An Examination of the Local Economic Impacts of Military Base Closures

Abstract

The 1988 Base Closure and Realignment Act empowered the Commission on Base Realignment and Closure (BRAC) to undertake five rounds of significant military base closure and realignment. Investigating the economic impacts associated with base closures provides the first step in not only determining whether the BRAC Commission has effectively minimized negative impacts on affected communities but also provides guidance to policy makers in targeting future base closures, and helps affected communities prepare for the economic shocks associated with these events. This article examines (a) the local economic impacts associated with general changes in military base employment; (b) if the first four, completed rounds of BRAC closures have a significant differential impact on county-level employment; (c) whether these BRAC closures exhibit economic spillover effects on the employment of neighboring counties; and (d) if the economic impact associated with BRAC losses differs for enlisted and civilian personnel reductions.

Keywords

economic development government policy job creation military base closures

Introduction

Entering the late 1980s and early 1990s, the U.S. Government increasingly began to turn its attention to reducing and restructuring its existing domestic military forces, both to increase its capabilities to meet post-Cold War threats and as a cost-savings measure. Although the idea of force reduction and reconfiguration was not new, a decade-long stalemate between the military and Congress, whose approval was required for realignments eliminating a significant number of civilian personnel,¹ had prevented a single major base closure since the Carter administration. This situation changed with the establishment of the Commission on Base Realignment and Closure (BRAC) and the 1988 passage of the Base Closure and Realignment Act, authorizing the Commission to make comprehensive recommendations to Congress regarding base closure and realignment and freeing it of many of the constraints imposed by prior legislation. To date, the BRAC Commission² has engaged in five rounds (1988, 1991, 1993, 1995, 2005) of base closure and realignment, with its recommendations taking into account eight criteria, including the military value, budgetary savings, and local economic impacts in targeting each closure.³

Although the process has avoided much of the stalemate and acrimony characterizing previous national efforts at force restructuring, base closures under BRAC have nevertheless met with considerable opposition from local communities where elimination of military facilities and personnel are being contemplated. The expectation for residents and leaders of these communities is that base closures will negatively affect local economies as military and civilian positions are eliminated. Although the impacts of base closure and restructuring may be far less severe from a macro perspective, as many military personnel and civilian employees from closed bases can be restationed or employed in government positions at other locations, respectively, this is cold comfort for local communities faced with the uncertainty that base closure may bring about in terms of its effects on local businesses, employment, and economic growth.

Adding to this uncertainty and confusion is the variety of conclusions reached by past research on the economic impact of base closures under the BRAC Commission. Many early studies, including U.S. Department of Defense estimates, suggested local impact multipliers of greater than one, indicating that jobs lost through base closure would also result in additional employment losses in the surrounding community. Conversely, although few and far between, a handful of more careful econometric analyses showed base closures to be associated with much smaller local economic impacts, with some analyses’ evidence suggesting that closures may stimulate local economic activity in the long run. Given the relative lack of research, and especially of recent econometric analysis in this area, the question of the economic impact of base closures under the BRAC policy remains an open one. Reexamination of this policy’s effects is especially salient considering that inquiry in recent years has shifted away from estimation of local economic impacts to examination of local communities’ base reuse strategies, leaving the results of the handful of previous large-sample empirical base closure studies unconfirmed and unreflective of the impact of the latest two rounds of BRAC closures and realignments.

Determining whether local economies experience negative impacts as a result of BRAC base closure and estimations of the magnitude of those impacts provides the first step in determining whether the BRAC Commission has effectively met its objective of minimizing negative impacts on affected communities, potentially guides the Commission and other national-level policy makers in targeting future base closures, and helps communities affected by both current and future BRAC restructuring prepare for and minimize the potential economic shocks associated with these events in both the short and long terms. Using annual data from 1977 to 2005 on U.S. domestic military base personnel levels, combined with county-level employment information, this study attempts to take such a first step at informing future policy decisions by answering several important questions. First, what local economic impacts, if any, are associated with general changes in military base employment? Second, do BRAC closures have a significantly different impact on changes in county-level employment? Third, do BRAC closures exhibit economic spillover effects on the employment outcomes of counties neighboring those in which a base has closed? Finally, does the economic impact of BRAC reductions in civilian military employees differ from the impact of BRAC reductions in enlisted personnel?

The “Background and Previous Research” section of this article provides an overview of the scope of base closures and personnel reductions under BRAC, discusses previous research into the impacts of such closures, outlines this article’s contributions to that literature, and presents this study’s research questions in greater detail. The “Data” section discusses the data used in this analysis, whereas the “Method” section formally presents the regression models used in the article. Regression results are presented and discussed in the “Results” section, while the final section concludes with the policy implications of the findings and suggests refinements and expansions for future research.

