The Rise of Income and Wealth Inequality of America

Page 1


FALL 2020

NEW YORK UNIVERSITY

SCHOOL OF LAW

“The

Rise of Income and Wealth Inequality of America: Evidence from Distributional Macroeconomic Accounts.”

University of California, Berkeley

October 13, 2020

Via Zoom

Time: 2:00 – 3:50 p.m. EST

Week 8

SCHEDULE FOR FALL 2020 NYU TAX POLICY COLLOQUIUM

(All sessions meet online on Tuesdays, from 2:00 to 3:50 pm EST)

1. Tuesday, August 25 – Steven Dean, NYU Law School. “A Constitutional Moment in Cross-Border Taxation.”

2. Tuesday, September 1 – Clinton Wallace, University of South Carolina School of Law. “Democratic Justice in Tax Policymaking.”

3. Tuesday, September 8 – Natasha Sarin, University of Pennsylvania Law School. “Understanding the Revenue Potential of Tax Compliance Investments.”

4. Tuesday, September 15 – Adam Kern, Princeton Politics Department and NYU Law School. “Illusions of Justice in International Taxation.”

5. Tuesday, September 22 – Henrik Kleven, Princeton Economics Department. “The EITC and the Extensive Margin: A Reappraisal.”

6. Tuesday, September 29 – Leandra Lederman, Indiana University Maurer School of Law. “Of Risks and Remedies: Best Practices in Tax Rulings Transparency.”

7. Tuesday, October 6 – Daniel Shaviro, NYU Law School. “What Are Minimum Taxes, and Why Might One Favor or Disfavor Them?”

8. Tuesday, October 13 – Gabriel Zucman, University of California, Berkeley. “The Rise of Income and Wealth Inequality in America: Evidence from Distributional Macroeconomic Accounts.”

9. Tuesday, October 20 – Michelle Layser, University of Illinois College of Law. “How Place-Based Tax Incentives Can Reduce Economic Inequality.”

10. Tuesday, October 27 – Steve Rosenthal, Urban-Brookings Tax Policy Center. “Tax Implications of the Shifting Ownership of U.S. Stock.”

11. Tuesday, November 10 – Owen Zidar, Princeton Economics Department. “The Tax Elasticity of Capital Gains and Revenue-Maximizing Rates.”

12. Tuesday, November 17 – Abdoulaye Ndiaye, NYU Stern Business School. “Redistribution With Performance Pay.”

13. Tuesday, November 24 – Lilian Faulhaber, Georgetown Law School. “Searching for Coherence: The Overuse of Excess Returns and Excess Profits.”

14. Tuesday, December 1 – Erin Scharff, Arizona State Sandra Day O’Connor College of Law. “Revisiting Local Income Taxes.”

TheRiseofIncomeandWealthInequalityinAmerica: EvidencefromDistributionalMacroeconomicAccounts

August26,2020

Abstract

ThispaperstudiesinequalityinAmericathroughthelensofdistributionalmacroeconomic accounts—comprehensivedistributionsoftheaggregateamountofincomeandwealth recordedintheofficialmacroeconomicaccountsoftheUnitedStates.Weusethesedistributionalmacroeconomicaccountstoquantifytheriseofincomeandwealthconcentration sincethelate1970s,thechangeintaxprogressivity,andthedirectredistributiveeffects ofgovernmentinterventionintheeconomy.Between1978and2018,theshareofpre-tax incomeearnedbythetop1%rosefrom10%to19%andtheshareofwealthownedbythe top0.1%rosefrom7%to19%.In2018,thetaxsystemwasregressiveatthetop-end;the top400wealthiestAmericanspaidaloweraveragetaxratethanthemacroeconomictax rateof28%.Weconfrontourmethodsandfindingswiththoseofotherstudies,pinpoint theareaswhereadditionaldataandmoreresearchisneeded,anddescribehowadditional datacollectioncouldimproveinequalitymeasurement.

∗ EmmanuelSaezandGabrielZucmanareProfessorsofEconomics,bothattheUniversityofCalifornia, Berkeley,California.Theiremailaddressesaresaez@econ.berkeley.eduandzucman@berkeley.edu.Wethank editorsGordonHansonandEnricoMoretti,andTimTaylorfordetailedcomments.FundingfromtheCenterfor EquitableGrowthatUCBerkeley,theSandlerfoundation,andtheStonefoundationisthankfullyacknowledged.

1Introduction

Forthemeasurementofinequalityofincomeandwealth,thereisnoequivalentofGrossDomestic Productstatistics—thatis,nogovernment-runstandardized,documented,continuallyupdated, andbroadlyrecognizedmethodologysimilartothenationalaccountswhicharethebasisfor GDP.Startinginthemid-2010s,wehaveworkedtoaddressthisshortcomingbydeveloping “distributionalnationalaccounts,”statisticsthatprovideconsistentestimatesofinequality capturing100percentoftheamountofnationalincomeandhouseholdwealthrecordedinthe officialnationalaccounts.

Thiseffortismotivatedbythelargeandgrowinggapbetweentheamountofincomerecorded inthedatasetstraditionallyusedtostudyinequality—householdsurveys,incometaxreturns— andtheamountofnationalincomerecordedinthenationalaccounts.Forinstance,thefraction ofnationalincomethatisreportedinindividualincometaxdatahasdeclinedfrom70%inthe late1970stoabout60%in2018.ThegapisevenlargerinsurveydatasuchastheCurrent PopulationSurveywhichcannotcapturetopincomeswell.Thisgapmakesithardtoaddress questionssuchas:whatfractionofnationalincomeisearnedbythebottom50%,themiddle 40%,andthetop10%ofthedistribution?Whohasbenefitedfromeconomicgrowthsincethe 1980s?HowdoesthegrowthexperienceofthedifferentgroupsofthepopulationintheUnited Statescomparetothatseeninothercountries?

Distributingthetotalityofincomeandwealthallowstocomputeincomegrowthratesforthe differentsocialgroupsconsistentwiththeofficialmacroeconomicgrowthrates,thusbridging thegapbetweenmacroeconomicanalysisandthestudyofinequality.Itreducesarbitrarinesscomparedtoapproachesthatfocusonnarrowernotionsofeconomicresources.Inaddition,becausethemacroeconomicaggregatesaredefinedandestimatedfollowingharmonized, internationally-agreedconceptsandmethods,distributionalnationalaccountsshouldmaximize thecomparabilityofinequalityovertimeandacrosscountries.

Piketty,Saez,andZucman(2018)presentaprototypeofdistributionalnationalaccounts forUnitedStates.Theseseriesaresupplementedbyasetofpubliclyavailablemicro-filesrepresentativeoftheUSpopulation,inwhicheachvariablecorrespondsandaggregatestoanational accountconcept(suchascompensationofemployees,corporateprofits,orincometaxespaid) andeachobservationisasyntheticindividualcreatedbycombiningtax,survey,andotherpubliclyavailabledatasources.ThesemicrofilesallowanyonetoreproduceallourfindingsonUS inequality—includingthosedescribedinthisarticle—andtocomputeotherstatisticsofinterest.Inthesamewayasthenationalaccountsareconstantlyupdated,revised,andrefined,we

regularlyupdateourseriesandmicro-fileswhenevernewdatabecomeavailableandimproved estimationtechniquesaredesigned.Theserevisionsaredocumentedinmethodologicalnotes thatexplainthechangesmadeandtheireffectonpreviouslyreportedstatistics.Followingharmonizedandregularlyupdatedguidelines(Alvaredoetal.,2017),similarmethodsareapplied toconstructprototypedistributionalnationalaccountsinagrowingnumberofcountries,includingFrance,India,China,andBrazil.TheseriesaremadeavailableontheWorldInequality Database http://WID.world,alongwithallcomputercodeandtechnicalappendices.Because thecodeandrawdataaregenerallypubliclyavailable,alternativemethodologiescanbetested.

Intime,wehopethatourprototypedistributionalnationalaccountswilleventuallybe takenoverbygovernmentandpublishedaspartoftheofficialtoolkitofgovernmentstatistics. Inequalitystatisticsaretooimportanttobelefttoacademics,andproducingtheminatimely fashionrequiresresourcesthatonlygovernmentandinternationalagenciespossess.Asimilar evolutionhappenedforthenationalaccountsthemselves,whichweredevelopedinthefirsthalf ofthetwentiethcenturybyscholarsintheUnitedStates(suchasSimonKuznets),theUnited Kingdom(suchasJamesMeadeandRichardStone),France(suchasDug´edeBernonville),and othercountries,beforebeingtakenoverbygovernmentagencies.

Itmaytakedecadesbeforewegetthere.Economicstatisticslikeaggregateoutputorconcentrationofincomearenotphysicalfactslikemassortemperature.Instead,theyarecreations thatreflectsocial,historical,andpoliticalcontext.Howthedatasourcesareassembled,what conceptualframeworkisusedtocombinethem,whatindicatorsaregivenprominence:allof thesechoicesreflectobjectivesthatmustbemadeexplicitandbroadlydiscussed.Beforerobust distributionalnationalaccountsarepublishedbygovernmentagencies,therearestillmanydecisionstobetakenandagreedonbytheacademicandstatisticalcommunity.Aspartofthat process,ourprototypecanbeusedtocharacterizetheriseofinequalityintheUnitedStates, toconfrontourmethodsandfindingswiththoseofotherstudies,andpinpointtheareaswhere moreresearchisneeded.

