Download - Ganong Slides - Brookings Institution
Why Has Regional Income Convergence in the U.S.Declined?
Peter Ganong (UChicago) and Daniel Shoag (Harvard)
July 2016
AL
AZ
AR
CA
CO
CT
DE
FL
GA
ID
IL
IN
IA
KSKY
LA
ME
MDMA
MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WVWI
WY
11
.52
5 6 7Log Income Per Capita, 1940
1940-1960, Coef: -.488 SE: .019
Change in Log Income Per Capita
AL
AZ
AR
CA
CO
CT
DE
FL
GA
ID
IL
IN
IA
KSKY
LA
ME
MDMA
MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WVWI
WY
11
.52
5 6 7Log Income Per Capita, 1940
1940-1960, Coef: -.488 SE: .019
AL
AZ
AR
CA
CO
CT
DE
FLGAID
ILIN
IAKS
KY
LA
MEMD
MA
MI
MN
MS
MO
MTNE
NV
NHNJ
NM
NYNC
ND
OH
OK
ORPA
RI
SC
SD
TN
TXUT
VT
VAWAWV
WI
WY
.4.9
1.4
9.4 10.2Log Income Per Capita, 1990
1990-2010, Coef: -.198 SE: .057
Change in Log Income Per Capita
-.6
-.4
-.2
0
1950 1980 2010
Coef: dInc = b*inc_t-20
Timeseries of Coefficients - Convergence
AL
AZ
AR
CA
CO
CT
DE
FL
GA ID IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NYNC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WV
WI
WY
0.5
1
5 6 7Log Income, 1940
1940-1960, Coef: .33 SE: .07
Change in Log Population
AL
AZ
AR
CA
CO
CT
DE
FL
GA ID IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NYNC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WV
WI
WY
0.5
1
5 6 7Log Income, 1940
1940-1960, Coef: .33 SE: .07
AL
AZ
AR CA
CO
CT
DE
FLGA
ID
IL
IN
IA
KSKY
LA ME
MD
MAMI
MN
MSMO
MT
NE
NV
NH
NJ
NM
NY
NC
ND OH
OK
OR
PARI
SC
SD
TN
TX
UT
VT
VA
WA
WV
WI
WY
0.5
1C
ha
ng
e in
Lo
g P
op
ula
tio
n,
19
90
-20
10
9.4 10.2Log Income, 1990
1990-2010, Coef: -.09 SE: .12
Change in Log Population
-.6
-.4
-.2
0.2
.4
1950 1980 2010
Coef: dInc = b*inc_t-20 Coef: dPop = b*inc_t-20
Timeseries of Coefficients - Convergence and Pop
-.6
-.4
-.2
0.2
.4
1950 1980 2010
Coef: dInc = b*inc_t-20 Coef: dPop = b*inc_t-20
Timeseries of Coefficients - Convergence and Pop
-.6
-.4
-.2
0.2
.4
1950 1980 2010
Coef: dInc = b*inc_t-20 Coef: dPop = b*inc_t-20
Timeseries of Coefficients - Convergence and Pop
Outline
1 Facts About Housing Prices
2 How Does Housing Supply Affect Convergence? New Measure ofSupply Restrictions
Outline
1 Facts About Housing PricesI Differences in Housing Prices Have Grown Relative to Differences in
IncomesI Housing Prices Have Lowered the Returns to Living in Productive
Places For Unskilled WorkersI Migration Flows Respond to Skill-Specific Gains Net of Housing Prices
2 How Does Housing Supply Affect Convergence? New Measure ofSupply Restrictions
Differences in Housing Prices
AL
AR
AZ
CA
CO
CT
DEFL
GAIA
ID
IL
IN
KSKY
LA
MA
MD
ME
MIMN
MO
MS
MT
NC
NDNE
NH
NJ
NM
NVNY
OH
OK
ORPA
RI
SC
SDTN
TX
UT
VA
VT
WAWI
WV
WY
10.5
1111
.512
Log
Hous
ing
Valu
e, 1
960
9.2 9.4 9.6 9.8 10Log Income Per Cap, 1960
1960, Coef: .95 SE: .08
AL
AR
AZ
CA
CO
CT
DE
FL
GA
IA
ID
IL
IN KSKY
LA
MA
MD
ME
MI
MN
MO
MS
MT
NC
ND
NE
NH
NJ
NM
NV
NY
OH
OK
OR
PA
RI
SC
SD
TNTX
UT
VA
VT
WA
WI
WV
WY
11.5
1212
.513
Log
Hous
ing
Valu
e, 2
010
10.3 10.5 10.7 10.9 11.1Log Income Per Cap, 2010
2010, Coef: 2.04 SE: .25
Changes in Returns to Migration0
.51
1.5
Coe
f
1940 1960 1970 1980 1990 2000 2010
Unskilled HH Skilled HH
Effect of $1 of Statewide Inc on Skill-Specific Inc Net of Housing
Migration Flows – 2000
-40
48
Net
Mig
ratio
n as
% P
op
10.8 11 11.2 11.4 11.6 11.8Log Nominal Income
Low Skill Coef: -2.17 SE: 1.00
-40
48
Net
Mig
ratio
n as
% P
op
10.8 11 11.2 11.4 11.6 11.8Log Nominal Income
High Skill Coef: 4.07 SE: .69
-40
48
Net
Mig
ratio
n as
% P
op
10.4 10.6 10.8 11 11.2Log (Inc-Housing Cost) for Low Skill
Low Skill Coef: 4.30 SE: 2.00
-40
48
Net
Mig
ratio
n as
% P
op
11 11.2 11.4 11.6 11.8Log (Inc-Housing Cost) for High Skill
High Skill Coef: 4.71 SE: .89
Outline
1 Facts About Housing Prices
2 How Does Housing Supply Affect Convergence? New Measure ofSupply Restrictions
A New Regulation Measure
ConstructionI Number of hits per capita from state appeals courts for “land use”I Omnibus measure. Captures many different anti-development tactics.