Background and Previous Research

In many ways, the siting of military facilities can be seen as a form of geographically targeted economic development policy at the federal level and military bases have been highly sought after for their economic impact since the 1930s, if not before (Wright, 1988). The economic impact of such facilities on nearby local communities is not inconsequential. Early research on the impact of the U.S. Navy in San Diego suggested that just a decade after acquiring a base, nearly one third of that city’s economy was Navy-dependent (Lotchin, 1982). More recently, the 1986 selection of Corpus Christi, Texas as one of the Navy’s four new major homeports was projected to annually bring more than $100 million in wages and salaries, $160 million in new spending on goods and services, and $12 million in state and local tax revenues to that community, as well as providing short-term employment and wage benefits to the local construction industry (Wright, 1988). Even increases in personnel levels for existing bases have been linked with significant economic impacts: The 1980s relocation of the U.S. Army’s 10,000-member 10th Light Infantry Division to Fort Drum was linked to the creation of 6,000 new jobs in nearby Watertown, New York, and viewed by many in the local community as a boon for the economically depressed area (Weisskopf, 1985). Empirical research conducted by Poppert and Herzog (2003) on a nationwide sample of counties also indicated that counties with military facilities exhibit a significantly higher annual rate of employment growth, estimated at an additional 1,580 to 1,630 positions per year, relative to nonmilitarized locations. Although the distribution of such economic activity may potentially be inefficient from a national perspective, and while there are compelling theoretical reasons to suspect that military employees’ spending may not exert as great an impact as might be expected on local economies,⁴ given the positive economic outcomes associated with at least some base location decisions, it is not unreasonable to expect negative impacts on nearby areas from base closures.

Yet despite the potential impact of base closures on local economies and the large number of closures in recent years resulting from the five BRAC rounds, scholarly inquiry into the effects of these policies remains relatively underdeveloped. Much of the early work in this area was confined to case studies examining individual locations or a handful of nearby base closures (e.g., Beyard, 1987; Bradshaw, 1999; Glassberg, 1995; Hartwig, 1989; Hill, 2000; Laubernds, 1989). The generalization of such studies’ conclusions, and even examples of more empirically sophisticated analyses such as Dardia McCarthy, Malkin, and Vernez (1996), can be questioned on the grounds of the small samples used in their analyses. Another important division in the existing literature is that between predictive work and post hoc evaluations. Although many evaluations of potential closure sites, especially those commissioned by targeted communities, have produced conspicuously large impact multipliers,⁵ even more careful multiplier analysis indicates at least a modest negative impact, falling in a range such as 1.18 to 1.68 (Mueller, Hutchinson, Goldstone, & Moore, 1993), a figure in agreement with the multiplier of 1.51 produced by the Office of the Secretary of Defense’s (1998) own analysis of potential BRAC impacts. Although these estimates may quell some trepidation over the potentially catastrophic impacts of base closure, they nevertheless differ from many post hoc evaluations, again mostly studies of single locations, which often present examples of communities where lost military jobs were replaced by private industry activity on a greater than one-to-one basis (Spiegel, 1997), where unemployment levels decreased (Laubernds, 1989), or where the industrial composition of the local economy diversified (Beyard, 1987). Although Bradshaw (1993, 1999) provides a persuasive rationale for expectations that negative base impacts may be somewhat muted, without a large-sample post hoc evaluation of base closures, it is difficult to determine whether the modest losses predicted by the multipliers noted above or locations that seemingly rebound from base closures without missing a beat characterize the bulk of communities’ base closure experiences.

A handful of more recent studies have attempted to remedy this deficiency by using panel data covering multiple locations, with results generally supporting those of earlier multiplier-based examinations, indicating small but significant negative economic effects. Isserman and Steinberg (1994) concluded that when compared with similar Florida counties and to the state as a whole, the employment growth rates of Florida counties with base closures displayed only modest decreases. Using microlevel data, Krizan’s (1998) work on base closures and realignments in California from 1989 and 1996, comparing BRAC-targeted locations with other, unaffected bases and with comparable nonmilitarized California counties, found that base employment increases had a significant but small (0.6%) impact on local establishments’ growth rates. Using a national data set containing counties experiencing military base closures from 1977 to 1994, a difference-in-difference analysis by Hooker and Knetter (1999) of county-level employment, and income associated with base closures when compared with own-county baselines found similarly significant but small negative impacts of closure-related personnel losses on employment and no significant effect on personal income. Likewise, Hooker and Knetter’s simple multiplier estimates indicate 1.09 jobs lost per position eliminated due to base closure, although when base personnel decreases are differentiated by civilian and military job types, only the multiplier (1.24) associated with the loss of military personnel remains significant. Their findings also indicated that job losses are confined to years immediately prior to, and including, the year of closure, suggesting that most civilian employees displaced by base closure may ultimately be reemployed in the local labor force. Using similarly sophisticated methodology and panel data on all U.S. counties from 1978 to 1999, Poppert and Herzog (2003) conclude that each position eliminated through BRAC force reductions (including both base closures and realignments) is associated with a county-level employment loss of 1.01 to 0.60 positions, depending on model specification. However, combining the negative effects associated with job loss with positive effects linked to base reuse and time since BRAC announcement, Poppert and Herzog find that in the long term, personnel reductions under BRAC are actually associated with county-level employment growth.

Nevertheless, a number of areas remain where existing analysis of the economic impacts of base closures under BRAC can be refined and extended. First, the question of base closure impacts has not been revisited using econometric techniques and a large-scale data set since Poppert and Herzog (2003), which used data through 1997. Roughly a decade’s worth of additional BRAC activity has occurred since this time, including full completion of all base closures designated by the 1995 BRAC committee.