2TheRiseofWealthInequality

AfirststeptowardsthecreationofdistributionalnationalaccountswastakeninSaezand Zucman(2016),whoproducedestimatesofUSwealthinequalityallocating100percentofthe householdwealthrecordedintheFinancialAccounts,theofficialUSmacroeconomicbalance sheet.Householdwealthincludesallthenon-financialassets(suchasrealestate)andfinancial assets(suchasequities,bonds,andpensionwealth,whetherheldinindividualretirement

accountsorthroughpensionfunds)ofUShouseholds,netofdebts.

In2019,theFederalReservereleaseditsownofficialDistributionalFinancialAccounts, paintingasimilarpictureofalargeriseinwealthconcentration.

2.1MeasuringWealthWhenThereisNoAdministrativeDataon Wealth

BecausethereisnoadministrativedataonwealthintheUnitedStates,SaezandZucman (2016)useanindirectmethod,knownastheincomecapitalizationtechnique,toestimatewealth inequality.TheideaistoallocatetheFinancialAccountsaggregatestotheincomeflowsthat theseassetsgenerate:thus,interest-bearingassetsarelinkedtointerestpayments,corporate equitiesarelinkedtodividendsandcapitalgains,businessassetsarelinkedtobusinessprofits, andsoon.Concretely,iftheratiobetweenthestockofinterest-bearingassetsintheFinancial Accountsandthetotalflowofinterestincomereportedintaxreturnsis50,thensomeonewith $1,000ininterestisassigned$50,000inbonds,savingaccounts,andotherinterest-generating assets.Wealth,inotherwords,isestimatedbycapitalizingincome;intheprecedingexample, interestiscapitalizedusingacapitalizationfactorof50,orequivalentlyaninterestrateof2%.

Becausenotallassetsgeneratetaxableincome(pensions,mostimportantly,donot),taxdata needtobesupplementedwithotherdatatocaptureallformsofwealth.

Thebasiccapitalizationmethodissimpleandtransparent,anditdeliversresultsconsistent withotherevidenceaboutUSwealth.In2016,accordingtothebasiccapitalizationmethod, billionairesowned$3.1trillioninwealth,anumberclosetotheoneimpliedbythe Forbes annual listofthe400wealthiestAmericans,$3.0trillion.Taxunitswithlessthan$1billionandmore than$50millioninnetwealthowned$9.2trillion,anumbernotdissimilartotheonefoundin theSurveyofConsumerFinances,$10.2trillion.

Initssimplestform,thecapitalizationmethodreliesontheassumptionthatwithinanasset class,thelinkbetweenincomereportedintaxreturnsandwealthisthesameacrossindividuals; inotherwords,thatpeoplehavethesamerealizedrateofreturntowealth.Butofcourse,not everybodyactuallyhasthesamerealizedrateofreturnandtherateofreturnsmayevenbe positivelycorrelatedwithwealth.InSaezandZucman(2016)wearguedthattheassumption ofconstantrealizedreturnswithinassetclassappearedreasonable,basedondatafromestate taxreturnsmatchedtotheincometaxreturnofthedecedenttheyearbeforedeath;theSurvey ofConsumerFinances;andtaxreturnsfromfoundations.Inparticular,andsomewhatcounterintuitively,weshowedthatthecapitalizationtechniqueworkswellforUSfoundationsdespite thefactthatthewealthiestfoundations—withsophisticatedinvestmentsinprivateequityand

hedgefunds—havehighertotalratesofreturnsthanlesswealthyfoundations.Thereasonfor thisapparentparadoxisthatthehightotalreturnsoftopfoundationsstemfromhighunrealized capitalgains,nothighrealizedincome(interest,dividends,realizedcapitalgains)relativeto wealth.Whatmattersforthecapitalizationmethodisthat,withinanassetclass,theflowof realized incomebeproportionaltowealth,whichseemstogenerallybethecase.

However,wealsofound,inmatchedestate-incometaxdata,aninterestratepremiumseemed toappearamongtherichstartingaroundthetimeoftheGreatRecessionof2008–2009,and noted(p.550)thatthispatternshouldbewatched.Subsequentresearchsuggeststhatthe interestratepremiumoftherichhasbecomeafixtureofthepost-GreatRecessionera.Inthe SurveyofConsumerFinances,thetop1percentrichesthouseholdshaveahigherinterestrate thanaverageinthe2010,2013and2016wavesofthesurvey,byafactorof1.3(Brickeretal., 2018;SaezandZucman,2019).Inmatchedestates-incometaxdata,estatesabove$10million havecontinuedtoexhibitahigherinterestratesthanaverageAmericansafter2012,thelast yearinSaezandZucman,2016(Smith,Zidar,andZwick,2020,AppendixFigureA.15PanelA).

Thus,assumingthatallAmericanshavethesameinterestrateexaggeratestheinterest-bearing assetsofthewealthyinthepost-GreatRecessionperiod.

Forequitywealth,thecapitalizationmethodinfersassetsbasedondividendsandrealized capitalgains,andthusitcannotcapturethewealthofsomeonewhoreceivesnodividendand barelyrealizesanycapitalgains.AstrikingexampleisgivenbyWarrenBuffett,themain shareholderofBerkshireHathaway,acompanythatdoesnotpaydividends.In2016,Buffett disclosedhehadadjustedgrossincomeof$11.5millionin2015,anegligiblerealizedreturn relativetothevalueofhisstakeinBerkshireHathaway,whichamountedtoabout$60billion. Sixofthetop10wealthiestAmericans—JeffBezos(Amazon),MarkZuckerberg(Facebook), WarrenBuffett(BerkshireHathaway),SergeyBrin(Alphabet),LarryPage(Alphabet),and ElonMusk(Tesla),collectivelyworthmorethan$500billioninAugust2020—arethemain shareholdersofcorporationsthatdonotpaydividends.Indeed,bytriangulatingtheavailable sourcesonthereportedincomesoftheultra-wealthy,SaezandZucman(2019)estimatethatthe top400wealthiestAmericansasawholeearnlesstaxableequityincome(dividendsandrealized capitalgains)relativetotheirequitywealththantherestofthepopulation,byafactorofabout 2.AssumingthatallAmericanshavethesamerealizedreturnonequitiesunderestimatesthe equityassetsofbillionaires—aproblemthatisbecomingmoreacutewiththegrowthofgiant techcompaniesthatdonotdistributedividends.

Capturingthesetrendscallsforimplementingamoresophisticatedversionofthecapitalizationmethod.TheAugust2020updateoftheSaezandZucman(2016)estimatesofwealth

inequality,publishedontheWorldInequalityDatabaseandalsopresentedinthispaper,incorporatetheinterestratepremiumoftherichseeninmatchedestates-incometaxdata.They alsoupgradetheequitywealthofbillionairessothatthetotalnetworthofbillionairekeeps matching Forbes. Thesechangesdonotsignificantlyaffecttheleveloftopwealthsharesnor theirtrend,butbringassetcompositioninlinewiththeexistingevidence.Intheupdatedcapitalizedincomeseries,interest-bearingassetsaccountfor25percentofthewealthofthetop0.1 forexample,in2018,consistentwiththeassetcompositionseenintheofficialFederalReserve dataonwealthinequality.1

2.2DistributionalFinancialAccounts:ALandmark

In2019,theFederalReservereleaseditsownDistributionalFinancialAccounts.Itwasthefirst timethattheFederalReservepublishedstatisticsonwealthcoveringtheentirepopulation— fromthebottom50percentuptothetop1percent—consistentwithitsownofficialmacroeconomicbalancesheets.2

LikeinSaezandZucman(2016),theDistributionalFinancialAccountsstartfromtheFinancialAccountsaggregateandallocatethesetotalsacrossthepopulation.Methodologically, thetwoapproacheshavesomedifferences.TheFederalReservereliesontheSurveyofConsumerFinancessupplementedwiththe Forbes 400toallocatetheFinancialAccountaggregates; itdoesnotuseincometaxdata.TheSurveyofConsumerFinancesisahigh-qualitywealth surveythatover-samplestherich.However,thesurveyisonlyconductedtriennially,starting in1989.Thus,theDistributionalFinancialAccountsstartin1989andthedataisinterpolated betweeneachwaveoftheSCF.Likeallhouseholdsurveys,theSCFreliesonself-reportedinformationandsuffersfromsmallsamplesizesatthetop.Inthelatestwaveofthesurvey,about 6,200familiesweresampled.

TheFederalReserveincludesconsumerdurablesandunfundedpensionsinitsdefinitionof wealth,incontrasttoSaezandZucman(2016).Althoughincludingdurablesandunfunded

1 Smith,ZidarandZwick(2020)alsomodifythebenchmarkSaezandZucman(2016)capitalizationmethod. HowevertheyassumethewealthiestAmericansearnaninterestratehigherthanwhatisseeninthedatasets wherebothincomeandwealthcanbeobserved.Thisleadsthemtounder-estimatetheinterest-bearingassetsof thewealthy.Smith,ZidarandZwick(2020)alsoinferequitywealthbasedondividendincome,despitethefact thatthewealthiestAmericansoftenownequitiesthatdonotpaydividends.Asaresulttheycaptureonly57 percentofthebillionairewealthestimatedby Forbes.Oncethecorrectinterestrateisusedtocapitalizeinterest andequitywealthisfixedtomatchtheestimatesofbillionairewealthfrom Forbes,theSmith,ZidarandZwick (2020)estimatesareveryclosetotheSaezandZucman(2016)updatedseries;seeSaezandZucman(2020).

