Political Predators: Transnational Terrorists Strategic Target Selection as a Mechanism for Political Mobilization

Transcription

1 Political Predators: Transnational Terrorists Strategic Target Selection as a Mechanism for Political Mobilization Graig Klein Binghamton University Working Paper Prepared for International Studies Association Annual Conference New Orleans, Louisiana Abstract: To maintain relevance and make demands, terrorists must recruit. The question we must ask is: how do terrorists gain recruits? In order to subsidize recruitment costs, transnational terrorists seek to motivate violent and indiscriminate counterterrorism responses through strategic and deliberate target selection. When counterterrorism increases grievances or generates vengeance, terrorists have an easier and less costly means of recruiting. In an effort to maximize recruitment benefits, transnational terrorists target states when leaders face pre-existing diversionary conflict incentives. Targeting leaders in need of diversion increases the likelihood of retaliation, thus maximizing recruitment benefits through a combination of attack success and grievances created by the counterterrorism response. Empirical results provide robust evidence that transnational terrorists, from , indeed target OECD countries facing high unemployment rates and upcoming elections in support of a new mechanism for better understanding the disproportionately high rate of terrorists targeting democratic states.

2 Terrorist groups have three primary goals gaining policy concessions, maintaining ideological relevance and recruiting new members (Crenshaw 1981; Kydd and Walter 2006). Without new recruits, the first two goals are unachievable. Recruitment is an intermediate goal because supporters constitute a primary resource for the organization (Carceles-Proveda and Tauman 2011, 50). Recruitment is a political game; governments, dissident group and terror organizations are constantly competing for supporters. One means of recruiting is exploiting grievances generated by counterterrorism retaliation. By offering an opportunity for vengeance for collateral damage from counterterrorism, transnational terrorist cells are capable of attracting new members. Attacks serve two purposes; one is to maintain relevance and credibility to audiences at home and abroad, and the second, is recruitment. If terrorist attacks are planned correctly, they subsidize recruitment costs. Successfully motivating counterterrorism operations that are violent and indiscriminate in nature as such responses impose higher costs and casualties on terrorists and innocent civilians the targeted state s retaliation enables terrorist groups to more effectively recruit. Recruitment is costly; one unit of resource expenditure does not generate one new recruit. Terrorists can expend resources to gain popular support, but must retain sufficient resources to perpetuate attacks (Enders and Sandler 2005). Without attacks, ideological relevance may wane and bargaining power and policy concessions are impossible, therefore, terrorists are faced with a puzzle. They must recruit, but not expend too many resources, which will sacrifice their ability to attack. Understanding the impact of retaliation strategies on terrorists ability to increase popular support and gain new recruits is of elemental importance. If a regime inadvertently rewards terrorism in any way, the benefits for terrorists increase (Sederberg 1995). Terrorists face a political mobilization problem and use collateral damage caused by counterterrorism operations to solve that problem. This forces us to consider attacks in a new light. Attacks are deliberate, strategic decisions by terrorist leadership aimed at motivating violent, and preferably, indiscriminate retaliation by the targeted state. To subsidize recruitment costs, attacks must be effective in targeting the correct government that is susceptible to engaging in violent, and from the terrorists perspective, over zealous, retaliation. 1 Klein, Working Paper - ISA Conference 2015

3 Such a perspective of transnational terrorist attacks begs a new question: What makes a country, from a terrorist leader s perspective, a profitable target? Scholarship evidences that democracies disproportionally suffer from terrorism compared to autocracies (see Ross 1993; Eubank and Weinberg 2001; Li and Schaub 2004; Li 2005; Chenoweth 2010). But not all democracies suffer from terrorist attacks at a similar rate; there is clearly something beyond being democratic that motivates targeting. It is the transparency of political debate and public opinion in democracies that results in both the overall increased rate of terrorism in democracies and the variation in attack frequency and size amongst democracies. Democratic transparency encourages elected leaders to engage in aggressive foreign policy, or political uses of force, in an attempt to divert the public s attention toward a more favorable source of evaluation when confronted with negative domestic economic and political conditions. But, diversion is only possible if appropriate targets exist. Because of the inherent transparency of democracy, potential targets know when the likelihood of being targeted increases and strategically alter behaviors to avoid being the target of aggression (see Fordham 2005). Leaders of foreign countries are not the only observers of the daunting domestic conditions in democracies; terrorist leaders are also attentive. What makes terrorist leaders drastically different is that, rather than averting conflict with leaders in need of aggression, terrorists deliberately engage in attacks against democracies where they perceive leaders are desperate for diversion. Terrorist leadership adds an additional incentive to the political use of force calculus by attacking countries where leaders are already searching for diversionary conflict. By targeting such states, terrorists are maximizing the possibility of retaliation, which in turn creates or increases grievances and vengeance within the population terrorists recruit from. If successful, counterterrorism operations subsidize recruitment costs, if not, the attack, at least, maintains the terrorist group s relevance. In the remainder of the article, I examine why violent indiscriminate counterterrorism benefits transnational terrorists, how it assists their recruitment efforts and propose a new argument underlying transnational terrorists target selection. After articulating a theory of conflict seeking behavior, I empirically test the resulting hypotheses on transnational terrorism activities in OECD countries from In 2 Klein, Working Paper - ISA Conference 2015

