Evaluating cities' vitality and identifying ghost cities in China with

emerging geographical data

YingLong,AssociateProfessorSchoolofArchitecture,TsinghuaUniversity,China

November14,2016

IncollaborationwithXiaobin Jin,WeiSun,Yuyin Lu,Xuhong Yang

RapidurbanexpansioninChinaduringlastdecades(around30%residentialdevelopments)

Search “Ghost Cities/鬼城” at Baidu

The existing rankings of ghost cities in China

Housing vacancy rate has been a national secret in each city.

TheBeijingCityLab(BCL)isaresearchnetwork,dedicatedtostudying,butnotlimitedto, China’scapitalBeijing.TheLabfocusesonemployinginterdisciplinarymethodstoquantifyurbandynamics,generatingnewinsightsforurbanplanningandgovernance,andultimatelyproducingthescienceofcitiesrequiredforsustainableurbandevelopment.Thelab'scurrentmixofplanners,architects,geographers,economists,andpolicyanalystslendsuniqueresearchstrength.

http://www.beijingcitylab.com

MethodologyIdentifyingghostcitiesthroughthelensofthevitalityofresidentialdevelopments

• KevinLynchbelievesthattheprimarycriterioninthequalityassessmentofurbanspaceformisthevitality,whichisdefinedasasettlement(thedimensionurbanmorphology)thatsupportsthevitalfunctions(thedimensionurbanfunction)andthebiologicalrequirementsandcapabilitiesofhumanbeings(thedimensionurbansociety),and howtoprotectthecontinuationofthespecies(Lynch,1984).

Three components of urban vitality

Urbanform Urbanfunction Urbanactivity

Urbanvitality

• Urbanform=Roadjunctiondensity(J)• Urbanfunction=Pointsofinterest(POIs)density(P)• Urbanactivity=LBSdensity(L)• Urbanvitality(V)= ln (J*P*L)

Calculating urban vitality for each place

Roadjunction Pointsofinterest LBS

Urbanvitality

Ghost cities are always associated with new developments

No urban fabric/form on urbanized/developed areas

No urban function on existing urban form

No urban activities in functioned urban areas

• G:Thedegreeonacitybeingaghostcity

• Thedegreetowhichacitybelongstoaghostcityisrelatedtovitalityinnewurbanareasandthedifferenceinvitalitybetweenoldandnewurbanareas.Thelargerthedifferenceandthelowerthevitalityinthenewresidentialprojects,themorelikelythecitytobeaghostcity.

Identifying ghost cities via aggregating the vitality of residential projects

G = 1 / (Vnew/Vold * Vnew)

Theaveragevitalityofresidentialdevelopmentsinnewurbanareas(Vnew)andresidentialdevelopmentsinoldurbanareas(Vold)canbecalculatedforeachcitybyoverlayingresidentialtransactionswiththeadministrativeboundariesofcities.

Data used for this study

• Thereare653Chinesecitiesin2014.• OnthebasisoftheChineseadministrativesystem,therearemainlyfivelevelsofcitiesclassifiedinthisway,including:• municipalities(MD)directlyledbythenation(with4cities,tier1)

• sub-provincialcities(SPC)(with15cities,tier2)

• otherprovincialcities(OPCC)(with17cities,tier3)

• prefecture-levelcities(PLC)(with250cities,tier4)

• county-levelcities(CLC)(with367cities,tier5)

The Chinese city system

AdministrativeareasofChinesecities

• Weassumethattheformationofghostcitiesisgenerallyduetothenewlydevelopedareas,thusmakingitnecessarytodefinetherelevantboundaries.

Urban areas in 2000 for differentiating new and old developments

Theurbanareasareinterpretedfromremotesensingimages,andtheoverallaccuracyis94.3%.Thereare9,128patcheswithin33,148km2 intotal,andthemeanpatchsizeis3.6km2.

Residential projects as the core data for identifying ghost cities

Availableatthegovernmentalwebsite

• Intotal,thereare535,523residentialprojectswithinatotalareaof7770.2km2from2002-2013• EachprojectisstoredasapointinGIS

