ABSTRACT COMMITMENT AND RADICAL RELIGIOUS SECTS: AN EMPIRICAL EXAMINATION OF THE ROOTS OF TERRORISM. by Kendrick Morales

Transcription

1 ABSTRACT COMMITMENT AND RADICAL RELIGIOUS SECTS: AN EMPIRICAL EXAMINATION OF THE ROOTS OF TERRORISM by Kendrick Morales This paper provides an empirical examination of the observed use of two technologies that foster commitment (Sacrifice and Stigma (S&S); Social Service Provision (SSP)) to explain the incidence of terrorism. Proxies for these technologies were identified to evaluate the existence of such a link at the country-level. The cross-sectional analysis in this paper is consistent with the conclusions expressed in the extant theoretical literature that the commitment inducement (or identification) theory could be correct. Specifically, 3 out of the 4 proxies for S&S were robust predictors of terrorism and 6 out of the 7 proxies for both S&S and SSP were significant predictors of overall violence at the.01 level. These findings suggest that radical religious sects are remarkably successful in the production of coordinated violence and (when suitable) terrorist activity due to their exceptional ability to form a coalition of highly committed members.

2 COMMITMENT AND RADICAL RELIGIOUS SECTS: AN EMPIRICAL EXAMINATION OF THE ROOTS OF TERRORISM A Thesis Submitted to the Faculty of Miami University in partial fulfillment of the requirements for the degree of Master of Arts Department of Economics by Kendrick T. Morales Miami University Oxford, Ohio 2014 Advisor: Reader: Reader: George K. Davis Prosper Raynold Jing Li

3 Table of Contents INTRODUCTION... 1 PART I... 3 REVIEW OF THE THEORETICAL LITERATURE... 3 Commitment... 3 Commitment-Fostering Technologies... 4 I. Sacrifice and Stigma... 5 II. Social Service Provision... 7 REVIEW OF THE EMPIRICAL LITERATURE PART II DATA JUSTIFICATION OF PROXIES FOR SACRIFICE AND STIGMA Government Restrictions Index (GRI) Social Hostilities Index (SHI) Group Grievance (GG) and Intergroup Cohesion (IC) JUSTIFICATION OF PROXIES FOR SOCIAL SERVICE PROVISION (SSP) Corruption Perceptions Index (CPI) Refugees and Internally Displaced Persons (RIDP) Uneven Economic Development (UED) PART III EMPIRICAL RESULTS AND INTERPRETATION ROBUSTNESS Switching the Economic Variable Instrumental Variables Estimation Lag Dependent Variable Link with Global Peace Index PART IV DISCUSSION OF RESULTS POLICY IMPLICATIONS FUTURE RESEARCH BIBLIOGRAPHY ii

4 List of Tables Table 1 - Descriptive Statistics 15 Table 2 - Components of the Government Restrictions Index and Social Hostilities Index 16 Table 3 - The Costs associated with Non-compliance to Burqa Ban: Before and After 18 Table 4 - Terrorism, Country Characteristics with S&S/SSP Separately 24 (OLS w/ Heteroskedasticity-Robust Standard Errors) Table 5 - Terrorism, Country Characteristics with S&S/SSP Combined 25 (OLS w/ Heteroskedasticity-Robust Standard Errors) Table 6 - Terrorism, Country Characteristics with S&S/SSP and HDI 27 (OLS w/ Heteroskedasticity-Robust Standard Errors) Table 7 - Terrorism, Country Characteristics with S&S/SSP Separately 29 (IV w/ Heteroskedasticity-Robust Standard Errors) Table 8 - Terrorism, Country Characteristics with S&S/SSP Combined 30 (IV w/ Heteroskedasticity-Robust Standard Errors) Table 9 - Terrorism, Country Characteristics with S&S/SSP Combined and Lag Dependent 32 (OLS w/ Heteroskedasticity-Robust Standard Errors) Table 10 - Terrorism, Country Characteristics with Only S&S/SSP Combined and Lag Dependent 33 (OLS w/ Heteroskedasticity-Robust Standard Errors) Table 11 - Terrorism, Country Characteristics with S&S/SSP Combined and Lag Dependent 34 (IV w/ Heteroskedasticity-Robust Standard Errors) Table 12 - Violence, Country Characteristics with S&S/SSP Combined 36 (OLS w/ Heteroskedasticity-Robust Standard Errors) Table 13 - Violence, Country Characteristics with S&S/SSP Combined 37 (IV w/ Heteroskedasticity-Robust Standard Errors) iii

