Weapon of the Weak?:
Assessing the Effects of State Capacity on Terrorism1
Cullen S. Hendrix
Department of Government
College of William & Mary
chendrix@wm.edu
and
Joseph K. Young
School of Public Affairs
American University
jyoung@american.edu
Abstract
Conventional wisdom suggests that dissident groups use terrorism when they face an
overwhelming more powerful state. Given this core belief, cross-national studies of terrorism
find mixed results for how common measures of state capacity influence terrorism. We argue
that these indeterminate findings are due in part to a partial understanding of both what
constitutes state capacity and how different aspects of state strength or weakness relate to the
propensity of groups to use terrorism. We decompose state capacity into two dimensions that we
theorize are particularly relevant to dissident groups: military capacity, or the ability to project
conventional military force and bureaucratic/administrative capacity, or the ability to collect and
manage information. Our analysis supports the claim that terrorist attacks are more frequently
targeted at states with large, technologically sophisticated militaries, but less frequently targeted
at states with higher bureaucratic and administrative capacity. States can be capable in different
ways, and these various capabilities create differing incentives for using terror as a strategic and
tactical tool.
Keywords
Terrorism, state capacity, violence, military capacity, bureaucracy
1
A version of this paper was presented at the 2012 Midwest Political Science Association Annual Meeting,
Chicago, IL. The authors would like to thank Chris Gelpi, Andrew Boutton, and Mike Findley for helpful
comments and Alex Testa for research assistance.
Introduction / Problem
As Crenshaw (1981, 387) observed decades ago, “[t]errorism is a logical choice when
oppositions have such goals and when the power ratio of government to challenger is high. The
observation that terrorism is a weapon of the weak is hackneyed but apt.” Crenshaw’s
“hackneyed” observation has remained a conventional wisdom for scholars and policy makers.
In an Op-Ed to the New York Times, former US President, George W. Bush argued that,
Poverty does not transform poor people into terrorists and murderers. Yet, poverty,
corruption and repression are a toxic combination in many societies, leading to weak
governments that are unable to enforce order or patrol their borders and are vulnerable to
terrorist networks and drug cartels.2
In Bush’s formulation, state weakness allows groups to operate with impunity and thus the
factors that make a state weak can influence terrorism both internally and abroad. In contrast to
Crenshaw, this view suggests state weakness, rather than state strength, promotes terrorism.
Kofi Annan (2006, 15), the former Secretary-General of the United Nations, argued for
increasing state capacity to prevent terrorism. Similar to Bush, his concern was related to weak
states serving as a base of operations for terrorist groups “to fund, organize, equip and train their
recruits, carry out their attacks, and hide from arrest.” If terrorism is a weapon of the weak, then
we should expect to see increasing amounts against more capable states. Yet, as Bush and
Annan argue, weaker states may be bases and targets of groups as well. The empirical findings
linking state capacity to terrorism are similarly mixed. GDP per capita, a measure used in civil
war studies to proxy state capacity (Fearon and Laitin 2003, Sambanis 2004, Hegre and
Sambanis 2006), is sometimes negatively related to frequency of terrorism (Li 2005), sometimes
positively (Findley and Young 2011, Walsh and Piazza 2011) and sometimes the relationship is
2
George W. Bush, “Securing Freedom’s Triumph.” New York Times Op-ed, September 11, 2002.
2
indeterminate (Young and Dugan 2011). Measures of state material capabilities, indicators that
capture actual and latent capacity to wage conventional war, are mostly positively associated
with the frequency of terrorism (Li 2005, Koch and Cranmer 2007, Young and Findley 2011,
Findley, Piazza, and Young 2012). Lai (2007) finds that weaker states (as measured by low
GDP) produce more terrorist attacks that are exported abroad. Taken together, these results are
inchoate and often contradictory, suggesting a more complex understanding of the effect of state
capacity on terrorism is warranted. In this paper, we discuss different dimensions 3 of state
capacity and how they influence the likelihood of terrorism. We also consider how these
dimensions of state capacity influence domestic, transnational, and exported terrorism. We
argue for dividing the concept of state capacity into two dimensions: military capacity, or the
ability to project conventional military force and bureaucratic/administrative capacity, or the
ability to collect and manage information. We hypothesize that these separate dimensions have
countervailing impacts on both the incentives for and opportunity to engage in terrorism.
After developing a consistent set of hypotheses deriving from a multidimensional state
capacity approach, we test these claims using data on domestic and transnational terrorism.
Consistent with our argument, we find that while military capacity is positively associated with
the frequency of terror attacks, bureaucratic/administrative capacity is associated with negatively
associated with the frequency of terror attacks. In the conclusion, we discuss the implications for
policy makers and for the scholarly study of terrorism.
Goertz (2006, 4) states that “good concept[s] draw distinctions that are important in the behavior of an object.”
What he refers to as the basic level of the concept, or the main definition, is the noun to which we attach adjectives.
State capacity is the noun where we will add modifiers, such as bureaucratic or military. In Goertz (2006) verbiage,
these secondary level dimensions then offer causal mechanisms or how a multidimensional concept relates to other
concepts. Munck and Verkuilen (2002) offer a similar schema but refer to a concept (Goertz’s basic level) and its
attributes (Goertz’s secondary level). We adopt Goertz’s language to explain the multidimensional nature of the
concept of state capacity.
3
3
How State Capacity Influences Political Violence
How the state structures incentives for dissident violence is a critical portion of the story
for why rebellion, and political violence more generally, occurs (Skocpol 1979, Gurr 1988,
Goodwin 2001).
Given this long tradition of state-centered analyses of political violence
(Skocpol 1979, Evans et al. 1985), relatively less attention has been paid to how states help
structure incentives for oppositional terror.4 Some insights from the civil war literature may be
useful. Within the civil war literature, state capacity is a key component of political opportunity
structure that affects potential rebels’ decisions to fight (Tilly 1978, Hendrix 2010, Sobek 2010).
The decision to rebel takes into account the government’s capacity to repress and to
accommodate. States with considerable repressive capacity can impose more significant costs on
potential dissidents and thus can deter rebellion. States capable of accommodating grievances via
redistribution, the granting of autonomy rights, or the incorporation of dissident movements
within the party system will be more successful at placating restive groups and less likely to face
armed dissent.
State capacity, however, is a multifaceted concept (Hendrix 2010, Kocher 2010). In
general, military capacity – operationalized as military personnel per capita – is associated with a
lower likelihood of onset, higher likelihood of war termination, and shorter war duration (Mason
and Fett 1996, Balch-Lindsay and Enterline 2000, DeRouen and Sobek 2004, Buhaug 2010).