2 Whenweproducedourwealthinequalityestimates,wehadafruitfulexchangewiththeresearchersat theFederalReservewhoproduceandanalyzetheSurveyofConsumerFinances.Theseexchanges,sometimes vigorous(Brickeretal.2016,Bricker,Henriques,andHansen2018),helpednurturethecreationoftheFederal ReserveDistributionalFinancialAccounts,akeyandwidelyaccessibletool.

pensionscanbeappropriateforsomepurposes,itraisessomeissues.Durablesarenotassets intheUNSystemofNationalAccounts(UnitedNations,2009);othercountriesdonotinclude theseitemsintheirestimatesofaggregatehouseholdwealth(PikettyandZucman2014).Unfundedpensions—99percentofwhichinvolvepromisestogovernmentemployees(in2018)—are notbackedbyactualwealth.Includingunfundedpensionsinwealthwouldcalllogicallyforalso includingpromisesoffutureSocialSecuritybenefitsandpromisesofotherfuturegovernment benefits(suchasMedicare,futurespendingoneducation,etc.,netoffuturetaxes),whichneithertheFederalReserve,norSaezandZucman(2016),norothercountriesdo.Forinternational comparabilityandconceptualconsistency,durablesandunfundedpensionsarebestleftoutof wealth.

Oneimportantbutsubtleissueinthinkingaboutinequalityiswhethertomeasurethe distributionofeconomicresourcesacrosshouseholds,astheFederalReservedoes,oracross adultindividualsortaxunits,asinSaezandZucman(2016).Therearemoretaxunits(180 millionin2016)thanhouseholds(126million),becauseroommatesformseparatetaxunitsbut onehousehold,asdoparentslivingwithanadultchild,andunmarriedpartners.Webelieve thatdatausersshouldchoosetheunitofobservationthatfitsthequestiontheyareasking.

Forinstance,ifoneisinterestedintaxreforms,liketheintroductionofawealthtax,thenthe taxunitistheproperunitofobservation.Inthemicro-filesofPiketty,SaezandZucman(2018), onecanlookatthedistributionofwealthacrosstaxunitsoracrossadultindividualswiththe assumptionthatwealthisequallysplitamongmarriedspouses.This“equal-splitadult”approachisinpracticeclosetotheapproachbasedontaxunits:itassumesthatthereisafull sharingofresourcesbetweenmarriedspouses—albeitnotbetweenunmarriedpartners,incontrasttothehousehold-basedapproach.Onemeritofusingequal-splitadultsisthatitimproves thecomparabilityofinequalitystatisticsovertimeandacrosscountries,sincethedefinition ofadult(inourcaseanindividualaged20ormore)isfixed,whiledefinitionsofhouseholds andtaxunitscanvary.However,equal-splitadultstatisticswillunderstateinequality,because notallhouseholdwealthisequallysharedamongmarriedspouses.InFrance,Fremeauxand Leturcq(2020)findthatagrowingfractionofwealthisindividualized,asopposedtojointly ownedbetweenspouses.Animportantareaforfutureresearchinvolvescollectingmoredataon thedivisionofwealthbetweenmarriedspouses.ItwouldalsobehelpfuliftheFederalReserve alloweduserstolookatthedistributionofwealthacrossindividualsandtaxunits.

Inthemeantime,wecanconverttheFederalReserveDistributionalFinancialAccountsfrom householdstotaxunitsourselves,andcomparetheresultingdistributionstoSaezandZucman

(2016).3 Oncethesameunitofobservationandthesamedefinitionofwealthareused,the FederalReserveDistributionalNationalAccountsareveryclosetotheSaezandZucman(2016) estimates.AsshowninFigure1,inbothcases,thetop10percentwealthiesttaxunitsowned about78percentofwealthin2018,anincreaseof10pointssince1989.Inbothcases,thetop 1percentwealthiesttaxunitsowned38percentofwealthin2018,alsoanincreaseof10points since1989.

Overall,whetheronelooksattheabsolutelevelofwealthatthetop,thesharesofwealth ownedbythetopgroups,theportfolioallocationofthewealthy—andhowallofthishasevolved since1989—theDistributionalFinancialAccountsandtheSaezandZucman(2016)estimates paintthesamepicture.Byconstruction,totalwealthisthesameinthesetwodatasets(equal totheFinancialAccountsaggregate).In2018,the1percentrichesttaxhadabout38timesthe averagewealthpertaxunitof$482,000thatyear—thatis,about$18milliononaverage.In termsofportfoliocomposition,interest-bearingassetsaccountforaquarterofthenetwealth ofthetop1percentinbothdatasets,andpensionassetsfor10percent.

TheDistributionalFinancialAccountschoosesnottoreportwealthstatisticsforthetop0.1 percentorsmallergroups.ButwecanapplytheFederalReservemethodologyandcomputethe top0.1percentwealthshareinthatway.AsshowninPanelCofFigure1,theFederalReserve dataagainappearconsistentwiththeSaezandZucman(2016)estimates,althoughtheincrease inthetop0.1percentwealthshareisslightlymorepronouncedincapitalizedincomestatistics. Giventhelimitationsofthecapitalizationmethod,theSaezandZucman(2016)seriesmight overestimatetheriseofthetop0.1percentwealthshare.Butitisatleastequallylikelythat theSurveyofConsumerFinancesunder-estimatestheriseofthistopshare,becausetheSCF doesnotcapturethefullextentoftheriseofincomeinequalityatthetopendofthescale.4

The Forbes 400ranking,whichroughlycorrespondstothewealthiest0.00025percenthouse-

3 Forallintentsandpurposeshigh-endfamiliesarethesameashigh-endtaxunits.InSaezandZucman (2016),“thetop1percent”includes1.8milliontaxunitsin2016,whileintheFederalReservedata,“thetop1 percent”includes1.26millionhouseholdsandaround1.26milliontaxunits,i.e.,capturesonlythetop0.7percent wealthiesttaxunits.StandardPareto-interpolationtechniquesimplythattheshareofthetop0.7percentwithin thetop1percentis( 0 7 1 ) a 1 a where a istheParetocoefficient,equaltoabout1.3intheDistributionalFinancial Accounts.Thereforeoneneedstomultiplytheshareofwealthownedbytherichest1percenthouseholdsby 1.08tocapturetheshareofwealthownedbytherichest1percenttaxunits.Excludingconsumerdurablesand unfundedpensions,thetop1percentwealthiesthouseholdshave35.4percentoftotalwealthintheFederal ReserveDistributionalFinancialAccountsin2018,hencethetop1percentwealthiesttaxunitshave38percent oftotalwealth,anumberidenticaltotheSaezandZucman(2016)estimate.

4 RespondentstotheSurveyofConsumerFinancesareaskedabouttheirincomeasreportedontheirtax return.ButaspointedinSaezandZucman(2016),thetop0.1percentcapitalincomeshareroselessintheSCF thanintherealworldtaxdatafrom1989to2016.Brickeretal.(2016,p.290)arguethatthisgapmayowe toincomemisclassification:SCFrespondentsmay,forexample,callwageswhatinfactisbusinessincome.But theshareof totalincome earnedbythetop0.1percentalsoroselessintheSCFthanintheentirepopulation, suggestingthattheSCFdoesnotcapturethefullextentoftheriseinthetop0.1percentwealthshare.

holds,canbeusedtofocusonmuchnarrowerslicesofthewealthdistribution.Thesedataconfirmthesurgeofwealthconcentrationseenintaxdata:thetop0.00025percentwealthshare accordingto Forbes hasincreasedevenfasterthanthetop0.1percentwealthshareaccording tothetaxdata.Tobesure, Forbes isfarfromanidealdatasource.Itmaymisspeoplewho ownwealthindiversifiedportfoliosofstocksandbonds(forwhichnopublicinformationexists) andoverstatethevalueofprivatebusinesses.Toalleviatesomeoftheseconcerns,wecanfocus ontheverytopoftheranking,thetop0.00001percentwealthiestAmericans,agroupthat includes17taxunitstodayand10in1982,thefirstyearthatthe Forbes 400waspublished.It isnotunreasonabletoassumethatinagivenyearthe10or20wealthiestpeopleinthecountry arecorrectlyidentifiedby Forbes andtheirholdingsbroadlyaccuratelyestimated.Thisgroup isamixofmajorshareholdersofbig,publiclylistedcompanies(in2020,Amazon,Facebook, Google,Walmart,Microsoft,BerkshireHathaway;in1982,GettyOil,StandardOil,Hewlett Packard,etc.,in1982),forwhichvaluationsareobservable;andgiantprivatebusinesses(Koch IndustriesandBloombergLPtoday;MarsandHuntOilCompanyin1982)thatattractpublic scrutiny.AsshownbythePanelDofFigure4,theshareofwealthownedbythiselitegroup hasrisenfrom0.13percentoftotalUSwealthin1982to1.3percentin2020,atenfoldincrease.

AlthoughitisnotablethatthetwokeysourcesusedtoestimateUSwealthinequality— incometaxreturnsandtheSurveyofConsumerFinance—deliverconsistentresults,itwould beamistaketoexaggerateourabilitytomeasuretop-endwealth.Changesintaxavoidance, thegrowthofwealthheldinfoundations,andtheglobalizationofwealthmanagementpose formidablechallenges(fordiscussion,seeZucman,2015).Itisafailureofpublicstatisticsthat theonlyinformationonbillionairewealthcomesfrommagazines.Wecouldandshoulddo bettertomeasurewealthinequalitythanrelyonasurveyof6,200familiesorindirectlyinfer assetownershipbasedonincomeflows.