5 Introduction Radical religious sects are potent contributors to the incidence of terrorism. Radical religious sects are exceptionally adept in the art of mobilization where their success can be attributed to their use of technologies that build commitment (Berman and Iannaccone, 2006; Raynold, 2013). If the theory is correct, these commitment inducement (or identification) technologies should have explanatory power for the incidence of terrorism. This paper is an attempt to test the hypothesis that observed use of two of these technologies, Sacrifice and Stigma and Social Service Provision, explains terrorist activity. While Berman and Laitin (2005; 2008) have examined the statistical link between commitment inducement (or identification) technologies and terrorism at the organizational level, the empirical literature has neglected to investigate whether a link exists at the country level. Yet this topic warrants meaningful consideration. If the successful recruitment of terrorists requires a certain kind of environment, then it would suggest that counter-terrorism policies should focus on creating an unfavorable environment for radical religious groups in contrast with attempting to eradicate every potential terrorist. This paper is the first of its kind to invoke country-level data in an empirical examination of the effect of commitment inducement (or identification) technologies on terrorism. Still, the nature of this topic immediately raises the following question: what is the theoretical basis for why radical religious sects (or more generally, religious sects) are far more successful at building commitment than other groups? Berman and Iannaccone (2006) conclude that religious sects enjoy a comparative advantage. More precisely, they benefit from an extraordinary ability to screen out free-riders and generate collective production. In addition, exploiting the afterlife also contributes to their success but to a much lesser extent. For instance, if supernatural or theological motivations were an adequate explanation for the successful nature of a religious sect, then you would expect to observe this advantage across all religious groups. A religious sect, on the other hand, is the only type of group who demonstrates the capacity to inspire remarkably high levels of conviction, solidarity, and commitment. 1

6 Yet Raynold (2013) reasons that the failure to account for afterlife considerations prevents religious sects from altering the overall distribution of commitment levels; thus, religious sects are only permitted to identify the highly committed. Expectations of temporal rewards such as the religious experience or the provision of essential services are not the primary engine of commitment. Rather, commitment is driven predominately by expectations of supernatural rewards. Thus, by explicitly incorporating a supernatural motive in the analysis, religious sects are not only granted the capacity to identify the highly committed but also to induce high levels of commitment. Whether the technologies employed by religious sects have the capacity to induce or simply identify commitment is worth considering. For the purposes of this paper, however, these differences are not essential given that the objective is to determine if the mere use of these technologies leads to terrorism. Although this paper does not focus on delineating between the effect of commitment inducement and identification technologies, an intended objective is to motivate further study involving the importance of the afterlife in the commitment-building process. Is it really just a coincidence that many of the most lethal organizations in the production of terrorism are explicitly connected to religion? Are afterlife considerations really as benign as they are purported to be or do they deserve further examination in understanding terrorist activity? A prerequisite to answering these questions and more is to thoroughly understand what technologies radical religious groups employ and how the use of these technologies affects terrorism. Only then will evaluating the importance of the afterlife in the employment of these strategies warrant further consideration. In Part I, the relevant theoretical and empirical literature is discussed. Part II reports the data used and justifies the proxies for two commitment-building technologies (Sacrifice and Stigma; Social Service Provision). Part III discusses the empirical model and the subsequent results. Part IV draws the main conclusions and discusses the policy implications and opportunities for future research. 2

7 Part I Review of the Theoretical Literature Commitment A central component to understanding the connection between radical religious groups and terrorism begins with an understanding of religious affiliation. Consistent with fundamental economic theory, Raynold (2013) developed a theory of religious affiliation where an explicit link between commitment and utility is articulated. For example, affiliation with a religious organization requires that rational agents exhaust some amount of resources (which is costly). Yet the fact that many rational agents become members suggests that the benefits derived from membership exceed the associated costs. Commitment to the organization, therefore, may be defined in terms of utility where it can be precisely described as the maximum utility the agent is willing to forego to maintain membership in the organization and that this level of utility cannot exceed the utility the agent expects to derive from his/her membership Raynold (2013; pp. 5). In other words, how tolerant a member is to a loss in utility reflects their level of commitment to an organization. This implies the existence of an inseparable link from the costs of membership to the degree of commitment. The conclusions of Berman and Laitin (2008) are consistent with Raynold s definition of commitment. When radical religious groups turn to terrorism, the operations that members conduct are both lethal and ubiquitous. The highly illegal and deadly nature of their activities results in perceived defection costs that are extremely high. For instance, participation in terrorist activities brings forth the wrath of the authorities if perpetrators are caught or identified. Additionally, members have inside information that could be relayed to the authorities where knowledge of organizational strategies and targets are valuable (especially when the target is in wealthy nations). The decision to remain a member, therefore, is exceedingly costly for individuals (suicide attack being the extreme case). As it turns out, the most effective way to address this problem is to only grant admission to members who are highly committed, suggesting that radical religious groups will impose high costs on members. 3

8 Raynold (2013) provides additional insights into the theoretical link between commitment and terrorism by explicitly accounting for a supernatural motive. When supernatural motives are taken into account, asymmetric information becomes an issue for members and religious intermediaries alike. Although there are ways for both parties to reduce the impact of asymmetric information, religious intermediaries are essentially free to guide their members in any direction they desire. As a result, if radical religious groups are effective at building commitment in the midst of the inevitable asymmetric information problem, the likelihood of religious intermediaries directing members in a manner which is conducive to violence is much higher; thus, the incidence of terrorism is likely to become more prevalent. When the stakes are exceptionally high as they are in producing terrorism, fostering commitment and thereby controlling defection is crucial to the organization s long-term survival. As such, it is necessary to build off the existing theoretical literature and, specifically, the technologies that are employed by organizations to build commitment. The next section discusses two of these technologies. Commitment-Fostering Technologies Radical religious groups appear to play a crucial role in the production of terrorists but what methods are they employing to form a coalition of such committed members? What kind of motivations do these groups evoke in the mind of an attacker which somehow grants themselves permission to harm others and perhaps even themselves? The breakthrough in understanding terrorism and, more specifically, the ability of radical religious groups to foster commitment stems from an application of the knowledge acquired through exhaustive research on relatively peaceful (albeit strict) religions, cults, and other sects. Namely, radical religious groups are exceptionally strong at screening out halfhearted members and in the production of collective goods and services. These characteristics are essential in controlling defection which allows radical religious groups to become remarkably effective in the development of lethal terrorists. The rest of this section is devoted to discussing two commitment-fostering technologies employed by these groups. 4

9 I. Sacrifice and Stigma A critical input in the commitment identification process is Sacrifice and Stigma (i.e. strictness) which evolved from economic and sociological foundations. Rational choice theory, on the surface, appears to predict that ceteris paribus rational agents will always select into moderate religious groups as opposed to selecting into more extreme alternatives. A stricter group imposes additional costs on its members by preventing them from engaging in behavior that they would otherwise prefer to experience. Yet it is an empirical fact that the stricter religions experience higher participation rates and grow much faster than their moderate counterparts (Kelly, 1986; Iannaccone, 1992). In the wake of these empirical findings, the literature became dominated by the following explanation: religious groups are best thought of as robust social clubs. Laurence Iannaccone (1992) led the charge in the development of the club good approach with his seminal work on Sacrifice and Stigma (i.e. strictness) which is widely acknowledged as the benchmark for the rational choice approach. The benefit of membership to a religious group is not merely based on individual participation but it also depends on the participation of fellow members experiencing the religious activity collectively is a crucial part of the attraction. Religions, therefore, suffer from a free-rider problem which threatens to dilute the religious experience of committed members. In the extreme, the free-rider problem leads to the club s total elimination as the incentive for members to participate continually wanes. Nevertheless, it is inherently difficult for clubs to screen out free-riders. Having to continually monitor a member s level of commitment, enthusiasm, effort, etc. is a lofty and perhaps unattainable goal. Since monitoring members and directly rewarding participation is impractical, religious groups often settle for imposing prohibitions on normal activities opting to observe and penalize mere involvement opposed to having to monitor the level of involvement in those groups Iannaccone (1992; pp. 275). The fact that members are required to behave in a manner that is inconsistent with normal behavior serves to mark members as relative pariahs. Consequently, free-riders are more likely to be screened out due to the increase in the cost of extra-denominational activities, resulting in more satisfying experiences for committed members. Subsequent research has offered additional support for the club good approach as a viable method to explain the process via which individuals select into differing religious bodies. Eli 5

10 Berman (2000) famously applied the club model to explain the behavior of Ultra-Orthodox men living in Israel. Berman posits that these men study full-time in Yeshiva much longer than their North American counterparts (despite many being in poverty with families) due to the desire to sufficiently signal commitment to their community. For instance, Israel complements Yeshiva attendance with a draft deferment. This policy acts like a subsidy which reduces the costs of membership and thereby reduces the potency of the commitment signal for Ultra-Orthodox men. Therefore, the draft deferment compels the Ultra-Orthodox men in Israel to sacrifice much longer than their North American counterparts to generate a similar signal. Michael McBride also applies the club model (with slight modifications) to the Mormon Church where he explicitly examines their use of the menu-monitoring approach. Consumers of club goods, like other goods, have a variety of options to which they may indulge. McBride asserts that the Mormon Church exploits this reality by imposing strict behavioral requirements while, at the same time, allowing for the possibility of free-riding (particularly, in the short-run). This strategy provides members with the opportunity to make an informed, unadulterated decision with regards to their consumption decision. In addition, the relative value that the Mormon Church provides members (as a direct result of their strictness ) improves the chances that rational individuals will select into their group. All of the aforementioned outcomes are expected to result in the formation of a strong, cohesive group with a consistent inflow of highly committed members. Unlike the previous club model literature, Raynold (2013) develops a theory of strictness that allows religious groups to induce commitment as opposed to merely identifying the highly committed. By including a supernatural motive in the analysis, the members and religious intermediaries are presented with a complication in the form of asymmetric information. From the member s perspective, they must ingratiate themselves with the supernatural being to achieve their goals. The religious intermediary, however, is the agent who informs members of the will of the supernatural being which provides them with an opportunity to influence the cost of membership in a desirable way. Moreover, the way in which members are required to sacrifice may be delineated into two forms: fungible payments and non-fungible payments. Fungible payments may be characterized as marketable (e.g. money or real-estate) whereas non-fungible payments are thought to have limited marketability potential (e.g. dress codes or dietary 6