PropertiesI correlated with cross-sectional measures
F American Institute of Planners, 1975F Wharton Residential Land Use Regulatory Index, 2005
I correlated with prices, conditional on state and year fixed effects
Look at patterns separately for low elasticity and high elasticity states
01
23
Cas
es P
er M
illion
Peo
ple
1940 1950 1960 1970 1980 1990 2000 2010
Land Use Cases Per Million People
How Regulations Affect Convergence
The previous graphs split states at the median within each year.We can instead exploit the regulation measure to fix a cutoff level ofregulation.
Regs,t =Percentile{LandUseCasesstPopst
}
Ys,t = αt +αtRegs,t +β Incs,t +βhigh regInc i ,t−1×Regs,t + εs,t
Dependent variables: log housing prices, population, human capital(growth accounting measure from a Mincerian regression), andincome.
How Regulations Affect Convergence
Dep Var lnphouse d lnpop d lnHumanCapital d ln Inc
Log Income 0.77***(0.11)
Log Income 0.83****Reg (0.26)
N 384
Indicators for Year and Year*Reg in all specifications.
How Regulations Affect Convergence
Dep Var lnphouse d lnpop d lnHumanCapital d ln Inc
Log Income 0.77*** 1.69**(0.11) (0.64)
Log Income 0.83*** -1.88****Reg (0.26) (0.61)
N 384 2,448
Indicators for Year and Year*Reg in all specifications.
How Regulations Affect Convergence
Dep Var lnphouse d lnpop d lnHumanCapital d ln Inc
Log Income 0.77*** 1.69** -0.043***(0.11) (0.64) (0.007)
Log Income 0.83*** -1.88*** 0.040***Reg (0.26) (0.61) (0.016)
N 384 2,448 288
Indicators for Year and Year*Reg in all specifications.
How Regulations Affect Convergence
Dep Var lnphouse d lnpop d lnHumanCapital d ln Inc
Log Income 0.77*** 1.69** -0.043*** -2.03***(0.11) (0.64) (0.007) (0.10)
Log Income 0.83*** -1.88*** 0.040** 1.30****Reg (0.26) (0.61) (0.016) (0.39)
N 384 2,448 288 2,448
Indicators for Year and Year*Reg in all specifications.
Regime One
Dep Var lnphouse d lnpop d lnHumanCapital d ln Inc
Log Income 0.77*** 1.69** -0.043*** -2.03***(0.11) (0.64) (0.007) (0.10)
Log Income 0.83*** -1.88*** 0.040** 1.30****Reg (0.26) (0.61) (0.016) (0.39)
N 384 2,448 288 2,448
Indicators for Year and Year*Reg in all specifications.
Regime Two
Dep Var lnphouse d lnpop d lnHumanCapital d ln Inc
Log Income 0.77*** 1.69** -0.043*** -2.03***(0.11) (0.64) (0.007) (0.10)
Log Income 0.83*** -1.88*** 0.040** 1.30****Reg (0.26) (0.61) (0.016) (0.39)
N 384 2,448 288 2,448
Indicators for Year and Year*Reg in all specifications.
Omitted Variable Bias
More generally, could there be a post-1980 shock that raisedregulation levels and lowered convergence?Use pre-1980 measures of housing supply elasticity
Omitted Variable Bias
Dep Var: Log Income Per Cap t - Log Income Per Cap t-20
Constraint: Regs in 2005 Regs in 1965 Buildable LandPre Post Pre Post Pre Post
Log Income -1.93*** -1.80*** -2.05*** -1.97*** -2.49*** -1.20***(0.11) (0.33) (0.15) (0.47) (0.06) (0.08)
Log Income 0.22 2.01*** 0.20 1.91*** -0.09 0.71****Constraint (0.27) (0.66) (0.27) (0.69) (0.10) (0.17)
N 1,248 1,200 1,248 1,200 8,413 9,194Geo State State State State County County
Indicators for Year and Year*Constraint in all specifications.
Conclusion
Weakening of directed migration and convergence in last 30 yearsDirected migration drove convergence:
I Regional labor markets clear through skill-sorting rather than netmigration. Can be explained by supply shifts.
I Continued convergence in places with unconstrained supply
State-level panel measure for housing regulations.I We are happy to share this with other researchers.
Model Overview
Model with multiple skill types, endogenous migration, andpotentially restricted housing supply.Proposition 1: Directed migration drives convergenceProposition 2: Housing prices affect migration differently by skill level
Substantive Expositional
Regional labor demand is Y = AL1−α
downward sloping
Land is an Inferior Good within a City U = cβ (h−H)1−β
Migration is Costly
Simulation
Income Convergence Rate
Mig Rate from South to North(Skilled & Unskilled)
Reg ↑ Begins(Unanticipated)
Reg ↑ Complete
Mig Rate (Skilled)
Mig Rate (Unskilled)
0R
ate
of C
onve
rgen
ce /
Mig
ratio
n
t0 t1 t2Time