CONTACT
Alain [email protected]://urbanizationproject.org
WORKI NG PAPER #38 / DECEMBER 1, 2016
AFFORDABILITY: HOUSEHOLDS’ INCOME, REGULATIONS, AND LAND SUPPLYPart I+ ALAIN BERTAUD
ABSTRACT
Governments try to formulate housing policies that will provide socially acceptable housing standards at an
affordable price for all. However, governments may exacerbate the high cost of housing in a city by limiting the
supply of housing through regulations and underinvesting in urban expansion. In their search for solutions,
urban managers often ignore that households’ housing choice is driven by a combination of three attributes:
floor area, location, and price of land and construction per square meter. Because floor area and construction
quality are the most visible of the three attributes, planners tend to concentrate on improving the design and
increasing the area of dwellings when drafting a housing policy. They tend to ignore housing location and its
corollary: access to city’s labor markets. When low housing standards are largely due to poverty, ignoring location
to provide larger homes might devastate the very population the policy is supposed to help.
This working paper was prepared as part of a forthcoming book about urban planning, tentatively titled “Order Without Design.”
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage1of21
Contents
CHAPTER6–AFFORDABILITY:HOUSEHOLDS’INCOME,REGULATIONS,ANDLANDSUPPLY 2
1. “Theneedtodosomething”:affordablehousing..........................................................................................2
2. Definingandmeasuringhousingaffordability..............................................................................................3Housingaffordabilityisdifferentfromtheaffordabilityofanyotherconsumerproduct.................................................3
Asimple“affordability”index:thePrice/Incomeratio...........................................................................................................................5WhatdoesitmeanforacitytohaveaPIRabove8?..........................................................................................................................6ThePIRisausefulindextoidentifyanaffordabilityproblembutitistoocrudetoidentifyapolicysolution.........9
HowdohouseholdsadjusttounaffordablePIR!.........................................................................................................................................9InformalsubdivisionofapartmentsinChina......................................................................................................................................10Youngpeoplelivingwiththeirparents..................................................................................................................................................10Reducing“minimumsociallyacceptablehousingstandardsMicroapartmentsinNewYork......................................11InformalsubdivisionsinNewYorkCity:creatingaffordablehousingbelowtheminimumsociallyacceptable
standard...................................................................................................................................................................................................................................13
Whenthepoorareunabletosubstitutecapitalforland.......................................................................................................................13
MeasuringIncomedistributioninrelationtohousingconsumptionisindispensableforpolicyformulation..................15UsingCities’householdsincomedistribution......................................................................................................................................15HousingStockandflow:thetrickledowntheory..............................................................................................................................16Whathappenwhenincomeincreaserapidly?....................................................................................................................................17Incomedistributionrelatedtohousingtypology..............................................................................................................................18
Relatingincomedistributionwithhousingconsumption....................................................................................................................20
ListofFiguresFigure1:Price/incomeratioofaselectionofworldcities(2015)..............................................................6Figure2:Price/incomeratioandaffordabilityin10selectedcities..........................................................7Figure3:PercentofrentinghouseholdsspentonrentsandPrice/IncomeRatioofowners...........8Figure4:PosterinasuburbofBeijingadvertisingaroomtorentinasubdividedapartment...10Figure5:Shareofyoungpeopleaged25to34livingwiththeirparentsinEuropeandintheUS
.................................................................................................................................................................................................................11Figure6:NewYorkCity"minimumacceptablehousingstandardssince1860..................................12Figure7:MumbaiNorthernsuburbs-Informalandformalsettlement.................................................14Figure8:Shanghaihouseholds'incomedistributionin1998.....................................................................15Figure9:Shanghaichangesinincomedistributionbetween1998and2003.....................................18Figure10:Hanoiincomedistributionrelatedtotypology...........................................................................19Figure11:Hanoi–Householdsincomedistributionandfloorconsumption.......................................21
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage2of21
CHAPTER6–Affordability:
Households’Income,Regulations,andLandSupply(firstpart)
1. “Theneedtodosomething”:affordablehousing“…amajorimpedimenttoamoreefficientspatialallocationoflaborarehousingsupplyconstraints.TheseconstraintslimitthenumberofUSworkerswhohaveaccesstothemostproductiveofAmericancities.Ingeneralequilibrium,thislowersincomeandwelfareofallUSworkers.”Chang-TaiHsiehandEnricoMoretti1.
Wehaveseenthatprosperouscitiesdependonwell-functioninglabormarkets.Hsiehand
Moretti,twoeconomists,foundthatthehighpriceofhousinginsomeotherwiseextremelysuccessfulUScitiesdistortsthespatialallocationoflabornationwide.Theycalculatethecostofthismisallocationtoabout9.4%oftheUSGDP.Housingaffordabilityisthereforenotatrivialissue.HsiehandMorettiarguethatregulatoryhousingsupplyconstraintscontributeheavilytothehighpriceofhousing,apositionwithwhichIconcurandsupportinthischapter.Someaffordabilityproblemsareduetopoverty,butinmostcasestheyarecreatedorexacerbatedbyman-madeconstraintsonthesupplyoflandandfloorspace.
Forlabormarketstowork,householdsandfirmsmustfindanaffordablespaceinwhichto
locate.Whenselectingthisaffordablespace,theymustmaketrade-offsbetweenrent,floorareaandlocation.Theirfinallocationchoicewillreflectthetrade-offthatmaximizestheirwelfare.Locationisofcourseextremelyimportant,asthelocationprovidesaccesstotherestofthecityanditslabormarket.Thewell-wornrealestatedevelopers’cliché“Location,location,location”reflectsarealityandawisdomthatmanygovernmenthousingaffordabilityexpertstendtoforget.Thefloorarea,locationandpricepersquaremeterofahousehold’shousingunitconstituteitscurrent“affordability”.Thiscurrentlyoccupied“affordablehousingunit”representsthehousehold’sbestpossiblechoiceamongallotherhousingchoicesofferedbythemarket.
However,eveninafreemarket,lowerincomehouseholds’optimalhousingchoiceoftendoesnotmeetsociallyunacceptablestandards.Inlow-andmoderate-incomecountries,thesehomesareoftenpoorlyconstructed,lackstandardaccesstowaterandsanitation,andprovidelittlefloorspaceperperson.Inhigh-incomecountries,thehousingqualityisusuallyacceptable.However,householdsmightconsumeverylittlefloorspaceperpersonrelativetotheirneighborsandpaymuchmoreinrentthanthe30%ofincomethatisconsiderednormal.Thelowhousingstandardsandthehighrentthataffectthelowerincomepopulationwilllegitimatelysoonattractpublicattention.Socialpressurewilleventuallyforcegovernmentsto“dosomethingabouthousing”.
Thisneedto“dosomething”pushesgovernmentstoformulatenewhousingpoliciesthatwillprovidesociallyacceptablehousingstandardsatanaffordablepriceforall.AngusDeaton,inhisbook,“TheGreatEscape”writes“Theneedtodosomethingtendstotrumptheneedtounderstandwhatneedstobedone.Andwithoutdata,anyonewhodoesanythingisfreetoclaimsuccess.”Thisperfectlycharacterizesthedesignofmanyhousingpolicies.
Householdsconsumedeficienthousingwhentheycannotaffordthehighcostoflandandconstructioninalargecity.However,governmentsmayexacerbatethehighcostofhousinginacitybylimitingthesupplyofhousingthroughregulationsandunderinvestinginurbanexpansion.Improvinglow-incomehouseholds’housingstandardsrequiresidentifyingtherelativeroleofbothfactors:povertyandinflatedhousingpricescausedbysupplyconstraints.
1WhyDoCitiesMatter?LocalGrowthandAggregateGrowth”,byChang-TaiHsiehandEnricoMoretti,
NBERWorkingPaperNo.21154,May2015,RevisedJune2015
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage3of21
Intheirsearchforsolutions,urbanmanagersoftenignorethathouseholds’housingchoiceis
drivenbyacombinationofthreeattributes:floorarea,location,andpricepersquaremeter.Becausefloorareaandconstructionqualityarethemostvisibleofthethreeattributes,plannerstendtoconcentrateonimprovingthedesignandincreasetheareaofdwellingswhendraftingahousingpolicy.Theytendtoignorehousinglocationanditscorollary:accesstocity’slabormarket.Whenlowhousingstandardsarelargelyduetopoverty,ignoringlocationtoprovidelargerhomesmightdevastatetheverypopulationthepolicyissupposedtohelp.Someexamplesbelowwillillustratethispoint.
Inthischapter,Iwilldiscusshousingaffordabilitypoliciesinseveralcitiesandshowhowthesepoliciesimpactedthefourattributes:floorarea,landarea,andpriceoflandandconstructionpersquaremeter.
Forhouseholds,thepriceofhousingPdependsonthevalueoffourobservablevariables:P=(landareaXpriceofland)+(floorareaXcostofconstruction)Therentpaidwillalsoberelatedtothesefourvariables.Thepriceoflanddependsonlocation;alocationwithahighjobaccessibilityorclosetohigh
qualityamenitieswillcorrespondtohighpriceofland.Ingeneral,ahighlydesirablelocationhasahighlandprice.