11 restrictions). The religious intermediary ceteris paribus prefers fungible payments as it provides them with the freedom to re-allocate funds in a way that promotes their agenda. In contrast, members perceive religious intermediaries who primarily demand non-fungible payments as more credible. Consequently, if the religious intermediary places greater weight on demanding non-fungible payments as opposed to fungible payments, this choice would not only screen out halfhearted members but it would also induce higher levels of commitment from the existing distribution of members. Although the extant literature focuses on understanding strictness at the group level, the theoretical implications suggest that the analysis may be extended to the country level. Recall that the defining feature of a sect is their uncompromising stance to distance themselves from society typically achieved through placing rigid behavioral requirements on its members (Dynes 1955). The key insight in linking sacrifice to commitment then is to imagine how the leaders of a sect would accomplish their goals; particularly, how do they build commitment? Given that strictness is an effective strategy to induce (or identify) commitment, the employment of this technology suggests groups will demand that members not comply with societal norms. This theoretical insight provides an inventive outlook for quantifying the effect of strictness at the country level. Part II presents a more complete justification for the strictness proxies. II. Social Service Provision Another commitment-fostering device employed by radical religious groups is the provision of collective goods and services which is perhaps the most famous policy application evolving from the club model approach. The concept of Social Service Provision asserts that the combined failure of the government and market to deliver essential services to households offers radical religious groups a chance to provide these services and thereby foster commitment. Berman and Laitin (2008), most notably, promote this idea where they find that radical religious organizations provide benign public goods to identify highly committed members and gain support from the surrounding community. Given that controlling defection is essential to survival, providing social and economic services to citizens is an important complement in the commitment identification process. Indeed, a prerequisite to the employment of strictness is often the provision of public goods where many radical religious groups such as the Taliban, Hamas, and Hezbollah apply this strategy so that they may have the opportunity to observe 7

12 signals of commitment (the act of observing signals of commitment is where strictness comes in). For instance, the radical religious group must persuade the population to tolerate their presence (at minimum). Without the provision of social services, the population would be unlikely to even entertain the notion of becoming a member (much less actually becoming a member) for they would not be benefiting from their presence in some way. What social service provision achieves is an increase in committed members (since they are able to observe more signals of commitment even if many more uncommitted members are quickly screened out via strictness) and a surrounding population that is more friendly to them (e.g. they will be more willing to look the other way if they see illegal activity or not cause any other problems for the radical group). In contrast to Berman and Laitin (2008), Raynold (2013) develops a theory of commitment inducement in which the provision of social services has the capacity to induce commitment as opposed to merely identifying commitment. Similar to the strictness strategy articulated by Raynold (2013), religious affiliation possesses an inherent asymmetric problem where it is impossible to verify the quality of the actual religious product (since it s safe to assume that nobody ever reemerges after dying and then provides evidence about the validity of various religious claims). Despite this shortcoming, it is in the best interest of both potential members of a religious organization and the faith intermediaries to provide essential services. From the potential member s perspective, they desire compliance with the will of a supernatural being. If the religious organization is competent in one area of production (the provision of collective goods), they are more likely to be competent in accurately informing members on how to comply with the will of a supernatural being. Thus, social service provision signals to potential members that the religious intermediaries are credible. From the religious intermediary s perspective, providing a signal of their credibility is crucial as the organization s livelihood relies on maintaining the quality of their product. Moreover, these signals of credibility not only have the capacity to identify commitment, they can also induce commitment out of their members. If the circumstances are such that potential members view the benefit of the religious products to outweigh the costs, then religious organizations will be able to use the provision of social services to induce commitment. 8

