Global Extreme Poverty - Our World in Data

Global poverty is one of the very worst problems that the world faces today. The poorest in the world are often hungry, have much less access to education, ... InthenewsExploredataforcontextofthewarinUkraineGlobalExtremePovertyOverviewofthisentryIstheworldontracktoendextremepovertyby2030?Extremepovertyinthebroadercontextofwell-beingExtremepovertyaroundtheworldtodayTheevolutionofextremepoverty,countrybycountryExtremepovertyinahistoricalperspectiveHistoricalpovertyaroundtheworldHistoricalpovertyintoday’srichcountriesTheevolutionofpovertybyworldregionsGlobalpovertyrelativetohigherpovertylinesHowmuchdoesthereductionoffallingpovertyinChinamatterforthereductionofglobalpoverty?The(mis)perceptionsaboutpovertytrendsThedemographicsofextremepovertyHowmanypoorpeopleliveineachcountry?Whoarethepeoplelivinginextremepoverty?Howmanychildrenliveinextremepovertyaroundtheworld?Howdoespovertyamongchildrencomparetopovertyamongadults?PovertyacrossmultipledimensionsHowcanwemeasurepovertybeyondincomeandconsumption?DeterminantsandconsequencesThelinkbetweeneconomicgrowthandpovertyPovertytrapsEvidenceonspecificstrategiestoreducepovertyCross-countrycorrelatesMeasurementandDataQualityWhatarethemainindicatorsusedtomeasurepoverty?Thedifferencebetween‘absolute’and‘relative’povertyHowdoresearchersreconstructhistorical povertyestimates?HowdoestheWorldBankestimateextremepoverty?WhatarethemainlimitationsofWorldBankpovertyestimates?Howproblematicaredatalimitations?Whatalternativesaretheretoestimatemonetarypoverty?Whatisthecostofendingextremepoverty?DataSourcesEndnotesLicenceCitationContentsAllourchartsonGlobalExtremePoverty Comparisonofextremepovertyestimates,PovcalvsBradshaw&Mayhew(2011)Datadeprivation:numberofpovertysurveysperdecadeavailableviatheWorldBankDisposablehouseholdincome:IncomecutofftothepoorestdecilevsmeanincomeDistributionofpopulationbetweendifferentpovertythresholdsPovcaldataExtremepovertyratevs.povertygapindexGDPpercapitaWorldBankGDPpercapitaininternationalandmarketdollarsHowdoestheinternationalpovertylinecomparetomultidimensionalpoverty?MDG1.A:Shareofpopulationinextremepoverty(lessthan$1.25perday)Nationalpovertylinesvs.GDPpercapitaPopulationlivinginextremepovertyPovertygapindexat1.90int.-$perdayPovertyheadcountratio(Povcalvs.WDI)Povertyvs.LifeexpectancyPovertyvs.educationalattainmentProjectednumberofpeopleinextremepoverty,byregionCrespo-Cuaresmaetal.(withprojections)ProjectedshareofthepopulationinextremepovertyReconstructionofhistoricalglobalextremepovertyratesShareinpovertyrelativetodifferentpovertythresholdsSharelivingonlessthan3.20int.-$perdayShareofglobalpopulationlivinginextremepovertyincludingandexcludingChinaShareofpeoplelivinginextremepovertyEuropeancountries(BradshawandMayhew,2011)ShareofpopulationinextremepovertyShareofpopulationlivinginmultidimensionalpovertyShareofpopulationlivingwithlessthan1.90$and3.20$perdayTheglobalpovertygapThepovertygapTheshareofpeoplelivinginextremepovertyvsGDPpercapitaTotalnumberofpeopleinextremepovertyTotalpopulationlivinginextremepovertybyworldregionWorldpopulationlivinginextremepovertyWorldBank&BourguignonandMorrisson OverviewofthisentryMostpeopleintheworldliveinpoverty.85%oftheworldliveonlessthan$30perday,two-thirdsliveonlessthan$10perday,andeverytenthpersonlivesonlessthan$1.90perday.Ineachofthesestatisticspricedifferencesbetweencountriesaretakenintoaccounttoadjustforthepurchasingpowerineachcountry.Theresearchhereisconcernedwiththelivingconditionsoftheworstoff:thosewholivein‘extremepoverty’asdefinedbytheUnitedNations.TheWorldBank,whichispartoftheUN,isthemainsourceforglobalinformationonextremepovertytodayanditsetsthe‘InternationalPovertyLine’.Thepovertylinewasrevisedin2015—sincethen, apersonisconsideredtobeinextremepovertyiftheyliveonlessthan1.90internationaldollars(int.-$)perday.Thispovertymeasurementisbasedonthemonetaryvalueofaperson’sconsumption.Incomemeasures,ontheotherhand,areonlyusedforcountriesinwhichreliableconsumptionmeasuresarenotavailable.Akeydifficultyinmeasuringglobalpovertyisthatpricelevelsareverydifferentindifferentcountries.Forthisreason,itisnotsufficienttosimplyconvertthe consumption levelsofpeopleindifferentcountriesbythemarketexchangerate;itisadditionallynecessaryto adjustforcross-countrydifferencesinpurchasingpower.Thisisdonethrough PurchasingPowerParityadjustments(explainedbelow).Itisimportanttoemphasizethatthe InternationalPovertyLineisextremelylow. Indeed,‘extremepoverty’isanadequatetermforthoselivingunderthislowthreshold.Focusingonextremepovertyisimportantpreciselybecauseitcapturesthosemostinneed.However,itisalsoimportanttopointoutthatlivingconditionswellabovetheInternationalPovertyLine canstillbecharacterizedbypovertyandhardship.Accordingly,inthisentrywewillalsodiscusstheglobaldistributionofpeoplebelow povertylinesthatarehigherthantheInternationalPovertyLineof1.90int.-$.Butrelyingonlyonhigherpovertylineswouldmeanthatwearenotkeepingtrackoftheverypoorestpeopleintheworldandthisisthefocusofthisentry.Povertyisaconceptintrinsicallylinkedtowelfare–andtherearemanywaysinwhichonecantrytomeasurewelfare.Inthisentrywewillfocusmainly(thoughnotexclusively)onpovertyasmeasuredby‘monetized’consumptionandincome,followingtheapproachusedbytheWorldBank.Butbeforewepresenttheevidence,theintroductorysub-sectionhereprovidesabriefoverviewoftherelevanceofthisapproach.Globalpovertyisoneoftheveryworstproblemsthattheworldfacestoday.Thepoorestintheworldareoftenhungry,havemuchlessaccesstoeducation, regularlyhavenolightatnight, and sufferfrommuchpoorerhealth.Tomakeprogressagainstpovertyisthereforeoneofthemosturgentglobalgoals.Theavailablelong-runevidenceshowsthatinthepast,onlyasmalleliteenjoyedlivingconditionsthatwouldnotbedescribedas‘extremepoverty’today.Butwiththeonsetofindustrializationandrisingproductivity,theshareofpeoplelivinginextremepovertystartedtodecrease.Accordingly,theshareofpeopleinextremepovertyhasdecreasedcontinuouslyoverthecourseofthelasttwocenturies.Thisissurelyoneofthemostremarkableachievementsofhumankind.Closelylinkedtothisimprovementinmateriallivingconditionsistheimprovementofglobal healthandtheexpansionof globaleducation thatwehaveseenovertheselasttwocenturies.Wealsodiscussthelinkbetweeneducation,health,andpovertyinthisentry.Duringthefirsthalfofthelastcentury,thegrowthoftheworldpopulationcaused theabsolutenumberofextremelypoorpeopleintheworldtoincrease,eventhoughtheshareofpeopleinextremepovertywasgoingdown.Afteraround1970,thedecreaseinpovertyratesbecamesosteepthattheabsolutenumberofpeoplelivinginextremepoverty startedfallingaswell.Thistrendofdecreasingpoverty—bothinabsolutenumbersandasashareoftheworldpopulation—hasbeenaconstantduringthelastthreedecades.Butaswehighlightinthefirstsectionofthisentryitisunfortunatelynotwhatwecanexpectforthecomingdecade.Itisthefactthatstillalmosteverytenthpersonlivesinextremepovertyandtheslowingprogressagainstextremepovertythatmotivatethisentry.Relatedwriting:Howdoweknowthehistoryofextremepoverty?Economicgrowth–Howdoeconomiesbecomemoreproductive?Understandinghowandwhencountriesachievedeconomicgrowthiscrucialtounderstandhowsomecountrieslefttheworstpovertybehindandhowothercountriescanfollow.Incomeinequality–Itisnotjusttheaverageincomethatmattersforwhetherornotpeopleliveinpovertybuthowincomesaredistributed.Globaleconomicinequality–Ourentryontheglobaldistributionofincomes.Istheworldontracktoendextremepovertyby2030?GrowthoftheglobalmiddleclassandfallingextremepovertyOverthecourseofthelastgenerationmorethanabillionpeopleleftthemostdestitutelivingconditionsbehind.Canweexpectthisprogresstocontinueoverthecomingdecade?Theworldeconomyisgrowing.Inlessthanagenerationthevalueoftheyearlyglobaleconomicproductionhasdoubled.1Increasingproductivityaroundtheworldmeantthatmanylefttheworstpovertybehind.Morethanathirdof theworldpopulationnowliveonmorethan10dollarsperday.Justadecadedecadeagoitwasonlyaquarter.Inabsolutenumbersthismeantthenumberofpeoplewholiveonmorethan10dollarsperdayincreasedby900millioninjustthelast10years.2Thisexpansionoftheglobalmiddleclasswenttogetherwithprogressinreducingglobalpoverty–nomatterwhatpovertylineyouwanttocompareitwith,theshareoftheworldpopulationbelowthispovertylinedeclined.3In1990internationalorganizationsadoptedadefinitionofpovertyinlinewiththepovertylinesinlow-incomecountries.Inthelatestadjustmenttheinternationalpovertylineissettothethresholdoflivingonlessthan$1.90perday.Thatisaverylowpovertylineandfocussesonwhatishappeningtotheverypoorestpeopleontheplanet.4Thesameinternationalorganizationsthatsetthepovertylinemadeitaglobalgoaltoendextremepoverty.GoalnumberoneoftheSustainableDevelopmentGoals(SDGs),agreedonbyallnationsintheworld,isthe“eradicationofextremepovertyforallpeopleeverywhere”.Thedeadlineforachievingthisgoalis2030.Canweexpecttoachievethis?Halfabillionprojectedtoliveinextremepovertyin2030ResearchteamsfromtheWorldBank,ODI,theIHME,andBrookingsjointlywiththeWorldDataLabmadeindependentprojectionsforwhatwecanexpectforglobalpovertyduringtheSDGera.Whiletheprojectionsdifferinmethodologyandunderlyingassumptions,it’sstrikinghowmuchtheyalignintheirprojectionforwhattoexpectinthecomingdecadeiftheworldstaysoncurrenttrajectories.Allexpectsomepositivedevelopment–thenumberofpeopleinextremepovertyisexpectedtocontinuetodecline–butallalsoagreeonthebadheadline:theworldisnotontracktoendextremepovertyby2030.ThechartshowstheprojectionmadebythedevelopmentresearchteamattheWorldBank.Thisprojectionanswersthequestionofwhatwouldhappentoextremepovertytrendsiftheeconomicgrowthofthepastdecade(2005–15)continueduntil2030:5Thenumberofpeopleinextremepovertywillstagnateatalmost500million.Thisisnotbecauseitisnotpossibletoendextremepoverty.Inmorethanhalfofthecountriesoftheworldtheshareofthepopulationinextremepovertyisnowlessthan3percent.6Inthesamecountriesthehugemajority–evenintoday’srichestcountries–livedinextremepovertyjustafewgenerationsago.Infact,thebigsuccessoverthelastgenerationwasthattheworldmaderapidprogressagainsttheveryworstpoverty.Thenumberofpeopleinextremepovertyhasfallenfromnearly1.9billionin1990toabout650millionin2018.7Thiswaspossibleaseconomicgrowthreachedmoreandmorepartsoftheworld.8InEthiopia,India,Indonesia,Ghana,andChinamorethanhalfthepopulationlivedinextremepovertyagenerationago.Butaftertwodecadesofgrowththeshareinextremepovertymorethanhalvedinallthesecountries.PovertywasnotconcentratedinAfricauntilrecently.In1990morethanabillionoftheextremelypoorlivedinChinaandIndiaalone.Sincethenthoseeconomieshavegrownfasterthanmanyoftherichestcountriesintheworldanddidmuchtoareductionofglobalinequality.Theconcentrationoftheworld’spoorestshiftedfromEastAsiainthe1990stoSouthAsiainthefollowingdecade.NowithasshiftedtoSub-SaharanAfrica.Theprojectionssuggestthegeographicconcentrationofextremepovertyislikelytocontinue.AccordingtotheWorldBankforecasts87%oftheworld’spoorestareexpectedtoliveinSub-SaharanAfricain2030ifeconomicgrowthfollowsthetrajectoryovertherecentpast.Povertydeclinedduringthelastgenerationbecausethemajorityofthepoorestpeopleontheplanetlivedincountrieswithstrongeconomicgrowth.Thisisnowdifferent.StagnationforthepoorestManyoftheworld’spooresttodayliveincountriesthathadveryloweconomicgrowthinthepast.9ConsiderthecaseofMadagascar:Inthelast20yearsGDPpercapitahasnotgrown;andthenumberinextremepovertyincreasedalmostone-for-onewithtotalpopulation.Developmenteconomistshaveemphasizedthisforsometime:Theverypoorestpeopleintheworlddidnotseetheirmateriallivingconditionsimprove.10Thisfactissurelyoneofthebiggestdevelopmentfailuresofourtime.Yetthestagnationoftheworld’spoorestcountriesisnotaswidelyknownasitshouldbe–onereasonisthatwearenotpayingattentiontopovertylineslowenoughtofocusonwhathappenstotheverypoorest.Thisisanimportantreminderthatonepovertylineisnotenoughandweneedtorelyonseveralpovertylines–higherandlowerthantheinternationalpovertyline–tounderstandwhatishappening.Arisingglobalmiddleclassandstagnationoftheworld’spoorestwillalsomeanthatanewdivideatthelowestendoftheglobalincomedistributionisopeningup.Wemissthisifweonlyfollowwhatishappeningtotherapidlyemergingglobalmiddleclassorifwerelyonglobalpovertylinesthatarenotcapturingwhatishappeningtothepoorest.Theprojectionssuggestthatoverthecomingdecadethestagnationatthebottomwillbecomeveryclear.Themajorityoftheworld’spooresttodayliveineconomiesthatarenotgrowingandhalfabillionfacetheprospecttoremainstuckinextremepoverty.Thisisterriblenews.PolicyandgrowthTheseprojectionsdescribewhatwehavetoexpectoncurrenttrends.Butcurrenttrendsdon’thavetobecomefuturetrends:allcountriesthatleftextremepovertybehindhadamomentatwhichtheybrokeoutofstagnation.Thesecondbiglessonfromthehistoryofextremepovertyisthatitisthegrowthofanentireeconomythatliftsindividualsoutofpoverty.Keyforendingextremepovertygloballywillbethatthepoorestcountriesachievethedifficulttaskofeconomicgrowth.Butit’snotonlyaboutmacroeconomicperformance.Socialpolicyanddirecthousehold-levelsupport,too,makeanimportantdifference.Eveninverypooreconomiesthereisscopefortargetedpoliciestosupporttheverypoorest.Inananalysisofhowtoday’srichestcountriesleftextremepovertybehindMartinRavallionemphasizestheroletheexpandedsocialprotectionpoliciesplayedatthetime.11Themostimportanttaskinourtimeistoensurethatthelivingconditionsoftheworld’spoorestimproveandtoendextremepoverty.Weknowthatitispossible;wehavedoneitmanytimesinthepast.Thebigsuccessofthelastgenerationwasthatglobalextremepovertydeclinedrapidly.Butmanyarestillverypoorandprogressagainstextremepovertyisurgentlyneeded.However,wearecurrentlyfarofftracktoendingextremepoverty–weexpecthalfabillionpeopletostillliveonlessthan$1.90perdayby2030.Toensurethat‘nooneisleftbehind’astheSDGagendapromises,thisiswhereweneedtofocusourefforts.It’snotjustaconcernuntil2030:withoutrisingincomesintheworst-offplacesextremepovertywillremainarealityformillions.Extremepovertyinthebroadercontextofwell-beingIspovertyonlyaboutconsumption?Therearemanywaysinwhichresearchersandpolicymakerstrytomeasurewelfare.Inthisentrywefocusmainlyonwelfareasmeasuredby‘monetized’consumptionandincome,followingtheapproachusedbytheWorldBank.However,asweemphasizethroughout,thisisonlyoneofmanyaspectsthatweneedtoconsiderwhendiscussingpoverty.InotherentriesinOurWorldInDatawediscussevidencethatallowstrackingprogressinotheraspectsofwelfarethatarenotcapturedbystandardeconomicindicators.Thisbroadperspectiveonglobaldevelopmentisattheheartofourpublication.Thepracticeofmeasuringwelfareviaconsumptionandincomehasalongtraditionineconomics.Manyclassictextbooksandpapersprovidedetailsregardingtheconceptualframeworkbehindthis(forabasictechnicaloverviewseeDeatonandZaidi2002);12andbynowthereisalsoanextensiveliteraturediscussingvariousimportantpointsofcontention(seeCh2.inAtkinson2016forabriefrecentoverview).13Alternativestartingpointsformeasuringwelfareincludesubjectiveviews(e.g.self-reportedlifesatisfaction),basicneeds(e.g.caloricrequirements),capabilities(e.g.accesstoeducation),andminimumrights(e.g.humanrights).Thesealternativenotionsofwelfareplayanimportantroleinacademiaandpolicy,anditisnecessarytobearinmindthattheyareinterrelated.Indeed,asweexplainbelow,manyoftheseconceptsindirectlyenterthemethodologyusedbytheWorldBanktomeasurepoverty;forexample,byhelpingsetthepovertylinesagainstwhichmonetizedconsumptionisassessed.Thistable,fromAtkinson(2016)providesacomparisonofthe‘money-metricapproach’usedbytheWorldBankvis-à-visthemostcommonalternatives.Whatdopoorpeoplethinkaboutpoverty?Themostimportantconclusionfromtheevidencepresentedinthisentryisthatextremepoverty,asmeasuredbyconsumption,hasbeengoingdownaroundtheworldinthelasttwocenturies.Butwhyshouldwecare?Isitnotthecasethatpoorpeoplemighthavelessconsumptionbutenjoytheirlivesjustasmuch—orevenmore—thanpeoplewithmuchhigherconsumptionlevels?Onewaytofindoutistosimplyask.Subjectiveviewsareanimportantwayofmeasuringwelfare.ThisiswhattheGallupOrganizationdid.TheGallupWorldPollaskedpeoplearoundtheworldwhattheythoughtabouttheirstandardofliving—notonlyabouttheirincome.Thefollowingchartcomparestheanswersofpeopleindifferentcountrieswiththeaverageincomein thosecountries.Itshowsthat,broadlyspeaking,peoplelivinginpoorercountries tendtobelesssatisfiedwiththeirlivingstandards.Thissuggeststhateconomicprosperityisnotavain,unimportantgoalbutratherameansforabetterlife.Thecorrelationbetweenrisingincomesandhigherself-reportedlifesatisfaction isshowninourentryonhappiness.Thisismorethanatechnicalpointabouthowtomeasurewelfare.Itisanassertionthatmattersforhowweunderstandandinterpretdevelopment.First,thesmoothrelationshipbetweenincomeandsubjectivewell-beinghighlightsthedifficultiesthatarisefromusingafixedthresholdabovewhichpeopleareabruptlyconsideredtobenon-poor.Inreality,subjectivewell-beingdoesnotsuddenlyimproveaboveanygivenpovertyline.Thismakesusingafixedpovertylinetodefinedestitutionasabinary‘yes/no’problematic.Therefore,whiletheInternationalPovertyLineisusefulforunderstandingthechangesinlivingconditionsoftheverypoorestoftheworld,wemustalsotakeintoaccounthigherpovertylinesreflectingthefactthatlivingconditionsathigherthresholdscanstillbedestitute.Andsecond,thefactthatpeoplewithverylowincomestendtobedissatisfiedwiththeirlivingstandardsshowsthatitwouldbeincorrecttotakearomanticviewonwhat‘lifeinpoverty’islike.Asthedatashows,thereisjustnoempiricalevidencethatwouldsuggestthatlivingwithverylowconsumptionlevelsisromantic.Adisregardforordisinterestinpovertyestimatesthatarecalculatedonthebasisoflowconsumptionandincomelevelsispartlyexplainedbythefactthatitcanbeverydifficultforpeopletoimaginewhatitisliketo livewithverylittle.Eveneconomistswhothinkalotaboutincomeandpovertyfinditdifficulttounderstandwhatitmeanstoliveonagivenincomelevel. Itisjusthardtopicturewhatlifeislikewhenallyouknowisa “dollar-per-day”figure.Toaddressthis,AnnaRoslingRönnlundputtogether acaptivating,visualprojectatGapminder.orginwhichsheportraysthelivingconditionsofpeoplelivingatdifferentincomelevels.InDollarStreetyoucanfindportraitsoffamiliesandseehowtheycook,whattheyeat,howtheysleep,whattoiletstheyhaveavailable,whattheirchildren’stoyslooklike,andmuchmore.DissatisfactionwithstandardoflivingvsGDPpercapita14 ExtremepovertyaroundtheworldtodayInthissectionTheevolutionofextremepoverty,countrybycountryTheevolutionofextremepoverty,countrybycountryExtremepoverty:Theshareofpeoplelivingonlessthan1.90int.-$perdayThemoststraightforwardwaytomeasurepovertyistosetapovertylineandtocountthenumberofpeoplelivingwithincomesorconsumptionlevelsbelowthatpovertyline.Thisistheso-calledpovertyheadcountratio.Measuringpovertybytheheadcountratioprovidesinformationthatisstraightforwardtointerpret;bydefinition,ittellsustheshareofthepopulationlivingwithconsumption(orincomes)belowsomeminimumlevel.TheWorldBankdefinesextremepovertyaslivingonlessthan1.90int.-$.Inthemapweshowavailableestimatesoftheextremepovertyheadcountratio,countrybycountry.Themapshowsthe latestavailableestimatesbydefault,butwiththeslider(immediatelybelowthemap)youcanexplorechangesovertime.Youcanalsoswitchtothe‘chart’tabtoseethechangeovertimeforindividualcountriesorworldregions;oryousimplyclickonacountrytoseehowthepovertyheadcountratiohaschanged.Estimatesareagainexpressedininternationaldollars(int.-$)using2011PPPconversionrates.Thismeansthatfiguresaccountfordifferentpricelevelsindifferentcountries,aswellasforinflation. Clicktoopeninteractiveversion Theshareofpeoplelivingonlessthan3.20int.-$perdayExtremepoverty,asdefinedbytheWorldBank,isindeedextreme–livingon$1.90perdayisverydifficult.Hence,itisbothinterestingandimportanttomeasurepovertywithhigherpovertylines.TheWorldBankalsoreportspovertyheadcountratiosusingahigherlineat3.20int.-$,andthemapshowstheseestimates. Clicktoopeninteractiveversion ExtremepovertyinahistoricalperspectiveInthissectionHistoricalpovertyaroundtheworldHistoricalpovertyintoday’srichcountriesTheevolutionofpovertybyworldregionsGlobalpovertyrelativetohigherpovertylinesHowmuchdoesthereductionoffallingpovertyinChinamatterforthereductionofglobalpoverty?The(mis)perceptionsaboutpovertytrendsHistoricalpovertyaroundtheworldBelowwesummarizehowpovertyhaschangedoverthelasttwocenturies.Howhistoriansknowaboutthehistoryofpovertyisthefocusofalongertextthatyoufindhere:Howdoweknowthehistoryofextremepoverty?TheshareofpeopleinextremepovertyoverthepasttwocenturiesTheWorldBankonlypublishesdataonextremepovertyfrom1981onwards,butresearchershavereconstructedinformationaboutthelivingstandardsofthemoredistantpast.Theseminalpaperonthiswaswrittenby BourguignonandMorrisonin2002.15Inthispaper,thetwoauthorsreconstructmeasuresofpovertyasfarbackas1820.Thepovertylineof1.90int.-$perdaywasjustintroducedin2015,sothe2002paperusesthemeasureof‘onedollarperday’.Thisdifferenceinthedefinitionofpovertyshouldbekeptinmindwhencomparingthefollowinggraphtothosediscussedinothersectionsofthisentry.In1820,thevastmajorityofpeoplelivedinextremepovertyandonlyatinyeliteenjoyedhigherstandardsofliving.Economicgrowthoverthelast200yearscompletelytransformedourworld,withtheshareoftheworldpopulationlivinginextremepovertyfallingcontinuouslyoverthelasttwocenturies.Thisisevenmoreremarkablewhenweconsiderthatthepopulationincreased7-foldoverthesametime.Inaworldwithouteconomicgrowth,anincreaseinthepopulationwouldresultin lessandlessincomeforeveryone.A7-foldincreaseintheworldpopulationwouldbepotentiallyenoughtodriveeveryoneintoextremepoverty.Yet,theexactoppositehappened.Inatimeofunprecedented populationgrowth,wemanagedtoliftmoreandmorepeopleoutoftheextremepovertyofthepast.Itisverydifficulttocompareincomeorconsumptionlevelsoverlongperiodsoftimebecausetheavailablegoods andservicestendtochangesignificantly,totheextentwhereevencompletelynewgoodsandservicesemerge.Thispointissosignificantthatitwouldnotbeincorrecttoclaimthateverypersonintheworldwasextremelypoorinthe19thcentury.NathanRothschildwassurelytherichestmanintheworldwhenhediedin1836. Butthecauseofhisdeathwasaninfection—aconditionthatcannowbetreatedwithantibioticssoldforlessthanacoupleofcents.Today,onlytheverypoorestpeopleintheworldwoulddieinthewaythattherichestmanofthe19thcenturydied.Thisexampleisagoodindicatorofhowdifficultitistojudgeandcompare levelsofprosperityandpoverty,especiallyforthedistantpast.Thetrendovertimebecomesmoreclearifonecomparestheavailabilityofnecessitieslikefood,housing,clothing,andenergy.Asmoreandmorecountriesindustrializedandincreasedtheproductivityofwork,theireconomiesstartedtogrowandpovertybegantodecline.AccordingtotheestimatesbyBourguignonandMorrison—showninthevisualization—onlyalittlemorethanaquarteroftheworldpopulationwasnotlivinginpovertyby1950.From1981onwards,wehavebetterempiricaldataonglobalextremepoverty.TheBourguignonandMorrisonestimatesforthepastarebasedonnationalaccountsandadditionalinformationonthelevelofinequalitywithincountries.Thedatafrom1981onwardscomefromtheWorldBank,whichbasestheirestimatesonhouseholdsurveys.