Spatial Competition with Heterogeneous Firms Jonathan Vogel - - PowerPoint PPT Presentation

spatial competition with heterogeneous firms
SMART_READER_LITE
LIVE PREVIEW

Spatial Competition with Heterogeneous Firms Jonathan Vogel - - PowerPoint PPT Presentation

Spatial Competition with Heterogeneous Firms Jonathan Vogel November 2007 Jonathan Vogel () Spatial Competition 11/14 1 / 32 Introduction I model endogenous product dierentiation with heterogeneous rms Jonathan Vogel () Spatial


slide-1
SLIDE 1

Spatial Competition with Heterogeneous Firms

Jonathan Vogel November 2007

Jonathan Vogel () Spatial Competition 11/14 1 / 32

slide-2
SLIDE 2

Introduction

I model endogenous product di¤erentiation with heterogeneous …rms

Jonathan Vogel () Spatial Competition 11/14 2 / 32

slide-3
SLIDE 3

Introduction

I model endogenous product di¤erentiation with heterogeneous …rms Two branches of product di¤erentiation literature

Jonathan Vogel () Spatial Competition 11/14 2 / 32

slide-4
SLIDE 4

Introduction

I model endogenous product di¤erentiation with heterogeneous …rms Two branches of product di¤erentiation literature Economists tend to hold product characteristics …xed when considering pricing decisions and …rm behavior more generally = ) endogeneity bias

Jonathan Vogel () Spatial Competition 11/14 2 / 32

slide-5
SLIDE 5

Introduction

Motivating example

Estimate the change in domestic-…rm pro…t resulting from an increase in a tari¤

Jonathan Vogel () Spatial Competition 11/14 3 / 32

slide-6
SLIDE 6

Introduction

Motivating example

Estimate the change in domestic-…rm pro…t resulting from an increase in a tari¤ First step 2 4 market shares prices product characteristics 3 5 = ) demand system marginal costs

  • Jonathan Vogel

() Spatial Competition 11/14 3 / 32

slide-7
SLIDE 7

Introduction

Motivating example

Estimate the change in domestic-…rm pro…t resulting from an increase in a tari¤ First step 2 4 market shares prices product characteristics 3 5 = ) demand system marginal costs

  • Counter-factual exercise

2 4 demand system NEW marginal costs FIXED product characteristics 3 5 = ) market shares prices

  • Jonathan Vogel

() Spatial Competition 11/14 3 / 32

slide-8
SLIDE 8

Introduction

Endogenous di¤erentiation and …rm heterogeneity

Markets are rarely perfectly competitive —–Spence (1976), Dixit Stiglitz (1977), Salop (1979) Firm productivity di¤ers signi…cantly both within and across industries —–Jovanovic (1982), Hopenhayn (1992) Models studying …rm heterogeneity in monopolistically competitive industries abstract from or treat as exogenous product placement —–Melitz (2002), Syverson (2004), Melitz Ottaviano (2005)

Jonathan Vogel () Spatial Competition 11/14 4 / 32

slide-9
SLIDE 9

Introduction

Spatial competition

Spatial competition models are ideally suited to answer: How does …rm heterogeneity a¤ect product placement in product space or …rm location in geography? Spatial competition literature dates back to Hotelling (1929)

Two-stage model of Bertrand competition in which location di¤erentiates otherwise homogeneous goods

Jonathan Vogel () Spatial Competition 11/14 5 / 32

slide-10
SLIDE 10

Introduction

While a spatial competition framework would be ideal, …nding equilibria in "simple" symmetric-…rm Hotelling-style models has proven di¢cult

Jonathan Vogel () Spatial Competition 11/14 6 / 32

slide-11
SLIDE 11

Introduction

While a spatial competition framework would be ideal, …nding equilibria in "simple" symmetric-…rm Hotelling-style models has proven di¢cult

Hotelling was wrong

Jonathan Vogel () Spatial Competition 11/14 6 / 32

slide-12
SLIDE 12

Introduction

While a spatial competition framework would be ideal, …nding equilibria in "simple" symmetric-…rm Hotelling-style models has proven di¢cult