Thecostofconstructiondependsonthequalityofconstruction.Itispossibletobuildaninformalsheltermadeoflumber,plasticandcorrugatedironroofforaslittleasUS$25persquaremeter;whilethepriceofconstructionforanapartmentfullyequippedwithkitchenandbathroomsmaycostseveralthousandUS$persquaremeters(aboutUS$2500US$/m2inNewYorkCityin2013forresidentialbuilding3to7stories).
Therefore,householdssearchingforhousingatagivenpricewouldhavetomakeatrade-offbetweenlocation,landandfloorarea,andqualityofconstruction.AttimeIwilluselocationasaproxyforthepriceofland,andqualityofconstructionasaproxyforpriceofconstruction.
Fordevelopers,thecostcomponentsofproducinghousingunitsaremuchmorecomplex.Inadditiontothephysicalcostdescribedabove,developers’costwillincludefinancialcosts,overhead,managementsanddesigncosts.Theratiobetweenthelandandthefloorareawillbeusuallyconstrainedbyregulations.Thecostofconstructionwillalsodependinpartonregulations.However,forhouseholds,thepriceoflandandthepriceofconstructionaggregateallthesecostcomponents.
Inthefollowingchapter,wewillbeconcernedmostlybyhouseholds’choiceswhenselectinghouses.Thesechoiceswillthenbedrivenbyfoureasilyobservableparameters:location,landandfloorarea,andqualityofconstruction.
Aswewillsee,thehomesthatfamiliesendupoccupyingarestarklydifferentfromthosethey
wouldchooseiftheirincomehadsimplyincreasedbytheimplicitsubsidytheyreceive.Iwilljudgethemeritsofvarioushousingpoliciesbycomparingthehomestheyendupwithtothosetheywouldhavechosenwithanincomesubsidy.
2. Definingandmeasuringhousingaffordability
Housingaffordabilityisdifferentfromtheaffordabilityofanyotherconsumerproduct“Affordable”meanssomethingdifferentwhenitisusedforhousinginsteadofanotherobject,
say,acellphoneoracar.Apersonwhocannotaffordacellphoneoracardoesnothaveone.However,whenhousingissaidtobeunaffordabletohouseholdsbelowanincomeofX,itdoesnotmeanthatallhouseholdsbelowanincomeofXarehomeless.Itonlymeansthatthesehouseholdsarelivinginhousingunitsthatareunacceptableinquality,floorarea,and/orthatthesehouseholdsarespendinganunacceptablyhighproportionoftheirincomeonrentormortgagepayment.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage4of21
Housingaffordabilitythereforesayswhetherhousingis”sociallyacceptable,”notwhetheror
notahouseholdoccupiesahousingunit.WhenwereadthathousingisunaffordabletohouseholdsbelowincomeX,itmeansthatthetrade-offsnecessaryforhouseholdsbelowincomeXtorenttheircurrentdwellingunitareinadequateintermsofrentpaidinproportionofincome,floorarea,qualityofconstructionorlocation.
Thesociallyacceptedminimumhousingstandardsineachcitydonotcorrespondtoa
scientificallyaccepteduniversalnorm.Inthisitdiffersfrommanyothernorms.Forinstance,minimumnutritionaldailyintakeisauniversalnormdefinedforallhumanbeings.MostairpollutionnormsareestablishedbytheWorldHealthOrganizationandareacceptedasuniversal.Bycontrast,minimumacceptablehousingstandardsarerelatedtotheprevailingstandardsinthecitywheretheyareapplied.ThesociallyacceptableminimumhousingstandardsinStockholmareverydifferentfromthestandardsinDhaka.Thisisduetodifferencesinclimateandcultureinadditiontodifferencesinhouseholds’incomebetweenthetwocities.
WhilemanyhouseholdsinDhakamighthappilyliveinhomesmeetingtheminimumnormsprevalentinStockholm,thereisnoevidencethatDhaka’shouseholdssufferirreparabledamagesbylivinginhousesofsignificantlylowerstandards.Minimumhousingstandardsarethereforealwaysarbitrary.Thesestandardsmightbeusefulasabenchmark,butwhentheybecomeenshrinedinlawsandregulationstheycandogreatharmtotheverypopulationtheyaresupposedtohelp,aswewillseebelow.
Inmanycountries,suchasSouthAfrica,thegovernmentidentifiesasetofminimumhousingstandardsdefininganationalminimumhousingnorm.Settingminimumhousingstandardsisapoliticalact.Governmentstendtoselecthighstandardsasanoptimisticsignalforthefutureofthecity,thesortofoptimisticprojectionthatpoliticiansareallobligedtomake.Statisticiansthencomparethestandardsoftheexistingurbanhousingstock(obtainedthroughsurveysandcensus)withthenationalminimumhousingnormestablishedbythegovernment.Thenumberofexistingdwellingunitsbelowthenationalhousingnormissaidtoconstituteahousing“backlog”.Toeliminatethisbacklog,thegovernmentcommitsitselftobuildingenoughhousingunitseachyeartoclearthebacklogwithinagivenperiod,say,10years.Notethatthehousingprogramisdefinedonlythroughtwoattributes:priceandphysicalhousingstandard.Locationisabsentfromthepolicy.Inanycase,itwouldbedifficulttodefinealocation“standard”atthenationalscale.SouthAfrica’shousingprogramprovidesawarningoftheadverseconsequencesofignoringlocationwhendefininghousingaffordability.
Howgovernmentsdefinehousingaffordabilityisthereforeveryimportantwhendevelopinghousingprogramstohelpthepoor.Governmentbureaucratstendtomakedifferenttrade-offsthanthehouseholdswouldmakeforthemselveswhenchoosingbetweenprice,location,areaandquality.Ifthetrade-offsmadebygovernmentdifferalotfromthosethathouseholdswouldmake,thenthehousingprogramwillfaildespitethemoneyinvestedandthegoodintentionsoftheexpertdesigners.Urbanplannersdonothaveenoughinformationtoenablethemtoselectan“optimum”combinationofrent,floorareaandlocationforeachhouseholdandfirm.Thechoiceofthequantityoflandandfloorareaconsumedinaspecificlocationisthereforebetterlefttotheend-userwheneverpossible.
Ihaveseenmanygovernmentsimplement“slumrelocation”programsthatsendhouseholdsfromslumstohighqualityandsubsidizedformalhousingunitsinaremotelocation.Tothedismayofgovernmentofficials,theformerslumdwellersoftenabandontheirformalhousingtoreturntoaslumwherebuildingqualityislowerbutaccesstothejobmarketisbetter.Thisreturnisoftenattributedtoslumdwellers’lackofjudgement.Thisisnotthecase.Theresidentsreturnbecausetheypreferawell-locatedhomeoflowerqualitythantoapoorly-locatedhomeofhighquality.Policy-makersfailedtochoosethebesttrade-offbetweenrent,locationandhousingstandards.Affordabilityshouldnotbedefinedbymerelywhetherahouseholdcanaffordtopayforitsdwelling.Affordabilitymustalsoconsiderwhetherthathomeisthebestpossiblecombinationofsize,quality,locationandprice.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage5of21
Thisisnotmeanttoimplythatdeficienciesinhousingqualityareagovernmentinvention.
Manylow-incomehouseholdsaspiretomuchhigherqualityhousingthanwhattheycancurrentlyafford.Inmanycases,theinsalubrityorbadlocationoftheirdwellingslowslow-incomehouseholds’integrationinthemoreproductivepartoftheurbaneconomy.
Governments’affordablehousingprogramsareneverdesignedtoenticehouseholdswholiveinadifferentcityorruralareatomovetothecityprovidingthehousing.Tothecontrary,socialhousingprogramsusuallyspecifythatpotentialbeneficiariesmusthaveresidedinthecityseveralyearstoqualifyforgovernmenthelp.Thisresident-onlypolicyismeanttopreventanimmigrationstampedetothecity.Itisimportanttodefinehowcitiesidentifythehouseholdswhoarefacinganaffordabilityissue,or,inotherwords,householdswhomadeatrade-offthatresultsinunacceptablelowhousingstandardsintheviewofthecommunity.Because“affordable”housingprogramsareusuallydesignedforpeoplealreadylivinginacity,unaffordablehousingisdefinedbyunacceptablysmallsizeorlowqualityofconstruction,notasadistantlocationthatmakesforanunacceptablylongcommute.
“Affordablehousing”policyaimstoincreaselow-incomehouseholds’housingconsumptionuntiltheyhavereachedasociallyacceptablelevel.Todesignthispolicy,weneedtoquantify,first,theminimumsociallyacceptablehousingconsumptionlevel,and,second,thenumberofhouseholdswhoconsumelesshousingthanthislevel.Oncethesetwonumbersareidentifiedmunicipalgovernmentscanhaveaninformeddiscussionaboutwhatitmightdotoaddresstheaffordabilityissue.Shoulditbuild“affordablehousing”atorabovetheminimumstandardandthensellorrentthishousingbelowmarketprices?Openupnewareasforurbandevelopmenttoincreasehousingsupplyandlowermarketprices?Reviseregulationsthatrestrictdevelopersfromprovidinghousingthatmeetstheminimallyacceptablestandard?Expandthefinancialsectortoprovidemortgagestolowerincomehouseholds?Ordirectlysubsidizehouseholds’incomesothattheycanaffordahigherqualitydwellinginalocationoftheirchoice?
