Economic Sanctions and the Dynamics of Terrorist Campaigns

Transcription

1 Economic Sanctions and the Dynamics of Terrorist Campaigns Elena V. McLean Texas A&M University 4348 TAMU Allen Building 2010 College Station, TX Kaisa H. Hinkkainen University of Lincoln Baryford Pool, Lincoln LN67TS United Kingdom Luis De la Calle Centro de Investigación y Docencia Económicas (CIDE) Carretera Mexico-Toluca 3655, Lomas de Santa Fe Mexico City Luis.delacalle@cide.edu Navin A. Bapat The University of North Carolina - Chapel Hill 304 Hamilton Hall Chapel Hill, NC bapat@unc.edu Abstract: Although states rarely use economic s specifically to combat transnational terrorism, potential targets of s often face terrorist campaigns within their territory. States may avoid using s against states with terrorists for fear of weakening target states excessively, thereby indirectly strengthening terrorist groups. However, this argument has not been subjected to rigorous empirical testing. This study presents a theoretical and empirical examination that explores how the imposition of s affects the dynamics of ongoing terrorist campaigns in the targeted state. We argue that, while targets may use the terrorist threat to avoid economic s ex ante, target governments have incentives to either disarm or negotiate with domestic terrorists following imposition. However, when terrorists operate transnationally, s do weaken states ability to engage in effective counterterrorism. We also demonstrate that s directed against states serving as home bases can coerce sponsors or passive hosts to crack down on terrorists.

2 In July 2002, over one hundred states participated in the founding of the International Criminal Court (ICC). The goal of the court was to provide an international tribunal to prosecute individuals that committed crimes against humanity, genocide, or other crimes during war. Although agreeing in principle with the creation of the tribunal, the U.S. refused to ratify the treaty. Both the Clinton and Bush administrations argued that American soldiers serving oversees might be accused of war crimes and subject to arrest. Since there was no credible guarantee that American soldiers would be afforded the same level of protection guaranteed by the U.S. constitution, the U.S. refused to ratify the treaty. However, the U.S. took an additional step. In 2002, the American Congress passed the American Servicemembers Protection Act, which threatened to suspend military aid to any country that was party to the International Criminal Court. In response, numerous states signed agreements barring them from turning over U.S. forces to the ICC. Yet, several states faced punitive s, including many key U.S. allies that were fighting alongside American forces in the war on terror. The U.S. therefore faced a suboptimal outcome: the threat of s was instrumental to preventing states from cooperating with the ICC, but imposition could undermine these states efforts against hostile terrorist groups. As in this case, s are commonly used for reasons unrelated to terrorism. Yet, if s undermine economic relations between states and the international community, it follows that s should harm the counterterrorism efforts of states that are fighting terrorist groups. Although this claim appears logical, we presently have no evidence to indicate if s truly do harm counterterrorism efforts. We further have no indication of how harmful s are relative to other factors affecting states efforts to terminate terrorist campaigns, nor do we know how particular types of s influence terrorists ability to persist and actively pursue their objectives. The empirical question motivating this study is: how do s and associated

3 costs imposed on the target state 1 affect the success of the government s anti-terrorist efforts and, ultimately, on terrorist groups ability to survive and continue their activities? We argue that the effect of s depends both on the type of instrument utilized by the sender 2 and the type of terrorist group a target government is facing. Recent literature suggests that states may have incentives to leverage the threat from terrorists to gain more favorable trade terms or increase levels of foreign aid. Following this argument, we hypothesize that s against states facing domestic terrorists might encourage these states to disarm their groups. This may occur because terrorism becomes less lucrative, or because the loss of gains from trade may render a government weaker in the future, thereby making it more vulnerable to terrorism. In either case, states would have incentives to eliminate terrorists immediately. However, the literature also demonstrates that transnational terrorists tend to be more deadly and motivated by interstate rivalries. We therefore hypothesize that s against states that are fighting transnational terrorists are likely to prolong terrorist campaigns and worsen the problem. Finally, we argue that s against supporters of terrorists can be effective, but that excessively harsh s may encourage greater terrorism. The rest of the paper is organized as follows. After a brief review of the relevant literature that has examined the duration and collapse of terrorist campaigns, as well as the literature exploring effects of economic s on targeted states, we present our theoretical argument that links different types of s with terrorist groups survival prospects. We then conduct empirical analyses of the hypotheses derived from our argument using data from the updated version of the Threat and Imposition of Sanctions (TIES) dataset and from the Jones and Libicki dataset on terrorist campaigns. Finally, we discuss our findings and conclude by highlighting the 1 We refer to ed states as targets. 2 We refer to ing states as senders. 1