Onemeritofawell-administeredwealthtaxisthatitwouldprovidebetterinformationon thedistributionofwealth,oneofthemosthotlydebatedissuesindemocraticsocieties.Even withoutawealthtax,governmentscouldcollectinformationonassetsanddebtsfromthird parties(banks,pensionfunds,brokers,andothers),astheyalreadydoforincome.Thesedata couldbeusedtoimprovetaxenforcement—ascurrentlydoneinDenmark—andalsomeasure individualwealthbetterandallowfortheconstructionofmoreaccurateDistributionalFinancial Accounts.

Likeallimportanteconomicstatistics,eventheFinancialAccountsthemselveshavelimitationsandremain,decadesaftertheircreation,aworkinprogress.Onechallengeinvolvesthe valuationofprivatebusinessassets,whichtendstobeconservativeintheseaccounts.Another

relatestooffshorewealth:foreignbankaccounts,portfoliosofequitiesandbondsheldthrough foreignfinancialinstitutions,andholdingsofforeignmutualfunds(includinghedgefunds)that arenotintermediatedthroughaUSbrokerarenotcapturedintheFinancialAccounts(Zucman 2013).Theformsofwealththatarebroadlysharedtendtobeaccuratelymeasured,whilethe morecomplexinvestments,involvinglegalandfinancialintermediariesinforeigncountrieswith agreatdealofsecrecy,arelesswellcaptured.Theestimatesofwealthconcentrationwehave today,whichbyconstructionareanchoredtotheFinancialAccountstotals,shouldbeseenas lowerbounds.

3.1BridgingtheGapBetweentheStudyofInequalityandMacroeconomics

Therehasbeenasurgeofresearchinrecentyears,towhichwehavecontributedourshare, measuringincomeequalityusingadministrativetaxtaxdataratherthanself-reportedhousehold surveydata.However,thisworkhasmadeusawareofthelargeandgrowinggapbetween nationalincomeandtaxableincome.Onthelaborside,untaxedlaborincomeincludestaxexemptemploymentbenefits(contributionsmadebyemployerstopensionplansandtoprivate healthinsurance),employerpayrolltaxes,thelaborincomeofnon-filers,andunreportedlabor incomeduetotaxevasion.Thefractionoflaborincomewhichistaxablehasdeclinedfrom80 percent-85percentinthepost-WorldWarIIdecadestojustunder70percentin2018,dueto theriseofemploymentfringebenefitsandinparticulartheriseofemployercontributionsfor healthinsurance,particularlyexpensiveintheUnitedStates.Moststudiesonwageinequality ignorefringebenefitseventhoughtheyarealargeandgrowingfractionoflaborcosts.Asfor capital,onlyone-thirdoftotalcapitalincomeisreportedontaxreturns.Untaxedcapitalincome includesundistributedcorporateprofits,theimputedrentsofhomeowners,capitalincomepaid topensionaccounts,anddividendsandinterestretainedintrusts,estates,andfiduciaries. Whentakingintoaccountbothlaborandcapitalincome,thefractionofnationalincomethatis reportedinindividualincometaxdatahasdeclinedfrom70percentinthelate1970stoabout 60percenttoday.

Piketty,SaezandZucman(2018)estimatethedistributionof100percentofnationalincome bycombiningnationalaccounts,tax,andsurveydata.AsFigure2shows,inbothfiscalincome andnationalincomestatistics,theshareofincomeearnedbythetop1percentwashighbefore the1930s,fellfromthe1930stothe1970s,beforerisingagainfromthelate1970son.This

U-shapedevolutionofincomeconcentrationisabitlessspectacularwhenonelooksatnational incomeratherthanfiscalincome,mainlybecauseonlythefractionofcorporateprofitspaidout asdividendsareincludedinfiscalincomestatistics,whileallcorporateprofitsareincludedin nationalincome.Accountingforthetotalityofcorporateprofitsgenerallyincreasesthetop1 percentincomeshare,buttheeffectisstrongerinthepost-WorldWarIIyears,atimebefore theriseofpensionplanssomewhatbroadenedequityownership.

Onevirtueofdistributionalnationalaccountsisthattheyarenotaffectedbylegalchanges inbusinessorganization.IntheUnitedStates,agrowingnumberofbusinesseshavebeen organizedas“pass-through”entitiessincethelate1980s.Theincomeofpass-throughentities— partnerships,S-corporations,soleproprietorships—isnotsubjecttothecorporateincometax; instead,alltheincomeofthesebusinessesispassedtotheirindividualownersandsubjectto theindividualincometaxonly.Whenmorebusinessesoperateaspass-throughs,moreincome mechanicallyshowsuponindividualincometaxreturns,especiallyatthetop-endoftheincome distribution.Inourdistributionalnationalaccounts,allcorporationsaredefactotreatedas pass-throughentities,nomattertheirlegalstatus.Inthesamewayaspartnershipincome isallocatedtopartners,corporateincomeisallocatedtoshareholders.Inthesamewayas partnerspaytheindividualincometaxontheirshareofpartnerships’income,shareholderspay thecorporatetaxontheirshareofcorporations’income.Thisseemsalogicalwaytoallocate thecorporatetax.

BecausethereisnoadministrativedataintheUnitedStatesontheownershipofnon-passthroughcorporations,wemustmakeassumptionstoallocatetheportionofcorporateprofit thatisnotpaidoutasdividendsacrosstheincomedistribution.Inourdistributionalnational accounts,weallocate50percentofundistributedprofitsproportionallytodividendsand50 percentproportionallytorealizedcapitalgains.Thismethodisfarfromideal.Inthereal world,somepeoplewithlittledividendsandrealizedcapitalgainsaremajorshareholdersof corporationswithlargeundistributedprofits.Foradeeperunderstandingofincomeinequality, thegovernmentshouldcollectinformationabouttheownershipofcorporations.Thisinformationexistsinprivatefinancialinstitutions,suchastheDepositoryTrustCompany,thecentral securitiesdepositoryoftheUnitedStates,which defacto actsastheultimatebookkeeperfor theownershipofsecurities.

5 Inapapertitled“Capitalistsinthe21stcentury”Smithetal.(2019)findthat“thetypicaltopearner derivesmostofherincomefromhumancapital.”Theyobtainthisfindingbynotingthatpass-throughbusiness incomeisakeysourceofincomeatthetopofthefiscalincomedistributionandestimatingthat75%ofthis

3.2FromMacroeconomicGrowthtoPeople’sGrowth

Anadvantageofdistributingthetotalityofnationalincomeisthatitallowsforapples-to-apples comparisonsofinequalityacrosscountries,becausenationalincomeisdefinedandcomputed inthesamewayinternationally.(NationalincomeisequaltoGDPminuscapitaldepreciation plusnetincomereceivedfromabroad.)Ourbenchmarkstatisticsusetheequal-splitadultas theunitofobservation.Ourbenchmarkdefinitionofincome,pre-taxnationalincome,includes allpensionincome(fromSocialSecurityandprivatepensions)andsubtractallcorresponding pensioncontributionstocontrolforagingtrends.

Forexample,compareFranceandtheUnitedStates.IntheUnitedStates,nationalincome reached$17.5trillionin2018,closeto$72,500onaverageamongthe242millionadultswholived intheUS.Thebottom50percentearned12.5percentofnationalincome,whichmeansthat membersofthebottom50percentearnedaquarteroftheaverageincomeintheeconomy,or about$18,500onaverage.InFrance,usingpurchasingpowerparityexchangeratestoconvert eurosintoUSdollars,nationalincomeperadultwas$53,000—substantiallylessthaninthe US.Butthebottom50percentearned22.5percentofnationalincome,orabout$23,400on average.Eventhoughaverageincomeis37%higherintheUSthaninFrance,themarket delivershigherincomestothebottom50percentinFrancethanitdoesintheUnitedStates. TheFrenchwelfarestateisnotresponsibleforthisfeat,aswearetalkinghereaboutpre-tax nationalincome(beforegovernmenttaxesandtransfersotherthanSocialSecurity).Moreover, theincomecomparisondoesnotincludethebetterhealthoutcomesandmoreextensiveleisure timeinFrance.

Distributingthetotalityofnationalincomealsoallowsforrigorouscomparisonsofincome overtime.Figure3showsthegrowthrateofincomeforeachpercentileoftheincomedistribution from1946to1980and1980to2018.From1946to1980,averageperadultnationalincome rose2.0percentayear,oneofthehighestgrowthratesrecordedoveragenerationinacountry attheworld’stechnologicalfrontier.Moreover,thisgrowthwaswidelyshared,withonlythe incomeofthetop1percentgrowingabitlessthanaverage.Oneeasilyunderstandswhymany economistschose,duringthisperiod,totreattheUSdistributionofincomeasaconstantover time.

From1980to2018,averagegrowthinperadultnationalincomefallsto1.4percentayear. incomeislaborincomeratherthancapitalincome.However,fiscalincomemissestwothirdsoftotalcapital incomeandinparticularprofitsfromcorporationsthatdonotpaydividends(suchasAmazon,Facebook,or Google).SaezandZucman(2020)discussthisindetailandestimatethatcapitalincomeisslightlymorethan halfofincomeforthetop1%andabouttwothirdsforthetop.1%oftheUSnationalincomedistribution.

