Policy Research Working Paper 10612 Impacts of Extremist Ideologies on Refugees’ Integration Evidence from Afghan Refugees in Tajikistan Laurent Bossavie Sandra V. Rozo María José Urbina Development Economics A verified reproducibility package for this paper is Development Research Group available at http://reproducibility.worldbank.org, November 2023 click here for direct access. Policy Research Working Paper 10612 Abstract This paper examines the effect of exposure to extremist ide- characteristics, are less integrated into their host communi- ologies on the integration of Afghan refugees in Tajikistan, ties than other refugees. They also show lower educational using a 2023 census following the Taliban’s takeover in levels and more mental health problems. However, there Afghanistan. It finds that Afghan refugees born in prov- is no observed impact of this exposure on their income, inces with increased Taliban territorial control between consumption, or employment. 2017 and 2021, despite having comparable pre-migration This paper is a product of the Development Research Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at sandrarozo@worldbank.org. A verified reproducibility package for this paper is available at http:// reproducibility.worldbank.org, click here for direct access. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Impacts of Extremist Ideologies on Refugees’ Integration: Evidence from Afghan Refugees in Tajikistan* Laurent Bossavie † Sandra V. Rozo‡ Mar´ e Urbina§ ıa Jos´ Keywords: Refugees, Female Education, Extremism JEL Classification: F22, D74, J15, I21, I10 * The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. All errors are our own. The team is grateful to the project “Enhancing Government capacity in hosting temporarily Afghan refugees”, funded under the Rapid Social Response Program (RSR) multi-donor Trust Fund of the World Bank, for funding data collection and allowing us to use the Afghan refugee survey as one of the multiple sources of data for the paper. The purpose of the project and the data it collected was to identify opportunities for refugee integration in Tajikistan. We acknowledge financial support from the Research Support Budget at the World Bank. We also thanks Daniel Garrote for his initial support on instrument design and Merve Demirel for supporting data collection. † World Bank, e-mail: lbossavie@worldbank.org ‡ World Bank, Dev. Research Group, e-mail: sandrarozo@worldbank.org § World Bank, e-mail: murbinaflorez@worldbank.org “The pen is mightier than sword...The extremists are afraid of books and pens, the power of education frightens them.” Malala Yousafzai, UN Speech, July, 2013 I INTRODUCTION How does exposure to extremism affect refugees’ integration? Extremist groups typi- cally control their territories by limiting access to information (Naylor 2008, ICG 2008) and education (HRW 2006, Giustozzi 2010, Giustozzi and Franco 2011), using violence to undermine democratic institutions (Condra et al. 2018) and fostering a sense of distrust towards outsiders (Pourghadiri 2020). Consequently, as conflicts escalates, many refugees are compelled to flee their places of origin in an attempt to escape widespread extremism. However, there is a significant knowledge gap on the impact of exposure to extremism, beyond the migration episode, on the subsequent trajectories of refugees’ lives. Assess- ing these crucial impacts could provide valuable insights to effectively support refugees’ resilience in hosting locations. We examine this question in the context of Afghanistan’s conflict. As many refugees in the world, Afghan refugees left their home country escaping extremist ideologies and regimes. The Taliban, an extremist Islamic group, ruled Afghanistan from 1996 to 2001 imposing strictly the mandates of Sharia law, the Islamic legal system.1 Under their gov- ernment, they issued numerous edicts to control every aspect of Afghan’s life in the public and private spheres (CFR 2020). For example, internet access, entertainment, and music were banned (Monitor 2023Mehran 2022),2 and women were forbidden from receiving secondary or higher education, taking employment, and appearing in public without the appropriate custom and/or a male relative (U.S Department of State 2009). After the September 2001 terrorist attacks, the United States launched the Operation Endur- 1 The sharia, means “the clear path” in Arabic, is derived from a univocal interpretation of the Quran and rules over almost every aspect of the daily life of a Muslim. 2 During this period, the majority of Afghans could only access the internet through telephone lines in neighboring Pakistan. Moreover, when the Taliban took over Kabul in 1996, Afghanistan’s only radio station was ransacked and its library of music was destroyed. 2 ing Freedom, whose objective was to break apart the Taliban and Al Qaeda groups in Afghanistan. The operation encompassed two decades of intense conflict between the U.S. allies and the Taliban militias until 2021, when the U.S. government withdrew its troops from Afghanistan. Seizing this opportunity, the Taliban abruptly regained com- plete territorial control of Afghanistan in 2021. Their rule, once again, restricted access to information, and women’s access to education and labor markets (Amnesty International 2023, United Nations 2023a, United Nations 2023b). To examine the impacts of extremism exposure on refugees’ integration, we utilize as one of our sources of information a census of Afghan refugees living in Tajikistan in 2023.3 Tajikistan has an open border with Afghanistan, and has accommodated nearly 9,532 Afghan refugees and asylum-seekers after the events of 2021 (UNHCR 2023).4 For this purpose, the census interviewed all Afghan refugee households registered with the United Nations Refugee Agency office in Tajikistan.5 Our empirical strategy compares the integration outcomes of refugees who experienced varying exposure to extremism generated by the exogenous and sudden dramatic shift in Taliban’s territorial control in their province of birth between 2017 and 2021. Dur- ing this period, the Taliban’s territorial control abruptly shifted from 19 to 100 percent of the Afghan territory. Specifically, we compare the integration outcomes of refugees living in Tajikistan in 2023, that had similar characteristics before the migration episode, but who were born in provinces with more or less Taliban territorial control between 2017 and 2021. To do this, our main specification controls for a rich set of demographics using retrospective information for the pre-migration period, while refugees still lived 3 The census was collected by the project Enhancing Government capacity in hosting temporarily Afghan refugees with focus on socio-economic resilience and gender issues, under the Rapid Social Response Pro- gram (RSR), multi-donor Trust Fund of the World Bank. The objective of that project was to identify oppor- tunities for refugee integration in Tajikistan. 4 The census was collected with the objective of understanding the profile of Afghan refugees, their current socio-economic outcomes, as well as their needs and potential vulnerabilities to support their inte- gration. 