Präsentationsvorlage
FIRM LOCATION CHOICE
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
2
the idea
I COURSE COMPONENTS
▪ Block I
▪ Introduction to Urban and Regional Economics and Course Overview
▪ Topic I: Regional and urban concentration forces
▪ Topic II: The empirics of agglomeration
▪ Topic III: Costs and benefits of agglomeration
▪ Block 2
▪ Topic IV: Monocentric city I (household location choice)
▪ Topic V: Monocentric city II (household location choice)
▪ Topic VI: Firm location choice
▪ Topic VII: The urban economy in general equilibrium
▪ Block 3
▪ Topic VIII: The vertical dimension of cities
▪ Topic IX: Suburbanization and gentrification
▪ Topic X: Hedonic analysis
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
3
roadmap
I INTRODUCTION
▪ Last time: The monocentric city model II
▪ 1) Equilibrium conditions
▪ Additional restrictions
▪ 2) Comparative statics: Income
▪ What happens if people get richer?
▪ Where do the rich and the poor live in cities?
▪ 3) Comparative statics: Transport cost
▪ What happens if transport gets cheaper?
▪ Did cities decentralize over time?
▪ 4) Other predictions
▪ Distinguishing between open-city and closed-city case
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
4
roadmap
I INTRODUCTION
▪ This time: Firm location choice
▪ 1) Firms in the monocentric city model
▪ Firm bid-rent
▪ Land-use segregation
▪ 2) Agglomeration and decentralization
▪ Endogenous agglomeration
▪ Multiple equilibria
▪ 3) Emergence of new clusters
▪ Sub-centres
▪ Edge cities
▪ Historic anchoring
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
LSE © Gabriel M Ahlfeldt5
firms in the MCM
II THE CBD IN THE MCM
Q: Why do firms concentrate in the CBD?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
6
roadmap
II INTRODUCTION
▪ This time: Firm location choice
▪ 1) Firms in the monocentric city model
▪ Firm bid-rent
▪ Land-use segregation
▪ 2) Agglomeration and decentralization
▪ Endogenous agglomeration
▪ Multiple equilibria
▪ 3) Emergence of new clusters
▪ Sub-centres
▪ Historic anchoring
▪ Edge cities
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
7
firms in the MCM
II FIRM BID-RENT FUNCTION
▪ To rationalize firm concentration in the CBD in the MCM, we need a
force that attracts firms to the CBD
▪ High cost of distance, so that firms outbid residents
▪ Some fundamental location factor
▪ Firms need to be close to a natural harbour (trading cities)
▪ Firms sell goods at a market place (von Thünen)
▪ An agglomeration effect declining in distance from the CBD
▪ Being located centrally in a labour pool (MAR, (topic I)
▪ Knowledge spillovers (MAR, topic I)
Transport cost for goods decreased over time (topic I): Plausible in the past
Knowledge spillovers very localized (topic II): More plausible today
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
8
firms in the MCM
II FIRM PROFITS
▪ Use the Ahlfeldt & Wendland (2013), henceforth AW, framework
to illustrate the firm bid rent
▪ Identical services firms at location i maximize profits 𝜋𝑖
▪ Firms occupy land directly: No developers
▪ Capital broadly defined, includes workers, machines, computers,
and building structure
▪ Could introduce developers providing space in perfect analogy
to household bid-rent model
𝜋𝑖 = 𝜒𝑖 − 𝐾𝑖 − 𝜓𝑖𝐿𝑖
Output = revenue, since
price of good set to 1
Cost of land
L, Ψ is the
land rentCost of capital K, price
of capital set to 1
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
9
firms in the MCM
II PRODUCTION FUNCTION
▪ The production technogloy is Cobb-Douglas
▪ Similar results as long as (diminishing MPs)
▪ Productivity spillovers 𝑨𝒊 decline in distance from CBD
𝜒𝑖 = 𝐴𝑖𝐾𝑖
𝛼𝐿𝑖