Second, prior studies have not considered the possibility of geographic spillover effects of base closures. As neighboring counties often share similar labor and input markets and have similar industrial compositions, it is reasonable to expect that the effects of base closures, positive or negative, do not stop at political boundaries, and that employment in neighboring counties may be affected by BRAC policy in addition to the effects anticipated for counties containing bases. Analyses failing to consider the possibility of BRAC-related economic spillover effects in neighboring counties may thus miss, and perhaps substantially underestimate, the total impact of changes in military employment on local employment.

Third, although analysis of base closures’ economic impacts does need to compare observed growth with the counterfactual of growth expected in the absence of base closures, to date, no study has done so adequately. When analyzing the impact of economic development policy on local economic outcomes, “[m]eaningful difference in difference estimates depend on the construction of proper comparison groups. An appropriate set of comparison [areas] should have characteristics that make them likely candidates to be designated” as a target for similar policy treatments (Greenbaum & Engberg, 2004, pp. 324-325). Based on this rationale, the most appropriate comparison group would be other counties potentially targeted for base closure under BRAC, and which would be affected by that policy in a similar manner due to a high degree of economic dependence on military activity and high structural incorporation of that activity into a location’s economy. That is, the comparison group should consist of other counties with military bases (and by the argument above, neighboring areas) only. Although some past studies have included independent variables controls for economic similarity, no study to date has made the comparison of BRAC-targeted and neighboring counties to non-BRAC base counties and surrounding areas only. Isserman and Steinberg’s (1994) study, for example, is limited to Florida bases and does not make the comparison of BRAC base counties to non-BRAC base counties. Hooker and Knetter’s (1999) comparison of base counties with their own (and state-level) preclosure growth rates assumes that such growth would be expected to persist, an assumption which may not be accurate and is unnecessary with the presence of a control group of non-BRAC base counties with similar economic dependence on their military facilities. Poppert and Herzog’s (2003) sample, in contrast, is too large, encompassing all U.S. counties and drawing comparisons not only between BRAC base counties and non-BRAC base locations but also between BRAC-affected areas and sites with no economic dependence on military activity, and which may react to changing economic conditions quite differently than locations where a military presence is embedded in the local economic structure. And although Krizan’s (1998) research represents the best example of establishing such a counterfactual, those results deal solely with California base closures, leaving open the questions of whether their findings are generalizable to bases in other locations.

To address these shortcomings, this study adopts the following approach: Using a panel of economic and military personnel data from 1977 to 2005 with annual observations for all counties containing military bases and counties neighboring those base counties, the impact of BRAC base closure decisions on changes in employment will be estimated using an equation containing time and location-specific fixed effects and controls for other socioeconomic factors. Closure decisions will be modeled by creating a series of dummies and interaction terms, similar to Poppert and Herzog (2003). As in Hooker and Knetter (1999), force levels will also be decomposed into civilian personnel and military personnel to determine if each type of force reduction exhibits differential impacts on the local economy. This estimation strategy is discussed in greater detail in the following section.

Data

The data used in this article are compiled from two primary sources. Information on military force levels are drawn from the U.S. Department of Defense’s annual Distribution of Personnel by State and by Selected Locations reports for the years 1977-2005. These reports present data on total military (enlisted) personnel and civilian personnel employed in each major domestic U.S. military base for a given year, covering a total of 948 individual Department of Defense bases in the 50 states and the District of Columbia that were active during this 27-year span. Each of these installations was then identified by location (state and county) and military employment was aggregated at the county level, producing observations for 510 individual counties containing one or more military bases active from 1977 to 2005. Additionally, all neighboring counties containing one or more military bases were also identified, assigned information on the total number of military and civilian employees located on bases in those neighboring counties, and added to the data set. This process produced annual observations for 1,721 individual counties (510 counties containing one or more bases and 1,211 counties without a base but neighboring one or more counties with bases). A dummy variable indicating whether or not the county in question was a “base county,” containing one or more military bases or a “neighboring county” was also created for all observations.

These data were combined with information on locations affected by a base closure as a result of BRAC Commission actions in its 1988, 1991, 1993, or 1995 rounds, summarized in Table 1 by closure round. Closures were identified using U.S. Department of Defense’s (2004) Section 2912 Report, used to identify Department of Defense–designated “major” bases closed during the 1988, 1991, 1993, and 1995 rounds. Counties were coded using a dummy variable indicating whether a location was affected (i.e., was in the year of closure designation or the subsequent 6-year window⁶ allowed for closure implementation) by one or more BRAC closures from 1991-1995 rounds in the current year or prior years

Table 1.

Major Base Closures by BRAC Round, 1988-1995

BRAC round	Number of base closures^a
1988	39
1991	36
1993	49
1995	24
Total	148

Note. BRAC = Base Realignment and Closure.

Does not include closures of reserve centers or housing-only facilities.