5 After their arrival, refugees are entitled to primary and secondary education, employment on legal basis, healthcare, and residence in Tajikistan until the final consideration of their migration status case. 3 in Afghanistan prior to the Taliban’s take over. They include individual (age, gender, employment status, student status, and number of friends in Tajikistan before migrat- ing) and household demographics (encompassing 34 variables characterizing households size, dwelling characteristics, income, and asset ownership). We explore the effects of extremism on refugee integration which is our primary outcome. We examine impacts on both social and economic integration, using the rich set of out- come variable collected by the census. We also complement this analysis examining the effects of extremism on secondary outcomes including educational attainment, mental health, income and consumption, and labor market-related variables. For each outcome, we estimate effects on individual related variables and also combine them in an index outcome variable. Our data on Taliban’s territorial control between 2017 and 2021 comes from the Foun- dation for Defense of Democracies’ Long War Journal. It encompasses information on the districts controlled by the Taliban, U.S. allies, and contested territories between the two factions. The data set combines information from NATO, press reports, government agencies, and the Taliban itself. From these, we create district-yearly data that reflect the territorial control in Afghanistan from 2017 to 2021, classifying each district into three cat- egories: Taliban-controlled, U.S. allies-controlled, and contested territories between both groups. Using this categorization we estimate the share of the territory controlled by each of the factions by province and year in Afghanistan between 2017 and 2021. We then uti- lize the census data to determine the province of birth for each Afghan refugee living in Tajikistan in 2021 and merge this information to calculate our three treatment including: the average share of province that was Taliban-controlled, U.S. allies-controlled, or con- tested between the two factions for each individual’s province of birth between 2017 and 2021. The intuition behind analyzing these three treatment variables is to carry out a test of 4 consistency. Given the actions taken by the Taliban in Afghanistan, we hypothesize that the effects of exposure to Taliban territorial control should be negative on the outcomes we examine. Moreover, we expect that the impacts of exposure to the territorial con- trol of U.S. allies should be at least neutral and possibly positive in some of the out- comes we study (such as educational attainment) considering that more liberal education policies were established promoting information access and supporting female education (Alvi-Aziz 2008, AREU 2016). Moreover, we also expect to observe detrimental effects of refugees’ exposure to contested territories, as documented extensively in the conflict ´ 2012, Diwakar 2015). Char- literature (Bertoni et al. 2019, Dabalen and Paul 2012, Leon acterizing the size of these effects would be informative on their relevance to refugees’ lives. Our results suggest that refugees’ exposure to Taliban’s territorial control has detrimen- tal consequences for their integration in host communities; these effects are substantial. Our findings indicate that for every one-unit increase in the average share of exposure to the Taliban regime, the integration index of refugees is reduced by 0.56 standard devia- tions. This index is derived from six variables that gauge the refugees’ sense of belonging in their host communities and the number of friends they have in Tajikistan. Notably, individual estimations of the effects of Taliban exposure in each of the six variables con- sistently point to negative impacts across all the studied factors. We also document neg- ative effects of refugees’ exposure to contested territories and positive effects of refugees’ exposure to U.S. allies territorial control. Yet, these effects are not statistically significant. Concerning the effects of Taliban, U.S. allies, and contested territories exposure on the secondary outcomes we are only able to distinguish statistically significant effects of our treatment variables on educational attainment and mental health. Specifically, in terms of educational attainment, an increase in the average share of exposure to Taliban presence by one unit (i.e., Taliban has 100% territorial control in the province of birth of the refugee 5 between 2017 and 2021), results in a decrease of 18.5 percentage points in the likelihood that the refugee has received some level of education. Furthermore, an examination of the heterogeneous effects of Taliban’s exposure by gender reveals that these negative impacts are observed solely among women. The findings align with reductions in the likelihood of being literate and having primary or secondary education, and is also reflected on an index that combines all of these education-related variables. Our analysis reveals, as ex- pected, that exposure to contested territories has negative effects on refugee educational attainment. However, the magnitude of these effects is approximately half the size of the impacts observed for exposure to the Taliban’s territorial control. On the other hand, the effects of exposure to U.S. allies on refugee educational attainment are distinctly positive across all the variables related to educational outcomes that we examined. Lastly, concerning the impacts of Taliban exposure on mental health, our estimates reveal that for every one-unit increase in the average share of exposure to Taliban presence, the likelihood that a refugee reports mental health issues increases by 7.5 percentage points. These effects are large and correspond to an increment of 46 percent of the mean mental health issues observed for the refugees in our sample. The effects are also robust and con- sistently indicate a higher prevalence of mental health problems across all the variables we use as proxies for mental health status. Our main findings on the detrimental impacts of exposure to Taliban territorial control on refugees’ outcomes remain robust across a number of robustness tests. For example, we prove that our results hold after correcting for multiple hypothesis testing and a variety of control-variable restrictions. We excluded variables such as employment and education status while in Afghanistan, and added controls for measurements of social desirability bias. Moreover, we also applied multiple sample restrictions such as excluding from the sample individuals born in provinces where the Taliban had more than 50 percent of territorial control before 2017. Our underlying idea was to clean any endogeneity coming 6 from the location of choice of the migrant’s families prior to the individual’s birth. Despite these tests and adjustments, the results consistently support the validity of our findings. Our results are informative for hosting countries, humanitarian, and development orga- nizations supporting refugees after forced migration has taken place. Our conclusions suggest that refugees who have been exposed to extremism might be in need tailored support to improve their integration in hosting communities. Moreover, they might be disproportionately affected by mental health issues and will require additional guidance to improve female education. Related literature This document contributes to the strands of literature highlighted below. Conflict and integration. There is a growing literature studying the effect of conflict on refugees’ social capital. These studies document that individuals exposed to conflict tend to be more altruistic and prosocial towards their neighbors (Becchetti, Conzo and Romeo 2014, Bowles 2009, Voors et al. 2012, Bauer et al. 2016, Jakiela and Ozier 2019, Cassar, Healy and Von Kessler 2017), more egalitarian towards in-group members (Bauer et al. 2014), and tend to increase their social participation by joining more local social and civic groups or taking on more leadership roles in their communities (Bellows and Miguel 2006, Bellows and Miguel 2009, Gilligan, Pasquale and Samii 2014, Bauer et al. 2016, Blattman and Annan 2010).6 However, few studies examine how refugees’ previous conflict expo- sure can affect their willingness to integrate in a new country, despite the importance of refugees’ integration on their economic assimilation in the host country. We advance this agenda by examining the effects of exposure to extremism, conflict, and liberal ideologies on refugees’ integration sentiments in their host communities. Conflict and education. Multiple studies document that conflict is detrimental for educa- 6 In contrast Alesina and La Ferrara (2002) found that Americans who had a traumatic experience are much less likely to claim they “trust” others. 