(Seebelowformoreonwherehistoricalpovertyestimatescomefrom).Accordingtothesehouseholdsurveys, 44%oftheworldpopulationlivedinextremepovertyin1981.Sincethen,theshareofextremelypoorpeopleintheworldhasdeclinedveryfast—infact,fasterthan everbeforeinworldhistory.In32years,the shareofpeoplelivinginextremepovertywasdividedby4,reachinglevelsbelow10%in2015.Thereisalsoaninteractiveversionofthisvisualizationhere.ShareoftheWorldPopulationlivinginAbsolutePoverty,1820-201516ThenumberofpeopleinextremepovertyoverthepasttwocenturiesWehaveseenthatthechanceofbeingbornintoextremepovertyhasdeclineddramaticallyoverthelast200years.Butwhatabouttheabsolutenumberofpeople livinginextremepoverty?Thevisualizationcombinestheinformationontheshareofextremepovertyshowninthelastchart,withthenumberofpeoplelivingintheworld. Fortheyearspriorto1980, weusethemid-pointoftheestimatesfromBourguignonandMorrison(2002)asshowninthepreviouschart;from1981, weusetheWorldBankestimates.Aswecansee,in1820therewerejustunder1.1billionpeopleintheworld,ofwhichthelargemajoritylivedinextremepoverty.Overthenext150 years,thedeclineofpovertywasnotfastenoughtooffsettheveryrapidriseoftheworldpopulation,sothenumberofnon-poorandpoorpeopleincreased.Sincearound1970,however,wearelivinginaworldinwhichthenumberofnon-poorpeopleisrising,whilethenumberofextremelypoorpeopleisfalling.Accordingtotheestimatesshownhere,therewerecloseto2billionpeoplelivinginextremepovertyintheearly1980s,andtherewere735millionpeoplelivinginextremepovertyin2015.In 1990,therewere1.9billionpeoplelivinginextremepoverty.Withareductionto735millionin2015,thismeansthatonaverage,everydayinthe25yearsbetween1990and2015,128,00fewerpeoplewerelivinginextremepoverty.17Oneverydayinthelast25yearstherecouldhavebeenanewspaperheadlinereading,“Thenumberofpeopleinextremepovertyfellby128,000sinceyesterday”.Unfortunately,theslowdevelopmentsthatentirelytransformourworldnevermakethenews,andthisistheveryreasonwhyweareworkingonthisonlinepublication.Recentlythisdeclinegotevenfasterandinthe7yearsfrom2008to2015theheadlinecouldhavebeen“Numberofpeopleinextremepovertyfellby192,000sinceyesterday”.Intherecentpastwesawthefastestreductionofthenumberofpeopleinextremepovertyever.Whilethereductionto10%isamajorachievementofhumanityitisstillabsolutelyunacceptablethateverytenthpersonintheworldlivesinextremepoverty.Whatourhistoryshowsusisthatitispossibletoreduceextremepovertyitisonustoendextremepovertyassoonaspossible. Clicktoopeninteractiveversion Historicalpovertyintoday’srichcountriesWehavealreadypointedoutthatinthethousandsofyearsbeforethe beginningoftheindustrialera,thevastmajorityoftheworldpopulationlivedinconditionsthatwewouldcallextremepovertytoday. Productivitylevelswerelowandfoodwasscarce—materiallivingstandardsweregenerallyverylow.Thefirstcountriesinwhichpeopleimprovedtheirlivingconditionswerethosethatindustrializedfirst.Thechartshowsthedeclineofextremepovertyinthesecountries.TheseestimatescomefromRavallion(2015).18Theyuseapovertylineof1.25 int.-$in2005prices,andtheyrelyonincomesmeasuredfromnationalaccounts.The‘nationalaccounts’methodtoestimatepovertyisbasedonacademicstudiesthatreconstructhistoricalincomelevelsfromcross-countrymacroestimatesoneconomicoutputandinequality.(Seebelowformoreonthe‘nationalaccounts’methodtoestimatepoverty).Twopointsareworthemphasizing.First,wecanseethatextremepovertywasverycommonintoday’srichcountriesuntilfairlyrecently;infact,inmostofthesecountriesthemajorityofthepopulationlivedinextremedeprivationonlyafewgenerationsago.Progresswasmadeatafastpace—insomecasesevenataconstantpace.Wecandefinitelyendextremepovertyinlowincomecountries,andwecandoitsoon.Othercountrieshavedoneitbefore.Second,wecanalsoseefromthischartthatdespiteremarkableprogress,insomerichcountries—notablytheUnitedStates—afractionofthepopulationstilllivesinextremepoverty.Thisistheresultofexceptionallyhighincomeinequality.(Seebelowformoreonextremepovertyinrichcountries).Thereductionofextremepovertyincountriesthatarerichtoday,1820-200019TheevolutionofpovertybyworldregionsTotalpopulationinextremepoverty,byworldregionsAbove,weprovidedanoverviewofrecentpovertytrendscountrybycountry.Herewefocusontrendsfromaregionalperspective.Thefirstchartprovidesregionalestimatesofpovertycounts–thetotalnumberofpeoplelivingbelowtheInternationalPovertyLineineachworldregion.Thesecondchartprovidesregionalestimatesofpovertyrates–theshareofpopulationineachregionlivingbelowtheInternationalPovertyLine.FigurescorrespondtotheInternationalPovertyLine,at1.90int.$in2011PPPprices.Aswecansee,globally,thenumberofpeoplelivinginextremepovertyfellbymorethan1billionduringtheperiod;from1.9billionin1990to0.74billionin2015.Onaverage,thenumberofpeoplelivinginextremepovertydeclinedby47millioneveryyearsince1990.Onanyaveragedaythenumberofpeopleinextremepovertydeclinedby130,000people.InSub-SaharanAfricahoweverthenumberofpeopleinextremepovertyhasincreasedandweexplainedatthebeginningofthisentryvariousprojectionsexpectthatextremepovertywillbeincreasinglyconcentratedinAfrica.Thefollowingchartshowsthattheshareofpeoplelivinginextremepovertyhasfallenevenfaster.ThisverypositivedevelopmenthasbeenpossibleinpartduetotheremarkableimprovementsinEastAsiaandthePacific,wherepovertyrateswentfrom81%in1981to2.3%in2015. Clicktoopeninteractiveversion Clicktoopeninteractiveversion GlobalpovertyrelativetohigherpovertylinesTheInternationalPovertyLinethatinternationalorganizationsliketheUNrelyoncorrespondsto1.90international-dollars(int.-$)perpersonperday.Byanystandardthisisanextremelylowpovertyline–theterm“extremepoverty”ismorethanappropriate.Becauseofthisitisimportant tomeasurepovertynotjustbyoneverylowpovertyline,butmanyotherpovertylinesaswell.Arepovertyratesreallygoingdown?Thevisualizationshowstheglobalincomedistributionin2003and2013(belowwewilllookatalongertimeperiod).Itismeasuredininternational-$whichmeansitisadjustedforpricedifferencesbetweencountries,asweexplainhere.Itisofcoursealsoadjustedforpricechangesovertime(inflation).Anincomeover14,500int-$in2013meantthatpersonwaspartoftheglobaltop10%,andintheworld’srichestcountriesthemajorityofpeoplehaveanincomethatisoftenmuchhigherthanthan.Whatthisdistributionshowsisthatglobalincomeinequalityisextremelyhigh.Thetop10%cut-offis30-timeshigherthanthecut-offforthepoorest10%.Herethedataisplottedonalogarithmicy-axistofocusonthechangeofincomesbetweenthemajorityoftheworld’spopulation,hereisthesamedataplottedonalinearaxis.Toreadthechartbelow,choose alevelofannualincomeonthey-axisandthenusetheblue2013-lineandthered2003-linetofindthecorrespondingshareoftheworldpopulationlivingwithlessthanthatincome onthex-axis.Thefirstthingthatthischartshowsisthatalargeshareoftheworldpopulationlivesonverylowincomes.Themedianincomein2013was2,010int.-$.IfyouwanttoconsiderapovertylinehigherthantheInternationalPovertyLine,youcouldchosealineofint.-$1,000peryearinsteadandseethatin2003,48%oftheworldpopulation wasbelowthispovertyline;tenyearslater,in2013,29%werebelowthisline.Thiswasadeclineof20percentagepointsinonedecaderelativetothishigherpovertyline.Ifyouthinktheinternationalpovertylineshouldbemuchhigherandshouldinsteadbe4,000 int.-$,thenyouseethatin2003,80%oftheworldpopulationwasbelowthatpovertyline.10yearslater:67%.Adeclineof13percentagepointsinadecade.Thereisabsolutelynoreasontobecomplacentaboutpovertytoday–itremainsoneoftheworld’sveryworstproblems.Butitisclearthattheworldhasmadeprogressagainstit. Whatthischartshowsisthat,nomatterwhatglobalpovertylineyouchoose,theshareofpeoplebelowthatpovertylinehasdeclined.Theglobalincomedistributionin2003 and201320TheshareoftheworldpopulationrelativetovariouspovertylinesThestudybyMauroandHellebrandt21onwhichtheabovechartisbasedonlyhasdatafrom2003onwards. But thereissomegooddatathatallowsustogofurtherbackintime,aswellaslookingatabsolutenumbersofpeopleinpoverty(ratherthanshares).Thevisualizationshowsabreakdownofthepopulationbypercapitahouseholdconsumption.Afterseeingthedatafor2003to2013,thedatashownhereshouldnotbesurprising:Globallytheshareofpeoplebelowanypovertyline–$1.90,$3.20,$5.50,$10–isdeclining.Progressagainstapovertylineofint.-$10perdayisaveryrecentachievement. Morethanathirdoftheworldpopulationnowliveonmorethan10dollarsperday.Untiladecadedecadeagoitwasonlyaquarter.Themajorityoftheworldpopulationisstillverypoor.Whatthecutoffforextremepovertyishelpfulforistofocustheattentiontothosewhoaretheverypoorest.Thiswouldnotbepossibleifwewouldonlyrelyonmuchhigherpovertythresholds.Apovertylineofint.-$10perdaywouldincludetheverypoorest(onlessthan$2)andthosethatare5-timesricherandwouldtherebyhideimportantdifferences. Clicktoopeninteractiveversion Ratesaregoingdown–whataboutthetotalnumberofpeopleinandoutofpoverty?Thechartbelowanswersthequestionofhowthenumberofpeople belowdifferentpovertylinesischanging.The numberofpeoplelivingonlessthan$1.90perdayhasdeclined.Andthenumberofpeopleabovethepovertylinehasincreasedrapidly.Whenwelookathigherpovertylinesweseeadifferentpicture:From1990to2005weseethatthenumberofpeoplelivingonlessthan$10perdayincreased.Butsincethen,thenumberlivingonlessthan$10stagnatedwhilstthenumberabovethispovertylineincreasedrapidly.Thenumberofpeoplewholiveonmorethan10dollarsperdayincreasedby900millioninthelast10years.Onanyaveragedayinthelastdecadethenumberofpeoplelivingonmorethan$10increasedbyalmostaquarterofamillion(246,500).22Halfabillionprojectedtoliveinextremepovertyin2030Theworldismakingprogressagainstallpovertylinesandwithrapidgrowthinmanymiddleincomecountrieswecanhopethatthisprogressagainstpovertyrelativetohighpovertylineswillcontinue.Butourfocusshouldbeonthoseintheveryworstpoverty.Asthepreviouschartshows,therearestill730millionpeoplelivingonlessthanint-$1.90perday.Andthebadnewsisthatwecannotexpectthisprogresstocontinue:AsIhaverecentlywritten,becausetheworld’sverypooresteconomiesarestagnatinghalfabillionareexpectedtobeinextremepovertyin2030. Clicktoopeninteractiveversion HowmuchdoesthereductionoffallingpovertyinChinamatterforthereductionofglobalpoverty?Inrecentdecades,theshareinextremepovertyhasdeclinedfasterthaneverbeforeinhumanhistory.Acommonresponsetothisfactis‘Yes,butthisisonlybecauseofChina.’Thispostaskswhethersuchremarksaretrue.Is thesubstantialdeclineofglobal povertyonlyduetothepovertydeclineinChina?ThehistoricaldeclineofextremepovertyinChinaFirst,letuslookatthehistoricalevolutionofpovertyinChina.ShownindarkblueisthedecliningshareoftheChinesepopulationlivingbelowtheInternationalPovertyLine(1.90int.-$),accordingtoWorldBankestimates.In1981around88%oftheChinesepopulationlivedinextremepoverty.Accordingtothelatestestimates, extremepoverty–measuredinthesameway–hasdeclinedtobelow1%inChina.Thedecliningshareofpeoplebelowhigherpovertyisalsoshowninthisvisualization. Clicktoopeninteractiveversion ComparingaworldwithandwithoutChinaThedeclinefromalmosteveryChinesepersonlivinginextremepovertytoalmostnoChinesepeople livinginextremepovertyisofcourseanexceptionalachievement.Butisthistheentirestoryoffallingglobalpoverty?TofindtheanswerwerecalculatedtheshareofpeoplelivinginextremepovertygloballyanddisregardedChinaentirely–sothatwe compareaplanetwithChinatoaplanetwithoutChina.23Thechartshowstheresults.Inblueisthedeclineofglobalpoverty,inredthedeclineofpovertyexcludingChina.Weseethatthereductionofglobalpovertywas verysubstantialevenwhenwedonottakeintoaccountthepovertyreductioninChina.In1981almostonethird(29%)ofthenon-Chineseworldpopulationwaslivinginextremepoverty.By2013thissharehadfallento12%.ExtremepovertydeclinedinChinaandintherestoftheworld.Whatisalsointerestingtoseeinthechart isthatuntil2005,theinclusionofChinaincreasedtheshareoftheworldpopulationlivinginextremepoverty;butsincethen, thishasreversed,andtheinclusionofChinaisnowreducingtheglobalpovertyheadcountratio.Thisisbecausein2005China’spovertyratiofellbelowtheworldpovertyratio.Additionally, itisofcoursesillyanywaytosay‘thedeclineofglobalpovertyisonlybecauseofChina’. Wecareaboutpeople–notaboutcountries,andsincemorethanevery5thpersonintheworldisChinese,itisareally importantachievementfortheworldthatextremepovertyhas decreasedsosubstantiallyinChina. Clicktoopeninteractiveversion The(mis)perceptionsaboutpovertytrendsMisperceptionintheUKDespitetheclearevidence,manypeoplearenotawareofthefactthatextremepovertyisdecliningacrosstheworld.Thechartshowstheperceptionsthatsurvey-respondentsintheUKhaveregardingglobalachievementsinpovertyreductions.Whiletheshareofextremelypoorpeoplehasfallenfasterthaneverbeforeinhistoryoverthelast30years,themajorityofpeopleintheUKthinksthattheoppositehashappened,andthatpovertyhasincreased.TheextentofignoranceintheUKisparticularlybadifwetakeintoaccountthattheshownresultcorrespondstoapopulationwithauniversitydegree.SeetheGapminderIgnoranceProject formoreevidence.SurveyresponseintheUKtothequestion howglobalpovertyhaschanged24GlobalmisperceptionsNotonlyintheUKaremanywronglyinformedabouthowpovertyischangingglobally.Acrossalargenumberofcountries,themajorityofpeople–52%–believethattheshareofpeopleinextremepovertyisrising.Theshareofcorrectanswersdifferssubstantiallyacrosscountries.ThecountriesImarkedwithastararethosethatwerealow-incomeorlower-middle-incomecountriesagenerationago(in1990).Inthesepoorercountriesmorepeopleunderstandhowglobalpovertyhaschanged.Peopleinrichcountriesontheotherhand–inwhichthemajorityofthepopulationescapedextremepoverty somegenerationsago –haveaparticularlywrongperceptionaboutwhatishappeningtoglobalpoverty.ThesameresultwasalsofoundinasurveycommissionedbyOxfam25andOxfam,alargecharitythatfocusesonthealleviationofglobalpoverty,warnedthat“publicpessimismandmisunderstandingcouldunderminethefightagainstglobalpoverty”. ThedemographicsofextremepovertyInthissectionHowmanypoorpeopleliveineachcountry?Whoarethepeoplelivinginextremepoverty?Howmanychildrenliveinextremepovertyaroundtheworld?Howdoespovertyamongchildrencomparetopovertyamongadults?Howmanypoorpeopleliveineachcountry?Theglobalincidenceofextremepovertyhasgonedownfromalmost100%inthe19thcentury,to10.7%in 2013.Whilethisisagreatachievement,thereisabsolutelynoreasontobecomplacent:apovertyrateof10.7%meansatotalpovertyheadcountof746 millionpeople.Wheredotheylive? Thefollowingvisualizationprovidesabreakdownofthisfigurebycontinentandcountry.Thesefigurescomefrommultiplyingestimatesofpovertyratesbythecorrespondingestimatesoftotalpopulationinthosecountries.ThepovertyrateestimatescomefromtheWorldBank(2016PovCalrelease,using2013householdsurveydata);26andtotalpopulationestimatescomefromtheWorldDevelopmentIndicators.AsusualwithWorldBankestimates,povertymeasuresareadjustedtoaccountfordifferencesinpricelevelsbetweencountries.Thisisreflectedinthe‘internationaldollar’metricusedtomeasureincomes.Aswecansee,today,Africa isthecontinentwiththelargestnumberofpeoplelivinginextremepoverty.Thebreakdownbycontinentisasfollows:383MillioninAfrica327MillioninAsia19MillioninSouthAmerica13MillioninNorthAmerica2.5MillioninOceania0.7MillioninEuropeWecanalsoseethatIndiaisthecountrywiththelargestnumberofpeoplelivinginextremepoverty(218millionpeople),withNigeriaandtheCongo(DRC)followingwith86and55millionpeople,respectively.Thesefiguresaretheresultofimportantchangesacrosstime.Aswementionedaboveinourdiscussionofregionaltrends,in1990Asiawastheworldregionwiththelargestnumberofpoorpeople(505millioninSouthAsia,plus966millioninEastAsiaandthePacific). However,withrapideconomicgrowth inAsiaoverthepasttwodecades,povertyinAsiafellmorerapidlythaninAfrica.Whoarethepeoplelivinginextremepoverty?TheWorldBankGrouprecentlypublishedanewsetofpovertyestimates,aspartoftheirreportPovertyandSharedProsperity(2016).Theseestimates,explainedindetailintworelatedbackgroundpapers(Newhouseetal.2016andCastanedaetal.2016)27areconsistentwiththeofficialWorldBankpovertyfigurespublishedinPovcalandtheWorldDevelopmentIndicators,buttheyaredisaggregatedbykeydemographiccharacteristicssuchasageandeducationalattainment.Inordertoproducedisaggregatedestimates,theWorldBankreliedonnewdatafromtheGlobalMicroDatabasethataugmentssurveydatain89countries,byprovidingasetofharmonizedhouseholdcharacteristics,enrichingtheothersurveyinstrumentsusedbytheWorldBanktomeasurepoverty.AccordingtotheWorldBank,thesampleof89countriesincludedintheGlobalMicroDatabasecontainsanestimated84.2percentofthepopulationinlowandmiddleincomecountries,and82.1percentofthechildpopulation.28Inthismapyoucanseeexactlywhichterritoriesarecovered.Astheauthorspointout,whilenoteverycountryiscovered,thisnewsetofestimatesisthemostupdatedandcomprehensivesourcecurrentlyavailabletoresearchersandpolicymakerstryingtounderstandthedemographicsofpoverty.Thefollowingvisualizationusesthissourcetoprovideacharacterizationofthosewholiveinextremepoverty.Aswecansee,acrossallworldregionsthepoortendtobeyoungandliveinruralareas.Inthebackgroundpaperaccompanyingthedata,Castanedaetal.(2016)providesimpleregressionresultsandconcludethat“Afterconditioningonotherindividualandhouseholdcharacteristics,havingfewerthanthreechildren,havinggreatereducationalattainment,andlivinginanurbanareaarestronglyandpositivelyassociatedwitheconomicwell-being”.Interestingly,andperhapsalsosurprisingly,wecanseefromthisvisualizationthatthosewithnoeducationarenowadistinctminorityofthepopulation.29Onethingthisshowsisthat,despiteimprovements,theexpansionofeducationaroundtheworldinthelastdecadeshasstillnotbeenenoughtoliftmanyhouseholdsoutofpoverty.Howmanychildrenliveinextremepovertyaroundtheworld?Globalestimatesofchildpovertyareunfortunatelynotavailable.However,aswementionedabove,wecanhaveareasonablepictureofthisissuebylookingattheestimatesrecentlypublishedbytheWorldBankusingtheGlobalMicroDatabase.Formeasurementpurposes,childrenareconsideredtobepooriftheyliveinapoorhousehold(i.e.allchildreninpoorhouseholdsareassumedtobepoor,whileallchildreninnon-poorhouseholdsareassumedtobenon-poor).Ahouseholdisconsideredpoor,inturn,ifthepercapitaconsumptionofitsmembers(orpercapitaincome,dependingonthecountry),fallsbelow1.90int.-$.ThisisthestandarddefinitionofabsoluteextremepovertyusedbytheWorldBank.Thefollowingchartsummarizestheavailabledata.Theheightofeachbarinthisplotshowstheshareofpeoplelivinginextremepovertybyagegroup,whilethewidthofthebarsreflectsthetotalsizeofeachagegroupintheoverallpopulation.Theareaofeachbar(heighttimeswidth)givesthenumberofindividualslivinginextremepovertywithineachagebracket—thesearethenumberswritteninsideeachbar.Aswecansee,povertyisparticularlyhighamongchildren:inlowandmiddleincomecountriesmorethan20%ofchildrenunder10yearsofagelivewithlessthan1.90int.-$perday.Foradults,thecorrespondingfiguresaremuchlower:lessthan10%ofadultslivewithcomparablylowconsumptionlevels.Bylookingatthetotalnumberofpeopleinextremepoverty(areaofthebars)wecanalsoseeanotherimportantfact:virtuallyhalfofthepeoplelivinginextremepovertyareunder18yearsofage.Thisisalargeshareifweconsiderthatthoseunder18accountforonlyaroundathirdofthegeneralpopulation(asshownbythewidthofthebars).Extremepovertyinlowandmiddleincomecountriesbyagegroup(2013)30Howdoespovertyamongchildrencomparetopovertyamongadults?Theabove-mentioneddatafromtheGlobalMicroDatabaseallowsustostudypovertyacrossagegroupsforvariouspovertylines—notjusttheInternationalPovertyLine.Thefollowingchartshowsthecumulativedistributionofwelfarefordifferentagegroups.Eachofthelinesinthisplotshows,foreachagegroup,theshareofthepopulationlivingbelowagivenlevelofpercapitadailyincomeorconsumption(afteraccountingfordifferencesinpricesacrosscountries).Ifyoulocatetheverticallinepassingthrough$1.9inthehorizontalaxis,youwillseethatitcutstheseriesforadultsataround9%.Thismeansthataround9%oftheadultpopulationliveswithconsumption(orincome)levelsbelowthe1.90int.-$povertyline.Followingthislogic,wecanreadthepovertyratesatanypovertyline.Aswecansee,thedistributionofconsumptionforadultsisalwaystotherightofthedistributionforchildren.Ineconomicslingo,whatweobserveisthatthedistributionforadultsstochasticallydominatesthatofchildren.Thismeansthatpovertyratesforchildrenarehigheratanypovertyline.It’simportanttomentionthattheseresultsdonotreflectthefactthatadultstendtogeneratemoreincomethanchildren.Bearinmindthattheseareestimatesofhouseholdpercapitaincome.Thatmeansthatchildrenlivinginhouseholdswithrichadultsarealsoassumedtoberich.Percentofpeoplelivingbelowdifferentlevelsofconsumptionorincomeinlowandmiddleincomecountries,byagegroup(2013)–UNICEF(2016)31 PovertyacrossmultipledimensionsInthissectionHowcanwemeasurepovertybeyondincomeandconsumption?Howcanwemeasurepovertybeyondincomeandconsumption?ThemethodologyusedbytheWorldBanktomeasurepovertyreliesonincomeandconsumption.Whileinformative,thismethodologycertainlyleavesoutmanyimportantaspectsofwelfare.AtOurWorldinData,webelievethatitisimportanttotrackprogressindimensionsofwell-beingspanningbeyondstandardeconomicindicators.Thisiswhywemakeanefforttostudyawiderangeofaspects,includingeducation,health,humanrights,etc.Ifyouareinterestedinunderstandingpovertythroughtheseotherlenses,youarewelcometoexploreourwebsite—thecontentmenuatthetopofthepagelinkstoallofourentriesonthesetopics.Trackingvariousindicatorsofwell-beingindependentlycanmakecomparisonsdifficult,sincedifferentindicatorsmoveindifferentdirectionsacrosstimeandspace.Becauseofthis,researchersandpolicymakersoftenconstructsyntheticindicatorsthataggregatevariousdimensionsofdeprivation,byattachingwelfareweightstoasetofkeyunderlyingmetricsofwell-being.TheMultidimensionalPovertyIndex(MPI)publishedbytheOxfordPoverty&HumanDevelopmentInitiative(OPHI),isonesuchefforttoaggregatevariousaspectsofwell-beingintoasinglemetric.DifferentfromotherindexesliketheHumanDevelopmentIndex,theMPIisnotaggregatedatthecountrylevel,butinsteadattheindividuallevel—itmeasureshowoneandthesameindividualisdeprivedindifferentdimensions.OPHI’sMPIiswidelyusedaroundtheworld,andcurrentlycoversover100lowandmiddleincomecountries.TheMPIisconstructedfromtenindicatorsacrossthreecoredimensions:health,educationandlivingstandards.Thistablespecifieshowthedifferentindicatorsaredefinedandaggregated.TheMPIisconstructedusingthreemaindatasets:theDemographicandHealthSurvey(DHS),theMultipleIndicatorsClusterSurvey(MICS),andtheWorldHealthSurvey(WHS).YoucanfindfurtherdefinitionsandexplanationsintheMPI’sdocumentation.AndyoucanfindamoretechnicaldiscussionoftheMPIanditspropertiesinAlkireandFoster(2011).32TheMPIistypicallyusedtoassessdeprivationattheindividuallevel:ifsomeoneisdeprivedinathirdormoreoftheten(weighted)indicators,theindexidentifiesthemas‘MPIpoor’.Inthefollowingmap,weshowtheshareofMPIpoorpeoplecountrybycountry(i.e.themultidimensionalpovertyheadcountratios).Aswecansee,thisalternativemetricshowsthatpovertyisalsoparticularlyacuteinsub-SaharanAfrica. Clicktoopeninteractiveversion Howwelldoincomeandconsumptioncapturemultidimensionalpoverty?Aswementionedabove,povertyismultidimensionalinnature,anditisthereforeusefultotrytomeasurepovertythroughalternativeinstrumentsthatcapturedeprivationbeyondincomeandconsumption.TheMultidimensionalPovertyIndex(MPI)—shownintheworldmapaboveandpublishedbytheOxfordPoverty&HumanDevelopmentInitiative(OPHI)—isthemostcommoninternationalinstrumentusedinthiscontext.Thefollowingchartplotstheshareofpeoplelivinginextremepovertyasmeasuredbyconsumptionandincome,againsttheshareofpeoplelivingin‘multidimensionalpoverty’accordingtotheMPI.Theformeristhesamemetricwehavediscussedextensivelythroughoutthisentry.Thelatterisametricbasedonunmetneeds:theMPI’sdefinitionstipulatesthatsomeonelivesin‘multidimensionalpoverty’iftheyaredeprivedinathirdormoreofthetenweightedindicators(suchas,forexample,nutrition,electricity,orschooling)thatcomposetheindex.Aswecansee,thereisapositivecorrelationbetweenthesetwomeasuresofdeprivation,buttheyareclearlynotidentical.KenyaandChadhavesimilarmonetarypovertyrates(about35%ofthepopulationlivebelowtheInternationalPovertyLinein2015),buttheyhaveextremelydifferentmultidimensionalpovertyrates(around40%inKenyain2014,comparedto87%inChadin2015werelivingin‘multidimensionalpoverty’).Thishighlightstheusefulnessoftrackingdeprivationacrossmultipledimensionsofwell-being,includingbothstandardandnon-standardeconomicindicators. Clicktoopeninteractiveversion DeterminantsandconsequencesInthissectionThelinkbetweeneconomicgrowthandpovertyPovertytrapsEvidenceonspecificstrategiestoreducepovertyCross-countrycorrelatesThelinkbetweeneconomicgrowthandpovertyTheshareofthepopulationinextremepovertyislowwhenaverageincomesarehighNationalprosperityisastrongpredictorofextremepovertyattheindividuallevel.Thefollowinggraphshowsthisrelationshipbetweenaverageincomes(GDPpercapita)andtheshareofthepopulationlivinginextremepoverty.ThechartshowsthattodaythereisnocountrywithaGDPpercapitahigherthan15,000int.-$inwhichmorethan6%ofthepopulationlivesinextremepoverty.AndinmostcountrieswithGDPpercapitabelow4,000int.