4 support of conflict seeking behaviors and terrorists attempts to maximize counterterrorism benefits, transnational terrorists indeed strategically select targets to attack. Why Collateral Damage Matters Irregular conflicts, i.e. counterterrorism, implicate civilians in a direct way; civilian zones are integrated into the battlefield (Kalyvas and Kocher 2007). As the lines between conflict and civilian zones become increasingly blurred, noncombatants find it difficult to ensure their safety. Terrorists demonstrate their desire for civilian casualties by intentionally blurring the boundaries between conflict and civilian zones. For example, by hiding in urban areas, terrorists make collateral damage from retaliation nearly unavoidable (Sandler 2005). When it is difficult for the retaliating state to isolate terrorists or training grounds outside of populated areas, indiscriminate violence becomes increasingly inevitable. The inability of noncombatants to avoid victimization alters the cost-benefit analysis of joining the terrorists (Kalyvas and Kocher 2007). An indiscriminately repressive response is likely to backfire because non-elites can no longer secure their own safety (Francisco 1995, 1996) so the costs of joining the opposition are drastically reduced causing active support for the opposition to increase (Mason and Krane 1989). A goal of counterterrorism is to punish terrorists and discourage active support, but indiscriminate retaliation motivates physically insecure and/or vengeful individuals to become active supporters (Gurr 1970; Lichbach 1987; Ondetti 2006). Indiscriminate counterterrorism also influences active supporters to become more determined adherents when violence appears arbitrary and brutal (Mason and Krane 1989). Counterterrorism policies that do not properly gauge the impact of collateral damage create unintended consequences (Sandler and Enders 2004; Rosendorff and Sandler 2004) such as increasing sympathy for the terrorists cause, increasing recruitment of previously nonsympathetic civilians and making active supporters more determined to strike back. Civilians are also driven to join terrorists in an attempt to seek revenge for collateral damage from counterterrorism. Revenge is a strong motivational force and 3 Klein, Working Paper - ISA Conference 2015

5 drives passive bystanders to become recruits and active supporters (Altran 2003; Pape 2003; Jacobson and Kaplan 2007). Terrorist groups can harness the desire for vengeance by offering opportunities to strike back through future attacks. Additionally, the perception of the retaliator as evil becomes increasingly easy for the terrorists to narrate as civilian casualties mount. Counterterrorism policy decisions must consider terrorists retaliation preferences. Targeted governments typically respond in order to maintain legitimacy, protect its citizens and/or avoid a loss of popular support, but if the targeted government responds too harshly it encourages nonparticipants to join the terrorist group (Wilkinson 2001). Violent retaliation for a terror attack is never a guaranteed response because counterterrorism provides a collective good by diminishing terrorists assets and reducing the threat to all states; states have an incentive to free-ride on others counterterrorism (Lee 1988; Sandler 2003; Arce and Sandler 2005). Retaliation may deflect future targeting (Lee 1988), but, typically, countries that engage in counterterrorism absorb the majority of terrorist targeting (Arce and Sandler 2005). A state that does not violently retaliate limits terrorists recruitment payoffs. Therefore, terrorists must find a way to motivate violent counterterrorism. Terrorists deliberate and strategic target selection and the scale of an attack are mechanisms to motivate violent retaliation from the targeted state. Terror groups have incentives to expend resources strategically to maximize the probability of provoking a response (Enders and Sandler 2006). For example, large death tolls from terrorist attacks are more likely to provoke harsher government responses (Quillen 2002). Terrorists cannot control the level of recruitment provided from a target s retaliation, but terrorists can maximize the probability of higher recruitment by strategically attacking appropriate targets. The recruitment benefits, i.e. increased grievances and vengeance resulting from counterterrorism, underlie the importance for terrorists to successfully select targets; the necessity of success induces the strategic conflict seeking behaviors of transnational terrorists. 4 Klein, Working Paper - ISA Conference 2015