Likewise, states with greater bureaucratic/administrative capacity – operationalized alternately
by level of economic development (Fearon and Laitin 2003), survey measures of bureaucratic
quality and expropriation risk (DeRouen and Sobek 2004, Fearon 2005) and indicators of natural
resource dependence or revenue-generating capacity (Collier and Hoeffler 2004, Humphreys
4
The one area that has received extensive attention is the study of regime type and terrorism (e.g. Eubank and
Weinberg 1994, Sandler 1995, Weinberg and Eubank 1998, Li 2005, Young and Dugan 2011, Aksoy, Carter and
Wright 2012).
4
2005, Thies 2010, Buhaug 2010, Besley and Persson 2010) – are less likely to experience
conflict onset and more likely to endure shorter conflicts. Bureaucratic and military capacities
tend to be positively correlated, in part due to the interrelationship between warfare and the
development of hierarchical, bureaucratically organized state institutions (Tilly 1975). Some
states, however, intermingle low levels of bureaucratic capacity with high levels of military
capacity (present-day Russia, Egypt) and comparatively high levels of bureaucratic capacity with
low levels of military capacity (Costa Rica, Namibia). In short, these two dimensions of capacity
can diverge. Moreover, the two may have differential effects on civil conflict onset. When
analyzed together, Hendrix (2011) finds that bureaucratic capacity – operationalized as tax
capacity – is negatively associated with conflict onset while larger militaries and more military
spending are positively related to civil war onset.
How the state structures incentives for violence or non-institutional political participation
is part of a larger political opportunity model for behavior (e.g. Eisinger 1973, Tilly 1978,
Tarrow 1994, McAdam 1999). The political opportunity model of violent mobilization contends
that the decision to mobilize, as well as the subsequent choices of targets and tactics, takes into
account the state’s ability to repress and accommodate challenges (Tilly 1978). Thus, state
capacity is central. Mann (1984) claims that there are two dimensions of state power: despotic
power and infrastructural power. The first is analogous to what previous state theorists might
term autonomy from civil society. In Mann’s (1984, 113) words, despotic power is “the range of
actions which the elite is empowered to undertake without routine, institutionalized negotiation
with civil society groups.” In contrast, Mann (1984, 113) suggests infrastructural power is the
capacity of the state “to implement…political decisions throughout the realm.” The difference
between these two is illustrated by exploring some cases. For example, the advanced capitalist
5
democracies have strong powers to tax and thus have infrastructural capacity, but they cannot
alter fundamental societal rules and thus do not have despotic capacity. Mann’s (1984) two-part
distinction serves as an example for a division of state capacity, but there are many others (see
Hendrix 2010). Most interpretations of capacity assume that these dimensions are either fused or
privilege one dimension over the other. We make some amendments to Mann (1984) as despotic
power is a narrower concept than more broad notion many scholars term repressive or military
capacity (Goodwin 2001, Hendrix 2010). Similar to Mann’s infrastructural power, bureaucratic
power suggests the state has a professional bureaucracy that can see its population (Scott 1998,
Hendrix 2010). This can deter or mitigate violence both by channeling dissent as well as
providing the state the ability to organize a coherent response to dissent. Moreover, more
bureaucratically capable states are better able to credibly negotiate with dissidents because they
can actually follow through on their commitments (McBride, Milante and Skaperdas 2011) and
thus address one of the common causes of conflict between state and non-state actors (See Lake
2003).
Some states seek to deter violent challenges via investments in repressive capacity; others
provide institutionalized channels for the expression of grievances, devolve political authority to
restive minorities, or incorporate dissident movements into the party system in order to diminish
the relative expected gains from violent tactics.
Like Goodwin (2001) and other political
opportunity theorists, we focus on how the state structures incentives for certain forms of
violence. Our focus here is on state repressive capacity and its incentive effects for engaging in
terrorism. Terrorism is the threat or the use of violence against noncombatants to influence an
audience for political purposes (For detailed discussions of how to define terrorism see:
6
Weinberg et al. 2004, Jongman and Schmid 2005, Hoffman 2006). 5 As policymakers and
scholars have suggested, the conventional wisdom is that terrorism is tactic of a weak opponent
facing a stronger state.
Assuming that a dissident group has decided to use violence,6 its choice of tactics can
range from open armed rebellion to low-level, hit-and-run insurgency and terrorist tactics such as
bombings, assassinations, hijackings, and kidnappings. These tactical considerations are
informed by dissident beliefs about the viability of these different tactics given the repressive
capacity of the state. As Crenshaw (1982) notes,
The attractiveness of terrorism to insurgents who lacks means is the reason it is
often called the ‘weapon of the weak’ and many strategic models of insurrection
situate it as the first phase in the conflict, followed respectively by guerrilla and
then conventional warfare as the insurgents grow stronger (387).
Terrorism is a relatively more attractive alternative for dissidents facing more militarily capable
states. It allows dissidents to avoid direct, costly strikes on government forces, who are typically
superior in numbers and weaponry.7 Dissidents can partially substitute soft targets (undefended
civilian populations, infrastructure, transportation and commercial hubs) for hard targets
(military and police installations, government buildings) where engagement by state forces
would be more likely (Berman and Laitin 2006). If terrorism is a tactical response to
preponderant repressive capacity on the part of the state, then states with more repressive
capacity should experience more terrorist attacks – this is the standard weapon of the weak
hypothesis (See Lake 2002).
5
We are examining oppositional terror in this study. State terror is both real and understudied (Stohl 1983).
Unfortunately, there is insufficient cross-national events data to examine the relationship between state terror and
state capacity.
6
For a recent study of the use of nonviolence as a dissident strategy, see Chenoweth and Stephan (2011).
7
Against guerrilla fighters, the typical advantages of large, mechanized fighting forces are diminished due to
guerrilla’s avoidance of direct assault and the tactical environments, such as mountains or closed terrain, in which
guerrillas prefer to operate (Hendrix 2011). Lyall and Wilson III (2009) find that mechanized armies are less likely
to defeat insurgencies.
7
By contrast, bureaucratic/administrative capacity, or the state’s ability to collect and
manage information on potential dissidents may deter or repress these activities via nonmilitarized means (Goodwin 2001, Hendrix 2010). While military capacity may incentivize
terror tactics because it makes other means of violent contestation relatively more costly,
bureaucratic/administrative capacity should diminish the use of terror tactics by hampering the
ability of terrorist groups to mobilize and conduct attacks. More bureaucratically capable states
will be better at identifying and tracking dissidents and more likely to interdict attacks. This
multidimensional conceptualization of state capacity suggests two competing effects that state
capacity should have on terrorism. At the secondary level, state capacity as divided into military
and bureaucratic/administrative capacity will have opposing effects. This leads to the following
hypotheses:
H1: Indicators of military capacity will be positively related to the number of terrorist attacks.
H2: Indicators of bureaucratic/administrative capacity will be negatively related to the number
of terrorist attacks.