7 tional attainment (Ichino and Winter-Ebmer 2004, Blattman and Annan 2010, Shemyak- ´ 2012, Justino, Leone and Salardi ina 2011, Shemyakina 2011, Dabalen and Paul 2012, Leon 2014, Verwimp and Bavel 2014, Diwakar 2015, Swee 2015, Diwakar 2015, Singh and She- myakina 2016, Bertoni et al. 2019, Efendic, Kovac and Shapiro 2022). Mechanisms driv- ing these effects include the deterioration of school infrastructure, worse mental health ¨ outcomes (Bruck, Di Maio and Miaari 2019), substitution effects between education and income (Justino, Leone and Salardi 2014), and child recruitment for war (Swee 2015). We add to this literature by analyzing the effects of being exposed to conflict and to extrem- ism ideologies separately. It aids in identifying the persistence of these effects after the individuals moved to a country that is not affected by conflict, where there is no school infrastructure deterioration. Conflict and mental health. Exposure to armed conflict has been widely associated with negative mental health consequences (Boscarino 2006). Studies have documented that individuals exposed to violence show more symptoms of anxiety (Moya 2018), depression (Londono, Romero and Casas 2009), and post-traumatic stress disorder (Yehuda 2002, e et al. 2012), relative to the individuals who have not been exposed to violence.7 Espi´ This paper contributes to this literature by examining the role of extremist ideologies on the mental health of forced migrants. Impacts of forced displacement. This group of studies document that forced migrants usually live in conditions inferior to their pre-migration situation (Ruiz and Vargas-Silva 2013, Ib´ ˜ and Moya 2010).8 An important number of these studies have focused on the im- anez pact of the European forced migration from events related to WWII instead of the promi- aki nent civil wars in the African continent or violence in the Middle East (Matti Sarvim¨ 7 See Mesa-Vieira et al. (2022) for a detailed literature review on the effect of pre-migration exposure to armed conflict on migrant’s mental health. The migrant’s income in the host country and the migrant’s characteristics (Favara et al. 2022, Murthy and Lakshminarayana 2006) are correlated with the prevalence of mental health disorders. Particularly, poor, younger and female migrants have a higher propensity to experience mental health disorders. 8 See Ruiz and Vargas-Silva (2013) for a detailed literature review. 8 2009, Kondylis 2010 Falck, Heblich and Link 2011, Bauer, Braun and Kvasnicka 2013). This paper contributes to this literature by providing evidence of the impacts of extrem- ism for forcibly displaced migrants in a developing country. Afghanistan’s conflict. This literature examines the effects of the insurgent violence on elec- toral outcomes (Condra et al. 2018, Beath, Christia and Enikolopov 2017), economic risk, mobile money, saving preferences (Callen et al. 2014, Blumenstock et al. 2021, Blumen- stock, Callen and Ghani 2018), international and government aid (Child 2019), security transition dynamics (Trebbi and Weese 2019), and the rise of opium production (Lind, Moene and Willumsen 2014). We contribute to this line of work, by examining how ex- posure to extremism and Western values impacts Afghans integration behaviors, income and consumption, labor market variables, educational attainment, and mental health.9 II CONTEXT: AFGHAN REFUGEES IN TAJIKISTAN II.A Afghanistan’s Conflict: 2001-2021 The United States and its allies began their military presence in Afghanistan in October 2001, after the 9/11 terrorist attacks on the United States (CFR 2021). In response to the at- tacks, President George W. Bush launched the combat mission Operation Enduring Free- dom to dismantle Al-Qaeda and remove the Taliban regime from power. Over the next two decades, the U.S. government led a coalition in an effort to stabilize Afghanistan, establish a democratic government, and develop Afghan security forces capable of main- taining stability in the country (Bereiter 2020). As the years passed, the conflict in Afghanistan proved to be a protracted and challenging endeavor. The insurgency led by the Taliban persisted, despite the considerable invest- ment in terms of troops, resources, and financial aid by the allied nations. As a result, in 9 Some other studies in Afghanistan have concentrated on evaluating which methodologies are better to use for program targeting, and for the detection of unobserved coalitions of militants in conflict areas (Aiken et al. 2023, Trebbi and Weese 2019); the role of government’s information campaigns on civilian security cooperation (Aiken et al. 2023), and electoral fraud (Callen and Long 2015), and the effects of differences in electoral systems on voter behavior, political selection, and policy outcomes (Beath et al. 2016) 9 February 2020, the United States and the Taliban signed the Doha Agreement, which out- lined a framework for the withdrawal of U.S. and NATO forces from Afghanistan (CFR 2020). The agreement stated that the U.S. would gradually reduce its military presence, with a complete withdrawal planned by May 2021, conditional upon the Taliban meet- ing certain commitments such as reducing violence and engaging in peace talks with the Afghan government. In line with the Doha Agreement, the U.S. and its NATO allies initiated the process of withdrawing their troops from Afghanistan in May 2021. The withdrawal involved a phased approach, with coalition forces handing over security responsibilities to Afghan forces (CFR 2020). However, as the withdrawal progressed, the Taliban launched offen- sives and rapidly captured territory across the country. By mid-August 2021, the situation on the ground deteriorated significantly, and the Afghan security forces faced numerous challenges. The Taliban’s advances were swift, and they eventually took control of Kabul, the capital city, on August 15, 2021 (CFR 2023). The collapse of the Afghan government occurred much faster than anticipated, and President Ashraf Ghani fled the country. The swift takeover of Afghanistan by the Taliban marked a significant turning point, as they regained the control of the country after being ousted by the U.S.