6 Strategic Conflict Avoidance vs. Strategic Conflict Seeking Theory The crux of strategic conflict avoidance theory is that potential target states behave less confrontationally and more cooperatively when domestic conditions in a rival state provide reason for the rival state to engage in political use of force (Fordham 2005). We should find more conflict when states do not face diversionary incentives since diversion reduces the costs of inducing conflict. For example, rivals prefer to quarrel with the United States when diversion is not a prominent motivation; rivals respond less negatively to the United States policies or hostility when unemployment in the United States is high (Fordham 2005; Clark et al. 2011). Leeds and Davis (1997) provide evidence for this dynamic beyond the United States; states are less likely to make demands against advanced economic countries faced with an underperforming domestic economy. International leaders observe a rival s diversionary incentives and change their conflict behaviors accordingly (Smith 1996, 1998). Essentially, two effects stimulate the lack of observations of conflict or diversionary use of force the simple true lack of political use of force and the dynamic process where strategic conflict avoidance produces no political use of force (Clark 2003). The opportunity to use force may not be randomly generated; the opportunity is shaped by rationality expressed through strategic conflict avoidance behaviors (Leeds and Davis 1997, Clark et al. 2011). At the core of the strategic conflict avoidance theory is that the costs of war have to be considered, this includes the domestic costs for both the diverting leader and the targeted leader and the ability for leaders to observe and assess an opponent s costs. For some non-state actors, the logic of strategic conflict avoidance theory is applicable, for others, it is not. Terrorists benefit from being the target of political uses of force as they use retaliation and government violence to reinforce their anti-regime rhetoric and propaganda; effective counterterrorism must balance the negative externalities for the state and positive externalities for terrorists (see Rosendorff and Sandler 2004; Siqueria and Sandler 2007; Arce and Sandler 2010). In general, democracies tend to be more frequently targeted by terrorist groups (see Li 2005; Chenoweth 2010; Eubank and Weinberg 1994; Pape 2003, 2005; Enders 5 Klein, Working Paper - ISA Conference 2015

7 and Sandler 2000), but the casual mechanism(s) for this statistical correlation remains unclear. Strategic conflict seeking theory details a casual mechanism that enhances our understanding of why democracies are targeted by terrorists more frequently than nondemocracies. Democratic leaders respond to electoral incentives and thus when attacked, the government cannot remain passive, the state has incentives to react. Successful terrorist attacks demonstrate the government s inability to provide safety and security to the population, making leadership vulnerable to voters fears and dissatisfaction. Democratic leaders, aware of the looming guillotine, are forced to respond to terrorist attacks to compensate for the perceived inability to protect the homeland. This response dynamic is the keystone to the strategic conflict seeking theory. Terrorists seek a response from the target; random targeting is not efficient, but strategic selection of democracies where leaders face diversionary incentives is efficient. Terrorism works best when it alters the target s behavior (Tilly 2004) and targeting states where leaders are in need of diversionary conflict is most likely to alter behavior and motivate retaliation. Provoking the targeted government into a counterterrorism response assists terrorists recruitment efforts because the population absorbing counterterrorism is more likely to be radicalized by the violence (Kydd and Walter 2006). Provocation is a key goal of terrorist attacks (Crenshaw 1981). The material success of an attack does not directly enter the objective function of a terrorist s victory because attacks are, in part, used to mobilize potential supporters; objective success is a combination of material damage and recruitment payoffs (Siqueira and Sandler 2006, 884). Recruitment depends on terrorists success, the spectacular or symbolic nature of the attack and the target s retaliatory response (Rosendorff and Sandler 2004). As retaliation helps maximize terrorists expected utility from an attack, states that are more likely to retaliate are more likely to be targeted. Leaders in need of diversion may be overly ambitious when presented with an opportunity to retaliate against terrorists making them ideal targets. Democratic transparency allows transnational terrorists to observe signals of political leaders vulnerability to popular demands. When adverse domestic conditions present in a potential target, transnational terrorists, rather than avoid confrontation, fast 6 Klein, Working Paper - ISA Conference 2015

8 track plans against a politically vulnerable target. I am not suggesting that terrorists begin formulating plans once a potential target attains a pre-determined level of domestic political liability; rather, I suggest that as plans are being considered and developed, terrorists withhold action until presented with the opportunity to maximize their risky and expensive efforts. By observing the domestic political environment in potential targets, transnational terrorists can selectively implement attacks in a strategic manner. The political use of force literature points to three observable conditions high unemployment, high inflation and upcoming federal elections that are likely to motivate aggressive foreign policy posturing by vulnerable democratic leaders. By attacking vulnerable democracies, transnational terrorists strive to purchase diversionary action by adding to the list of incentives for aggressive foreign policy. As Meernik (1994) states, a sagging image may be strengthened by forceful action abroad (130); transnational terrorist contribute to a leader s sagging image by strategically timing attacks to coincide with a downtrodden economy. If the goal of aggressive foreign policy is to build political capital (McAvoy 2008), then it is reasonable to expect leaders to pursue political capital when it is most valuable as elections approach. As the shadow of the future losing an election grows shorter, political leaders are more likely to act inline with diversionary logic (Levy 1989, Ostrom and Job 1986, Russett 1989). Counterterrorism operations are likely to be increasingly aggressive and large when political capital needs to be created and is most valuable. Thus, transnational terrorists time their attacks to coincide with election cycles in order to maximize the probability of large-scale counterterrorism retaliation. More formally: Hypothesis 1: The probability of being the target of a transnational terrorist attack increases as a country s national election approaches. Elected leaders are expected to act in a predictable manner when confronted with a sagging economy (Baker and Oneal 2001; Druckman and Holmes 2004; Trager and Vavreck 2011). By tracking two economic indicators in particular unemployment and inflation rates observers can hypothesize how elected leaders will respond to 7 Klein, Working Paper - ISA Conference 2015