These hypotheses are general and do not distinguish between types of terrorism (domestic vs.
transnational, suicide vs. non-suicide). Work by Abadie (2004) and Young and Dugan (2011)
suggests that domestic vs. transnational terrorism may have different logics of violence. We will
return to this issue in the conclusion. Additionally, countries that export terrorism may also have
a more complicated relationship between state capacity and terrorism. In the next section, we
8
discuss how to test these hypotheses derived from a multidimensional conceptualization of state
capacity.
Research Design
To test the hypotheses derived from a multidimensional conceptualization of state
capacity, we create a time series cross-national dataset that brings together measures of state
capacity and terrorist attacks. The temporal domain of our study is 1984-2007. Our critical
measures of state capacity with sufficient cross-national coverage begin in the mid 1980s thus
limiting the study to this period. Since this period includes some pre-Cold War observations, the
rise of Islamic extremist terrorism, and the decline of the Marxist wave of terrorism (Rapaport
1984), we are more confident that our inferences are not necessarily confined to a specific time
period and/or type of terrorist organization.
Dependent Variable
The annual count of terror attacks committed by dissident groups originating in a
particular country is from Enders, Sandler and Gaibulloev (2011), who decompose the Global
Terrorism Database into transnational and domestic terror events. See LaFree and Dugan (2007)
for a thorough description of the data. As a robustness check, we estimate models using the
annual count of transnational terror attacks targeting a country. These data are from the
ITERATE project (Mickolus et al. 2003). Because ITERATE omits cases of homegrown terror,
the GTD and ITERATE counts of events are only moderately positively correlated (r = 0.43). As
our theoretical model does not posit differential effects of state capacity for domestic and
transnational terror, as do arguments revolving around domestic political institutions (Young and
9
Dugan 2011), we expect the hypothesized relationships to hold across specifications of the
dependent variable.
Independent Variables
Our central independent variables are measures of two aspects of state capacity:
bureaucratic/administrative capacity and military capacity. Both have been measured in various
ways: Hendrix (2010) identifies ten different operationalizations of bureaucratic/administrative
and military capacity commonly used in the civil conflict literature, while Van de Walle (2005)
identifies five measures of bureaucratic/administrative capacity used in the public administration
literature. However, many of these operationalizations are characterized by weak construct
validity and/or not being available for a large sample of countries.
We construct bureaucratic/administrative capacity from a factor analysis of two
variables: the International Country Risk Guide (ICRG) Bureaucracy Quality and Law and
Order variables (Copling, O’Leary and Sealy 2007). Bureaucracy Quality is measured on a 0–4
scale and reflects expert assessments of the degree to which the country’s bureaucracy is
characterized by (1) regular, meritocratic recruitment and advancement processes, (2) insulation
from political pressure, and (3) the ability to provide services during government changes
(Knack, 2001). Law and Order is an assessment of (1) the strength and impartiality of the
judicial system, and (2) popular observance of the law. The two variables are positively and
highly correlated. The resulting variable, bureaucratic/administrative capacity, has a mean of
zero and a standard deviation of 0.82. Our second measure is relative political reach (RPR)
(Tammen and Kugler, ed. 2012). RPR is the ratio of actual participation in the formal economy –
the economy that is taxed and directly supported by public infrastructure – to expected
10
participation in the formal economy, estimated as a linear function of the structure, size, and
degree of social spending in the national economy.8 Given similar structural features, states in
which a larger proportion of the population participates in the formal economy will have a
greater capacity to monitor and collect information on societal actors. As Arbetman-Rabinowitz
et al. (2011) note, “Reach establishes the degree to which the government influences and
penetrates into the daily lives of individuals” (2). Thus, the measure is conceptually close to the
ability of the state to see its population (Scott 1998). The variable has a mean of one and a
standard deviation of 0.29.
As Hendrix (2010) notes, some caution is warranted when using expert measures of
governance quality to predict political violence. The ICRG measures, which are solicited to
generate predictive models of governance crises and the security of foreign investments, will
likely be highly sensitive to information about violent unrest, such as terrorist attacks. In the case
of Rule of Law, this may be definitional. For these reasons, it will be important to model past
political violence explicitly in order to mitigate concerns about endogeneity.
To operationalize military capacity, we use both the Correlates of War (COW)
Composite Index of National Capability (CINC) and a factor analysis-based composite indicator
of military capability. CINC the most widely used measure of a state’s material capacity to wage
war. The composite indicator reflects both actual military capacity (military spending, military
personnel) and fungible economic and demographic capacity (iron production, energy
consumption, and population), and is expressed as the share of power held by country i in year j
RPR is based on the calculation of activity rate as follows: Activity Rate/Population = α + β1+(time) + β2
(bureaucracy) + β3 (Education) + β4 (Pop Age) + β5 (Social Security) + β6 (Urbanization) + β7 (GDPpercap)
+β8(Unemployment)+ε (Arbetman-Rabinowicz et al. 2011). RPR is the actual value for activity rate/predicted value
generated by this equation.
8
11
(Singer 1987). The value ranges from zero to one; we rescale it to range from 0-100 and log it
due to the presence of outliers.
Despite its widespread use, CINC may not be an ideal operationalization of existing
military capacity, which we argue influences the tactical choice to use terror. By including
demographic and economic variables, CINC establishes an upper bound of a state’s military
capacity assuming total mobilization for war, and not the degree of extant mobilization, which
would be the more relevant factor affecting dissident group behavior. 9 Alternately, we use
military capacity, which is the result of factor analysis conducted on CINC’s measures of
military spending and military personnel. The resulting variable, military capacity, has a mean of
0.05 and a standard deviation of 0.31.
Figure 1 plots the scores for bureaucratic/administrative capacity and military capacity
for all country-years, 1984-2007. The two measures are positively correlated as the trend line
indicates. As all four quadrants are represented, however, there are country-year observations
that represent the four possible combinations (high-high, high-low, low-high, low-low). Some
states are highly capable both ways (the United Kingdom, USA), while others lack both
bureaucratic capacity and the ability to project military force (Nicaragua, Mozambique). The offdiagonal cases are perhaps more theoretically interesting: Costa Rica and Namibia have
relatively capable bureaucracies and minimal or nonexistent militaries; Russia, Iran, and Egypt
combine relatively weak bureaucratic capacity with strong conventional militaries.
[INSERT FIGURE 1 HERE]
9
Hence the inclusion of population, urban population, and steel/iron production, which reflect the types of fungible
assets that were/are key to military power in the age of mechanized warfare. Another limitation of CINC is that it is
a proportion of the state’s power in the international system. Since we are more interested in the military ability of a
state vs. dissidents rather than other states, it is not the optimal measure.
12
Table 1 provides some prima facie evidence for the relationships hypothesized in the
previous section. The low-high cutoff is zero (≈ mean) for both axes. As hypothesized, countryyears with higher military capacity see more terror attacks (country-year mean of 22.4) than
those with lower military capacity (6.1, t-test significant at p < 0.001). However, country-years
with higher bureaucratic effectiveness see fewer attacks (14.2) than those with lower
bureaucratic capacity (25.0, t-test significant at p < 0.001). The lowest average number of attacks
is found in country-years that combine high bureaucratic capacity with low military capacity; the
highest are found in countries with strong conventional militaries but weak bureaucracies.