Almostalways,affordabilityissuesrequireseveralsimultaneousactionsinvolvinginvestmentprogramsandregulatoryreforms.Nosilverbulletcaneasilysolvehousingaffordability.However,governmentscannotdesignacrediblepolitywithoutclearmeasuresofboththethresholdbelowwhichhousingstandardsareunacceptableandthenumberofhouseholdswhofallbelowthisthreshold.Therefore,beforediscussingspecificpoliciesindetailIwillfirstdiscussthevariousmethodsofmeasuringtheaffordabilitythresholdandthenumberofhouseholdsthatfallbelowit.Fuzzydataonhouseholds’incomeandcurrenthousingstandardsisasignificantimpedimenttocreatingsensible“affordablehousing”policies.
Asimple“affordability”index:thePrice/Incomeratio
ThePrice/IncomeRatio(PIR)measureshousingaffordabilityinacitybycomparingthemedianpriceofadwellingwiththemedianhouseholdincome.Thissimpledefinitionmakesitiseasytocomparethepriceofhousingindifferentcitiesthathavedifferentincome.However,thisindexdoesnotsayanythingabouthowmuchhousingahouseholdgetsforthemedianpriceorwherethisdwellingislocated.ThePIRalsoappliesonlytosalesandnotrentals,althougharenttoincomeratiocouldbedevelopedusingthemedianincome.
The“DemographiaInternationalHousingAffordabilitySurvey”2,issuedeveryyearsince2004,comparesthePIRbetweensome367metropolitanmarketsinninedevelopedcountries(Australia,Canada,HongKong,Ireland,Japan,NewZealand,Singapore,theUnitedKingdomandtheUnitedStates).Amongthesecities,87metropolitanareashaveapopulationlargerthanonemillion.Because
2The12thAnnualDemographiaInternationalHousingAffordabilitySurveycovers87majormetropolitan
markets(morethan1,000,000population)inAustralia,Canada,HongKong,Ireland,Japan,NewZealand,Singapore,theUnitedKingdomandtheUnitedStates.http://www.demographia.com/dhi.pdf
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage6of21
theindexconsistentlyusesthesamemethodology,itprovidesaninvaluabletooltocomparePIRsbetweencitiesaswellastoseehowtheseratiosevolvesovertime.
Letuslookattheprice/incomeratioindexforaselectionof26citiesintheyear2015(Figure1).ThecitiesselectedarerepresentativeofthevariationsofPIRshownintheentireDemographiasurvey.Amongtheselectedcities,AtlantahasthelowestPIR(3.1),whileSydneyhasthehighest(12.2).Whyaretheresuchlargevariationsinaffordability?Wenoticethatmanyofthecities—SanFrancisco,Auckland,VancouverandSydney—withhighPIRhaveadifficult(thoughbeautiful)topography.Themixofwaterandlandmakesforattractivecitiesbutremovesmuchofthelandavailablefordevelopment.Thistopographicalconstraintonthelandsupplyislikelytohaveanimpactonlandpricesandthereforehousingprices.ButwhiletopographycertainlyexplainssomeofthevariationsinPIR,itisnottheentirestory.CitieslikeChicago,WashingtonDCandTokyo-YokohamaalsohaveimportantwaterareasclosetotheirCBDbuthavesuccessfullymanagedtheseissues.ThosecitieshavePIRsthatarelessthanSydney.WewillseebelowthatlandusepolicyandregulationsconstrainingcityexpansionareoftenlargelytoblameforahighPIR.
Figure1:Price/incomeratioofaselectionofworldcities(2015)
WhatdoesitmeanforacitytohaveaPIRabove8?Intuitively,wefeelthathouseholds’welfareshouldbehigherinacitywithalowPIRthanina
citywithahigherone.Iflessofincomeisspentonhousing,morecanbespentonotheritems.However,averylowPIRmightindicateeconomicstress.IntheDemographiasurveyfor2015,Detroit’sPIRisalow2.8.SomeRussiancitieswithheavypopulationlossescanhavehomepricesclosetozerointheabsenceofdemand.Obviously,thePIRneedsinterpretation.Whilelowhousingcostcomparedtoincomeisgenerallyagoodthing,itmightalsoindicateotherproblems.ItwouldobviouslybeabsurdtouseDetroit’sPIRtojustifyusingthatcityasamodelofgoodhousingpolicyandaffordablehousing.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage7of21
WhatPIRvaluewouldindicateanaffordablehousingstock?Demographiasuggeststhat
housingisaffordableincitieswherethePIRisequalorbelow3.NoneofthecitieswhosePIRisshownonFigure1qualifies,althoughAtlantawithaPIRof3.1comesclose.Demographia’scomplete“degreeofaffordability”categoriesareasfollows:
Householdsusuallyborrowmoneytobuytheirfirstdwellingsoletuscalculatethemortgage
paymentsassociatedwithvariousPIR.Figure2relatesthecostofhousingasapercentofyearlyincomefordifferentvaluesofPIRunderthreepossibleborrowinginterestrates,5,7and9%,(over25yearswithadownpaymentof20%).Mortgagelendersusuallywouldprovideloanstohouseholdswhentheirmonthlypaymentdoesnotexceed30%oftheirincome,andthehorizontaldottedlineonFigure2representsthisaffordabilitythreshold.Whentheinterestrateis5%,onlyinAtlanta,Houston,TokyoandSingaporewillthemedianhouseholdbeabletoobtainamortgageforthemedianpricedhome.Withahigherinterestrateof9%,onlyinAtlantaandHoustonwillthemedianincomehouseholdbeabletoobtainamortgageforthemedianpricedhome.Whatwouldhappentohouseholdsintheothercities,wherethehighPIRimpliesthathouseholdsatthemedianincomewouldnotbeabletoaffordamortgagetobuyamedianpriceddwelling?
Figure2:Price/incomeratioandaffordabilityin10selectedcities
SomehouseholdsmighthaveboughtadwellingsomeyearsbeforewhenthePIRwasstillintheaffordablerange.Thesehouseholdsthenliveinahousethattheycouldnotaffordtobuynowwiththeirincomealone,buttheincreaseinPIRmeansthattheircapitalassetshaveincreased.Theycould,however,affordtobuyanewhousebysellingtheircurrenthouse,eventhoughthePIRshowsthatanewhousewouldbeunaffordable.ThesehouseholdsarethereforeprobablyquitesatisfiedbytheincreasingPIRvalue,evenifitshowsthathousingisunaffordabletothem.Thisfactmayexplainthe
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage8of21
regulatorypolicyofsomecitiesthatseemsintendedtoconstantlyincreasehousingpricesbyrestrictingnewsupply.
However,householdsthathavenotbenefitedfrompreviousPIRincreasesbutliveinahighPIRcitycannotaffordtobuyanewdwelling.SomemightdecidetomovetoadifferentcitywithalowerPIR,butchangingcityhassocialandfinancialcostsandrisks.Morelikely,thenewhouseholdwilltrytofindanalternativetobuyingahome.Forinstance,newlyformedhouseholdsmightrentratherthanbuyadwelling.Often,inhighPIRcities,monthlyrentsarelowerthanmortgagepaymentsforequivalentdwellings.IhavecomparedthePIRofnewownerstothepercentageofincomespentonrentinasampleof10UScities(Figure3).WhilethepercentofincomespentonrenttendstoincreaseincitieswithahigherPIR,therentalmarketstendtobemoreaffordablethanbuyingdwellings.SanFranciscoisanoutlierwithaveryhighPIRof9.4butrentsataratheraffordable32%ofrenterincome3.
However,thenumberofdwellingunitsavailableforrentmightdecreasewhenthePIRishigh,aslandlordswouldhaveanincentivetosellduetohighhomepricesandlowrents.
