Participation in the international coalition against Daesh and the rise of foreign fighters

The flow of foreign fighters leaving for Iraq and the Syrian Arab Republic has slowed, but they often still pose a serious threat, either by encouraging others toward violence or by directly assisting themselves in a terrorist attack after their return. This article studies the effect of a country’s active involvement in a conflict zone on the flow of foreign fighters. Specifically, we test whether a nation’s participation in the international coalition against Daesh influences its number of foreign fighters. Despite the small sample size resulting from limited official data on foreign fighters, we report several interesting insights for cautious interpretation and only regarding the countries included. Findings from a negative binomial model suggest that a country’s active international role against Daesh also increases the foreign fighters coming from that country. Hence, it is important to keep in mind that the cost of a military intervention can be higher than the cost of the operation itself. Policymakers should also account for the cost of the increased number of foreign fighters and the resulting threat.

I n the wake of the Arab Spring of 2011, the European Counter-Terrorism Coordinator noted that "a significant number of radicalized people travel from the EU to conflict areas ... and pose a clear threat to internal security." By the end of 2015, the number of foreign fighters in Iraq and Syria was estimated at between 27,000 and 31,000 people. 1 Since then the number of foreign fighters leaving for Iraq and the Syrian Arab Republic has slowed, but they still pose a serious threat in at least two ways. First, while abroad they can assist and encourage others to execute attacks in their country of residence. Second, they can be directly involved in terrorist attacks when they return. The attack on the Jewish Museum in Brussels in 2014, but also the attacks in Paris in 2015 and those in Brussels in 2016 constitute painful anecdotic evidence of this. Recent estimates suggest that about 30 percent of 5,000 European foreign fighters have returned home. Moreover, returning foreign fighters are more effective terrorists than non-veterans. While abroad, they have often built a social network and gained experience on the battlefield. As Braithwaite and Chu note, foreign fighters present an important form of trained human capital to perform attacks at home. Their study shows that having a significant number of foreign fighters abroad increases the likelihood of terrorism in the home country, at least when the conflict is won by the rebels. 2 To respond appropriately to this threat, a thorough understanding of the triggers for foreign fighters to leave their country of residence is necessary. Official data on the topic is limited but clearly shows that the number of foreign fighters is not simply related to a country's population (nor to the composition thereof). Recent literature identifies multiple variables that play a role in explaining the flow of foreign fighters. Among others, a country's economic prosperity and population are believed to significantly affect the number of foreign fighters, as are the percentage of Muslims and the distance to Syria. 3 In this article we test whether military counter-terrorism policy, expressed by a partnership in the international coalition against Daesh, has an impact on the number of foreign fighters leaving the country. While data availability limits the scope of our tests, we are able to include from 24 to 49 countries, depending on model specifications. Including the usual set of control variables in our negative binomial model shows that, among the countries included, coalition members have substantially higher numbers of foreign fighters leaving to fight in Iraq and Syria as compared to noncoalition countries. This finding has important policy implications since it suggests that military interventions in Iraq and Syria have backlash effects in the home country. The additional costs need to be accounted for when deciding whether to intervene in the conflict area. When a decision to intervene is made, it needs to be complemented with considerable preventive policies as well. 4 The article is organized as follows. First we review the relevant literature on radicalization as the literature on foreign fighters per se is rather scarce. In addition, the section develops the research question. Subsequently, the data collection process and the descriptive statistics are presented. This is followed by a description of the model to be estimated and a discussion of the main results. The final section concludes.