Hotelling was wrong D’Aspremont, Gabszewicz, and Thisse (1979) prove that no pure-strategy equilibrium exists to a standard Hotelling model

Jonathan Vogel () Spatial Competition 11/14 6 / 32

slide-13
SLIDE 13

Introduction

While a spatial competition framework would be ideal, …nding equilibria in "simple" symmetric-…rm Hotelling-style models has proven di¢cult

Hotelling was wrong D’Aspremont, Gabszewicz, and Thisse (1979) prove that no pure-strategy equilibrium exists to a standard Hotelling model Salop (1979) and Syverson (2004) abstract from product placement

Jonathan Vogel () Spatial Competition 11/14 6 / 32

slide-14
SLIDE 14

Introduction

While a spatial competition framework would be ideal, …nding equilibria in "simple" symmetric-…rm Hotelling-style models has proven di¢cult

Hotelling was wrong D’Aspremont, Gabszewicz, and Thisse (1979) prove that no pure-strategy equilibrium exists to a standard Hotelling model Salop (1979) and Syverson (2004) abstract from product placement Lancaster (1979) assumes that product placement and prices are chosen simultaneously

Jonathan Vogel () Spatial Competition 11/14 6 / 32

slide-15
SLIDE 15

Introduction

While a spatial competition framework would be ideal, …nding equilibria in "simple" symmetric-…rm Hotelling-style models has proven di¢cult

Hotelling was wrong D’Aspremont, Gabszewicz, and Thisse (1979) prove that no pure-strategy equilibrium exists to a standard Hotelling model Salop (1979) and Syverson (2004) abstract from product placement Lancaster (1979) assumes that product placement and prices are chosen simultaneously

Either assume that …rms are homogeneous or abstract from location choice

Jonathan Vogel () Spatial Competition 11/14 6 / 32

slide-16
SLIDE 16

Introduction

I allow …rms to randomize over prices

Jonathan Vogel () Spatial Competition 11/14 7 / 32

slide-17
SLIDE 17

Introduction

I allow …rms to randomize over prices

Nevertheless, strategies are pure along equilibrium path

Jonathan Vogel () Spatial Competition 11/14 7 / 32

slide-18
SLIDE 18

Introduction

I allow …rms to randomize over prices

Nevertheless, strategies are pure along equilibrium path

Tractability of framework allows me to answer questions of the form:

Jonathan Vogel () Spatial Competition 11/14 7 / 32

slide-19
SLIDE 19

Introduction

I allow …rms to randomize over prices

Nevertheless, strategies are pure along equilibrium path

Tractability of framework allows me to answer questions of the form:

Will a …rm locate closer to its relatively less productive neighbor?

Jonathan Vogel () Spatial Competition 11/14 7 / 32

slide-20
SLIDE 20

Introduction

I allow …rms to randomize over prices

Nevertheless, strategies are pure along equilibrium path

Tractability of framework allows me to answer questions of the form:

Will a …rm locate closer to its relatively less productive neighbor? Does opening the black box of di¤erentiation yield new insight into the mechanism linking productivity to pro…t and market share?

Jonathan Vogel () Spatial Competition 11/14 7 / 32

slide-21
SLIDE 21

Introduction

I allow …rms to randomize over prices

Nevertheless, strategies are pure along equilibrium path

Tractability of framework allows me to answer questions of the form:

Will a …rm locate closer to its relatively less productive neighbor? Does opening the black box of di¤erentiation yield new insight into the mechanism linking productivity to pro…t and market share? How does the productivity of direct competitors a¤ect outcomes such as pro…t, market share, and the ease with which consumers substitute between goods?