-$,betweenonequarterandthreequartersofthepopulationlivesinextremepoverty.Thescatterplotisinteractive—bymovingthetimesliderundertheplot,youcanseethechangeovertime. Clicktoopeninteractiveversion ExtremepovertydecreaseswhenaveragenationalincomesgrowHowpovertychangesisnotonlyaconsequenceofeconomicgrowth,italsodependsonthedistributionofincomesandhowthisinequalitychangesduringthegrowthprocess.Ifgrowthonlyliftstheincomesatthetop,povertylevelswillremainunchanged.Ontheotherhand,ifgrowthisinclusiveandliftsallboats,theeconomyisabletoreduceabsolutepovertyovertime.Asdiscussedinourentryonincomeinequality,incomeinequalityhasdevelopedquitedifferentlyindifferentcountries.InIndia,forexample,inequalityhasincreased,whileinmostLatinAmericancountries,inequalityhasfallen.Researchershavecomparedhowmuchchangesininequalitymatterforpovertyreductionrelativetoeconomicgrowth.DavidDollarandAartKraaystudiedthislinkbetweengrowth,inequalityandpovertyreductioninawidelycitedpaperin2002.33Thetitleoftheirpaperisthesummaryoftheirfinding:‘Growthisgoodforthepoor’.Theauthorsfindthattheshareofincomeofthepoorestquintiledoesnotvarysystematicallywithaverageincome—or,inotherwords,thattheincomesofthepooronaverageriseproportionatelywithaverageincomes—andthatconsequently,“growthonaveragedoesbenefitthepoorasmuchasanyoneelseinsociety”.Therefore,theauthorsrecommendthat“growth-enhancingpoliciesshouldbeatthecenterofanyeffectivepovertyreductionstrategy.”Theauthorsemphasizethattheirfindings“donotimplythatgrowthisallthatisneededtoimprovethelivesofthepoor”orthattheirfindingswould“suggesta‘trickle-down’processorsequencinginwhichtherichgetricherfirstandeventuallybenefitstrickledowntothepoor”.TwelveyearslaterthesametwoauthorsandTatjanaKleinebergrevisitedthequestionontheconsequencesofgrowthandchangesininequality.Intheirnewerpaper,theybroadenedthescopeoftheresearchquestiontostudysocialwelfare.Thisapproach—usingtheconceptofsocialwelfare—takesintoaccountnotjustpoverty,butalsothechangeinlivingstandardsofindividualsabovethepovertyline.Asintheirearlierresearch,Dollar,Kleineberg,andKraay(2014)34studiedalargenumberofcountriesoverthepast40years.Thethreeauthorssummarizetheirresearchbyconfirmingtheirfindingfrom2002:“Mostofthecross-countryandover-timevariationinchangesinsocialwelfareisattributabletogrowthinaverageincomes.Incontrast,thecontributionofchangesinrelativeincomestosocialwelfaregrowthisonaveragemuchsmallerthangrowthinaverageincomes,andmoreoverisonaverageuncorrelatedwithaverageincomegrowth.”Thefollowingchartfocusesonthepopulationlivinginextremepoverty.Itplotsthechangeofnationalaverageincomeagainstthechangeinextremepovertylevelsovertime.Eachcountryisshownhereoverasuccessionofpoints,oneforeachyearlyobservationofGDPandpoverty.AscountrieslikeIndia,Brazil,Indonesia,andChinagotricher,theshareoftheirpopulationlivinginextremepovertyhasfallen.Onewaytothinkaboutthisistoconsiderhowlowprosperityisbeforeaneconomyachievessustainedeconomicgrowththatliftsthemajorityofthepopulationoutofpoverty.Indiain1983hadaGDPpercapitaof1,070int.-$.Attheendoftheperiodintheconnectedscatterplot,averageincomewasmorethan4-timeshigherat4,560int.-$.Overtheperiodshownintheconnectedplot,Brazil’saverageincreased3-foldandChina’saverageincomeincreasedeven6-fold.Persistenteconomicgrowthreallyisaverypowerfulforce,andthefindingsofDollar,Kleineberg,andKraayandthechartmakethisveryclear.Whatistruefortherecentdecadesisalsotrueforthelong-runperspectiveonaglobalscale.Withouttheincreaseinproductivitythatbroughteconomicgrowth,itwouldnothavebeenpossibletolifthundredsofmillionsofpeopleoutofextremepoverty.Withoutlarge-scaleeconomicgrowth,manymorepeoplewouldstillliveattheverypoorlevelsofmaterialwell-beingthatcharacterizedourancestors’existenceformillennia.Seenfromthelonghistoricalperspective,itisclearthatcountrieshavetobeextraordinarilyrichtohavethepossibilitytoendextremepovertyforthemajorityoftheirpopulation.PovertytrapsWhatare‘povertytraps’?Economistsusetheterm‘povertytrap’todenoteasituationinwhichindividualsarestuckindeprivationoverlongperiodsoftime,andthereisnothingtheycandobythemselvestoescapetheirsituation.Theideaissimple:povertytodaycausespovertyinthefuture,sohouseholdsthatstartpoor,remainpoor.Insufficientnutrition,forexample,canleadtoapovertytrap.Moreprecisely,ifphysicalcapacitytoworkincreasesnonlinearlywithfoodintakeatlowlevels(i.e.ifthefirstcaloriesthatweconsumeareusedbyourbodytosurvive,ratherthantoprovidethestrengthrequiredtowork),itispossiblethatthoseinextremepovertygetstuckinaperverseequilibriumcharacterizedbylowincomesandlownutrition:poornutritionthenbecomesboththecauseandconsequenceofpoorincomes.Conceptually,povertytrapscanalsotakeplaceatacollective‘macro’level.Forexample,low-incomecountriesmightlackthegoodgrowthfundamentals(e.g.technology,education,infrastructure,etc.)thatarenecessaryforthehighsavingrateswhichleadtoproductivitygainsandrisingnationalincomes.Theconceptofpovertytrapsisimportantinthecontextofpolicy,sinceitimpliesthatone-offpolicyeffortsthatmakeitpossibleto‘escapethetrap’havepermanentpositiveeffects.Thisistherationaleoftenusedtoarguefor‘bigpush’macropoliciessuchastheexpansionofmicro-financeinlow-incomecountries.Suchpoliciesaremeanttotriggeravirtuouscycleofmoresavings,moreinvestment,andeconomicgrowth.Aswediscussbelow,althoughunidimensionalpovertytrapssuchasthosecausedbysinglefactorsareconceptuallyappealing(e.g.nutrition-basedtraps,orcountry-levelsavingstraps),thereislittleempiricalevidencesupportingtheirpracticalrelevance.Theevidencesuggeststhatmulti-prongedinterventionsaimedatrelievingmultiplejointconstraintsatthehousehold-level,aremorelikelytoreducepovertythan‘bigpush’policiesonthemacro-level.Howcommonarepovertytraps?Asmentionedabove,a‘povertytrap’isasituationwhereincomesarestagnantoverlongperiods,because‘povertytodaycausespovertyinthefuture’.Thefollowingchartprovidessomeevidenceregardingthecross-countryevolutionofincomesovertime.Itplots,foreachcountry,thenationalincomein1960againstthecorrespondingnationalincomein2014.GDPpercapitaisusedtomeasurenationalincomes,andfiguresareexpressedin‘realterms’,whichmeanstheyareadjustedforinflation.Inthischart,countrieswithstagnantincomesareclosetotheblue45°line,whilecountrieswithincomesthatrosebetween1960and2014areabovethe45°line.Thelatterarethecountrieswhichexperiencedincomegrowthoverthese54years.Aswecansee,somecountriessuchasMadagascar,Chad,Senegal,andNicaraguaexperiencedincomestagnation—theyarerightonthe45°degreeline.AndacoupleofcountriessuchasNigerandtheDemocraticRepublicofCongohaveevenexperiencednegativegrowthoverthereferenceperiod.Butthelargemajorityofcountries,allthoseabovetheblueline,haveexperiencedgrowth.Thosecountriesthatarefarabovethebluelinehadthestrongestgrowth.Botswana(38-foldincrease),SouthKorea(30-fold),Romania(15-fold),China(11-fold),andThailand(18-fold)aresomeofthecountrieswiththestrongestgrowthoverthese54years.Acloserlookatthedatasuggeststhatthetypicalpoorcountrygrewatleastasfastastheglobalaverageoverthisperiod.35Ofcourse,whatweseeinthischartisonlypartofthestory,sincethemicroandmacrodynamicsofincomescanbeverydifferent.Itispossible,forexample,thatcountry-levelaverageincomesarenotstagnant,buthousehold-levelincomeslagforparticularsegmentsofthepopulationwithinthosecountries.Indeed,intheUSthereisevidenceofstagnatingincomesforthoseatthebottomofthedistribution.Thus,apropertestfortheexistenceofpovertytrapsrequiresamoresophisticatedeconometricanalysis.KraayandMcKenzie(2014)37providesuchananalysisinaninterestinganddetailedreviewoftheavailablestudiestestingfortheexistenceofmechanismsleadingtopovertytraps.Theyarguethatthereislimitedevidenceformostofthemechanismswhenoperatinginisolation;exceptperhapsforspatialpovertytraps(individualsbeingtrappedinlow-productivitylocations),andbehavioralpovertytraps(individualsbeingstuckinsituationswheretheydevotethemostmentalefforttomeetingdailyneeds,leavinglittleattentionalresourcesforsolvingotherproblemsthatcouldraisetheirincomes).Theimplicationofthisevidenceshouldnotbethatthereisnoroleforpolicy;butratherthattraditional‘bigpush’macropoliciesareperhapsnotthebestapproachtoreducingpoverty.Other,lesstraditionalpoliciesmightworkbetter.Belowwediscusssomeexamples,suchasencouragingmigration,andimplementingmultifacetedprogramsthatrelievejointconstraintsatthehouseholdlevel.RealGDPpercapita,1960vs201436EvidenceonspecificstrategiestoreducepovertyMultifacetedhousehold-levelinterventionsAroundtheworld,mostgovernmentprogramshopetoreducepovertythroughshort-terminterventionsthathavelastingeffects.Whilethisisnotaneasytask,thereisconcreteevidencesuggestingthatitispossible.Insixdifferentcountries,amultifacetedprogramofferingshort-termsupportalongvarioushouseholddimensionshasbeenshowntocauselastingprogressfortheverypoor.Theinterventioninquestionconsistsofsixelements:(1)aproductiveassetgrant,(2)temporarycashconsumptionsupport,(3)technicalskillstraining,(4)highfrequencyhomevisits,(5)asavingsprogram,and(6)healtheducationandservices.Thefollowingvisualizationsummarizestheevidence.Thelightbluebarsshowtheimpactofthisintervention,measuredbytheyearlyaverageincreaseinhouseholdconsumption,threeyearsaftertheproductiveassettransferandoneyearaftertheendoftheprogramintervention.Thedarkbluelinepresentsthe‘netpresentvalue’oftheseeffects—thatis,thevalueofthebenefitsassumingthesegainslastforever,minustheprogramcosts(discountingthebenefitsandcostsbyhowfarinthefuturetheyoccur).Althoughthecostsofthisinterventionaresubstantial,wecanseethatthenetbenefitsarestillpositiveandlarge—preciselybecauseimpactsaresustainedintothefuture.TheshownresultscomefromRandomizedControlledTrials(RCTs).Thisevaluationtechniqueconsistsinadministeringthepolicyinterventiontoarandomgroupofindividuals(the‘treatmentgroup’)andevaluatingtheeffectbycomparingoutcomesagainstanothergroupofindividualswhowerenotaffectedbythepolicy(the‘controlgroup’).Thisisalsotheideabehindmedicaltrials,andhasbecomeincreasinglypopularindevelopmentresearch.ThefullstudyandresultsareexplainedinBanerjeeetal.(2015).38Theyreporttheimpactsonconsumption,foodsecurity,productiveandhouseholdassets,financialinclusion,timeuse,incomeandrevenues,physicalhealth,mentalhealth,politicalinvolvement,andwomen’sempowerment.Theyfindstatisticallysignificantimpactsonalloftheseoutcomes.MigrationasawaytobreakoutofextremepovertyTheevidencemostconsistentwithpovertytrapscomesfrompoorhouseholdsinremoteruralregions—thesearehouseholdsthataretrappedinlow-productivitylocations,butwhichcouldenjoyarisingstandardoflivingiftheyweresomehowabletoleave(seeKraayandMcKenzie201439forareviewoftheevidence).Howdopoorhouseholdsget‘trapped’inlow-productivitylocations?Therearemanypossiblemechanisms—oneisthelackoffinancialresources.Bryan,Chowdhury,andMobarak(2013)40arguethathouseholdsclosetosubsistenceareoftenunwillingtotaketheriskofmigration;buttheybecomemorewillingtodosoifinsuredagainstthisrisk.Thisrelaxestheliquidityconstraintandopensawindowofpossibilityforpoliciesaimingtopromotemigration,bothwithinandacrosscountries.HowlargearethepotentialgainsfrommigrationtoahighproductivitycountrysuchastheUnitedStates?Clemens,Montenegro,andPritchett(2016)41offeratentativeanswer.Specifically,theyprovidealowerboundestimateontheannualwagegainoflow-skilledmaleworkersmigratingtotheUnitedStatesfromvariouslow-incomecountries.Thefollowingvisualizationplotstheirresults,andcomparesthemtothebenefitsfromthesuccessfulmultifacetedanti-povertyinterventionwediscussedabove.Aswecansee,theeffectofmigrationforthepoorisremarkablyhigh.Thesefiguressuggestthatthetotallifetimevalueofthemostsuccessfulanti-povertyprogramislessthanaquarterofthegaineveryyearfromlettingaworkerworkinahighproductivityenvironment,inthiscasetheUnitedStates.Ofcourse,fromasocialwelfarepointofview,theseeffectshavetobeconsideredinconjunctionwiththeeffectson‘native’workersinthenewhostenvironments.Tothisend,OttavianoandPeri(2011)42estimatethatovertheperiod1990–2006,immigrationtotheUnitedStateshadatmostamodestnegativelong-runeffectontherealwagesoftheleasteducated‘natives’.Astheauthorsexplain,thisispossiblebecausetherearecomplementaritiesamongdifferenttypesofworkers:‘natives’and‘immigrants’ofsimilareducationandagehavedifferentskills,oftenworkindifferentjobsandperformdifferentproductivetasks.Givingpeoplemoney–conditionalcashtransfersTargetedtransferprogramshavebecomeanincreasinglypopularpolicyinstrumentforreducingpovertyinlow-incomecountries.Theyareanobviousinstrumenttoconsider,sincetransferringcashisperhapsthemoststraightforwardwayofraisingincomes;andwhencoupledwithwell-designedconditionalities,transferscanhelp‘nudge’participantswhoarecaughtupin‘psychologicalpovertytraps’(seeourdiscussionofpovertytrapsabove).Gentilinietal.(2014)43reportthat119developingcountrieshaveimplementedatleastonetypeofunconditionalcashassistanceprogram,and52countrieshaveconditionalcashtransferprogramsforpoorhouseholds.Cashtransferprogramshavebeenshowntoreducepovertyacrossavarietyofcontexts.FiszbeinandSchady(2009)44provideacomprehensiveanalysisoftheevidence.Theyconcludethat“Byandlarge,[ConditionalCashTransfers]haveincreasedconsumptionlevelsamongthepoor.Asaresult,theyhaveresultedinsometimessubstantialreductionsinpovertyamongbeneficiaries—especiallywhenthetransferhasbeengenerous,welltargeted,andstructuredinawaythatdoesnotdiscouragerecipientsfromtakingotheractionstoescapepoverty.”Asthelastpartoftheconclusionabovenotes,acommonconcernamongpolicymakersisthatwelfareprogramscanpotentiallydiscouragework—infact,thisisaconcernthatissharedbypolicymakersinbothlow-andhigh-incomecountries.Banerjeeetal.