Theoretical framework and research question
The literature on radicalization 5 A review of the literature on terrorism reveals that the answer to how and why an individual engages in terrorism often boils down to finding the reasons how and why someone radicalizes. Radicalization can indeed be a pathway to terrorism but is neither a necessary nor a sufficient condition. Many individuals with radical ideas never turn to terrorism and foreign fighter groups in Iraq or Syria also contain terrorists who are not driven by a radical belief, but are rather motivated by the opportunity to escape a life seen as meaningless. A universal terrorist profile does not exist: Terrorists' motives and roles are heterogeneous. 6 To understand the flow of foreign fighters, then, we need to draw on the literature on radicalization. This literature shows that the factors causing someone to evolve toward violent terrorism are not only inherent to the individual (e.g., perceived deprivation or personal grievance) or related to the group or the direct environment of the person (meso-level determinants) but that macro-level determinants stemming from society also play an important role in the radicalization process. 7 Globalization and modernization as well as foreign policy of some (Western) countries constitute typical examples of macro-level determinants which can initiate or advance radicalization. Globalization can threaten group identity and reinforce an us-versus-them way of thinking. A black and white view of the world is easy and offers a feeling of security. Globalization also results in higher mobility of people, leading to ever-increasing numbers of refugees and international migrants worldwide. Migration politics also has a significant effect on terrorism events in the home country. 8 The geopolitical policy of a country likewise can affect the likelihood of terrorism. Some Muslim groups experience the Western way of life as a threat to their personal lifestyle and some interpret Western geopolitical policy as a threat to the Muslim community at large. One of the defenders of this thesis is the French Professor François Burgat who explains that the vulnerability of France to terror attacks partly stems from its colonial past and its geopolitical policy. Others posit that suicide bombers against Western targets often are driven by nationalist motives. If we extrapolate this reasoning to the number of foreign fighters, we expect a positive relationship between the number of foreign fighters leaving from a country to fight in Iraq and Syria and the foreign policy of this country with respect to the conflicts in Iraq and Syria. If active involvement in the Syrian conflict is interpreted by Muslim society as a threat to their lifestyle, for instance, this could motivate more people to leave and join the fight on the other side, i.e., on the side of Daesh. 9 Empirical evidence documenting the relationship between military deployment and terror attacks already exists. Foreign military interventions which support and help the government and which involve a large number of ground troops increase the incidence of suicide attacks performed by regime challengers. This is partly explained by the phenomenon that military interventions strengthen the power of the local government and increase the defense of the targets. Thus, insurgents resort to nonconventional, more lethal tactics such as suicide attacks. The military presence of a foreign country also significantly increases the probability of a suicide terrorist targeting the police. Military interventions do not only seem to affect the tactics used in the conflict zone but also in the country which deploys the troops. For instance, U.S. military involvement in different conflict zones has made it more attractive for international terror plots, a finding later confirmed for all NATO countries (over the period 1998-2007): Military deployment to conflict areas results in a significant increase in the probability of a terror attack in the deploying country. And although other researchers initially found that the deployment of U.S. troops decreases the number of terror attacks affecting the host country, the effect fades after controlling for strategic goals (in casu, oil). 10 Clearly, a range of evidence suggests that the total cost of military deployment to a conflict area surpasses the explicit, budgetary cost of sending troops. Among the implicit costs are the reaction to military interventions in terms of increased terror attacks. We study whether military interventions also lead to an increased number of foreign fighters leaving from the troop-deploying country. If this is the case, then an increase in the flow of foreign fighters is an implicit cost of military counter-terrorism policy. Hence our research question: Does participation in the Global Coalition Against Daesh in Iraq and Syria lead to a larger flow of foreign fighters?

This article reports that countries that have joined the Global Coalition Against Daesh (also known as IS, ISIS, or ISIL) experience statistically significant increases of their citizens joining as foreign fighters for Daesh
, and also posing a risk of backlash terror attacks for the sending country. The policy implication is that in addition to explicit budgetary costs, policymakers in sending countries should take into account the eventual cost of possible backlash terror attacks in their home countries.