1−𝛼 0<α<1 is capital share at inputs (constant in Cobb-Douglas) 𝐴𝑖 = 𝑎𝑖𝑒 −𝜃𝐷𝑖 Fundamental amenity, let’s assume it is “flat” Distance from the CBD Θ>0 Determines
the decay of the
spillover
1
D
Exponential function 𝑒−𝜃𝐷𝑖
returns one for D=0 and
approaches zero for D=∞
Productivity shifter
𝝌′ > 𝟎, 𝝌′′ < 𝟎 I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary 10 firms in the MCM II PROFIT MAXIMIZATION ▪ Firms choose inputs to maximize profits ▪ First-order conditions give ▪ Recall Micro 101 𝜋𝑖 = 𝐴𝑖𝐾𝑖 𝛼𝐿𝑖 1−𝛼 − 𝐾𝑖 − 𝜓𝑖𝐿𝑖 Combined profit and production function 𝐾𝑖 𝐿𝑖 = 𝛼 1 − 𝛼 𝜓𝑖 Capital density increases in land rent Similar to developers‘s problem, topic IV 𝜕𝜋𝑖 𝜕𝐾𝑖 𝜕𝜋𝑖 𝜕𝐿𝑖 = 1 − 𝛼 𝛼 𝐾𝑖 𝐿𝑖 = 𝜓𝑖 𝑝𝑘 =1 Check a micro textbook if unclear I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary 11 firms in the MCM II SPATIAL EQUILIBRIUM ▪ In spatial equilibrium profits are equalized ▪ Zero profits due to perfect competion, free entry and exit ▪ Plug in first-order condition: 𝜋𝑖 = 𝐴𝑖𝐾𝑖 𝛼𝐿𝑖 1−𝛼 − 𝐾𝑖 − 𝜓𝑖𝐿𝑖 = 0 𝜓𝑖 = 𝛼 𝛼𝐴 𝑖 1 1−𝛼 = 𝑐𝑒 − 𝜃 1−𝛼 𝐷𝑖 Corresponds to fixed utility in household bid-rent model ln𝜓𝑖 = 𝑐 − 𝜃 1 − 𝛼 𝐷𝑖 𝐾𝑖 = 𝛼 1−𝛼 𝜓𝑖 𝐿𝑖 𝜕𝜓𝑖 𝜕𝐷𝑖 < 0 Firm bid-rent decreases in distance from the CBD Higher office rents compensate for greater productivity near CBDc summarizes constants I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary 12 firms in the MCM II LAND PRICE GRADIENTS BY USE x=distance from CBD Bid rent for land Residential land price gradient (topics IV & V) Commercial land price gradient with large Θ Manufacturing land price gradient with small Θ Q: Land use pattern? I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary 13 firms in the MCM II LAND PRICE GRADIENTS BY USE x=distance from CBD Bid rent for land ManufacturingResidentialCBD URBAN RURAL Residential land price gradient (topics IV & V) Commercial land price gradient with large Θ Manufacturing land price gradient with small Θ Land goes to the highest bidder Land price gradient is the envelope of price gradients by use => convexity due to sorting
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
14
II PRIME LOCATION GRADIENTS IN 125 CITIES
Ahlfeldt et al. (2018)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
15
II PRIME LOCATION GRADIENTS IN 125 CITIES
Ahfleldt et al. (2020)
Largest prime location in city: Identified as
concentration of office buildings held by REITS &
Starbucks franchises:
Note: SNL Data AVAILABLE to REEF students
Commercial used
dominates
Office rents decline!
Structural density declines !
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
LSE © Gabriel M Ahlfeldt16
agglomeration and decentralization
III EMPLYOMENT CONCENTRATION
Q: Can the spatial structure of a city be changed?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
17
roadmap
III AGGLOMERATION AND DECENTRALIZATION
▪ This time: Firm location choice
▪ 1) Firms in the monocentric city model
▪ Firm bid-rent
▪ Land-use segregation
▪ 2) Agglomeration and decentralization
▪ Endogenous agglomeration
▪ Multiple equilibria
▪ 3) Emergence of new clusters
▪ Sub-centres
▪ Historic anchoring
▪ Edge cities
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
18
agglomeration and decentralization
III CBDs DOMINATED BY PRIME SERVICES
Knowledge-basedtradable
services highly
concentrated in CBDs
Manufacturing
employment much more
decentralized
Fundamentals not
obviously relevant for
prime services
Ahlfeldt et al. (2020)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
19
agglomeration and decentralization
III TRADABLE SERVICES
▪ An agglomeration economy emerging from the CBD is a
simplfying assumption
▪ More realistically firms benefit from each other and, therefore,
cluster in the CBD
▪ To capture the idea AW introduce a bidirectional spillover
CBD is endogenous!
