The Costs of Remoteness: Evidence from German Division and Reunification Stephen Redding London School of Economics Daniel Sturm University of Munich
Jan 13, 2016
The Costs of Remoteness: Evidence from German Division and
Reunification
Stephen ReddingLondon School of Economics
Daniel SturmUniversity of Munich
2
Motivation
• What determines the spatial distribution of economic activity?
• Main competing explanations:– Institutions– Natural Advantage– Market Access
• Very difficult to empirically disentangle the effects of these factors
3
This Paper
• Exploit German division and reunification as a natural experiment
• Division of Germany exogenously reduced the market access of West German cities close to the new East-West border relative to other West German cities
• Reunification reverses this loss to some extent
• Basic empirical strategy is diff-in-diffs. Compare:
a) Population growth in W German cities close to the new E-W German border to other W German cities
b) Both before and after division.
4
Map 1 – Germany in its pre - World War II borders
5
An Economic Geography Model
• There are N locations which are endowed with an exogenous stock of housing (immobile resource in fixed supply)
• Firms produce Dixit-Stiglitz varieties of a manufacturing good with IRS and use labor as the only input
• Manufacturing varieties are subject to (iceberg) transport costs between locations
• Consumers demand both housing and the manufacturing varieties
• Population is perfectly mobile across locations and migration equalizes real wages
6
Empirical Implication
• Suppose the N locations are divided by a closed border
• Cities near to the new border experience a reduction in market and supplier access relative to other cities
• This reduction in market and supplier access will reduce their equilibrium size relative to cities further from the border
7
-10,
000
-5,0
000
5,0
00M
ean
Sim
ula
ted
- A
ctua
l 193
9 P
opul
atio
n
<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
8
-8,0
00-6
,000
-4,0
00-2
,000
0M
ean
Sim
ula
ted
- A
ctua
l 193
9 P
opul
atio
n
Pop < 1939 median Pop >= 1939 median
within 75km of E-W border, city population above and below 1939 medianFigure 2: Simulated Change in West German City Population
9
Empirical Strategy
• Difference in Differences Estimation:– Compare population growth in West German cities
close to the E-W border with West German cities distant from the E-W border both before and after division
• Examine the treatment effect of re-unification separately
• Baseline Specification:
• Basic specification considers a 75km border zone
Popgrowct Borderc Borderc Division t d t ct
10
• We focus on a sample of West German Cities– We consider all West German cities which had a
population larger than 20,000 in 1919– We aggregate cities that merge during the sample
period
• This results in a sample of 119 West German cities of which 20 are within 75 km of the East-West German border.
• Time Periods– Pre – WW II Germany : 1919, 1925, 1933, 1939– Cold-War : 1950, 1960, 1970, 1980, 1988– Reunification : 1992, 2002
Description of the Data
11
11
.21
.41
.61
.8
1920 1930 1940 1950 1960 1970 1980 1990 2000year
Non-border index Border Index
Inde
x (1
919
=1)
Figure 3: Indices of Border & Non-Border City Population
12
-0.3
-0.2
-0.1
0.0
1920 1930 1940 1950 1960 1970 1980 1990 2000year
Bor
der
Inde
x -
Non
-bor
der
Ind
exFigure 4: Difference in Population Indices, Border - Non-border
13
(1) (2) (3) (4) (5)
Population Growth
Population Growth
Population Growth
Population Growth
Population Growth
Border × Division -0.746*** -0.746*** -0.746***(0.182) (0.183) (0.196)
Border × Year 1950-60 -1.249***(0.348)
Border × Year 1960-70 -0.699**(0.283)
Border × Year 1970-80 -0.640*(0.355)
Border × Year 1980-88 -0.397***(0.147)
Border 0-25km × Division -0.702***(0.257)
Border 25-50km × Division -0.783***(0.189)
Border 50-75km × Division -0.620*(0.374)
Border 75-100km × Division 0.399(0.341)
Border 0.129 0.129 0.325*(0.139) (0.139) (0.187)
Year Effects Yes Yes Yes Yes YesState ("Länder") Effects YesCity Effects Yes
Observations 833 833 833 833 833R-squared 0.21 0.21 0.21 0.22 0.32
Table 2 - Baseline Empirical Results
14
The Role of Market Access
• The decline of the border cities is consistent with our model
• Institutions or natural endowments can not explain the decline
• However, there are at least three other possible explanations for the decline:– Fear of further armed conflict– Differences in industrial structure– Differences in war related destruction
• There are several pieces of evidence that the decline is driven by a reduction in market access
15
(1) (2) (3) (4) (5) (6) (7)Population
GrowthPopulation
GrowthPopulation
GrowthPopulation
GrowthPopulation
GrowthPopulation
GrowthPopulation
Growth
Border × Division -0.746*** -0.052 -0.434 0.095 0.079 -1.097*** -0.384(0.182) (0.208) (0.307) (0.366) (0.399) (0.260) (0.252)
Market Potential Index 4.790***(0.490)
Eastern Market Potential Loss -0.344 -1.176*** -1.205***(0.227) (0.277) (0.367)
Border 0.129 0.230* 0.129 0.129 0.043 0.233 -0.009(0.139) (0.121) (0.139) (0.139) (0.184) (0.215) (0.148)
Year Effects Yes Yes Yes Yes Yes Yes Yes
Year Sample
1919-1939 & 1950-1988
1919-1939 & 1950-1988
1919-1939 & 1950-1988
1919-1939 & 1950-1970
1919-1939 & 1950-1970
1919-1939 & 1950-1988
1919-1939 & 1950-1988
City Sample All Cities All Cities All Cities All CitiesCities within 150km of the E-W border
Small Cities Large Cities
Observations 833 833 833 595 230 420 413R-squared 0.21 0.30 0.21 0.11 0.24 0.23 0.30
Table 4 - Market Potential and Size Effects
16
Conclusions
• Negative treatment effect of proximity to the East – West border on city development in market-based West Germany
• The evidence suggests that this treatment effect can be largely explained by the change in market access
• While institutions and natural advantage are certainly also important, market access plays a substantial role in determining economic prosperity
17
Thank You!
18
Non-parametric estimates-1
01
2
0 100 200 300Distance to the East-West German border (km)
Est
ima
ted
Div
isio
n T
rea
tmen
tFigure 5: Non-parametric Division Treatment Estimates
19
American and Russian Advance
20
Pre-WWI German and post-WWI Border Changes
21
Theoretical Model(Henderson 1974, Helpman 1998)
ccc axFl
Producer Behaviour
ccc wap
1 )1( Fx
1Hc
Mcc CCU
Consumer Behaviour
Trade costs
)1/(1)1/(11)( cicii iMc SATpnP PcH 1 Ec
Hc
22
Equilibrium Nominal and Real Wages
1 1/a cMAc
SAc /1 1 / LcHc
1 1/a i MA i
SA i /1 1 / L iHi
1
Factor Mobility
11
1Hi
Mi
i
Hc
Mc
c
PP
w
PP
w
Equilibrium Wages
/11c
cc MA
aw
1
acwc 1
x i
w iLi P iM 1Tci 1 1
x MAc