4 role that exogenous economic shocks in the form of various s can play in the conflict between governments and terrorist groups, and policy implications of our findings that emphasize unintended consequences of policies. Survival of Terrorist Groups What explains the duration of terrorist campaigns and their collapse? Most studies of terrorism have focused on the occurrence and lethality of terrorist attacks. As a result, the literature on the demise of terrorist organizations is still in a nascent stage. The lack of attention is unfortunate for two reasons. First, there is a significant variation in the duration of these groups. In 1992, David Rapoport reported that nearly 90% of terrorist organizations do not reach their first anniversary. Of the few that reached this milestone, half did not last a decade. Others have noted that Rapoport s numbers are overly pessimistic (Vittori 2009) 3, but the fact still remains that most terrorist organizations fail early on, while some groups such as FARC (Fuerzas Armadas Revolucionarias de Colombia) and the ETA (Euskadi Ta Askatasuna) last for generations. Second, for populations targeted by these groups and policy-makers that seek to address the problem of terrorism, the investigation of factors that increase the probability of terminating a terrorist campaign is just as vital as predicting the level and frequency of terrorist attacks. The main contribution of early studies was to provide a typology for the common ways in which terrorist groups collapsed (Crenshaw 1991; Cronin 2006, 2009a, 2009b, 2012; Gupta 2008; Jones and Libicki 2008; Ross and Gurr 1989; USIP 1999). However, these early theories often did not provide causal mechanisms for failed terrorist campaigns, and the evidence was often qualitative in nature and relied on a select number of cases. The earliest example of comparative research on the decline of terrorist groups is Ross and Gurr (1989). Using lessons drawn from the cases of the Canadian group FLQ (Front de Liberation du Quebec) and the American group The 3 Vittori found that almost half of terrorist campaigns in her sample lasted at least four years. 2

5 Weather Underground, they argue that terrorist groups ability to survive relies on their organizational strength. The study defines two dimensions for organizational strength: political vs. military, and internal versus external. Military strength could be weakened by removal of leaders or target hardening, referred to as preemption (external), or through deterrence, (internal) counterterrorism strategies intended to increase the cost of terrorist membership through punishment. Political strength in turn could be reduced through burnout, (internal), i.e., the loss of passion or commitment amongst the group s members, or backlash, (external), i.e., the loss of widespread public support either in response to the group s tactics or a lack of resonance with the group s message. However, the study concludes that political strength has greater explanatory power because in almost all cases terrorist groups are in a position of military weakness relative to their targets. A more complex explanation of why terrorist groups fail is offered in Crenshaw (1991). The study suggests that no single factor can spell the end for terrorist groups. Instead, terrorist groups fail when three conditions are present in some degree: effective counterterrorism strategies by the government, a deliberate decision by the group to end their activities, and organizational disarray. However, Cronin (2006, 2009a, 2009b, 2012) has pointed out that terrorist groups do not always quit attacks because they have collapsed or failed in some manner. She develops six reasons to explain why terrorist groups quit terrorist attacks: decapitation (capture or assassination of key leaders), internal failure (inability to spread to new generation, splintering, burnout, marginalization by public), victory (achievement of goals), transition to politics (negotiation, formation of political party), repression (government crackdowns destroys group) and transition to other violent methods (criminal networks, insurgency or warfare). However, none of these studies has subjected theoretical arguments to any sort of quantitative testing. The first effort to study patterns of terrorist groups survival through the use of 3

6 quantitative evidence was undertaken by Jones and Libicki (2008). After analyzing 648 terrorist groups that existed between 1968 and 2006, they found that almost all terrorist groups ended through one of the two mechanisms: the capture or killing of key officials by police (43%) or entrance into legitimate political processes (40%). In fact, few terrorist groups ceased their attacks because they achieved their goals (10%) and even fewer were rooted out through military defeat (7%). Moreover, the study suggests that the main difference between the groups that ended through a political exit and those that were defeated is the extent of the groups demands. Groups with limited demands were able to enter the political space, while those with more ambitious goals were not. This first set of quantitative results was followed by studies that seek to control for socioeconomic, political and geographical variable in statistical tests. Recently, a number of such tests were conducted cross-nationally (Blomberg, Engel and Sawyer 2010; Blomberg, Gaibulloev, and Sandler 2011; Gaibulleov and Sandler 2013; Miller 2012; Young and Dugan 2010). 4 The tests include two types of variables to analyze the end of terrorist campaigns: characteristics of the state and the terrorist group. With respect to state characteristics, studies demonstrate that differences in national income, level of violence, regime type, and geography can explain a substantial amount of the variation in the duration of terrorist groups (Blomberg, Engel and Sawyer 2010; Blomberg, Gaibulloev, and Sandler 2011; Gaibulleov and Sandler 2013). To sum up this research, rich, populous, less militarily capable, coastal, democratic, moderately diverse, Middle Eastern states with fewer terrorist groups host more durable terrorist groups. Some of these findings have been challenged: for instance, Young and Dugan (2010) find that democracy has no effect on 4 There have also been formal and mathematical models that seek to predict the survival of terrorist organizations. See Faria and Arce (2012); Feinstein and Kaplan (2011); Gutfraind (2009); Sandler (2013) for examples. 4

