%jtdsjnjobujpoboeqvcmjdqpmjdjft/2441/1iuripn2a39q5bd6h... · 2021. 1. 28. · jtbusjdf +...

1

25

R2

Figure 1.1: Added-variable plot

Figure 1.2: Rank-ordered logit on gender of candidates by year

Figure 1.3: Low-skilled workers Figure 1.4: High-skilled workers

Figure 1.5: Low-skilled workers Figure 1.6: High-skilled workers

2

.1.2

.3.4

.5R

atio

of f

orei

gner

s in

squ

ads

1981/82 2008/091994/95Season

England Other European Countries

Notes: Other European Countries are Denmark, France, Germany, Greece, Italy, Netherlands, andSpain. The vertical axis indicates the pre-Bosman ruling year.

Figure 2.2: Club'sWage Bills

Elasticity = .17 (.002)

R-squared = .91

1315

1719

Clu

b's

wag

e bi

ll (in

logs

)

1981 1984 1987 1990 1993 1996 1999 2002 2005 2008Notes: Each dot stands for one club. The regression line is depicted. Elasticity coefficient from theOLS regression of the log-club's wage bill on year dummies is reported with standard error inparenthesis.

English First League (1981-2008)

Figure 2.3: AverageWage Bill and Team Success

AAAAAAAAAAAAAAAAAAAAAAAAAAAA

A A A A A A A A A A A A A A A A A A A A A A A A A A A

By

BghBghBghBghBghBghBghBghBgh

BuBuBuBuBuBuBuBuBuBuBuBuBuBuBu

BBBBBBBBBB

BddBdd

Bgh & Hv ABgh & Hv Ahhhhhhhhhhhh

hhhhhhhhhhhhhhhhh

vyvyvyvyvyvyvyvyvyvyvyvyvyvyvyvyvyvyvyy y y y y y y uyy uyy uyy uyy uyy uyy uyy uyy uyy uyy uy

vvvvvvvvvvvvvvvvvvvvvvvvvvvv

uhuhuhuhuhuhuhuh

Hu

whwhwhwhwhwhwhwhwhwh

d dd dd dd dd dd dd dd dd dd dd dd dd dd dd d

vvvvvvvvvvvvvvvvvvvvvvvvvvvv

uuuuuuuuuuh yh yh yh yh yh yh yh yh yh yh yh yh yh yh yh yh y

ddughddughddughddughddughddughddughddughddughddughddughddughddughddughddughddughw w

wwwwwwwwwwwwwwwwwwwww

wh ywh ywh ywh ywh ywh ywh ywh ywh ywh ywh ywh ywh y

gh gh gh gh gh gh gh gh gh gh gh gh gh gh gh gh

uy uy uy uy

dh Ah dh Ah dh Ah d dd dd d

uhuhuhuhuhuhuh

dgdghd dhd dhd dhd dhd d

hd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddyhd ddy

uhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuhuh

udduddudduddudduddudduddudduddudd

w yw y

wd

hhhhhhhhhhhhhhhhhhhhhhhhhhh

dddddddd

Bwh Bwh Bwh Bwh Bwh Bwh Bwh Bwh Bwh

H H H H H H H H H H H H H H H H H H H H H H

gggg

d d d d d d d d d d d d d d

vv

-ud .85

h d

0.2

.4.6

.81

Av

g

g

-1.3 -.7 -.1 .5Avg v wg ( g)

Notes: Avg g h vg u' 1981 2008. u i ud h y t [(g - )/ ( - )], whg d h w d hgh g t.Avg v wg h vg u' v wg 1981 2008. hv wg u i y t h g d h u' wg .v h u vg.

gh gu (1981-2008)

Table 2.1: Individual Differences inMeans: Black vsWhite English Players

c a

b a

a a a

a a a

a b c

Table 2.2: ClubDifferences inMeans

a

a

a

a

it = αi + β1( it − t)

+ β2( it − t)

+ β3( it − t) + ϵit,

it

(Rankingit−mint

maxt−mint

)Rankingit

i t

αi i

ϵit

( it − t)