Residential projects as the core data for identifying ghost cities

Year CountsumType

AreasuminhaType

Market# Social# Marketinha Socialinha

2002 2,576 2,182 394 4,042.1 3,726.4 315.7

2003 3,194 2,887 307 6,149.2 5,715.5 433.7

2004 10,840 3,965 6,875 13,839.8 11,641.6 2,198.2

2005 7,321 2,933 4,388 10,435.4 8,755.3 1,680.1

2006 7,880 3,897 3,983 10,270.5 9,232.9 1,037.7

2007 48,910 32,924 15,986 65,496.5 54,676.9 10,819.6

2008 43,545 29,682 13,863 60,112.3 48,901.4 11,210.9

2009 73,678 54,849 18,829 85,472.7 71,338.1 14,134.6

2010 87,017 78,234 8,783 127,848.2 109,779.3 18,068.9

2011 86,999 77,888 9,111 133,875.5 108,372.4 25,503.1

2012 72,618 62,861 9,757 116,444.5 87,714.5 28,730.0

2013 90,945 87,141 3,804 143,030.6 136,326.1 6,704.5

Sum 535,523 439,443 96,080 777,017.2 656,180.2 120,837.0

Road junctions in 2014

8.24millionroadjunctionsforthewholeChina

Points of interest in 2014

10.6millionPOIsforthewholeChina

• Wehavegathered1km*1kmLBSdatafromBaidu,oneofthelargestinternetcompaniesinChina,andthedatacoverawholeweekinJulyof2015• Consideringthedatacoverofabout800millionusersinChina,thedatacanwellreflecthumanactivitiesinChina

LBS data in 2015

LBSdensityof asmallareainJiangsuprovince

Results

Urban form dimension

Urban function dimension

Urban activity dimension

Urban vitality reflecting the previous three components

We find the average vitality of residential projects in new

urban areas is only 8.8% of that in old urban

areas, denoting the potential existence of ghost cities in newly developed areas in Chinese cities.

• 1MD=citiesdirectlyledbythenation(with4cities,tier1),2SPC=sub-provincialcities(with15cities,tier2),3OPCC =otherprovincialcities(with17cities,tier3),4PLC=prefecture-levelcities(with250cities,tier4),and5CLC=county-levelcities,and6OthertownsmeansthattheprojectsarebeyondtheadministrativeboundariesofcitiesinChina

The urban vitality and its components by each level of cities

• TheresultsindicatethatthecitiesinthemiddleregionofChinahavethegreatestvitalityvaluesandthoseinnortheasternChinaandwesternChinaarewithlowervalues.

The average vitality of residential projects (RPs) in new urban areas for all Chinese cities

Variable Vitality Junctions POIs LBSsURBAN 0.126 0.352 0.307 0.286YEAR 0.008 0.003 0.040 0.015

CENTER -0.025 -0.103 -0.056 -0.089LEVEL -0.132 -0.353 -0.057 -0.336AREA -0.041 -0.051 -0.082 -0.064

AdjustedR2 0.045 0.353 0.123 0.274

The regression model for residential project vitality suggests that residential projects in old urban areas, being close to the city center in an administratively higher tier city, with a smaller development size and more recent development, positively contribute to a higher level of vitality.

Note: All variables are significant at the 0.05 level. All coefficients have been normalized in the linear regression process. When we usethe logarithm value of vitality as the dependent variable, the adjusted R2 and the normalized coefficients do not change much.

The ghost index (G) for all Chinese cities

Thirty ghost cities are identified accordingto the G indicator evaluation results, andmost of these cities are distributing innorth eastern China, particularly inShandong and Anhui provinces.

We have also benchmarked our identified ghost cities against existing rankings, the Baidu search engine, and night-time light images.

Although we admit that ghost cities may exist in the particular urbanizing phase in China, and some cities that are ghost cities now may be well developed in future, this study provides a thorough evaluation on the ghost city condition in China.

Our profiling results illustrate a big picture for Chinese past residential developments and then the ghost cities.

This may shed light on policy implications for Chinese urban development.

• Borrowthetheoryofurbanvitalitytoidentifyandevaluateghostcities• Threecomponents:form,functionandactivity• LBSshouldnotbetheonlydatafortheghostcitystudy

• Evaluateeachcityinabottom-upmannerbyusingubiquitousandincreasinglyavailablebig/opendatafromofficialwebsites,commercialinternetcompaniesandsocialnetworks• Both“ghost”leveland“ghost”sizeinourstudy

• Developtheoreticalmodelstoexplainthecontributingfactorsonvitalityattheresidentialprojectlevel• ProvidingthepossibilityforproposingnecessarypoliciestofightagainstghostcitiesinmodernChina

• TheproposedghostindicatorG considersnotonlythevitalityinnewurbanareas,butalsothegapbetweennewandoldurbanareas• Makingitpossibletoovercomethegapofurbanform,functionandinformationcommunicationtechnology(ICT)penetrationacrossvariouscities

Highlights of this study

• Theaccesstotherealboundaryofeachresidentialcommunitymayleadtomoreobjectiveevaluationandidentificationresults.• Theinformalresidentialdevelopmentsalsocontributetotheformationofghostcities.Thisshouldbeaddressedinournextstudyaswell.

Potential biases of this study

• Batty,M.(2016).Emptybuildings,shrinkingcitiesandghosttowns.EnvironmentandPlanningB:PlanningandDesign,43,3-6.

• Chen,Q.(2014).Willthenumberofghostcitiescontinueincreasing?. http://www.zhihu.com/question/26193868/answer/33116107

• Chi,G.,Liu,Y.,&Wu,H.(2015).GhostCitiesAnalysisBasedonPositioningDatainChina.arXiv preprintarXiv:1510.08505.