Figure3:PercentofrentinghouseholdsspentonrentsandPrice/IncomeRatioofowners
SomehouseholdsmightchoosetoleavethecityandlookforacitywithalowerPIR,evenifthis
meanshavingalowerincome.Butthemajorityofhouseholdswillhavetwooptions.Eitheradjusttheirlivingstandardsandoptforlowerstandardhousing,orspendamuchhigherpartoftheirincomeonhousing.Ifhouseholdsoptforthefirstsolution,householdswithmedianincomewillopttobuyadwellingwhosepriceislowerthanthemedianprice,outbiddingintheprocesshouseholdswithincomeslowerthanthemedian.Householdsatthebottomoftheincomedistributionwillnotbeabletooutbidanybodyandwillprobablybeforcedtosubdivideexistingdwellingstobeabletoaffordanewhome.InAuckland,NewZealand,withaPIRof9.7in2015,itisreportedthatanumberofhouseholdsarelivingingarages,trailersortheirparents’home.Thesearehouseholdsnearthebottomoftheincomescalethatcannotoutbidanylowerincomegroup.Inacitywithahigh,PIRhouseholdsarelikelytodevoteahighpercentageoftheirincometohousingandalsoreducetheirhousingconsumptioncomparedtowhatitwouldbeinacitywithalowerPIR.Reducinghousingconsumption,inthiscase,mightinvolvereducinganyorseveralofthehousingpricecomponents.Movingtoalessfavorablelocationtoreducelandprice;movingtoasmallerapartment;movingtoadwellingwithalowerstandardofconstruction,asthehouseholdsinAucklandwhohavemovedtotrailersandgarages.In
3Thepercentofincomespentonrentiscalculatedusingthemedianincomeofrenters,whichison
averageabout65%lowerthanthemedianincomeofthetotalpopulation.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage9of21
Mumbai,anumberofmedianincomehouseholds,whiletheirincomehavebeenrising,havebeenforcedtomoveintoslumsasthePIRhasincreasedfasterthantheirincome.
SomehouseholdsconsideringmovingtoacityforajobmightwellchooseadifferentjobwithalowersalaryinacitywithalowerPIR.ThisiswhatwasisimpliedbyChang-TaiHsiehandEnricoMoretti,quotedatthebeginningofthischapter.TheproductivityoftheseusuallyhighlyskilledhouseholdswouldbecomethereforelessthanwhatitcouldhavebeeniftheyhadmovedtothecitywithahighersalarybuthigherPIR.Thereisalossofwelfareforeverybody.
However,itseemsthatonlyhigherincomehouseholdswillmakethischoice.Giventheterribleconditionsintheslumsofmanyeconomicallydynamiccities,likeRiodeJaneiroorMumbai,itseemsthatevenmiddle-incomehouseholdsseemtopreferaneconomicallydynamiccitytolowerhousingprices.Thepossibilityofparticipatinginadynamiclabormarketattractsverylow-incomehouseholdstoexpensivecitieslikeRiodeJaneiroorMumbai,despitethedeplorablehousing.Tobeableparticipateintoadynamiclabormarket,theyarereadytoreducetheirhousingconsumptiontoabareminimum,ortochooselocationswithverylongcommute,ortoincreasetheshareoftheirincomethatwilldecreasetheirspendingonhealthoreducation.ThewelfareeffectsofahighPIRarenottrivial.Planners,giventheirresponsibilityforconstraininglandsupply,shouldmonitorPIRsregularlyandactwhenthePIRincreases.
ThePIRisausefulindextoidentifyanaffordabilityproblembutitistoocrudetoidentifyapolicysolutionThePIRisausefulandeasilyunderstoodindextoidentifyanaffordabilityprobleminhigh-
incomecities.However,itrelatesonlythemedianincometothemedianhomeprice.Ithasnothingtosayaboutthequalityorlocationofthehouseatthemedianprice.WhilethePIRissimpleanduncontroversial,collectingincomeandhomepricedatacanbedifficultincountriesthatdonothaveasystematicsaleregistrationsystemandincitiesthathavealargeinformalsector.ThisiswhyDemographiadoesnotyetcoverdevelopingcountries.Findingthemedianhousingpriceimpliesthatalltransactionsareequallywellknown.Inmanydevelopingcountries,iteasiertofindthepricesatthehighendofthehousingmarketthanatthelowend.Itisalsoeasiertofindpricesofnewhousingthanthatofexistinghouses.Inmanycities,thedatarequiredtocalculateacrediblePIRdoesnotexist.
HowdohouseholdsadjusttounaffordablePIR!
IncitieswithahighPIR,housingisassumedtobeunaffordablenotonlytothepoorbuttothemiddleclass.However,wedonotseepeopleleavinghighPIRcitiesenmasseformoreaffordablecities.ItseemsthatincitieslikeSydney,VancouverorSanFrancisco,lifegoesonasusualdespiteaveryhighPIR.ThesamecouldbesaidofcitiesforwhichnoPIRcanbecalculatedbuthavenotoriouslyhighrealestateprices,suchasMumbai,LagosandJakarta.Obviously,thevastmajorityofhouseholdsadaptto“unaffordable”pricesbychoosingnottoleavetheircurrentcity.Weevenseethatthepopulationkeepsgrowingincitieswithunaffordablepricesbecauseofmigrationandnewhouseholds’formation.
Buthighrealestatepricesareanythingbutbenign.Thisapparent“businessasusual”responsetorapidlyincreasinghousingpricesmighthideadeterioratingqualityofurbanlifeforallbutthemostaffluentresidents.Householdsadapttohousingpricesthatrisefasterthantheirincomebyconsuminglessfloorspaceandspendingahighershareoftheirincomeonrent.
Highpricesorabsolutepovertyforcepoorerhouseholdstoconsumelesshousingthantheminimum“sociallyacceptable”levelsetbyregulationsasmentionedabove.Fallingbelowthisminimumwillfurtherdecreasethehousingstandardsofthepoor.Consuminglesshousingthanwhatisprescribedas“sociallyacceptable”oftenpreventstheirhousingfromattaininglegalstatusandpermanence.Thiswillcompoundtheirmiserythroughaviciouscycle:povertycauseslowhousingconsumptionwhichcausesmorepoverty.
Inmiddleandhighincomecountries,manylessaffluentpeoplecanrespondtohighpricesbysubdividingexistingdwellingintosmallerunits,formallyorinformally.Inothercases,newhouseholdscohabitwiththeirparentsorotherrelativesmuchlongerthaneitherwouldwish.Ineithercase,high
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage10of21
housingcostsresultsinlowerhousingconsumption.Twocasestudiesdiscussedbelowillustratetheseforcedadjustments:thesubdivisionofapartmentsinBeijingsuburbsandthecohabitationofadultchildrenwiththeirparentsinEurope.
Finally,sometimescitiesrevisetheir“sociallyacceptable”minimumhousingconsumptiontoreflectthedemandfromachangingsocio-economicgroup.ThisoccurredinNewYorkCityin2016.There,theexistenceofalargenumberofsinglepersonhouseholdsconvincedcityregulatorstolowertheminimumhousingstandard.Thiscaseisalsodiscussedmoreindetailbelowasitillustratesthefutilityofsettingminimumsociallyacceptablestandardsinthefirstplace.
InformalsubdivisionofapartmentsinChinaChinesecitieshaveveryfewidentifiableinformalsettlements.However,thehousing
consumptionoflow-incomehouseholdsisoftendifficulttomeasure.ManyofthenewapartmentsbuiltinChinesecities’peripheryaretoolargetobeaffordabletolow-incomehouseholds.Asaresult,low-incomeresidentsaffordhousingbyrentingaroominasubdividedapartment.Thestreetposterappearingin2013inanorthernsuburbofBeijing(Figure4)advertisean18squaremetersroomtoberentedwithinalargerapartmentwithakitchenandbathroomtobesharedbyothertenants.Fortunately,Beijing’sgovernmenttoleratesthispractice,althoughotherapartmentownersinthesamecondominiumcomplexprotestthepracticeandroutinelytrytoconvincethemunicipalitytobanit.Fromahousingsupplypointofview,thisadjustmentisdesirablebecauseittransformsthebuilthousingstockintohousingthatisaffordable.Onceablockofapartmentshasbeenbuiltitisverydifficulttoreducethesizeofunitstomeetthedemandforsmallerunits.Informallysubdividingexistingapartmentsisthefastestwaytomatchsupplyanddemand.Thissituationneednotbepermanent.Overtime,thesupplyofnewlybuilthousingwillbettermatchingdemand,andthepracticeofsubdividingapartmentswilldisappearbyitself.
Figure4:PosterinasuburbofBeijingadvertisingaroomtorentinasubdividedapartment
Governmentshouldmonitorbutnotbanthepracticeofsubdividingapartments.Ifthepracticeofsubdividingapartmentspersistsovertime,regulationsmaybeblame.Forexample,anarbitraryminimumapartmentsize,oramaximumnumberofdwellingperblockmightberesponsibleforthemismatchbetweensupplyanddemand.Removingtheseregulations,whichhavenodiscernablebenefits,wouldallowthehousingmarkettorespondtochangingconsumerdemand.
YoungpeoplelivingwiththeirparentsInaffluentcountries,assessingtheimpactofhighprice-to-incomeratiosonthehousing
consumptionofspecificincomegroupcanbedifficult.A2013PewResearchCentersurveyshowstheproportionofpeopleaged25to35wholivewiththeirparentsinEuropeancountriesandtheUS
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage11of21
(Figure5).Thispercentagevariesfrom1.8percentforDenmarkto56.6percentfortheCzechRepublic.Culturalfactorsmayexplainsomeofthedifferencebetweencountries.However,economicfactors,liketheemploymentrate,andhousingsupplyissuesalsoaffecttherate.Whateverthereasonsfortheinternationaldifferences,mypointistoillustratethathousingconsumptionadjustswhenthereisadiscrepancybetweensupplyanddemandforhousing.