7 organizational survival, and that poorer nations are more likely to host more durable terrorist groups. With respect to terrorist groups characteristics, it appears that larger, more diversified (with respect to tactics), religiously oriented and multinational groups are likely to last longer (Blomberg, Engel and Sawyer 2010; Blomberg, Gaibulloev, and Sandler 2011; Gaibulleov and Sandler 2013). In addition, Miller (2012) finds that there is a correlation between factors present at the groups birth and current chance of survival. She shows that groups that were more violent at their outset and experience more rapid ascents in power are more likely to fail. Finally, Blomberg et al. (2010) consider the issue of time dependence and report that it is conditional: there is negative time dependence as younger groups experience a greater hazard rate, but more violent groups experience positive time dependence (i.e., older groups are more likely to survive in the next period). Yet, all these statistical models tend to be first cuts at the data that look for correlations between basic characteristics and group duration and are not theoretically motivated. There are four notable exceptions (Carter 2012; Daxecker and Hess 2013; Phillips 2013; and Vittori 2009). Carter (2012) demonstrates that state sponsorship can lead to a greater chance of failure if sponsors provide a home base since targeted states can pressure or reward the sponsor to give privileged information on the group s activities. Governmental repression is more effective for eliminating terrorist groups in autocracies than democracies, due to the potential of public backlash in democracies (Daxecker and Hess 2013). Using network analysis, Phillips (2013) shows that groups with more ties to other terrorist networks survive longer due to increased resources and more effective attacks, which in turn leads to better recruitment. Finally, groups with more resources are more likely to survive due to their increased capabilities and autonomy (Vittori 2009). While this overview suggests that economic factors, such as access to resources and consequently capabilities of both sides, affect terrorist groups survival, the extant research has not investigated the effect of exogenous economic shocks one form of which is the economic 5

8 . One exception is Bapat and Carter (2013): the study investigates the impact of s (trade, aid and financial s, more specifically) on states counterterrorism efforts. The empirical findings suggest that trade and investment restrictions tend to weaken the government, while aid cuts incentivize governments to fight terrorist groups. Although this study identifies the effect of different types of s imposed against the host state on the government s willingness and ability to defeat terrorist groups, it does not differentiate between domestic and transnational groups. However, s are likely to present governments dealing with transnational groups with a more challenging dilemma: s may provide powerful incentives to governments to eliminate terrorist threats, but transnational groups will remain beyond these governments reach given that their support bases are located abroad. Hence, the same economic punishment may result in different outcomes in terms of terrorist group survival. Therefore, we need to consider the impact of different types of terrorist threats on consequences of economic s for governments ability to end terrorist campaigns. In the next section, we discuss how s shape outcomes of interactions between governments and different types of terrorist groups. How do Sanctions Affect the Fate of Terrorist Campaigns? Sanctions are a coercive policy, which aims to extract concessions from the target government through threatened or actual restrictions on some or all economic exchanges between the sender and target. The coercive power of s lies in their ability to generate significant economic costs to the target country; when such costs (either anticipated or actual) are greater than the cost of policy concessions, the target will acquiesce to the sender s demands. Therefore, s can be viewed as an economic form of cost imposition in international disputes (Morgan and Schwebach 1997:33). Sanctions typically create costs by introducing market imperfections that create barriers to commerce between individuals and companies within the 6

9 sender and the target (e.g., Morgan and Schwebach 1997; McLean and Whang 2010). For example, if a sender passes a law that fines individuals $1,000 per hour if they visit a target country, this may dissuade the sender s individuals from touring the target or entering into contracts with the target s firm. Other s, such as cuts in foreign aid or the imposition of blockades, impose costs more directly using the government s resources. Theoretically, increasing the cost of economic transactions indirectly or reducing inflows of capital directly should weaken the target s economy. The amount of weakening is largely a function of the type of (blockades are likely more costly than asset freezes) and a sender s level of enforcement. By reducing the gains from free commerce between itself and the target, the sender reduces the resource pool available to the target, which in turn may face tradeoffs that affect its population. For example, with fewer resources, a target state might have to choose between repairing roads and policing them. The target s population likely prefers that their government engage in both activities, but the target will need to make a choice if it loses some of its resources. Sanctioners often hope that the population of target states will respond to these costs by pressuring their leaders into compliance with the senders demands. In the absence of a terrorist group, the political pressure caused by the reduction in resources and subsequent tradeoffs in government spending should encourage targets to comply with the sender s demands. However, with a terrorist group, the lack of resources may force targets into tradeoffs that require a reduction in counterterrorism efforts. If it is true that s will reduce the pool of resources for counterterrorism, and senders prefer less terrorism to more, target states may leverage the threat of terrorism to avoid s altogether. Keeping the threat of terrorism alive therefore provides a defense against other states seeking to coerce the target (Bapat 2011). The target may also use the threat of terrorism to engage in a wide variety of other offensive behaviors, such as restricting democratic rights and freedoms, election rigging, or human rights abuses. In all 7