Jonathan Vogel () Spatial Competition 11/14 7 / 32

slide-22
SLIDE 22

Introduction

Technical contributions

1

A set of SPNE to a standard Hotelling-style model generalized in two ways:

1

…rm heterogeneity

2

horizontal and vertical di¤erentiation (vertical not in presentation)

2

Firms use pure strategies along the equilibrium path

3

There is a unique economic outcome in any strict SPNE under a simple re…nement

Jonathan Vogel () Spatial Competition 11/14 8 / 32

slide-23
SLIDE 23

Setup

Consumers

A mass L of consumers uniformly distributed along a unit circumference

Jonathan Vogel () Spatial Competition 11/14 9 / 32

slide-24
SLIDE 24

Setup

Consumers

A mass L of consumers uniformly distributed along a unit circumference Each consumer inelastically demands one good

Jonathan Vogel () Spatial Competition 11/14 9 / 32

slide-25
SLIDE 25

Setup

Consumers

A mass L of consumers uniformly distributed along a unit circumference Each consumer inelastically demands one good A consumer located at point z buys from …rm i if pi + t kz ik min

j

fpj + t kz jkg where t > 0

Jonathan Vogel () Spatial Competition 11/14 9 / 32

slide-26
SLIDE 26

Setup

Consumer preferences

A B 1 t z

p A p B

A graphical representation of consumer preferences

Jonathan Vogel () Spatial Competition 11/14 10 / 32

slide-27
SLIDE 27

Setup

Firms: costs

Firm i is associated with a constant marginal cost of production ki

Jonathan Vogel () Spatial Competition 11/14 11 / 32

slide-28
SLIDE 28

Setup

Firms: costs

Firm i is associated with a constant marginal cost of production ki Additionally, …rm incurs a "shipping cost" of 2τd, with τ 2 [0, t), to ship a good to a consumer located a distance d from its location

Jonathan Vogel () Spatial Competition 11/14 11 / 32

slide-29
SLIDE 29

The game

Firms play a two-stage game of complete information

Jonathan Vogel () Spatial Competition 11/14 12 / 32

slide-30
SLIDE 30

The game

Firms play a two-stage game of complete information

1

Location stage

Jonathan Vogel () Spatial Competition 11/14 12 / 32

slide-31
SLIDE 31

The game

Firms play a two-stage game of complete information

1

Location stage

2

Price stage

Jonathan Vogel () Spatial Competition 11/14 12 / 32

slide-32
SLIDE 32

The game

Stage one: location stage

There is a set of n 2 …rms The vector of marginal costs (k1, ..., kn) is common knowledge All …rms simultaneously choose locations along the circumference of the circle

Jonathan Vogel () Spatial Competition 11/14 13 / 32

slide-33
SLIDE 33

The game

Stage two: price stage

All locations and marginal costs are common knowledge at the beginning of the price stage All …rms simultaneously choose their prices

Jonathan Vogel () Spatial Competition 11/14 14 / 32

slide-34
SLIDE 34

No SPNE

A simple game without a simple solution

p A

p B A B

pA

D

pC zBC zAB zAB

D

C

Market share is discontinuous in price

Jonathan Vogel () Spatial Competition 11/14 15 / 32

slide-35
SLIDE 35

No pure-strategy equilibrium

Pro…ts are not globally continuous or quasi-concave

pB

D = pA ? td

pB

DD = pA + td

pB

^B

Firm B’s pro…t as a function of its price (with n = 2)

Jonathan Vogel () Spatial Competition 11/14 16 / 32

slide-36
SLIDE 36

Mixing

Outline

For any subgame, there exists a mixed-strategy equilibrium - Reny (1999)

Jonathan Vogel () Spatial Competition 11/14 17 / 32

slide-37
SLIDE 37

Mixing

Outline

For any subgame, there exists a mixed-strategy equilibrium - Reny (1999) Can’t solve directly for pro…t with n asymmetric …rms randomizing

  • ver prices - Osbourne and Pitchik (1987)

Jonathan Vogel () Spatial Competition 11/14 17 / 32

slide-38
SLIDE 38

Mixing

Outline

For any subgame, there exists a mixed-strategy equilibrium - Reny (1999) Can’t solve directly for pro…t with n asymmetric …rms randomizing

  • ver prices - Osbourne and Pitchik (1987)