13 Although the theoretical basis for this paper is predominately derived explicitly from the extant literature, the unprecedented task of attempting to measure the provision of social services at the country level requires a precise delineation of the criteria used to proxy this concept. The criteria below are asserted to be implicit in the extant literature: 1) Government and Markets are failing to deliver essential services to households 2) Breakdown in the scientific link (i.e. effort not yielding expected results) 3) Contempt for other people or groups: the basis for which stems from others benefiting in disproportionate or seemingly unfair amounts. The first criterion expresses a clarification of the literature where the mere provision of social services by radical religious groups is not a sufficient condition they must also surpass the government and market in these services. For instance, if the government or market is willing and able to deliver essential services to households most effectively, then this would hinder the ability of radical religious groups to observe signals of commitment or demonstrate their competency. Therefore, in order for this strategy to be an effective input in the commitment inducement (or identification) process, radical religious groups must become the dominant provider of social services. Before proceeding, it is useful to elaborate on the meaning of the scientific link. Specifically, the scientific link essentially connects current actions with future outcomes on the basis of natural laws. As an example, if a job candidate successfully develops the appropriate skills for a particular position and is objectively the best candidate, then he/she will be offered the job. The second criterion stems from the following assertion: a market and government failing to deliver essential services to households in a country is quite distinct from saying that a country is impoverished. Poverty has proven to be inadequate in explaining the incidence of terrorism, yet there are a variety of factors beyond poverty that could explain the failure of markets and governments. The breakdown in the scientific link constitutes one of these factors. A prime example of this phenomenon is corruption which rewards citizens who ingratiate themselves with people in power. Corruption leads to poor economic outcomes due to a misallocation of time and resources which inhibits markets or governments from providing social services to 9

14 citizens. This creates opportunities for radical religious groups to induce (or identify) commitment. The third criterion is a byproduct of a breakdown in the scientific link. When citizens feel they are being shortchanged in the sense that their effort is not producing the expected results (e.g. corruption), a result is contempt for certain people or groups. Continuing with corruption as an example, the groups who devote their effort and resources to appease those in power are unfairly benefitting at the expense of more deserving citizens (the meaning of fairness and deserving in this context is defined as effort which results in the preservation of the scientific link). As a result, feelings of dissatisfaction will not only be directed towards the agents who allow the breakdown in the scientific link (e.g. governments and markets) but also towards other groups who are disproportionately benefitting from the circumstances. Review of the Empirical Literature Berman and Laitin (2008) link the use of commitment inducement (or identification) technologies to terrorism by appealing to two comparative advantages of radical religious groups: 1) They employ strictness as a means to recruit members who signal high levels of commitment or unfavorable outside options 2) They supply essential services to members and the surrounding population to provide additional incentive not to defect These advantages help foster commitment and thereby reduce defection which lays the groundwork for radical religious groups to produce terrorism. Moreover, these claims are subject to empirical examination where the above hypotheses were conflated and subsequently verified at the organizational level. Using data from Israel and Lebanon, the organizations that provided social services (i.e. strong clubs) were much more lethal in producing terrorist attacks and conducting suicide operations than the deficient providers of social services (i.e. weak clubs). Aside from examinations of the connection between commitment-fostering technologies and terrorism at the organizational level, the empirical literature has neglected to explore this 10

15 link at the country level. Abadie (2004) established a benchmark for examining the incidence of terrorism through various country characteristics where the introduction of the 2003/4 World Market Research Center s Global Terrorism Index (WMRC-GTI) allowed the opportunity. The conclusions exhibit a non-linear relationship between terrorism and political rights in the sense that countries with extremely low levels of political rights may experience a temporary increase in terrorist activity. These findings were also consistent with previous research which concluded economic variables such as education or poverty are inadequate predictors of terrorism (Kreuger and Laitin (2003); Krueger and Maleckova (2004)). Many other extensions to Abadie (2004) have been conceived to explain terrorism at the country level. Goldstein (2005) demonstrated that the use of adult unemployment as a proxy for the economic variable renders it a significant predictor of terrorism (although the magnitude was minimal). Similar to a lack of participation in the government, unemployment leads to a lack of participation in the economy which could inspire more terrorist acts. Moreover, Das et. al (2007) estimated the effects of corruption on terrorism where they conclude that corruption within the private and public sectors generates more terrorist activity in that country. Consistent with previous research, their findings suggest that it is an inability of citizens to shape the political or economic landscape that ultimately leads to increased levels of terrorism. The literature that focuses on country-level data, however, has overlooked the effect of both strictness and social service provision on terrorism. To the best of my knowledge, no one has attempted to proxy the use of these commitment-fostering technologies at the country level and subsequently conduct an empirical investigation for the existence of such a link. This paper attempts to fill in these gaps. Beyond the empirical focus of this paper, an examination of the link between the use of commitment-fostering technologies and terrorism also motivates further study in understanding just how important religion or supernatural motivations are in this process. In an attempt to compare insurgency to suicide attacks, Berman and Laitin (2005) found that the perpetrators of suicide attacks in Israel/Palestine are overwhelmingly directed towards victims with a different religion than theirs. Moreover, suicide attacks are disproportionately linked with radical religious groups. These findings are then interpreted in a manner which downplays the effect of supernatural motivations and instead appeal to the supreme ability of radical religious groups to 11