(2015)45analyzethedatafromsevenrandomizedcontrolledtrialsofgovernment-runcashtransferprogramsinsixdevelopingcountriesindifferentworldregionsandfindnosystematicevidencethatcashtransferprogramsdiscouragework.Thechartprovidesagraphicalsummaryoftheirmainfindings.Inthetoppanel,theauthorsgraphtheemploymentrateforalleligibleadultsinboththecontrolandtreatmentarmsforeachevaluation.Thebottompanelreplicatestheoneabove,butforhoursofwork.Aswecansee,theoverallfiguresforbothemploymentandhoursofworkaresimilaracrosstreatmentandcontrolinalloftheevaluatedprogramsanddonotstatisticallydiffer.Experimentalestimatesoftheeffectofcashtransfersonworkoutcomes–Banerjeeetal.(2015)Offshoringoflow-skilledjobsGrowinginternationaltradehaschangedourworlddrasticallyoverthelastcoupleofcenturies.Oneparticulareffecthasbeenasubstantialincreaseinthedemandforindustrialmanufacturingworkersinlowincomecountries,mainlyduetotheriseinoffshoringoflow-skilledjobs.Acommonargumentputforwardisthattheseindustrialmanufacturingjobsareapowerfulinstrumentforreducingpoverty,evenifsalariestendtobeverylowbythestandardsofrichcountries.Amorecarefulanalysisoftheargumentrevealsacomplexreality.Ontheonehand,lowskilledindustrialjobsdoprovideaformal,steadysourceofincome,soitispossiblethattheyraiseincomesandreducepoverty.Yet,ontheotherhand,thesejobstendtobeunpleasantandverypoorlypaidopportunitiesevenbythestandardsoflowincomecountries.So,whatistheimpactofthesejobsonthewelfareoftheworkersdoingthem?Toanswerthisquestion,BlattmanandDercon(2016)46ranapolicyexperimentinEthiopia.Theywereabletoconvincefivefactoriestohirepeopleatrandomfromagroupofconsentingparticipants,andthentrackedtheeffectsontheirincomesandhealth.Theyfindthattheselow-skillindustrialjobspaidmorethanthealternativesavailabletoasubstantialfractionofworkers;butatthesametime,theyhadadversehealtheffectsanddidnotofferalong-termsolution—mostapplicantsquittheformalsectorquickly,findingindustrialjobsunpleasantandrisky.(Youcanreadmoreaboutthisstudyandtheauthors’interpretationoftheresultsinthispressreleasefromvox.com).Thisevidenceispartial,sinceitdoesnotaccountfor‘generalequilibriumeffects’—thatis,thepotentiallypositivelong-termeffectsthatnewmanufacturingjobshaveviamorecompetitionandhighersalariesinothersectorsoftheeconomy.Butitdoessuggestthatwhilelow-skilledindustrialjobsmayimproveconsumptionopportunities,providingashort-termsafetynet,theymaydosoatimportantcostsinotherdimensionsofwell-being.Thisreaffirmstheimportanceofmeasuringpovertybeyondjustincomeandconsumption,andofmaintaininganuancedunderstandingofhowgloballivingconditionschange.Cross-countrycorrelatesPovertycorrelateswithhealthoutcomesCountrieswheremorepeopleliveinextremepovertytendtohaveparticularlybadhealthoutcomes.Thefollowingvisualizationprovidesevidenceofthisrelationship.Itshowslifeexpectancyatbirthontheverticalaxis,againstpovertyrates(forapovertylineequivalentto3.20int.-$perday)onthehorizontalaxis.Thebuttonatthebottomallowsyoutochangethereferenceyears,sothatyoucanseehowthesetwovariablescovaryacrosstime.Aswecansee,thereisaclearnegativerelationship:peopletendtolivelongerincountrieswherepovertyislesscommon.Yetthecorrelationisfarfromperfect—somecountriessuchasSouthAfricahavearelativelylowlifeexpectancyincomparisontoothercountrieswithsimilarpovertyrates.Thisreinforcestheimportanceofthinkingaboutdeprivationbeyondincomeandconsumption. Clicktoopeninteractiveversion PovertycorrelateswitheducationoutcomesAboveweshowedthatpovertycorrelateswithhealth.Here,weprovideevidenceofanotherimportantcorrelate:education.Thefollowingvisualizationplotsmeanyearsofschoolingagainstpovertyrates(againusingapovertylineequivalentto3.20int.-$perday).Asbefore,thebuttonatthebottomallowsyoutochangethereferenceyears,sothatyoucanseehowthesetwovariablescovaryacrosstime.Aswecansee,thereisonceagainaclearnegativerelationship:povertytendstobemorefrequentincountrieswhereeducationislessdeveloped.Aswediscussedabove,thereisalsohousehold-levelevidenceofthiscorrelation—schoolingisoneofthestrongestpredictorsofeconomicwell-being,evenaftercontrollingforotherhouseholdcharacteristics. Clicktoopeninteractiveversion MeasurementandDataQualityInthissectionWhatarethemainindicatorsusedtomeasurepoverty?Thedifferencebetween‘absolute’and‘relative’povertyHowdoresearchersreconstructhistorical povertyestimates?HowdoestheWorldBankestimateextremepoverty?WhatarethemainlimitationsofWorldBankpovertyestimates?Howproblematicaredatalimitations?Whatalternativesaretheretoestimatemonetarypoverty?Whatisthecostofendingextremepoverty?Whatarethemainindicatorsusedtomeasurepoverty?The‘povertyheadcountratio’Themoststraightforwardwaytomeasurepovertyistosetapovertylineandcountthenumberofpeoplelivingwithincomesorconsumptionlevelsbelowthatpovertylineanddividethenumberofpoorpeoplebytheentirepopulation.Thisisthepovertyheadcountratio.Measuringpovertythroughtheheadcountratioprovidesinformationthatisstraightforwardtointerpret;ittellsustheshareofthepopulationlivingwithconsumption(orincomes)belowthepovertylineare.Butmeasuringpovertythroughheadcountratiosfailstocapturetheintensityofpoverty–individualswithconsumptionlevelsmarginallybelowthepovertylinearecountedasbeingpoorjustasindividualswithconsumptionlevelsmuchfurtherbelowthepovertyline.Thepovertygapindexisanalternativewayofmeasuringpovertythatconsiderstheintensityofdeprivation.The‘povertygapindex’Thepovertygapindexisameasureofpovertythatreflectsboththedepthandprevalenceofpoverty.Itisdefinedasthemeanshortfallofthetotalpopulationfromthepovertyline(countingthenon-poorashavingzeroshortfall),expressedasapercentageofthepovertyline. Itisworthunpackingthatdefinitionalittle.Forthosebelowthepovertyline,theshortfallcorrespondstotheamountofmoneyrequiredinordertoreachthepovertyline.Forthoseatorabovethepovertyline,theshortfalliscountedaszero.Theaverageshortfalliscalculatedacrossthetotalpopulation–bothpoorandnon-poor–andthenexpressedasashareofthepovertyline.Unlikethemorecommonly-usedmetricoftheheadcountratio,thepovertygapindexisthussensitivenotonlytowhetheraperson’sincomefallsbelowthepovertylineornot,butalsobyhowmuch–i.e.tothedepthofpovertytheyexperience.Thedifferencebetween‘absolute’and‘relative’povertyAbsolutepovertyismeasuredrelativetoafixedstandardofliving;thatis,anincomethresholdthatisconstantacrosstime.Absolutepovertymeasuresareoftenusedtocomparepovertybetweencountriesandthentheyarenotjustheldconstantovertime,butalsoacrosscountries.TheInternationalPovertyLineisthebestknownpovertylineformeasuringabsolutepovertyglobally.Somecountriesalsouseabsolutepovertymeasuresonanationallevel.Thesemeasuresareanchoredsothatcomparisonsrelativetoaminimumconsumptionorincomelevelovertimearepossible.RelativePoverty,ontheotherhand,ismeasuredrelativetolivingstandardsinaparticularsociety,andvariesbothacrosstimeandbetweensocieties.Theideabehindmeasuringpovertyinrelativetermsisthatthedegreeofdeprivationdependsontherelevantreferencegroup;hence,peoplearetypicallyconsideredpoorbythisstandardiftheyhavelessincomeandopportunitiesthanotherindividualslivinginthesamesociety.Inmostcases,relativepovertyismeasuredwithrespecttoapovertylinethatisdefinedrelativetothemedianincomeinthecorrespondingcountry.Thispovertylinedefinespeopleaspooriftheirincomeisbelowacertainfractionoftheincomeofthepersoninthemiddleoftheincomedistribution.Becauseofthis,relativepovertycanbeconsideredametricofinequality—itmeasuresthedistancebetweenthoseinthemiddleandthoseatthebottomoftheincomedistribution.Relativepovertycanbemeasuredusingthepovertyheadcountratioandthepovertygapindex.Indeed,theseindicatorsarecommoninEurope.47However,itisimportanttobearinmindthatthesearenotcomparabletotheestimatespublishedbytheWorldBank—thenatureoftheInternationalPovertyLineisdifferent.Howdoresearchersreconstructhistorical povertyestimates?Historicalestimatesofpovertycomefromacademicstudiesthatreconstructpastincomeandconsumptionlevelsbyestimatingeconomicoutputandinequalityforthetimebeforehouseholdsurveysbecameavailable.Aseminalpaperfollowingthisapproachandestimatingglobalpovertyfiguresfrom1820onwardisBourguignonandMorrison(2002).48Theirworkisthesourceofthepovertyestimatesforthetime1820to1970shownabove.BourguignonandMorrison’sstartingpointistoestimate theglobaldistributionofincomesovertime.Thechangeinextremepovertyisthencalculatedviachangesintheshareoftheworldpopulationwithincomesbelowthepovertyline,accordingtothecorrespondingestimateddistributionofincomes.BourguignonandMorrison(2002)relyonthreetypesofdatainordertoestimatethedistributionsofincome:economicoutput(realGDPpercapita),population,andinequality.Thefirsttwosourcesprovideinformationregarding‘thesizeofthepie’,whilethethirdoneprovidesevidenceregardingthedistributionofthatpie.Theapproachoutlinedaboveleadstoanaturalquestion:Howcanresearchersconstructeconomicoutputforthedistantpast?FouquetandBroadberry(2015)49provideadetailedaccountofhoweconomichistoriansconstructtheseestimates.Itispainstaking workwithwhichresearchersoccupythemselvesforyears.Thegenerallypreferredapproachtoestimatingnationalincomeistheoutputapproach,whichreliesonhistoricalrecordsbyeconomicsector.Forexample,foragriculturalproduction,researchersusechurchrecordsfortheestatesoffarmers,aswellasaccountingdocumentsproducedbyfarmersandkeptinlocalrecordoffices.Agriculturaloutputsarethencalculatedbymultiplyingtheacreageforeachcropbytheyieldperacre.Oncethisisestablished,pricesforindividualcropsandanimalproductsareusedtoconverttheoutputintocurrentpricesandcreateweightsforan‘agriculturalrealoutputindex’.Outputsrelatedtoothersectors,suchasleatherandfoodprocessing,areestimatedusingasimilarapproachappliedtothespecificsofeachsector.Finally,whentheoutputofallsectorsisreconstructed,thesevariousseriesarebroughttogetherand—usingasetofsectoralweightsthatcapturethechangingstructureoftheeconomy—anestimateofthetotalhistoricoutputoftheproductiveworkofthepopulationisreached.HowdoestheWorldBankestimateextremepoverty?TheWorldBankisthemostimportantinstitutionmeasuringtheextentofglobalpovertyforthetimesince1981.TheWorldBankestimatesofpovertyarepublishedviaPovcalNetandalsointheWorldDevelopmentIndicators.TheWorldBankestimatesareproducedfromthreekeyingredients:householdsurveysprovidingevidenceabouthouseholdconsumptionperhead(or,insomecasesaswewillsee,incomeperhead);domesticpriceindexesandpurchasingpowerparityrates;andanInternationalPovertyLinebasedonnationallinesinthepoorestcountriesforwhichsuchlinesareavailable.Belowweprovideanoverviewofeachoftheseingredients.Ferreiraetal.