I prove there exists an upper bound on a …rm’s pro…t in any subgame in which there is no pure-strategy equilibrium in prices

Jonathan Vogel () Spatial Competition 11/14 17 / 32

slide-39
SLIDE 39

Mixing

Outline

For any subgame, there exists a mixed-strategy equilibrium - Reny (1999) Can’t solve directly for pro…t with n asymmetric …rms randomizing

  • ver prices - Osbourne and Pitchik (1987)

I prove there exists an upper bound on a …rm’s pro…t in any subgame in which there is no pure-strategy equilibrium in prices Suppose …rm i unilaterally deviates in the location stage from conjectured equilibrium and in subsequent price stage there exists no pure strategy equilibrium in prices

Jonathan Vogel () Spatial Competition 11/14 17 / 32

slide-40
SLIDE 40

Mixing

Outline

For any subgame, there exists a mixed-strategy equilibrium - Reny (1999) Can’t solve directly for pro…t with n asymmetric …rms randomizing

  • ver prices - Osbourne and Pitchik (1987)

I prove there exists an upper bound on a …rm’s pro…t in any subgame in which there is no pure-strategy equilibrium in prices Suppose …rm i unilaterally deviates in the location stage from conjectured equilibrium and in subsequent price stage there exists no pure strategy equilibrium in prices Upper bound on i’s pro…t strictly less than pro…t had it not deviated

Jonathan Vogel () Spatial Competition 11/14 17 / 32

slide-41
SLIDE 41

Mixing

Auxiliary game

p B

A B pA

D

pC

zBC

zAB

D

zAB

v

pA

v

C

Jonathan Vogel () Spatial Competition 11/14 18 / 32

slide-42
SLIDE 42

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies

Jonathan Vogel () Spatial Competition 11/14 19 / 32

slide-43
SLIDE 43

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies Let πA0

i

(E [π0

i]) denote …rm i’s pro…t in the auxiliary game (expected

pro…t in the real game) if i unilaterally deviates

Jonathan Vogel () Spatial Competition 11/14 19 / 32

slide-44
SLIDE 44

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies Let πA0

i

(E [π0

i]) denote …rm i’s pro…t in the auxiliary game (expected

pro…t in the real game) if i unilaterally deviates I prove that there exists a φ > 0 s.t. if ki 2 [k, k + φ] for all i:

Jonathan Vogel () Spatial Competition 11/14 19 / 32

slide-45
SLIDE 45

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies Let πA0

i

(E [π0

i]) denote …rm i’s pro…t in the auxiliary game (expected

pro…t in the real game) if i unilaterally deviates I prove that there exists a φ > 0 s.t. if ki 2 [k, k + φ] for all i:

1

πA

i

= π

i

Jonathan Vogel () Spatial Competition 11/14 19 / 32

slide-46
SLIDE 46

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies Let πA0

i

(E [π0

i]) denote …rm i’s pro…t in the auxiliary game (expected

pro…t in the real game) if i unilaterally deviates I prove that there exists a φ > 0 s.t. if ki 2 [k, k + φ] for all i:

1

πA

i

= π

i

2

No pro…table dev. in auxiliary game: πA

i

πA0

i

(with strict inequality if τ > 0)

Jonathan Vogel () Spatial Competition 11/14 19 / 32

slide-47
SLIDE 47

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies Let πA0

i

(E [π0

i]) denote …rm i’s pro…t in the auxiliary game (expected

pro…t in the real game) if i unilaterally deviates I prove that there exists a φ > 0 s.t. if ki 2 [k, k + φ] for all i:

1

πA

i

= π

i

2

No pro…table dev. in auxiliary game: πA

i

πA0

i

(with strict inequality if τ > 0)

