Bye Bye, G.I. - The Impact of the U.S. Military Drawdown on Local German Labor Markets Jan Peter aus dem Moore1
Alexandra Spitz-Oener2
1 Humboldt Universität Berlin 2 Humboldt Universität Berlin, IAB, IZA, CASE
Tagung des SFB 649 "Ökonomisches Risiko" Motzen, 1 July 2011
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
1 / 24
Motivation & Research question
Bases of U.S. Forces in Germany
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
2 / 24
Motivation & Research question
Bases of U.S. Forces in Germany
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
2 / 24
Motivation & Research question
U.S. Forces in Germany - Historical Evolution
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
3 / 24
Motivation & Research question
Motivation - Economic impact of military base closures Base closures constitute a large, exogeneous shock at local level
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
4 / 24
Motivation & Research question
Motivation - Economic impact of military base closures Base closures constitute a large, exogeneous shock at local level Strong public interest (e.g. subsidies for regional development)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
4 / 24
Motivation & Research question
Motivation - Economic impact of military base closures Base closures constitute a large, exogeneous shock at local level Strong public interest (e.g. subsidies for regional development) But: challenges for empirical research ▸ Identication of causality (endogeneity bias) ▸ Data availability
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
4 / 24
Motivation & Research question
Motivation - Economic impact of military base closures Base closures constitute a large, exogeneous shock at local level Strong public interest (e.g. subsidies for regional development) But: challenges for empirical research ▸ Identication of causality (endogeneity bias) ▸ Data availability The drawdown of U.S. Military Forces in Germany provides a unique "natural experiment"-type setting: ▸ Exogeneous as drawdown decisions follow U.S. military rationale ▸ Strong regional variation in withdrawal 'treatment' intensity and timing ▸ Separation of direct vs. indirect (spill-over) eects
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
4 / 24
Motivation & Research question
Motivation - Economic impact of military base closures Base closures constitute a large, exogeneous shock at local level Strong public interest (e.g. subsidies for regional development) But: challenges for empirical research ▸ Identication of causality (endogeneity bias) ▸ Data availability The drawdown of U.S. Military Forces in Germany provides a unique "natural experiment"-type setting: ▸ Exogeneous as drawdown decisions follow U.S. military rationale ▸ Strong regional variation in withdrawal 'treatment' intensity and timing ▸ Separation of direct vs. indirect (spill-over) eects
Research Question
What is the impact of the U.S. drawdown on the regional labor market? ☀ Spill-over eects into private sector employment ☀ Spill-over eects into private sector wages aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
4 / 24
Outline
Outline Motivation & Research Question Related Literature Historical Background - U.S. Forces Stationing and Withdrawal in Germany Empirical strategy Data sources Results Robustness Checks Conclusion & Future Work
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
5 / 24
Related Literature
Related Literature 1
Adjustment to regional shocks
Blanchard/Katz (1992), Decressin/Fatas (1995), Moretti (2010, 2011)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
6 / 24
Related Literature
Related Literature 1
Adjustment to regional shocks
2
Impact of Military Base Realignment and Closures (BRAC)
Blanchard/Katz (1992), Decressin/Fatas (1995), Moretti (2010, 2011)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
6 / 24
Related Literature
Related Literature 1
Adjustment to regional shocks
2
Impact of Military Base Realignment and Closures (BRAC)
Blanchard/Katz (1992), Decressin/Fatas (1995), Moretti (2010, 2011) ▸
Germany ☀ Some early descriptive/case studies for U.S. withdrawal in Germany: Bebermeyer/Thimann (1990), Eisbach et al. (1991), Blien et al. (1992, 1993), Gettmann (1993), BICC (1995) ☀ Eects of Bundeswehr base realignments on local economy, crime: Paloyo et al. 2010a, 2010b
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
6 / 24
Related Literature
Related Literature 1
Adjustment to regional shocks
2
Impact of Military Base Realignment and Closures (BRAC)
Blanchard/Katz (1992), Decressin/Fatas (1995), Moretti (2010, 2011) ▸
▸ ▸
Germany ☀ Some early descriptive/case studies for U.S. withdrawal in Germany: Bebermeyer/Thimann (1990), Eisbach et al. (1991), Blien et al. (1992, 1993), Gettmann (1993), BICC (1995) ☀ Eects of Bundeswehr base realignments on local economy, crime: Paloyo et al. 2010a, 2010b US - Krizan (1998), Hookers/Knetters (2001), Poppert/Herzog (2003) Sweden - Andersson et al. (2007)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
6 / 24
Related Literature
Related Literature 1
Adjustment to regional shocks
2
Impact of Military Base Realignment and Closures (BRAC)
Blanchard/Katz (1992), Decressin/Fatas (1995), Moretti (2010, 2011) ▸
▸ ▸
3
Germany ☀ Some early descriptive/case studies for U.S. withdrawal in Germany: Bebermeyer/Thimann (1990), Eisbach et al. (1991), Blien et al. (1992, 1993), Gettmann (1993), BICC (1995) ☀ Eects of Bundeswehr base realignments on local economy, crime: Paloyo et al. 2010a, 2010b US - Krizan (1998), Hookers/Knetters (2001), Poppert/Herzog (2003) Sweden - Andersson et al. (2007)
Other specic/exogeneous regional economic shocks ▸ ▸
Regional impact of coal boom and bust: Black et al. (2002, 2003, 2005a, 2005b) Spill-over eects from million dollar plants: Greenstone/Moretti (2003), Greenstone/Hornbeck/Moretti(2010)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