Foralmost90percentofthepopulation,growthhasbeenbelow—oftenmuchbelow—thatfigure. Forthebottom50percentasawhole,growthinpre-taxincomehasbeenonly0.2percenton averageperyear.Thisquasi-stagnationisnotduetopopulationaging,sincepre-taxincome includesSocialSecurityandotherretirementbenefits.Excludingtheelderly(aged65ormore), averagebottom50percentpre-taxincomehasslightlydeclinedsince1980.Duringthelastfour decades,macroeconomicgrowthhasnotbeenrepresentativeofthegrowthexperienceofthe vastmajorityofthepopulation.

Weneedadifferentmeasureofeconomicgrowthtocapturethelivedrealityofgrowthin aneraofrisinginequality.SaezandZucman(2019b)propose“people’sgrowth,”whichisthe arithmeticaverageofthegrowthrateofeachpercentileoftheincomedistribution.Itcaptures howincomegrowsonaverageacrosspeople,asopposedtohowtheaverageincomegrows.From 1946to1980,people’sgrowthandnationalincomegrowthcoincidedintheUnitedStates(2.0 percentayear).From1980to2018,people’sgrowthhasbeenonly0.65percentayear,less thanhalfofthe1.4percentmacroaverage.

Withafullpictureofthedistributionofnationalincomeovertime,wecanaskhowincome wouldhavegrownacrosstheincomedistributionifgrowthhadbeenequitablydistributed.If macrogrowthhadbeenequitablysharedfrom1980to2018,theaveragepre-taxincomeofthe working-classinthebottomhalfoftheincomedistributionwouldhavebeen57percenthigher in2018. 6 Forthemiddleclassfromthe50thtothe90thpercentileofincome,iftheseincomes hadgrownatthesamepaceasaverageincomeperadult(amodestgrowthrateof1.4percent), averageincomewouldhaverisenby16percent.However,fortheuppermiddleclassfromthe 90thto99thincomepercentile,averageincomeswouldhavefallenby8percentfrom1980to 2018,andfortherichinthe99thpercentile,averageincomeswouldhavedeclinedby36percent.

Tobesure,thiscounterfactualanalysishaslimitations.Perhapswithlessinequality,average growthmighthavebeenlower(therewouldperhapshavebeenlessinnovationifmillion-dollar earnershadnotbeenabletoearnthesumstheydid)orhigher(theremighthavebeenmore innovationifcredit-constrainedhouseholdshadbeenabletoearnmorethantheydid).Butthe counterfactualdoesillustratevividlytheshiftinincomedistribution.

3.3PitfallsofPersonalIncomeDistributions

InMarch2020,theUSBureauofEconomicAnalysisreleasedaprototypedistributionofpersonalincome—oneoftheaggregatemeasuresofincomeusedintheUSnationalaccounts.These

6 Thiscomputationusesthenationalincomedeflatortoconvertpastdollarsintocurrentdollars,thesame priceindexastheoneusedtocomputerealnationalincomeandallourseries.UsingtheCPI-U-RS,non-

dataprovideanimportantsteptowardthecreationofofficialdistributionalnationalaccounts. Buttherearestrongreasonstoprefernationalincomeoverpersonalincome.

First,personalincomeisspecifictotheUSnationalincomeandproductaccounts.Itisnot computedinothercountriesandinfactdoesnotexistintheUNSystemofNationalAccounts. Thismakesitimpossibletocompareinequalityinternationally.

Second,personalincomeisamixtureofpre-taxandpost-tax,andforthatreasonnot asatisfactorydefinitionofincomeconceptually.Personalincomesdoesnotsubtractpayroll taxesorindividualincometaxes,butitincludesallindividualizedgovernmenttransfers,such asSocialSecuritybenefits,welfareassistance,Medicare,andMedicaid.Therefore,personal incomedoublecountssomeformsofincome.

Third,personalincomedoesnotincludecorporateprofits;itonlyincludestheportionof corporateprofitsdistributedasdividends.Asaresult,personalincomeisaffectedbybusinesses’ choicesoforganizationalform.Ifapersonoperatesasapass-throughentity,allherincomegets countedinpersonalincome.Ifthesamepersonoperatesasacorporation,herincomecanbe zero.WarrenBuffethasbillionsinpre-taxnationalincome;hispersonalincomeissmallerbya factorof1,000.Unsurprisingly,theinequalityofpersonalincomeislowerthantheinequalityof nationalincome.Thedistributionofpersonalincomeislikelytobecomeapoorerandpoorer indicatorofincomeinequalityifmorebusinessesincorporatetotakeadvantageoftheverylow federalcorporatetaxratesince2018.Whenevertherichincorporate,personalincomeandits concentrationfall,eventhoughnothingelsechangesthanthetaxformusedbythebusiness owner.

TheBureauofEconomicAnalysisjustifiesthechoiceofpersonalincomebystatingthat thisaggregate“isclosesttothemeasureofeconomicresourcesavailabletohouseholdstopurchasegoods”andthat“[s]tartingwithpersonalincomewillallowfurtheranalysisofdisposable personalincome(aftertaxes)andabettercomparisontoconsumption(Fixleretal.2020,p. 3).Consistentwiththisview,BEAusesthehouseholdastheunitofobservation,notthe adultindividualasinPiketty,SaezandZucman(2018).7 Ourownviewisthatincomeand consumptionbothdeservetobestudied,butseparatelysincetheyaredistinctconcepts.

Nationalincomeisamoremeaningfulconcepttostudyincomeinequalitythanpersonal income,becauseitincludesalltheformsofincomethataccruetoindividuals,nomatterthe supervisorywageshavegrown0.33percentayearbetweenDecember1979andDecember2018.UsingtheCPIU,non-supervisorywageshavegrown0.2percent.

7 TheBEAalsouseshouseholdequivalencescales,inwhichhouseholdincomeisdividedbythesquareroot ofhouseholdsize,asanadjustmentfordifferencesinhouseholdsize,whichmakesitimpossibletodrawdirect connectionsfromdistributiontomacroeconomicgrowth.

specificwaysinwhichthisincomeisearned,consumed,orsaved.Thenotionofpersonalincome waspopularamongBEAstatisticiansinthe1950s;thefirstdistributionsofpersonalincome werecomputedatthattime.Inthe1950s,whenlargecorporationswerecontrolledbymultiple stakeholders,whathappenedintherealmofcorporationscouldfeeldisconnectedfromwhat happenedintherealmofhouseholds.Today,shareholdersexercisefullcontroloftheirfirms; thefrontierbetweencorporationsandhouseholdsisfuzzy;thefictionthatwhathappensinthe corporateworldhasnothingtodowithincomeinequalityisnolongertenable.Lookingforward, itisessentialforBEAtodistributenationalincome.

4HowGovernmentTaxesandSpendingAffectInequality

IntheUnitedStates,federal,state,andlocalgovernmentscollectabout28percentofnational incomeintaxesandspendmorethan28percentofnationalincome.InDenmarkandFrance, taxesandgovernmentspendingreach50percentofnationalincome.Ourdistributionalnational accountscanbeusedtoaskquestionslike:Dohigh-incomepeoplepaymoreorlessthantheaveragemacroeconomictaxrate?Howdocashtransferscomparetotaxesforlow-incomegroups? Aremiddle-classincomeshigheraftertaxesandtransfersthanbeforetaxesandtransfers?

Accordingtoawidespreadview,agovernmenttransferissimplyataxwithaminussignand allthatmattersispeople’sbudgetsets,netofalltaxesandtransfers.Weemphasize,however, thattaxesandtransfersaredistinctobjectsthatmustbestudiedassuch.Forexample,taxes reducecashincomebutmostgovernmenttransfersdonotincreasecashincome.Thebulkof governmenttransfersarein-kind—suchashealthinsuranceforthepoorandtheelderly,MedicaidandMedicare—ortaketheformofcollectiveconsumption,suchasspendingoneducation, police,anddefense.Evenwhentransfersaremonetary,it’simportanttorecognizethattaxes areoftenpaidcashonthenail,whiletransfersaregenerallyreceivedwithalag.Forapoor, credit-constrainedworker,paying$100inpayrolltaxeseachandeverymonthandreceivinga one-timecheckof$1,200thefollowingyearisnotequivalenttozerotaxandnotransfer.With anextra$100amonth,peoplearelesslikelytodefaultonamonthlyrentorinterestpayment; theyaremorelikelytobeabletoaffordanemergencyexpense,suchasavisittothedoctor, andtoaffordbasicdailyconsumptionneedssuchasfoodfortheirfamily.

4.1TheCollapseofTaxProgressivity

ThereisalongtraditionofresearchonthedistributionofUStaxpayments,pioneeredby ColmandTarasov(1940),Musgraveetal.(1951),andPechmanandOkner(1974).This traditionhasbeenrefinedbygovernmentagencies.Ourdistributionalnationalaccountsmake fourmaindeparturesrelativetotheanalysiscarriedoutbyUSagencies—mostprominentlythe CongressionalBudgetOffice—andthink-tanks.