( it − t)

i t

it − t

i t

β3

β3

ijt = ξij + β1 ( it/ jt)

+ β2( it − jt)

+ β3( it − jt) + eijt,

ijt i j

t(

it/ jt

)

it− jt

ξij eijt

αi

Table 2.3: DiscriminationMarket-test: League Success and Long Span

a a a b

a a a a

a b

R2

(Rankingit−mint

maxt−mint

)Rankingit

i ta b

ij i j ji

∆ β3

β1

[(∆ ∗ β3/β1) + ]−

Table 2.4: DiscriminationMarket-test: Match Success and Short Span

c c b a

a a a b

a b

a b c

Table 2.5: Post-Bosman: Are non-EU Black Players Discriminated? (1996-2008)

a a b

a a a a

a a c c

(Rankingit−mint

maxt−mint

)Rankingit

i ta b c

= B

B

− A

A

.

Figure 2.4: Relative Turnover of Black English Players

−.05

0.0

5.1

Turn

over

of b

lack

w.r.

t. w

hite

pla

yers

1980 1985 1990 1995 2000 2005 2010Year

Table 2.6: DiscriminationMarket-test: League Success and Long Span - IV and GMM

c a a

a a c

b b

(Rankingit−mint

maxt−mint

)Rankingit

i ta b c

Table 2.7: Share of Black Players and Corporate Control

a

a b c

Table 2.8: DiscriminationMarket-test: Effect of Corporate Control

a a b b

a a a a

b c

a

(Rankingit−mint

maxt−mint

)Rankingit

i ta b c

Table 2.9: DiscriminationMarket-test: Heterogeneity of clubs

(Rankingit−mint

maxt−mint

)Rankingit

i ta b c

A

Figure A.1: AverageWage Bill and Team's Quality

ArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenalArsenal

Aston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston VillaAston Villa

Barnsley

BirminghamBirminghamBirminghamBirminghamBirminghamBirminghamBirminghamBirminghamBirmingham

BlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburnBlackburn

BoltonBoltonBoltonBoltonBoltonBoltonBoltonBoltonBoltonBolton

BradfordBradford

Brighton & Hove AlbionBrighton & Hove Albion

CharltonCharltonCharltonCharltonCharltonCharltonCharltonCharltonCharltonCharltonCharltonCharlton

ChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelseaChelsea

CoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCoventryCrystal PalaceCrystal PalaceCrystal PalaceCrystal PalaceCrystal PalaceCrystal PalaceDerby CountyDerby CountyDerby CountyDerby CountyDerby CountyDerby CountyDerby CountyDerby CountyDerby CountyDerby CountyDerby County

EvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEvertonEverton

FulhamFulhamFulhamFulhamFulhamFulhamFulhamFulham

Hull

IpswichIpswichIpswichIpswichIpswichIpswichIpswichIpswichIpswichIpswich

Leeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds UnitedLeeds United

LeicesterLeicesterLeicesterLeicesterLeicesterLeicesterLeicesterLeicesterLeicesterLeicesterLeicester

LiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpoolLiverpool

LutonLutonLutonLutonLutonLutonLutonLutonLutonLuton Manchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityManchester CityMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbroughMiddlesbrough

Millwall FCMillwall FC

NewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastleNewcastle

Norwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich CityNorwich City

Nottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham ForestNottingham Forest

Notts CountyNotts CountyNotts CountyNotts CountyOldham Athletic FCOldham Athletic FCOldham Athletic FC

Oxford UnitedOxford UnitedOxford UnitedPortsmouthPortsmouthPortsmouthPortsmouthPortsmouthPortsmouthPortsmouth QPRQPRQPRQPRQPRQPRQPRQPRQPRQPRQPRQPRQPR

ReadingReadingSheffield UnitedSheffield UnitedSheffield UnitedSheffield UnitedSheffield United

Sheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield WednesdaySheffield Wednesday

SouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonSouthamptonStokeStokeStokeStokeStokeSunderlandSunderlandSunderlandSunderlandSunderlandSunderlandSunderlandSunderlandSunderlandSunderlandSunderland