9 confrontation. For example, when industrialized democracies experience a declining economy, elected officials increase aggressive foreign policy rhetoric and engage in confrontation with the goal of avoiding punishment for the economy and gaining reward for foreign policy (Russett 1989). Gambling for resurrection (Downs and Rocke 1994) can become a rational behavior that is exploitable by rivals (transnational terrorists) in search of conflict (aggressive counterterrorism). This leads to two additional hypotheses: Hypothesis 2: The probability of being the target of a transnational terrorist attack increases as a country s unemployment rate increases. Hypothesis 3: The probability of being the target of a transnational terrorist attack increases as a country s inflation rate worsens. Together, strategic conflict seeking theory and these hypotheses, force a reconsideration of what motivates transnational target selection. Explicit causal mechanisms that are specific to democracies contributes to a better understanding of why democratic states are disproportionately targeted by terrorists; it is not because democracies are democracies, it is because the transparency of democratic processes and economic conditions allows transnational terrorists to adequately speculate about when the timing of attacks best maximizes the probability of motivating retaliatory counterterrorism that subsidizes their recruitment costs. The more recruitment costs are subsidized, the more effective a terrorist organization is and, potentially, the longer the organization is capable of lasting. Modeling Strategic Conflict Seeking Behavior: Data and Methodology I focus my analysis on transnational terrorist events. Terrorism is transnational through its victims, targets, institutions, supporters, or terrorists, transnational terrorism involves more than one country (Sandler 2003, 781). My analysis is restricted to OECD 8 Klein, Working Paper - ISA Conference 2015

10 countries from 1990 to The unit of observation (UoO) is country-month; both unemployment and inflation data are available on a monthly basis providing sufficient variation in my key explanatory variables to justify a country-month UoO. Dependent Variable I test the implications of conflict seeking behavior using Enders, Sandler and Gaibulloev s (2011) systematic separation of transnational terrorist attacks from the Global Terrorism Database. 1 The theory points to transnational terrorists strategically selecting targets based on the target s potential need for diversion; thus suggesting that the quantity of interest is not the number of attacks per month, but rather the strict probability of being targeted. As terrorists seek to maximize recruitment, striking a country once can be sufficient if that country responds in the terrorist s desired manner, but if the country does not, it may be deemed a waste of resources to repeatedly attack such a country. Terrorists are not seeking to initiate an ongoing conflict or war, but rather motivate a response that produces positive externalities, i.e. recruitment benefits. Therefore, I transform my dependent variable, Terror Attack, into a binary measure; if a country experiences a transnational terrorist attack in the observation month, Terror Attack assumes a value of 1, if no attack occurs the value is 0. I treat the dependent variable as a binary measurement of terrorists target selection and use a logit regression technique. In employing logit regression analyses, I use two different procedures to improve model specification. First, as presented in Table 3, I use robust standard errors clustered by country code. In Table 4 I use fixed effects by country and robust standard errors. 2 This allows country specific characteristics that are not included as variables in the regression models to shape the probability of a specific country being targeted. Additionally, the assessment of unemployment rates is likely to vary by country as higher rates of unemployment are more routine in some OECD countries than others; using country specific fixed effects helps account for this dynamic. The reference country is the United States. Independent Variables 1 GTD and ITERATE are the two main sources of data on terror events. ITERATE only 2 I exclude Japan from the sample since Japan s military is constitutionally restricted to be a defensive force and thus reduces the direct benefit of targeting only allies can come to Japan s defense and go on the counterterrorism offensive. 9 Klein, Working Paper - ISA Conference 2015