-led coalition over 20 years. The fall of the Afghan government led to a significant humanitarian crisis, with many Afghans fearing reprisals from the Taliban and attempting to flee the country. II.B The Taliban’s Ideology and Policies The Taliban, an Islamist fundamentalist group historically engaged in the pursuit of gov- ernance in Afghanistan, ascended to power in 1996, establishing the Islamic Emirate of Afghanistan (NCTC 2023). During their administration, Mullah Mohammad Omar, as the head of the state, oversaw the implementation of a stringent interpretation of Sharia law. Following the expulsion of the Taliban from power by the United States in 2001, their pri- mary focus shifted towards territorial expansion, involving confrontations with foreign 10 troops and government forces. This expansionist agenda was pursued through tactics such as assassinations, bomb attacks, and other coercive measures aimed at compelling civilian compliance (Clark and Bjelica 2018). Between 2003 and 2005, the Taliban strategically aimed to mobilize support to expand and popularize an insurgency across Afghanistan, progressively gaining control, influ- ence, and support. This calculated effort was designed to erode the connections between communities and the government (Clark and Bjelica 2018). In 2006, the Taliban intro- duced their inaugural code of conduct, characterized by egregious human rights abuses, restrictions on personal freedoms, and the deprivation of fundamental rights to women and minorities (U.S Department of State 2009). Notably, the group’s governance is dis- tinguished by the imposition of a severe interpretation of Islamic law, leading to various restrictive measures, some of the key features of their governance include: Restrictions on Education and Health Care: The Taliban declared teaching in government schools illegal, imposing harsh punishments. Teachers faced warnings, and if deemed necessary, physical beatings. A stringent directive stipulated that those who continued to instruct contrary to Islamic principles could face severe consequences, including lethal measures (Clark 2010). Although education was allowed, it was constrained to mosques or similar institutions, utilizing specific textbooks from the jihad or Emirate era, taught by individuals with religious training. Schools, if deemed necessary, were subject to closure and even destruction. In contrast to education, the Taliban did not view the health service as a political issue, recognizing the utility of clinics to their cause, unlike state schools (Clark and Bjelica 2018). Restrictions on Service Delivery: The Taliban instituted parallel governmental structures, es- tablishing commissions for education, health, agriculture, trade and commerce, financial affairs, and NGOs. These entities sought to co-opt the services of government, NGOs, and private companies at the local level, employing financial influence and control to align 11 them with Islamic ideology (Clark and Bjelica 2018). They appropriated aid resources, imposing their ideological mandates. Restrictions for Women: The Taliban enforced a strict dress code mandating women to cover themselves entirely in public. Women were prohibited from pursuing education beyond a specified age and grade level, with restrictions on subjects such as science, mathematics, and certain social sciences. Additionally, girls were banned from attend- ing schools and universities, and women were barred from most professions, including teaching. Institutions catering to girls faced severe consequences, including closures, de- struction, and harassment. Restrictions on Personal Freedom: The Taliban regulated various facets of daily life, insti- tuting bans on activities such as listening to music, watching television, flying kites, and playing certain sports. Religious Police: The Taliban maintained a dedicated religious police force tasked with enforcing their interpretation of Islamic law. This force played a crucial role in ensuring strict adherence to their moral code. The U.S. presence between 2001 to 2021 played a significant role in improving the life of the Afghan population that resided in districts controlled by the U.S. allies. With the support of international aid and organizations, numerous schools and universities were established or reopened, providing educational opportunities for girls and women (Amnesty 2004). The number of girls attending school increased significantly, and fe- male literacy rates showed improvement. Women also gained access to higher education and pursued various academic disciplines (UNESCO 2023). Additionally, initiatives were launched to train and support female teachers, providing them with the necessary skills and resources to educate girls effectively (PBS 2006). These efforts aimed to empower women through education and promote gender equality in Afghanistan. However, the restrictions persisted in areas where the Taliban kept control or held in- 12 fluence. In rural areas where the population was more conservative, the societal bar- riers limited girls and diminished minorities’ rights. Even more, opposition groups in Afghanistan attempted to frustrate educational development efforts (Narang and Stanton 2017). For example, the Taliban waged an active campaign of violence against educational institutions. The attacks were justified by the arguments that the educational curriculum in schools was influenced by the Western ideology, teachers delivered political lectures, imposed secularization, penetration of Christianity, and failure to enforce the veil (HRW 2006, Jackson and Giustozzi 2012). II.C Tajikistan’s Policies toward Afghan Refugees The Republic of Tajikistan is a signatory of the 1951 Convention relating to the Status of Refugees and its 1967 Protocol, as well as the Convention against torture and other inhuman and degrading forms of treatment and punishment (UNHCR 2021). The coun- try has also enacted a Refugee Law, initially adopted in 1994 and subsequently revised in 2014, which aligns with most international standards. Notably, it includes provisions against penalizing individuals who enter the nation without legal authorization to seek asylum. Moreover, it commits to not returning individuals to a jurisdiction where they would encounter severe harm. However, the present national legislation does impose limitations on the freedom of movement for refugees within Tajikistan. Resolutions 325 and 328, for instance, limit the residency options for refugees, primarily restricting them to areas within the Districts of Republican Subordination, such as Vahdat, Rudaki, Hissor, and Kushoniyon, along with the Jabbor Rasulov district in the Sughd region. It is impor- tant to note that refugees and asylum seekers who arrived after the year 2000 face further restrictions. They are prohibited from settling in Gorno-Badakhshan province and in spe- cific districts, including Varzob, Jirgital, Pianj, Jaikhun, Farkhar, Hamadoni, Shamsiddini Shohin, Shaartuz, Kabodion, Jilikul, as well as several cities, including Dushanbe, Khu- jand, Bokhtar, Tursunzade, Ragun, Levakand, Kulob, Nurek, Buston, Guliston, Taboshar, Konibodom, Isfara, and Bobojon Ghafurov. These resolutions are strictly enforced by the 13 government, and non-compliance can result in the rejection of asylum applications, the de facto revocation of refugee status, detention, and administrative penalties, including deportation. However, the implementation of these policies, aligned with international conventions, has not been consistently applied. Particularly concerning is the Tajik government’s deci- sion to implement a closed border policy in response to the worsening crisis in 2022. This policy restricts access for asylum-seekers with valid Tajik visas. Additionally, attempts to seek asylum through illegal border