You are on page 1of 43

Chapter4

CitySize

McGraw-Hill/Irwin Copyright 2012 by The McGraw-Hill Companies, Inc. All rights reserved.
WhyDoCitiesVaryinSizeandScope?
WhileNYChasapopulationofmorethan18million,the
smallest urbanareaintheU.S.hasonly13,000.

Awellobservedregularitywithinagivenregion:
Thereareafewlargecities,
amoderatenumberofmediumcities,
andalargenumberofsmallcities!

4-2
WhyDoCitiesVaryinSizeandScope?

Urban areas inSpain


Greaterthan4million 2
1millionto4million 2
500.000to1million 8
100.000to500.000 41

4-3
CitiesinEurope
A Principal Metropolises

B Regional Centres

C Smaller Centres

D Towns & Cities of the


Lagging Regions

Urban Audit (EU)

Urban Audit (non-EU)

Size of circle is relative to population


in core city* in 2004
10,000,000
1,000,000
500,000
100,000

*Paris: Kernel

4-4
What arethe economic forces behind the development of
cities ofdifferent size?

4-5
UtilityandCitySize
Agglomerationeconomies(localizationandurbanization
economies) causefirmstocluster,increaseproductivity,
andraise wages.
Asacitygrows,thebenefitsofhigherwagesare atleast
partially offsetbyseveraldisamenitieslike:
Greaterdensities
Congestionandpollution
Longercommutingtimes decreasingleisuretime

4-6
UtilityandCitySize
Howdoesanincreaseincitysizeaffecttheutilityofatypicalworker?

Wagesreflectlaborproductivityandarehigherinlargercities.

Thecostsofthecitysizeareassociatedwithanincreasingrateof
commutingcosts (simplifiedmodel).

4-7
Thereisanutilityoptimum
regardinggrowingcities.

Aslongasagglomeration
economiesarestrongerthan
diseconomiesofscale
(commutingcost),utility
increases.

Whenagglomeration
economiesareweakerthan
thediseconomiesfrom
commuting,utilitydecreases.

2012TheMcGrawHillCompanies.AllRightsReserved
4-8 8
LocationalEquilibriumWithinaCity
Wenowconsiderlocationdecisionsoftypicalworkers
withinacity(2million inhabitants).
Workers commute from different residential locations to
the citys production centre.
Workers differ intheir commuting costs.
1staxiom:Prices adjust toachieve locational equilibrium

Landrentswilladjusttomakeallworkersindifferent
amongtheirresidential location.

4-9
LocationalEquilibriumWithinaCity

Differencesincommutingcostsgeneratedifferencesinlandrent.
Differencesincommutecostoffsetbydifferencesinlandrent(C).
Whogetstherentfromland?
Equalsharesoflandrent,averaging$15(E).
Utility=Laborincome+rentalincome commutecost rentpaid

4-10
ASystemofCities
Howisaregionsworkforcedistributedamongitscities?
Aretheremanysmallcitiesorafewbigcities?
Assumptions:
Regionwith6millionworkersandthreepossible
configurations
Sixcities,eachwith1millionworkers

Threecities,eachwith2millionworkers

Twocities,eachwith3millionworkers

4-11
Figure 4-2 Cities May Be Too Large, but not too small

Findingtheequilibrium(we
startwithsixcities)
Isthisastableequilibrium?

2012TheMcGrawHillCompanies.AllRightsReserved
4-12
12
Figure 4-2 Cities May Be Too Large, but not too small

Findingtheequilibrium(we
startwithsixcities)
Isthisastableequilibrium?

Selfreinforcing effects generate


extremeoutcomes.

Alongthepositivelysloped
portionoftheutilitycurve,
changesinpopulationareself
reinforcing.

Utilitygapopens.

2012TheMcGrawHillCompanies.AllRightsReserved
4-13
13
Figure 4-2 Cities May Be Too Large, but not too small

Wenowstartwithtwolarge
cities(3millioneach).

Is this astable equilibrium?