16 screen out free-riders and inspire collective production. What is less clear, however, is an understanding of why radical religious groups are much more effective at fostering commitment than other groups. Considering that groups who are unaffiliated with religion possess a similar capacity to match the skills of radical religious groups (in the absence of supernatural or theological motivations), the assertion that religion is inconsequential to understanding terrorism raises many questions. Daniel Chen (2010) addresses this issue by attempting to explain religious intensity. By studying the price shocks brought on by the 1998 financial crisis in Indonesia, he was able to conclude that a heightened level of economic distress leads to citizens more frequently engaging in religious activities. In the context of this paper, a financial crisis is an example of a breakdown in the scientific link which compels citizens to turn to religion more frequently. Radical religious groups are more likely to exploit this situation by fostering higher levels of commitment, rendering terrorism a more attractive option. In essence, these results suggest that religiosity may play an important role in linking Social Service Provision to terrorism. Moreover, these findings are consistent with the overt incorporation of a supernatural motive specified in Raynold s theory of commitment inducement, yet they are not consistent with the club model s theory of commitment identification. Although the empirical contribution of this paper does not delineate between the effects of these two theories, it provides a more complete understanding of the explanatory power of both technologies. Thus, the conclusions of this paper motivate future research to empirically investigate the effect of religion on the production of terrorists. Part II Data All of the variables included in the analysis are based on Abadie (2004) with the exception of an alternative (albeit similar) dependent variable, some additional control variables, and (when possible) the use of 2010 measures. Table 1 provides the details and descriptive statistics of each variable. The dependent variable employed by Abadie (2004) was the World Market Research 12

17 Center s Global Terrorism Index (WMRC-GTI) 2003/4 dataset which measured the risk of both transnational and domestic terrorism in 186 countries. This dataset, however, was measured only once in 2003/4 and, to the best of my knowledge, is inaccessible. Consequently, this paper utilizes a comparable and more up-to-date proxy for terrorism: the Institute of Economics and Peace s Global Terrorism Index (IEP-GTI). The IEP-GTI measures the extent to which 158 countries are impacted by both transnational and domestic terrorism for the year The factors within the IEP-GTI score include: the number of incidents, fatalities, injuries, and property damage resulting from terrorism along with a five-year weighted average to account for the lingering psychological damage caused by terrorism. The IEP-GTI score ranges from 0-10 with higher scores reflecting a higher impact of terrorism. In an attempt to investigate the testable implications of the extant theoretical literature at the country level, this paper includes proxies for Sacrifice and Stigma (S&S) and Social Service Provision (SSP). The S&S proxies consist of the Pew Forum s Government Restrictions Index (GRI) and the Social Hostilities Index (SHI) as well as the Failed State Index s Group Grievance (GG) and the Indices of Social development s Intergroup Cohesion (IC) measures. The Government Restrictions Index captures the ways that national and local governments restrict religion whereas the Social Hostilities Index reflects the ways in which private individuals and social groups infringe on religious beliefs and practices. Table 2 provides the entire questionnaire for both the GRI and SHI. In addition, the Group Grievance measure represents the tension and violence existing between groups. Similarly, the Intergroup Cohesion measure is inversely related to the Group Grievance measure where it mirrors the level of cooperation and respect between identity groups in a society. The Social Service Provision variables consist of the Failed State Index s Uneven Economic Development (UED) and Refugees and Internally Displaced Persons (RIDP) measures as well as the Transparency International s Corruption Perceptions Index (CPI). The UED variable mirrors the citizen s devotion to uphold the social contract. The RIDP variable represents the strain on public services brought on by pressures associated with population displacement. The CPI reflects the perceived corruption levels of the public sector. 13

18 The original control variables in Abadie (2004) were updated for the year 2010 except for the geography and fractionalization variables since those measures are neither calculated annually nor expected to significantly change. 14

19 Table 1 - Descriptive Statistics Variable Definition Obs Mean S.D. Min Max Terrorism: Impact of Terrorism IEP Global Terrorism Index Impact of Violence IEP Global Peace Index Economic Variables: GDP per Capita GDP per capita for 2010 in current USD HDI U.N. Human Development Index, Political Freedom: Lack of Political Rights Freedom House's Index of Political Rights, Fractionalization: Linguistic Linguistic Fractionalization Ethnic Ethnic Fractionalization Religious Religious Fractionalization Geography and Climate: Country Area Country area in million square kilometers Elevation Average elevation above sea level in hundred meters Tropical Area Fraction of country area in tropical weather Landlock Fraction of country area beyond 100 km of ice-free coast Social Service Provision: Corruption Corruption Perception Index, Disparities in Gov t Services Uneven Economic Development Index, Population Displacement Refugees and IDPs Index, Sacrifice and Stigma: Religious Demands from Gov t Pew Forum s Index of Government Restrictions involving Religion, 2010 Religious Demands from Society Pew Forum s Index of Social Hostilities involving Religion, 2010 Tension between Groups Group Grievance Index, Cooperation between Groups Intergroup Cohesion Index,