11 Unemployment Rate and Inflation Rate are my key predictor variables. Even for OECD member countries, these are not always available, restricting my analysis to twenty-eight member states. 3 Unemployment Rate is measured as the harmonized monthly rate of all persons and Inflation Rate is measured as changes in the consumer price index from the previous month (stats.oecd.org). Both variables are expected to have a positive effect on the probability of being attacked by transnational terrorists. To test hypothesis 1, I include a measure of the time since the previous election. I operationalize this in two ways, the first is Months Since Executive Election, and the second is Months Since National Election. Both are ascending counts of the number of months since the last election. National elections include legislative elections in countries where the executive is directly elected outside of a parliamentary setting. Routine set elections are not the norm in many OECD countries as they are parliamentary systems; operationalizing the election variables in this way accounts for the fluidity in election timing. While parliamentary systems establish a set time period within which new elections must be held, there is always a chance for snap elections to be called. By accounting for the time since the last election, the election variables mimic the signals and information both elected officials and observers benefit from as the longer it has been since the previous election the more proximate either a set new election or the more likely a snap election will be called. As months since either type of election increases, the probability of being targeted by transnational terrorists is expected to increase. Election dates are taken from the IAEP dataset (Regan and Clark 2005) and supplemented by hand-coding to extend the variable to Control Variables That democratic states disproportionately suffer from terrorism is taken as an article of faith in the terrorism literature. Therefore, I control for the level of Democracy in each country as measured by Polity IV (Marshall and Jaggers 2013); I rescale the original coding to range from 0 to 21. While all countries in OECD are distributed among non-authoritarian values of the Polity IV spectrum, there is variation in the level of democracy amongst OECD countries. 3 See Table 2 for a list of OECD member states included in the analysis. 10 Klein, Working Paper - ISA Conference 2015

12 Freedom of speech and press may result in terrorist attacks being over-reported in democracies compared to non-democracies. To account for this possibility, I include Free Speech using the CIRI coding of Freedom of Speech (Cingranelli and Richards 2013). Similarly to democracy, there is substantial variation in Freedom of Speech; 51.6% of the sample fully respects Freedom of Speech. Lastly, to account for time dependence in terrorists target selection and to control for A.R. 1 type errors, I include a one-month lag of the dependent variable. 4 [Insert Table 1 Here] Results The results suggest that transnational terrorists are conflict seeking and target OECD countries led by elected officials that are prone to engage in diversionary conflict and are thus most susceptible to being motivated to engage in aggressive counterterrorism operations. Hypothesis 2 finds strong support across both the clustered standard errors and fixed effects models. The coefficient associated with Inflation Rate is in the expected direction, but does not achieve statistical significance, thus hypothesis 3 is not supported and the null cannot be rejected. Hypothesis 1 is not supported. Transnational terrorists act strategically and attempt to maximize an attack s recruitment benefits by striking countries with high unemployment rates. Based on Model 2, as Unemployment Rate increases, countries are increasingly likely to be targeted by transnational terrorists. The direction and statistical significance (>.05, p <.05) of the Unemployment Rate clearly illustrates this relationship. On average, a country in the sample has a 25% probability of being targeted by transnational terrorists if the country was targeted in the previous month. In countries with a fully free press, the probability of being targeted increases to 28%, this suggests that terrorists are cognizant of the likelihood of the attack being reported and thus fulfilling advertising, signaling and communication needs. The negative and significant coefficient associated with the level of democracy suggests in support of, amongst others, Eyerman (1998), Li (2005) and 4 This also accounts for unemployment rates reflecting trends across time. 11 Klein, Working Paper - ISA Conference 2015

13 Abadie (2006) that middling democracies are more likely to be targeted than entrenched democracies. [Insert Table 3 Here] Holding variables constant at their mean (months until election and democracy) or mode (free speech), when a country was attacked in the previous month and the Unemployment Rate shifts from the sample s mode (4.6%) to one standard deviation above the mode (8.4%), the predicted probability of being targeted increases to 32%. When the Unemployment Rate reaches one standard deviation above the mean (11.59%), the predicted probability of being targeted by transnational terrorists increases to 36%. If the country did not experience a transnational terrorist attack in the previous month, the base predicted probability of being targeted is 3%, when there is full freedom of speech, the probability increases to 4%, and, when the Unemployment Rate is one standard deviation above the sample s mean or mode, the probability of being targeted is 5%. Figure 1 clearly illustrates this relationship in support of hypothesis 2. [Insert Figure 1 Here] Models in Table 4 include country fixed effects; the United States is omitted from the countries listed as it serves as the comparison country in the models. Even when accounting for country specific characteristics and the variance in routine unemployment statistics, as a country s unemployment rate increases, the probability of being targeted by transnational terrorists increases across all OECD countries in the sample. That the finding holds in direction and significance (>.08, p<.001) across country fixed effects estimates indicates that the relationship between high unemployment and increased probability of transnational terrorist targeting is strong and provides robust support for hypothesis 2. As the comparison country, the United State s probability of being targeted is a function of the effect of each coefficient on the average value of the variable for the United States plus the constant. The coefficients listed for other countries in the sample 12 Klein, Working Paper - ISA Conference 2015