crossings have become virtually impossible and are subject to penal punishment. These challenges stem from existing contradictions between the Refugee Law and the country’s Criminal Code (UNHCR 2023). In the context of Tajikistan’s humanitarian efforts, Afghani refugees are granted access to primary and secondary education, contingent to a completion of a language program transitioning from their Farsi to Tajik language. Additionally, they are afforded to engage in lawful employment, receive essential healthcare services, secure temporary residency within Tajikistan pending the resolution of their asylum applications, and establish con- nections with organizations dedicated to providing complimentary assistance to refugees and asylum seekers. Furthermore, they receive support in initiating appeals to the Prose- cutor’s Office in cases involving mistreatment or human rights violations, as documented by the United Nations High Commissioner for Refugees in 2021 (UNHCR 2023). How- ever, it is worth noting that there exist certain constraints pertaining to private enterprises, which are obligated to incur substantial financial obligations for the sponsorship of work visas on behalf of refugee beneficiaries. III DATA III.A Data for Afghan Refugees in Tajikistan The primary dataset is a detailed census of Afghan refugees in Tajikistan carried out by the World Bank at the end of 2022, after the Taliban regained complete control of 14 Afghanistan in 2021. The sampling frame for the census is the administrative database of Afghan refugees registered with the United Nations Refugee Agency (UNHCR) in Tajik- istan. All households recorded in the registry were interviewed, leading to a total of 1,958 Afghan refugee households and 9,763 individuals. Figure 2 displays the dates of arrival of the refugees in the census. As shown in the figure, the vast majority of migrants arrived to Tajikistan in 2021 when the Taliban gained complete control of Afghanistan. The questionnaire administered to refugee families was quite comprehensive; it collected detailed information on the socio-economic background, migration history, labor market outcomes and current socio-economic situation of Afghan refugee household and their members, and some measures of socio-economic integration. The approximate duration of the interviews was about two hours. We first use these data to describe the main socioeconomic characteristics of the popu- lation of Afghan refugees in Tajikistan and illustrate some general patterns in Appendix A. As shown in Figure A.1, the refugee population is disproportionately young and in working age. In terms of individual characteristics, the average age of refugees is ap- proximately 25 years—with males being younger. Refugees have literacy rates close to 90 percent overall, but literacy is higher among males than females. Refugees completed 6 years of formal education on average, with about 1 year of difference between males and females (see Figure A.2). We also employ the data to characterize the labor market situation of refugees before and after migrating to Tajikistan in Figure A.3. The data suggests that refugees are less integrated in Tajik labor markets relative to their previous situation in Afghanistan. This is a typical characteristic of forcibly displaced populations as it naturally take time for them to recover their livelihood. Moreover, females have lower labor market participation rates, hours worked, and wages relative to men. We also see that student enrollment for individuals of schooling age is dramatically lower in Tajikistan relative to Afghanistan, 15 but interestingly, there are no gender gaps in this variable for individuals of schooling age. Finally, not surprisingly, household size for refugees is smaller in Tajikistan, which originates from family separation due to forced displacement from Afghanistan. Outcome variables. Using the census data, we construct five index variables for each of the outcomes of interest: refugee integration, educational attainment, mental health, income and consumption, and labor market outcomes. Each index combines answers to the questions related to the outcomes of interest using the methodology outlined by Kling, Liebman and Katz (2007). Appendix B.A describes in detail all the variables included in each of the indexes. In sum, each of the indexes comprises the following: Integration index: it includes four variables that measure the feelings of belonging of refugees in Tajikistan. These are measured using a Likert scale. It also includes two addi- tional variables that record the number of local and refugee friends that refugees have in Tajikistan. Because some variables are continuous the index is constructed following the methodology outlined in Kling, Liebman and Katz (2007). Educational attainment index: average of four dichotomous variables: being literate, having some formal education, completing primary, and completing secondary. Mental health index: average of four dichotomous variables that take the value of one if the individual feels little interest in doing things, had sleeping troubles, had poor appetite or over eating, had suicidal thoughts, or received mental health assistance. Income and consumption index: includes the logarithm of monthly household income and consumption. Because these variables are continuous the index is constructed following the methodology outlined in Kling, Liebman and Katz (2007). Labor market index: includes a dichotomous variable for employment, and the logarithm of monthly wages and weekly hours worked. Descriptive statistics for each of the outcomes and their components variables are de- 16 scribed in Table 1. III.B Data for the Taliban’s Geographical Presence We scraped online data from the Foundation for Defense of Democracies’ Long War Jour- nal on the districts controlled by Taliban, by U.S. Allies, and contested territories be- tween the two groups.10 The data includes a yearly evolution of the territorial control in Afghanistan. The classification is based on open-source information from NATO’s data in Afghanistan, press reports, information provided by government agencies, and the Tal- iban. To categorize the districts, they identify who provides government services, ensures security within the district, administers the district openly, and oversees local courts (See Appendix B.B for a detail description of the variable’s construction).11 Figures 3 to 5 report province-yearly changes in the share of districts controlled by the Taliban, the U.S. allies, and the contested districts in dispute between the Taliban and the U.S. allies.12 As seen in the figures, Taliban control was relatively stable from 2017 through 2019, after which drastic shifts took place across regions, in parallel with the U.S. withdrawal from Afghanistan. IV EMPIRICAL STRATEGY IV.A Measures of Taliban, U.S. Allies, and Contested Territories Exposure For each individual in our sample, we construct a measure of exposure to Taliban (ET ), exposure to U.S. allies (EA), and exposure to contested territories between the two sides of the conflict (EC ) using the data from 2017 until 2021. For this purpose, we employ the district of origin and the location of the Taliban, U.S. allies, and contested territories 10 The Foundation for Defense of Democracies is Washington, DC-based nonpartisan research institute focusing on national security and foreign policy. It aims to strengthen US national security. 