6 / 24
Historical Background
U.S. Forces in Germany - Stationing and Drawdown Process Initial stationing locations after World War II
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
7 / 24
Historical Background
U.S. Forces in Germany - Stationing and Drawdown Process Initial stationing locations after World War II ▸ American occupation zone ▸ Availability of installations/barracks ▸ NATO, "Forward strategy": two major lines of defense against potential Soviet invasion
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
7 / 24
Historical Background
U.S. Forces in Germany - Stationing and Drawdown Process Initial stationing locations after World War II ▸ American occupation zone ▸ Availability of installations/barracks ▸ NATO, "Forward strategy": two major lines of defense against potential Soviet invasion CFE treaty process with USSR: negotiations on limits to total troop levels in Western Europe
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
7 / 24
Historical Background
U.S. Forces in Germany - Stationing and Drawdown Process Initial stationing locations after World War II ▸ American occupation zone ▸ Availability of installations/barracks ▸ NATO, "Forward strategy": two major lines of defense against potential Soviet invasion CFE treaty process with USSR: negotiations on limits to total troop levels in Western Europe Fall of the Berlin wall, German reunication, CFE treaty
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
7 / 24
Historical Background
U.S. Forces in Germany - Stationing and Drawdown Process Initial stationing locations after World War II ▸ American occupation zone ▸ Availability of installations/barracks ▸ NATO, "Forward strategy": two major lines of defense against potential Soviet invasion CFE treaty process with USSR: negotiations on limits to total troop levels in Western Europe Fall of the Berlin wall, German reunication, CFE treaty Selection criteria for U.S. base closures (i) Ensuring that the forces would meet military and operational requirements (ii) Decreasing support costs and increasing eciency of base operations (iii) Minimizing personnel moves (iv) Reducing environmental impact (v) Considering the proximity of training areas, the quality of housing and facilities, the local political and military environment, the concerns of host nations, and the base's proximity to road and rail networks.
(Source: U.S. General Accounting Oce 1991, 1994) aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
7 / 24
Historical Background
U.S. Forces in Germany - Structure of demand in the German economy 1
Bases as direct employers of German civilian workers (≈71,000 in 1990)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
8 / 24
Historical Background
U.S. Forces in Germany - Structure of demand in the German economy 1 2
Bases as direct employers of German civilian workers (≈71,000 in 1990)
Demand by U.S. bases for goods and services from German companies (≈4.2 bn DM in 1986) ▸ Construction, repair, maintenance ▸ Real estate (rented homes & apartments) ▸ Food, gasoline, mail and railway, services etc.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
8 / 24
Historical Background
U.S. Forces in Germany - Structure of demand in the German economy 1 2
3
Bases as direct employers of German civilian workers (≈71,000 in 1990)
Demand by U.S. bases for goods and services from German companies (≈4.2 bn DM in 1986) ▸ Construction, repair, maintenance ▸ Real estate (rented homes & apartments) ▸ Food, gasoline, mail and railway, services etc. U.S. Forces as private consumers in the local economy (≈2.1 bn DM in 1989) ▸ Hotels, restaurants, bars, entertainment ▸ Rental cars, car dealerships, gas stations ▸ Groceries, daily necessities
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
8 / 24
Data
Data sources 1
Data on U.S. Army Stationing and Withdrawal in Germany ▸ ▸ ▸ ▸ ▸ ▸
Number of U.S. personnel (Soldiers, U.S. Civilian, German employees) All US Army and Airforce Bases in West-Germany Reporting dates: 1988, 1990, 1992, 1995, 1999, 2001, 2002-2009 Exact geographic base location (geocoded) Information on date(s) of announcement of base reduction/closure, actual nal closure (exact to the day) Some add. information on type of base, area size, estd. replacement value etc.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
9 / 24
Data
Data sources 1
Data on U.S. Army Stationing and Withdrawal in Germany ▸ ▸ ▸ ▸ ▸
2
▸
Number of U.S. personnel (Soldiers, U.S. Civilian, German employees) All US Army and Airforce Bases in West-Germany Reporting dates: 1988, 1990, 1992, 1995, 1999, 2001, 2002-2009 Exact geographic base location (geocoded) Information on date(s) of announcement of base reduction/closure, actual nal closure (exact to the day) Some add. information on type of base, area size, estd. replacement value etc.
IAB employment and wage data from IAB Beschäftigungshistorie and IAB Betriebshistorikpanel (BHP)
(BeH) ▸ ▸ ▸
100% sample of individual employment spells for 4 Bundesländer (Hessen, Rheinland-Pfalz, Bayern, Baden-Württemberg, Bayern) BeH individual variables: info on start/end of employment spells, age, nationality, education, full-time/part-time, occupation, gross daily wage BHP employer variables: location (district), industry, rm size
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
9 / 24
Data
Data sources 1
Data on U.S. Army Stationing and Withdrawal in Germany ▸ ▸ ▸ ▸ ▸
2
▸
Number of U.S. personnel (Soldiers, U.S. Civilian, German employees) All US Army and Airforce Bases in West-Germany Reporting dates: 1988, 1990, 1992, 1995, 1999, 2001, 2002-2009 Exact geographic base location (geocoded) Information on date(s) of announcement of base reduction/closure, actual nal closure (exact to the day) Some add. information on type of base, area size, estd. replacement value etc.