These county-level data on military base employment and BRAC closures were then paired with information on changes in annual employment, total employment, and share of county employment arising from service activities from the U.S. County Business Patterns data set for the period 1977-2005. After data cleaning, the combination of the above data sources produced a panel of 46,216 county–year observations containing annual information on own-county and neighboring-county military base employment, nonmilitary employment, other measures of economic activity, and socioeconomic controls covering 948 different domestic military installations spanning 509 individual “base counties” and 1,211 discrete “neighboring counties” (Table 2).

Table 2.

Descriptive Statistics for Selected Variables

Variable	Mean	SD	Minimum	Maximum
ΔTotal base personnel, own county	−8.727	604.444	−30,148	37,003
ΔTotal base personnel, neighboring county	−40.254	4,456.856	−440,436	440,925
ΔTotal base enlisted personnel, own county	−2.715	496.784	−27,296	34,995
ΔTotal base enlisted personnel, neighboring county	−16.828	1,586.105	−89,399	94,567
ΔTotal base civilian personnel, own county	−5.874	236.557	−13,990	14,105
ΔTotal base civilian personnel, neighboring county	−22.977	1,207.265	−109,493	109,544
Total employment	48,434.33	156,762.500	0	3,895,886
Percentage service	0.500	0.136	0	1.667
ΔTotal employment	950.757	5,874.011	−164,203	274,133
ΔTotal wages	73.577.56	187,000,000.00	−29,500,000,000	29,500,000,000

Method

This article uses a general method of estimation where annual change in employment in a particular county is expressed as a function of annual changes in military base employment within that county, annual changes in military base employment in neighboring counties, a vector representing local economic controls, a vector representing base closures designated through BRAC policy, a county-specific fixed effect, and a time-specific fixed effect. This article confines itself to examining only those counties containing an active military base for one or more years during the period 1977-2005 and all counties neighboring those counties with military bases active for one or more years during this period. In order to appropriately assess the impact of BRAC base closures, it is necessary to establish the counterfactual of the local economic growth expected in the absence of such policy interventions. This study does so by selecting a comparison group of similar counties that were also likely candidates to be affected by the policy in question, but were not so designated. Such a group includes only other counties with military bases and neighboring counties rather than the full population of all U.S. counties for two important reasons. First, only counties with existing military bases would be candidates for BRAC base closures—a county can only be directly affected by a base closure if that county has a base to close. Second, counties with bases may have very different economies than counties without bases if the economic activities and impacts of bases are integrated into local markets for goods, services, and labor. Local economies with military bases are likely far more similar to other counties with bases than they are to counties that do not depend on base activity for their economic fortunes. This argument can be extended to counties neighboring those containing military bases. If these nearby locations participate in the same local market for goods, labor, and so on, as those neighboring counties with bases, any disruption of the economic activity generated by those bases could produce a spillover effect felt in the economies of geographically proximate locations. For these reasons, the population of counties considered in this article is thus confined to base-affected counties; that is, counties containing a military base and counties neighboring a county containing a base.

An additional question is whether local economies respond differently to force-level changes resulting from base closures under BRAC policy than they do to force-level changes occurring under other, more ordinary circumstances. BRAC-related personnel reductions might elicit a different local economic response for a number of reasons. First, such changes are highly visible in nature. Whereas ordinary rises and falls in base employment might otherwise fly under the collective radar of the general public, and thus may not be fully responded to, BRAC base closures are policy events of which local communities, especially those directly affected, are likely to be highly aware. Second, although the permanence of ordinary gains and losses in base employment are likely to be uncertain—it may be just as likely that a reduction in employment for an active base will be followed by an employment increase in the next year as it would be by additional losses—one cannot make a similar argument for force-level drops associated with base closures. Once a base is closed, it is certain those jobs will not be filled by future military employment within that community. Thus, given the differential expected “permanence” of ordinary military job losses versus those occurring under BRAC base closures, local economies may respond to these two types of losses in very different fashions, and BRAC-related losses may precipitate a similarly permanent reduction in local economic activity that depends on base employees as customers. Closures may also create a “chilling effect” on growth whereby potential entrants into the local market may be less likely to locate within a community due to the removal of the market demand created by a base and/or its employees.

To measure the effect of changes in military base employment levels on local economic activity and account for unique characteristics of individual counties and years that may not be sufficiently captured by other independent variables included in the model, this article uses panel data techniques incorporating both year and county-specific fixed effects. As this estimation deals with a specific set of units, the entire population of U.S. counties containing military bases or neighboring counties with bases, rather than drawing a random sample of such counties, and as the impact of changes in base employment, both BRAC and non-BRAC in nature, are only intended to be generalized to other locations containing or neighboring bases (i.e., other observations within this population), the use of fixed effects is a more appropriate approach than the alternative of modeling through random effects (Baltagi, 2002; Hsiao, 2003).

This approach produces the following model to be estimated:

Δ EM P_{i, t + 1} = β_{0} + β_{1} Δ BAS E_{i, t} + β_{2} Δ BASE N_{i, t} + β_{3} Δ BRA C_{i, t} + β_{4} Δ BRAC N_{i, t} + β_{5} Δ EM P_{i, t + 1} + β_{6} Δ PCTSV C_{i, t} + 1 + α_{i} + γ_{t} + e_{i, t} + 1 .