10 of these cases, a domestic threat from terrorism can be somewhat beneficial, particularly if we consider that most terrorists do not generate significant human and economic costs. We therefore see that states may have a reason to keep terrorist groups active to avoid international punishments. However, once an international punishment is applied, the target state may no longer have incentives to protect its terrorists. Therefore, if a sender can credibly impose s against a target state, the target may choose to disarm its terrorist group, simply because the threat of the group is no longer providing protection. The target therefore has no reason to continue allowing terrorists to operate, given that the threat of these groups has not stopped the sender from imposing a punishment. Hypothesis 1. Punitive s make domestic terrorist group collapse more likely. Additionally, we might expect that if s weaken target states, and this weakness may worsen over time, target states may face a commitment problem with their terrorists. If the target will continue to weaken, terrorists may make excessive demands in the future. Sanctions may therefore encourage target states to co-opt their terrorists immediately by allowing them to enter normal politics. Hypothesis 2. Punitive s make domestic terrorist group negotiation more likely. This argument suggests that targets may use the threat of terrorism to avoid s, but will choose to disarm their terrorists if a is applied. However, this argument assumes that the target is capable of addressing the terrorists demands. While this may be true for many domestic groups, it is less likely to be true for transnational groups. The literature demonstrates that groups operating from foreign territory tend to be more violent, more extreme, and less prone to collapse (Byman et al 1999; Byman 2005). If this is the case, any weakening of the target state may make it more vulnerable to terrorism and less capable of disarming the group. We therefore expect 8

11 that while s may encourage targets to both disarm and co-opt domestic groups, s will make target states more vulnerable to transnational terrorists. Hypothesis 3. Punitive s make transnational terrorist group collapse less likely. Hypothesis 4. Punitive s make transnational terrorist group negotiation less likely. This argument suggests that s against target states that face transnational terrorist attacks from abroad may make counterterrorism more difficult. An alternative possibility is for senders to use s against the home country of transnational terrorists. For example, although Hezbollah engages in attacks in Lebanon, the source of its power is in Iran. Therefore, if s can reduce Iran s resource pool, they may force Iran into a choice between providing private goods for government allies or maintaining support for Hezbollah. If this is the case, we should theoretically expect s against the home bases of terrorist organizations to weaken these groups and increase the likelihood of their collapse. Hypothesis 5. Sanctions against home bases of transnational terrorists make collapse more likely. Data and Method To test the hypotheses formulated in the previous section, we create a dataset that combines information from two primary sources: the updated TIES dataset and Jones and Libicki s (2008) data on terrorist groups. The TIES dataset provides information on the time frame of each episode, as well as various characteristics of s: the type of imposed s, specific measures that were utilized against the ed state, the scope of s, as well as the level of target costs. The episodes included in the TIES dataset were initiated between While the total number of observations in TIES is 1412, s were in fact imposed in 845 cases. We identify punitive s as those that are aimed at weakening the target s capabilities. These include import restrictions, export restrictions, cuts in foreign aid, total 9

12 embargoes, and blockades. 5 To sharpen our results, we examine how s generally affect terrorist campaigns, and subsequently disaggregate to test each type individually. We drop episodes of threatened s that were never imposed due to either the target s decision to concede or the sender s reluctance to follow through on its threat, because our theoretical argument suggests that s affect the survival of terrorist groups through costs generated by the implementation of s. The Jones and Libicki dataset contains variables that identify terrorist groups and countries, in which the groups operate, the years of operation, the type of each group and its main objective, and the manner in which the group ended its existence. The number of groups, for which we have this information, is 648, and their operation years range from 1866 to A significant share of these groups 235 either collapsed or entered regular politics. [Figure 1 about here] Note that terrorist groups included in the Jones and Libicki dataset often operate in more than one country at the same time. Also, not all surviving groups are active until the last year of their terrorist campaigns. Therefore, we modified the Jones and Libicki dataset using information from the Global Terrorism Database (GTD) and the Terrorist Organization Profiles (TOPs). 6 First, we coded attack locations for each group, as well as countries that serve as home bases. If a group conducted attacks only in the country that served as its home base, such a group is coded as a domestic group; otherwise, we code it as a transnational group. Second, the year of the last attack 5 If we use information on target economic costs provided in the TIES dataset, we can differentiate the punitive effect of various instruments. On average, the least costly measures are import restrictions (the mean value of target costs equals 1.0), export restrictions and aid cuts (1.4 each), while the most costly s are blockades (1.7) and total embargoes (1.8). 6 The GTD is available at the TOPs database is available at 10

13 recorded in the GTD is treated as the final year of the terrorist campaign. Third, for transnational groups with multiple attack locations and/or home bases, we needed to choose one country in each set in order to construct control variables for country-specific determinants of terrorist groups survival. We assume that a group s ability to carry out attacks is the greatest in the weakest (in terms of its capabilities) country. To identify the weakest country, we compare GDP per capita of all attack locations for each terrorist group and choose the one with the lowest value. 7 We repeated the same coding procedure for home bases to choose the weakest home base, since the weakest states should be least successful in their counterterrorist efforts. This procedure also helped us eliminate most of the cases, in which major powers are either attack locations or home bases of transnational groups. Major powers tend to be most successful in eliminating terrorist threats within their borders, while being targets of frequent s episodes. 8 Subsequently, we treat the weakest country as the main country, for which all country-specific variables are obtained. As a result, we have two sets of control variables for transnational terrorist groups: one for the weakest attack location, and the other for the weakest home base. The objective of this paper is to evaluate factors determining the survival of terrorist groups. Since Figure 1 suggests that terrorist campaigns that collapse tend to do so quite quickly, we test our hypotheses using the Weibull survival model. Each model clusters standard errors by country. To estimate a survival model, we rely on three variables. Group collapse is coded as 1 in the year when a group is eliminated through the use of military or police force, and 0 otherwise. Group politics is coded as 1 in the year when a group enters regular politics, and 0 otherwise. We also 7 We can use states CINC (Composite Index of National Capability) scores for this coding procedure instead; our main findings remain unaffected. 8 We can further drop all cases of groups whose hosts are major powers from the sample; our results are largely unaffected. 11

