Cities in the EU: economic performance and resilience in the aftermath of the financial crisis

65 Pages·2016·1.84 MB·English

FAC-SIMILE THESIS 4-YEAR BACHELOR’S DEGREE AND MASTER’S DEGREE FRONTISPIECE
Department of Economics and Finance
Cities in the EU: economic performance and
resilience in the aftermath of the financial crisis
SUPERVISOR
Prof. Giovanna Vallanti
Valentin Vander Borght
Number: 677241
CO-SUPERVISOR
Prof. Salvatore Niticò
ACADEMIC YEAR 2015 2016
Acknowledgements
IwouldliketoexpressmygratitudetomysupervisorsProfessorGillesVanHammeandProfessor
Giovanna Vallanti. At first, thank you for your patience and flexibility that allow me to do this
master thesis at Universite´ Libre de Bruxelles and Libera Universita` Internazionale degli Studi
Sociali. Thankstoyou,Ihaveenjoyedthisgreatexperienceofdoubledegree. Then,Iwouldliketo
thankyouforbeingalwaysavailableforanyquestions. Finally,Iwouldliketothankyouforyour
guidanceandexpertisethatallowmetodothismasterthesis.
Inaddition,thismasterthesisisalsothefulfillmentoffivefantasticyears. Iwouldliketothank
myfriendsandtheAnalysisteamforhelpingmetoaccomplishmydegree.
Lastbutnotleast,Iwouldliketothankmyparentfortheirunconditionalsupportduringmy
studies and my girlfriend for her encouragement. I would not have been able to complete this
thesiswithoutthem.
Cities in the EU: economic performance and resilience in
the aftermath of the financial crisis
MasterThesis
Valentin Vander Borght
2015–2016
Abstract
ThecityisdefinedasafunctionalurbanareabasedonEurostatdataatNUTS3level. Ifirst
regresstheeconomicperformanceofEuropeancitiesoneconomicstructure,includinghuman
capital, andsize-basedtypologyacrosstheperiod2003–2013usingfixedeffectspaneldata.
Atfirst,IfindevidenceofconvergencebetweenEuropeancitiesoverthestudiedperiod. But,
thedivergenceoccursduringthecrisis. Secondly,Ifindthatmetropolitancitiesoutperform
thesmallestcities. Thirdly,constructionsectorandhumancapitalaretheengineofeconomic
growthduringthestudiedperiod. IthenregresstheeconomicperformanceofEuropeancities
duringthecrisisandpost-crisisperiodoninitialeconomicstructure,includinghumancapital,
andsize-basedtypologyusingrobustOLSmodel. Giventheirinitialfeatures,manufacturing,
administrative,financialandadvancedserviceshaveanegativeimpactonthecrisiseconomic
growth. Besides,thenegativeeffectofmanufacturingandadministrationpersistduringthe
post-crisis. Onlyhumancapitalhasapositiveimpactonbothcrisisandpost-crisiseconomic
growth.