The dependent variable for Equation (1) is the annual change from period t to period t + 1 in employment for county i. Importantly, as County Business Patterns data are used as the source of county employment information, employment in this case will not include public-sector employment activity such as enlisted and military base employment. The marginal impact of changes in base employment in this case is thus interpreted as the resulting change in local economic activity in addition to a one-unit change in employment for a given base.

The independent variables in the above equation are the change in total base employment from period t − 1 to period t for counties containing one or more military bases (ΔBASE_{i, t}); the change in total base employment from period t − 1 to period t for counties neighboring one or more counties containing military bases (ΔBASEN_{i, t}; this variable thus represents the change in aggregate base employment across all such neighboring counties); an interaction term, BRAC_{i, t} equal to ΔBASE_{i, t} if the county in question contains one or more bases during year t in the process of closing⁷ under the 1988, 1991, 1993, or 1995 rounds of BRAC base closures and equal to 0 otherwise; a second interaction term, BRACN_{i, t} equal to ΔBASEN_{i, t} if the county in question neighbors counties containing one or more bases during year t in the process of closing⁸ under the 1988, 1991, 1993, or 1995 rounds of BRAC base closures and equal to 0 otherwise; EMP_{i, t + 1}, the total county employment in period t + 1 (again, excluding on-base and other public-sector employment); PCTSVC_{i, t + 1}, the percentage of total county employment arising from service sector activities in period t + 1; a county-specific fixed effect, α_i; and a year-specific fixed effect γ_t.

Annual changes in total base employment are also separated into two terms, changes in total military (enlisted) personnel employed at a given base and changes in total civilian base employment, as employment shifts in these two subgroups may result in different impacts on nonbase local employment totals. Enlisted employees may exit a local economy completely if a base closes or decreases its force levels, as there are likely to be few if any perfect local employment substitutes for active military duty, with this exit precipitating a decrease in the local demand for goods and services that their presence created. Civilian employees, on the other hand, may be more closely tied to the community rather than the base. If off-base employment is viewed as an acceptable substitute for eliminated on-base positions, they may remain in a location, contributing to that local economy even if their on-base employment is terminated. A reduction in civilian employees may not be associated with the same magnitude of loss, if any, in local demand for goods and services, and drops in on-base civilian employment may actually increase off-base employment if base closures cause at least some civilian labor to seek employment with other local, nonmilitary firms. Moreover, if local civilian and enlisted employment impacts are of opposite sign and similar magnitude, examining the effects associated with aggregate base employment only may mask or otherwise miss these counterbalancing yet important impacts (Greenbaum & Engberg, 2004; Peters & Fisher, 2002).

Such differentiation of Equation (1)’s base employment terms produces the following model:

Δ EM P_{i, t + 1} = β_{0} + β_{1} Δ BASE M_{i, t} + β_{2} Δ BASE C_{i, t} + β_{3} Δ BASM N_{i, t} + β_{4} Δ BASEC N_{i, t} + β_{5} BRAC M_{i, t} + β_{6} BRAC C_{i, t} + β_{7} BRACM N_{i, t} + β_{8} BRACC N_{i, t} + β_{9} EM P_{i, t + 1} + β_{10} PCTSV C_{i, t + 1} + α_{i} + γ_{t} + e_{i, t + 1} .

In each instance, the individual terms associated with Equation (1)’s coefficients β₁ through β₄ have been split into two terms to model the impacts of changes in bases’ military (enlisted) employment (β₁, β₃, β₅, and β₇) and changes in bases’ civilian employment (β₂, β₄, β₆, and β₈) in Equation (2). Apart from this disaggregation, the general definitions of these variables and the construction of the interaction terms remain the same as in Equation (1).

Results

Tables 3 and 4 present the results from the estimation of Equations (1) and (2), respectively. Turning first to Table 3, the effects associated with changes in total military base employment and BRAC policy overall on annual changes in annual county employment, there is a statistically significant and positive association between changes in the number of base employees and county-level employment, with each increase in base employment associated with a 0.14 job increase in other county employment. Changes in base employment levels are not, however, associated with any significant effect on employment in neighboring counties, suggesting that, at least considering changes in overall base employment, economic impacts may be confined to the immediate county in which the base is located. Interestingly, the insignificant coefficients associated with the BRAC interaction terms indicate that bases closed under those BRAC policy rounds respond to increases and decreases in base personnel no differently than other non-BRAC bases gaining or losing employment during the 1977-2005 period.

Table 3.

Regression Results for the Effect of Change in Total Military Base Employment on Change in County Employment, 1977-2005

Independent variable^a	Coefficient value (p value)
ΔTotal base personnel, own county	0.142 (.004)
ΔTotal base personnel, neighboring county	−0.002 (.645)
BRAC interaction, own county	−0.092 (.286)
BRAC Interaction, Neighboring County	−0.046 (.167)
Total employment	−0.046 (.000)
Percentage service	−287.737 (.467)
N = 46,216	df = 44,462	F = 111.58 (0.000)	Adjusted R² = .229

Note. BRAC = Base Realignment and Closure. Entries in boldface signify that the coefficient is significant at a p < .05 level or higher.