Swansea CitySwansea City

Swindon

TottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenhamTottenham

WatfordWatfordWatfordWatfordWatfordWatfordWatfordWatfordWest BromwichWest BromwichWest BromwichWest BromwichWest BromwichWest BromwichWest BromwichWest BromwichWest Bromwich

West HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest HamWest Ham

WiganWiganWiganWiganWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FCWimbledon FC

WolvesWolves

R-squared = .81 Manchester United

-1.3

-1-.7

-.4-.1

.2.5

.8Av

erag

e re

lativ

e w

age

bill

(in lo

gs)

-20 -10 0 10 20Average relative quality level

Linear fit

Notes: Average relative wage bill is the average club's relative wage bill from 1981 to 2008.The relative wage bill of a club i in year t is the log difference of the club's wage bill relative to theannual average. Average relative quality level is the average club's quality from 1981 to 2008. Theclub's relative quality level is computed as the difference of the club's quality level relative to theannual average.

English First League (1981-2008)

R

L A

B B

λ

δ

r

ω t

ω = t

ω = kt

0 < k ≤ 1

(1−k)t

λγ γ

c(t)

JI

(1− k)t δ

λγ

rJI = (1− k)t− δJI − λγ(JI + c(t)),

⇔ JI =(1− k)t− λγc(t)

r + δ + λγ.

t JI JP

ω = t

rJP = t− ω − δ(JP ) ⇔ JP = 0.

JP > JI ⇔ 0 >(1− k)t− λγc(t))

r + δ + λγ,

⇔ c(t) >(1− k)t

λγ.

γ

c c

α ≥ 0

G(c) =(cc

)−α, c ∈ [c,∞],

α

c α

c

c(t) = ct γ

γ(t) =

(cλγ

1− k

)α

=

(cλ

1− k

) α1−α

.

γ λ

k λ

λ

k

λθ 0 ≤ θ ≤ 1

θ < 1 γ

θ

λ

λ

λ

λ

γ λ

γ

B B

B

R Ω C

c n

T

T ≡ (n− 1)t+ t.

n − 1 t

t φ n − 1

λ A B

n − 1

T

t

R = Ω + C = (n− 1)[φkt+ (1− φ)t] + t+ (n− 1)φλ(θµγB + (1− µ)γA)ct

= (n− 1)t[φ(k + cλ(θµγB + (1− µ)γA)) + 1− φ] + t,

φ ω < t µ

φ

c

c ≥ (1−k)λθγB

t φ = 0

(1−k)λγA

≤ c ≤ (1−k)λθγB

φ

µ

t = R− µt× (n− 1)(k + λθγBcB)− (1− µ)t× (n− 1).

λ

c ≤ (1−k)λγA

φ = 1

t = R− (n− 1)t[k − 1 + λc(θµγB + (1− µ)γA)]λ

λ

T

λ

µ

R = Ω + (n − 1)tµλθγBc

⎧⎪⎪⎨

⎪⎪⎩

T = Ω

T = Ω + µt(N − 1) ∗ (1− k)

λ

Ω λ

Ω µ

T

Figure A.2: Pre-Bosman Figure A.3: Post-Bosman

Table A.1: DiscriminationMarket-test: Club's League Success in France and Belgium

a a a a

a a

c

b c a a

a b c

Table A.2: DiscriminationMarket-test: Match Success and Long Span -Within andWithin-IV

a a a

a a a a

a a

a

Table A.3: First Stages: League Success and Long Span

c

a a

b

a a

a a

χ2 a a

χ2

a b c

Table A.4: First-Stages: Match Success and Long span

a

a a

a b

a a

a a

χ2 a a

χ2

a b

Table A.5: Structural Break: Long Span

a a b a

a b b b

a b b b

a a a c

(Rankingit−mint

maxt−mint

)Rankingit

i ta b

Table A.6: Structural Break: Short Span

b a

c

c c

b a

a a

a b

c

3

j Uij i

Uij = qij + µgigj

qij

µ gi

gj

µ

1st 2nd 2nd 3rd

i

M i′ M − 1

Uij qij

qij = xijβ + ηj + ϵij

xij ηj ϵij

q∗ij k

rij i k ϵij

Pr(ϵij < u) = e−e−u

jth

Pr(qr1jj > qr2jj > . . . > qrMjj)

lj(β) =M−1∏

i:q∗ij=1

(xijβ + µgigj)∑Mi′:ri′j≥rij (xi′jβ + µgi′gj)