𝐴𝑖 = 𝑎𝑖Λ𝑖Γ𝑖
Λ𝑖 = 𝑒
−𝜃𝐷𝑖
Γ𝑖 = 𝑒
𝛽𝑍𝑖; 𝑍𝑖 = σ𝑗 𝜒𝑗𝑒
−𝜏𝑑𝑖𝑗
Productivity shifter
Exogenous „CBD spillover“
Endogenous bidirectional
agglomeration force. Depends
on nearby output, weighted by
distance
Agglomeration elasticity
(strength of the effect)
Agglomeration
decay
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
20
agglomeration and decentralization
III BID-RENT WITH BILATERAL AGGLOMERATION
▪ Commercial bid rent
𝐴𝑖 = 𝑎𝑖Λ𝑖Γ𝑖
Λ𝑖 = 𝑒
−𝜃𝐷𝑖
Γ𝑖 = 𝑒
𝛽𝑍𝑖; 𝑍𝑖 = σ𝑗 𝜒𝑗𝑒
−𝜏𝑑𝑖𝑗
Commercial bid rent decline in distance from CBD
𝜓𝑖 = 𝛼
𝛼𝐴
𝑖
1
1−𝛼
Commercial bid rent inreases in local economic density
If bilateral agglomeration economies are strong,
commercial land prices will be highly spatially correlated!
𝜒𝑗 =
1
1−𝛼
𝜓𝑗𝐿𝑗
Zero-profit condition and
first-order condition at location j give
Market potential measure
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
21
agglomeration and decentralization
III BID-RENT WITH BILATERAL AGGLOMERATION
▪ Commercial bid rent
𝐴𝑖 = 𝑎𝑖Λ𝑖Γ𝑖
Λ𝑖 = 𝑒
−𝜃𝐷𝑖
Γ𝑖 = 𝑒
𝛽𝑍𝑖; 𝑍𝑖 = σ𝑗 𝜒𝑗𝑒
−𝜏𝑑𝑖𝑗
𝜓𝑖 = 𝛼
𝛼𝐴
𝑖
1
1−𝛼 𝜒𝑗 =
1
1−𝛼
𝜓𝑗𝐿𝑗
Market potential measure
ln𝜓𝑖 = c +
𝛽
1−𝛼 2
σ𝑗𝜓𝑗𝐿𝑗𝑒
−𝜏𝑑𝑖𝑗 −
𝜃
1−𝛼
𝐷𝑖
Distance-weighted land value
Q: Changes over time?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
22
agglomeration and decentralization
III CBD VS. BILATERAL SPILLOVERS IN BERLIN I
▪ AW analyse a commercial land values in 1890-1936 Berlin
▪ Rapid urban growth: Population tripled from 1.5M to 4.5M
▪ Transition from “historic” to “modern” CBD / economy
Transition resembles many cities in
developing world today
Bilateral agglomeration should be
increasingly important
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
23
agglomeration and decentralization
III CBD VS. BILATERAL SPILLOVERS IN BERLIN I
▪ AW provide a multivariate estimation of strength of CBD
spillover and bidirectional agglomeration forces over time
𝜓𝑖 = 𝑐𝑒
−
𝜃
1−𝛼
𝐷𝑖
ln𝜓𝑖 = ln𝑐 −
𝜃
1−𝛼
ln𝐷𝑖
Exogenous „CBD spillover“
becomes less important
Transition towards “knowledge” economy
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
24
agglomeration and decentralization
III CBD VS. BILATERAL SPILLOVERS IN BERLIN II
▪ AW provide a multivariate estimation of strength of CBD
spillover and bidirectional agglomeration forces over time
Exogenous „CBD spillover“
becomes less important
1890 1936Transition to “knowledge” economy
Agglomeration elasticity 𝛽
increases from 3.5% to 8.3%
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
25
agglomeration and decentralization
III IMPLICATION FOR URBAN STRUCTURE I
With strong CBD spillover
New cluster emerges in
periphery (e.g. start-ups)
Or shock relocates existing
firms (e.g. disaster)
Exogenous CBD spillover
pulls firms into the CBD
City structure remains stable
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
26
agglomeration and decentralization
III IMPLICATION FOR URBAN STRUCTURE II
With bidirectional
agglomeration force
New cluster emerges in
periphery (e.g. start-ups)
Or shock relocates existing
firms (e.g. disaster)
New cluster generates own
agglomeration economies
City structure can change
=> Multiple equilibria!
