The Costs of Remoteness: Evidence from German Divisionand Reunification
Stephen ReddingLondon School of Economics and CEPR
Daniel SturmLondon School of Economics and CEPR
1
The Costs of Remoteness - 2 - Stephen Redding, Daniel Sturm
Motivation
• What determines the spatial distribution of economic activity?
• There are several competing explanations:
– Institutions
– Natural Advantage
– Culture
– Market Access
• Very di±cult to empirically disentangle the eÆects of these factors.
The Costs of Remoteness - 3 - Stephen Redding, Daniel Sturm
This Paper
• We exploit German division and reunification as a naturalexperiment to provide evidence for the importance of marketaccess.
• Key Idea: Division and Reunification exogenously changed therelative market access of West German cities.
The Costs of Remoteness - 4 - Stephen Redding, Daniel Sturm
The Costs of Remoteness - 5 - Stephen Redding, Daniel Sturm
Plan of the Presentation
1. Sketch of the theoretical model
2. Empirical strategy
3. Basic results
4. Further evidence
5. Conclusion
The Costs of Remoteness - 6 - Stephen Redding, Daniel Sturm
Theoretical Model I
• We consider a standard new economic geography model based onHelpman (1998).
• There are N locations (here cities) which are endowed with animmobile resource (housing).
• Consumers:
– Spend a share µ of their income on manufacturing varieties andthe remaining income on the immobile resource.
– Have CES preferences with an elasticity of substitution æ overmanufacturing varieties.
– Inelastically supply one unit of labor.
The Costs of Remoteness - 7 - Stephen Redding, Daniel Sturm
Theoretical Model II
• Manufacturing firms have IRS, use labor as the only input and aremonopolistically competitive.
• Manufacturing varieties are subject to iceberg transport costs T ,which are in turn a function of distance (Tij = dist¡ij).
• In the long-run population is perfectly mobile across locations andmigration equalizes real wages.
The Costs of Remoteness - 8 - Stephen Redding, Daniel Sturm
Calibration
• We use central values from the existing literature for the three keyparameters of the model (æ = 4, µ = 2/3 and ¡ = 1/3)
• We calibrate the stock of the immobile resource in each city sothat the 1939 distribution of population across cities in pre-warGermany is the (unique) equilibrium of the model.
• We simulate the division of Germany and allow the population ofthe West German cities to adjust to this exogenous shock.
The Costs of Remoteness - 9 - Stephen Redding, Daniel Sturm
−20
−15
−10
−50
5M
ean
Sim
ulat
ed C
hang
e (%
)
<25 25−50 50−75 75−100 100−150 150−200 >200
By distance in km from the East−West BorderFigure 1: Simulated Change in West German City Population
The Costs of Remoteness - 10 - Stephen Redding, Daniel Sturm
−15
−10
−50
Mea
n Si
mul
ated
Diff
eren
ce (%
)
Pop < 1919 median Pop >= 1919 median
within and beyond 75km of E−W border for small and large West German citiesFigure 2: Differences in Simulated Population Changes
The Costs of Remoteness - 11 - Stephen Redding, Daniel Sturm
Data
• We focus on a sample of West German cities which had at least20000 inhabitants in 1919.
• We aggregate cities that merge during the sample period.
• Observations:
– Pre-war: 1919, 1925, 1933, 1939
– Division: 1950, 1960, 1970, 1980, 1988
– Reunification: 1992, 2002
The Costs of Remoteness - 12 - Stephen Redding, Daniel Sturm
Basic Empirical Strategy
• DiÆerence-in-DiÆerences Estimation:
– Compare population growth in West German cities close to theEast-West border with other West German cities both beforeand after division.
• Baseline Specification
Popgrowthct = Ø Borderc +∞ (Borderc £Divisiont)+dt +"ct
VOL. 98 NO. 5 1779REDDING AND STURM: THE COSTS OF REMOTENESS
Our key coefficient of interest g on the border 3 division interaction is negative and highly statistically significant, consistent with the predictions of the theoretical model. Division leads to a reduction in the annualized rate of growth of the cities along the East-West German border relative to other West German cities of about 0.75 percentage points. This estimate implies a decline in the population of treatment cities relative to control cities over the 38-year period from 1950 to 1988 of around one-third.