14 create a count variable (Group time) that counts the number of years since the beginning of the group s life cycle until a given year, before the group s terrorist campaign ends. Explanatory Variables To test the hypotheses that link group and government characteristics and the use of s to the likelihood of terrorist groups collapse, we rely on three sets of variables. First, Jones and Libicki s dataset also provides information on group size (Group size), coded as an ordinal variable ranging from 0 when a terrorist group has less than 100 members, to 3 when a group has more than 10,000 members. We also count the number of attack location and home base states for each terrorist group to create the Number of homebases/attack locations variables. Values of Number of attack locations range from 1 to 10 the maximum number of states corresponds to Black September, a clandestine wing of al-fatah, which carried out a number of attacks in the Middle East, North Africa, Western Europe and North America. Values of Number of homebases range from 1 to 7: the groups with the highest number of home bases are Aum Shinrikyo and Islamic Movement of Uzbekistan. Second, to measure the government s capacity to fight and defeat terrorist groups, we rely on the following variables. The Regime type variable is a country s polity2 score. The Regime durability variable (logged) gauges how durable the country s regime is: it is a count of the number of years since the last political transition. These two measures were extracted from the Polity IV dataset. Third, we construct several variables that capture the use of s and measure various characteristics of s imposed on the country, in which one or more terrorist groups operate. The source of information on s is the TIES dataset. First, we code basic measures of imposition: New is equal to 1 in the year when a was imposed and 0 otherwise; New s total counts the number of new s imposed on a country in a given 12

15 year and ranges from 0 to 11. Second, we generate a set of variables to capture types in order to differentiate their punitive impact. We rely on the Sanction Type variable of the TIES dataset to code the following dummies: Total embargo equals 1 if Sanction Type equals 1 (i.e., total economic embargo) and 0 otherwise; Import equals 1 if Sanction Type equals 3 (i.e., import restriction) and 0 otherwise; Export equals 1 if Sanction Type equals 4 (i.e., export restriction) and 0 otherwise; Blockade equals 1 if Sanction Type equals 5 (i.e., blockade) and 0 otherwise; Aid cuts equals 1 if Sanction Type equals 7 (i.e., termination of foreign aid) and 0 otherwise. Theoretically, the most costly forms of economic s are embargoes and blockades, given that both essentially shut down commerce between the sender and the target. On the other hand, import and export restrictions represent lower levels of punitive action, given that these s do not completely sever economic ties. The foreign aid s are an interesting case. These s tend to harm the ruling coalition of the state more than the population. Foreign aid cuts are punitive in the sense that they punish the rulers of the state and their coalition, but may be less damaging to the population at large. Note that the Sanction Type variable of the TIES dataset codes all applicable types for each episode; therefore, the resulting type dummies are not mutually exclusive. In addition, we code IO as a dummy that takes the value of 1 when an international organization backs s, and 0 otherwise. Fourth, we utilize a number of control variables identified in previous studies as important determinants of terrorist campaign duration. The impact of economic s on a state can be exacerbated or weakened by the presence of other factors influencing its vulnerability to economic s. We include several such variables to capture relevant economic and geographical aspects of target states. We create a measure of a state s trade dependency by calculating the percentage of trade over total GDP (Trade dependence). The higher the percentage of trade as a share of GDP, the more 13

16 dependent the host state is on economic interactions with other states. Sanctions imposed against such states should be more economically damaging to the government and, thus, should positively affect the duration of terrorist campaigns. In addition to international economic linkages, the civil war literature has identified numerous factors influencing states propensity to experience violent rebellions. Many previous studies have linked larger populations to a higher risk of conflict (Fearon and Laitin 2003; Hegre and Sambanis 2004). Scarcity theories suggest that more populous states have more competition for scarce resources. Similarly, lower levels of GDP per capita and GDP growth are often linked with instability and increase in support for terrorist groups. Lower GDP per capita and GDP growth indicate lower levels of economic opportunities and greater poverty, and hence more significant motivations for violence. The source of data for these variables (Trade dependence, Population, GDP pc, and GDP growth) is the World Bank s World Development Indicators and we use a logarithmic transformation of the trade ratio, GDP per capita and population variables in our analyses. Geographical factors can also be linked to support for terrorist groups, as well as the government s ability to defeat them. Non-contiguous and mountainous territories tend to serve as safe shelters for terrorist groups and are more difficult for governments to subdue and control. We use two measures from Fearon and Laitin s (2003) study of states vulnerability to domestic militant challenges: the logged share of the country s territory that is mountainous (Nested mountains), and a dummy variable (Non-contiguous) that takes the value of 1 if the country has a non-contiguous territory, and 0 otherwise. Finally, to control for the effect of the Cold War rivalry between the US and the USSR on terrorist group survival, we code a dummy variable. Cold War takes the value of 1 for the years between 1947 and 1991, and 0 otherwise. 14