3

Either πA0

i

E

  • π0

i

  • r π

i > E

  • π0

i

  • Jonathan Vogel

() Spatial Competition 11/14 19 / 32

slide-48
SLIDE 48

Proof strategy

Let π

i (πA i ) denote …rm i’s pro…t in the real game ("auxiliary"

game) if …rms follow eqm strategies Let πA0

i

(E [π0

i]) denote …rm i’s pro…t in the auxiliary game (expected

pro…t in the real game) if i unilaterally deviates I prove that there exists a φ > 0 s.t. if ki 2 [k, k + φ] for all i:

1

πA

i

= π

i

2

No pro…table dev. in auxiliary game: πA

i

πA0

i

(with strict inequality if τ > 0)

3

Either πA0

i

E

  • π0

i

  • r π

i > E

  • π0

i

  • =

) Either π

i > E [π0 i] or π i = πA i

πA0

i

E [π0

i]

Jonathan Vogel () Spatial Competition 11/14 19 / 32

slide-49
SLIDE 49

SPNE

De…nition

Firm i’s strategy space is Ωi and a strategy is ωi 2 Ωi Let Ωn Ω1 ... Ωn and denote ~ ω 2 Ωn by a strategy vector

Jonathan Vogel () Spatial Competition 11/14 20 / 32

slide-50
SLIDE 50

SPNE

Proposition: existence

Proposition

Suppose τ 0. For any set of parameters θ (n, t, τ, L) and k 0 there exists a φ (θ, k) > 0 such that if ki 2 [k, k + φ (θ, k)] for all i, then there is a non-empty set O 2 Ωn such that any ~ ω 2 O is a SPNE and strategies are pure along the equilibrium path for all ~ ω 2 O.

Jonathan Vogel () Spatial Competition 11/14 21 / 32

slide-51
SLIDE 51

SPNE

Proposition: characterization

Proposition

For an arbitrary order in which …rms locate, label any …rm 0 and label subsequent …rms in a clockwise direction (to …rm n 1). This order corresponds to an equilibrium in O. For any ~ ω 2 O the distance between each pair of neighbors, …rms i and i + 1, is d

i,i+1 = 1

n + 2 3t + 2τ

  • ¯

k ki + ki+1 2

  • Firm i’s price, market share, and pro…t are

p

i

= (t + τ) 1 n + 2 3t + 2τ ¯ k

  • +

t 3t + 2τki x

i

= 1 n + 2 3t + 2τ (¯ k ki) π

i

= Lt (x

i )2

Jonathan Vogel () Spatial Competition 11/14 22 / 32

slide-52
SLIDE 52

Equilibrium description

Distance adjusts

Suppose there are four …rms: two relatively unproductive …rms and two productive …rms

Jonathan Vogel () Spatial Competition 11/14 23 / 32

slide-53
SLIDE 53

Equilibrium description

Distance adjusts

Suppose there are four …rms: two relatively unproductive …rms and two productive …rms The two productive …rms could be separated by the unproductive …rms:

X X

X X Jonathan Vogel () Spatial Competition 11/14 23 / 32

slide-54
SLIDE 54

Equilibrium description

Distance adjusts

Suppose there are four …rms: two relatively unproductive …rms and two productive …rms The two productive …rms could be separated by the unproductive …rms:

X X

X X

The two productive …rms could neighbor each

  • ther:

X X

X X Jonathan Vogel () Spatial Competition 11/14 23 / 32

slide-55
SLIDE 55

Equilibrium description

1

Isolation between two neighbors is strictly decreasing in their average marginal cost ki +ki+1

2

Jonathan Vogel () Spatial Competition 11/14 24 / 32

slide-56
SLIDE 56

Equilibrium description

1

Isolation between two neighbors is strictly decreasing in their average marginal cost ki +ki+1

2

2

More productive …rms have larger market shares; a …rm’s market share is greater than average if and only if ki < ¯ k Novel mechanism linking productivity to …rm size

Jonathan Vogel () Spatial Competition 11/14 24 / 32

slide-57
SLIDE 57

Equilibrium description

1

Isolation between two neighbors is strictly decreasing in their average marginal cost ki +ki+1

2

2

More productive …rms have larger market shares; a …rm’s market share is greater than average if and only if ki < ¯ k Novel mechanism linking productivity to …rm size