Again,migrationopensupthe
utilitygap.
Alongthenegativelysloped
portionoftheutilitycurve,
migrationisselfcorrecting.
Migrationwillbereversed!

2012TheMcGrawHillCompanies.AllRightsReserved
4-14
14
Figure 4-2 Cities May Be Too Large, but not too small

Whyisthesituationwithsmall
citiesunstableandthesituation
withlargecitiesstable?
Smallcities:positivelysloped
utilitycurve.
Agglomerationeconomiesare
strongerthandiseconomies.
Utilityofmigrantsincreases.
Utilityofthosethatstaydecreases.
Largecities:negativelysloped
utilitycurve.
Agglomerationeconomiesare
weakerthandiseconomies.
Utilityofmigrantsdecreases.
Utilityofthosethatstayincreases.
Inecientlylargeci esmay
persist.

2012TheMcGrawHillCompanies.AllRightsReserved
4-15
15
SpecializedandDiverseCities
Docitiesspecializeinanarrowsetofeconomicactivitiesor
dotheycreate diverse environments?
WefindBOTHtypesofcities:Thetypicalregioncontainsa
widevarietyofcities,fromhighlyspecializedtohighly
diverseones(Henderson,1988).

Specializationhappensbecauseoflocalizationeconomies
Diversificationhappensbecauseofurbanizationeconomies

Thecitytypesarecomplementaryservingdifferentfunctions
inamarketeconomy.

4-16
SpecializedandDiverseCities
Twotypesofcitiesarecomplementary.
Duranton andPuga(2001) Nursery Cities:Urban diversity,
process innovation,andthe life cycle ofproducts',
AmericanEconomic Review.
Manyfirmsstartindiversecity,whichfosternewideas.

Maturingfirmsrelocatetospecializedcitiestoexploitlocalization
economies.

Where should firms startupthere business (searching for a


prototype andthe best production process for anew
product)?

4-17
AModelofLaboratoryCities
(Duranton andPuga,2001)

Setting1: Afirmexperimentsinadiversecityandlater
movestoaspecializedarea.
Assumptions:
Sixpotentialproductionprocesses
Firmobservesotherfirmsindiversecitytofindidealproduction
process(betterlearningopportunities)
Prototypeisfoundandproducedbyimitation
Discoveryofidealprocesstakesanaverageofthreeyears
Afterthediscovery,thefirmwillmovetoaspecializedcity switch
tomassproductionandstartmakingprofit

4-18
AModelofLaboratoryCities
(Duranton andPuga,2001)

Setting2:Afirmexperimentsinaspecializedarea.
Assumptions:
Therearetradeoffsresponsibleforthedecisionsonwheretolocate:
Good lowerprototypecosts:Eachcityhasspecializedinputsforthe
productionprocess,soinputswillbecheaper.
Bad highermovingcosts:Searchforidealprocessrequiresmoving
around.Onaverage,afirmwillneedthreemoves.

Ifmovingcostsarelargerelativetothesavingsin
prototypecosts,profitwillbelower,whenthefirm
experimentsinspecializedcities.
Startinginthediversecity=1move
Startinginthespecializedcity=3movesonaverage

4-19
Example:TheRadioIndustryinNew York

Vernon(1972):Earlyfirmsweresmall,numerous,agile,
nervous,andheavilyreliantonsubcontractors.
NYCprovidedawidevarietyofintermediateinputsand
workers.
Oncetechnologysettled,productionbecamestandardized,
Producersgainlessfromdiversifiedcities

firmsrelocatedtoeconomizeonlaborcost,landrent,localisation
economies (lower production costs).

4-20
SpecializedandDiverseCities
Empirical evidence for France(Duranton andPuga(2001):
Frenchfirms:7of10relocationsfromdiversetospecializedcity.
Firmsmoveastheymature.
Mostinnovativefirmshavehighestfrequencyofmovesfrom
diversetospecialized.
pharmaceuticals andcosmetics,business services,printing and
publishing,aerospace,equipment andelectronic equipment.
Low frequency:Less innovative branches like clothing andfurniture.
Empirical evidence for Portugal:Holl (2004)Startupsand
relocations: ManufacturingplantlocationinPortugal,Papersin
RegionalScience

4-21
DifferencesinCitySize

Whydocitiesdifferinsizeandscope?
Differencesinlocalization&urbanizationeconomies.
Introductionofconsumer goodsamplifiesdifferencesin
size.