ηj

J∑

j=1

M−1∑

i:q∗ij=1

xijβ + µgigj −J∑

j=1

M−1∑

i:q∗ij=1

(M∑

i′:ri′j≥rij

(xi′jβ + µgi′gj))

β µ µ

i i′

Uij ≥ Ui′j

qi ≥ qi′ − µgigj + µgi′gj

qi

xi ϵij ∼ N(0, σ)

i

i′ Y

Pr(Y = 1|x) = Pr(U(i) ≥ U(i′))

Pr(xiβ + µgigj + ϵij ≥ xi′β + µgi′gj + ϵi′j)

Pr(β(xi − xi′) + µ(gi − gi′)gj ≥ ϵi′j − ϵij)

= Φ(Xβ + µGgj)

X = xi − xi′ G = gi − gi′

gi = gi′ G = 0 µ

χ2

Table 3.1: Some descriptive statistics

Table 3.2: Recruitments and gender

Table 3.3: Descriptive statistics on hires and applicants

Table 3.4: Descriptive statistics on dyads

xi

Table 3.5: Descriptive statistics on academic connections

Table 3.6: Effect of the reform on themean share of women jurors and number of female presidents

Table 3.7: Effect of the reform on themean h-index of jurors

Table 3.8: Correlation of the gender of jurors on the probability of women being first-ranked

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.9: Correlation of the gender of jurors and the rank of candidates

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.10: Correlation of the gender of jurors and the probability that a woman is better rankedwithin a dyad

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.11: Correlation of the gender of jurors and the probability that a woman is better rankedwithin a dyad. Split by sub-

ject

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.12: Regression of the gender composition of applicant pools on the gender composition of the recruitment committee

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.13: Effect of the reform on the candidate pool

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.14: Effect of the quota on the rank of female candidates

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.15: IV estimate of the effect of the increase in women jurors on the probability that a woman is better rankedwithin a

dyad

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 3.16: Rank-Ordered Logit using the quota: top 3 ranks only

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Figure 3.1: Rank-ordered logit on gender of candidates by year

Figure 3.2: Effect of the reform on the ranks of women by discipline. Disciplines most affected by the reform are ordered

from left to right.

Table 3.17: Effect of the reform by gender of the jury president

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

B

Wij = Xβ +m(Ggj) + ϵij

Wij

β

β

gj

Figure B.1: Semi-parametric estimation of the relationship of interest

Uij

Uij = βqij + µgigj + ϵij

gi gj qij

β µ

Uij

rij

Uij

µ/β

µ

ϵij

Table B.1: Power of estimationmethods

≤

≥

µ/β β

β µ

Xβ + µGij

Table B.2: Estimates from simulations

µ β

Table B.3: Estimates from simulations: large parameters

µ β

Table B.4: Probability thatW is more highly ranked

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.5: Probit results, Difference-in-Difference

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.6: OLS results, Difference-in-Difference

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.7: First stage: IV

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.8: Rank-Ordered Logit using the quota

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.9: Rank-Ordered Logit using the quota

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.10: IV Probit on gender of first-ranked candidate

ρ

σ

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table B.11: Rank-Ordered Logit using the quota: Other publicationmeasures

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

4

Figure 4.1: Low-skilled workers Figure 4.2: High-skilled workers

Figure 4.3: Black workers/Total population perMSA in 1980

Figure 4.4: Black workers/Total population perMSA in 2000

Figure 4.5: Low-skilled workers Figure 4.6: High-skilled workers

i

Hit Lit

Yit = (θHitHρit + θLitL

ρit)

1ρ

ρ 1 − 1σHL

σHL

θ

Hit Lit Hit = (ηtHWit + HB

it ) Lit = (φtLWit + LB

it)