3

Firm i earns more pro…t than average if and only if ki < ¯ k

Jonathan Vogel () Spatial Competition 11/14 24 / 32

slide-58
SLIDE 58

Uniqueness

A SPNE is strict if a unilateral deviation along the equilibrium path by …rm i strictly decreases …rm i’s pro…t

This is not the standard de…nition of strict. A more accurate term would be "strict along the equilibrium path"

Jonathan Vogel () Spatial Competition 11/14 25 / 32

slide-59
SLIDE 59

Uniqueness

A SPNE is strict if a unilateral deviation along the equilibrium path by …rm i strictly decreases …rm i’s pro…t

This is not the standard de…nition of strict. A more accurate term would be "strict along the equilibrium path"

Proposition

If τ > 0 and ki 2 [k, k + φ (θ, k)] then ~ ω is a strict SPNE if and only if ~ ω 2 O.

Jonathan Vogel () Spatial Competition 11/14 25 / 32

slide-60
SLIDE 60

Uniqueness

Auxiliary game and re…nement

Given locations, …rm’s i’s best-response in prices is 2 (τ + 2t) (t + τ) pi = pi1 + pi+1 + t (di1,i + di,i+1) + 2t t + τki

Jonathan Vogel () Spatial Competition 11/14 26 / 32

slide-61
SLIDE 61

Uniqueness

Auxiliary game and re…nement

Given locations, …rm’s i’s best-response in prices is 2 (τ + 2t) (t + τ) pi = pi1 + pi+1 + t (di1,i + di,i+1) + 2t t + τki This implies the system A~ p0 =~ b0 where A 2 6 6 6 4

2(2t+τ) t+τ

1 1 1

2(2t+τ) t+τ

1 ... ... ... ... ... 1 1

2(2t+τ) t+τ

3 7 7 7 5 and bi t (di1,i + di,i+1) + 2t t + τki

Jonathan Vogel () Spatial Competition 11/14 26 / 32

slide-62
SLIDE 62

Uniqueness

Auxiliary game and re…nement

In the auxiliary game …rm i’s price is: pi = β1 (di1,i + di,i+1) + β2 (di2,i1 + di+1,i+2) + ... +δ0ki + δ1 (ki1 + ki+1) + ... Its market share and pro…t are xi = 1 2t (pi1 + pi+1 2pi + t (di1,i + di,i+1)) πi = L

  • xi (pi ki) τ
  • x2

i,i1 + x2 i,i+1

  • Jonathan Vogel

() Spatial Competition 11/14 27 / 32

slide-63
SLIDE 63

Uniqueness

Auxiliary game and re…nement

In the auxiliary game …rm i’s price is: pi = β1 (di1,i + di,i+1) + β2 (di2,i1 + di+1,i+2) + ... +δ0ki + δ1 (ki1 + ki+1) + ... Its market share and pro…t are xi = 1 2t (pi1 + pi+1 2pi + t (di1,i + di,i+1)) πi = L

  • xi (pi ki) τ
  • x2

i,i1 + x2 i,i+1

  • Re…nement intuition: want to be "centered in market share"

Jonathan Vogel () Spatial Competition 11/14 27 / 32

slide-64
SLIDE 64

Extensions

Consider both horizontal di¤erentiation and (arbitrarily many dimensions of) vertical di¤erentiation pi + t kz ik

K

k=1

k,i min j

  • pj + t kz jk

K

k=1

k,j

  • Jonathan Vogel

() Spatial Competition 11/14 28 / 32

slide-65
SLIDE 65

Extensions

Consider both horizontal di¤erentiation and (arbitrarily many dimensions of) vertical di¤erentiation pi + t kz ik

K

k=1

k,i min j

  • pj + t kz jk

K

k=1

k,j

  • Allow consumers to vary in value they place on quality, θ, where

θ 2 [θL, θH]: pi + t kz ik θzqγ

i min j

n pj + t kz jk θzqγ

j

  • Jonathan Vogel

() Spatial Competition 11/14 28 / 32

slide-66
SLIDE 66

Extensions

Consider both horizontal di¤erentiation and (arbitrarily many dimensions of) vertical di¤erentiation pi + t kz ik