20 Table 2 - Components of the Government Restrictions Index and Social Hostilities Index Government Restrictions Index Does the constitution, or law that functions in the place of a constitution (basic law), specifically provide for "freedom of religion" or include language used in Article 18 of the United Nations Universal Declaration of Human Rights? Does the constitution or basic law include stipulations that appear to qualify or substantially contradict the concept of religious freedom? Taken together, how do the constitution/basic law and other national laws and policies affect religious freedom? Does any level of government interfere with worship or other religious practices? Is public preaching by religious groups limited by any level of government? Is proselytizing limited by any level of government? Is converting from one religion to another limited by any level of government? Is religious literature or broadcasting limited by any level of government? Are foreign missionaries allowed to operate? Is the wearing of religious symbols, such as head coverings for women and facial hair for men, regulated by law or by any level of government? Was there harassment or intimidation of religious groups by any level of government? Did the national government display hostility involving physical violence toward minority or nonapproved religious groups? Where there instances when the national government did not intervene in cases of discrimination or abuses against religious groups? Does the national government have an established organization to regulate or manage religious affairs? Did the national government denounce one or more religious groups by characterizing them as dangerous cults or sects? Does any level of government formally ban any religious group? Were there instances when the national government attempted to eliminate an entire religious group s presence in the country? Did any level of government ask religious groups to register for any reason, including to be eligible for benefits such as tax exemption? Did any level of government use force toward religious groups that resulted in individuals being killed, physically abused, imprisoned, detained or displaced from their homes, or having their personal or religious properties damaged or destroyed? Do some religious groups receive government support or favors, such as funding, official recognition or special access? Does the country's constitution or basic law recognize a favored religion or religions? Do all religious groups receive the same level of government access and privileges? Does any level of government provide funds or other resources to religious groups? Does any level of government provide funds or other resources for religious education programs and/or religious schools? Does any level of government provide funds or other resources for religious property (e.g., buildings, upkeep, repair or land)? Does any level of government provide funds or other resources for religious activities other than education or property? Is religious education required in public schools? Does the national government defer in some way to religious authorities, texts or doctrines on legal issues? Note: Questions obtained from Pew Forum s Rising Tide of Restrictions on Religion Social Hostilities Index Were individuals assaulted or displaced from their homes in retaliation for religious activities, including preaching and other forms of religious expression, considered offensive or threatening to the majority faith? Did individuals face harassment or intimidation motivated by religious hatred or bias? Was property damaged as a result of religious hatred or bias? Were there detentions or abductions motivated by religious hatred or bias? Were individuals displaced from their homes because of religious hatred or bias? Were there physical assaults motivated by religious hatred or bias? Were there deaths motivated by religious hatred or bias? Were there acts of sectarian or communal violence between religious groups? Was there mob violence related to religion? Were religion-related terrorist groups active in the country? Was there a religion-related war or armed conflict in the country? Did violence result from tensions between religious groups? Did organized groups use force or coercion in an attempt to dominate public life with their perspective on religion, including preventing some religious groups from operating in the country? Did religious groups themselves attempt to prevent other religious groups from being able to operate? Did individuals or groups use violence or the threat of violence, including so-called honor killings, to try to enforce religious norms? Were there crimes, malicious acts or violence motivated by religious hatred or bias? Were women harassed for violating religious dress codes? Were there incidents of hostility over proselytizing? Were there incidents of hostility over conversions from one religion to another? 16

21 Justification of Proxies for Sacrifice and Stigma Government Restrictions Index (GRI) Some governments obstruct citizens from adhering to the religion of their choice. The extent to which government officials successfully use their power or influence to prevent citizens from freely selecting into the various religions is captured by the GRI. A high GRI indicates that the government is able to successfully restrict the citizens religious freedoms and a low GRI indicates that the government is either unwilling or unable to restrict the religious freedoms of citizens. Before proceeding, a clarification is useful. When the government restricts freedom of religion, they are essentially making an endorsement to citizens about how they should live their lives. Not only do citizens prefer to avoid punishment from the government but compliance with the restriction becomes relatively less expensive due to network externalities. These outcomes more closely align the government-endorsed behaviors with normal behavior as the citizens have a stronger incentive to comply with the restriction than before. Thus, noncompliance with the restriction as opposed to compliance will face greater tension with the rest of society which renders this action more taboo than before. This is the rationale for using religious norms and religious restrictions interchangeably. The key insight in linking religious restrictions to commitment via Sacrifice and Stigma is to recall from Part 1 that how costly membership into a radical religious sect is will determine the commitment level of members. Considering that the defining feature of a religious sect is to demand noncompliance with religious norms, the costliness of noncompliance reflects the strength of the signal of commitment. For instance, the implementation of a harsher religious restriction (i.e. increase in GRI) is unambiguously costly for members of radical religious sects who choose noncompliance. This provides an opportunity for the leaders of radical religious sects to foster higher levels of commitment by demanding strictness. To illustrate this idea, Table 3 compares noncompliance costs before and after the implementation of the French Burqa Ban that prevents citizens from wearing a burqa in public. 17