φt ηt

Pt

WHWit =

∂PtYit

∂HWit

= PtθHitH

ρ−1it Y 1−ρ

it ηt

wHWit = ln(

WHWit

Pt) = ln(θHit ) + (ρ− 1)ln(Hit) + (1− ρ)ln(Yit) + ln(ηt)

θit = F (Hit, Lit)

∆wHWit = γHHW∆ln(Hit) + γHLW∆ln(Lit) +∆ln(ηt) +∆ϵθHW

it

∆wHBit = γHHB∆ln(Hit) + γHLB∆ln(Lit) +∆ϵθHB

it

∆wLWit = γLHW∆ln(Hit) + γLLW∆ln(Lit) +∆ln(φt) +∆ϵθLWit

∆wLBit = γLHB∆ln(Hit) + γLLB∆ln(Lit) +∆ϵθLBit

∆ϵθit

Tit = WWLψW

it WBLψB

it exp(ϵTit)

∆ln(Tit) = ψW∆wLW + ψB∆wL

B +∆ϵTit

ψ < 0

HD

P hit = f(CCit, PL(HD))

P hit CCit

PL

P hit =

Riti i

∆ln(Rit) = µi∆ln(HDit) +∆ϵCCit

µ µ0+µlandxlandi +µregxreg

i

HD

HD =∑

z ζzZz

itWzit Zit ζz

ϵCCit

µ

j

U zijt = ln(G1−ζz

jt ) + ln(Qζz

jt ) + uj(Dit)

Gjt ∗ Pt +Qjt ∗Rit ≤ W zit

ζ Qjt

Gjt Pt

D

ζ

j vjit

vjit = wzit − ζzrit + uj(Dit)

wzit = ln(W z

it/Pt) rit = ln(Rit/Pt) W zit = sWW z

it +

sLTit sW sL

uj(Dit)

uj(Dit) = aitβz,at + βz,st

t xz,stj + βz,div

t xz,divj + ωz(

Hzit + Lz

it

Hit + Lit) + ϵz,City

it + ϵz,Aijt

βstxst βdivxdiv

(aitβz,at )

ωz(Hzit+Lz

itHit+Lit

i

δ

δzit = λz(wzit − ζzrzit) + ωz(

Hzit + Lz

it

Hit + Lit) + βz,aait + ϵz,City

it

vjit = δzit + βz,stt ∗ xz,st

j + βz,divt xz,div

j + ϵAijt

vjit ≥ vjkt∀k ϵAijt

Pr(vjit ≥ vjkt∀k) =exp(δzit + βz,st

t xz,stj + βz,div

t xz,divj )

1 + Σni=1exp(δ

zit + βz,st

t xz,stj + βz,div

t xz,divj )

δz

δz

∆δzit = λz(∆wzit − ζz∆rzit) + ωz∆(

Hzit + Lz

it

Hit + Lit) + βz,a∆ait +∆ϵz,City

it

ω

σzj

λz

∆ait = βHL∆ln(Hit

Lit +Hit) + βBW∆ln(

Bit

Wit + Bit) +∆ϵEA

it

Bit = HBit + LB

it Wit = HWit + LW

it

∆wzit = γHWZ∆lnHit + γHLZ∆lnLit +∆ηzt +∆ϵwZ

it

Zit = Σj∈zexp(δzit + βz,st

t xz,stj + βz,div

t xz,divj )

1 + Σni=1exp(δ

zit + βz,st

t xz,stj + βz,div

t xz,divj )

∆δZit = λzit(sW∆wzit + sT ∗∆tit − ζz∆rit) + ωz∆(

Hzit + Lz

it

Hit + Lit) + βz,a

t ∆ait +∆ϵz,Cityit

∆ait = βHL∆ln(Hit

Lit) + βBW∆ln(

Bit

Wit) +∆ϵEA

it

∆ln(Rit) = µi∆ln(HDit) +∆ϵCCit

∆ln(Tit) = ψW∆wLW + ψB∆wL

B +∆ϵTit

Table 4.1: Amenity index

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Bzit =

N∑

r=0

(wzrt − wz

rt−1)Zirt−1

Zit−1

i

R2

Table 4.2: Are the Bartik shocks good instruments?