K

k=1

k,i min j

  • pj + t kz jk

K

k=1

k,j

  • Allow consumers to vary in value they place on quality, θ, where

θ 2 [θL, θH]: pi + t kz ik θzqγ

i min j

n pj + t kz jk θzqγ

j

  • Prove that there exist equilibria when the cost of transportation is

convex (concave) that limit to my class of equilibria as the convexity (concavity) limits to linearity

Jonathan Vogel () Spatial Competition 11/14 28 / 32

slide-67
SLIDE 67

Empirical implementation

Central prediction is that the distance between two neighbors is a decreasing function of their average marginal cost ki +ki+1

2

Jonathan Vogel () Spatial Competition 11/14 29 / 32

slide-68
SLIDE 68

Empirical implementation

Central prediction is that the distance between two neighbors is a decreasing function of their average marginal cost ki +ki+1

2

Empirically testing this prediction requires a measure of physical productivity and a measure of distance

Jonathan Vogel () Spatial Competition 11/14 29 / 32

slide-69
SLIDE 69

Empirical implementation

Central prediction is that the distance between two neighbors is a decreasing function of their average marginal cost ki +ki+1

2

Empirically testing this prediction requires a measure of physical productivity and a measure of distance Can be tested in two types of industry:

Jonathan Vogel () Spatial Competition 11/14 29 / 32

slide-70
SLIDE 70

Empirical implementation

Central prediction is that the distance between two neighbors is a decreasing function of their average marginal cost ki +ki+1

2

Empirically testing this prediction requires a measure of physical productivity and a measure of distance Can be tested in two types of industry:

1

homogeneous good industry in which …rms are di¤erentiated by location

Jonathan Vogel () Spatial Competition 11/14 29 / 32

slide-71
SLIDE 71

Empirical implementation

Central prediction is that the distance between two neighbors is a decreasing function of their average marginal cost ki +ki+1

2

Empirically testing this prediction requires a measure of physical productivity and a measure of distance Can be tested in two types of industry:

1

homogeneous good industry in which …rms are di¤erentiated by location

2

di¤erentiated good industry

Jonathan Vogel () Spatial Competition 11/14 29 / 32

slide-72
SLIDE 72

Empirical implementation

Central prediction is that the distance between two neighbors is a decreasing function of their average marginal cost ki +ki+1

2

Empirically testing this prediction requires a measure of physical productivity and a measure of distance Can be tested in two types of industry:

1

homogeneous good industry in which …rms are di¤erentiated by location

2

di¤erentiated good industry

Examples of industries: —–ready-mixed concrete (Syverson (2004) and Collard-Wexler (2006)) —–movie theaters (Davis (2005)) —–motels (Mazzeo (2002)) —–video retail (Seim (2001)) —–eyeglass retail (Watson (2004))

Jonathan Vogel () Spatial Competition 11/14 29 / 32

slide-73
SLIDE 73

Spatial price discrimination

Framework di¤ers from previous in two respects

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-74
SLIDE 74

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-75
SLIDE 75

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Mill pricing: …rm charges one price to all consumers and consumers pay the cost of transportation

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-76
SLIDE 76

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Mill pricing: …rm charges one price to all consumers and consumers pay the cost of transportation Spatial p.d.: …rm chooses a price schedule that lists the prices that the …rm charges consumers at each location in space

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-77
SLIDE 77

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Mill pricing: …rm charges one price to all consumers and consumers pay the cost of transportation Spatial p.d.: …rm chooses a price schedule that lists the prices that the …rm charges consumers at each location in space

2

Identity of the agent that incurs the cost of transportation

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-78
SLIDE 78

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Mill pricing: …rm charges one price to all consumers and consumers pay the cost of transportation Spatial p.d.: …rm chooses a price schedule that lists the prices that the …rm charges consumers at each location in space