14 indicate differences in the intercept, or baseline probability, of being attacked compared to the United States. Those that are statistically significantly more likely to be targeted include France, Germany, Austria, and Greece; the only country in the sample statistically significantly less likely to be targeted by transnational terrorists is Poland. Using Model 4 s results, if the United States was not attacked in the previous month, the probability of being attacked when Unemployment Rate is at its mean is 4.8%; if Unemployment Rate increases by one standard deviation, the probability of being targeted increases to 5.7%. If attacked in the previous month, the United States probability of being targeted when unemployment is held constant at the mean increases to 21.5% and when Unemployment Rate increases by one standard deviation, the probability increases to 25.4%. The effects of executive and federal elections on transnational terrorist targeting suggest that states which recently held elections are more vulnerable to transnational terrorists than when a new round of elections is in the near future. This opposes the logic of conflict seeking behaviors. Transnational terrorists may be seeking to assess the likelihood of a new leader s counterterrorism response mechanisms or efforts in order to update information and future benefits of targeting the new government. The predicted probabilities suggest that previous targeting is particularly important in the strategic calculations of transnational terrorists. This could be the result of several dynamics that must be untangled; in the following section I return to this and discuss one dynamic in particular. Regardless of whether a country was attacked or not in the previous month, as the unemployment rate increases, a country is more likely to be targeted by transnational terrorists. Discussion & Extensions The results suggest that transnational terrorists observe a leader s incentives for diversionary conflict behavior, but unlike most traditional targets (nation states) of diversionary behavior, transnational terrorists strategically target such leaders rather than alter their conflict behavior to avoid confrontation. Transnational terrorists selectively 13 Klein, Working Paper - ISA Conference 2015

15 attack leaders facing diversionary incentives in an attempt to motivate counterterrorism retaliation to maximize an attack s recruitment benefits. The results indicate that the probability of being targeted is a function of previous targeting. While this could be an econometric association motivated by the likelihood of high unemployment in one month preceding high unemployment in the following month, there is also a possible theoretical implication here that requires further exploration. If a terrorist group successfully motivates a state to engage in a proactive counterterrorism response and gains recruitment benefits, it is reasonable to assume that the targeted state will be attacked again. Therefore, if a leader does in fact satisfy his diversionary needs through counterterrorism retaliation, future attacks become more likely. Countries that have demonstrated a willingness to engage in counterterrorism operations that best subsidize a terrorist group s recruitment costs are most likely to be targeted in the future, especially when the leaders of such countries face domestic diversionary incentives. A majority of terrorist groups have a lifespan of less than one year, but those that last beyond a year typically remain a threat for a decade (Arce and Sandler 2010). If terrorist groups can be defined by such a dichotomy, it begs an important question related to the results presented here. Are terrorist groups that last beyond the one-year lifespan more effective in strategic target selection; are these the terrorists exploiting the dynamics of diversionary incentives to subsidize their recruitment costs and preserve their stockpile of finite resources longer? This is certainly an interesting and imperative extension of the relationship uncovered here. The literature to date assumes that democracies are more prone to terrorist attacks than non-democracies, but the arguments focus on structural or institution components of democracy. Little attention is given to leader incentives. This article suggests that transnational terrorists attempt to exploit diversionary incentives to motivate counterterrorism operations that serve both terrorists and leaders needs maximizing recruitment benefits and diversionary conflict, respectively. The results presented here suggest that this gap in the literature is worth exploring further. Democracies are not targeted more than non-democracies simply because they are democracies. The results point to an underlying dynamic. Democratic transparency provides opportunities to observe adverse domestic conditions and assess a leader s 14 Klein, Working Paper - ISA Conference 2015

16 incentives to engage in diversionary foreign policy. Transnational terrorists take advantage of this in order to maximize the recruitment benefits from counterterrorism operations. By targeting leaders that stand to gain political capital useful for offsetting poor domestic conditions, transnational terrorists maximize the likelihood of motivating aggressive counterterrorism that provides benefit to transnational terrorists by subsidizing future recruitment costs. This is especially important as terrorists have a finite stockpile of resources; counterterrorism operations that assist recruitment efforts reduce terrorists costs potentially extending a terrorist group s lifespan. 15 Klein, Working Paper - ISA Conference 2015

17 Table 1: Descriptive Statistics Variable N Mean Std. Dev. Min Max Terror Attack Unemployment Rate Inflation Rate Democracy Free Speech Months Until Executive Election Months Until Federal Election Terror Attack (t-1) Table 2: Countries in Sample Country Observations Country Observations Australia 263 Japan 263 Austria 228 Luxembourg 263 Belgium 263 Mexico 263 Canada 263 Netherlands 263 Czech Republic 168 Norway 263 Denmark 263 Poland 180 Estonia 168 Portugal 263 Finland 263 Slovak Republic 168 France 263 Slovenia 192 Germany 252 Spain 263 Greece 165 Sweden 263 Hungary 192 Turkey 84 Ireland 263 United Kingdom 263 Italy 263 United States Klein, Working Paper - ISA Conference 2015