∆ ∆

× ∗∗∗ ∗∗∗

× ∗∗∗

∗∗∗ ∗∗∗

∗∗∗

R2

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

η φ

ζ

ζHW = 0.37, ζLW = 0.40, ζHB = 0.35, ζLB = 0.45

ζ

ζ

ζ

ρ

11−ρ

Table 4.3: Labour demand parameters

ρ

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 4.4: Transfers andHousing Supply

µ0

µlandavailability

µregulation

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01 µ0

γ

ζ

Table 4.5: Sensitivity to wages and rents

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

ζ

ω

Table 4.6: Homophily and amenities

ωHB

ωLB

ωHW

ωLW

ω

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 4.7: Network amenities

δ

ωz

δzit = λz(wzit − ζzrzit) + ϵit

ϵit

δ

R2

Table 4.8: Racial wage gap

Table 4.9: Racial real wage gap

Figure 4.7: Difference between the true and counterfactual populations with no homophily, as a function of the initial shares

of black andwhite workers.

δit+1

δ

δzit = λzit(sW × (wzit−1 + Bz

it) + sT × tit − ζzrit−1) + ωz(Hz

it−1 + Lzit−1

Popit−1) + ϵit−1

ϵit−1 tit = tit−1 + (ψW + ψB)BLit

δ

ϵCCit µ

rit = µ× ln(HDit) + ϵCCit

rit

ˆHDit =∑

z

ζzZit(wzit−1 + Bz

it)

rit = µ× ln( ˆHDit) + ϵCCit

Zit

δ

δzit = λzit(sW × (wzit−1 + Bz

it) + sT × tit − ζz rit) + ωzln(Hz

it−1 + Lzit−1

Popit−1) + ϵit−1

Zit δ

Hzit + Lz

it

immit + ΣzZit

δ

δzit = λzit(sW × (wzit−1 + Bz

it) + sT × tit − ζz rit) + ωzln(Hz

it + Lzit

immit + ΣzZit

) + ϵit−1

Table 4.10: Response to shocks of the share of white workers in a city.

∆ ∆ ∆∗ ∗ ∗∗

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table 4.11: Response to shocks of the share of black workers in a city.

∆ ∆ ∆∗∗∗ ∗∗∗ ∗∗

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

C

Figure C.1: Low-skilled workers Figure C.2: High-skilled workers

Yit = (θhtHρit + θltL

ρit)

1ρ

Figure C.3: Low-skilled workers Figure C.4: High-skilled workers

H = (θhbtHBξit + θhwtH

W ξit )

1ξ , L = (θlbtL

Bφit + θlwtL

Wφit )

1φ

H L

θ ρ ξ φ

ρ = 1− 1/σHL

∂Yit

∂Hit= θhtH

ρ−1it Y 1−ρ

t

∂Hit

∂HWit

= θhwtHW ξ−1it H1−ξ

it

wHWit =

∂Yit

∂Hit

∂Hit

∂HWit

= θhtHρ−ξit Y 1−ρ

t θhwtHW ξ−1it

ξ

Table C.1: %of residents who are black in 1980

ln(wHB

it

wHWit

) = ln(θhbtθhwt

) + (ξ − 1)ln(HB

it

HWit

)

Table C.2: %of residents who are black in 2000

ξ φ

ξ φ

ξ − 1 = − 1σHWB

Table C.3: Estimation of elasticities of substitution

ln(wB/wH) ln(wB/wH

∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Table C.4: Are the Bartik shocks good instruments?

∆wHW ∆wLW ∆wHB ∆wLB ∆rit

∗∗∗ ∗∗∗

× ∗∗ ∗∗∗

×

× ∗∗

×

× ∗∗ ∗ ∗

× ∗∗∗ ∗∗∗ ∗∗ ∗∗∗

× ∗∗∗ ∗∗∗ ∗∗ ∗∗ ∗∗

R2