• Doll,C.N.H.,Muller,J.-P.,&Morley,J.G.(2006).Mappingregionaleconomicactivityfromnight-timelightsatelliteimagery.EcologicalEconomics,57,75-92.

• Elvidge,C.D.,Baugh,K.E.,Dietz,J.B.,Bland,T.,Sutton,P.C.,&Kroehl,H.W.(1999).RadianceCalibrationofDMSP-OLSLow-LightImagingDataofHumanSettlements.RemoteSensingofEnvironment,68,77-88.

• Jacobs,J.(1961).TheDeathandLifeofGreatAmericanCities.NewYork:Randomhouse.

• Lynch,K.(1984).GoodCityForm.Cambridge:TheMITPress.

• Kuang,W.,Liu,J.,Dong,J.,Chi,W.,&Zhang,C.(2016).TherapidandmassiveurbanandindustriallandexpansionsinChinabetween1990and2010:ACLUD-basedanalysisoftheirtrajectories,patterns,anddrivers.LandscapeandUrbanPlanning,145,21-33.

• Liu,X.,&Long,Y.(2016).AutomatedidentificationandcharacterizationofparcelswithOpenStreetMap andpointsofinterest.EnvironmentandPlanningB:PlanningandDesign,43,341-360.

• Liu,Z.,He,C.,Zhang,Q.,Huang,Q.,&Yang,Y.(2012).ExtractingthedynamicsofurbanexpansioninChinausingDMSP-OLSnighttime lightdatafrom1992to2008.LandscapeandUrbanPlanning,106,62-72.

• Long,Y.(2016).RedefiningChinesecitysystemwithopendata.BeijingCityLab.

• Long,Y.,Gu,Y.,&Han,H.(2013).Spatiotemporalheterogeneityofurbanplanningimplementationeffectiveness:Evidencefromfiveurban masterplansofBeijing.LandscapeandUrbanPlanning,108,103-111.

• Long,Y.,Han,H.,Tu,Y.,&Shu,X.(2015).Evaluatingtheeffectivenessofurbangrowthboundariesusinghumanmobilityandactivityrecords.Cities,46,76-84.

• Long,Y.,&Wu,K.(2016).ShrinkingcitiesinarapidlyurbanizingChina.EnvironmentandPlanningA,48,220-222.

• Ma,T.,Zhou,Y.,Zhou,C.,Haynie,S.,Pei,T.,&Xu,T.(2015).Night-timelightderivedestimationofspatio-temporalcharacteristicsofurbanizationdynamicsusingDMSP/OLSsatellitedata.RemoteSensingofEnvironment,158,453-464.

• Montgomery,M.R.(2008).TheUrbanTransformationoftheDevelopingWorld.Science,319,761-764.

• Nie,X.-y.,&Liu,X.-j.(2013).Typesof“GhostTowns”intheProcessofUrbanizationandCountermeasures.JournalofNantongUniversity(SocialSciences),29,111-117. (inChinese)

• O'Callaghan,C.,Boyle,M.,&Kitchin,R.(2014).Post-politics,crisis,andIreland's‘ghostestates’.PoliticalGeography,42,121-133.

• Shepard,W.(2015).GhostCitiesofChina:TheStoryofCitieswithoutPeopleintheWorld'sMostPopulatedCountry.ZedBooks.

• Su,X.(2014).Rankofghostcitiesin2014. http://house.ifeng.com/detail/2014_10_12/50060123_0.shtml

• Tian,G.,Liu,J.,Xie,Y.,Yang,Z.,Zhuang,D.,&Niu,Z.(2005).Analysisofspatio-temporaldynamicpatternanddrivingforcesofurbanlandinChinain1990susingTMimagesandGIS.Cities,22,400-410.

• Wang,L.,LI,C.,Ying,Q.,Cheng,X.,Wang,X.,Li,X.,Hu,L.,Liang,L.,Yu,L.,Huang,H.,&Gong,P.(2012).China’surbanexpansionfrom1990to2010determinedwithsatelliteremotesensing.ChinSciBull,57,2802-2812

• Xiao,P.,Wang,X.,Feng,X.,Zhang,X.,&Yang,Y.(2014).DetectingChina’sUrbanExpansionOverthePastThreeDecadesUsingNighttime LightData.IEEEJournalofSelectedTopicsinAppliedEarthObservationandRemoteSensing,7,4095-4106.

• Zijderveld,A.C.(1998).ATheoryofUrbanity:TheEconomicandCivicCultureofCities (TransactionPublishers,NewBrunswick,NJ)

References

Thanks

ylong@tsinghua.edu.cn

ThisworkissupportedbygrantsfromtheNationalNaturalScienceFoundationofChina(No.41340016and51408039).TheauthorswouldliketothankMr/MsXuefeng HuangandJinyuan Xie fortheirassistanceon

dataprocessing.OurthanksarealsogiventoMsTraceyTaylorforeditingthelanguageofthispaper.