IAB employment and wage data from IAB Beschäftigungshistorie and IAB Betriebshistorikpanel (BHP)
(BeH) ▸ ▸
3
▸
100% sample of individual employment spells for 4 Bundesländer (Hessen, Rheinland-Pfalz, Bayern, Baden-Württemberg, Bayern) BeH individual variables: info on start/end of employment spells, age, nationality, education, full-time/part-time, occupation, gross daily wage BHP employer variables: location (district), industry, rm size
Regional statistical data from Federal and Regional Statistical Oces ▸
Area, Total population/population density, employment, unemployment rate, net migration, classication of area types (BBR)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
9 / 24
Empirical strategy
Empirical approach (1) - Employment 1
Panel Di-in-Di logYkt αk δt TIk
=
0
Year k
U .S .Mil
1990−
U .S .Mil
population
aus dem Moore (HU Berlin)
= αk + δt + β × TIk × 1 [t > Year0k ]) + kt
1990
2002
-
(1)
District xed eects Year xed eects Measure of treatment intensity Year of 1st announcement of withdrawal in district
U.S. Drawdown & Local Labor Markets
01-07-2011
10 / 24
Empirical strategy
Empirical approach (1) - Employment 1
Panel Di-in-Di logYkt αk δt TIk
=
U .S .Mil
1990−
= αk + δt + β × TIk × 1 [t > Year0k ]) + kt
U .S .Mil
population
1990
2002
-
(1)
District xed eects Year xed eects Measure of treatment intensity Year of 1st announcement of withdrawal in district
0 ⇒ Inclusion of State-by-Year-Eects Year k
logYkt = αk + δt +ηst +β × TIk × 1 [t > Year0k ]) + kt
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
(2)
10 / 24
Empirical strategy
Empirical approach (1) - Employment 1
Panel Di-in-Di logYkt αk δt TIk
=
U .S .Mil
1990−
= αk + δt + β × TIk × 1 [t > Year0k ]) + kt
U .S .Mil
population
1990
2002
-
(1)
District xed eects Year xed eects Measure of treatment intensity Year of 1st announcement of withdrawal in district
0 ⇒ Inclusion of State-by-Year-Eects Year k
logYkt = αk + δt +ηst +β × TIk × 1 [t > Year0k ]) + kt
(2)
⇒ Inclusion of linear/quadratic district specic time trends
logYkt = α0k + α1kt + α2kt + δt + ηst + β × TIk × 1 [t > Year0k ]) + kt α0k α1k α2kt
aus dem Moore (HU Berlin)
-
(3)
District specic intercept District specic coecient on linear time trend District specic coecient on quadratic time trend U.S. Drawdown & Local Labor Markets
01-07-2011
10 / 24
Empirical strategy
Empirical approach (2) - Employment 2
Dynamic specication ▸
Exploiting regional variation in timing of 1st announcement date to explore pattern of lead/anticipatory and lagged eects: +6
logYkt = α0k +α1kt +α2kt +δt +ηst + ∑ (τs × TIk × 1 [t = Year0k + s ])+kt s =−5
τs
aus dem Moore (HU Berlin)
-
(4)
Period treatment eects
U.S. Drawdown & Local Labor Markets
01-07-2011
11 / 24
Empirical strategy
Empirical approach (3) - Wages 1
Micro estimation at individual level logWikt
= α0k +α1kt +α2kt +δt +ηst +β × TIk × 1 [t > Year0k ])+ Xikt γ +ikt Xikt
-
aus dem Moore (HU Berlin)
Individual level covariates (age, age2 , nationality, education, rm size, occupation)
U.S. Drawdown & Local Labor Markets
01-07-2011
(5)
12 / 24
Empirical strategy
Empirical approach (3) - Wages 1
Micro estimation at individual level logWikt
= α0k +α1kt +α2kt +δt +ηst +β × TIk × 1 [t > Year0k ])+ Xikt γ +ikt Xikt
2
-
Individual level covariates (age, age2 , nationality, education, rm size, occupation)
2-step estimation at District x Industry level Step 1. logWijkt j
ηjkt
-
(5)
(6)
= ηjkt + Xijkt γ + µijkt
Industry index (i=1,⋯, 8) District X Industry eects (conditional on individual level covariates)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
12 / 24
Empirical strategy
Empirical approach (3) - Wages 1
Micro estimation at individual level logWikt
= α0k +α1kt +α2kt +δt +ηst +β × TIk × 1 [t > Year0k ])+ Xikt γ +ikt Xikt
2
-
Individual level covariates (age, age2 , nationality, education, rm size, occupation)
2-step estimation at District x Industry level Step 1. logWijkt j
ηjkt
-
(5)
(6)
= ηjkt + Xijkt γ + µijkt
Industry index (i=1,⋯, 8) District X Industry eects (conditional on individual level covariates)
Step 2. ηˆjkt = α0k + α1kt + α2kt + δt + ηst + βj × TIk × 1 [t > Year0k ]) + ξjkt (7)
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
12 / 24
Empirical strategy
Treatment vs. control districts & treatment intensity
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
13 / 24
Empirical strategy
Sample specication Employment ▸
▸ ▸ ▸ ▸
