 but, for a given country, takes the 2020 deviation from projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is thestandardizedweightedaverageofVerticalFiscalStimulusofacountry’stradingpartners. Total ExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. Theconstantandcontrolsareomittedfromthetableforbrevity. Robuststandarderrorsareinparentheses. **p<0.01,**p<0.05,*p<0.1 33

Table.12. Robustnesstest: FiscalStimulusandCoreInflation AlternativedefinitionofFiscalStimulus: basedon2020governmentspendingonly. (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Core Core Core Core Core Core Core Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 0.75*** 0.66** 0.63** 0.65** 0.65** Stimulus (0.34) (0.32) (0.30) (0.31) (0.32) VerticalExposure 0.72** 0.64** -0.28 toForeignStimulus (0.30) (0.29) (0.46) HorizontalExposure 1.01*** 0.94*** 1.17** toForeignStimulus (0.29) (0.28) (0.47) TotalExposure 0.76** toForeignStimulus (0.30) R2 0.11 0.12 0.23 0.24 0.31 0.32 0.25 Observations 42 42 42 42 42 42 42 Note: Excess Core Inflation is computed by subtracting February 2022 12-month core inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation from projected spending shown in Figure 3 but, for a given country, takes the 2020 deviation from projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is thestandardizedweightedaverageofVerticalFiscalStimulusofacountry’stradingpartners. Total ExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. Theconstantisomittedfromthetableforbrevity. Robuststandarderrorsareinparentheses. **p<0.01,**p<0.05,*p<0.1 34

Table.13. Robustnesstest: FiscalStimulusandCoreInflation,withcontrols AlternativedefinitionofFiscalStimulus: basedon2020governmentspendingonly. (1) (2) (3) (4) (5) (6) (7) Excess Excess Excess Excess Excess Excess Excess Core Core Core Core Core Core Core Inflation Inflation Inflation Inflation Inflation Inflation Inflation DomesticFiscal 0.65** 0.67** 0.61* 0.62* 0.68** 0.79** 0.74** Stimulus (0.32) (0.33) (0.33) (0.34) (0.32) (0.33) (0.35) TotalExposure 0.76** 0.70** 0.72** 0.62* 0.70** 0.81*** 0.73** toForeignStimulus (0.30) (0.29) (0.31) (0.32) (0.29) (0.29) (0.31) Controls SumofNegativeGrowth ✓ ✓ IP ✓ ✓ Mobility TotalGrowth ✓ ✓ IP ✓ ✓ Mobility R2 0.25 0.26 0.26 0.23 0.26 0.29 0.25 Observations 42 39 41 38 39 41 38 Note: Excess Core Inflation is computed by subtracting February 2022 12-month core inflation from its2015-2019average. DomesticFiscalStimulusisconstructedusingasimilardefinitionofdeviation from projected spending shown in Figure 3 but, for a given country, takes the 2020 deviation from projected fiscal spending. The variable is then standardized by dividing by the variable’s standard deviation. Vertical Exposure to Foreign Stimulus is the standardized weighted average of Domestic Fiscal Stimulus of a country’s trading partners, and Horizontal Exposure to Foreign Stimulus is thestandardizedweightedaverageofVerticalFiscalStimulusofacountry’stradingpartners. Total ExposuretoForeignStimulusisthestandardizedsumofVerticalandHorizontalExposure. Theconstantandcontrolsareomittedfromthetableforbrevity. Robuststandarderrorsareinparentheses. **p<0.01,**p<0.05,*p<0.1 35