In column 2 we augment our baseline specification and examine heterogeneity over time in the treatment effect of division. Instead of considering a single interaction term between the border dummy and a dummy for the period of division, we introduce separate interaction terms between the border dummy and individual years when Germany was divided. These interaction terms between division years and the border dummy are jointly highly statistically significant
1920 1930 1940 1950 1960 1970 1980 1990 2000
Treatment group
Control group
Inde
x (1
919 5
1)
1.8
1.6
1.4
1.2
1.0
Year
Figure 3. Indices of Treatment and Control City Population
0.0
20.1
20.2
20.3
1920 1930 1940 1950 1960 1970 1980 1990 2000
Tre
atm
ent g
roup
– C
ontr
ol g
roup
Year
Figure 4. Difference in Population Indices, Treatment–Control
The Costs of Remoteness - 14 - Stephen Redding, Daniel Sturm
−0.3
−0.2
−0.1
0.0
1920 1930 1940 1950 1960 1970 1980 1990 2000year
Trea
tmen
t Gro
up −
Con
trol G
roup
Figure 4: Difference in Population Indices, Treatment − Control
Population G
rowth
Population G
rowth
Population G
rowth
Population G
rowth
Population G
rowth
(1)(2)
(3)(4)
(5)
Border × D
ivision -0.746***
-1.097***-0.384
(0.182)(0.260)
(0.252)B
order × Year 1950-60
-1.249***(0.348)
Border × Y
ear 1960-70 -0.699**
(0.283)B
order × Year 1970-80
-0.640*(0.355)
Border × Y
ear 1980-88 -0.397***
(0.147)B
order 0-25km × D
ivision -0.702***
(0.257)B
order 25-50km × D
ivision -0.783***
(0.189)B
order 50-75km × D
ivision -0.620*(0.374)
Border 75-100km
× Division
0.399(0.341)
Border
0.1290.129
0.233-0.009
(0.139)(0.139)
(0.215)(0.148)
Year Effects
Yes
Yes
Yes
Yes
Yes
City Sam
ple A
ll Cities
All C
itiesA
ll Cities
Small C
itiesLarge Cities
Observations
833833
833420
413R
-squared0.21
0.210.21
0.230.30
Notes: Standard errors are heteroscedasticity robust and adjusted for clustering on city. * denotes significance
at the 10% level; ** denotes significance at the 5%
level; *** denotes significance at the 1% level.
Table 2 - B
asic Results on the Im
pact of Division
Yes
The Costs of Remoteness - 16 - Stephen Redding, Daniel Sturm
−4−2
02
4
0 100 200 300Distance to the East−West German border (km)
Estim
ated
Div
isio
n Tr
eatm
ent
Figure 5: Non−parametric Division Treatment Estimates
The Costs of Remoteness - 17 - Stephen Redding, Daniel Sturm
Is it Really Loss of Market Access?
• The decline of the cities along the East-West border is consistentwith our model.
• There is no simple explanation for the decline in terms ofinstitutions, endowments or culture.
The Costs of Remoteness - 17 - Stephen Redding, Daniel Sturm
Is it Really Loss of Market Access?
• The decline of the cities along the East-West border is consistentwith our model.
• There is no simple explanation for the decline in terms ofinstitutions, endowments or culture.
• However, there are other possible explanations for the decline:
– DiÆerences in industrial structure
– DiÆerences in war-related disruption
– Western Economic Integration
– Fear of further armed conflict
The Costs of Remoteness - 18 - Stephen Redding, Daniel Sturm
Quantitative Analysis of the Model
• Can the model not only qualitatively, but also quantitativelyaccount for the decline of the cites along the East-West border?
The Costs of Remoteness - 18 - Stephen Redding, Daniel Sturm
Quantitative Analysis of the Model
• Can the model not only qualitatively, but also quantitativelyaccount for the decline of the cites along the East-West border?