For ease of presentation, results for county and year fixed effects have not been reported here.

Table 4.

Regression Results for the Effect of Change in Military Base Employment (by Type) on Change in County Employment, 1977-2005

Independent variable^a	Coefficient value (p value)
ΔTotal base enlisted personnel, own county	0.190 (.002)
ΔTotal base civilian personnel, own county	−0.226 (.052)
ΔTotal base enlisted personnel, neighboring county	0.090 (.000)
ΔTotal base civilian personnel, neighboring county	−0.015 (.441)
BRAC enlisted interaction, own county	0.124 (.264)
BRAC civilian interaction, own county	−0.865 (.001)
BRAC enlisted interaction, neighboring county	−0.012 (.847)
BRAC civilian interaction, neighboring county	−0.405 (.001)
Total employment	−0.046 (.000)
Percentage service	−293.345 (.458)
N = 47,216	df = 44,462	F = 101.49 (.000)	Adjusted R² = .231

Note. BRAC = Base Realignment and Closure. Entries in boldface signify that the coefficient is significant at a p < .05 level or higher.

For ease of presentation, results for county and year fixed effects have not been reported here.

For reasons noted above, it is also important to disaggregate the impact of changes in on-base civilian and on-base enlisted employment. As suspected, these results (Table 4) suggest that local economic responses to enlisted base employment and civilian base employment are quite different, both for non-BRAC employment changes and for responses to BRAC-precipitated force-level shifts. Here, changes in enlisted base employees are associated with significant and positive impacts on county employment for both base employment changes in one’s own county, where each additional enlisted job gain is linked to 0.19 additional nonbase jobs, and for base employment changes in neighboring counties, where the incremental enlisted job is associated with a 0.09 job gain in nonpublic-sector employment. Although not significant at the 5% level, the coefficients for job change associated with both own-county and neighboring-county civilian base employment are negative in sign, suggesting that whereas enlisted on-base employment is a complement to nonbase county economic activity, on-base civilian employment is at least a partial substitute for employment outside the base. Notably, the magnitude of the coefficients linked to own-county shifts in base employees are larger than the magnitude of the coefficients linked to neighboring-county shifts in base employees; employment shifts in nearby bases thus appear to have a greater impact on local economic conditions than employment changes in neighboring-county bases further away.

In contrast, when changes in civilian and enlisted base employees are disaggregated as in Table 4, a number of additional significant effects for 1988-1995 BRAC closures manifest themselves. Beyond the impacts associated with ordinary changes in civilian on-base employment, the incremental civilian base job gain in BRAC 1988-1995 closure counties is linked to a 0.87 job loss in nonbase employment beyond the effects otherwise associated with changes in civilian base employment, whereas each additional civilian base job gained in neighboring counties affected by a 1988-1995 BRAC closure is linked to a 0.41 job loss in own-county employment. Given that these interaction terms are additive in nature and that that the sign of the incremental BRAC 1988-1995 round effects “agree” (when significant) with the sign of the effects associated with “ordinary” changes in base employment, the interpretation of these findings is that force-level changes under the 1988-1995 BRAC rounds exacerbate the county employment effects associated with ordinary changes in civilian base employment; no additional impact is observed, however, for BRAC-based changes in enlisted personnel beyond the effect associated with ordinary, non-BRAC changes in such employment.

As the coefficients presented in Tables 3 and 4 represent nonpublic-sector employment gains and do not include changes in base employment, and given that the BRAC-related interaction terms are additive in nature, for ease of interpretation, Tables 5 and 6 present the expected net change in total local employment resulting from the marginal on-base job. The figures in Table 5 are calculated by adding the one-job gain in base employment to the appropriate marginal gain in off-base employment indicated by the coefficients in Table 3 (if significant) to produce the “active base” value in the first column. The interaction effect indicated by the coefficient for bases affected by 1988-1995 BRAC closures is added to this value to produce the “Bases Closed by BRAC Policy” value. Table 6 presents similar values for the disaggregated impacts of the marginal one-unit gain in enlisted and civilian base personnel calculated from the coefficients contained in Table 4. To illustrate, Table 6 indicates that an increase of one enlisted base job will result in a total gain of 1.190 jobs, whereas an increase of one civilian base job for a location with one or more bases closed under the 1988-1995 BRAC rounds will yield a total gain of 0.135 jobs. As Tables 5 and 6 present the net change in total county employment expected from a one-job gain in base employment, the expected change in total county employment associated with a job (total, enlisted, or civilian) lost, as would likely result from BRAC closure policy, reverses the signs in these tables.

Table 5.

Predicted Impact on County Employment per Unit Change in Total Military Employment Under Various BRAC Policy Environments, 1977-2005

	Active bases	Bases closed by BRAC policy, 1988-1995
Per unit change in total base personnel, own county	1.142	1.142
Per unit change in total base personnel, neighboring county	1.000	1.000

Note. BRAC = Base Realignment and Closure.

Table 6.