[TABLE 1 HERE]
However, these patterns could be spurious. For instance, more populous countries tend to
have larger armed forces, and thus population, rather than military capacity, could explain this
relationship. In order to eliminate alternative explanations for these observed relationships, we
conduct more rigorous statistical modeling. Since our dependent variable is an overdispersed
count of events, we estimate negative binomial models with a full set of control variables.10
Control Variables
We include a battery of control variables in order to militate against the possibility that
the relationships between the two components of repressive capacity and terror events are
10
Because there may be systematic underreporting of terrorist activity in highly authoritarian regimes, we also
estimate zero-inflated negative binomial models with a measure of democracy (the Polity2 scale) in the logistic
regression stage that corrects for over-reporting of zeroes (Drakos and Gofas 2006, Findley and Young 2012).
Vuong tests confirm that the zero-inflated models are a better fit for the models (Vuong 1989). The findings,
reported in Table A1 in the appendix, are similar to those reported here.
13
spurious. Because present evaluations of bureaucratic/administrative capacity may be sensitive to
past levels of terror activity, we include the panel average of terror attacks over the panel
duration. 11 We control also for log GDP per capita. Controlling for level of development is
standard in conflict studies, though there is disagreement about what the variable proxies. For
some, log GDP per capita is a measure of the state’s bureaucratic and military capacity (Fearon
and Laitin 2003); for others, it proxies economic grievances and/or opportunity cost to
participating in violence (Collier and Hoeffler 2002, 2004, Miguel and Blattman 2010). As we
model state capacity more directly, we interpret any significant finding on GDP per capita as
evidence of an income-grievance effect. We control also for log population. Controlling for
population allows us to mitigate the possibility that our measures of repressive capacity simply
capture variation in the population size of states (more people provide more opportunities to
repress).
We control also for the incidence of domestic armed conflict. One criticism of the GTD is
that it captures a diverse set of actors and targets, including acts such as bombings against
civilians in Israel, rebel attacks against the government in Colombia, and acts of sabotage against
infrastructure by the African National Congress in South Africa. As such, it is a broad measure
that captures some violence that may be part of ongoing an ongoing insurgency. Our measure,
armed conflict incidence, indicates whether there was an active intrastate war in the previous
year; data are from Gleditsch et al. (2002). The inclusion of this control also mitigates concerns
about the endogeneity of military capacity to expectations about facing armed resistance, i.e.,
states that expect to fight insurgents/dissidents will invest more in military capacity. The armed
conflict incidence variable controls for this explicitly.
11
As a robustness check, we estimate all models using the one-year lag of terror attacks, the number of active terror
organizations, and the panel average of active terror organizations. Results did not deviate materially from those
shown here.
14
We include three controls intended to model regime type and durability. The first two
model distinct aspects of democracy: checks and balances and the extent of participation and
fortunes of small parties in the electoral arena. The first, constraints on the executive, comes
from the Polity IV dataset, and captures the degree to which executive authority is constrained by
formal or informal institutional checks (Marshall, Jaggers and Gurr 2011). The second,
Vanhanen Democracy Index, is an equally weighted composite of the proportion of votes won by
smaller parties and the proportion of the population that actually voted in elections. Past research
suggests that these variables should be positively associated with terror attacks (Li 2005, Young
and Dugan 2011). Regime durability is the number of years since the most recent regime change
(a three-point change in the Polity score over three years or less) or the end of a period of regime
instability
(such
as
foreign
occupation)
(Marshall,
Jaggers
and
Gurr
2011).
If
bureaucratic/administrative capacity and military capacity both require the investment of
significant societal resources over extended periods of time, the confounding effect of regime
durability on these factors must be addressed.
We include two variables to address temporal issues with the data. We include a linear
time trend in order to capture any secular trends either in the reporting of terror attacks or actual
prevalence of terror attacks. Lastly, a dummy indicator for the period before/after 1998 is
included in the models run on the GTD outcome variable to account for a systematic change in
the way that GTD reports terror events from the first to second phase of data collection (see GTD
2011).
Results
15
Table 2 reports the results of our base models. Models 1-4 use the GTD outcome
measure; models 5-8 use ITERATE. Models 1 and 5 include our measures of
bureaucratic/administrative capacity and military capacity, models 2 and 6 substitute the log
CINC score as a measure of military capacity. Models 3-4 and 7-8 substitute our alternate
measure of bureaucratic/administrative capacity, Political Reach (RPR).
---------------TABLE 2 HERE
Table 2 provides strong support for H2. Countries with higher bureaucratic/administrative
capacity
experience
fewer
terror
attacks.
The
coefficients
on
both
measures
of
bureaucratic/administrative capacity, B/A capacity and Political Reach, are negative and
significant across all eight specifications, though Political Reach is only significant at the p <
0.10 level in models 3 and 7.
Our findings provide less consistent support for H1: countries with more conventional
military capacity experience more terror attacks. The coefficients on military capacity are
positive and statistically significant under all four specifications, though they are only significant
at the p < 0.10 level in the models including Political Reach. The coefficients on log CINC,
however, are negative and statistically significant in models 4 and 8. Thus, the two measures of
military capacity do not demonstrate a consistent relationship with terrorist attacks. We point out
that on theoretical grounds, our military capacity measure is superior to log CINC as a measure
of actual repressive capacity, which we argue influences decisions about the use of terrorism as a
strategic and tactical doctrine. Log CINC conflates potential military capacity with actual
military capacity and state capacity versus other states.
16
The control variables performed largely as expected. More developed countries with
larger populations that previously experienced attacks and are embroiled in civil conflict
experience more terrorist attacks. Interestingly, once state capacity is modeled explicitly, the
positive relationship between political democracy and terrorist attacks is no longer present. If
anything, greater constraints on the executive are associated with fewer attacks. While no linear
time trend appears in the GTD models, there are systematically fewer terrorist attacks in the
ITERATE data over time.
Figure 2 presents the substantive significance of the most robust independent variables. A
one standard deviation increase in B/A capacity from the mean, equivalent to going from the
Philippines in 2006 to France in 2001, is associated with a 23.5% decrease in the expected count
of GTD attacks and a 20.7% decrease in the expected count of ITERATE attacks. The
substantive effect of a similar shift in Political Reach is roughly half that magnitude. The largest
substantive effects are for military capacity: a similar shift in military capacity (Ethiopia in 2007
to Colombia in 2001) is associated with a 66.2% increase in the expected count of GTD attacks
and a 59.9% increase in the expected count of ITERATE attacks.
Another way of considering these effects is to focus on particular cases. Peru and
Colombia have been among the most terrorism-afflicted countries in the post World War II era.