4-22
DifferencesinCitySize

The roleoflocalization economies andurbanization


economies indetermining the size ofacity:

4-23
Utility curvefor three types of
cities inaregion.
First,the curvefor an industry
for which localization ecnomies
areexhausted with arelatively
small workforce.
Diseconomies ofcommuting
quickly outplace agglomeration
benefits
Optimum city size is small.

4-24
Next:aspecializedcitywith
largerlocalizationeconomies.
Optimumcitysizeislarger.

4-25
Utilitycurveforacitythat
experienceslarge
urbanizationeconomies
Largeoptimumsize.

4-26
Locational equilibrium requires that workers inthe region areindifferent between
the three cities! achieve same utility level!
Assume:the region hasatotalpopulation of10million:s,m,b possible
equilibrium
This is astable equilibirum because each city is on the negatively sloped portion of
its utility curve.

4-27
LocalGoodsandCitySize
Weswitchthefocustotheconsumersideofanurban
economy.
Distinguishbetweenemploymentinindustriesthatexport
theiroutputtopeopleoutsidethecity(cars)and
employmentinindustriesthatselltheirgoodslocally
(coffeeshop).

4-28
LocalGoodsandCitySize
Goodsarelocallysoldif:
Percapitademandlargerelativetoscaleeconomiesinproduction

Somelocalgoods(haircuts,groceries,pizza).E.g.abarbercanbesupportedbyonlyafew
hundredpeople

Willbesoldinallcities,large&small(evensmallcitieswillgeneratesufficientdemandtosupport
atleastonefirm

Localemploymentroughlyproportionaltopopulation

Goodsareexported(orregionallysold)if:
Percapitademandsmallrelativetoscaleeconomiesinproduction

Soldonlyinlargecities(brainsurgery,opera)E.g.ittakesmanypeopletosupportoperasand
theaters>theywillonlysurviveinbigcities.

Localemploymentconcentratedinlargercities

Largercitieshavewidervariety:pizzas,haircuts,opera,brainsurgery

4-29
LocalGoodsandCitySize
Thisgeneratesspecificpatternsoftravelbehavior:

Sinceallgoodsareavailableinbigcities comparedtoonlyasmall
varietyinsmallcities consumerstraveltothemtogethighorder
products(surgery,luxurygoods).

Residentsfrombiggercitieswillnottraveltoasmallcityforapizza.

Export employment supports localemployment.Abig city supports a


wider range ofconsumer goods

4-30
Introducing localconsumer
goods:
Smallcity increase by half
one export job 0.5jobs inthe local
industry

Mediumcity doubles
one export job 1job inthe localindustry

Large city tripeles


on export job 2jobs inthe localindustry

Aslarger cities cansupport a


wider range ofconsumer
goods!

Differences incity size get


amplified!

4-31
The size distribution ofcities

Ranksize distribution is the


distribution ofsize by rank,
indecreasing order ofsize.

4-32
4-33
TheRankSizeRule
Withinaregion,thesizeofacityspopulationisinversely
proportionaltoitsrankonanorderedlist:e.g.
thelargestcityis#1
the2nd largestis#2withthepopulationof#1,
the3rd largestcityis#3with1/3thepopulationof#1
Forexample:Ifthelargestcityhasapopulationof24million(rank1),thesecond
largestcitywillhave12millionandthethirdlargestwillhave8million
inhabitants.

Thisrelationshipappliestolargeurbansystemsaroundtheworld
andoverlongperiodsoftime.