Figure5:Shareofyoungpeopleaged25to34livingwiththeirparentsinEuropeandintheUS
Reducing“minimumsociallyacceptablehousingstandardsMicroapartmentsinNewYorkInNewYork,a1987cityzoningregulationspecifiedthattheareaofapartmentsshouldbeat
least37.2squaremeters(400squarefeet).Butthesupplyofapartmentsofthissizeisconstrainedbyanotherzoningregulationsthatputsamaximumlimitonthenumberofdwellingunitsperacre,thusimplicitlyreducingthenumberofsmallapartmentsthatcouldbebuiltinablock.
However,thedemandforsmallapartmentshasbeenincreasingasthenumberofpeopleperhouseholdshasdecreasedinthelastfiftyyears.In2015,thenumberofnon-familyhouseholds,i.e.householdsformedbysinglepersonorunrelatedindividuals,represented38%ofallhouseholds.Recognizingthisproblem,in2015,thezoningboardallowedtheconstructionof55“mini-apartments”rangingfrom24to33squaremeters,inasingleninestorybuildingontheEastSideofManhattan.Thiswasatimidsteptowardusingacommon-senseapproachtorepealminimumsociallyacceptablestandards.
Whenthe55mini-apartmentsappearedonthemarket,therewere4,300applicantsforeachapartment!Thisdemonstratedthelargedemandforsmallunits,whichisarbitrarilyconstrainedbytheminimumapartmentsizeregulation.TheapartmentbuildingiscentrallylocatedwithexcellentaccesstoNewYork’slabormarket.Loweringtheminimumapartmentsizeregulationgaveindividualsthefreedomtomaketheirownchoiceinthetrade-offbetweenlargerapartmentsinlesscentrallocationsandcentrallylocatedsmallerapartments.
However,ifthemunicipalityremovedtheapartmentminimumsizeconstraintfortheentirecity,developerswillstillnotbuildmoremini-apartments.Anotherregulationlimitingthenumberofdwellingunitsperblockwillpreventthat.Limitingthenumberofdwellingunitsperblockwasmeanttolimitresidentialdensity.Butsincetheregulationwasputinplace,residentialdensityhasfallenastheaveragehouseholdsizehasfallen.Layersofregulations,whoseoriginalobjectivehasoftenbeenforgotten,preventhousingsupplytomatchdemand.Thesemultiplelayersofregulationsmustberepealedforhousingsupplytoberesponsivetohousingdemand.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage12of21
Figure6:NewYorkCity"minimumacceptablehousingstandardssince1860
InNewYorkCity,minimumhousingregulationshaveevolvedovercenturies.Figure6showsexamplesoffloorplansofapartmentsofthesmallestacceptablesizeatdifferentdates.In1860,housingconstructionstandardswerepracticallyunregulated.Developersrespondedtohousingdemandfromallsocio-incomegroups.ThetypicalfloorplanofatenementshownasAontheleftofFigure6,builtin1860,showseachfloorhas4apartmentsof3roomseach.Onlyoneroomineachapartmenthaswindows.Bathroomsweresharedamongallbuildingtenantsandprovidedonthegroundfloorinthebackyard.Theapartmentsaredesignedsuchthathouseholdsmayoccupyonlyoneroomorseveralconnectingrooms.Atthetime,householdswereverylarge,oftensixorsevenpersonsperhousehold.Thepopulationdensityintenementneighborhoodswereabout660peopleperhectarein18604.Thedensitypeakedat1530p/hain1910andfellto390p/hain2010.
TheminimumsociallyacceptablehousingstandardevolvedovertimeandareformmovementresultedintheTenementHouseActin1879(planBonFigure6)whichrequiredroomstohaveaccesstoventilationshafts.Inaddition,regulationsrequiredabathroomandtoiletoneveryfloor.In1987,theminimumapartmentsizeallowedbyregulationswas37.2squaremeters.Theplanofastudioofthissize,builtin2016isshowninC.Finally,theplanofoneofthe55“micro-apartments”of28squaremeterstobebuiltinthemiddleofManhattanmentionedaboveisshowninD.
Theseexamplesillustratethefutilityofcontrollingmaximumdensitiesorminimumfloorarea
perpersonthroughregulations.Theveryhighdensitiesofthetenementsinthe1860swerenotgeneratedbydesignorregulationsbutbythemarket.Thetenements’excellentlocationandthetenants’verylowincomecreatedthehighdensity.
Manyofthese“oldlowtenements”survivetothisdayinManhattan.AstudybyStephenSmithandSandipTrivedipublishedintheNewYorkTimesin2016showsthatabout40percentoftheexistingbuildingsinManhattancouldnotbebuilttodaybecauseofthecompoundeffectofoverlappingregulations!Itisdifficulttounderstandtherationaleforsuchregulations,althoughpracticallyeverycityintheworldhassimilarrules.
4“TheRiseandFallofManhattan'sDensities,1790–2010”bySollyAngelandPatrickLamson-Hall,5-Dec-2014,MarronInstituteofManagement,WorkingPaperSeries;18
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage13of21
InformalsubdivisionsinNewYorkCity:creatingaffordablehousingbelowtheminimumsociallyacceptablestandard
Subdividinglargeapartmentsintoroomsindividuallyrentedtopeoplewhoshareakitchenandbathroomisusuallylegalinmanycities.NewYorkCity,however,limitsthisrighttonomorethanthreeunrelatedindividuals.Subdividingapartmentsorhousesintoindividualindependentunitswiththeirownkitchenandbathroomisillegalinmostcities.
InNewYork,arecentstudy5showedthatbetween1990and2000about114,000newhousingunitswereillegallycreatedbysubdividingexistinghousesandbytransformingbasementsandgaragesintonewunits.TheseunauthorizeddwellingsaccountedforhalfofthehousingstockaddedinNewYorkCityinthe1990s.Thedwellingsrepresentedaboutfourpercentofthetotalhousingstock,andprovidedsheltertoabout300,000to400,000people.
Subdividinghomeslikethisisillegal,buttheactioncreatesaffordablenewunitswithoutanygovernmentsubsidies.AnotherstudydescribestheplightofnewemigrantsfromBangladeshwhosettleinNewYorkCity.Mostofthesehouseholdshaveverylowincomeandwouldbeunabletoaffordanylegallybuiltdwelling.Severalfamiliespooltheirresourcestobuyadetachedhouseinalow-incomepartofQueens.Theythensubdivideitintoseveralindependentunitswheretheylive.Theseillegalunitsenterthemarketandaresubsequentlyeithersoldorrented.
Thecitysendsinspectorstopreventthisfromhappening.Theargumentagainsttheseinformalsubdivisionsisthattheyoverloadtheutilitysystem,urbantransportandschoolsandbecauseofthehigherdensitytheycreate.However,itisunlikelythattheutilitysystemisreallyaffectedbecauseofthedecreaseinhouseholdsizesinthelastthirtyyears.However,schoolmightindeedbecomeovercrowdedbecauseimmigrantstendtohavemorechildrenthannativebornhouseholds.However,aprimaryfunctionofamunicipalityistoprovideschoolspacetoallthecity’schildren.Itdoesnotmakesensetopreventfamiliesfromsettlinginaneighborhoodunderthepretextthattheexistingnumberofclassroomsisinsufficient.Theinterdictionagainstsubdividingisusuallyapretexttohidethemunicipality’sinabilitytoprovideadequatenumberofclassroomtoitsresidents.Manyzoningregulationsareestablishedtopreventchangesofanynature,includingpreventinglowerincomefamiliesfromlivinginmiddleincomeneighborhoods.InthecaseoftheBangladeshimigrants,theyoutbidtheirmoreaffluentneighborsbyconsuminglessfloorspacethantheexistingresidents.Itistheoppositeofgentrification.Whilethecityplanningdepartmentclaimssocialinclusion,i.e.neighborhoodswithmixedhouseholdsincome,asitsobjective,itspreventtheemergenceofmixedincomeneighborhoodsbyzoninglawsthatpreventlowerincomehouseholdstoaffordtoliveinhigherincomeneighborhoods.
Theaboveexamplesshowhowhouseholdsadjusttohighpricesbyconsuminglesshousing.Ideally,therewouldbeamatchbetweenhousingsupplyanddemand.Becauseoftheinevitablelagbetweendemandandsupplychanges—forinstancewhenhouseholdssizedecreases—regulationsshouldallowtheseinformaladjustmentstooccurlegally.
Whenthepoorareunabletosubstitutecapitalforland
Ascitiesexpand,centrallylocatedlandbecomemoreexpensive.Householdsandfirmsrespondtothisbymovingintomultistorybuildings—apartmentsandofficetowers—whichreducestheirlandconsumption.Bythisaction,theysubstitutecapitalforland.Bybuildingmulti-storeystructures,theycanincreasetheirconsumptionoffloorspacewhiledecreasingtheirconsumptionofland.By
5RobertNeuwirth,andChhaya“NewYork’sHousingUnderground:ARefugeandaResource”Pratt,
ChhayaCenterforEconomicDevelopmentCDC(2008).
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage14of21
substitutingcapitalforlandandthereforeconsuminglesslandperdwellingunit,lowerincomehouseholdscancompeteforthesamelandwithhigherincomehouseholds.