(2016)50providefurtherdetails.MeasuringconsumptionfromhouseholdsurveysConsumptionpercapitaisthepreferredwelfareindicatorfortheWorldBank’sanalysisofglobalpoverty.Consumptionisdefinedas“theuseofresourceswhetheracquiredthroughpurchase(expenditure)orthroughhouseholdproductionorprovidedfromoutsidethehousehold,suchasbyrelatives,charities,orthegovernment”.51Inprinciple,onecouldusehouseholdsurveystoestimate(i)resourceoutflows(monetaryexpenditures,homeproductionandtransfers);(ii)resourceinflows(earningsandothernon-marketsourcesofincomesuchas,again,homeproductionandtransfers);and(iii)changeinassetsbetweenthebeginningandendoftherelevantperiod(includingsavings,owneddurablegoods,etc.).Givenallthisinformation,consumption,asperthedefinitionabove,couldbeestimateddirectlyfrom(i),orasthedifferencebetween(ii)and(iii).Intheory,bothapproachesshouldgivethesameresult.Inpractice,however,surveysonexpendituresaredifferentfromsurveysonincomes(moreonthisbelow).Forthemajorityofcountries,theWorldBankestimatesconsumptiondirectlyfromhouseholdsurveysonexpenditures.Forasignificantminorityofcountries,however,WorldBankestimatesarebasedonincomesurveys.Notably,inbothcases,theestimationmethodologydoesinclude homeproductionandtransfers,byattachingmonetaryvaluestosuchnon-markettransactions.Howaremonetaryvaluesplacedonthingslikefoodgrownathomeandgiftsfromrelatives?Onecommonapproachistoasksurvey-respondentsabouttheamountofsuchresourcesconsumedoveragivenreferenceperiod.Theaimistothenascribeamonetaryvaluetothereportedconsumption. Thisisdoneby multiplyingtheconsumed amountsbyextrapolatedmarketprices.A secondapproachaskshouseholdsdirectlyabouttheirownvaluationoftheamountofmoneytheywouldexpecttopayiftheyhadboughtsuchitemsthemselves,or,theamountofmoneytheywouldexpecttoreceiveiftheyhadsoldtheseitems.Thesecondapproachiscommonlyusedtoestablisharentalequivalentforhousinganddurablegoodsownedbythehousehold.Howareincomeandexpendituresurveysactuallyconducted?Differentcountriesusedifferentsurveyinginstruments,andwhilethereismuchscopeforharmonization(seeBeegleetal201252),therearesomebasiccommonfeaturesthatallowforcross-countrycomparisons.Inmostcases,surveysarerepresentativeatthenationallevelandrecordresponsesprovidedby‘primaryrespondents’suchastheheadofthehousehold.Respondentsreportexpenditures(orincomes)eitherbyansweringquestionsfrommemory(the‘retrospectiverecallmethod’)orbyrelyingonwrittenrecords(the‘diarymethod’).Inthecaseofexpenditures,differentreferenceperiodsareusedtorecordresponsesacrossdifferentcategoriesofgoods,withlongerperiodsforgoodsorservicesthattendtobeacquiredlessfrequently.AdjustingfordifferentpricelevelsindifferentcountriesIncomeandconsumptionmeasuresavailablefromnationalhouseholdsurveysaredenominatedinlocalcurrencyunits.Thismeansthatinordertomakemeaningfulcross-countrypovertycomparisons,itisnecessarytotranslatefiguresintoacommoncurrency—i.e.useaconsistent‘unitofmeasure’.Onepossibilitywouldbetosimplyusetheexchangeratesfromcurrencymarketstotranslateallnationalfiguresintoonecommoncurrency—suchas,forexample,theUS-dollar.Thisapproach,however,wouldfailtoaccountfordifferencesinpricelevels:oneUSdollarallowsyoutoachievehigherconsumptioninIndiathanitdoesintheUS.Ifweareinterestedinmaterialdeprivation,anymonetaryincomeshouldbeconsideredinrelationtotheamountofgoodsandservicesthatitcanbuylocally.Forthisreason,theWorldBank’sinternationalpovertycomparisonsdonotrelyonmarketexchangeratesbutonexchangeratesthatarebasedonpurchasepowerparityconversionfactors.AngusDeatonexplainsitasfollows:“Purchasingpowerparityexchangerates,orPPPs,arepriceindexesthatsummarizepricesineachcountryrelativetoanumerairecountry,typicallytheUnitedStates.Thesenumbersareusedtocomparelivingstandardsacrosscountries,byacademicsinstudiesofeconomicgrowth,particularlythroughthePennWorldTable,bytheWorldBanktoconstructmeasuresofglobalpoverty,bytheEuropeanUniontoredistributeresources,andbytheinternationaldevelopmentcommunitytodrawattentiontodiscrepanciesbetweenrichandpoorcountries.”PPPexchangeratesallowustotranslatemonetaryincomes(orconsumption)fromlocalcurrencyunitsinto‘internationaldollars’.Theideaisthatagivenamountofinternationaldollarsshouldbuyroughlythesameamount(andquality)ofgoodsandservicesinanycountry.AsthegraphshowsforGDPpercapita,assessinglivingstandardsusingPPPadjustedinternationaldollarsratherthanUSmarketdollarscanmakeahugedifference.WhenpricelevelsinacountryaremuchlowerthanintheUS,usingUSdollarsatmarketexchangerateswillsignificantlyunderestimatethevalueofincomes.PPPfactorsareestimatedbytheInternationalComparisonProgramme(ICP).ThetwolastroundsofPPPfactorsestimatedbytheICParefrom2005and2011andthenextoneisscheduledfor2017.YoucanreadmoreaboutPPPadjustmentsinourdedicatedblogentryonthistopic. Clicktoopeninteractiveversion SettingtheInternationalPovertyLineToday,theInternationalPovertyLineis1.90internationaldollars(at2011PPPprices).Wheredoesthisnumbercomefrom?ThepioneeringworkthatsetouttocountthenumberofpeopleinpovertyusingacommonglobalstandardwaspublishedbyMontekAhluwalia,NicholasCarter,andHollisCheneryin1979.53ThethreeauthorsbasedtheirestimatesofglobalpovertyfiguresontheIndianpovertylineatthetime.Torelyonthenationalpovertylineofalow-incomecountryisstillthebasicideaonwhichtheInternationalPovertyLineisbased.ButtodayitisnotjustthepovertylineofIndiathatistakenintoaccountrather,aswecanseeinthefollowingtable,itisbasedonthenationalpovertylinesof15differentlow-incomecountries.Therewereseveralmajorrevisionsbetweenthefirstformulationofaglobalpovertylinein1978andtoday.Thetableshownhere,takenfromFerreiraetal.(2016),54showsindetailhowtheInternationalPovertyLinehasbeenupdatedovertime.Aglobal‘dollar-a-day’povertylinewasintroducedintheWorldDevelopmentReportin1990,andwassubsequentlyusedfortheformulationoftheMillenniumDevelopmentGoalsandthecommitmentto“halve,between1990and2015,theproportionofpeoplewhoseincomeislessthan$1aday.”TheInternationalPovertyLineisintendedtobeaglobalpovertylineforabsolutemeasurementofdeprivation—soitisnotrecurrentlyadjustedaslow-incomecountriesgrowricher.However,itisimportanttobearinmindthattheInternationalPovertyLineissometimesupdated;in2015,forexample,thelinewasupdatedfrom1.25internationaldollars(at2005PPPprices),to1.90internationaldollars(at2011PPPprices).Thislastupdatewasmadeinordertoincorporatenewevidenceonrelativepricelevels,ratherthantochangetheunderlyingrealwelfarestandardusedtodefinedeprivation.Inthelastcoupleofrevisions,theguidingprinciplehasbeentoincorporatenewdataonrelativepriceswhileattemptingtominimizechangestotherealwelfarevalueoftheline,soastoavoid“shiftingthegoalposts”.ThecurrentmethodologyforchoosingthesetofcountriesusedtodefinetheInternationalPovertyLinewasfirstproposedbyChenandRavallion(2001).55Theycompiledasetofnationalpovertylinesforlow-incomecountries—drawingfromtheWorldBank’scountry-specificPovertyAssessmentsandfromthePovertyReductionStrategyPaperscompiledbythegovernmentsofthecountriesconcerned—andfoundthatwhiletherewasapositiveassociationbetweenthevalueofthepovertylineandnationalpercapitaconsumptionforthelargemajorityofcountries,therelationshipwasflatforthefifteenpoorestcountriesintheirsample.Inotherwords:theyfoundthatthepoorestfifteencountriesintheirsampleusedaroughlysimilarabsolutepovertyline,independentofdifferencesintheirpercapitaconsumptionlevels.Thesearethefifteencountriesthatwerechosenasreference.Thecurrentmethodologyhasbeencriticizedbecauseoflackofcomparabilityintheunderlyingsetofnationalpovertylinesthatwereusedtochoosethefifteenreferencecountries.JolliffeandPrydz(2016)56addressthisissueofcomparabilitybyestimatingthenationalpovertylinesthatareimpliedbypovertyrates.57ThesetofnationalpovertylinesestimatedbyJolliffeandPrydzsuggests,incontrasttoearlierfindingsbyChenandRavallion,thatthereissubstantialvariationinpovertylinesevenamongthepoorestcountries.However,thisvariationdoesnotseemtocontradictthechoiceoftheInternationalPovertyLine:ifweorderthepovertylinesofthepoorest25percentofcountries,thevalueinthemiddleiscloseto1.90int.-$.ThisisinlinewithabroaderpointmadebyFerreiraetal2016:alternativeapproachesthatwereproposedforupdatingtheInternationalPovertyLineto2011PPPsendupgeneratinglinesthatareeitherexactlyorverycloseto1.90int.-$aday.EvolutionoftheWorldBank’sInternationalPovertyLine–Ferreiraetal.(2016)58  1979“Indialine”1990“Dollar-a-day”2001"1.08/day"2008"1.25/day"2015"1.90/day" SourceAhluwaliaetal.(1979)WorldBank(1990)ChenandRavallion(2000)RCS(2009)Ferreiraetal.(2016) Relativepricelevels(ICPdata)1975PPPs1985PPPs1993PPPs2005PPPs2011PPPs Nationalpovertylinesused(#)1(India)8countries10countries15countries15countries(same linesas2008) MethodIndia'spovertyline(46thpercentile)Inspection(rounded)MedianMeanMean(rounded) Povertyline(ICPbase yearUSD)$0.56$1.01($1.00)$1.08$1.25$1.88(1.90) Povertyline(Constant 1985USD)$1.12$1.01$0.8$0.69$0.91 Nationalpovertylinesused(countries)IndiaBangladesh, Indonesia, Kenya, Morocco, Nepal, Pakistan, Phillippines, Tanzania,Bangladesh, China,India, Indonesia, Tanzania, Thailand, Tunisia,Nepal, Pakistan, ZambiaChad, Ethiopia, TheGambia, Ghana, Guinea-Bissau, Malawi,Mali, Mozambique, Nepal,Niger, Rwanda, SierraLeone, Tajikistan, Tanzania, UgandaSameas2008 HowdoestheInternationalPovertyLinecomparetonationalpovertylines?ThefollowingvisualizationshowshownationalpovertylinesindifferentcountriescomparetotheInternationalPovertyLine.ThefigurescomefromJolliffeandPrydz(2016),59andcorrespondtothepovertylinesthatareimpliedbynationalpovertyhead-counts.60Ascanbeseen,thereisacleargradient:poorercountriestendtouselowerpovertylines.Importantly,thischartalsoshowsusthatalthoughtheInternationalPovertyLineisverylow,itisstillhigherthantheofficialpovertylinesusedbymanylow-incomecountries.InMalawi,forexample,thenationalpovertylineis1.27int.-$perday.Forreference,inthischartwehaveincludedalsoOECDrelativepovertylines.Itshouldbenotedthat,bydefinition,thesepovertylineschangeovertimesincetheyaredefinedrelativetothemedianincome.Theyarehoweverincludedtogiveanideaofthedegreeofvariationinstandardsusedbycountriestomeasurepoverty.WhileinMalawithenationalpovertylineisequivalentto1.27int.-$perday,inNorwayitisequivalentto35.10int.-$perday.ApersondefinedaspoorinNorwaycanbe27-timesricherthanapersondefinedaspoorinMalawi,acountryinwhichGDPpercapitais58-timeslowerthaninNorway.Mostlow-incomecountriesdefinetheirnationalpovertylinesbyrelyingonaversionofthe‘costofbasicneeds’method.Thisapproachfirststipulatesaconsumptionbundlethatisdeemedadequateforbasicconsumptionneedsinthelocalcontext,andthenestimatesthecostofthisspecificbundle.Whatisanadequateconsumptionbundle?Onecommonstartingpointistorelyonagenericfoodrequirement,suchas2,100caloriesperpersonperday,andthenincludeanonfoodcomponentthatisaddedtoreflectcostsforhousing,clothing,electricity,andsoon.Anotherapproach—lesscommonbutalsoemployedinpractice—istosetabsolutelinesbasedonaskingpeoplewhatminimumconsumptionorincomeleveltheyneedjusttomakeendsmeet.Above,weshowthatthereisindeedacloserelationshipbetweentheself-assessmentoflivingconditionsandthemeanincomeinthatsociety,bothbetweenandwithincountries.YoucanreadmoreaboutnationalpovertylinesintheWorldBank’sresearchreport,“AMeasuredApproachtoEndingPovertyandBoostingSharedProsperity:Concepts,Data,andtheTwinGoals”,pages37and38. Clicktoopeninteractiveversion WhatarethemainlimitationsofWorldBankpovertyestimates?Above,wediscussedthemethodologyusedbytheWorldBanktomeasureextremepoverty.Here,wefocusonthevariouslimitationsofthismethodology.WefollowthepointsdiscussedbyFerreiraetal.(2016).61DatadeprivationOneofthekeyingredientsfortheWorldBank’smeasurementofpovertyarehouseholdsurveysprovidingevidenceabouthouseholdconsumptionperhead(or,insomecases,incomeperhead).ThefollowingmapshowstheavailabilityofsurveysusedtoconstructtheWorldBank’spovertyestimates.Forallcountriesshowningreyinthismap,thereisnotasinglesurveyavailabletotheWorldBankinthelastthreedecades.Manyofthesecountriesarerichcountriesinwhichextremepovertyisverylow.Butthereisalsomissingdataforsomepoorercountries,inwhichsurelyaconsiderableshareofthepopulationislivinginextremepoverty.Aswecanalsoseefromthismap,therearesomecountrieswithveryfewobservations.ThisisthecaseformanyAfricancountries,wherethereisonlyonesurveyavailableinthelastdecade.Thisisextremelyinfrequent,evenincomparisontoLatinAmericaandCentralAsia,wheremanycountrieshavealmostannualsurveys.Bymovingthetimesliderbelowthemap,youcanseehowmanysurveysareavailableineachdecade.Forindividualcountries,theWorldBankpublishespovertyestimatesonlyforyearsinwhichhouseholdsurveydataisavailable.Butforregionalandglobalestimates,theWorldBankpublishesestimateseverythreeyears.Clearly,sincenotallcountrieshavesurveydataforallyearsinwhichregionalandglobalestimatesareproduced,theWorldBankmustrelyonapproximations.Intheirownwords,they‘lineup’countryestimatestoareferenceyearinordertoproduceregionalandglobaltotals.Theprocessofliningupestimatesreliesoninterpolationforcountriesinwhichsurveydataarenotavailableinparticularyears,butareavailableeitherbeforeorafter(orboth).YoucanreadmoreaboutthisprocessinPovcalNet.Thebottomlineisthattheaccuracyoftheseapproximationsreliesheavilyontheavailabilityofsurveydata—themoresurveyyearsareavailableforacountry,themoreaccuratetheapproximation.62Forlowandmiddleincomecountrieswithoutreliablesurveydatainanyyear(mainlycountriesintheMiddleEastandNorthAfrica),theWorldBankchoosesnottopublishcountry-specificestimates,butstillincludesanapproximatednumberintheregionalandglobaltotals,byrelyingonalternativestatisticaltechniques.63 Clicktoopeninteractiveversion ComparabilityofincomeandconsumptiondataConsumptionpercapitaisthepreferredwelfareindicatorfortheWorldBank’sanalysisofextremepoverty.ButnotallnationalstatisticalagenciesreportconsistentestimatesofconsumptionbasedonexpendituresurveystotheWorldBank.Theissueisthatnationalstatisticalagenciesdesignandexecutesurveystoservethespecificneedsandinterestsofeachparticularcountry;whichmeansthatdifferentcountriesusedifferentconcepts,methods,andquestionnairedesigns.Ferreiraetal.