Full-time, private sector employment Age 25-55 Education variable improved via imputation procedures proposed by Fitzenberger et al. (1999, 2006) Crosssamples for reporting date 30/06 in each year, collapsed on district level Employment spell is recorded at employer's location
Wages ▸
▸ ▸
10 percent subsample for males age 25-55 Gross real daily wages, deated by common price index for West Germany Imputation of right-censored wages separately by education groups (Gartner, 2005) with full set of available covariates
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
14 / 24
Empirical strategy A
Table Appendix
Descriptive Statistics
Table 1: Selected district characteristics by treatment status 1990
2002
Treatm. (1)
Contr. (2)
Diff. (3)
Treatm. (4)
Contr. (5)
Diff. (6)
195,130 (21,082)
125,259 (7,377)
69,871*** (22,336)
207,210 (21,373)
138,093 (8,061)
69,118*** (22,843)
703 (97)
347 (54)
356*** (111)
729 (98)
370 (54)
359*** (112)
24,451 (1,165)
22,002 (1,057)
2,449 (1,573)
30,079 (1,492)
27,206 (1,451)
2,874 (2,081)
5.3 (0.2)
4.8 (0.2)
0.5 (0.3)
7.6 (0.2)
6.6 (0.3)
1.0*** (0.4)
Urban
.429 (.059)
.250 (.056)
.179** (.082)
.429 (.059)
.250 (.056)
.179** (.082)
Conurban
.443 (.060)
.583 (.064)
-.140 (.087)
.443 (.060)
.583 (.064)
-.140 (.087)
Rural
.129 (.040)
.167 (.048)
-.038 (.063)
.129 (.040)
.167 (.048)
-.038 (.063)
Hesse
.229 (.050)
.067 (.032)
.162*** (.060)
.229 (.050)
.067 (.032)
.162*** (.060)
Rhineland-Palatinate
.171 (.045)
.183 (.050)
-.012 (.068)
.171 (.045)
.183 (.050)
-.012 (.068)
Baden-Wuerttemberg
.214 (.049)
.233 (.055)
-.019 (.074)
.214 (.049)
.233 (.055)
-.019 (.074)
Bavaria
.386 (.058)
.517 (.065)
-.131 (.087)
.386 (.058)
.517 (.065)
-.131 (.087)
70
60
70
60
Demographics Population Population density (inhabitants per sqkm) Socio-economic outcomes GDP per capita† (EUR) Unemployment Rate Area type
Geographic distribution
N
Notes: †Due to data limitations, GDP per capita reported in 1990 column are 1992 values. * Significant at 10%, ** at 5%, *** at 1%.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
trend
01-07-2011
15 / 24
Results
Employment
Impact of withdrawal on employment (baseline) Table 4: Estimated impact of U.S. military withdrawal on total district employment, 1975-2002 Dep. variable: Total employment (log)
(1)
(2)
(3)
(4)
U.S. WD treatment (%)
-.015** (.006) .987
-.018*** (.006) .989
-.015** (.006) .987
-.017*** (.006) .989
-.016** (.008) .985
-.021*** (.008) .986
-.011*** (.003) .997
-.010*** (.002) .998
No No No
Yes No No
Yes Yes No
Yes Yes Yes
3,640
3,640
3,640
3,640
A. - All
2
R
-.009*** (.003) .998
-.007*** (.002) .998
B. - Male U.S. WD treatment (%) R2
-.008*** (.003) .997
-.005** (.002) .998
C. - Female U.S. WD treatment (%) 2
R Other covariates: State by year dummies District x time trends District x time2 trends N
Notes: Each cell reports the coefficient on the treatment variable for one regression. All regressions include district and year fixed effects. Robust std. errors clustered at district level in parentheses. * Significant at 10%, ** at 5%, *** at 1%.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
16 / 24
Results
Employment
Dynamic pattern of withdrawal eect on employment Table 8: Dynamic pattern of impact of U.S. military withdrawal on total employment at district level Dep. variable: Total employment (log) WD announcement𝑡−5 WD announcement𝑡−4 WD announcement𝑡−3 WD announcement𝑡−2 WD announcement𝑡−1 WD announcement𝑡0 WD announcement𝑡+1 WD announcement𝑡+2 WD announcement𝑡+3 WD announcement𝑡+4 WD announcement𝑡+5 WD announcement𝑡+6𝑓 𝑜𝑟𝑤𝑎𝑟𝑑 Other covariates: State by year dummies District x time trends District x time2 trends R2 N
aus dem Moore (HU Berlin) Notes:
(1)
(2)
(3)
(4)
-.002 (.002) -.003 (.003) -.005 (.003) -.005 (.003) -.006 (.004) -.009* (.005) -.011** (.005) -.012** (.006) -.015** (.006) -.017** (.007) -.019** (.008) -.019** (.008)
-.002 (.002) -.004 (.003) -.005* (.003) -.006 (.004) -.007* (.004) -.010** (.005) -.012** (.005) -.015*** (.006) -.018*** (.006) -.020*** (.007) -.022*** (.008) -.023*** (.009)
-.002 (.001) -.003** (.002) -.005** (.002) -.006** (.003) -.006* (.003) -.009** (.004) -.012*** (.004) -.014*** (.005) -.017*** (.005) -.019*** (.006) -.022*** (.006) -.022*** (.007)
-.001 (.001) -.003* (.001) -.004** (.002) -.005* (.002) -.005* (.003) -.008** (.004) -.010** (.004) -.012*** (.005) -.015*** (.005) -.017*** (.006) -.019*** (.006) -.018*** (.007)
No No No
Yes No No
Yes Yes No
Yes Yes Yes
.987 3,640
.989 3,640
.998 3,640
.998 3,640
U.S. Drawdown & Local Labor Markets All regressions include district and year fixed effects. Robust std.