Predicted Impact on County Employment per Unit Change in Military Employment (by Type) Under Various BRAC Policy Environments, 1977-2005

	Active bases	Bases closed by BRAC policy, 1988-1995
Per unit change in enlisted base personnel, own county	1.190	1.190
Per unit change in civilian base personnel, own county	1.000	0.135
Per unit change in enlisted base personnel, neighboring county	1.090	1.090
Per unit change in civilian base personnel, neighboring county	1.000	0.595

Note. BRAC = Base Realignment and Closure.

The expected value for each cell in Tables 5 and 6 would be 1.00 if there was no significant impact on local nonbase employment—adding one job to base employment would simply add that one job to total county employment. Positive values greater than one indicate that the creation of economic activity beyond the base job added, that adding a base job has a multiplier effect on total community employment, or alternately, each enlisted job lost due to a force reduction removes not only that job from the community but also “costs” a fraction of an additional job as well. Positive values less than one suggest a mild substitution effect between on-base and off-base civilian employment, with losses of on-base civilian jobs mitigated by the fact that some (but not all) of those employees may find other employment within the local economy. The addition of enlisted employees to any military base thus creates an own-county spillover effect of sorts, adding employment beyond the single job added to the base, although no such additional impact is associated with the marginal civilian employee. In cases where bases are closed due to BRAC policy from 1988-1995, each civilian job eliminated is linked to a decrease of 0.14 jobs, suggesting that at least some civilian base employees are able to find other local, nonmilitary employment.

Conclusion

The results presented above suggest that changes in military base employment do indeed have a statistically significant impact on local nonbase employment, both within the same county and in neighboring counties. In both cases, additional enlisted military personnel are linked to a positive economic impact resulting in job creation extending beyond their own position, and this economic impact also has a positive and significant, if somewhat smaller spillover effect on employment in neighboring counties. It is worth noting the degree to which these findings confirm those of previous studies, specifically the work of Hooker and Knetter (1999) in that the impacts observed are confined to enlisted rather than civilian employees when base employment is disaggregated. The magnitude of the observed effect (1.19) is also quite close to Hooker and Knetter’s observed multiplier of 1.24. As well, the impact of BRAC-related employment gains and losses for civilian employees are notably different from the impacts of changes in base employment occurring under other circumstances.

The contributions of this article are thus threefold. First, the work provides empirical confirmation of the impact of changes in military base employment on other local economic activity. Second, it provides evidence that changes in military base employment under the U.S. Department of Defense’s extremely visible BRAC policy process produce impacts differing from those occurring under other circumstances. And third, it improves on previous research, extending the available data by 8 years (including post-drawdown data for all bases closed under the 1995 BRAC round, unexamined by previous research), constructing a more appropriate comparison group of only counties containing or neighboring locations with a military base and considers potential spillover effects of military base activity on neighboring counties’ economies in addition to own-county effects.

Nevertheless, although the findings of this article suggest that policy learning may not have occurred over the four BRAC rounds examined, that BRAC base closures are still associated with important and potentially negative consequences, and that future policy might therefore be improved, these findings do not answer the equally important question of exactly how to achieve such improvement. Moreover, although this research indicates that locations experiencing base closures might expect negative local economic consequences, it does not provide prescriptions for how those locations can minimize or avoid such losses. It is also important to note that this article’s model does not explicitly account for local policies designed to reuse base infrastructure or otherwise blunt the negative impacts associated with base-related employment losses, and a better modeling of local responses is certainly one manner in which the current model might be extended. Costs accruing to locations experiencing BRAC-related employment losses may also be partially counterbalanced by other benefits, such as compensatory treatment and a higher rate of award for programs intended to mitigate the impact of base closures or other types of federal economic development assistance. Although this article’s objective is to quantify employment gains or losses associated with military force reductions in order to inform future policy-making efforts, it is likewise important to note that this article does not itself aim to undertake such a complete cost/benefit analysis. Consideration of whether locations targeted for base closures are given preference in other policy areas relative to counties not experiencing base closures is thus left to future work.

As well, several important questions remain for future research. First, this article does not consider the impacts associated with the 2005 round of BRAC base closures. Given that bases have a 6-year window in which to cease operations, many bases affected by the 2005 closures are still in the process of drawing down force levels. The full local economic impacts associated with these changes, positive or negative, may not yet be realized at the time of this writing, and the use of currently available data to consider the impacts associated with bases that have already drawn down their force levels fully may be unrepresentative of the overall local economic effects for all bases closed in this latest policy round. Nevertheless, there may be compelling reasons to expect the impacts of BRAC policy from the first four rounds (1988-1995) to differ from the impacts associated with the most recent 2005 round of closures. One might expect that the experience of earlier closure rounds would allow policy “learning” in later rounds for both national actors and local community leaders. Alternately, if earlier BRAC committees selected “easy” cases for closure first, 2005 BRAC closures may be drawn from remaining bases, where higher levels of negative economic consequences might be expected.⁹ Although it is unclear which of these scenarios applies in practice, both possibilities present a persuasive argument as to why economic impacts might differ by round of BRAC closures, and why differences in these rounds should thus be examined by future research to determine whether the experience of prior rounds left later BRAC Committees better able to minimize negative economic impacts in their selection of base closures and enabled affected communities to better prepare in realistically anticipating and dealing with associated negative shocks to the local economy.