18 Table 3: Determinants of Transnational Terrorists Target Selection in OECD Countries, Logit Regressions Model 1 Model 2 Model 3 Model 4 Dependent Variable = β β β β Terror Attack Unemployment Rate.050* (.028).055* (.028).050* (.029).054* (.029) Inflation Rate.073 (.140).089 (.143) Democracy -.170** (.067) -.182** (.068) -.158** (.068) -.173** (.069) Free Speech.325 (.359).294 (.353).345 (.359).319 (.353) Months Until Executive Election -.014*** (.004) -.013*** (.004) Months Until Federal (.004) (.004) Election Terror Attack (t-1) 2.32*** (.357) 2.31*** (.345) 2.36*** (.359) 2.34*** (.347) Constant (.792) (.861) (.787) (.869) N Model 1 Wald chi2 = , Prob. > chi2 = ; Model 2 Wald chi 2 = , Prob. > chi2 = ; Model 3 Wald chi2 = , Prob. > chi2 = ; Model 4 Wald chi2 = , Prob. > chi2 = Robust standard errors in parentheses. Japan is not included in the sample. ***p<.001 **p<.01 *p< Klein, Working Paper - ISA Conference 2015

19 Figure 1: Predicted Probability of a Transnational Terrorist Attack Across Changes in the Unemployment Rate With 95% Confidence Intervals Probability of Targeting Monthly Unemployment Rate 23.2 Not Attacked in Previous Month Attacked in Previous Month 18 Klein, Working Paper - ISA Conference 2015

20 Table 4: Determinants of Transnational Terrorists Target Selection By Country Fixed Effects of OECD Members, Logit Regression Model 5 Model 6 Model 7 Model 8 Dependent Variable = Terror Attack β β β β Unemployment Rate.084** (.031).105*** (.031).084** (.031).105*** (.031) Inflation Rate.141 (.102).157 (.101) Democracy -.182*** (.051) -.196*** (.087) -.190*** (.051) -.206*** (.048) Free Speech.929*** (.168).897*** (.170).949*** (.171).923*** (.173) Months Until Executive Election -.010** (.004) -.010** (.004) Months Until Federal Election -.007* (.004) (.004) Terror Attack (t-1) 1.47*** (.149) 1.48*** (.146) 1.49*** (.148) 1.49*** (.145) Canada (.656) (.657) (.654) (.655) Mexico.645 (.420).744* (.412).525 (.417).636 (.408) United Kingdom.399 (.362).374 (.363).446 (.368).413 (.368) Ireland (.536) (.543) (.533) (.540) Netherlands.525 (.370).548 (.370).576 (.372).591 (.373) Belgium.471 (.407).405 (.408).540 (.408).471 (.409) France.780* (.368).668* (.367).711* (.368).595* (.367) Spain.205 (.468) (.477).263 (.471) 051 (.480) Germany 2.18*** (.359) 2.09*** (.360) 2.25*** (.360) 2.16*** (.361) Poland -2.17* (1.05) -2.38* (1.06) -2.21* (1.05) -2.41** (1.06) Austria.953* (.435).961* (.437).850* (.436).867* (.438) Hungary (.568) (.576) (.568) (.576) Italy.568 (.362).502 (.362).624 (.366).552 (.366) Greece 1.16** (.401) 1.05** (.398) 1.23*** (.402) 1.11** (.398) Sweden (.428) (.429).063 (.430).021 (.432) Norway (.508) (.507) (.510) (.510) Denmark (.502) (.502) (.503) (.503) Turkey.612 (.652).502 (.647).649 (.654).540 (.648) Australia (.415) (.414) Constant (1.09) (1.03) (1.09) (1.03) N Model 5 Wald chi2 = , Prob > chi2 = ; Model 6 Wald chi2= , Prob. > chi2 = ; Model 7 Wald chi2 = , Prob. > chi2 = ; Model 8 Wald chi2 = , Prob. > chi2 = Robust standard errors in parentheses. ***p<.001 **p<.01 *p<.05. Luxembourg, Portugal, Czech Republic, Slovak Republic, Slovenia, Estonia and Finland omitted from table because failure is predicted perfectly [coefficient =0 with no standard error]. 19 Klein, Working Paper - ISA Conference 2015