11 The LJW data aligns with the territorial control analyses conducted by SIGAR and U.S. Forces- Afghanistan, which effectively tracks Taliban-controlled and contested districts by evaluating 5 indicators of stability, including governance, security, infrastructure, economy, and communications (Roggio 2017). 12 More granular data is illustrated in Figures A.5 and A.6, which depict the district-yearly geographical distribution of Taliban controlled, U.S. allies controlled, and contested districts in Afghanistan. 17 during this period.13 Consequently, we construct ET , EA, and EC as: 2021 ETi = [0.2 × Share of province of birth controlled by Taliban in y]i (1) y =2017 2021 EAi = [0.2 × Share of province of birth controlled by US-Allies in y]i (2) y =2017 2021 ECi = [0.2 × Share of province of birth composed of contested districts in y]i (3) y =2017 Panel A of Table 1 presents descriptive statistics for these variables. The variables were constructed so that they can only take a maximum value of one and a minimum value of zero. For the case of Taliban exposure, for example, ETi can be interpreted as the aver- age share of territorial control that the Taliban had in the province of birth of individual i during the five years between 2017 and 2021. The refugees in our sample have had more exposure to U.S. allies territorial control (mean: 0.60), than Taliban exposure (mean: 0.25), and exposure to contested territories (mean: 0.16). This is intuitive as most refugees are fleeing the Taliban and its policies, hence we expect to have fewer refugees from regions that have been dominated by the Taliban completely throughout the period of study. Fig- ures A.7 and A.9 show histograms for the treatment variables and maps of the location of origin of refugees in our sample. They confirm that there is sufficient variation in the treatment variables as well as on the province of origin in Afghanistan of the refugees surveyed in the census used for this study. The vast majority of changes in Taliban and U.S. allies territorial control occurred during 2020 and 2021 (see Figures 2 and 3). As such, the variation in our measure of exposure comes primarily from these two years. 13 Appendix B.B describes in detail how we construct each of the treatment variables. 18 IV.B Identification Strategy We examine the impacts of Taliban exposure, U.S. allies exposure, and contested territo- ries exposure on refugees’ outcomes using the following specifications: Yij = α0 + α1 ETi + Xi β ′ + Zj Γ + ϵij (4) Yij = γ0 + γ1 ECi + Xi β ′ + Zj Γ + ψij (5) Yij = β0 + β1 EAi + Xi β ′ + Zj Γ + ϕij (6) where Yij represents our outcomes of interest including measures of i) integration (our primary outcome), ii) educational attainment, iii) mental health, iv) income and consump- tion, and v) labor market outcomes. For each outcome, we evaluate the effects on multiple related variables and also combine them in an index. The variables included in each of the indexes are summarized in Panels B, C, and D of Table 1, and described in Appendix B.A.14 By comparing a measure of territorial control from 2017 to 2021 for all the refugees, in the same time period, we isolate endogeneity coming from refugees’ migration date to Tajikistan. Moreover, to account for potential differences in individual characteristics that could bias our results, we control for a rich set of demographics Xi and Zi . Those include variables not affected by the migration shock and collected using retrospective informa- tion on the pre-migration period, while in Afghanistan before the Taliban’s take over. Xi 14 Indexes for dichotomous variables were constructed as the average of all outcomes. Indexes that in- clude continuous variables were constructed standardizing each variable, averaging all variables, and stan- dardizing the average once again. The methodology to construct indexes was adopted from Kling, Liebman and Katz (2007). 19 is a vector of individual demographics which includes age, gender, employment status in Afghanistan, student status in Afghanistan, and number of friends in Tajikistan before migrating. Zi is a vector of household demographics listed in Panel B of Table 2. Those in- clude households size, dwelling characteristics, income, and asset ownership. Our sam- ple of interest is restricted to the forced migrants that only lived in the province where they were born and includes roughly 5,590 individuals. The intuition for restricting the sample to these individuals is to isolate endogeneity coming from migration decisions while in Afghanistan. The next section presents multiple tests to assess the robustness of our results. V IMPACTS OF EXTREMISM ON REFUGEES’ OUTCOMES Our main results are presented in Tables 3 through 7. The reported estimates correspond to the coefficients of interest α1 , β1 , and γ1 of the specifications described in equations (4), (5), and (6). Each coefficient corresponds to an independent regression. In each table column (1) presents the estimated coefficients for the general index constructed based on all the remaining outcomes listed in the horizontal axis. All the tables have three panels. Panel A presents the results for equations (4) examining the effects of exposure to the Taliban presence. Panel B presents the results from estimating equation 5 that assesses the effects of exposure to contested territories. Finally, Panel C presents the results for equation (6) examining the effects of exposure to U.S. allies. V.A Impacts of Exposure to Taliban Territorial Control on Refugees’ Integration Table 3 summarizes the effects of Taliban, contested territories, and U.S. allies exposure on refugees’ integration in Tajikistan. All estimates show strong adverse effects of Taliban exposure on refugee integration. These effects are robust to multiple hypothesis testing, as illustrated by the False Discovery Rate (FDR) q-values reported in brackets (see Panel A). Specifically, the estimates presented in Panel A and column 1 indicate that a one-unit increase in the average share of exposure to Taliban presence (i.e., 100% presence in the 20 province of birth between 2017 and 2021) leads to a reduction of 0.56 standard deviations in the integration index of refugees. We also document mixed results, some of which are not statistically significant, regarding the effects of exposure to contested territories, as reflected in Panel B. The coefficients show both positive and negative values. Interestingly, we find that all coefficients flip sign and become positive when examining the effects of refugee exposure to U.S. allies’ presence. However, it is important to note that these effects are not statistically significant (see Panel C). Robustness and Heterogeneous Effects. Table C.1 includes estimates of the effects of Taliban exposure on refugee integration while incorporating an additional control measuring so- cial desirability, as assessed by Crowne (1964). Our main results remain robust even after including this additional control in the analysis. Moreover, the main results are robust to excluding the provinces that had more than 50 percent of Taliban control in 2017 (Table C.3). The idea behind this exercise is to rule out that the main results effects are driven by the endogenous selection of the migrant’s families in regions with more or less Taliban exposure before 2017.15 Finally, Table D.2 investigates potential heterogeneous effects of Taliban exposure by gender. We do not however find evidence for differential effects by gender groups. V.B Impacts of Exposure to Taliban Territorial Control on Educational Attainment Table 4 reports estimates for the effects of refugees’ exposure to Taliban, contested terri- tories, and U.S. allies on refugees’ educational attainment. First, this reveals that the im- pact of Taliban exposure is negative across four variables we examined: literacy, formal schooling attendance, primary education completion, and secondary education comple- tion. However, it is worth noting that the estimated effects do not reach statistical signif- icance for the completion of primary education. Nevertheless, when we combine these 15 We also carry out the same test for the outcomes of educational attainment and mental health. All of the estimates effects remain unchanged. 