1
Contents
1 Introduction 13
2 LiteratureReview 15
2.1 Triumphofmetropolitancities? Atheoreticalperspective . . . . . . . . . . . . . . . 15
2.1.1 Agglomerationeconomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.2 Networkparadigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Fromthebottomtothetop: doEuropeancitiesconvergeacrosstime? . . . . . . . . 17
2.3 Resilience: fromdefinitiontoevidence . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Data 21
3.1 Citiesasfunctionalurbanareas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 Citiestypology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4 Descriptivestatistics 25
4.1 Theconvergenceofcitiesacross2003–2013 . . . . . . . . . . . . . . . . . . . . . . . 26
4.2 Relationshipbetweentypologyandeconomicperformance . . . . . . . . . . . . . . 27
4.3 Resilience: afirstdescriptiveassessment . . . . . . . . . . . . . . . . . . . . . . . . . 28
5 Modelspecifications 33
5.1 AnalysisofeconomicperformanceofEuropeancitiesovertheperiod2003–2013 . . 33
5.2 AnalysisofEuropeancitiesresilience . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6 Results 38
6.1 EconomicperformanceofEuropeancities2003–2013 . . . . . . . . . . . . . . . . . 38
6.2 Citiesresilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7 Conclusion 44
2
List of Tables
1 DescriptionvariablesdatasetatNUTS3level . . . . . . . . . . . . . . . . . . . . . . 22
2 Citiestypology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3 AverageGDPgrowthratesbycitygroupings(in%) . . . . . . . . . . . . . . . . . . . 26
4 AverageGDPgrowthratesbytypology(in%) . . . . . . . . . . . . . . . . . . . . . . 27
5 Resultsoftheeconometricmodelusingthefixedeffectspaneldatamodel . . . . . . 39
6 Coefficientsignofinteractionbetweeneconomicstructureandtypologyresulting
fromtheequation(5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7 ResultsoftheeconometricmodelusingtherobustOLSmodel . . . . . . . . . . . . 43
8 Mainfunctionalurbanareas(numberofFUApercountry) . . . . . . . . . . . . . . . 50
9 Summarystatisticsvariables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
10 RelationshipbetweenaverageGDPgrowth2003–2007andtypology . . . . . . . . . 52
11 RelationshipbetweenaverageGDPgrowth2008–2010andtypology . . . . . . . . . 53
12 RelationshipbetweenaverageGDPgrowth2011–2013andtypology . . . . . . . . . 54
13 Average gross value added per sector growth rates by period (in %): financial
andadvancedservices(1),manufacturing(2),construction(3),administration(4),
agriculture(5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3
List of Figures
1 Relationshipbetweentheaveragepre-crisisshareoffinancialandadvancedservices
andtheaveragecrisis/post-crisisGDPgrowth . . . . . . . . . . . . . . . . . . . . . . 29
2 Relationshipbetweentheaveragepre-crisisshareofmanufacturingandtheaverage
crisis/post-crisisGDPgrowth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3 Relationshipbetweentheaveragepre-crisisshareofconstructionandtheaverage
crisis/post-crisisGDPgrowth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4 Relationship between the average pre-crisis share of agriculture and the average
crisis/post-crisisGDPgrowth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5 Relationshipbetweentheaveragepre-crisisshareofadministrationandtheaverage
crisis/post-crisisGDPgrowth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
6 EU-28GDPinpurchasingpowerstandardsperinhabitantatNUTS3level–2008.. 47
7 EUtypologyontechnologicallyadvancedregionsatNUTS2level–2007. . . . . . . 48
8 EUpopulationdensityclassificationatNUTS3level–2007. . . . . . . . . . . . . . 49
4
Thesis summary
Citiesareconsideredasthemajordriverforterritorialcohesionandeconomicgrowthaccording
to European policy. In line with this perspective, my thesis aims at analyzing the impact of the
financialcrisisonEuropeancities. Ontheonehand,Iregresstheeconomicperformanceofcities
ontheeconomicstructure,includinghumancapital,andpopulationsizebasedtypologyoverthe
period2003–2013. Ontheotherhand,Itrytoidentifythesourceofresiliencebyanalyzingthe
impactofinitialeconomicstructure,includinghumancapital,ofcitiesandpopulationsizebased
typologyonthecrisisandpost-crisiseconomicgrowth. Insummary,theobjectivesofthisstudy
areoutlinedasfollows:
1. IsthereaconvergencebetweenEuropeancities? Inwhichextentthefinancialcrisisalters
thistrend?
2. Whatisthestrongestcityintermsofeconomicperformanceovertheperiod2003–2013?
3. Whatarethedrivingfactorsofresilienceintheaftermathofthefinancialcrisis?
I develop three strands of literature that I deem relevant to answer to these questions: ag-
glomerationeconomicsandnetworkparadigm,convergencetheoriesandtheconceptofresilience.