In addition to their above-average populations and long history of civil conflict, Peru and
Colombia have had comparatively low levels of bureaucratic/administrative capacity (panel
17
averages for B/A capacity of -0.70 and -0.53, respectively) and, especially since 2000, relatively
large militaries (military capacity = 0.18 and 0.36).12
If Peru and Colombia were to have merely average levels of B/A capacity and military
capacity, implying an increase of the former and decrease of the latter, model 1 would predict a
40% and 52% decrease in the expected count of GTD attacks, respectively.
Robustness
As Table 2 indicates, our main findings regarding the state capacity variables are robust
across operationalizations of the dependent variable. To further probe the robustness of our
results, we estimate a series of models varying the inclusion of temporal controls and World
Bank income group indicator variables, as well as additional control variables that have been
shown to be robust predictors of terrorism: physical integrity rights and horizontal inequalities
(Walsh and Piazza 2010, Moore, Bakker and Hill, Jr. 2011). First, we include a measure of
horizontal inequality. Recent research on civil war indicates that horizontal inequalities –
inequalities between ascriptively defined groups, such as between the marginalized Fur and the
dominant Shaigiya, Ja’Alin and Danagla ethnic groups of the Republic of Sudan – are a source
of grievances that spur violent conflict (Østby 2008, Cederman, Weidmann, and Gleditsch 2011).
The measure, lineq2, captures income inequalities across ethnic groups; higher values indicate
larger degrees of inequality. The data are from Cederman, Weidmann, and Gleditsch (2011).
Second, we include a measure of state repression. Respect for physical integrity rights is believed
to suppress terrorist activity, as respecting these rights reduces grievances and makes collecting
12
This is partly a legacy of the historical strength of military institutions in recently established democracies (see
McClintock 1998, 43). In many Latin American states, transitions to democracy required a special acquiescence to
military privileges (Karl 1990).
18
intelligence on extremist groups easier (Walsh and Piazza 2010). We operationalize physical
integrity rights using the CIRI physical integrity rights scale (Cingranelli and Richards 2010).
The results of our robustness checks are generally consistent with the findings reported in
Table 2: while military capacity is associated with more frequent terrorist attacks,
bureaucratic/administrative capacity is associated with less. 13
Some of the models using
ITERATE as the dependent variable miss conventional levels of statistical significance,
especially when controlling for physical integrity rights. The expected directions are always
maintained, but the coefficient size is attenuated. Consistent with our expectations, log CINC is
the least robust indicator. When we estimate the same models using the zero-inflated estimator,
the results are more consistent and robust: bureaucratic/administrative capacity is negatively
associated with the frequency of attacks, while military capacity is positively associated with the
frequency of attacks.14 Given the many number of models estimated, and different variables
included, the results suggest support for the theoretical disaggregation of state capacity and its
competing effects on the number of terrorist attacks.
Conclusions
While conventional wisdom argues that strong states are the targets of terrorist violence,
we find that this relationship is complicated by how we conceptualize state capacity. By fusing
research on the state, civil war, and terrorism, and then examining time-series cross-national
data, we find that military capacity does seem to encourage terrorism as the weapon of the weak
13
The inclusion of physical integrity rights reduces the statistical significance of bureaucratic/administrative
capacity. This is due in part to the two variables being strongly positively correlated (0.58). However, there are
theoretical reasons to believe that the physical integrity measure may mediate some of the effect of
bureaucratic/administrative capacity: previous studies suggest that regimes lacking in bureaucratic/administrative
capacity respond to both terror campaigns and insurgency with more widespread repression. These regimes’
inability to identify and target dissidents directly causes them to resort to more collective forms of punishment,
inciting violence by dissidents (Mason and Krane 1989, Kalyvas 2006, Lyall 2009).
14
Results reported in Appendix A1.
19
hypothesis would suggest. Additionally, we find that bureaucratic/administrative capacity
actually is associated with less terrorism. We are confident in the findings, but there are some
important areas for extensions. First, the specific causal mechanisms are still somewhat opaque.
Bureaucratic/administrative capacity may depress terrorist attacks due to more effective policing,
or due to government capacity to address societal grievances via effective public policies.
Military capacity may encourage the use of terror as a tactic due to force preponderance, but
large, well-funded militaries may be a source of grievances themselves. Military spending is
positively correlated with political corruption (Gupta, de Mello and Sharan 2001), and military
spending may be perceived as a form of patronage politics that saps societal resources from other
uses (Bueno de Mesquita et al. 2003). Adjudicating between these competing causal mechanisms
would constitute an important contribution.
Second, we did not fully explore how state capacity relates to domestic vs. transnational
terrorism or suicide vs. non-suicide terrorism. As Abadie (2004), Young and Findley (2011) and
Findley, Piazza, and Young (2012) argue, there may be a different logic of terrorism between
dissidents inside a country contending with that state than between a group residing in a country
against another external state. Abadie (2004, 2) suggests:
Much of modern-day transnational terrorism seems to generate from grievances against
rich countries. In addition, in some cases terrorist groups may decide to attack property or
nationals of rich countries in order to gain international publicity. As a result,
transnational terrorism may predominantly affect rich countries. The same is not
necessarily true for domestic terrorism.
The proper research design to test these transnational violence processes may be directed dyads
(Young and Findley 2011) or another research design that takes into account multiple actors (see
Gelpi and Avdan 2011).
20
A third extension relates to states that export terrorism.
Lai (2007) provides an
explanation for why states might export terror, and state capacity is a prominent factor. Lai
suggests that terrorist organizations seek environments with low operating costs and use these
states to export terrorism to countries that have higher costs. Former Secretary of Defense,
Robert Gates (2009, 31), argued a similar point:
The recent past vividly demonstrated the consequences of failing to address adequately
the dangers posed by insurgencies and failing states. Terrorist networks can find a
sanctuary within the borders of a weak nation and strength within the chaos of social
breakdown. The most likely catastrophic threats to the U.S. homeland, for example, that
of a U.S. city being poisoned or reduced to rubble by a terrorist attack, are more likely to
emanate from failing states than from aggressor states.
Examining different dimensions of state capacity on this interaction could help unpack this
process further and test claims from both scholars and policymakers.
Finally, we only examine the direct effects that levels of state capacity have on levels of
terrorism. Other indirect relationships may explain how opportunities for terror change. Berrebi
and Ostwald (2011), for example, examine the effects natural disasters have on terrorism. They
find that certain types of disasters increase the frequency of terrorism, but that is effect is
concentrated in countries of low to middle GDP per capita.
Disasters and other similar
exogenous shocks likely influence state capacity and future work should examine whether they
influence each dimension of state capacity in similar or different ways and how this might
change how capacity and terrorism are related.
21
References
Abadie, Alberto. 2006. “Poverty, Political Freedom, and the Roots of Terrorism.” American
Economic Review 96(2): 50-56.