17 For robustness checks reported in the appendix, we created additional control variables. First, to capture group characteristics, we use Jones and Libicki s classification of terrorist groups by type: religious, nationalist, left- or right-wing. For each characteristic, we create a dummy variable that takes the value of 1 when this characteristic is present, and 0 otherwise. Second, we code two dummy variables for the issue type involved in a particular episode: these variables allow us to distinguish s motivated by political considerations and s driven by economic disagreements. The Political issue dummy equals 1 if the Issue variable of the TIES dataset takes the values of 1-11 (contain political influence, contain military behavior, destabilize regime, release citizens, property or material, solve territorial dispute, deny strategic materials, retaliate for alliance or alignment choice, improve human rights, end weapons/material proliferation, terminate support of non-state actors and deter or punish drug trafficking practices), and 0 otherwise. The Economic issue dummy equals 1 if the Issue variable of the TIES dataset takes the values of (improve environmental policies, trade practices, implement economic reform), and 0 otherwise. Third, we create a dummy variable that indicates whether or not the terrorist group controls some territory (Territorial control). Discussion of Empirical Results Tables 1 2 report the results for domestic groups, whereas Tables 3 6 focus on transnational groups. Note that the tables show coefficients, rather than hazard ratios. Models in Table 1 examine how s affect the likelihood that a domestic group will be forcibly terminated. The group s size has a negative and statistically significant effect on group termination across all models. Substantively, this indicates that larger terrorist groups are less likely to be defeated, which is not altogether surprising. However, since this effect is consistent across all specifications, it provides a baseline to compare the effects of s to. Across the eight models presented in Table 1, a one 15

18 unit decrease in the group s size (moving from a score of 2 to 1, or a score of 1 to 0) increases the likelihood that a group will be forced to end its attacks by approximately 67%. [Table 1 about here] We can now compare the impact of s imposition with the effect of the group s size. The coefficients on several variables -- i.e., imposition of s, the total number of new s, cuts in foreign aid, s supported by international organizations, blockades, and import s -- are positive and statistically significant. This indicates that these instruments increase the likelihood that a domestic group will be forcibly terminated. Paradoxically, this indicates that imposing s on a state with a domestic terrorist organization increases the odds that the state will destroy this group. Theoretically, this is consistent with arguments presented by Bueno de Mesquita and Smith (2007) and Bapat (2011), who argue that states with domestic terrorists may leverage the threat of terrorism to gain favorable economic concessions. This test demonstrates that if such governments lose these economic concessions, they may also lose the incentive to allow terrorists to persist and will subsequently take actions to disarm their domestic challengers. Figure 2 indicates the substantive effect of each type on the likelihood of group collapse. In particular, a new increases the probability that a group will be defeated by 93%. Furthermore, each new increases the probability of defeat by 17%. When we examine results for different types, we see that cuts in foreign aid increase the likelihood of group collapse by 210%; s backed by international organizations -- by 140%; and import s -- by 110%. Blockades appear to be the most effective type, but the large effect may be a function of the small number of cases in our dataset (<3%). Figure 3 illustrates these effects with survival graphs. Taken together, our findings suggest that several instruments appear to be quite effective in incentivizing states to tackle their domestic terrorists, particularly when compared to the effect of reducing the group s size. Reductions in foreign aid, s 16

19 imposed through international organizations, and s targeting imports clearly encourage target governments to take steps to disarm their domestic terrorist organizations. [Figures 2 and 3 about here] Table 2 examines how s affect governments willingness of states to allow domestic terrorists to enter politics. In these models, the coefficient on group size remains negative and significant, indicating that the likelihood of entering into politics is approximately 63-65% lower for larger groups. The only two specifications with statistically significant results are Models M3 (Aid cuts is the s regressor) and M6 (Blockade is the s regressor). The imposition of a blockade reduces the probability that a domestic terrorist group will enter regular politics nearly to zero. However, the coefficient on aid cuts is positive, and indicates that reductions in aid disbursements to states with domestic terrorists makes it over three times more likely that these groups will enter politics. This suggests that aid s make states more likely to try to disarm their terrorists, but also more likely to negotiate and accommodate these groups. This result is consistent with Bapat s (2011) argument that the provision of foreign aid to hosts of terrorists creates moral hazard and encourages these states to prolong terrorist campaigns. These results imply that the imposition of s can be an effective way of addressing the moral hazard problem: by denying foreign aid to states fighting terrorists, ers force such states to either repress domestic terrorist groups or negotiate with them to end inefficient conflicts. In sum, the imposition of s can incentivize states to disarm domestic terrorist groups, and the removal of foreign aid may discourage states from using domestic terrorists as bargaining leverage. [Table 2 about here] The next two sets of results focus on the duration of transnational terrorist campaigns, first in models with attack locations as targets, and then in models with home bases as targets. Table 3 examines the effect of s on the likelihood that transnational groups will be 17