2

Identity of the agent that incurs the cost of transportation

Relevance of frameworks to industries

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-79
SLIDE 79

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Mill pricing: …rm charges one price to all consumers and consumers pay the cost of transportation Spatial p.d.: …rm chooses a price schedule that lists the prices that the …rm charges consumers at each location in space

2

Identity of the agent that incurs the cost of transportation

Relevance of frameworks to industries

1

Mill pricing appropriate for modeling di¤erentiation in geographic and product-characteristics space

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-80
SLIDE 80

Spatial price discrimination

Framework di¤ers from previous in two respects

1

Manner in which …rms compete in prices in the second stage

Mill pricing: …rm charges one price to all consumers and consumers pay the cost of transportation Spatial p.d.: …rm chooses a price schedule that lists the prices that the …rm charges consumers at each location in space

2

Identity of the agent that incurs the cost of transportation

Relevance of frameworks to industries

1

Mill pricing appropriate for modeling di¤erentiation in geographic and product-characteristics space

2

SPD most appropriate for geographic di¤erentiation and for di¤erentiation of intermediate inputs that must be tailored to exact speci…cations of …nal good producers

Jonathan Vogel () Spatial Competition 11/14 30 / 32

slide-81
SLIDE 81

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-82
SLIDE 82

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-83
SLIDE 83

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

2

Firm’s neighbor has no stronger e¤ect on its market share and pro…t than a distant …rm

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-84
SLIDE 84

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

2

Firm’s neighbor has no stronger e¤ect on its market share and pro…t than a distant …rm

3

More productive …rms are more isolated in product or geographic space, all else equal

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-85
SLIDE 85

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

2

Firm’s neighbor has no stronger e¤ect on its market share and pro…t than a distant …rm

3

More productive …rms are more isolated in product or geographic space, all else equal

Di¤erences

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-86
SLIDE 86

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

2

Firm’s neighbor has no stronger e¤ect on its market share and pro…t than a distant …rm

3

More productive …rms are more isolated in product or geographic space, all else equal

Di¤erences

1

Results hold not only in a neighborhood of symmetry, but for arbitrary distribution of m.c.’s

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-87
SLIDE 87

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

2

Firm’s neighbor has no stronger e¤ect on its market share and pro…t than a distant …rm

3

More productive …rms are more isolated in product or geographic space, all else equal

Di¤erences

1

Results hold not only in a neighborhood of symmetry, but for arbitrary distribution of m.c.’s

2

A unique characterization of SPNE in undominated, pure strategies without imposing any assumptions on the allocation of transportation costs

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-88
SLIDE 88

Spatial Price Discrimination

SPD relative to mill pricing: results

Similarities

1

All economically relevant …rm outcomes are uniquely determined across all SPNE in undominated strategies

2

Firm’s neighbor has no stronger e¤ect on its market share and pro…t than a distant …rm

3

More productive …rms are more isolated in product or geographic space, all else equal

Di¤erences

1

Results hold not only in a neighborhood of symmetry, but for arbitrary distribution of m.c.’s

2

A unique characterization of SPNE in undominated, pure strategies without imposing any assumptions on the allocation of transportation costs

3

Equilibria with SPD are all welfare maximizing (solve social planner’s prob)

Jonathan Vogel () Spatial Competition 11/14 31 / 32

slide-89
SLIDE 89

Conclusions

Di¤erences in productivity are re‡ected in location decisions through isolation

Jonathan Vogel () Spatial Competition 11/14 32 / 32

slide-90
SLIDE 90

Conclusions

Di¤erences in productivity are re‡ected in location decisions through isolation This is an important margin that has been mostly ignored for technical reasons

Jonathan Vogel () Spatial Competition 11/14 32 / 32

slide-91
SLIDE 91

Conclusions

Di¤erences in productivity are re‡ected in location decisions through isolation This is an important margin that has been mostly ignored for technical reasons Whether predictions are borne out remains to be seen

Jonathan Vogel () Spatial Competition 11/14 32 / 32