Arbetman-Rabinowitz, Marina, Jacek Kugler, Mark Abdollahian, Kristin Johnson, and
Kyungkook Kang. 2011. Relative Political Capacity: Definitions & Calculation
Procedures.
Typescript.
Accessed
at
http://dvn4.hmdc.harvard.edu/dvn/dv/rpc/faces/study/StudyPage.xhtml?globalId=hdl:1902.1/16845
&studyListingIndex=0_fe472af8e9028fc7d42971b7e46e.
Aksoy, Deniz, David Carter, and Joseph Wright. 2012. “Terrorism in Dictatorships.” Journal of
Politics.
Annan,
Kofi. 2006. “Uniting Against Terrorism: Recommendations for a Global
Counterterrorism Strategy.” Report of the Secretary General, 27 April 2006, 1-32.
Balch-Lindsay, Dylan and Andrew J. Enterline. 2000. “Killing Time: The World Politics of Civil
War Duration, 1820-1992.” International Studies Quarterly 44(4): 615-642.
Berman, Eli, and David D. Laitin. 2006. “Hard Targets: Theory and Evidence on Suicide
Attacks.” NBER Working Paper 11740.
Besley, Timothy, and Torsten Persson. 2010. “State Capacity, Conflict, and Development.”
Econometrica 78(1): 1-34.
Berrebi, Claude and Jordan Ostwald. 2011. “Earthquakes, Hurricanes, and Terrorism: Do Natural
Disasters Incite Terror.” Public Choice 149(3/4): 383-403.
Blattman, Christopher, and Edward Miguel. 2010. “Civil War.” Journal of Economic Literature
48(1): 3-57.
Buhaug, Halvard. 2010. “Dude, Where’s My Conflict? LSG, Relative Strength, and the Location
of Civil War.” Conflict Management and Peace Science 27(2): 107-128.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson, and James D. Morrow. 2003.
The Logic of Political Survival. Cambridge: MIT Press.
Cederman, Lars-Erik, Nils B. Weidmann, and Kristian Skrede Gleditsch. 2011. “Horizontal
Inequalities and Ethnonationalist Civil War: A Global Comparison.” American Political
Science Review 105(3): 478-495.
Cingranelli, David L., and David L. Richards. 2010. “The Cingranelli and Richards (CIRI)
Human Rights Data Project.” Human Rights Quarterly 32(2): 395-418.
22
Collier, Paul, and Anke Hoeffler. 2004. “Greed and Grievance in Civil War.” Oxford Economic
Papers 56: 563–595.
Collier, Paul, and Anke Hoeffler. 2002. “On the Incidence of Civil War in Africa.” Journal of
Conflict Resolution 46(1): 13–28.
Coplin, William, Michael O’Leary, and Tom Sealy. 2007. A Business Guide to Political Risk for
International Decisions. Syracuse, NY: PRS Group.
Crenshaw, Martha. 1981. “The Causes of Terrorism.” Comparative Politics 13(4): 379-399.
DeRouen, Karl R., and David Sobek. 2004. “The Dynamics of Civil War Duration and
Outcome.” Journal of Peace Research 41(3): 303-320.
Drakos, Konstantinos, and Andreas Gofas. 2006. “In Search of the Average Transnational
Terrorist Attack Venue.” Defence and Peace Economics 17(2): 73-93.
Enders, Walter, Todd Sandler, and Khusrav Gaibulloev. 2011. “Domestic versus Transnational
Terrorism: Data, Decomposition, and Dynamics.” Journal of Peace Research 48(3): 319337.
Eubank, William L., and Leonard B. Weinberg. 1994. “Does Democracy Encourage Terrorism.”
Terrorism and Political Violence 6(4): 417-435.
Evans, Peter, Dietrich Reuschemeyer, and Theda Skocpol, eds.1985. Bringing the State Back in
New York: Cambridge University Press.
Fearon, James D., and David D. Laitin. 2003. “Ethnicity, Insurgency, and Civil War.” American
Political Science Review 97(1): 75-90.
Fearon, James D. 2005. “Primary Commodity Exports and Civil War.” Journal of Conflict
Resolution 49(4): 483-507.
Findley, Michael G., James A. Piazza, and Joseph K. Young. 2012. “Games Rivals Play:
Terrorism in International Rivalries.” Journal of Politics 74(1): 235-248.
Findley, Michael G., and Joseph K. Young. 2011. “Terrorism, Democracy, and Credible
Commitments.” International Studies Quarterly 55(2): 357-378.
Gates, Robert M. 2009. “A Balanced Strategy: Reprogramming the Pentagon for a New Age,”
Foreign Affairs (January–February).
Gelpi, Christopher and Nazli Avdan. 2011. “Democracy and the Multilateral Flow of
Transnational Terrorism, 1968-2007.” Working paper.
23
Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and Håvard
Strand. 2002. “Armed Conflict 1946-2001: A New Dataset.” Journal of Peace Research
39(5): 615-637.
Goertz, Gary. 2006. Social Science Concepts: A User’s Guide Princeton: Princeton University
Press.
Goodwin, Jeff. 2001. No Other Way Out: States and Revolutionary Movements, 1945-1991 New
York: Cambridge University Press.
Gupta, Sanjeev, Luiz de Mello and Raju Sharan. 2001. “Corruption and Military Spending.”
European Journal of Political Economy 17(4): 749-777.
Hegre, Håvard, and Nicholas Sambanis. 2006. “Sensitivity Analysis of Empirical Results on
Civil War Onset.” Journal of Conflict Resolution 50(4): 508-535.
Hendrix, Cullen S. 2010. “Measuring State Capacity: Theoretical and Empirical Implications for
the Study of Civil Conflict.” Journal of Peace Research 47(3): 273-285.
Hendrix, Cullen S. 2011. “Head for the Hills? Rough Terrain, State Capacity, and Civil War
Onset.” Civil Wars 13(4): 345-370.
Hoffman, Bruce. 2006. Inside Terrorism. New York: Columbia University Press.
Humphreys, Macartan. 2005. “Natural Resources, Conflict, and Conflict Resolution.” Journal of
Conflict Resolution 49(4): 508-537.
Internal Revenue Service. 2010.
Jongman, Albert and Alex Schmid. 2005. Political Terrorism: a New Guide to Actors, Authors,
Concepts, Data bases, Theories, & Literature Transaction Publishers.
Kalyvas, Stathis M. 2006. The Logic of Violence in Civil War. New York: Cambridge University
Press.
Karl, Terry Lynn. 1990. “Dilemmas of Democratization in Latin America.” Comparative Politics
23(1): 1-21
Knack, Stephen. 2001. “Aid Dependence and the Quality of Governance: A Cross Country
Empirical Analysis.” Southern Economic Journal 68(4): 310–329.