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
27
agglomeration and decentralization
III CAUSES OF EMPLOYMENT DECENTRALIZATION
▪ Manufacturing should decentralize given change in transport,
production, and inventory technologies
▪ Tradable services could decentralize since bidirectional
agglomeration forces imply multiple equilibria
▪ Driving forces for employment decentralization
▪ Change in transport technology
▪ Suburbanization population (jobs follow people)
▪ Fiscal and social problems (e.g. US)
▪ Ageing/redundant building stock
▪ Regulation (e.g. height constraints)
▪ …
Q: Is there evidence?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
28
emergence of new clusters
III GLAESER & KAHN (2001)
▪ US cities are mostly decentralized (NY is an exception)
▪ Share of employment within three miles 29% or less
▪ Employment share of central MSA county falling since 1950
▪ Hard to predict decentralization, but
▪ Manufacturing particularly decentralized
▪ Suburbanization associated with decentralized employment
▪ Cities specialized in services are relatively centralized
▪ Knowledge-based industries more likely in city centre
Evidence supports theoretical expectations
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
29
emergence of new clusters
III BAUM-SNOW ET AL (2017)
▪ Transport infrastructure in 1990-2010 China decentralizes
population and employment
▪ Radial highways decentralize services activity
▪ Radial railroad decentralize industrial activity
▪ 20% of industrial activity
▪ Ring roads decentralize both
▪ 50% of industrial activity
▪ Radial / ring roads displace 4% / 20% of central city
population
Evidence supports theoretical expectations
Identification strategy similar to
Baum-Snow (2006), see topic V
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
30
roadmap
IV AGGLOMERATION AND DECENTRALIZATION
▪ This time: Firm location choice
▪ 1) Firms in the monocentric city model
▪ Firm bid-rent
▪ Land-use segregation
▪ 2) Agglomeration and decentralization
▪ Endogenous agglomeration
▪ Multiple equilibria
▪ 3) Emergence of new clusters
▪ Sub-centres
▪ Historic anchoring
▪ Edge cities
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
31
emergence of new clusters
IV DECENTRALIZED VS. DISPEARSED EMPLOYMENT
▪ Employment decentralized in many cities
▪ Not necessarily the same as the dispearsed employment
Decentralised and dispearsed Decentralised and clustered
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
LSE © Gabriel M Ahlfeldt32
emergence of new clusters
IV SUBCENTERS
Q: Why, when, and where do cities develop
commercial clusters outside the CBD?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
33
emergence of new clusters
IV MONOCENTRIC VS POLYCENTRIC CITIES
▪ (Dis)advantage of large monocentric cities
▪ Large agglomeration economies in CBD
▪ Long commuting cost and higher wages as compensation
▪ High land prices
▪ Large polycentric cities
▪ Have one or multiple subcentres
▪ Subcentres are clusters of employment that resemble the CBD
▪ Offer some agglomeration benefits
▪ Reduce commuting cost and land prices
Polycentric cities combine benefits of large and small cities!
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
34
emergence of new clusters
IV LARGE MONOCENTRIC CITY
x=distance from CBD
Bid rent
for land
CBDx=distance from CBD
Residential land price gradient
Commercial land price gradient
with large Θ
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
35
emergence of new clusters
IV LARGE POLYCENTRIC CITY
x=distance from CBD
Bid rent
for land
CBDx=distance from CBD
Residential land price gradient
Commercial land price gradient
with large Θ
SubcentreSubcentre
In AW model, every location j in
the cits is treated as a subcentre
𝜓𝑖 = 𝑓
𝑗
𝜒𝑗𝑒
−𝜏𝑑𝑖𝑗
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
36
emergence of new clusters
IV WHEN DO CITIES DEVELOP SUBCENTRES?