22 Table 3 - The Costs associated with Non-compliance to Burqa Ban: Before and After Religious Behavior: Wear a Burqa in public Law: Government Ban on wearing a burqa in public Clothing Preference Costs Before the Law (wear a Burqa) None: citizens are free to choose whether to wear a burqa or not. Thus citizens are allowed to wear a burqa without incurring any costs. After the Law (still wear a Burqa) None: Those that still decide to wear a burqa do so because they preferred not to before the law. Thus, their clothing preference is still unabated by the restriction Network Externality Costs Depends on how universal wearing a burqa in public is prior to the Law. If most people do not wear a burqa, network externality costs are high (since this action is abnormal). If only a few people wear a burqa, social costs are relatively low. Higher than before Law: As it becomes more taboo to wear a burqa in public, the decision to ignore the requirement becomes more costly Punishment Costs None: the government does not punish citizens for wearing a burqa Higher than before Law: As the government punishes citizens for not wearing a burqa, the decision to wear the requirement becomes more costly Table 3 describes the before and after scenarios for a member of a radical religious sect when harsher restrictions on religious freedoms are implemented. The clothing preference costs remain the same as the member will simply ignore the ban and wear a burqa in both scenarios. In addition, those who do not comply with the religious restriction will inevitably be forced to incur higher costs in the form of network externality and punishment costs. Table 3 suggests that an increase in the GRI unambiguously increases the cost of noncompliance with the religious restriction which provides a better opportunity for radical religious groups to effectively induce (or identify) commitment. Moreover, Berman (2000) articulated a similar argument in explaining why Ultra- Orthodox men in Israel, on average, attend classes in yeshiva well beyond the time needed to acquire a draft exemption. The draft deferment policy in Israel reduced the cost of attendance and thereby weakened the commitment signal for Ultra-Orthodox males, compelling members to sacrifice much longer than would otherwise be necessary. The Israeli draft deferment can be thought of as a government subsidy which is inversely related to a religious restriction. For instance, the government policy in France served as a government tax in the sense that it raises the cost of noncompliance with religious norms (i.e. wearing a burqa in public) whereas the Israeli policy reduces the cost of noncompliance with religious norms (i.e. studying full-time in yeshiva). 18

23 Social Hostilities Index (SHI) Similar to government efforts devoted to obstruct the religious freedoms of citizens, societies within a country also possess the capacity to take it upon themselves to hinder citizens from partaking in the religious practices of their choice. How successful private individuals and social groups are in restricting the religious freedoms of citizens are captured with the Social Hostilities Index (SHI). A high SHI score in a country suggests that societies are infringing on the religious beliefs and practices of citizens to a substantial degree whereas a low SHI score suggests the opposite. Although harsher religious restrictions brought on by society are expected to affect the cost of noncompliance with religious norms in the same way as government sanctions, it is important to clarify this link. Specifically, the SHI essentially captures the level of tension between religious groups in a country which suggests that an increase in a country s SHI score threatens the religious practices of the minority more than the religious practices of the majority. For instance, since affiliation to mainstream religious groups provides shelter from oppression or attacks more effectively than minority religious groups (i.e. strength in numbers), there are fewer costs associated with adherence to religious norms. As a result, the costs of membership into a radical religious sect and thereby commitment levels of members will rise which suggests that the same logic articulated in the section on the GRI may be applied to the SHI. Group Grievance (GG) and Intergroup Cohesion (IC) The final two proxies for Sacrifice and Stigma are GG and IC which are similar measures and are inversely related in the sense that they both reflect how well groups in a country co-exist. GG measures the level of tension between groups and IC measures the level of cooperation between groups. Thus, a high GG and low IC score both suggest that groups within a country do not co-exist well which represents the Stigma component of Sacrifice and Stigma. Moreover, when there is tension and a lack of respect between groups, it provides a signal that most of the citizens are highly committed to being a member of a particular group (since it is an expression of allegiance to the group to disrespect another individual merely because they belong to a different group). This is the intended effect of groups requiring sacrifices in the first place which is to invite a stigma associated with such behavior in an attempt to screen out free-riders 19