Koch, Michael, and Skyler Cranmer. 2007. “Testing the ‘Dick Cheney’ Hypothesis: Do
Governments of the Left Attract More Terrorism than Governments of the Right?”
Conflict Management and Peace Science 24(4): 311-326.
24
Kocher, Matthew Adam. 2010. “State Capacity as a Conceptual Variable.” Yale Journal of
International Affairs 5(2): 137-145.
LaFree, Gary, and Laura Dugan. 2007. “Introducing the Global Terrorism Database.” Terrorism
and Political Violence 19(2): 181-204.
Lake, David. 2002. “Rational Extremism: Understanding Terrorism in the Twenty-first Century.”
Dialogue-IO 1(1): 15-29.
Lake, David. 2003. “International Relations Theory and Internal Conflict: Insights from the
Interstices,” International Studies Review 5(4): 81-89.
Li, Quan. 2005. “Does Democracy Promote or Reduce Transnational Terrorist Incidents?”
Journal of Conflict Resolution 49(2): 278-297.
Lyall, Jason, and Isaiah Wilson III. 2009. “Rage Against the Machines: Explaining Outcomes in
Counterinsurgency Wars.” International Organization 63(1): 67-106.
Lyall, Jason. 2009. “Does Indiscriminate Violence Incite Insurgent Attacks? Evidence from
Chechnya.” Journal of Conflict Resolution 53(3): 331-362.
Mann, Michael. 1984. “The Autonomous Power of the State: Its Origins, Mechanisms, and
Results.” European Journal of Sociology 25(2): 185-213.
Marshall, Monty J., Keith Jaggers, and Ted Robert Gurr. 2011. Polity IV Project: Political
Regime Characteristics and Transitions, 1800–2010. Version p4v2010. College Park,
MD: University of Maryland (http:// www.cidcm.umd.edu/inscr/polity/index.htm).
Mason, T. David. and Dale A. Krane. 1989. “The Political Economy of Death Squads: Toward a
Theory of the Impact of State-Sanctioned Terror.” International Studies Quarterly 33(2):
175-198.
Mason, T. David and Patrick Fett. 1996. “How Civil Wars End: A Rational Choice Approach.”
Journal of Conflict Resolution 40(4): 546-568.
McAdam, Doug. 1982. Political Process and the Development of Black Insurgency, 1930-1970.
Chicago, IL: University of Chicago.
McBride, Michael, Gary Milante, and Stergios Skaperdas. 2011. “Peace and War With
Endogenous Capacity.” Journal of Conflict Resolution 55(3): 447-468.
McClintock, Cynthia. 1998. Revolutionary Movements in Latin America Washington, DC:
United States Institute of Peace Press.
Mickolus, Edward F., Todd Sandler, Jean M. Murdock, and Peter A. Flemming. 2003.
International Terrorism: Attributes of Terrorist Events (ITERATE). Data Codebook.
25
Moore, Will H., Ryan Bakker, and Daniel W. Hill, Jr. 2011. How Much Terror? Dissidents,
Governments, Institutions and the Cross-National Study of Terror Attacks. Available at
SSRN: http://ssrn.com/abstract=1977262.
Munck, Gerardo and Jay Verkuilen. 2002. “Conceptualizing and Measuring Democracy:
Evaluating Alternative Indices.” Comparative Political Studies 35(1): 5-34.
Østby, Gudrun. 2008. “Polarization, Horizontal Inequalities and Violent Civil Conflict.” Journal
of Peace Research 45(2): 143-162.
Rapaport, David C. 2002. “The Four Waves of Revel Terror and September 11.” Anthropoerics
8(1), found online: http://www.anthropoetics.ucla.edu/ap0801/terror.htm
Sambanis, Nicholas. 2004. “What is a Civil War? Conceptual and Empirical Complexities of an
Operational Definition.” Journal of Conflict Resolution 48(6): 814-858.
Sandler, Todd. 1995. “On the Relationship between Democracy and Terrorism.” Terrorism and
Political Violence 7(4): 1-9.
Sobek, David. 2010. “Masters of their Domains: The Role of State Capacity in Civil Wars.”
Journal of Peace Research 47(3): 267-271.
Scott, James. 1998. Seeing Like a State: How Certain Schemes to Improve the Human Condition
Have Failed. New Haven: Yale University Press.
Singer, J. David. 1987. “Reconstructing the Correlates of War Dataset on Material Capabilities
of States, 1816–1985.” International Interactions 14(2): 115-132.
Tammen, Ronald, and Jacek Kugler, eds. 2012. Performance of Nations. Lanham, MD: Rowman
and Littlefield.
Thies, Cameron G. 2010. “Of Rulers, Rebels, and Revenue: State Capacity, Civil War Onset and
Primary Commodities.” Journal of Peace Research 47(3): 321-332.
Tilly, Charles. 1975. The Formation of National States in Western Europe. Princeton, NJ:
Princeton University Press.
Tilly, Charles. 1978. From Mobilization to Revolution. Reading, MA: Addison-Wesley.
Van de Walle, Steven. 2005. “Measuring Bureaucratic Quality in Governance Indicators.” Paper
for the 8th Public Management Research Conference, Los Angeles, CA.
Walsh, James I., and James A. Piazza. 2011. “Why Respecting Physical Integrity Rights Reduces
Terrorism.” Comparative Political Studies 43(5): 551-577.
26
Weinberg, Leonard, and William Eubank. 1998. “Terrorism and Democracy: What recent events
disclose.” Terrorism and Political Violence 10(1): 108-118.
Weinberg, Leonard, Ami Pedahur, and Sivan Hirsch-Hoefler. 2004. “The Challenge of
Conceptualizing Terrorism.” Terrorism and Political Violence 16(4): 777-794.
Young, Joseph K. 2008. “Repression, Dissent, and the Onset of Civil War.” PhD dissertation,
Florida State University, Tallahassee, FL.
Young, Joseph K., and Laura Dugan. 2011. “Veto Players and Terror.” Journal of Peace
Research 48(1): 19-33.
Young, Joseph K., and Michael G. Findley. 2011. “Promise and Pitfalls of Terrorism Research.”
International Studies Review 13(3): 411-431.