▪ Fujita & Ogawa (1982) model: Number of subcentres increases
▪ in population and commuting cost
▪ Subcentres emerge endogenously to keep the city efficient
▪ McMillen & Smith (2003) test the prediction for US cities
▪ Identify subcentres as peaks in local employment densities
▪ McMillen (2001) locally weighted regression approach
▪ Run Poisson regressions (count models)
▪ Anayse determinants of number of subcentres per city
Q: How do city size, congestion levels, and other factors affect the
number of subcentres?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
37
emergence of new clusters
IV DETERMINANTS OF SUBCENTRE FORMATION
Low-congestion city develop
1st / 2nd subcenter at a
population of 2.68M / 6.74M
High congestion cities develop
subcentres earlier
Older housing stock also make
subcentre formation more
likely, but effect smaller
Evidence supports theory!McMillen & Smith (2003)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
38
emergence of new clusters
IV WHERE DO SUBCENTERS EMERGE?
▪ Subcentres tend to be emerge:
▪ McMillen & Smith (2003)
▪ close to highway intersections
▪ In old satalite suburbs
▪ Garcia-López et al (2017)
▪ close to regional express rail
▪ Ahlfeldt & Wendland (2013)
▪ close to the CBD if it is
historically strong
Evidence from the US
Evidence from Paris
Evidence from Berlin
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
39
IV EMPLOYMENT DENSITY IN NYC
Q: Why did some cities (not) decentralize?
Ahlfeldt et al (2020)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
40
IV LA VS. NYC NYC
Ahlfeldt et al (2020)
LA
LA has many
more remote PLs
than NYC,
despite similar
population
Role for history?
NYC 1900 pop: 3.4M
LA 1900 pop: 100k
NYC built a much
larger transit network
much earlier (1868)
Does the evidence
generalize?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
41
IV EVIDENCE FROM 125 GLOBAL CITIES
Ahlfeldt et al (2020)
Historic density and subways achhor CBDs Disasters promote transition
Caloric potential
Subway potential
Role of agglomeration- and transport-induced persistence
shown in simulations within granular spatial model
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
LSE © Gabriel M Ahlfeldt42
emergence of new clusters
IV EDGE CITIES
Q: Where would a profit-maximizing developer
strategically develop a satellite city?
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
43
emergence of new clusters
IV EDGE CITIES
▪ Term coined by Washington Post journalist and author Joel Garreau
in his 1991 book Edge City: Life on the New Frontier.
▪ New cities in urban periphery built by one developer
▪ Built to host knowledge based tradable services
▪ Dominated by class A office stock (shiny offices)
▪ Net commuting destination (unlike a residential suburb)
▪ Proper city with employment, recreation, entertainment
▪ Popping up since 1965: A “new” urban lifestyle (?)
▪ 123 edge cities and 83 up-and-coming places listed by Garreau
▪ 24 in greater LA, 23 in metro D.C., and 21 in greater NY
Edge city developed by a single developer (unlike a subcentre)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
44
emergence of new clusters
IV WHY EDGE CITIES?
▪ Developer aims at creating an attractive edge city
▪ internalizing agglomeration spillovers
▪ avoiding social-fiscal problems (poverty, crime…)
▪ reducing capacity constraints (congestions).
▪ Typical „instruments“
▪ Mixed use space to minimize commuting
▪ Optimal road layout and parking capacity (no sunk cost)
▪ Non-distorionary taxes (little redistribution)
Developer solves a “coordination problem” inherent to agglomeration
Value of land is low before and high after development ⇒ Profits!