20 forced to end their attacks against the states targeted with s. As in the previous models, we see that larger groups face lower odds of collapse (+69-71%). Transnational groups ability to continue attacks is only affected by new s, blockades, and export s imposed against attack locations. However, the results indicate that unlike s against states with domestic terrorists, s against states attacked by transnational groups are counterproductive because they prolong terrorist campaigns. New s decrease the likelihood of group collapse by 65%, export s reduce it by 83%, and blockades all but eliminate the possibility that a group will be defeated. The results demonstrate an interesting conclusion: whereas s encourage states facing domestic terrorism to end terrorist campaigns, s undermine counterterrorist efforts of states that combat transnational terrorism. [Table 3 about here] This finding for transnational terrorist groups, however, does not imply that s only have the (unintended) positive consequence of reducing the duration of terrorist campaigns when it comes to domestic terrorism. Table 4 demonstrates that s against home bases of transnational terrorists can increase the probability that these groups will meet a forced end. Models reported in this table show that group size reduces the likelihood of a forced end by approximately 50%. On the other hand, the total number of s, the imposition of a total embargo, and import s all significantly increase the likelihood of group collapse. More specifically, a new imposition increases this likelihood by 84%, whereas total embargoes and import s increase it by 431% and 252%, respectively. However, blockades, again, all but eliminate the probability of a forced end. These findings suggest that s against home base states are likely to push terrorists to their demise, but excessive s, such as blockades, may be counterproductive. [Table 4 about here] The results paint a somewhat mixed picture about the effects of s on the duration of terrorist campaigns. We find that s against states facing domestic terrorists may encourage 18

21 these states to either disarm or negotiate with these groups. This is particularly true of s that cut foreign aid. Our analyses also show that s against terrorist groups home bases may encourage these states to disarm their groups, but that excessively punitive s such as blockades may be counterproductive. At the same time, our results clearly demonstrate that s against states where transnational terrorists carry out their attacks weaken the ability of these states to prevent such attacks and combat these groups. Finally, we find no convincing evidence that s imposed against either attack locations or home bases of transnational terrorist groups affect these groups entrance into regular politics (see Tables 5 and 6). Taken together, these findings support two policy recommendations. First, s can encourage states to disarm their domestic terrorists. Second, new, moderately punitive s may encourage home base states to reign in transnational terrorists, but excessively punitive s will not. [Tables 5 6 about here] Conclusions We began with the observation that policymakers are often wary of imposing s against targets that are combating terrorist organizations. The argument is that s damage the target s economy, which in turn decreases the target s ability to fight terrorist groups. Sanctions may therefore be counterproductive to the sender s strategic interests, particularly if the target s domestic instability represents a security threat. Although this reasoning justifies states refusal to impose s, as in the case of overturned U.S. s against Pakistan following Osama bin Laden s death, the argument has never been empirically tested. No previous study systematically examines the relative damage caused by s to counterterrorism efforts. This study represents an attempt to fill this void and examine when and if s truly do undermine the ability of target states to engage in counterterrorism. Our results yield several interesting and somewhat counterintuitive conclusions. First, we find that s imposed against 19

22 target states that fight domestic terrorists accelerate the demise of these groups. In this case, s are paradoxically helpful in terminating terrorist campaigns rather than harmful. We argue that this may occur for two reasons. First, targets may use the threat of terrorism to deter senders from imposing s. If senders impose s anyway, domestic terrorists lose their utility to the government, which then has no reason to allow the group to survive. Second, economic damage from s may weaken the state s counterterrorist capability in the future. Hence, this anticipation may encourage the target government to accelerate its efforts to disarm domestic terrorists before any decline in the state s capability occurs. Although the conventional wisdom that s harm counterterrorism appears incorrect in the case of domestic terrorists, empirical evidence supports this conclusion when we focus on transnational terrorism. We find that s imposed against targets that experience attacks carried out by transnational terrorist groups result in longer terrorist campaigns. Consistently with previous studies that indicate that transnational groups are more violent, our finding suggests that s are counterproductive when they target a state fighting transnational terrorists. Interestingly, our empirical analyses also show that s directed at home bases of terrorist organizations encourage active supporters to cease state sponsorship and encourage passive supporters to crack down on terrorist groups. In sum, these results indicate that s harm states that are attack locations of transnational terrorists, but that s are valuable if directed at transnational terrorists home bases. To use an example, it may be the case that s against Hamid Karzai s Afghanistan are counterproductive since this state is targeted by transnational terrorists, but s against Pakistan may be valuable since Pakistan serves as the home base of these terrorist groups. Results reported in this study provide several interesting insights and directions for future research. One implication is that if s encourage targets with domestic terrorists to engage in 20