21 References Altran, Scott Genesis of Suicide Terrorism. Science 299 (5612): Arce, Daniel G. and Todd Sandler Counterterrorism: A Game Theoretic Analysis. Journal of Conflict Resolution 49(2): Arce, Daniel G. and Todd Sandler Terrorist Signaling and the Value of Intelligence. British Journal of Political Science 37: Arce, Daniel G. and Todd Sandler Terrorist Spectaculars: Backlash Attacks and the Focus of Intelligence. Journal of Conflict Resolution 54(2): Carceles-Proveda, Eva and Yair Tauman Analysis of the War Against Transnational Terrorism. Economic Behavior 71: Chenoweth, Erica Democratic Competition and Terrorist Activity. Journal of Politics 72(1): Cingranelli, David L., David L. Richards, and K. Chad Clay The CIRI Human Rights Dataset. Version Clark, David H Can Strategic Interaction Divert Diversionary Behavior? A Model of US Conflict Propensity. Journal of Politics 65(4): Clark, David H., Benjamin O. Fordham and Timothy Nordstrom Preying on the Misfortune of Others: When Do States Exploit Their Opponents Domestic Troubles? Journal of Politics 73(1): Crenshaw, Martha The Causes of Terrorism. Comparative Politics 13 (4): Enders, Walter and Todd Sandler Is Transnational Terrorism Becoming More Threatening? A Time-Series Investigation. Journal of Conflict Resolution 44(3): Enders, Walter and Todd Sandler Transnational Terrorism : Thresholds, Persistence, and Forecasts. Southern Economic Journal 71(3): Enders, Walter and Todd Sandler Distribution of Transnational Terrorism Among Countries by Income Class and Geography After 9/11. International Studies Quarterly 50: Enders, Walter, Todd Sandler and Khusrv Gaibulloev Domestic Versus Transnational Terrorism: Data, Decomposition and Dynamics. Journal of Peace Research 48(3): Eubank, William and Leonard Weinberg Does Democracy Encourage Terrorism? Terrorism and Political Violence 6 (4): Eubank, William and Leonard Weinberg Terrorism and Democracy: Perpetrators and Victims. Terrorism and Political Violence 13: Findley, Michael G. James A. Piazza and Joseph K. Young Games Rivals Play: Terrorism in International Rivalries. Journal of Politics 74(1): Fordham, Banjamin O Strategic Conflict Avoidance and the Diversionary Use of Force. Journal of Politics 67(1): Jacobson, Daniel and Edward H. Kaplan Suicide Bombings and Targeted Killing in (Counter-) Terror Games. The Journal of Conflict Resolution 51 (5): Klein, Working Paper - ISA Conference 2015

22 Kalyvas, Stathis N. and Matthew Adam Kocher How Free; Is Free Riding in Civil Wars?: Violence, Insurgency and the Collective Action Problem. World Politics 59 (2): Kydd, Andrew H. and Barbara F. Walter The Strategies of Terrorism. International Security 31 (1): Leeds, Brett Ashley and David R. Davis Domestic Political Vulnerability and International Disputes. Journal of Conflict Resolution 41(6): Li, Quan Does Democracy Promote or Reduce Transnaitonal Terrorism Incidents? Journal of Conflict Resolution 49 (2): Marshall, Monty G. and Keith Jaggers Polity IV Project [Codebook and data files]. Last retrieved on November 10, Mason, David T. and Dale A. Krane The Political Economy of Death Squads: Toward a Theory of the Impact of State-Sponsored Terror. International Studies Quarterly 33: OECD Statistics: Labor Statistics. (January 13, 2015). Pape, Robert A The Strategic Logic of Suicide Terrorism. American Political Science Review 97 (3): Pape, Robert A Dying to Win. New York, NY: Random House. Regan, Patrick M. and David H. Clark The Institutions and Elections Project. Rosendorff, B. Peter and Todd Sandler Too Much of a Good Thing? The Proactive Response Dilemma. The Journal of Conflict Resolution 48 (5): Sandler, Todd Collective Action and Transnational Terrorism. The World Economy 26 (6): Sandler, Todd Collective Versus Unilateral Responses to Terrorism. Public Choice 124: Sandler, Todd and Daniel G. Arce Terrorism and Game Theory. Simulation and Gaming 34(3): Sandler, Todd and Walter Enders An Economic Perspective on Transnational Terrorism. European Journal of Political Economy 20: Sandler, Todd and Kevin Siqueira Global Terrorism: Deterrence Versus Pre- Emption. Canadian Journal of Economics 39(4): Siqueira, Kevin and Todd Sandler Terrorist Backlash, Terrorism Mitigation and Policy Delegation. Journal of Public Economics 91: Smith, Alastair Diversionary Foreign Policy in Democratic Systems. International Studies Quarterly 40(1): Smith, Alastair International Crises and Domestic Politics. American Political Science Review 92(3): Tilly, Charles Terror, Terrorism, Terrorists. Sociological Theory 22 (1): Wilkinson, Paul Terrorism versus Democracy: The Liberal State Response. New York, NY: Routledge. 21 Klein, Working Paper - ISA Conference 2015