01-07-2011
17 / 24
Results
Wages
Eect on industry wages
Table 1: Impact of U.S. military withdrawal on wages by industry Dep. Variable: Real wages (log) by industry
(1)
(2)
(3)
(4)
1 Basic materials U.S. WD treatment (%)
.006 (.005)
.004 (.004)
U.S. WD treatment (%)
.002 (.003)
U.S. WD treatment (%)
-.000 (.002)
.001 (.001)
U.S. WD treatment (%)
.001 (.002)
.002* (.001)
-.001 (.002)
-.001 (.002)
2 Investment goods .003 (.003)
.001 (.001)
.001 (.001)
3 Food and consumption goods -.001 (.001)
-.001 (.001)
4 Construction -.000 (.001)
-.001 (.001)
5 Retail/Repair U.S. WD treatment (%)
.000 (.002)
U.S. WD treatment (%)
.013* (.007)
U.S. WD treatment (%)
.006 (.009)
U.S. WD treatment (%)
.006 (.005) Yes
.000 (.001)
.001 (.002)
.002 (.002)
6 Transport/Information .010*** (.003)
-.001 (.002)
-.005*** (.002)
7 Corporate services .003 (.005)
-.004** (.002)
-.005** (.002)
8 Private household services
Individual level covariates
.004 (.004) Yes
.002 (.003) Yes
.001 (.003) Yes
Notes: Each cell reports the coefficient the treatment for one regression. aus dem Moore (HU Berlin) U.S. Drawdown &onLocal Laborvariable Markets
01-07-2011
18 / 24
Results
Wages
Eect on industry wages by rm size (1)
Table 1: Impact of U.S. military withdrawal on industry wages by firm size category Dep. Variable: Real wages (log) by industry
(1)
(2)
(3)
(4)
1 Basic materials U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
.012*** (.005) -.026*** (.006) -.019*** (.004)
.011*** (.004) -.026*** (.005) -.018*** (.004)
.004* (.003) -.020*** (.005) -.013*** (.004)
.004* (.002) -.020*** (.005) -.013*** (.004)
2 Investment goods U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
.006* (.003) -.032*** (.005) -.019*** (.003)
.007** (.003) -.031*** (.005) -.018*** (.003)
.004*** (.001) -.029*** (.004) -.016*** (.003)
.004*** (.001) -.029*** (.004) -.016*** (.003)
3 Food and consumption goods U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
.005** (.002) -.014*** (.005) -0.006 (.003)
.006** (.003) -.014*** (.005) -.006* (.004)
.019*** (.004) -.029*** (.006) -.014*** (.005)
.020*** (.005) -.029*** (.007) -.014*** (.005)
.004** (.002) -.013*** (.005) -.005 (.003)
.003* (.002) -.013*** (.005) -.005 (.004)
4 Construction U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
aus dem Moore (HU Berlin)
.017*** (.004) -.030*** (.007) -.015*** (.005)
.016*** (.004) -.031*** (.007) -.015*** (.005)
Continued on next page
U.S. Drawdown & Local Labor Markets
01-07-2011
19 / 24
Results
Wages
Eect on industry wages by rm size (2) 5 Retail/Repair U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
.011*** (.004) -.019*** (.005) -.009*** (.003)
.011*** (.004) -.019*** (.005) -.009*** (.003)
.012*** (.003) -.019*** (.005) -.009*** (.003)
.013*** (.003) -.019*** (.005) -.010*** (.003)
6 Transport/Information U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
.029*** (.008) -.052*** (.015) -.019** (.008)
.026*** (.007) -.050*** (.014) -.016** (.007)
.012*** (.004) -.047*** (.014) -.014* (.007)
.009*** (.004) -.047*** (.015) -.014* (.008)
7 Corporate services U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
.020** (.008) -.032*** (.005) -.025*** (.004)
.017*** (.006) -.030*** (.005) -.024*** (.004)
.008*** (.003) -.023*** (.005) -.019*** (.004)
.007** (.003) -.023*** (.005) -.019*** (.004)
8 Private household services U.S. WD treatment (%) U.S. WD treatment (%) X small U.S. WD treatment (%) X medium
aus dem Moore (HU Berlin)
.037*** (.005) -.057*** (.007) -.032*** (.007)
.035*** .028*** .026*** (.006) (.006) (.007) -.057*** -.050*** -.049*** (.008) (.008) (.008) -.032*** -.028*** -.028*** (.007) (.007) (.007) Continued on next page
U.S. Drawdown & Local Labor Markets
01-07-2011
20 / 24
Robustness Checks
Robustness checks Selection of treatment and control districts Local shocks coinciding with U.S. drawdown process ▸
▸ ▸
graph In-migration from Eastern Germany after the fall of the Berlin wall Regional subsidy programms
Bundeswehr reductions
Serial correlation - Alternative Std. Errors Heteroscedasticity - Weighting by regional Kreis size Level of aggregation/spatial nature Inuence of FX eect Heterogeneity of eect between U.S. Air Force vs. U.S. Army bases aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
21 / 24
Robustness Checks
Robustness checks
Table 9: Robustness analyses for impact of U.S. military withdrawal on total employment Dep. Variable: Total employment (log) - All 1. Baseline Table 4-A. estimates 2. Exclude treatment districts with pop. > most populous control district N=3,472, N(treatment)=64, N(control)=60
3. Exclude districts in urban areas N=2,380, N(treatment)=40, N(control)=45
4. Include only treatment districts with U.S. force presence N=1,960, N(treatment)=70, N(control)=0
5. Keep only treatment districts with complete closure by 1995 N=2,800, N(treatment)=40, N(control)=60
6. Include border districts
(1)
(2)
-.009*** (.003)
-.007*** (.002)
-.009*** (.003)
-.007*** (.002)
-.012*** (.003)
-.009*** (.003)
-.008*** (.003)
-.005*** (.002)
-.006** (.003)
-.004** (.002)
-.004 (.003)
-.004* (.002)
-.013** (.005)
-.008* (.004)
8. Weight by district population in 1990
-.006*** (.002)
-.006*** (.001)
9. Cameron-Gelbach-Miller two-way clustering
-.006*** (.002)
-.005*** (.002)
10. Cluster by labor market region †
-.009*** (.003)
-.007*** (.002)
-.010* (.006)
-.008 (.005)
-.009*** (.003)
-.005*** (.002)
-.009*** (.003) -.018*** (.006)
-.006*** (.002) -.021*** (.005)
Yes Yes No
Yes Yes Yes
N=5,180, N(treatment)=89, N(control)=96
7. Exclude districts with Bundeswehr reduction 1991-2001 N=1,372, N(treatment)=49, N(control)=20
11. Aggregate on level of labor market regions ‡ N=2,156, N(treatment)=48, N(control)=32
12. Include control for FX effect 13. Separate treatment group by "U.S. Army" vs "Air Force" districts U.S. WD treatment (%) - Army (N=67) U.S. WD treatment (%) - Air Force (N=3) State by year dummies District x time trends District x time2 trends