• To compare moments in the simulation and the data, weundertake a grid search over 21 values of each parameter:
– Elasticity of substitution (æ) from 2.5 to 6.5
– Share of tradeables in expenditure (µ) from 0.65 to 0.85
– Distance elasticity of transport costs (¡) from 0.10 to 1.10
The Costs of Remoteness - 19 - Stephen Redding, Daniel Sturm
Identification
• We first show that the relative decline of the East-West bordercities is a well-behaved function of two relationships:
– The strength of agglomeration and dispersion forces: æ(1° µ)
– The coe±cient on distance: (1° æ)¡
The Costs of Remoteness - 19 - Stephen Redding, Daniel Sturm
Identification
• We first show that the relative decline of the East-West bordercities is a well-behaved function of two relationships:
– The strength of agglomeration and dispersion forces: æ(1° µ)
– The coe±cient on distance: (1° æ)¡
• We pin down values for æ(1° µ) and (1° æ)¡ by comparing thepredictions of the model with our two key empirical findings:
– The relative decline of the East-West border cities.
– The more pronounced relative decline of smaller cities.
The Costs of Remoteness - 20 - Stephen Redding, Daniel Sturm
The Costs of Remoteness - 21 - Stephen Redding, Daniel Sturm
The Costs of Remoteness - 22 - Stephen Redding, Daniel Sturm
Small Cities Large Cities
Stre
ngth
of A
gglo
mer
atio
n
Figure 6: Contours of the Simulated Division Treatment
-1.097
-0.3
84
-1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6
1.05
1.1
1.15
1.2
1.25
1.3
Distance Coefficient
The Costs of Remoteness - 23 - Stephen Redding, Daniel Sturm
−4−2
02
4D
ivis
ion
Trea
tmen
t
0 100 200 300Distance to East−West German Border (km)
Simulated Treatment Estimated Treatment
Figure 7: Simulated and Estimated Division Treatments
The Costs of Remoteness - 24 - Stephen Redding, Daniel Sturm
City Structure
• Maybe the cities along the East-West border declined because theywere specialized in industries that declined after the war.
• To control for this possibility we match each treatment city to acontrol city that is as similar as possible in terms of observedcharacteristics.
Population Growth
Population Growth
Population Growth
Population Growth
(1) (2) (3) (4)
Border × Division -0.921*** -1.000*** -0.888*** -0.782***(0.218) (0.253) (0.247) (0.261)
Border 0.309* 0.338** 0.082 0.061(0.153) (0.156) (0.167) (0.194)
Year Effects Yes Yes Yes Yes
Matching on Population Total Employment
Employment in 28 sectors
Employment in 28 sectors and
geography
Observations 280 280 280 280R-squared 0.29 0.26 0.38 0.29
Notes: Standard errors are heteroscedasticity robust and adjusted for clustering on city. * denotes significance at the10% level; ** denotes significance at the 5% level; *** denotes significance at the 1% level.
Table 3 − Matching
The Costs of Remoteness - 26 - Stephen Redding, Daniel Sturm
War-related Disruption
• Could diÆerences in destruction or refugee flows have aÆectedcities post-war growth performance?
• To control for this possibility we include measures of the degree ofwar-related disruption in the regression and allow their eÆect tovary over time.
Population G
rowth
Population G
rowth
Population G
rowth
(1)(2)
(3)
Border × D
ivision -0.737***
-0.656*** -0.678***
(0.182)(0.191)
(0.211)B
order0.136
0.1290.029
(0.139)(0.146)
(0.167)W
ar Disruption × Y
ear 1919-25-0.014
-0.004 0.004
(0.011)(0.006)
(0.020)W
ar Disruption × Y
ear 1925-330.019
0.006 -0.018
(0.017)(0.007)
(0.019)W
ar Disruption × Y
ear 1933-39-0.001
0.004 0.064**
(0.023)(0.009)
(0.028)W
ar Disruption × Y
ear 1950-60 0.073***
0.033*** -0.056**
(0.015)(0.008)
(0.026)W
ar Disruption × Y
ear 1960-700.012
0.009 -0.006
(0.017)(0.007)
(0.026)W
ar Disruption × Y
ear 1970-80-0.014
0.004 0.062*
(0.025)(0.012)
(0.034)W
ar Disruption × Y
ear 1980-880.007
0.0020.009
(0.013)(0.006)
(0.020)Y
ear EffectsY
esY
esY
esW
ar Disruption M
easureR
ubbleD
wellings
Refugees
Observations
777756
833R
-squared0.24
0.240.24
Notes: Standard errors are heteroscedasticity robust and adjusted for clustering on city. *
denotes significance at the 10% level; ** denotes significance at the 5%
level; *** denotes significance at the 1%
level.