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
LSE © Gabriel M Ahlfeldt45
emergence of new clusters
IV TYSON’S CORNER
„Archetypal “ Edge City
Developed by Til Hazel (Hazel & Peterson)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
46
emergence of new clusters
IV TYSON’S CORNER
Largest retail area on the area on the east coast
south of NY, 3,400 hotel rooms, over 100,000 jobs,
over 28 million square feet of A-level office space
(relative to 17 million in Washington core city)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
47
emergence of new clusters
IV IRVINE
Developed by Irvine Co. (Hazel & Peterson)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
48
emergence of new clusters
IV IRVINE
Largest Edge City, over 160,000 jobs, over 33
million square feet of A-level office space
(relative to 25 million in LA core city)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
49
▪ Edge city developed by one profit-maximizing developer
▪ Returns from producing a tradabe services good
▪ Faces cost of providing space (constant captial cost pk)
▪ Faces cost of hiring labour: Wages
▪ increase in residential rent (workers require compensaiton)
▪ increase in number of workers in the metro (congestion)
▪ Developers chooses
▪ capacity K1 (floor space)
▪ workforce B
▪ location y defined in terms of distance from the port city (CBD)
emergence of new clusters
IV HENDERSON AND MITRA (1996)
Developer has monopsony power
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
50
Edge city spillover
Port city spillover
emergence of new clusters
IV HENDERSON AND MITRA (1996)
▪ Port city and edge city generate agglomeration economies
▪ Depend on population (A for port city, B for edge city)
▪ Spillover to each other, declining in distance
Historic city (port)
Distance y
Edge city
A
g
g
lo
m
e
ra
ti
o
n
e
c
o
n
o
m
y
Agglomeration effect
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
51
emergence of new clusters
IV HENDERSON AND MITRA (1996)
▪ Edge city developed by one profit-maximizing developer
▪ Must choose the location of the edge city
Historic city (port) Distance y
Edge city
Productivity increases
(CBD spillover)
Developer faces a trade-off when choosing distance from the CBD
But CBD also benefits
(increases competition for workers)
Land cost increases (MCM), higher
compensations to workers
Result: Small changes in CBD capacity K0 can trigger large effects on y
Location choice of developer can appear „chaotic“
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
52
emergence of new clusters
IV HENDERSON AND MITRA (1996)
▪ Edge city developed by one profit-maximizing developer
▪ Must choose capacity K1 and employment B of edge city
▪ Higher CBD capacity K0 reduces monopsony power
▪ CBD pays higher wages (greater agglomeration effect)
Developer needs to consider K0 when choosing edge city capacity
Result: Smaller edge city (K1 and B) with larger port city, but
K1 and B may even increase if distance y changes in response to K0
Model highly sensitive to parameter values („everything goes“)
„Chaotic“ patterns resemble Garreau‘s notion of „randomness
(but „randomness“ and „chaos“ are not the same!)
More in seminar…
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
53
▪ Firm bid-rent can be derived using zero-profit condition
▪ Historically, transport cost to CBD relevant determinant
▪ Agglomeration spillover more plausible determinant today
▪ Changes in transport technology lead to employment decentralization
▪ Manufacturing employment decentralizes due to lower transport cost
▪ Services can decentralize since agglomeration endogenous
▪ Decentralized employment not the same as dispearsed employment
▪ Subcentres emerge endogenously if cities get large
▪ Developers strategically build edge cities to make profits
▪ Next: Firm location
▪ Determinants of firm locations within cities
conclusion
SUMMARY
THANKS
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
55
▪ Core readings:
▪ Ahlfeldt, G., Wendland, N. (2011), How polycentric is a monocentric city?
▪ Glaeser, E. L., Kahn, M. E., „Decentralized employment and the transformation of the American city”. Brookings-
Wharton Papers on Urban Affairs, pp. 1-63.
▪ McMillen D. and S.C. Smith, 2003, “The number of subcentres in large urban areas”, Journal of Urban Economics
53, 321-338
▪ Henderson J.V. and A. Mitra, 1996, “The new urban landscape: Developers and Edge cities”, Regional Science
and Urban Economics 26, 613-643
▪ Complementary readings and references:
▪ Ahlfeldt, G., Albers, T., Behrens, K. (2020): Prime locations. CEP DP 1725.