27
Table 1: Bureaucratic/administrative capacity, military capacity, and the average number of
terror attacks, 1984-2007
Military Capacity
Low
High
Total
Low
8.5
40.1
25.0
Bureaucratic Effectiveness
High
2.1
17.4
14.2
Total
6.1
22.4
11.0
28
Table 2: Negative binomial estimates of the effects of state capacity on terrorist attacks, 1984-2007
VARIABLES
Past Attacks
log Real GDPpc
B/A Capacity
(1)
GTD
(2)
GTD
(3)
GTD
(4)
GTD
(5)
ITERATE
(6)
ITERATE
(7)
ITERATE
(8)
ITERATE
0.021***
(0.005)
-0.009
(0.140)
-0.324**
(0.143)
0.021***
(0.005)
0.349***
(0.108)
-0.479***
(0.170)
0.021***
(0.004)
0.085
(0.110)
0.022***
(0.005)
0.304***
(0.097)
0.119***
(0.022)
0.164*
(0.095)
-0.280**
(0.123)
0.119***
(0.024)
0.480***
(0.094)
-0.394***
(0.149)
0.125***
(0.024)
0.224**
(0.092)
0.123***
(0.024)
0.416***
(0.081)
-0.574*
(0.301)
0.973*
(0.576)
-0.998***
(0.343)
-0.388*
(0.218)
0.860*
(0.509)
-0.849***
(0.320)
Political Reach
Military Capacity
1.575**
(0.633)
log CINC
log Population
Exec. Constraints
VH Democ. Index
Regime
Durability
Internal Conflict
After 1998
Year
Constant
Observations
0.346***
(0.091)
-0.005
(0.003)
0.018*
(0.009)
-0.008***
(0.003)
1.535***
(0.214)
-1.095***
(0.225)
-0.029
(0.019)
55.074
(38.054)
2,344
-0.353
(0.255)
0.639***
(0.107)
-0.017***
(0.006)
0.013
(0.009)
-0.004
0.428***
(0.083)
-0.006**
(0.002)
0.011
(0.009)
-0.012***
1.456***
(0.464)
-0.589**
(0.286)
0.704***
(0.097)
-0.011***
(0.003)
0.004
(0.009)
-0.009***
0.130
(0.080)
-0.007*
(0.004)
-0.000
(0.006)
-0.003
-0.287
(0.241)
0.411***
(0.113)
-0.019***
(0.006)
-0.004
(0.007)
0.000
(0.003)
(0.003)
(0.003)
(0.002)
(0.002)
1.545***
1.644***
1.610***
0.814***
0.735***
(0.215)
(0.206)
(0.203)
(0.176)
(0.197)
-1.058***
-1.116***
-1.156***
(0.249)
(0.202)
(0.211)
-0.022
-0.021
-0.014
-0.084***
-0.078***
(0.019)
(0.017)
(0.017)
(0.010)
(0.010)
35.926
39.868
22.039
165.205*** 147.569***
(38.266)
(34.062)
(33.362)
(20.605)
(20.733)
2,421
2,877
2,990
2,564
2,665
Robust standard errors, clustered on countries, in parentheses
*** p<0.01, ** p<0.05, * p<0.1
0.194**
(0.086)
-0.008***
(0.003)
-0.002
(0.006)
-0.005***
-0.487**
(0.246)
0.461***
(0.111)
-0.015***
(0.004)
-0.005
(0.007)
-0.003*
(0.002)
0.836***
(0.166)
(0.002)
0.753***
(0.167)
-0.077***
(0.008)
149.598***
(16.752)
3,252
-0.077***
(0.008)
145.528***
(15.689)
3,437
29
Table 3: Robustness checks of the effects of state capacity on terror attacks with different estimators and controls
Dependent
Variable
GTD
ITERATE
GTD
ITERATE
GTD
ITERATE
GTD
ITERATE
Additional
Controls
No temporal
controls
No temporal
controls
Horizontal
Inequality
Horizontal
Inequality
WB Income
Group
dummies
WB Income
Group
dummies
Physical
Integrity
Rights
Physical
Integrity
Rights
B/A Capacity
Coefficient
w/Military
Capacity
(Standard
Error)
RPR
Coefficient
w/Military
Capacity
(Standard
Error)
RPR
Coefficient
w/log
CINC
(Standard
Error)
-0.374**
(0.150)
-0.285**
(0.121)
-0.384***
(0.139)
-0.298**
(0.133)
-0.284*
(0.159)
B/A
Capacity
Coefficient
w/log
CINC
(Standard
Error)
-0.447***
(0.151)
-0.349***
(0.132)
--0.557***
(0.168)
-0.444***
(0.158)
-0.469***
(0.178)
Military
Capacity
Coefficient
w/RPR
(Standard
Error)
log CINC
Coefficient
w/B/A
Capacity
(Standard
Error)
log CINC
Coefficient
(Standard
Error)
-0.647*
(0.342)
-0.453
(0.289)
-1.083***
(0.367)
-0.969***
(0.342)
-0.949***
(0.333)
Military
Capacity
Coefficient
w/ B/A
Capacity
(Standard
Error)
1.371**
(0.613)
1.103**
(0.471)
1.632**
(0.673)
1.087**
(0.466)
1.637**
(0.666)
-0.410
(0.364)
-0.109
(0.224)
-0.585*
(0.314)
-0.396*
(0.225)
-0.531*
(0.305)
0.856
(0.605)
0.563
(0.543)
1.126*
(0.625)
0.705
(0.528)
1.452**
(0.648)
-0.157
(0.274)
-0.134
(0.215)
-0.571**
(0.251)
-0.466**
(0.229)
-0.434
(0.267)
-0.421
(0.308)
-0.315
(0.235)
-0.804***
(0.308)
-0.722***
(0.243)
-0.593**
(0.286)
-0.256*
(0.135)
-0.405***
(0.151)
-0.325
(0.234)
-0.827***
(0.295)
1.374***
(0.519)
1.000*
(0.526)
-0.327
(0.241)
-0.582***
(0.213)
-0.206
(0.151)
-0.284*
(0.154)
-0.367
(0.277)
-0.599**
(0.285)
1.351**
(0.635)
1.015*
(0.589)
-0.204
(0.283)
-0.270
(0.304)
-0.230*
(0.128)
-0.201
(0.126)
-0.267
(0.217)
-0.349
(0.230)
1.554***
(0.489)
1.177**
(0.551)
-0.268
(0.233)
-0.290
(0.247)
30
Figure 1: Bureaucratic/Administrative Capacity and Military Capacity, 1984-2007
1
Military Capacity
0.75
-2
R² = 0.2523
0.5
0.25
0
-1.5
-1
-0.5
0
0.5
1
1.5
-0.25
-0.5
-0.75
-1
Bureaucratic/Administrative Capacity
31
Figure 2: Expected Increase in Terrorist Events Based on Changes in
Bureaucratic/Administrative Capacity and Military Capacity Measures
120.00%
100.00%
80.00%
60.00%
40.00%
+1 SD from Mean
25th – 75th Percentile Shift
20.00%
0.00%
-20.00%
-40.00%
The results are based on one standard deviation changes from the mean or changes from the 25 th to the 75th
percentile for the variables of interest. The dependent variable is either from the Global Terrorism Database (GTD)
as adjusted by Enders, Sandler, and Gaibulloev (2011) or from the International Terrorism: Attributes of Terrorist
Events (ITERATE) database. The measures for Bureaucratic/Administrative capacity are the factor score made
from ICRG variables (BA) or the Relative Political Capacity measure (RPR). The Military Capacity measure
(MilCap) is the factor score from military personnel and military spending.
32