23 counterterrorism, but weaken targets facing transnational terrorists, targets may have an incentive to misrepresent the nature of their terrorists. Targets may argue that their terrorists are supported and funded by other states and hence s would only make the problem of terrorism worse. This argument might help targets to avoid s, whereas this strategy would not work if it was known that terrorists were purely domestic. Another implication of our research is that governments of states serving as terrorists home bases may also have incentives to misrepresent their relationship with the group. To avoid s, these states may seek to mask their relationship with terrorists and claim to be attack locations, rather than home bases. In this way, plausible deniability is essential to furthering the survival of terrorist campaigns. 21

24 References Bapat, Navin A Transnational Terrorism, US Military Aid, and the Incentive to Misrepresent. Journal of Peace Research 48(3): Bapat, Navin and David Carter Transnational Terrorism and Economic Sanctions. Presented at the 2013 ISA conference. Blomberg, S. Brock, Rozlyn C. Engel and Reid Sawyer On the Duration and Sustainability of Transnational Terrorist Organizations. Journal of Conflict Resolution 54(2): Blomberg, S. Brock, Khusrav Gaibulloev, and Todd Sandler Terrorist Group Survival: Ideology, Tactics, and Base of Operations. Public Choice 149(3-4): Carter, David B A Blessing or a Curse? State Support for Terrorist Groups. International Organization 66(1): Crenshaw, Martha How Terrorism Declines. Terrorism and Political Violence 3(1): Cronin, Audrey Behind the Curve: Globalization and International Terrorism. International Security 27(3): Cronin, Audrey How al-qaida Ends: The Decline and Demise of Terrorist Groups. International Security 31(1): Cronin, Audrey. 2009a. How Terrorism Campaigns End. Leaving Terrorism Behind: Individual and Collective Disengagement. Eds. Bjorgo, Tore, and John Horgan. London: Routledge, Cronin, Audrey. 2009b. How Terrorism Ends: Understanding the Decline and Demise of Terrorist Campaigns. Princeton, NJ: Princeton University Press. Cronin, Audrey Ending Terrorism: Lessons for Defeating al-qaeda. London: Routledge. Daxecker, Ursula E., and Michael Hess Repression Hurts: Coercive Government Responses and the Demise of Terrorist Campaigns. British Journal of Political Science 43(3): Faria, Joao Ricardo and Daniel Arce A Vintage Model of Terrorist Organizations. Journal of Conflict Resolution 56(4): Fearon, James D. and David Laitin Ethnicity, Insurgency, and Civil War. American Political Science Review 97(1): Feinstein, Jonathan S. and Edward H. Kaplan Analysis of a Strategic Terror Organization. Journal of Conflict Resolution 54(2): Gaibulloev, Khusrav and Todd Sandler Determinants of the Demise of Terrorist Organizations. Southern Economic Journal 79(4):

25 Gupta, Dipak K Understanding Terrorism and Political Violence: The Life Cycle of Birth, Growth, Transformation and Demise. New York, NY: Routledge. Gutfraind, Alexander Understanding Terrorist Organizations with a Dynamic Model. Studies in Conflict & Terrorism 32(1): Jones, Seth G. and Martin C. Libicki How Terrorist Groups End: Lessons for Countering al Qa'ida. Washington, DC: Rand Corporation. LaFree, Gary and Erin Miller Desistance from Terrorism: What can we Learn From Criminology? Dynamics of Asymmetric Conflict 1(3): Marinov, Nikolay Do Sanctions Destabilize Country Leaders? American Journal of Political Science 49(3): McLean, Elena V., and Taehee Whang Friends or Foes? Major Trading Partners and the Success of Economic Sanctions. International Studies Quarterly 54: Miller, Erin Patterns of Onset and Decline among Terrorist Organizations. Journal of Quantitative Criminology 28(1): Morgan, T. Clifton, and Valerie L. Schwebach Fools Suffer Gladly: The Use of Economic Sanctions in International Crises. International Studies Quarterly 41(1): Phillips, Brian J Terrorist Group Cooperation and Longevity. International Studies Quarterly, Forthcoming. Rapoport, David C Fear and Trembling: Terrorism in Three Religious Traditions. American Political Science Review 78(3): Rapoport, David Terrorism. In Encyclopedia of Government and Politics, Vol.2, eds. M. Hawkeworth and M. Kogan. London, UK: Routledge, Ross, Jeffrey I. and Ted Gurr Why Terrorism Subsides: A Comparative Study of Canada and the United States. Comparative Politics 21(4): Sandler, Todd Winning the War on Terror: Supply-side Perspective. Defence and Peace Economics 24(2): United States Institute of Peace (USIP) How Terrorism Ends. Report 48. Washington, DC: United States Institute of Peace. Vittori, Jodi All Struggles Must End: The Longevity of Terrorist Groups. Contemporary Security Policy 30(3):

26 Young, Joseph K. and Laura Dugan Why do Terrorist Groups Endure? Paper presented at the Annual Meeting of the International Studies Association, New Orleans, LA, February

27 Figure 1: Distribution of Failed Terrorist Groups Duration Periods from Start to Collapse 25