Thisispossibleincitieswherelowerincomehouseholdscanaffordtheincreasedcostofmulti-storeyconstruction,astructureofatleastabout18squaremeters6inreinforcedconcrete,whichwouldbestructurallystrongenoughtosupportstakingupapartmentsoneaboveanother.Inthelowestincomecountries,whereconstructionisthecheapest,householdswouldneedtobeabletoaffordatleastUS$6,000forastudioof12m2.Inotherwords,substitutingcapitalforlandrequiresaminimumofcapital.Insomecities,thepooresthouseholdscannotaffordthisminimumcostthreshold.Becausetheyareunabletosubstitutecapitalforland,thefloorspacetheyconsumeisevensmallerthanthelandtheyoccupy.Theycanaffordashelterbyconsumingverylittlelandandevenlessfloorspace.Theextremelynarrowpassagewaysfoundinslumsinmanycitiesisnotdueto“poordesign”butisarationalchoiceforhouseholdswhodesperatelyneedmorefloorspaceandarereadytotrade-offstreetspaceforadditionalfloorspace,astheyaretoopoortobuildhigherhouses.
Thefollowingexample,takenfromBhayandarWest,aNorthernsuburbofMumbai,illustratestheconsequenceofbeingunabletosubstitutecapitalforland.Figure7showstwosettlementsbuiltsidebyside.Ontheleft,settlementAisaverylowincomecommunitylivinginaninformalsettlementwherehousesarebuiltofscavengedwoodandcorrugatediron,structurestooweaktobeextendedvertically.Ontheright,settlementB,amiddle-classcommunitymadeofapartmentsinsevenfloorsbuildings.CommunityAistoopoortosubstitutecapitalforland,communityBcanafforddoso.Letuscomparethewaytheirconsumptionoflandandfloorspacediffers,shownonthetabletotherightofFigure7.
Themiddle-classcommunityBconsumesanaverageof23m2offloorspaceperpersonwhilethepoorcommunityAconsumesonly3.5m2.However,thelandconsumptionofbothcommunitiesisrelativelyclose:4m2forthepooragainst6m2forthemiddleclass.ThepoorhouseholdswhocannotaffordtheminimumbuildingcostofUS$6,000foroneroominanapartmentblockareobligedtousemorevaluablelandperunitoffloorspacethanthewealthierhouseholdsintheformalsettlementB.ThepoorhouseholdsinsettlementAmustuse1.16m2oflandpersquaremeteroffloorspace,whilehouseholdsinBuseonly0.27m2oflandpersquaremeteroffloorspace.HouseholdsinB,becausetheycanaffordapartmentsinmultistoreybuildingscanalsoaffordtoallow46%ofthelandtoremainasopenspacecomparedto13.5%inthehorizontalsettlementA.
Figure7:MumbaiNorthernsuburbs-Informalandformalsettlement
Despiteconsistingofonlygroundfloorstructures,theresidentialdensityofthehorizontalsettlementAismuchhigherthantheverticalsettlementB.I
6Thisareaincludesstaircasesandcorridorsthatareindispensableformultistorystructures,itassumea
livingspaceofabout12squaremeters.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage15of21
ItseemsthataboutUS$6,000(in2016)isthecostthresholdbelowwhichpoorhouseholdswho
cannotaffordthissumarecondemnedtoliveinhorizontaldevelopment.Inlargecitieswherelandisexpensive,thepooresthouseholdsareoftenobligedtoconsumemorelandperunitoffloorspacethanhigherincomehouseholds.Thisresultsinextremelylowhousingconsumptionforlow-incomehouseholds.Bycontrast,incitieswherepoorhouseholdscanaffordtospendmorethanUS$6,000perhouse,thehorizontalslumsshownonFigure7tendtodisappearandbereplacedbymultistoreyapartmentsthatallowamuchhigherconsumptionoffloorspace.TheurbanvillagehousinginShenzhendiscussedbelowwillillustratethiscase.
Thedevelopmentanddiffusionofbuildingtechnology,likeprestressedsmallprefabricatedbeams,couldsubstantiallylowertheUS$6,000costforanapartmentinamultistoreybuilding,andwouldthereforeincreasethehousingconsumptionofthepoormuchbeyondthesavingsinconstructioncosts,becauseitwouldallowamuchlargernumberofhouseholdstosubstitutecapitalforland,ashigherincomegroupsareroutinelydoing.
Inaddition,constructingofmultistoryapartmentbuildingstypicallyrequiresfinancing.Itisnearlyimpossibleforhouseholdstoself-financesuchastructure,thewaytheydoitforhorizontalhousingwhichcanbeimprovedinstages.Acity’sfinancialsectormustthereforebeabletoprovidemortgagesaswellasconstructionfinancefordeveloperstoimprovelandefficiency.
MeasuringIncomedistributioninrelationtohousingconsumptionisindispensableforpolicyformulation
UsingCities’householdsincomedistributionUsingamedianincometomeasureaffordabilityisajustifiedsimplificationwhencomparing
differentcitiesorwhenlookingforatrendinatimeseries.Itisalsoanacceptablesimplificationforcitieswithlargemiddleclasswheremosthouseholds’incomesarecloselyclusteredaroundthemedianincome.However,whentryingtoimprovehousingaffordabilityinaspecificcityitisnecessarytolookattheincomedistribution:ahouseholdwithamedianincomemayrepresentonlyaverysmallsocio-economicgroup.Thisisparticularlytrueinlargecitiesofdevelopingcountrieswhereincomesaremorewidelydispersedthaninmoreaffluentcities.
Figure8:Shanghaihouseholds'incomedistributionin1998
ThegraphofFigure8showsthedistributionofhouseholds’incomeinShanghaiin1998.Theincomeisdisplayedatequalintervalsalongthehorizontalaxis.Thebluebarsshowthenumberofhouseholdsineachintervalusingtheleftaxis.Thedotedredcurvesuperimposedonthebarchartshowsthecumulativepercentofhouseholdswithineachincomeintervalusingtherightaxis.The
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage16of21
graphshowsvisuallythenumberofhouseholdsindifferentsocio-economicgroupsthatcompeteforlandandhousing.Thisgraphicrepresentationofallincomegroupsinacityconveysmuchmoreinformationthanusingmedianincomeortheimprecisetermsof“lowincome”,“mediumincome”andsoon.Forinstance,usingthegraphofFigure8,the180,000householdswithannualincomebelowRmb6,000haveaverydifferentaffordabilityproblemthanthe260,000withannualincomearoundRmb14,000.However,bothofthosegroupshadincomewellbelowShanghai’smedianincomeofaroundRmb21,000(horizontaldottedlineonFigure8).Acity’sincomedistributioncurveisanindispensabletooltoanalyzeandquantifyhousingaffordabilityissues.
HousingStockandflow:thetrickledowntheoryTheshapeoftheincomedistributioncurvemayalsohelpanticipatethepolicyimpactof
affordability.Theshapeofthegraphenablesthetestingofwhetherthe“trickle-down”affordabilitytheory7islikelytohappenofnot.Forinstance,imaginethatdevelopersincreaseby10%thenumberofnewhousingunitsaffordabletohouseholdswithanincomearoundRmb14,000.ThiswouldimproveaffordabilityforhouseholdswithincomelowerthanRmb14,000,asthenumberofdwellingsvacatedbythebeneficiarieswilllikelytrickledowntolowerincomegroupsandhaveasignificantimpactasthesegroupsarelessnumerousthantheoriginalbeneficiaries.However,ifthesame10%increaseinnewhousingunitsisbuiltforhouseholdswithincomearoundRmb36,000,theincreaseinnumberofhousingunitswillalsotrickledowntowardlowerincomegroupsbutwillsoonhaveaninsignificantimpactbecauseofthemuchlargernumberofhouseholdsamongthelowerincomegroup.Thetrickle-downeffectdoesoccurineverycase,butitseffectwillbecompletelydilutediftheincreaseindwellingunitsistargetedtohouseholdswhoseincomeismuchtotherightofthedistributionmode(inthecaseofShanghaiasshowninFigure8,themodecorrespondstohouseholdswithincomearoundRmb22,000).Ifthenumberofhouseholdsbyincomeintervalwereequal(ifthegraphwasshowingahorizontallineratherthanacurve)thenthetrickledownwouldworkperfectly.
Ofcourse,the“trickledown”effectcouldalsobecomea“trickleup”.Imaginethatagovernmentconstrainsthehousingsupplyofhigherincomegroupsandfavorexclusivelythebuildingoflowercosthousingunits(say,forincomearoundRmb12,000onthegraphofFigure8)intheabsenceofnewsupply,higherincomegroupswilloutbidthelowerincomegrouptooccupytheonlynewunitsonthemarkets.Thetrickle-downwillthenbecomeatrickle-up.Trickle-upmeansthathousingunitspreviouslyaffordabletolowerincomearebeingboughtbyupperincome(gentrification).Thishappenquiteofteningovernmentsubsidizedhousingwhentheoverallhousingmarketisheavilyconstrainedbylanduseregulationsor/andlackofinfrastructureexpansionthatconstrainslandsupply.Higherincomegroups“invade”thehousingstockofthelowerincomegroup.Theeffectisparticularlyseverewhenhigherincomegroupsacquireexistingdwellingsonlytoreassemblethemintolargerones.