4-34
TheRankSizeRule
Thehypothesizedrelationshipis:
R=C/Nb
R:Rank C:Constant
N:Populationofcity b:scalingexponent

Ranksizeruleholdsifb=1:RankN=C

Empiricalresults
Medianestimateb=1.09:Closetoranksizerule,butmoreevendistribution

Definitionofeconomiccity:b=1.02

4-35
TheSizeDistributionofCities
Alternative hierarchy:deviations from the theoretical distribution.

There aretwo main exceptions tothe rule:

1.Urban Primacy

2.Binary pattern

4-36
UrbanPrimacy
Aprimatecity is one that hasmorethan twice the population ofthe next biggest
city (always disporportionately larger).
An example is Lima(Peru)that is morethan tentimeslarger than the next
settlement.
Other examples:Examples ofprimatecities include ParisinFrance,Londoninthe
United Kingdom,andTokyo inJapan.
Countries with aprimatecity,acity that dominates inpopulation size and,usually,
economically,have adeficit ofintermediate size cities.
The history ofthese countries plays alarge roleinthe persistence oftheir primate
city.Particularly,the concentration ofpolitical power inone city early on hasa
large degree ofpath dependency (colonialcities).
The Puzzle of the Large Primary City

4-38
ReasonsforLargePrimateCities
Historicallygrownprimarytradingcities:largeeconomiesofscaleintrade
(indivisibilitiesinimport/exportfacilities ports,tradecenters) encourageone
largetradingcityinthecountry.

Disproportionateinvestmentsininfrastructure:Indevelopingcountries,major
investmentsoccuraroundthecapital,leavingcitiesattheperipherybehind.
Henderson(2002)suggeststhatcountrieswithlimitedresourcesareforcedto
concentrateinfrastructureinvestmentononeortwocitiesraisingurbanprimacy
(.p.99).

Theroleofpolitics politicalbias:AdesandGlaeser (1995)suggestthatcountries


runbydictatorshavelargerprimarycities.Resourcesaregatheredfromthe
hinterlandandconcentratedinthecapital.Thestudyacross85countriesfindsthat
citieswithinadictatorshipare45%largerthaninothercountries.

4-39
ReasonsforLargePrimateCities
Thereareseveraltheoriestodescribethelargeconcentrationofpopulationin
primarycities:
History:capitalofalargeimperiumthatdisappeared(Vienna).
Vienna Population
Year Pop. %
1754 175,46
1800 271,8 +54.9%
1850 551,3 +102.8%
1900 1,769,137 +220.9%
1910 2,083,630 +17.8%
1923 1,918,720 7.9%
1939 1,770,938 7.7%
1951 1,616,125 8.7%
1961 1,627,566 +0.7%
1971 1,619,885 0.5%
1981 1,535,145 5.2%
1985 1,494,874 2.6%
1990 1,492,636 0.1%
1995 1,542,667 +3.4%
2000 1,548,537 +0.4%
2005 1,632,569 +5.4%
2010 1,689,995 +3.5%
2015 1,794,770 +6.2%

4-40
Binarypattern

Two or morecities arelarger then the predicted size.

Australia:Sydney andMelbourneeach with above 4million


andabout 20%oftotalpopulation

Other examples?

4-41
Furtherreading
Duranton,G.andD.Puga (2013)Thegrowthofcities
http://real.wharton.upenn.edu/~duranton/Duranton_Papers/Current_Research/urban_growth.pdf
Au,CCandJVHenderson(2006)AreChineseCitiesTooSmall?,Reviewof
EconomicStudies,73:549576.
Desmet,KandERossiHansberg (2010)UrbanAccountingandWelfare,CEPR
DiscussionPaper8168.
PaulKrugman (1996)Confronting the Mystery ofUrban Hierarchy,Journal ofthe
Japanese andInternationalEconomies 10,399418.
XavierGabaix andYannis M.Ioannides (2004),The Evolution ofCitySize
Distributions,In:HendersonV.andJF.Thisse,eds,Handbook ofRegionaland
Urban Economics Vol.4,Chapter 49,NorthHolland

4-42
Homeassignment is due on MONDAY!

You might also like