Accordingtoagglomerationeconomicsandnetworkparadigm,ithasbeenarguedthatmetropoli-
tancitiesbenefitfromtheirpositionofcentralnodesintheworldeconomyandtheavailability
of a diversified labor pool. Globalization and the development of ICT accentuate this trend by
fostering the concentration of highly value added sector in metropolitan areas. Nevertheless,
negativeexternalitiesarisingfromcongestion,suchaspollutionandincreasingcommutingtime,
arelikelytocounterbalancethisphenomenoninmetropolitanareas. Tocapturetheimportanceof
agglomerationandnetworkeffect,Iusetheclassificationofcitybasedonpopulationsizedescribed
intable1. Secondly,Idrawadistinctionbetweentwotheoriesconcerningconvergence. Onthe
onehand,theclassicalconvergencetheoryexplainsthecatchingupprocessofpoorregionsasa
result of a differential marginal productivity between labor and capital intensive cities. On the
otherhand,thecumulativecausationtheoryexplainswhyregionsdoesnotfollowthepathrecog-
nizedbytheclassicsandhighlightstheimportanceoflong-termstructureinordertounderstand
regionaldivergence. Inotherwords,initialdisparitiesreproduce,ornot,spatialinequalitydue
to the cumulative consequences of the regional situation. To assess the extent of a convergence
betweenEuropeancities,Idefineconvergenceasaβ-convergence,thatis,thenegativecorrelation
5
betweentheeconomicgrowthandtheinitiallevelofincome. Thirdly,severalfactorsofresilience
areemphasizedintheeconomicliterature. Itcanbedividedintodifferentstrands. Atfirst, the
importanceofinstitutions,cultureandpolitical,toensuretheresilience. Then,theimportanceof
smallandmediumsizedcompaniesandacreativeclassintheeconomytodealwithanegative
shock. Finally,someauthorspointouttheimpactofspecializationinsomeactivitiesasashield
for the economy. I define resilience as a two-step process, that is, the role of initial features of
citiesthatmitigatethenegativeshockduringthecrisisperiodandfostertheadaptationduringthe
post-crisisperiod.
Table1: Citiestypology
Typology Criteria
Metropolitanarea FUApopulation>500000inhabitants
PolyFUAs 2metropolitanareaswiththeircenters<60kmapartand
laborbasinstouchingeachother;
2 large areas with their centers < 30 km apart and labor
basinstouchingeachother;
1metropolitanand1large/mediumareawiththeircenters
<30kmapartandlaborbasinstouchingeachother;
2metropolitanareaswiththeircenters<60kmapartand
laborbasinsseparatedonlybythelaborbasinofasmaller
FUAtouchingthebothofthem.
Largearea FUApopulation>250000inhabitants
Mediumarea FUApopulation>100000inhabitants
Smallarea FUApopulation>50000inhabitants
Otherarea FUApopulation<50000inhabitants
Mydatabaseconsistsonapaneldatacomposingof1515unitsofobservation,functionalurban
areasconsideredascities,comingfrom26countriesovertheperiod2003–2013. Iconstructthe
functional urban areas based on Eurostat data at NUTS 3 level. By constructing the functional
urban areas, I define the concept of city. A city is not only an administratively-delineated area
butaplacecharacterizedbyalaborpool. Inotherwords,acityiscomposedbythecityitselfand
the share of surrounding agglomeration which economically contributes to the city. The extent
ofalaborpoolassociatedwithaNUTS3unitisdeterminedbyacoefficientbasedoncommuting
statistics. Inthisanalysis,thecoefficientsareconsideredasgivenbymysupervisor,ProfessorG.
6
VanHamme. ThiscoefficientrepresentsthepercentagegivingwhatpartoftheNUTS3variable,
such as gross domestic product, is associated with a single functional urban area. The sum of
each adjusted NUTS 3 value corresponding to a functional urban area gives the variable value
ofthisfunctionalurbanarea. Todefinetheeconomicstructureofacity,Ibringtogetherasetof
indicatorthatproxytheimportanceoftheprimarysector,manufacturing,construction,financial
andadvancedservices,administrativesectorsandhumancapital. Thesevariablesaredescribed
in table 2. In addition, it is important to note that I use a restricted sample when I include the
variablerelatedtohumancapitalbecauseIhaveonlythedataforthemostimportantEuropean
cities. For this reason, I do interpret the result of the other variables when I use human capital
variable.
Table2: DescriptionvariablesdatasetatNUTS3level
Indicator Explanation Period Source
Economicgrowth BasedonGDPatcurrentmarketpricepur- 2003-2013 Eurostat
chasingpowerstandardinmillion€. No
dataofGDPatbasicpriceonEurostat.