Data and descriptive statistics
The dependent variable comes from a Soufan Center report and counts the number of foreign fighters going to Iraq and Syria. It captures official, and for some countries also unofficial, data. An update of this report was issued in 2017. As the update did not revise the n u mb e r s f o r c e r t a i n countries, and left out others, we opt for the 2015 edition. The official data stems from government estima tes regarding foreign fighters. Other measures are usually derived from UN reports or academic sources. We opt for the official data which leads to a sample of 49 countries (30 of which are coalition members). This limits the dataset by excluding countries for which official numbers are not available, such as Afghanistan, Kuwait, and Libya. While official data are more reliable than the unofficial numbers, we point out that even for the official statistics different countries use different measures. Hence, cautious interpretation of the results is warranted. 11 Our independent variable of interest is country participation in the Global Coalition Against Daesh. Established on 17 October 2014 "to formalize and combine ongoing military actions against the threat posed by IS in Iraq and Syria," the coalition is led by the United States and at the time of writing consists of 79 countries. Contributions to the coalition can take the form of military support but also of human and/or financial support. While the types of contributions differ, all types are "visible" and can induce radicalization in the contributing country. A dummy variable indicates country involvement in the coalition, irrespective of the type of contribution. Table 1 lists the number of foreign fighters in 2015, grouped by coalition and noncoalition countries. We only report countries for which we have official data. This does not mean that other countries have no foreign fighters but merely points to the absence of official statistics. 12  Table 1 shows no direct relation between the number of foreign fighters and the size of sending countries but several smaller countries have relatively high numbers of foreign fighters, especially when they are coalition members against Daesh. For noncoalition countries, the range is between 1 and 2,400 foreign fighters, and for coalition countries between 1 and 6,000. Russia is the source of by far the highest absolute number of foreign fighters among the noncoalition countries (2,400). The active role of the government of Russia in the conflict, albeit outside the U.S.-led coalition, potentially offers at least part of the explanation. Tunisia has most foreign fighters among coalition members (6,000), followed by Saudi Arabia (2,500), and Turkey (2,200). The median, and mean, for foreign fighters of coalition members is about double the value of that of noncoalition members.
On a per capita basis (in terms of the total population as well as in terms of the Muslim population), the Maldives and Lebanon show the highest relative numbers of foreign fighters with, respectively, 49 and 15 foreign fighters per 100,000 people. India and South Africa report the lowest relative numbers. Among coalition members, Tunisia has by far most foreign fighters, not only in absolute (6,000) but also in relative terms (53 per 100,000). Tunisia is followed by Jordan (22 per 100,000) and Kosovo (13 per 100,000). Finland has most foreign fighters relative to the size of its Muslim population.
Apart from coalition membership, our models include a set of control variables inspired by the literature. Countries' economic, social, and political characteristics are reported to affect the number of people that leave to fight for Daesh. We include GDP per capita as a measure of economic prosperity and development. Poor economic conditions can nourish feelings of economic deprivation and marginalization. People living in poorer areas have fewer possibilities to develop a prosperous future and could hence have a higher propensity of developing radical behavior. GDP per capita captures averages and thus hides information on the distribution of economic wealth. But wealth distribution can also play an important role as a poor individual in a poor country may be relatively happier than a poor individual in a rich country. To account for the potential effect of inequality we include countries' Gini coefficient which take a value between 0 and 100. (A value of 0 represents total equality; the higher the coefficient, the more unequal the distribution.) In addition, our models control for population size as it is reasonable to assume that more populous countries host a larger pool of potential foreign fighters. We also include the distance between Damascus and the capital of the sending countries, as being closer to Iraq or Syria presumably makes it easier to leave as a foreign fighter. 13 Further, we draw on the radicalization literature for indicators regarding feelings of injustice and deprivation. The perception of unfairness can be provoked by the fact that Muslim groups often represent a minority in Western countries. While the lifestyle is already different, every restriction posed can be perceived as a threat, providing a cognitive opening for radicalization. Daesh enlarges the differences between the two lifestyles and offers a radical rhetoric against the Western one. In addition, the group offers identity and a sense of belonging. Hence, the more restrictions a country imposes on the practice of religion, the more prone members of a minority religion may become for radical ideas. We thus include a Government Restriction on Religion Index (GRR), ranging from 0 (very low level of restrictions) to 10 (very restrictive). The potential for democratic participation needs to be accounted for as well. The more people can participate in public debate, the lower political frustration will be. As a measure for democracy, we opt for a Polity Index, ranging from -10 (strongly autocratic) to +10 (strongly democratic). Finally, to include a measure capturing the degree of homogeneity of the society in a country, we include the percentage of Muslims in the total population. As Daesh constitutes an Islamic organization, we expect a positive relationship between the size of the Muslim population and the number of foreign fighters. In addition, we account for ethnic, linguistic, and religious fractionalization, where higher levels for the respective indices represent more fractionalized societies. Table 2 provides descriptive statistics. 14 We again remark that the sample is limited, consisting of 49 countries (the countries for which we have official data on foreign fighters in 2015). And as we do not have Gini coefficients for a number of countries, including it further reduces the sample size. We thus conduct all tests with and without it.