Table 4 - C
ontrolling for War D
evastation
The Costs of Remoteness - 28 - Stephen Redding, Daniel Sturm
Western Integration
• West Germany experienced considerable economic integration withWestern Europe in the post-war period.
• Can Western integration (at least partly) explain the relativedecline of the cities along the East-West border?
Population Grow
thPopulation G
rowth
(1)(2)
Border × D
ivision -0.730***
(0.204)B
order0.045
(0.151)W
estern Border × D
ivision0.032
(0.226)W
estern Border
-0.162 (0.152)
Border 0-25km
× Division
-0.675**(0.297)
Border 25-50km
× Division
-0.756***(0.240)
Border 50-75km
× Division
-0.593 (0.403)
Border 75-100km
× Division
0.426(0.372)
Western B
order 0-25km × D
ivision0.421
(0.383)W
estern Border 25-50km
× Division
0.488*(0.289)
Western B
order 50-75km × D
ivision-0.375 (0.338)
Western B
order 75-100km × D
ivision-0.140 (0.351)
Border D
istance Grid C
ellsY
esW
estern Border D
istance Grid C
ellsY
esY
ear EffectsY
esY
esO
bservations833
833R
-squared0.21
0.23
Notes: Standard errors are heteroscedasticity robust and adjusted for clustering on city. * denotes
significance at the 10% level; ** denotes significance at the 5%
level; *** denotes significance at the 1%
level.
Table 5 - C
ontrolling for Western E
conomic Integration
The Costs of Remoteness - 30 - Stephen Redding, Daniel Sturm
Fear of Further Armed Conflict
• Several pieces of evidence suggest that fear of a further armedconflict cannot explain the decline of the East-West border cities:
– Di±cult to square with the larger decline of small cities and ourquantitative analysis.
– There is no evidence of a negative eÆect of proximity to theEast-West border in centrally planned East Germany.
– There is no evidence of stronger treatment eÆects close tostrategic points along the border (“Fulda Gap”).
– Nuclear deterrence made a small scale war very unlikely.
– No evidence that another war was an everyday concern.
The Costs of Remoteness - 31 - Stephen Redding, Daniel Sturm
Reunification
• Do we observe a reversal of fortune in the cities along theEast-West border after reunification?
• There are good reasons to be sceptical:
– The size and income of the area added is much smallercompared to division.
– Heavy subsidies for the border cities are rapidly discontinued.
– While division abruptly severed all links between East and WestGermany, the re-creation of such links after reunification is likelyto take time.
Population Growth
Population Growth
Population Growth
Population Growth
(1) (2) (3) (4)
Border × Division -0.477*** -0.127 -0.223 -0.007 (0.156) (0.128) (0.202) (0.136)
Border -0.141 -0.141 -0.236 -0.064 (0.106) (0.106) (0.168) (0.108)
Year Effects Yes Yes Yes YesCity Sample All All Small Cities Large Cities
Year Sample 1950-1988 & 1992-2002
1980-1988 & 1992-2002
1980-1988 & 1992-2002
1980-1988 & 1992-2002
Observations 595 238 120 118R-squared 0.30 0.15 0.21 0.14
Notes: Standard errors are heteroscedasticity robust and adjusted for clustering on city. * denotes significance at the 10% level; ** denotes significance at the 5% level; *** denotes significance at the 1% level.
Table 6 - The Impact of Reunification
The Costs of Remoteness - 33 - Stephen Redding, Daniel Sturm
Summary
• West German cities close to the East-West border substantiallydecline after division relative to other West German cities.
• The evidence suggests that this decline can be largely explained bythe change in market access of these cities.
• While institutions and natural advantage are certainly alsoimportant, market access plays a substantial role in determiningeconomic prosperity.