21 four variables into an education index (computed as their average), we observe enhanced precision, allowing us to document the significant negative effects of Taliban exposure on the education index. These effects are both substantial and meaningful. For instance, the coefficients in column (3) and Panel A indicate that a one-unit increase in the aver- age share of exposure to Taliban presence (representing 100% presence in the province of birth during the five years between 2017 and 2021) is associated with an 18.5 percentage point decrease in the share of individuals with some level of education. The estimates further demonstrate strong and adverse effects of refugee exposure to con- tested territories on educational attainment. However, one should note that the magni- tude of these effects is approximately half the size of those observed for Taliban exposure. This indicates that while both conflict and extremism exposure influence educational at- tainment negatively, the impact of extremism is significantly more detrimental than that of conflict. Furthermore, we find positive effects of U.S. allies exposure on all the education-related outcomes examined. This finding aligns with the inclusive education policies that the U.S. and its allies supported in the regions under their territorial control. Robustness and Heterogeneous Effects. All the aforementioned effects are robust to mul- tiple hypothesis testing, as evidenced by the False Discovery Rate q-values reported in brackets below each coefficient. As shown in Table C.2, the main results are also robust to excluding controls for student status and employment in Afghanistan, which could be highly correlated with educational outcomes under examination. Additionally, Table D.1 investigates potential heterogeneous effects of Taliban exposure by gender and con- firms that the negative effects are exclusively observed for women. This finding aligns directly with the restrictions imposed by the Taliban on women’s education. Lastly, in D.4, we conducted a replication of the estimates presented in Table 4, but this time, we split the sample into two groups: individuals below and above 25 years of age. The ratio- 22 nale for doing so is to investigate whether the effects are disproportionately observed for individuals of traditional schooling age. The results indicate that the negative effects are concentrated among individuals who are older than 25 years of age. This suggests that this particular group might have been obliged to drop or abandon educational programs in their pursuit to increase educational attainment after they fled Afghanistan (see Table D.4). V.C Impacts of Exposure to Taliban Territorial Control on Mental Health Table 5 provides a summary of the effects of Taliban, contested territories, and U.S. allies exposure on mental health. Larger and positive values of any variable in the table indicate a higher prevalence of mental health issues, while vice versa suggests fewer mental health problems. The results predominantly indicate adverse effects of exposure to Taliban pres- ence on refugee mental health. Importantly, these estimated effects remain robust even after undergoing multiple hypothesis testing. Additionally, the magnitudes of the effects are substantial and meaningful. For instance, the coefficients in column (1) and Panel A indicate that a one-unit increase in the average share of exposure to Taliban presence (representing 100% presence in the province of birth during the five years between 2017 and 2021) is associated with a 7.5 percentage point increase in the share of refugees reporting mental health issues. Moreover, the point estimates also suggest negative effects of refugee exposure to con- tested territories and positive effects of refugee exposure to U.S. allies on mental health. However, it is worth noting that these effects are not statistically significant. Robustness and Heterogeneous Effects. Table D.3 investigates potential heterogeneous effects of Taliban exposure by gender and indicates that there are no mental Health differential effects by gender. 23 V.D Income and Consumption and Labor Market Access Finally, we examine the effects of Taliban exposure, U.S allies exposure and contested territories in labor market-related and income-related outcomes in Tables 6 and 7, re- spectively. We were not able to identify and statistically significant effects on any of the outcomes that we examined. VI DISCUSSION This paper examines the effects of refugees’ exposure to religious extremism on refugees’ integration, educational attainment, mental health, income, and labor markets. For this purpose, we employ data from a census collected in Tajikistan which interviews all the Afghan refugees that fled their country after the Taliban took control of Afghanistan in 2021. Refugees’ exposure to Taliban presence is constructed as an average measure of the share of the territory controlled by the Taliban in the province of origin of each refugee between 2017 and 2021. We also contrast the effects of refugees’ Taliban exposure with that of U.S allies and contested territories exposure. Our findings suggest that refugee exposure to extremism has significant and adverse im- plications for their integration, educational attainment, and mental health. Conversely, exposure to U.S. allies tends to have positive effects on these outcomes, although the ef- fects are generally less precise. With regard to the detrimental effects of Taliban exposure on educational attainment, our expectations are confirmed, as these effects are primarily driven by a reduction in female education attainment. Furthermore, the negative impact is particularly relevant for women above the age of 25 years. Taken together, these results highlight the importance of understanding the differential effects of different types of exposure on refugees’ well-being and educational develop- ment, especially with regards to gender-specific and age-specific considerations. Such in- sights can inform policymakers and humanitarian efforts aimed at supporting refugees’ successful integration and well-being. 24 References Aiken, Emily L., Guadalupe Bedoya, Joshua E. Blumenstock and Aidan Coville. 2023. “Program targeting with machine learning and mobile phone data: Evidence from an anti-poverty intervention in Afghanistan.” Journal of Development Economics 161:103016. Alesina, Alberto and Eliana La Ferrara. 2002. “Who trusts others?” Journal of Public Economics 85(2):207–234. Alvi-Aziz, Hayat. 2008. “A progress report on women’s education in post-Taliban Afghanistan.” International Journal of Lifelong Education 27(2):169–178. Amnesty. 2004. “Women in Afghanistan: The Back Story.”. URL: https://www.amnesty.org.uk/womens-rights-afghanistan-history Amnesty International. 