Model and results
We opt for a negative binomial regression to study whether a country's coalition participation influences the number of foreign fighters that left it (in 2015), while controlling for other potential effects. In all, we have 11 variables in the dataset, namely coalition member, GDP per capita, Gini coefficient, population size, distance to Damascus, the GRR Index, Polity Index, Muslim percentage, and three fractionalization indices, ethnic, linguistic, and religious Our model choice results from the count nature of the dependent variable. A Poisson model is not appropriate because the assumption of equal mean and variance is rejected in our data.
Results are shown in Table 3. To allow for a more direct and intuitive interpretation, we report incidence rate ratios (IRR) rather than coefficients. As mentioned, we have a limited number of observations for the Gini coefficient. Hence, we run the test with and without this variable. For each of these options, we proceed with a backward step-down selection, that is, a stepwise elimination of the least significant variables. This entails that we start by including all variables for which we find support in the literature and then rerun the estimation, each time dropping the least significant variable. Such a procedure does not only limit the number of independent variables (increasing the degrees of freedom) but also allows us to check, to the extent possible, the robustness of the results for our variable of interest. Backward selection is not without limitations but nevertheless is among the most widely used techniques when it comes to model selection and verification. We thus remind readers once more that results should be interpreted with caution considering the limited official data on foreign fighters and the resulting small sample of countries included in the estimations.
Regarding the variable of interest-participation in the coalition against Daesh- Table 3 mostly reports a substantial influence. We find an extremely high impact when reviewing the test results, including the Gini coefficient on the side of the independent variables, with a reported IRR for coalition membership of 45.15. This would mean that participation in the coalition against Daesh increases the number of foreign fighters by a factor of 45.15 as compared to the average country that does not participate. However, the small sample size combined with the inclusion of the full set of independent variables substantially decreases the statistical reliability of these results. We thus focus on the outcome of both stepwise elimination processes (one starting from the full model, the other one directly leaving out the Gini coefficient) and notice a smaller, yet still substantial, impact with IRRs of 2.39 and 2.62 for coalition members. From these results, we conclude that being a member of the coalition against Daesh increases the foreign fighters in coalition countries by a factor of about 2.5, compared to the noncoalition countries in the dataset. It thus appears that coalition membership has a substantial influence on the number of foreign fighters leaving from a coalition country. We directly add that these results must be interpreted keeping in mind the small sample size. Due to the limited amount of official data available, the estimations include a cross-section of 49 countries. Hence, the results cannot be generalized to other countries. In addition, even though data on foreign fighters is drawn from official sources, we cannot fully exclude differences in measurement techniques, nor measurement errors.
We further find that richer countries have a slightly higher number of foreign fighters. As for population, only one out of four negative binomial tests reveals that a larger population also results in more foreign fighters. However, a substantial Pseudo-R2 0.0965*** 0.0447*** 0.0591*** 0.0501*** positive size effect is found regarding the variable measuring the Muslim population in a country. Interestingly, we find seemingly conflicting results for distance. The first model reports a positive relationship, meaning that countries further away from Iraq and Syria have more foreign fighters. This is somewhat counterintuitive as we expected that being closer to the Iraq and Syria would facilitate the flow of foreign fighters. The results from both stepwise eliminations, however, indeed report this negative relationship between distance and the number of foreign fighters. This is possibly explained by the set of countries included in the different models. By including the Gini coefficient, a very specific group of countries drops out. Several of these countries are among the ones that are closest to the conflict. 15 Often it is argued that the United States constitutes an influential statistical observation in the data. As the coalition against Daesh was initiated and lead by the U.S., the results could be driven by its presence in the dataset. Therefore, we also ran our regressions excluding U.S. data. This did not lead to substantially different conclusions. For the most part, the results are similar to the stepwise elimination exercise after initially excluding the Gini coefficient. The incidence rate ratio for coalition membership is 2.69, still pointing toward a substantial influence.