First,weincludetaxesatalllevelsofgovernment,insteadoffederaltaxesonly.Stateand localtaxesaresizable,about10percentofnationalincome(athirdoftotaltaxrevenue). Inaddition,stateandlocalgovernmentsoftenmakesubstantialuseofsalesandexcisetaxes thataregenerallyregressive,andsoignoringthesetaxesgivesamisleadingpictureofthereal progressivityofthetaxsystem. 8

Second,weconsidertaxesasashareofpre-taxnationalincome,thebroadestandmost consistentdefinitionofincome.Thisisparticularlyimportantbecauseasizablefractionof thetruepre-taxincomeofthewealthy—theirshareofcorporateprofitsthatisnotpaidout asdividendsorrealizedascapitalgainsinagivenyear—isnotsubjecttoindividualincome taxation.Becauseweincludealltaxesandallnationalincomeinouranalysis,theaveragetax rateinourstatisticsisequaltotheofficiallyrecordedmacroeconomictaxrate,28percentof USnationalincomein2018.

Third,wedonotshifttaxesfromonefactorofproductiontoanother.Inourstatistics, consumptiontaxesareassignedtoconsumers,labortaxesareassignedtothecorresponding workers(evenwhenemployersnominallypaythem),andcapitaltaxesareassignedtothe correspondingownersofcapital.Inparticular,thecorporatetaxisassignedtoshareholders, justliketheincometaxpaidontheprofitofpass-throughbusinessesisassignedtotheowners ofpass-throughbusinesses.Thisframeworkallowstoallocatealltaxeswhilekeepingnational incomeconstantinaconceptuallyconsistentmanner(asdiscussedinSaezandZucman,2019c). Italsomakesitpossibletomeasuretheeconomicallyrelevanttaxwedgeoneachfactorof production,suchasthegapbetweenwhatitcoststoemployaworkerandwhattheworker receives.

Forthemostpart,themethodologycurrentlyfollowedbygovernmentagenciestostudythe distributionoftaxesissimilartothemethodologyweuse.Itallocatesalllabortaxestoworkers, allconsumptiontaxestoconsumers,mostcapitaltaxestocapitalowners,andkeepsnational incomeconstant.However,itshiftspartofthecorporatetaxtopeopleotherthanshareholders.

8 TheInstituteonTaxationandEconomicPolicy(2018)istheonlyinstitutionthatprovidescomprehensive distributionalstateandlocaltaxanalysisstatebystate.

Forexample,theCongressionalBudgetOffice(2012)allocates25percentofthecorporatetaxto wageearners,and75percenttocapitalowners,includingownersofinterest-bearingassets.The corporatetaxisallocatedpartlytoworkersbecauseitisassumedtodepressdomesticcapital andreducewages.Thisprocedureisinconsistentwithmaintainingaconstantlevelofnational income.9

Finally,inouranalysisrefundabletaxcreditsaretreatedasgovernmenttransfers—notas negativetaxes.Inthenationalaccountsconcepts,paymentsmadebythegovernmenttopeople aretransfers,nomatterwhichadministrationisinchargeofsendingthesetransfers.Thatthe EarnedIncomeTaxCreditpaymentsareadministeredbytheInternalRevenueService(rather than,say,theSocialSecurityadministration)hasnoeconomicimplication.Inthemacroeconomicstatisticsoftaxrevenues(forinstance,thegovernmentrevenuestatisticspublishedby theOECD),therefundableportionoftheEITCisneversubtractedfromtaxes.Thesameis truefortherefundableportionoftheChildTaxCredit.TheCongressionalBudgetOfficeand somethink-tanksthatproducedistributionaltaxstatisticschoosetosubtractrefundabletax creditsfromtaxespaid.ProponentsoftheEITCfeltthattheprogramwouldbemoreacceptablepoliticallyifpresentedasataxreductionratherthanatransfer,andalargeportionofthe USpublichasbeenusedtothinkingabouttheEITCasanegativetax.Butwhatmaybeperceivedasgoodpoliticsdoesnotnecessarilycorrespondtowhatismostconceptuallyconsistent. Economically,theEITCisnodifferentfromothercashtransferstolow-incomefamilies.

Thechoiceswemakeinourdistributionalnationalaccountsareofcoursenottheonlypossibleones,butwestressthattheyaretheoppositeofarbitrary.Instead,theyfollowconsistent economicconceptsasdefinedandmeasuredbyinternationalguidelines.Inourview,thisapproachprovidesapromisingwaytotranscendsomeofthepoliticalandideologicalidiosyncrasies thatcanaffecttheanalysisofinequalityinapurelynationalcontext.

Whentakingacomprehensiveperspectiveontaxation,adramaticdeclineintheprogressivity oftheUStaxsystemappears.Figure4depictstheUSaveragetaxratebyincomegroupsfor variousyearsfrom1950to2018.Allfederal,state,andlocaltaxesareincludedandtaxesare expressedasafractionofpre-taxincome.P0-10denotesthebottom10percentofadultsinthe incomedistribution,P10-20thenext10percent,etc.Wesplitthetop10percentintosmaller groupsallthewaythetop400wealthiestAmericanspopularizedby Forbes.Takingalltaxes

9 Allocatingpartofthecorporatetaxtoindividualsotherthanshareholdersalsocreatesinconsistencies.For example,ifaC-corporationelectstobetreatedasanS-corporation(apass-throughbusiness),thenintheCBO treatmentpre-taxincomeinequalityincreases(incomethatwaspreviouslyassignedtoworkersisnowallocated toshareholders,whoarehigherupintheincomedistribution),thelaborshareofnationalincomefalls,andthe taxsystembecomesmoreprogressive(taxesthatusedtobepaidbyworkersarenowpaidbyshareholders), despitethefactthatnothingrealhaschangedintheeconomyorinthetaxsystem.

together,theUStaxsystemusedtobeslightlyprogressiveorroughlyproportionalforthe bottom99percentoftheincomedistribution,buthighlyprogressivewithinthetop1percent. In1950,forexample,theuppermiddleclass(thetop10percentexcludingthetop1percent) paidaveragetaxratesofaround25percent,whilethetop0.01percentpaidalmost70percent ofitsincomeintaxes.

Thetaxsystemwashighlyprogressiveinthe1950sbecausecorporateprofits,themain sourceofincomefortherich,wassubjecttoahigheffectivecorporatetaxrateof50percent. Veryhightopmarginalindividualincometaxrates(91percentuntil1963atthefederallevel) madeitimpossibleforbusinessownerstobypassthecorporatetaxbyusingpass-throughs,such aspartnerships.Moreover,highincomeswerehitbothbytheprogressiveindividualincome taxontheirdistributedcapitalincomeandbyaprogressiveestatetaxatthetimeofdeath. Thecombinationoftheincometax,thecorporatetax,andtheestatetaxmadethetaxsystem extremelyprogressiveandhardtoavoid(SaezandZucman2019).Low-incomehouseholdspaid lowertaxesthantodaybecausethepayrolltaxwaslowerinthepast.

In2018,thetaxsystemlookslikeagiantflattaxthatbecomesregressiveattheverytopend. Theworkingclassandthemiddleclasspaysubstantialtaxesbecausepayrolltaxesarelarge andstateandlocalsalesandexcisetaxesareregressive.Theverytoppaysloweffectivetax ratesbecauseofthedemiseofthefederalcorporatetax,whichin2018collectedonly1percent ofnationalincome,downfrom5-7percentinthe1950s.Theeffectiveindividualincometax ratefallsatthetop-endbecausetheveryrichearnincomethroughcorporationsandcanavoid reportingindividualincome.Theregressivityofthetaxsystemattheextremetopendin2018 isstriking—adirectconsequenceofthemassive2018cutinthecorporatetax.Butthefigure showsadecades-longshift,withaslowerosionofthecorporatetax,theestatetax,andlower progressivityoftheindividualincometaxatthetop.

Ifthelowcorporatetaxof21percentsetin2018continues,thereisarealriskthatthe wealthywillincorporate,earnincomethroughtheircorporations,andbypasstheprogressive individualincometaxbyretainingearningswithintheircorporations.Ifhelduntildeath,the capitalgainsgeneratedbysuchretainedearningswillneverbetaxed.

4.2HaveGovernmentTransfersOffsettheRiseofInequality?

Taxesareonlyhalfofthegovernmentequation.Onthespendingside,SocialSecurity(retirementanddisability)andunemploymentinsurancereplacelostlaborearningsduetoretirement, disability,orunemployment.TheseprogramsgrewfastafterWorldWarIItoabout6percent ofnationalincomeinthelate1970sandhavebeenstableafterwards.Weincludethesetrans-

fersinourmeasureofpre-taxincome.Theremainingformsofgovernmentspendingarepart ofpost-taxincome(butnotpre-tax)andcanbeclassifiedinthreecategories,fromeasiestto allocatetoindividualstohardest:cashtransfers,in-kindtransfers,andcollectiveconsumption.

Cash(orquasi-cash)individualizedtransfersincludewelfareassistanceandrefundabletax creditsforlow-incomefamilieswithchildren,foodstampsforthepoor,andsupplementalsecurityincomeforthelow-incomeelderlyandthedisabled.Thesetransferscomeclosesttopure redistributionasindividualscanfreely(oralmostfreelyinthecaseoffoodstamps)choosehow tospendthem,justlikeearnedincome.Cashtransfersaresmall,2-3percentofnationalincome withnocleartrendafterthemid-1970s.Morespecifically,refundabletaxcreditshavegrown butwelfareassistancehasfalleninthesameproportion.Cashtransfersgooverwhelminglyto thebottom50percent.