aus dem Moore (HUNotes: Berlin) U.S. Drawdown &fixed Local Labor Markets All regressions include district and year effects. Robust std. errors clustered at district 01-07-2011
22 / 24
Conclusion & Future Work
Conclusion Employment
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Conclusion Employment ▸
The U.S. drawdown in Germany is associated with signicant negative spill-over eects into local private sector employment.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Conclusion Employment ▸ ▸
The U.S. drawdown in Germany is associated with signicant negative spill-over eects into local private sector employment. The magnitude of the baseline eect is equivalent to a drop of ≈1-2 log points in employment growth for the full withdrawal in an average district.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Conclusion Employment ▸ ▸ ▸
The U.S. drawdown in Germany is associated with signicant negative spill-over eects into local private sector employment. The magnitude of the baseline eect is equivalent to a drop of ≈1-2 log points in employment growth for the full withdrawal in an average district. The heterogeneity of the eects conrms the higher vulnerability of young, low to middle educated workers in occupations/industries susceptible to suer most from a drop in local private demand.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Conclusion Employment ▸ ▸ ▸ ▸
The U.S. drawdown in Germany is associated with signicant negative spill-over eects into local private sector employment. The magnitude of the baseline eect is equivalent to a drop of ≈1-2 log points in employment growth for the full withdrawal in an average district. The heterogeneity of the eects conrms the higher vulnerability of young, low to middle educated workers in occupations/industries susceptible to suer most from a drop in local private demand. The dynamic eect pattern indicates that adverse eects persist even several years after the withdrawal.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Conclusion Employment ▸ ▸ ▸ ▸ ▸
The U.S. drawdown in Germany is associated with signicant negative spill-over eects into local private sector employment. The magnitude of the baseline eect is equivalent to a drop of ≈1-2 log points in employment growth for the full withdrawal in an average district. The heterogeneity of the eects conrms the higher vulnerability of young, low to middle educated workers in occupations/industries susceptible to suer most from a drop in local private demand. The dynamic eect pattern indicates that adverse eects persist even several years after the withdrawal. The eects are robust to a number of alternative specications.
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Conclusion Employment ▸ ▸ ▸ ▸ ▸
The U.S. drawdown in Germany is associated with signicant negative spill-over eects into local private sector employment. The magnitude of the baseline eect is equivalent to a drop of ≈1-2 log points in employment growth for the full withdrawal in an average district. The heterogeneity of the eects conrms the higher vulnerability of young, low to middle educated workers in occupations/industries susceptible to suer most from a drop in local private demand. The dynamic eect pattern indicates that adverse eects persist even several years after the withdrawal. The eects are robust to a number of alternative specications.
Wages (preliminary) ▸ ▸
The results suggest a downward adjustment of local industry wage growth primarily within small rms The eects are heterogeneous across industries
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
23 / 24
Conclusion & Future Work
Future work 1
Extensions/Robustness (for wage analysis) ▸ ▸ ▸ ▸
More disaggregate industries? Interaction by education group? Inuence of collective bargaining agreements/work councils? Dynamic pattern?
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Conclusion & Future Work
Future work 1
Extensions/Robustness (for wage analysis) ▸ ▸ ▸ ▸
2
More disaggregate industries? Interaction by education group? Inuence of collective bargaining agreements/work councils? Dynamic pattern?
Eects from base land use & location, later redevelopment
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Conclusion & Future Work
Future work 1
Extensions/Robustness (for wage analysis) ▸ ▸ ▸ ▸
More disaggregate industries? Interaction by education group? Inuence of collective bargaining agreements/work councils? Dynamic pattern?
2
Eects from base land use & location, later redevelopment
3
What are the eects of the stationing & drawdown on individual decision-making? ▸ ▸ ▸ ▸
Migration Participation Occupational trajectories, labor turnover Endogeneous skill aquisition
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Conclusion & Future Work
Future work 1
Extensions/Robustness (for wage analysis) ▸ ▸ ▸ ▸
More disaggregate industries? Interaction by education group? Inuence of collective bargaining agreements/work councils? Dynamic pattern?
2
Eects from base land use & location, later redevelopment
3
What are the eects of the stationing & drawdown on individual decision-making? ▸ ▸ ▸ ▸
4
Migration Participation Occupational trajectories, labor turnover Endogeneous skill aquisition
What are the (long-run) eects of the U.S. presence & drawdown on individual preferences, culture?
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Appendix
U.S. Forces in Germany - Total presence
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Appendix
Announcement dates
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Appendix
Descriptive Statistics (2) - Employment distribution Sex
Female Nationality
Foreign
Age groups
25-35 yrs 36-45 yrs 46-55 yrs
Education
High Medium Low
Firm size
100 N
aus dem Moore (HU Berlin)
Treatm.