Although this article examines the impact of changes in military base employment on local employment, other important indicators of local economic activity may also be affected by shifts in base job levels. Total firms and firm birth and death rates may be other important indicators in which the impacts of base employment and closures are felt if shifts in military employment precipitate a high rate of new business creation, firm turnover, or a mass exodus of establishments from a local economy. If base closures affect the consumption of goods and services and the creation of jobs associated with these sectors, base expansion and reduction may also shift local economic composition toward or away from a higher degree of service-sector firms and employment, suggesting another area for future examination. As well, whether or not shifts in base employment affect local housing markets might also be explored.

In light of recent shifts in America’s economic landscape, as well as the changing role of the U.S. military both domestically and internationally, the appropriate focus, size, and distribution of America’s military employment remains an open question, and one which is likely to be revisited in coming years. The evidence presented in this article suggests that the economic impacts of these choices are significant ones, and must continue to be examined and considered carefully if policy pitfalls are to be avoided and policy opportunities realized, moving forward.

Footnotes

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Baltagi

B. H.

(2002). Econometric analysis of panel data (2nd ed.). Chichester, England: John Wiley.

Beyard

(1987). When the military leaves town: New life for an old Air Force base. Urban Land, 46(6), 6-10.

Bradshaw

T. K.

(1993). Which impact?: The local impact of base closure needs closer examination. (Working Paper No. 602). Berkeley: University of California, Institute of Urban and Regional Development.

Bradshaw

T. K.

(1999). Communities not fazed: Why military base closures may not be catastrophic. Journal of the American Planning Association, 65, 193-206.

Dardia

McCarthy

Malkin

Vernez

(1996). The effects of military base closures on local communities: A short term perspective. Santa Monica, CA: RAND Corporation.

Glassberg

A. D.

(1995). Intergovernmental relations and base closing. Publius, 25(3), 87-98.

Greenbaum

R. T.

Bondonio

(2004). Losing focus: A comparative evaluation of spatially targeted economic revitalization programmes in the U.S. and the E.U. Regional Studies, 38, 319-334.

Greenbaum

R. T.

Engberg

(2004). The impact of state enterprise zones on urban manufacturing establishments. Journal of Policy Analysis and Management, 23, 315-339.

Hartwig

R. P.

(1989). Life after death of a military base: The case of Chanute Air Force Base. Illinois Business Review, 46(3), 8.

10.

Hill

(2000). Measuring success in the redevelopment of former military bases: Evidence from a case study of the Truman Annex in Key West, Florida. Economic Development Quarterly, 14, 267-275.

11.

Hooker

M. A.

Knetter

M. M.

(1999). Measuring the economic effects of military base closures. (NBER Working Paper No. 6941). Cambridge, MA: National Bureau of Economic Research.

12.

Hsiao

(2003). Analysis of panel data (2nd ed.). Cambridge, England: Cambridge University Press.

13.

Isserman

Steinberg

(1994, April). The recovery of rural economies from military base closures: Control group analysis of two decades of experience. Paper presented at the Southern Regional Science Association Meeting, Orlando, FL.

14.

Krizan

C. J.

(1998). Localized effects of California’s military base realignments: Evidence from multi-sector longitudinal microdata (Vol. 18, Issue 19, Center for Economic Studies discussion paper). Washington, DC: Center for Economic Studies, U.S. Bureau of the Census.

15.

Laubernds

W. L.

(1989). Opportunities abound at closed military installations. Real Estate Finance Journal, 4, 37-41.

16.

Lotchin

R. W.

(1982). The city and the sword in metropolitan California, 1919-1941. Urbanism Past and Present, 7(2), 1-16.

17.

Mueller

Hutchinson

R. A.

Goldstone

D. E.

Moore

W. B.

(1993). A method for estimating local impacts of cuts in defense spending. Bethesda, MD: Logistics Management Institute.

18.

Office of the Secretary of Defense. (1998). The report of the Department of Defense on base realignment and closure. Washington, DC: Author.

19.

Peters

A. H.

Fisher

P. S.

(2002). State enterprise zone programs–Have they worked? Kalamazoo, MI: W. E. Upjohn Institute for Employment Research.

20.

Poppert

P. E.

Herzog

H. W.

Jr. (2003). Force reduction, base closure, and the indirect effects of military installations on local employment growth. Journal of Regional Science, 43, 459-481.

21.

Spiegel

(1997, September 22). Close a base, create a job. Forbes, pp. 110-112.

22.

U.S. Department of Defense. (1979-2007). Distribution of personnel by state and by selected locations. Washington, DC: Author.

23.

U.S. Department of Defense. (2004). Report required by Section 2912 of the Defense Base Closure and Realignment Act of 1990, as amended through the National Defense Authorization Act for Fiscal Year 2003. Washington, DC: Author.

24.

Weisskopf

(1985, May 27). A small town’s Army bonanza. The Washington Post, p. A3.

25.

Wright

D. S.

(1988). Understanding intergovernmental relations (3rd ed.). Pacific Grove, CA: Brooks/Cole.