(2016),64showthatapproximately75%ofthecountriesinthePovcalNetdatabasehavedataonpercapitaconsumption,whiletheremainingcountries—mostlyinLatinAmericaandtheCaribbean—haveincomepercapita.Incomeandconsumption,asmeasuredbyhouseholdsurveys,arenotusuallyperfectlycomparable.Despiteeffortstobroadenthedefinitionofincomesforthepurposeofmeasuringpoverty,inmanycountriesstatisticalagenciesstillusedefinitionsthatfailtoaccountfortheconsumptionthatoccursoutof‘non-income’resourcessuchassavingsandassets,borrowing,andsomeformsofgovernmentwelfarebenefits.Theimplicationisthat,bydefinition,zeroincomeisafeasiblevalue,whilezeroconsumptionisnotafeasiblevalue—peoplewithzeroconsumptionwouldstarve.Asonewouldthenexpect,thisisreflectedinthedata.“Thereisessentiallynomasspointinanycountrywithzeroconsumption,butmanycountriesthatuseincomedatahaveasignificantmassofzeroincomesinthedata,allofwhicharetreatedasbeingpoor.LatinAmerica,inparticular,predominantlyusesincometomeasurepoverty,andinmanycountriesthereareatleastafewpercentoftheobservationsthatarezero”.65Inrichercountries,where‘non-income’resourcessuchassavings,borrowing,andgovernmentwelfarebenefitsarecommon,thisissuesofcomparabilitycanbesubstantial.Indeed,inrichcountriessuchastheUS,theproblemofcomparabilityissosubstantialthattheWorldBankdecidesnottoincludeestimatesofitsextremepovertyintheglobaltotals.Thisisapointwediscussbelowinmoredetail.ThefollowingchartsfromChandyandSmith(2014)66showhowincomeandconsumptionestimatesdifferfortheUSandforMalawi.Inthesecharts,eachdotrepresentsahousehold.Morespecifically,foreachhousehold,thechartcomparesincomeperday(horizontalaxis)againstexpendituresperday(verticalaxis).Bothmeasuresforeachhouseholdcomefromthesamesurvey.Ifincomesandexpendituresareidentical,thenweshouldseeallhouseholdslinedupalongthe45-degreeblueline(whereexpenditureandincomeareequal).Theorangeandredlinesplotthetrendthatbestfitsthedata(i.e.thelinesthatbestdescribesthecross-householdrelationshipbetweenincomeandexpenditure).Aswecanseen,intheUSthebest-fitlineissignificantlydifferenttotheblueline:atverylowlevelsofincome,expenditureissignificantlyhigherthanincome;andathighlevelsofincome,expenditureislowerthanincome.Incontrast,inMalawithebest-fitlineisclosetotheblueline:householdswithverylowincomehavecomparablylowexpenditure.ThisisbecauseunlikeintheUS,householdsinMalawicanrelylessonsavings,borrowing,andgovernmentwelfare.Post-taxincomeandconsumptionperdayacrosshouseholds,USandMalawi–ChandyandSmith(2014)67ComparabilityofsurveyquestionnairesandsamplingAbove,wepointedoutthatWorldBankpovertyestimatesforsomecountriescomefromincomedata,whileinothercountriestheycomefromconsumptiondata.Asitturnsout,comparabilityissuesariseevenamongcountriesthatrelyonconsumptiondatafromexpenditures,sincesurveyquestionnairesarenotalwaysstandardized.Beegleetal.(2012)68giveusaconcreteideaoftheextenttowhichsurveyquestionnairesmatter.TheyconductanexperimentinTanzania,inwhichtheyrandomlychoosehouseholdsandtesteightalternativemethodsofmeasuringhouseholdconsumption.Theyfindsignificantdifferencesbetweenconsumptionreportedbythebenchmarkpersonaldiaryandotherdiaryandrecallformats.Thetablesummarizesthedifferencesinmeasuredpovertythatarisefromthevarioussurveydesigns(youcanseeanoverviewofthedifferencesbetweenquestionnaireshere,withmoredetailsinthepaper).Aswecansee,thedifferencesaresubstantialforboththeheadcountratioandthepovertygapindex.Theseresultsareconsistentwithotherstudies.Researchershavefoundthattherecallperiodforfoodconsumptionmattersfortheassessmentoffoodconsumptioninapopulation(foranexampleonIndiaseeDeatonandKozel2005);69andithasalsobeenobservedthatquestionnaireswithmorefooditemslistedreporthigherfoodconsumptionwhencomparedwithaquestionnairewithfeweritems(foranexampleonElSalvadorseeJolliffe2001).70Studiesalsosuggestthatsurveydesignmattersforsamplingpurposes.ThereistheoreticalandempiricalevidencepresentedbyKorineketal.(2005)71showingthataspeoplebecomericher,theyarelesslikelytorespondtosurveys.Whenricherindividualsorhouseholdsarelesslikelytoanswersurveysthanpoorpeople,survey-basedestimatesofconsumptionandincomewillunderstatethemeanlevelofprosperityandoverstatetheshareofpeopleinpoverty.Povertystatisticsasmeasuredfromconsumption,bysurveydesign–ExperimentalresultsfromTanzania,fromBeegleetal.(2012)72Howproblematicaredatalimitations?TheabovediscussionofdatalimitationsinthecontextofWorldBankpovertyestimateshighlightsanimportantfact:anyestimateofpoverty—ofeitheritslevelorchangeovertime—issurroundedbyamarginoferror.KeepinginmindthattheWorldBankpovertyestimatesareonlyapproximationsisimportantwhenmakingpolicydecisions,suchastheallocationofinternationalaid.Indeed,reviewingthemanyproblemsofpovertymeasurement,Deaton(2010)73concludesthat“probablythemosturgentareaforthepovertycountsisnotthe[InternationalComparisonProgram],buttheimprovementintheconsistencyandtimelinessofhouseholdsurveys,andtheupgradingofnationalaccounts.”ThefactthattheWorldBankpovertyestimatesareonlyimperfectapproximationsdoesnotmeanthatthesearemeaninglessoruselessnumbers—itmeansthattheyshouldbeusedasonemoresourceofinformationtoassesslivingstandards.Aswepointoutabove,therearemanyothercomplementarywaysofmeasuringdeprivation.Byvirtueofbeingapproximations,theWorldBankpovertyestimatescanunderestimate,aswellasoverestimatethesizeoftheunderlyingproblem.It’seasytoforgetaboutthisandthinkthatinaccuratepovertyestimatesmustnecessarilyunderestimatetruepovertyfigures.Asamatteroffact,thereissomeevidencesuggestingthatWorldBankfiguresmightbeover-estimates.ThefollowingvisualizationfromtheWorldBank’AfricaPovertyReport(2016)plotsregionaltrendsfordifferentsubsetsofcountries.ThegreenseriesplotsallcountriesinthePovcalNetdataset—thisisthebenchmark.Theotherlinesexcludedifferentcountries,dependingonwhethertheyhavecomparableandgood-qualitydata.Theseriesfor“comparableandgood-qualitysurveysonly”,forexample,excludessomeofthesurveysfromBurkinaFaso,Mozambique,Tanzania,andZambia;andreplacesthepovertyestimatesofthetwocomparablebutpoorer-qualitysurveysofNigeria(NigeriaLivingStandardsSurveys2003/04and2009/10)withtheestimatefromtheGeneralHouseholdSurveyPanel2010/11,whichhasbeendeemedofgoodquality.Aswecanseen,the2012estimateofpovertyinAfricaincludingonly“comparableandgood-qualitysurveys”is6percentagepointslowerthanthePovcalNetestimate(37percentinsteadof43percent).AtOurWorldInDataweareparticularlyinterestedinhowlivingconditionschangeoverthelongrun.Theavailabledatasuggeststhatthedeclineofpovertyhasbeensolargeoverthelongrun,thatitcannotbetheresultofmeasurementerror.Evenifweconsiderrealisticconfidenceintervals,thetrendshold.Sodespitelimitations,theavailablepovertyestimatesarestillveryinformativeabouthowtheworldhaschanged.ShareofpopulationlivingbelowtheInternationalPovertyLineinAfrica,bysampleofsurveys,2012–WorldBank(2016)74Whatalternativesaretheretoestimatemonetarypoverty?ThepovertyfigurespublishedbytheWorldBankarebasedon‘microeconomicdata’—specifically,householdsurveys.Asecondwayofmeasuringpovertyistostartfrom‘macroeconomicdata’,particularlytheNationalAccounts,whichreporttheaggregatetotaleconomicactivityofanationincludingtheGDP.ThissecondapproachwasfirstexploredbyAhluwaliaetal.(1979),75andaswepointoutabove,historicalestimatesofpoverty,suchasthosefromBourguignonandMorrison(2002),alsofollowthisapproach.Intheory,measuresofpovertyderivedfromthe NationalAccountsshouldmatchmeasuresofpovertyderivedfromhouseholdsurveys,butinrealitythereareoftensubstantialdiscrepancies.Deaton(2005)reviewsthereasonsforthesediscrepancies.76Giventhatbothapproachesaresubjecttomeasurementerror,itisnaturaltowonderwhichofthetwomethodsissuperior.DoNationalAccountsorhouseholdsurveysgiveusabetterunderstandingofthelevelsandchangesofpoverty?EconomistsinrecentyearshavestartedtakingthisquestionseriouslyandaskedhowmuchweightshouldwegivetoNationalAccountestimatesvis-à-vishousehold-surveyestimates?PinkovskivandSala-i-Martin(2016)trytoanswerthisquestion.77Theybeginbynotingthateconomicactivityaroundtheworldhasbeenshowntocorrelatewithsatellite-recordeddataonnighttimelightsfromthesurfaceoftheEarth,whicharevisiblefromspace.Wediscussthisinmoredetailinourentryonlightatnightanditisshowninthetwosatelliteimages.Itshowshoweconomicactivity,ofwhichnighttimelightsareaproxymeasure,changedinSouthAsiabetweentheyears1994and2010.Wecanseethatnightlightsin2010coverareasthatwereunlitin1994;andthereisalsoasubstantialincreaseintheintensityoflightsinmajorcitiesoverthesameperiod.ThisisindicativeoftheunderlyinggrowthineconomicactivitythatSouthAsiaachievedduringthisperiod.Basedonthiscorrelationbetweennighttimelightsandeconomicactivity,PinkovskivandSala-i-MartinevaluatetherelativequalityofincomedataderivedfromhouseholdsurveysandNationalAccounts.Theintuitionfortheirapproachisthat,aslongasthemeasurementerrorinnighttimelightsisunrelatedtothemeasurementerrorsineitherNationalAccountsorsurveymeans,onecanusenightlightsasagaugetoseehowmuchweighttogiveNationalAccountsincomeestimatesrelativetohousehold-surveyincomeestimatesinmeasuringtrueincome.TheauthorsfindthatnighttimelightsaremuchmorecloselycorrelatedwithGDPpercapitathanwithsurveymeans.Thissuggests,undertheirassumptions,thattheoptimal‘weights’tocalculateaggregateincomeshouldbeverylargefornationalaccountsandverymodestforsurveymeans.Whatisthecostofendingextremepoverty?Afundamentalquestionthatcutsthroughthistopiciswhetherwehaveagoodsenseoftheactualcostofendingpoverty.Theanswertothatquestionisno.Thesizeofthepovertygapisonlyaroughestimate,andtheavailableempiricalevidencedoesnotprovideaclearideaofhow‘rough’is‘rough’.Anaccurateestimateofthecostofendingextremepovertyrequiresnotonlyhavinggooddataonpovertyratesfromacrosstheworld(includingincountrieswheretherearestillimportantgaps),butalsounderstandingtheinefficienciesthatredistributivetransferstypicallyentail.Whataretheseinefficiencies?First,transfersarehardtotarget(itishardtoreachthedesiredpopulation);andsecond,transfershaveknock-oneffectsoneconomicbehaviour(transferschangeincentivesandhencemayaffect,amongotherthings,baselineincomelevels).Undertheassumptionthattheseinefficienciesarenotgrowingovertime,thereductionofthetotalsizeofthepovertygapoverthelastfewdecadesisgoodnews.Butitisimportanttonotethatwedon’thaveenoughdatatoreallyunderstandhowstrongthisassumptionis.Inbrief:Itisimportanttohavemoreresearchontheinefficienciesthatarisefromredistributivetransfers. DataSourcesLong-termdevelopmentofglobalpovertyBourguignonandMorrisson(2002)Data:SeveralmeasuresofpovertyandinequalityGeographicalcoverage:Global–byworldregions/continentsTimespan:1820to1992Availableat:Theresearchpaperis:BourguignonandMorrisson(2002)–InequalityAmongWorldCitizens:1820–1992.InAmericanEconomicReview,92,4,727–744.Thesedatawereusedaboveinthegraphshowingthedecliningshareofpeoplelivinginpovertysince1820.EconomistsXavierSala-i-MartinandMaximPinkovskiyestimatedtheshareoftheworldpopulationlivinginabsolutepoverty.78AnimportantrecentpaperonabsolutepovertyisChenandRavallion(2010)–TheDevelopingWorldisPoorerthanWeThought,ButNoLessSuccessfulintheFightAgainstPoverty.InTheQuarterlyJournalofEconomics,125,4,1577–1625.DataonglobalpovertyinrecentdecadesWorldBankData:Severalmeasuresofabsolutepoverty.Geographicalcoverage:Global–bycountryandworldregion.Timespan:Since1980Availableat:WorldBank’sPovcalNet–aninteractivetoolwhichvisualizesabsolutepovertyandmakesthedataavailablefordownload.Dataonthesub-nationallevel(withhugecoverage!)isavailablefromtheWorldBank.Thesearedataonthepovertyheadcount–atnationalpovertyline,urbanpovertyline,andtheruralpovertyline. WordpressEditPageOurWorldinDataisfreeandaccessibleforeveryone.Helpusdothisworkbymakingadonation.Donatenow