1990 Contr.
.353 (.004) .089 (.005) .418 (.003) .291 (.001) .291 (.003) .055 (.005) .710 (.005) .211 (.005) .336 (.011) .200 (.005) .465 (.015) 70
.352 (.006) .082 (.005) .427 (.004) .286 (.001) .288 (.003) .042 (.003) .716 (.005) .222 (.006) .366 (.012) .202 (.005) .432 (.015) 60
Di.
Treatm.
2002 Contr.
.001 (.007) .008 (.007) -.009* (.005) .005*** (.002) .004 (.005) .014** (.006) -.007 (.007) -.010 (.008) -.030* (.016) -.002 (.007) .033 (.021)
.364 (.004) .088 (.005) .332 (.002) .392 (.001) .276 (.002) .093 (.007) .746 (.007) .125 (.003) .373 (.011) .212 (.004) .415 (.014) 70
.356 (.005) .083 (.005) .342 (.003) .389 (.001) .269 (.002) .073 (.005) .772 (.006) .124 (.004) .398 (.013) .209 (.005) .393 (.016) 60
U.S. Drawdown & Local Labor Markets
Di.
.007 (.007) .005 (.007) -.010*** (.004) .003 (.002) .007** (.003) .021** (.009) -.026*** (.009) .000 (.005) -.025 (.017) .003 (.007) .022 (.021) 01-07-2011
24 / 24
Appendix
Descriptive Statistics (3) - Employment distribution Occupation
1 Production, mining, basic materials workers 2 Craft/construction workers 3 Professionals 4 Associate Profes. /Technicians 5 Clerks and sales workers 6 Transport/Security profes. and workers 7 Research/Education profes. and artists 8 Elementary services workers N
aus dem Moore (HU Berlin)
Treatm.
1990 Contr.
.365 (.009) .080 (.004) .103 (.005) .136 (.003) .174 (.004) .091 (.002) .010 (.001) .040 (.001) 70
.383 (.010) .096 (.004) .091 (.004) .130 (.004) .158 (.004) .096 (.002) .007 (.001) .039 (.001) 60
Di.
Treatm.
2002 Contr.
-.017 (.013) -.015*** (.006) .012* (.007) .007 (.005) .016*** (.006) -.005* (.003) .002** (.001) .001 (.002)
.315 (.010) .060 (.003) .115 (.006) .154 (.003) .198 (.006) .099 (.002) .012 (.001) .047 (.002) 70
.339 (.010) .072 (.004) .106 (.005) .151 (.004) .173 (.004) .106 (.003) .009 (.001) .045 (.002) 60
U.S. Drawdown & Local Labor Markets
Di.
-.023 (.014) -.012** (.005) .010 (.007) .004 (.005) .024*** (.007) -.007* (.004) .003* (.002) .002 (.002)
01-07-2011
24 / 24
Appendix
Descriptive Statistics (4) - Employment distribution
Industry
1 Basic materials 2 Investment goods 3 Food and consumption goods 4 Construction 5 Retail/Repair 6 Transport/Info. 7 Corporate svcs. 8 Private HH. services N
aus dem Moore (HU Berlin)
Treatm.
1990 Contr.
.094 (.007) .271 (.015) .149 (.009) .100 (.004) .169 (.006) .052 (.003) .117 (.006) .047 (.003) 70
.097 (.012) .278 (.017) .167 (.011) .116 (.005) .158 (.006) .049 (.003) .089 (.005) .046 (.003) 60
Di.
Treatm.
2002 Contr.
-.003 (.014) -.007 (.023) -.018 (.014) -.016** (.007) .012 (.009) .003 (.004) .028*** (.008) .001 (.004)
.081 (.006) .238 (.014) .119 (.008) .079 (.004) .181 (.005) .062 (.003) .185 (.011) .055 (.003) 70
.085 (.011) .254 (.016) .129 (.009) .094 (.005) .182 (.007) .066 (.005) .136 (.007) .054 (.003) 60
U.S. Drawdown & Local Labor Markets
Di.
-.004 (.012) -.016 (.022) -.010 (.012) -.015** (.006) -.002 (.009) -.004 (.006) .049*** (.013) .001 (.004)
01-07-2011
24 / 24
Appendix
Descriptive Statistics (5) - Industry wages (in logs)
Industry
1 Basic materials 2 Investment goods 3 Food and consumption goods 4 Construction 5 Retail/Repair 6 Transport/Info. 7 Corporate svcs. 8 Private HH. services N
aus dem Moore (HU Berlin)
Treatm.
1990 Contr.
4.555 (.027) 4.611 (.016) 4.441 (.010) 4.423 (.011) 4.482 (.018) 4.406 (.031) 4.608 (.017) 4.249 (.034) 70
4.553 (.052) 4.552 (.015) 4.403 (.012) 4.394 (.009) 4.412 (.024) 4.351 (.010) 4.542 (.017) 4.159 (.048) 60
Di.
Treatm.
2002 Contr.
.002 (.059) .059*** (.022) .038** (.016) .029** (.014) .070** (.030) .054* (.032) .067*** (.024) .090 (.059)
4.569 (.022) 4.675 (.022) 4.425 (.014) 4.374 (.010) 4.422 (.017) 4.319 (.043) 4.538 (.030) 4.052 (.046) 70
4.567 (.052) 4.611 (.016) 4.384 (.011) 4.357 (.009) 4.423 (.039) 4.242 (.028) 4.412 (.048) 4.003 (.090) 60
U.S. Drawdown & Local Labor Markets
Di.