InChennai,inIndia,inthe70s,themunicipalgovernmenthadavigorousprogramtobuildsubsidizedpublichousingwhileconstrainingthedevelopmentoflandforallotherincomecategoriesthroughregulationsandinadequateinfrastructuredevelopment.However,low-incomehouseholds,whohadbeencarefullyselectedbasedonincometobenefitfrompublichousing,oftensubletorinformallysoldtheirapartmentstohigherincomehouseholds.Thegovernment,then,didnotreactbyadjustingitshousingpolicybyreleasingmorelandforhousing.Instead,itconcentrateditsactioninpreventingtrickle-upsalesorsubletting.Itrequiredallmembersofhouseholdsinpublichousingtohaveidentitycardswithphotographsthatcouldbepresentedtoinspectorswhoconductedrandomvisits.Thiswasanexampleofatrickle-upeffectthatisquitecommonwhenlanddevelopmentpolicy
7The“trickledown”affordabilitytheoryassumesthatanyincreaseinthehousingstock,nomatterat
whatunitprice,wouldeventuallyimprovethehousingsupplyofeveryhousehold,eventhepoorest.Householdsbenefitingfromthesupplyincreasewould“moveup”tothenewhousing,thusfreeinganequivalentnumberofunitsthatwouldbecomeaffordabletohouseholdswithincomelowerthanthatofthebeneficiariesofnewhousing.Eventually,themovingupintobetterhousingwillrepeatitselfandthebenefitswill“trickledown”tothelowestincomegroups.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage17of21
andregulationsareatoddswithhousingpolicy.Iwilldiscussthistopicmoreindetailbelowwhenlookingathousingpolicyoptions.ThereactionoftheChennaigovernmentisalsotypicalofgovernmentsinmanycountries.Whendatashowsthatapolicyisnotworking—beneficiariessellingtheirsubsidizeddwellingtohigherincomegroups—governmentstrytoforcesuccessthroughmoreregulations.
Householdincomedistributioncurvesshowthecomplexityofanyhousingpolicyaimedatinsuringasupplyofaffordablehousingtoallhouseholds,especiallywhenincomesvarywidely.Iwillusetheincomedistributioncurveasamajortooltotesthousingpolicyoptions.
Whathappenwhenincomeincreaserapidly?Figure8showsShanghai’s1998incomedistribution.Theshapeofthecurvewilllikelybequite
differentafteronlyafewyears.Newunskilledmigrantsmightincreasethenumberofverylowincomehouseholdsontheleftsideofthegraph,whiletheincomeofotherhouseholds,whohavebeenurbanizedforalongertime,mightincreaserapidlybecauseofincreasedproductivityandskills.Theincreaseinincomewilladdhouseholdsinthemiddleandrightsideofthegraph.Thechangeinhouseholds’incomedistributionwillchangethedemandforhousingandrequireanadjustmentinthesupplyofnewhousingunits.Thepriceandstandardsofnewhousingshouldadjusttothenewdemand.
LetuscomparetheincomedistributionprofileofShanghaiof1998withthatof2003(Figure9).DuringthisperiodShanghaimedianincomeincreasedfromRmb21,000toRmb32,000inrealterms,a58%increaseatanaverageannualrateof8.8%8.Thisveryhighincomegrowthrateisexceptional.InShanghai,itwasaperiodwhereboldeconomicreformsandlargeinfrastructureinvestmentsimplementedinthepreviousdecadedramaticallyincreasedurbanproductivity.Duringthesameperiodthenumberofhouseholdsincreasedby17%oranaverageof3.3%ayear,alsoanexceptionalgrowthrateforacitywithapopulationof15.5millionin1998.ThenaturaldemographicgrowthrateofShanghaiduringthisperiodwasslightlynegativeat-0.08%.Thepopulationgrowthratewasthereforeentirelyduetomigration.
WhilethepopulationandincomegrowthrateofShanghaiareexceptional,theyprovideinsightintohousingaffordabilityissuesthatemergewhenurbanincomedistributionsarechanging.ThechangesinShanghaiarecompressedoverashortperiodof5years.Inothercities,comparablechangesmaybespreadoveralongerperiod,say,10years,buttheyareneverthelessdauntingandneedtobeaddressed.Inmanagingacity,nothingismoredamagingthanassumingastaticsituation.
The58%increaseinthemedianincomedoesnotreflectauniformincreaseamongallincomeclasses.Thewayhouseholds’incomesaredistributedhasimportantimplicationsforhousingaffordability.Inspiteofthelargeincreaseinmedianincome,thenumberofhouseholdsintheverylowincomegroup,belowRmb6,000peryear,increasedby53%,representing70,600additionalhouseholds.Thisincreaseisconsistentwiththehighrateofmigration.Alargenumberofmigrantsarecomingfromthecountrysideandhavenotacquiredyettheskillsneededtoaccessproductiveurbanjobs.
Inthenextcategory,lowmiddle-incomefrom6,001to24,000Rmb,thenumberofhouseholdsdecreasesby1.9millionor-58%comparedtothenumberofhouseholdsinthiscategoryin1998!Bycontrast,theincomegroupaboveRmb24,000increasedby2.8millionhouseholds.
AccordingtotheShanghaiMunicipalStatisticBureau,153.8millionofsquaremetersofresidentialfloorareawerebuiltduringthisperiod,orabout165squaremetersperadditionalhouseholds.Inaggregate,itseemsthatthesupplyofhousinghasmorethankeptpacewiththegrowthofpopulation—aremarkableachievementgivenShanghai’sfastdemographicgrowth.However,theaggregateamountoffloorspacedoesnottellusthetotalnumberofunitsbuilt,howlargetheywere,theirpriceorlocationorwhatcategoryofhouseholdswereabletoaffordthem.
8http://www.inflation.eu/inflation-rates/china/historic-inflation/cpi-inflation-china-2000.aspx
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage18of21
Housingaffordabilityshouldnotbecalculatedinaggregatebutbyincomegroup.Squaremeters
offloorspacearenotsoldindividuallybutinlumpsasapartmentsinspecificlocationsthatdeterminetheirprice.Low-incomehouseholdsthereforemaynothaveaccesstoallthefloorspacebuilt.Affordabilityassessmentcannotbedoneinaggregatebycomparingnewhouseholdsformationtonewhousingunitsbuilt.Theflowofnewsupply,measuredinhousingunitsratherthanaggregatefloorspaceshouldbedisaggregatedbynumberofnewunitsputonthemarketsthatareaffordabletospecificincomegroup.Inconductingaffordabilityevaluationtotestapolicyitisnecessarytodisaggregatethenumberofunitsproducedwithprices/incomes,floorconsumptionandlocation.
Figure9:Shanghaichangesinincomedistributionbetween1998and2003
IncomedistributionrelatedtohousingtypologyIhaveshownthewidevarietyofincomethatexistswithinacity.Everyhouseholdwhose
incomeisrepresentedonanincomedistributioncurvelivesinsomekindofshelterthatitcanaffordundercurrentconditions.However,thequalityofthissheltermayrangefromtwosquaremetersofcardboardonasidewalktoaluxuryvillawithindoorswimmingpool.Toidentifytherealaffordabilityissuewewillhavetomatchincomedistributionwithshelterconsumptionandtodecideatwhatpointtheshelterconsumptionhavefallenbelowthesociallyacceptableminimum.Inlookingforapolicysolutionwewillhavetoknowhowmanyhouseholdscurrentlyliveinashelterbelowtheminimumacceptablehomequality.Thepolicyoptionswillbedifferentdependingonthenumberofhouseholdsthatfallsunderthisminimum.Imaginethatinacityofonemillionpeoplewhereonlyfivehundredpeoplearelivinginsheltersmadeofcardboardandplastic.Thesolutionisprobablyawelfarebudgetallocationtomovethesefivehundredhouseholdstoadequatesheltersinacentrallocation,whileprovidingthemwitheducationandtrainingsothattheyeventuallyintegrateintothecity’slaborforce.However,ifinthesamecity,thirtypercentofthepopulationliveincardboardandplastichouses,thepolicysolutionwillrequireaverydifferentapproachinvolvinglookingcarefullyatdemandandsupplyforlandandhousing.Thesolutiontothehousingproblemwillrequireamarketintervention,evenifsomedemandsubsidiesarealsoused.Developingahousingpolicythereforenecessarilyrelateshousingaffordabilitydeficienciestothenumberofhouseholdswhosufferthesedeficiencies.ThisiswhatIproposeinthefollowingsection.WhetherlookingathousingaffordabilityissuesinMumbaiorNewYorkCity,thehouseholds’incomedistributioncurvewillbethefirstbuildingblockindevelopingasolution.Itisnecessarytoquantifytheproblemintermsoftheproportionofhouseholdsthatfallbelowthesociallyacceptableminimumstandard.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage19of21
Households’incomedistributionshouldthenberelatedtohousingconsumptionbyincome
range;animportantdimensionthatismissingfromthePIRindex.Itisimportanttolinkhousingpaymentswithwhathouseholdsgetfortheirmoney.Theobjectiveofahousingpolicyistoincreasethehousingconsumptionofhouseholdswhoconsumeanunacceptablylowstandardofhousingconsumptionduetotheirlowincome.Therefore,ahousingpolicyshouldneveraimtojustsupplyacertainnumberofhousingunitsperyeartofilla“backlog”ofsubstandardhousing.