Next,in-kindindividualizedtransfers,suchaspublichealthinsurance(MedicaidandMedicare),housingassistance,andhighereducationtuitionsubsidies,havegrownfromalmostzero in1960toabout8percentofnationalincomein2018.Thisgrowthisoverwhelminglydrivenby MedicareandMedicaid,whichaccountforover90percentofallin-kindtransfersin2018.In ourdistributionalnationalaccounts,weallocatethesetransfersasalumpsumperbeneficiary; thatis,wedividethesumspaidonMedicarebythenumberofMedicarebeneficiaryandassign eachbeneficiarytheaverageMedicaretransfer.Alargefractionofin-kindtransfersgotothe bottom50percent.

Thelastcategoryofgovernmentspendingiscollectiveconsumptionexpenditure.Thiscategoryincludesgovernmentspendingoneducation,defense,publicorder(police,prisons,courts), andotherpublicgoods.Collectiveconsumptionislarge(about18percentofnationalincome) andhasbeenfairlystablesince1960.Spendingondefensehasshrunkwhileotherformsof collectiveconsumptionhaveincreased;spendingonprisonshasincreasedparticularlyfast,due toamassiveincreaseinincarcerationratesinthe1980sand1990s.Governmentspendingon educationhasbeenstableat5percentofnationalincomesince1970.Inourdistributionalnationalaccounts,weallocateallcollectiveconsumptionneutrally,sothatcollectiveconsumption doesnotaffectincomeinequality.Obviously,ifweweretoallocatecollectiveconsumptionona percapitabasis,thatwouldmakeinequalitylooklower.10

Howdoesincorporatingtaxesandtransfersaffectthedistributionofincome?Inthebig picture,thetaxsystemisapproximatelyaflattax—taxesareproportionaltoincome—while 10 ThemainreasonwhyAutenandSplinter(2019)findlowtopincomesharesonapost-taxbasisisbecause theyallocatehalfofgovernmentconsumptionpercapita.SeeSaezandZucman(2010)foracompletediscussion ofAutenandSplinter(2019).

thetransfersystemisclosertoaflatamountperperson.Thiscombinationreducesinequality: post-taxinequalityislessthanpre-taxinequality.

Withourdistributionalnationalaccounts,wecanalsoexaminewhetherchangesingovernmentinterventionintheeconomyhaveliftedincomesatthebottom.Theshortansweris“yes, butnotalot.”Theaveragepre-taxincomeforthebottom50percent,aswehaveseen,has almoststagnatedsince1980inrealterms:Itwas$17,500peradultin1980andis$18,500in 2018.Afterdeductingtaxesandaddingallformsofgovernmentspending,averagepost-tax incomehasincreasedby25percentsince1980.Thisisbetterthanquasi-stagnation,butstill lessthanthe70percentincreaseinaverageincomeperadultfrom1980to2018.Theriseof governmenttransferstothebottomhasoffsetroughlyone-thirdofthegrowthgapbetweenthe bottomhalfandtheaverageAmerican.

Mostofthegrowthinbottom50percentpost-taxincomesisdrivenbythesurgeinMedicaid andMedicare.Toseethis,itisusefultoconsideranarrowerdefinitionofpost-taxincome, disposablecashincome.Disposablecashincomeispre-taxincomeminusalltaxesplusallcash orquasi-cashtransfers;itexcludesin-kindtransfersandcollectiveconsumptionexpenditures. Thisnotionofdisposableincomeisclosetotheoneusedtomeasurethepovertyrate(USCensus Bureau,2019),withtheimportantdifferencethatwedeductalltaxesandaddrefundabletax creditsandfoodstamps.Fortheaverageadult,cashdisposableincomeperadulthasincreased aboutasmuchasnationalincomefrom1980to2018,bycloseto70percent(thanksinpart tobeingbolsteredbygrowingfederaldeficits).Figure5showsthatforthebottom50percent, disposablecashincomehasgrownverymodestlyoverthelastfourdecades:itwas$16,000in 1980and$18,600in2016,a16percentincreaseover36years.

Until2008,thebottom50percentpaidmoreintaxesthanitreceivedincashtransfers: pre-taxincomewashigherthancashdisposableincome.Thecashdisposableincomeofbottom 50percentadultswasliftedupbythelargegovernmentdeficitsrunduringtheGreatRecession. Since2012,cashdisposableincomeisalmostidenticaltopre-taxincome.Thus,thegainsin post-taxincomeforthebottom50percentoverthistimetaketheformofin-kindtransfers (primarilyMedicaid),andcollectivepublicexpenditures(education,defense,policeandprisons beingthemainitems).

4.3TheLimitsofPost-TaxIncome

Themodestgainsinpost-taxincomefortheworkingclassmustbeanalyzedwithcare,because allocatingin-kindtransfersandcollectiveconsumptiontoindividualsbasedontheircostfor thegovernmentishighlyproblematic.AllOECDcountrieshavedecidedthateverybodyshould

haveaccesstoqualityeducation.AllOECDcountriesexcepttheUnitedStateshaveanational programforfinancinghealthcare.Thecostofuniversalprovisionofeducationandhealthlooks likealargetransferrelativetoincomeforlow-incomefamilies.Butitisconceptuallyincorrect totreatthisfulltransferasincomeforitsrecipients.Afterall,iflow-incomefamiliesreceivedan equivalentamountincash,mostofthemwouldnotspenditallonhealthoreducation.Perhaps thebestconceptualalternativewouldbetoassigntheperceivedcashvalueofindividualized in-kindtransferstorecipients,whiletreatingtherestasacollectivepublicgood.

TheseconceptualproblemsareparticularlythornyforhealthtransfersintheUnitedStates. MedicaidtransfersarelargebecausethecostofhealthcareisextraordinarilyhighintheUnited States.Themoneyisnotflowingintothebankaccountsofbeneficiaries;insteadit’sflowing tothebankaccountsofhealthcareproviders,manyofwhicharenearthetopoftheincome distribution.Whatsensedoesitmaketorejoiceintheriseofworking-classpost-taxincomes, ifthisrisereflectstheriseintherentsearnedbythemedicalandpharmaceuticalsectors?

Asimilarissueariseswithgovernmentmandates,liketheruleintheAffordableCareActin 2010thatemployerswith50ormoreemployeesarelegallyrequiredtoprovidehealthinsurance totheirfull-timeworkers(orpayapenaltyof$3,000peremployee).Thecostofthismandatory privatehealthinsuranceislargeandgrowing;itisaheavyburdenonlow-paidworkers.In conceptualterms,partofthiscostshouldbeconsideredasataxonworkersthatthegovernment imposestoachievewiderhealthinsurancecoverage(SaezandZucman,2019b).Likeothertaxes, thiscostshouldbesubtractedfromincomeforthecomputationofpost-taxincome.

Inshort,thereisnoperfectmeasureofpost-taxincome.Tomeasuretheinequalityof incomeaftertaxesandtransfers,disposablecashincomeasdepictedisperhapsthemostmeaningfulstatistics.Disposablecashincomecapturesincomeavailableforsavingandconsumption, excludingthecollectiveconsumptionofserviceslikeeducationandhealthmandatedbythe government.Butdisposablecashincomedoesnotadduptonationalincome.Post-taxnationalincomecapturesallofnationalincomebydeductingalltaxesandaddingbackallforms ofgovernmentspendingandthegovernmentdeficit.Butcomputingpost-taxnationalincome requiresassigningcollectiveconsumptionexpendituresaswellasthecurrentgovernmentdeficit toindividuals.Thereisnoobvious,universally“correct”waytodosuchanimputation,and therewillneverbe.

Doesthismeanthatwecannotknowwhatishappeningtoinequality?Ofcoursenot.There arenorawfactsinthesocialsciences.Rather,thereareattemptsatdescribingrealitythrough moreorlesselaboratestatisticalframeworks.Theresultsoftheseattemptscanonlybeproperly understoodonceweknowhowthemeasurementtoolswork,whataspectsofrealitytheyaimto

capture,whatledtotheircreation,theobjectivesoftheircreators,theknowledgetheyembody, theaccountabilityoftheinstitutionsthatpublishthem,andthetheoriesthatunderpinthem.

Onceweunderstandhowdistributionalnationalaccountsareconstructed,areasonedused ofthesestatisticsbecomespossible—justlikeareasoneduseofGDPstatisticsbecomespossible onceweunderstandtheirstrengthsandlimitations.

Pre-taxnationalincome,whichcapturesincomeearnedfrommarketactivities,canbeused todecomposemacroeconomicgrowthandtocompareinequalityovertimeandacrosscountries.Cashdisposableincomecanbeusedtostudytheincomeavailableforsavingandprivate consumption;bysubtractingthesavingcomponent,itcanbeusedtostudyconsumptioninequality.Post-taxnationalincomecanbeusedtoestimatethetotaldirectdistributiveeffects ofgovernmentinterventionintheeconomy.Allofthesenotionshavemeritsanddemeritsand mustbestudiedjointly.Ultimately,thebestdatawouldbepublishedbygovernmentagencies, accountabletoelectedrepresentatives,discussedbythepressandpartieswithastakeintheir improvement,andbasedonaregularlyupdated,internationally-agreedconceptualframework. Thisistherecipethathasmadethenationalaccountssuccessful;thisisthewayforwardfor allthoseinterestedinimprovingthemeasurementofinequality.