.002 (.057) .064** (.028) .041** (.018) .017 (.014) -.001 (.043) .077 (.051) .126** (.057) .048 (.101)
01-07-2011
24 / 24
Appendix
Evolution of employment by treatment status, 1975-2004
back aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Appendix
Evolution of unemployment rate by treatment status, 1984-2004
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24
Appendix
Employment results (2) - Age groups Dep. Variable: Employment (log) by age group U.S. Military (%) 2
R
U.S. Military (%) 2
R
U.S. Military (%) 2
R
State by year dummies District x time2trends District x time trends N Other covariates:
aus dem Moore (HU Berlin)
(1)
(2)
25-35 yrs. -.016*** -.011*** (.004) (.003) .995 .998 25-35 yrs. -.015*** -.009*** (.005) (.003) .995 .997 25-35 yrs. -.016*** -.014*** (.004) (.003) .994 .997 Yes Yes Yes Yes No Yes 3,640 3,640
(3) (4) A. - All 35-45 yrs. -.007 -.003 (.005) (.004) .996 .997 B. - Male 35-45 yrs. -.007 -.001 (.005) (.004) .995 .997 C. - Female 35-45 yrs. -.006 -.006 (.004) (.004) .995 .997 Yes Yes Yes Yes No Yes 3,640 3,640
U.S. Drawdown & Local Labor Markets
(5)
(6)
45-55 yrs. -.002 -.005* (.003) (.003) .996 .998 45-55 yrs. .001 -.003 (.003) (.003) .996 .998 45-55 yrs. -.008** -.010*** (.003) (.003) .995 .998 Yes Yes Yes Yes No Yes 3,640 3,640 01-07-2011
24 / 24
Appendix
Employment results (3) - Education groups Dep. Variable: Employment (log) by education group U.S. Military (%) R
2
U.S. Military (%) 2 R
U.S. Military (%) 2
R
State by year dummies District x time2trends District x time trends N Other covariates:
aus dem Moore (HU Berlin)
(1)
(2)
High -.008 -.004 (.006) (.005) .996 .997 High -.008 -.005 (.006) (.005) .995 .997 High -.013 -.008 (.008) (.007) .993 .994 Yes Yes Yes Yes No Yes 3,640 3,640
(3)
(4) A. - All Medium -.008*** -.007*** (.002) (.002) .998 .998 B. - Male Medium -.008*** -.006*** (.003) (.002) .997 .998 C. - Female Medium -.006** -.008*** (.002) (.002) .998 .998 Yes Yes Yes Yes No Yes 3,640 3,640
U.S. Drawdown & Local Labor Markets
(5) -.010** (.004) .994 -.002 (.005) .991
(6) Low -.010*** (.003) .996 Low
-.004 (.004) .994
Low -.014*** -.014*** (.004) (.004) .995 .997 Yes Yes Yes Yes No Yes 3,640 3,640 01-07-2011
24 / 24
Appendix
Employment results (4) - Selected occupations Dep. Variable: Employment (log) in selected occupations U.S. Military (%) 2
R
U.S. Military (%) 2
R
(1)
(2)
Craft/construction workers -.011** -.010** (.005) (.004) .989 .993 Craft/construction workers -.011** -.010** (.005) (.004) .989 .993
(3) (4) A. - All Clerks/sales workers -.003 -.005** (.003) (.002) .997 .998 B. - Male Clerks/sales workers -.002 -.002 (.003) (.003) .995 .997 C. - Female Clerks/sales workers -.003 -.006** (.003) (.002) .998 .999 Yes Yes Yes Yes No Yes 3,640 3,640
State by year dummies District x time2trends District x time trends N
Craft/construction workers .004 .004 (.021) (.018) .940 .959 Yes Yes Yes Yes No Yes 3,640 3,640
aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
U.S. Military (%) 2
R Other covariates:
(5)
(6)
Elem. svcs. workers -.010* -.014*** (.005) (.005) .992 .995 Elem. svcs. workers -.011* -.021*** (.007) (.006) .984 .989 Elem. svcs. workers -.010* -.012*** (.006) (.006) .992 .994 Yes Yes Yes Yes No Yes 3,640 3,640 01-07-2011
24 / 24
Appendix
Employment results (5) - Selected industries Dep. Variable: Employment (log) in selected industries U.S. Military (%) 2
R
U.S. Military (%) 2
R
U.S. Military (%) 2
R
State by year dummies Kreis x time2trends Kreis x time trends N Other covariates:
aus dem Moore (HU Berlin)
(1)
(2)
Construction -.006 -.004 (.005) (.004) .989 .993 Construction -.007 -.004 (.005) (.004) .988 .993 Construction .005 .001 (.009) (.008) .989 .993 Yes Yes Yes Yes No Yes 3,640 3,640
(3)
(4) A. - All Retail/Repair -.001 -.001 (.004) (.003) .995 .997 B. - Male Retail/Repair .000 -.000 (.005) (.003) .993 .996 C. - Female Retail/Repair -.003 -.001 (.005) (.004) .995 .997 Yes Yes Yes Yes No Yes 3,640 3,640
U.S. Drawdown & Local Labor Markets
(5)
(6)
Private hh. services -.011* -.012* (.006) (.006) .994 .996 Private hh. services -.008 -.012 (.009) (.010) .991 .994 Private hh. services -.013** -.012** (.005) (.005) .993 .995 Yes Yes Yes Yes No Yes 3,640 3,640 01-07-2011
24 / 24
Appendix
Bundeswehr reductions 1991-2001
back aus dem Moore (HU Berlin)
U.S. Drawdown & Local Labor Markets
01-07-2011
24 / 24