AllthehouseholdsshownonthegraphofFigure8liveindwellingsthatcanbeobservedonthegroundorfromabovethroughsatelliteimageryoraerialphotography.Highresolutionsatelliteimagerycanbeanalyzedtoidentifyhousingtypesthatcanberankedbytheircostandphysicalcharacteristics.Theareaoccupiedbyeachhousingtypecanbemeasured.Theentireresidentialhousingsupplyofacitycanthenbedividedintohousingtypes.Censusdataandfieldsurveyscancomplementtheinformationobtainedbysatelliteimageinterpretation.Theentirepopulationofacitycanthenbedistributedamongdifferenthousingtypes.Eachhousingtypecorrespondstoahousingpriceorrentrangethatcanberelatedtoahouseholdincome.ItisthenpossibletomergethetypeofincomedistributiongraphshownonFigure8withthehousingtypologydatatoobtainarepresentationofthedistributionoftheentiresetofhouseholdsbyincomegroupandhousingtype.ThegraphofFigure10illustratesthismethod.ItshowsHanoi’shouseholdsincomedistributionontheleftsideandtheirhousingtypologysuperposedontheincomedistributionontherightside.Thegraphsshowswhattypeofhousingiscurrentlyaffordabletoeachincomegroup—astheyarecurrentlylivinginit—andthenumberofhouseholdsineachhousingtype.
Hanoi’spopulationhasbeendistributedamongeighthousingtypes.ThesehousingtypesarespecifictoHanoiandcaneasilybeidentifiedonasatelliteimagery.Thereisnostandardhousingtypologythatcanbeusedacrosscities;foreachcity,anewtypologyhastobedefined,reflectingthelocalhistoryandculture.InthecaseofHanoi,twohousingtypesarespecifictoVietnam—urbanizedvillagehousingand“tubehouses”.UrbanizedvillagehousingcorrespondstohousingunitsthatwereoriginallylocatedinvillagesthatwereonHanoi’speripherybuthavebeenabsorbedbythecity’sexpandingurbanfootprint.Thesevillagesretaintheiroriginalstreetlayoutandplotsizes.“Tubehouses”aretraditionalrowhouseswithafrontageofabout3.5metersandadepthof22meters.Theycansometimeshaveupto6or7floors.Theymaybeusedbyoneextendedfamilyorsubdividedintoapartmentsorevenrentedroombyroom.Theincomegroupsthatcanaffordtubehousecanthereforevarygreatlyfromneighborhoodstoneighborhoodsandovertime.
Figure10:Hanoiincomedistributionrelatedtotypology
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage20of21
Thechoiceofatypologyisimportantinanalyzinghousingpolicy.Thenumberofunitsofa
certainhousetypescanincreaseovertimewhileothersareboundtodecrease.Forinstance,thehousingstockconstitutedby“oldtubehouses”and“oldapartments”locatedinthehistoricalcoreofHanoicannotexpandandcouldonlyslowlydisappearthroughdemolitionandtransformationintoanothertypeofhousing,forinstance“newapartments,”whichwouldbeaffordabletoadifferent,usuallyhigher,incomegroup.
ThegraphsofFigure10showonlyasnapshotofthehousingsituationinHanoiatthedateofthesurvey.Theincomedistributionwillchangeandthehousingstockwillbetransformedbydemolition,reconstruction,andextensionintonewgreenfielddevelopment.Neighborhoodsrarelyremainstatic,theyaresubjectto“gentrification”oritsopposite“degentrification”.Ingeneral,whenincomesareincreasingrapidly,asinShanghaiinthe2000s,higherincomegroupstendtomovetowardnewlybuiltunits,whilelowerincomegroupsreplacethemintheolderunitstheypreviouslyoccupied.However,higherincomegroupsmayalsomovebackintorenovatedhousingunitsinolderneighborhoodswhentheseneighborhoodsareeitherwelllocatedorhaveanhistoricalcachet.
Forinstance,inBeijingthe“hutongneighborhoods”wereofteninhabitedbyhighandmiddleincomehouseholdsbeforetherevolution.TheyweresubdividedwhentheCommunistgovernmenttookoverin1947,resultingindensificationandsubsequent“degentrification”.Inthe80sthemunicipalgovernmentconsideredthedensehutongstobeslumsandbulldozedthemandreplacedthemwithhighriseapartments.Intheearly2000s,somehutongareasbecamepopularandweresubsequentlyre-gentrifiedintolow-densityonefamilycompoundsorintopriceyhotels.Thecyclebetweendegentrificationandre-gentrificationlastedabout50years.Mostcities’ancientneighborhoodshavesimilarstorieswithlongerorshortercyclesbetweengentrification,degentrificationandre-gentrification—forinstance,theWestVillageinNewYork,theMaraisinParisorSohoinLondon.
Themainlessontodrawfromtheconstanttransformationofhistoricalcitiesisthattheentirehousingstockmighttransform.Anaffordablehousingpolicyshouldthereforeprojectthelikelyhousingstockandflows.Thestockandflowapproachismoreusefulwhenappliedtoahousingtypology.Forinstance,inthecaseofHanoi,weknowthatthe“oldapartment”flowwillbebynecessitynegative,whilethevillasandnewapartmentsarelikelytohavepositiveflows.
Itisacommonmistaketolookonlyatasliceofthehousingmarket,suchaslow-incomeneighborhoodsandconcentrateonnewsupplythroughgreenfielddevelopment,whiletheentirehousingstockissubjecttotransformation.Inparticular,low-incomehouseholdsareusuallybetteroffmovingintoexistingcentrallylocatedneighborhoodsnewlyaffordabletothemthanmovingintonewlydeveloped“lowincomehousing”intheperipherywithlongandexpensivecommutingtrips.
Relatingincomedistributionwithhousingconsumption
Afterrelatinghouseholds’incometoahousingtypeitisnecessarytorelatehouseholds’incometoactualmeasuredhousingconsumption.Manyconsumptionindicatorscouldbeused:floorspaceperhousehold,landareaperhousehold,residentialutilityconsumptionlikewaterandelectricity,accesstotransportandcommunityfacilities.Wecouldalsouseacompositeindexthatreflectstheweightedaggregatehousingconsumptionofhouseholds,includingallofthecomponentsabove.
Whateverconsumptionmeasuresweselect,allhousingunitsaredistributedamonghouseholdsaccordingtotheirpricerank.Thispricewilltheoreticallybedirectlyrelatedtohouseholdsincome.Byrelatinghouseholdhousingconsumptiontoincomedistributionwecanidentifythegroupsthatareparticularlydeprivedanddevelopahousingpolicytoaddressthisdeprivation.
Hanoi’sincomedistributionisrelatedtofloorspaceconsumptionoffloorspaceinFigure11.
AB_Chapter_6_Affordability_2016_11_30JS2_ABFirstPart.DocxPage21of21
Figure11:Hanoi–Householdsincomedistributionandfloorconsumption
OnthegraphofFigure11IhavereproducedthehouseholddistributionofFigure10atthebottomofthefigureandIhaveaddedanewgraphabovewiththesamehorizontalaxiscorrespondingtohouseholdsincomebutwithaverticalaxisrepresentingfloorareaperhouseholdasitvarieswithincome.Thelineonthegraphisanaverageperincome;differenthouseholdsmayconsumedifferentquantityoffloorareasdependingontheircircumstancesandpreferencesandhomelocation.However,thesevariationsbetweenpeopleandincomegroupsaverageoutacrossincomegroups.Thisiswhytheconsumptioncanbeconvenientlyrepresentedbyacurveshowingtheaverageconsumptionperincomeintervalratherthanbyascatterplotincludingallthesurveyedcases.
Thetwographsshowhowmanyhouseholdsconsumehowmuchhousing.Mosthousingpoliciesinitialstepconsistsindefiningthesociallyacceptableminimumhousingconsumption,usuallyintermoffloorarea.Theuseofanincomedistributioncurvelinkedtoaveragehousingconsumptionperincomeintervalwouldallowonetoevaluatethenumberofhouseholdsthatarebelowasetconsumptionthreshold.Thepolicyandpossiblythethresholdcouldbeadjustedaccordingly.Theaveragefloorareaconsumptionperincomeintervalcouldalsobereplacedbyotherconsumptionindicatorslinkedtoincome,forinstancewaterconsumption,oranyotherindicator.
TherepresentationofthetwographsonFigure11isasimplificationofreality,butitisausefulonetounderstandanddiscusspolicyoptions,aswewillseeinthefollowingsections.