Notes:Alltheseriesusethesamedefinitionofhouseholdwealth(themarketvalueofallnon-financialandfinancialassetsnetofalldebts,excludingconsumerdurablesandunfunded pensions),havethesametotalwealth(theofficialFinancialAccountstotal,e.g.,$77.5trillioninmid-2016),thesametotalsassetclassbyassetclass(theofficialFinancialAccounts total,e.g.,$26.2trillionforpensionwealthinmid-2016)andusethesameunitofobservation(taxunits).TomovefromhouseholdstotaxunitsintheSCFandtheDistributional FinancialAccounts,weassumethateachtaxunitwithinthetop1percentcorrespondstoonehousehold,andmakenocorrectionforthenext9percent.TomaketheSCFcomparable totheothertwosources,weaddtheForbes400tothepublic-useSCFfilesandadjustreportedwealthtomatchtheFinancialAccountstotalsassetclassbyassetclass.

Saez-Zucman (2016)
Saez-Zucman (2016)

Figure2:Shareofincomeearnedbythetop1percent

Pre-tax national income Fiscal income

Notes:Thisfigurecomparestheshareoffiscalincomeearnedbythetop1percenttaxunits(whiteseriesfromPikettyandSaez,2003)totheshareof pre-taxnationalincomeearnedbythetop1percentequal-splitadults(blackseriesfromPiketty,SaezandZucman,2018)from1913to2016.Allseriesare availableontheWorldInequalityDatabase, http://WID.world.

Figure3:AverageandPeople’sAnnualIncomeGrowth

Average growth 1980-2018: 1.4%

People's growth 1980-2018: 0.65%

Notes:Thisfiguredepictstheannualrealpre-taxincomegrowthperadultforeachpercentileinthe1946-1980period(inblue)and1980-2018period(in red).Italsodepictstheaveragegrowthperadultandpeople’sgrowth(averageacrosspeoplewithnoincomeweighting).From1946-1980,growthwas evenlydistributedwithallincomegroupsgrowingattheaverage2percentannualrate(exceptthetop1percentwhichgrewslower).From1980to2018, growthhasbeenunevenlydistributedwithlowgrowthforbottomincomegroups,mediocregrowthforthemiddleclass,andexplosivegrowthatthetop creatingabiggapbetweenaveragegrowthandpeople’sgrowth.Source:SaezandZucman(2019b).

Figure4:TheDemiseofUSTaxProgressivity

tax rates by income group (% of pre-tax income)

Working class

Middle-class Upper middleclass The rich

Notes:ThefiguredepictstheUSaveragetaxratebyincomegroupsfrom1950to2018.Allfederal,state,andlocaltaxesareincluded.Taxesareexpressed asafractionofpre-taxincome.P0-10denotesthebottom10percentoftheincomedistribution,P10-20thenext10percent,etc.Takingalltaxestogether, theUStaxsystemusedtobeprogressivewithmuchhighereffectivetaxratesatthetop.Ithasslowlymorphedintoanalmostflattaxwithsimilarrates acrossincomegroups.SourceisDistributionalNationalAccounts,updated2020.

Figure5:TheEvolutionBottom50percentIncomes

Disposable cash income

Notes:Thefiguredepictstheevolutionoftherealincomesperadult(in2016dollars)forthebottomhalfoftheincomedistributionforthreeincome concepts:(1)pre-taxincomebeforedeductingtaxesoraddinggovernmenttransfers(conceptsumsuptonationalincome),(2)post-taxincomethatdeducts alltaxesandaddsalltransfers(cashandin-kind)andcollectivepublicexpendituresminusthegovernmentdeficit(alsosumsuptonationalincome),(3) disposablecashincomewhichispre-taxincomeminusalltaxespluscash(orquasi-cash)transfers,i.e.(3)doesnotincludein-kindtransfers(primarily medicaidandmedicare)orcollectivepublicexpendituresthatareincludedin(2).SourceisDistributionalNationalAccounts,updated2020.

References

Alvaredo,Facundo,AnthonyB.Atkinson,LucasChancel,ThomasPiketty,EmmanuelSaez, andGabrielZucman.2017.“DistributionalNationalAccounts(DINA)Guidelines:Concepts andMethodsusedin WID.world”,WIDworkingpaper.

AutenGerald,andDavidSplinter.2019.“IncomeInequalityintheUnitedStates:UsingTax DatatoMeasureLong-termTrends,”workingpaper.

Bourguignon,Fran¸cois.2011.“Non-anonymousgrowthincidencecurves,incomemobilityand socialwelfaredominance.” JournalofEconomicInequality,605627.

Bricker,Jesse,AliceHenriques,JacobKrimmel,andJohnSabelhaus.2016.“Measuringincome andwealthatthetopusingadministrativeandsurveydata.” BrookingsPapersonEconomic Activity,261-331.

BrickerJesse,LisaJ.Dettling,AliceHenriques,JoanneW.Hsu,LindsayJacobs,KevinB. Moore,SarahPack,JohnSabelhaus,JeffreyThompson,andRichardA.Windle.2017.“Changes inU.S.FamilyFinancesfrom2013to2016:EvidencefromtheSurveyofConsumerFinances,” FederalReserveBulletin, 103(3),1–39.

Bricker,Jesse,AliceHenriques,andPeterHansen.2018.“HowMuchhasWealthConcentrationGrownintheUnitedStates?ARe-ExaminationofDatafrom2001-2013.”Financeand EconomicsDiscussionSeries2018-024.BoardofGovernorsoftheFederalReserveSystem. CensusBureau.2019. IncomeandPovertyintheUnitedStates:2018,availableonlineat https://www.census.gov/library/publications/2019/demo/p60-266.html Colm,Gerhard,andHelenTarasov.1940.“Whopaysthetaxes.” TemporaryNationalEconomicCommittee,Washington,DC. CongressionalBudgetOffice.2012.“TheDistributionofHouseholdIncomeandFederalTaxes, 2008and2009.”availableonlineat https://www.cbo.gov/publication/43373. EconomicPolicyInstitute.2019.StateofWorkingAmericaDataLibrary,“Annualwagesby wagegroup”availableonlineat https://www.epi.org/data/#?subject=wagegroup. Fixler,Dennis,MarinaGindelsky,andDavidJohnson.2020.“MeasuringInequalityinthe NationalAccounts,”BEAworkingpaperseries,WP2020-3. Fremeaux,NicolasandMarionLeturcq.2020.“Inequalitiesandtheindividualizationof wealth,” JournalofPublicEconomics, 184,p.104–145. InstituteonTaxationandEconomicPolicy.2018. WhoPays:ADistributionalAnalysisofthe TaxSystemsinAll50States. 6thed.,WashingtonDC:ITEP. Musgrave,RichardA.,JohnJ.Carroll,LorneD.Cook,andLenoreFrane.1951.“Distribution oftaxpaymentsbyincomegroups:acasestudyfor1948.”NationalTaxJournal4(1):1-53. Pechman,JosephA.,andBenjaminA.Okner.1974. WhoBearstheTaxBurden?,(Washington D.C:BrookingsInstitution).

Piketty,Thomas,andEmmanuelSaez.2003.“IncomeInequalityintheUnitedStates,19131998.” QuarterlyJournalofEconomics 118,no.1:1-39.

Piketty,Thomas,andGabrielZucman.2014.“CapitalisBack:Wealth-IncomeRatiosinRich Countries1700–2010.” QuarterlyJournalofEconomics 129(3):1255-1310.

Piketty,Thomas,EmmanuelSaez,andGabrielZucman.2018.“DistributionalNationalAccounts:MethodsandEstimatesfortheUnitedStates.” QuarterlyJournalofEconomics 133, no.1:553–609.

Saez,EmmanuelandGabrielZucman.2016.“WealthInequalityintheUnitedStatessince 1913:EvidencefromCapitalizedIncomeTaxData.” QuarterlyJournalofEconomics 131,no. 2:519-578.

Saez,EmmanuelandGabrielZucman.2019.“ProgressiveWealthTaxation.” BrookingsPapers onEconomicActivity

Saez,EmmanuelandGabrielZucman.2019b. TheTriumphofInjustice:HowtheRichDodge TaxesandHowtoMakethemPay. NewYork:W.W.Norton.

Saez,EmmanuelandGabrielZucman.2019c.“ClarifyingDistributionalTaxIncidence:Who PaysCurrentTaxesvs.TaxReformAnalysis.”UCBerkeleyWorkingPaper2019.

Saez,EmmanuelandGabrielZucman.2020.“TrendsinUSIncomeandWealthInequality: RevisingAftertheRevisionists.”UCBerkeleyWorkingPaper2019.

Smith,Matthew,OwenZidar,andEricZwick.2019.“Capitalistsinthe21stCentury,” QuarterlyJournalofEconomics, 134(4)p.1675–1745.

Smith,Matthew,OwenZidar,andEricZwick.2020.“TopWealthinAmerica:NewEstimates andImplicationsforTaxingtheRich.”WorkingPaper. USCensusBureau.2019.“IncomeandPovertyintheUnitedStates:2018CurrentPopulation Reports,”PublicationP60-252,CensusBureau.

Zucman,Gabriel.2013.“TheMissingWealthofNations:AreEuropeandtheU.S.netDebtors ornetCreditors?” QuarterlyJournalofEconomics 128,no.3:1321-1364..

Zucman,Gabriel.2015. TheHiddenWealthofNations:TheScourgeofTaxHavens. Chicago: UniversityofChicagoPress.

Zucman,Gabriel.2019.“GlobalWealthInequality.” AnnualReviewofEconomics 11,109-138.

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.