Population Criteria used for the typology. I assume 2014 Eurostat
topology is constant for the studied pe-
riod.
Agriculture GrossvalueaddedatbasicpricesinAEu- 2003-2013 Eurostat
rostatactivityinmillion€. Itcorresponds
totheprimarysector.
Manufacturing GrossvalueaddedatbasicpricesinCEu- 2003-2013 Eurostat
rostatactivityinmillion€. Exceptionfor
PolandwhereIuseB-EEurostatcategories
inmillion€.
Construction GrossvalueaddedatbasicpricesinFEu- 2003-2013 Eurostat
rostatactivityinmillion€
Finance and ad- GrossvalueaddedatbasicpricesinK-N 2003-2013 Eurostat
vancedservices Eurostatactivityinmillion€. Exception
forUKwhereIuseonlyKEurostatactiv-
ityinmillion€.
Administration GrossvalueaddedatbasicpricesinO-U 2003-2013 Eurostat
Eurostatactivityinmillion€. Exception
forUKwhereIuseonlyO-QEurostatac-
tivityinmillion€.
Education Shareoftertiarydiplomaintheactivepop- 2001 Eurostat; Labour
ulation. I assume the share is constant ForceSurvey
throughtime.
Inthisthesis,Iusetwoempiricalapproaches. Atfirst,thespecificationisdesignedtoemphasize
7
theeconomicperformanceofcitiesacross2003–2013usingfixedeffectspaneldata. Takinginto
accountthedependentvariables,thespecificationofthemodelisasfollows:
GDPgrowthit =α+β1ln(GDPit−1)+β2ln(GDPit−1)×crisist+β3EcoStructurej +β4EcoStructurej ×crisist
it it
+β EcoStructure ×typology +β EcoStructure ×typology ×crisis +β education +(cid:15) (1)
5 j i 6 j i t 7 i it
it it
where(cid:15) =typology +λ +φ +u ;
it i t c it

 ji ==11,,......,,nk wwhheerree nk==5151n5umnbuemrboferecoofncoitmieiscstructurevariables
where
 ctr=is2is00=31,..w.,h20en13t=T20=081,1200n9u,m20b1e0rof;yeacrrsisis=0otherwise
The methodology I use to estimate this equation is the fixed effects panel data regression. This
modelallowstocontrolforomittedvariablesthatvaryeitheracrosstimebutdonotchangeacross
country/typologyoracrosscountry/typologybutdonotchangeovertime. Therefore,theerror
termcanbedecomposedintoacountryfixedeffect,φ ,atypologyfixedeffect,typology ,atime
c i
fixedeffect,λ ,andaresidualerrorterm,u . Giventhenatureofmydata,thismodelisthemost
t it
coherent. That is why I put aside the test for random effect model or pooled model. My first
specificationallowsmetodrawseveralinterestingresults.
Fromageneralpointofview,myanalysisshowsthatmetropolitancitiesperformsbetterthan
large, medium, small and other cities over the period 2003–2013. All things being equal, the
smallerthepopulationsize,theloweristheeconomicgrowthinthecity. Thisresultconfirmsthe
importanceoftheeffectsofagglomerationandnetworkthatcharacterizemetropolitanareas.
Then, I test the extent of β-convergence process between cities and I check the impact of
the financial crisis on this convergence process. I find a consistent result through the different
specifications. Thecoefficientassociatedwiththeinitiallevelofgrossdomesticproductisnegative
andsignificantatlevel1percent. Therefore,thisresultadvocatesthehypothesisofconvergence
betweencitiesbecauseacitywithhigherlevelofgrossdomesticproductexperiencelowereconomic
growth. Further,Itesthowthecrisishasaffectedtheconvergenceprocessbyaddinganinteraction
variablebetweentheinitiallevelofgrossdomesticproductandtheperiod2008–2010. Thisresult
bringsamorenuancedpointofviewonconvergence. Duringthefinancialcrisis,divergenceoccurs
between European cities. The estimate of convergence during the crisis period can be defined
8