https://cep.lse.ac.uk/_new/publications/abstract.asp?index=7318
▪ Baum-Snow N., Brandt L., Henderson V., Turner, M. and Zhang, Q. (2017): Roads, Railroads, and
Decentralization of Chinese Cities. The Review of Economics and Statistics, 99(3)
▪ McMillen, D. (2001): Non-parametric subcenter identification. Journal of Urban Economics, 50, 448-473
▪ Garcia-López, M., Hémet, C., Viladecans-Marsal, E. (2017): Next train to the polycentric city: The effect of
railroads on subcenter formation, Regional Science and Urban Economics, 67, Pages 50-63
READING
https://cep.lse.ac.uk/_new/publications/abstract.asp?index=7318
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
56
agglomeration and decentralization
III DETERMINANTS OF FIRM LOCATION CHOICE
▪ In choosig locations, firms consider various factors
▪ Natural amenities
▪ Proximity to transport, railway, highway
▪ Proximity to consumers and clients
▪ Proximity to complementary firms
▪ Proximity to workers
▪ Land cost and land availability
▪ Telecommunication (super-fast broadband)
▪ …
Importance of factors is specific to industry sectors, and periods
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
57
agglomeration and decentralization
III CHANGES IN LOCATION FACTORS
▪ Fundamental changes over the past centuries affect choices
▪ Natural amenities arguably less important
▪ E.g. role of natural habours, proximity to coal fields
▪ Transportation systems have improved
▪ Railways (more than 25,000 km of high-speed rail in China)
▪ Subways (Shanghai Metor system has a length of 644 km)
▪ Mass-produced cars (as of 2010 1.015 billion cars in the world)
▪ Highways (Interstate Highway System has a lenght of 77,556 km)
▪ Clustering of complementary firms more important due to
increasing knowledge intenstiy of prodction
Fundamental changes can reduce the importance of being close to CBD
Likely different reasons for manufacturing and tradable services
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
58
agglomeration and decentralization
III MANUFACTURING
▪ Historically attractive to be located in the CBD
▪ Being close to natural advantages (waterways)
▪ Transport systems (central rail station)
▪ Other firms
▪ Relatively small establishments, land was affordable
▪ Better transportation systems (more efficient and denser)
▪ Access to highways in urban periphery might even be better
▪ Change in production and storage technology
▪ Integrated horizontal assembly lines
▪ Inventory technology requries large, single-story structure
Production requires more land
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
59
agglomeration and decentralization
III HISTORIC BID-RENT MODEL
▪ How did the CBD come about?
x=distance from CBD
Bid rent
for land
ResidentialManufacturingCBD
Residential land price gradient
Manufacturers face high
transport cost, small plants
serving local markets
„von Thünen world“
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
60
agglomeration and decentralization
III CLASSIC TEXT-BOOK MODEL
▪ After emergence of an office sector
x=distance from CBD
Bid rent
for land
ResidentialManufacturingCBD
Residential land price gradient
Manufacturers face high
transport cost, small plants
serving local markets
Classic „AMM“ model from
1960/70s
Commercial land price gradient
with large Θ
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
61
agglomeration and decentralization
III MODERN TEXT-BOOK MODEL
▪ After emergence of an office sector
x=distance from CBD
Bid rent
for land
ManufacturingResidentialCBD
Residential land price gradient
Manufacturers face low transport
cost, large plants serving
national/global markets
Commercial land price gradient
with large Θ
Changes in transportation and
production technology lead to
decentralization of manufacturing
employment!
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
62
Firms make
sequential
location choices
to maximize
profits,
benefitting from
spillovers from
other firmst
agglomeration and decentralization
III APPENDIX: AGENT-BASED MODELLING
Firms
Amenity
Spillover
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
63
agglomeration and decentralization
III ILLUSTRATION USING AGENT-BASED MODELLING
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
64
agglomeration and decentralization
III ILLUSTRATION USING AGENT-BASED MODELLING
Firms have move
away from the
former CBD!
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
65
emergence of new clusters
IV A POLYCENTRIC MONOCENTRIC CITY
Before transition to
knowledge-based economy
CBD gradients dominates
spatial structure
After transition, there are
several specialised micro
agglomerations (e.g. for
banking, media, lobbying)
mostly close to the CBD
T
ra
n
s
itio
n
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
66
emergence of new clusters
IV SPATIAL CONCENTRATION VS 1900 POPULATION
“Hysteresis effect”: If cities were
large at time of transition (around
1900), economic activity remains
anchored close to CBD
Explains differences in city structure
New York vs. Los Angeles
East coast vs. west coast US cities
US vs. European & Asian cities
Assumption: In 1900 cities were monocentric (large city = large CBD)
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
67
emergence of new clusters
IV IN LINE WITH AGENT-BASED MODELLING?
“Small-city”
results
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
68
emergence of new clusters
IV IN LINE WITH AGENT-BASED MODELLING?
“Big-city”
results
I Introduction II Firm bid rent III Decentralization IV Subcenters V Edge cities VI Summary
69
emergence of new clusters
IV IN LINE WITH AGENT-BASED MODELLING?
“Big-city”
results
In large city,
remain anchored
close to historic
centre
Small city
relocation