Conclusion
Keeping in mind the small sample size due to limited official data on foreign fighters, we document a positive relationship between membership in the Global Coalition Against Daesh and the number of foreign fighters. For the average of the 30 countries included in the estimation, coalition membership results in about 2.5 times more foreign fighters as compared to the average of the 19 countries not in this coalition. If we can interpret the number of foreign fighters leaving from a country to fight at the side of Daesh as an indication of radicalization in a country, this study hence shows that coercive military counter-terrorism policy affects the level of radicalization. In other words, participation in the anti-Daesh coalition increases the support for Daesh in the home country, at least as expressed by the number of individuals leaving to fight.
This study is not free of limitations. Several questions remain, creating opportunities for further research. The conclusions apply, of course, only to the data used in this study and results should be interpreted with caution. First, official data regarding the number of foreign fighters is available only for a limited number of countries. In addition to a small sample size, we point out that our estimations concern only a crosssection. Even though we work with official data only, potential measurement errors cannot be excluded. A more extensive (number of countries as well as a longer period of time) and coherent dataset would thus be very valuable to have at hand for future research. Second, we have used a dummy variable to measure involvement in the coalition against Daesh. Clearly, there are different types of support a country can deliver to the coalition and hence potentially different effects may result from that. Future studies could examine whether there is a difference in the specific contribution a country delivers to the coalition, e.g., the effect of military support versus humanitarian support. Furthermore, it would be interesting to study the effect on the number of foreign fighters of different types of military intervention as well as of past grievances resulting from a country's involvement in other regions.
Despite these limitations, the main results for our sample document a strong effect and hold during robustness tests. The size of the effect certainly offers food for thought. Since security is a public good which surpasses country borders, we need to act as an international society. We cannot merely rely on other countries to go and fight terrorism while we only focus on our own protection. Hence, this study should certainly not be interpreted as a plea against military action. It does, however, argue that the costs related to military interventions surpass the direct budgetary costs. Since military operations increase the flow of foreign fighters from a country, one also needs to account for the societal cost of this increase.
Most countries use a broad spectrum of policies regarding counter-terrorism and counter-radicalization. Diverse policies should be seen as complements, not substitutes. Especially if military interventions increase radicalization, this study suggests that complementing this policy with alternatives focusing on the prevention of radicalization is of crucial importance. These policies aim to increase the opportunity cost of going to Syria (or Iraq) to fight. If these opportunity costs are sufficiently high, this can lead potential candidates to refrain from leaving. The focus should thus probably be on increasing the benefits of not leaving, having more to give up, answering the need to belong in an alternative way. In short, to provide a long-run response to the danger posed by both foreign fighters as well as by homegrown fighters, repression by use of military action is only one part of the answer and it needs to be complemented with preventive policies. 16

Notes
Suggestions made by participants at the International Conference on Economics and Security (27-30 June 2018 at Middle Eastern Technical University, Northern Cyprus) are gratefully acknowledged as are comments from anonymous reviewers.
15. Whereas these estimations study the absolute number of foreign fighters and report a substantial influence of participation in the coalition against Daesh, it is also interesting to study which determinants influence the relative numbers of foreign fighters (compared to the total population of a country and compared to a country's Muslim population). While these estimations necessitate a different econometric approach resulting from the changing nature of the dependent variable, they provide very similar results. To avoid undue repetition they are not reported here but are available upon request. Participation in the coalition against Daesh thus substantially increases the number of foreign fighters relative to a country's population as well as the number of foreign fighters relative to a country's Muslim population. The variable GDP/capita, however, loses significance in one test, namely when estimating the effect of coalition participation on the number of foreign fighters relative to a country's population.