2023. “Afghanistan 2022.” Amnesty International . URL: https://www.amnesty.org/en/location/asia-and-the-pacific/south-asia/afghanistan/report- afghanistan/ AREU. 2016. “The Political Economy Of Education and Health Service Delivery In Afghanistan.” Afghanistan Research and Evaluation Unit . URL: https://www.refworld.org/docid/56b1b2cf4.html a and Joseph Henrich. 2014. “War’s En- Bauer, Michal, Alessandra Cassar, Julie Chytilov´ during Effects on the Development of Egalitarian Motivations and In-Group Biases.” Psychological Science 25(1):47–57. a, Joseph Henrich, Edward Miguel Bauer, Michal, Christopher Blattman, Julie Chytilov´ and Tamar Mitts. 2016. “Can War Foster Cooperation?” Journal of Economic Perspectives 30(3):249–74. Bauer, Thomas K., Sebastian Braun and Michael Kvasnicka. 2013. “The Economic Inte- 25 gration of Forced Migrants: Evidence for Post-War Germany.” The Economic Journal 123(571):998–1024. Beath, Andrew, Fotini Christia, Georgy Egorov and Ruben Enikolopov. 2016. “Elec- toral Rules and Political Selection: Theory and Evidence from a Field Experiment in Afghanistan.” The Review of Economic Studies 83(3):932–968. Beath, Andrew, Fotini Christia and Ruben Enikolopov. 2017. “Direct democracy and re- source allocation: Experimental evidence from Afghanistan.” Journal of Development Economics 124:199–213. Becchetti, Leonardo, Pierluigi Conzo and Alessandro Romeo. 2014. “Violence, trust, and trustworthiness: evidence from a Nairobi slum.” Oxford Economic Papers 66(1):283– 305. Bellows, John and Edward Miguel. 2006. “War and Institutions: New Evidence from Sierra Leone.” American Economic Review 96(2):394–399. Bellows, John and Edward Miguel. 2009. “War and local collective action in Sierra Leone.” Journal of Public Economics 93(11):1144–1157. Bereiter, Gregory. 2020. “Operation Enduring Freedom.” Naval History and Heritage Command . URL: https://www.history.navy.mil/browse-by-topic/wars-conflicts-and-operations/middle- east/operation-enduring-freedom.html ` 2019. “Educa- Bertoni, Eleonora, Michele Di Maio, Vasco Molini and Roberto Nistico. tion is forbidden: The effect of the Boko Haram conflict on education in North-East Nigeria.” Journal of Development Economics 141:102249. Blattman, Christopher and Jeannie Annan. 2010. “The Consequences of Child Soldier- ing.” The Review of Economics and Statistics 92(4):882–898. 26 Blumenstock, Joshua, Michael Callen and Tarek Ghani. 2018. “Why Do Defaults Af- fect Behavior? Experimental Evidence from Afghanistan.” American Economic Review 108(10):2868–2901. Blumenstock, Joshua, Michael Callen, Tarek Ghani and Robert Gonzalez. 2021. “Violence and Financial Decisions: Evidence from Mobile Money in Afghanistan.” The Review of Economics and Statistics pp. 1–45. Boscarino, Joseph A. 2006. “Posttraumatic stress disorder and mortality among US Army veterans 30 years after military service.” Annals of epidemiology 16(4):248–256. Bowles, Samuel. 2009. “Did Warfare Among Ancestral Hunter-Gatherers Affect the Evo- lution of Human Social Behaviors?” Science 324(5932):1293–1298. ¨ Bruck, Tilman, Michele Di Maio and Sami H Miaari. 2019. “Learning The Hard Way: The Effect of Violent Conflict on Student Academic Achievement.” Journal of the European Economic Association 17(5):1502–1537. Callen, Michael and James D. Long. 2015. “Institutional Corruption and Election Fraud: Evidence from a Field Experiment in Afghanistan.” American Economic Review 105(1):354–81. Callen, Michael, Mohammad Isaqzadeh, James D. Long and Charles Sprenger. 2014. “Violence and Risk Preference: Experimental Evidence from Afghanistan.” American Economic Review 104(1):123–48. Cassar, Alessandra, Andrew Healy and Carl Von Kessler. 2017. “Trust, risk, and time preferences after a natural disaster: experimental evidence from Thailand.” World Development 94:90–105. CFR. 2020. “U.S.-Taliban Peace Deal: What to Know.” Council on Foreign Relations . URL: https://www.cfr.org/backgrounder/us-taliban-peace-deal-agreement-afghanistan-war 27 CFR. 2021. “Timeline: The U.S. War in Afghanistan.” Council on Foreign Relations . URL: https://www.cfr.org/timeline/us-war-afghanistan CFR. 2023. “Instability in Afghanistan: Global Conflict Tracker.” Council on Foreign Relations . URL: https://www.cfr.org/global-conflict-tracker/conflict/war-afghanistan Child, Travers Barclay. 2019. “Conflict and counterinsurgency aid: Drawing sectoral dis- tinctions.” Journal of Development Economics 141:102245. Clark, Kate. 2010. “The Layha:The Taleban Codes of Conduct in English.” Afghanistan Analysis Netwrok . URL: https://www.afghanistan-analysts.org/wp-content/uploads/downloads/2012/10/Appendix1 C odei nE nglis Clark, Kate and Jelena Bjelica. 2018. “One Land, Two Rules: Service delivery in insurgent-affected areas, an introduction.” Afghanistan Analysis Netwrok . URL: https://www.afghanistan-analysts.org/en/reports/economy-development- environment/one-land-two-rules-1-service-delivery-in-insurgent-affected-areas-an- introduction/ Condra, Luke N., James D. Long, Andrew C. Shaver and Austin L. Wright. 2018. “The Logic of Insurgent Electoral Violence.” American Economic Review 108(11):3199–3231. Crowne, D. P., Marlowe D. 1964. “The approval motive: Studies in evaluative depen- dence.” New York: Wiley. . Dabalen, Andrew and Saumik Paul. 2012. “Estimating the Effects of Conflict on Education ˆ d’Ivoire.” Journal of Development Studies 50. in Cote Diwakar, Vidya. 2015. “The Effect of Armed Conflict on Education: Evidence from Iraq.” The Journal of Development Studies 51(12):1702–1718. Efendic, Adnan, Dejan Kovac and Jacob N. Shapiro. 2022. “Exposure to Conflict, Mi- 28 grations and Long-run Education and Income Inequality: Evidence from Bosnia and Herzegovina.” Defence and Peace Economics 0(0):1–15. e, Emmanuelle, Val´ Espi´ erie Gaboulaud, Thierry Baubet, German Casas, Yoram Mouchenik, Oliver Yun, Rebecca F Grais and Marie Rose Moro. 2012. “Trauma-related psychological disorders among Palestinian children and adults in Gaza and West Bank, 2005-2008.” Int J Ment Health Syst 3. Falck, Oliver, Stephan Heblich and Susanne Link. 2011. “The Evils of Forced Migration: Do Integration Policies Alleviate Migrants’ Economic Situations?” Discussion Paper, IZA (5829). Favara, Marta, Annina Hittmeyer, Catherine Porter, Saurabh Singhal and Tassew Wold- ehanna. 2022. “Young people, mental health, and civil conflict: Preliminary findings from Ethiopia’s Tigray region.” Psychiatry Research Communications 2(1):100025. URL: https://www.sciencedirect.com/science/article/pii/S277259872200006X Gilligan, Michael J., Benjamin J. Pasquale and Cyrus Samii. 2014. “Civil War and So- cial Cohesion: Lab-in-the-Field Evidence from Nepal.” American Journal of Political Science 58(3):604–619. Giustozzi, Antonio. 2010. “Nation-Building Is Not for All: The Politics of Education in Afghanistan.” Afghanistan Analysts Network Report . Giustozzi, Antonio and Claudio Franco. 2011. “The Battle for the Schools The Taleban and State Education.” Afghanistan Analysts Network Report . HRW. 2006. “Lessons in Terror: Attacks on Education in Afghanistan.” Human Rights Watch 18(6). Ib´ ˜ anez, ıa and Andr´ Ana Mar´ es Moya. 2010. “Vulnerability of Victims of Civil Conflicts: Empirical Evidence for the Displaced Population in Colombia.” World Development 38(4):647–663. 29 ICG. 2008. “Taliban Propaganda: Winning the War of Words?” International Crisis Group . Ichino, Andrea and Rudolf Winter-Ebmer. 2004. “The Long-Run Educational Cost of World War II.” Journal of Labor Economics 22(1):57–87. Jackson, Ashley and Antonio Giustozzi. 2012. “Talking to the Other Side: Humanitarian Engagement with the Taliban in Afghanistan.” Humanitarian Policy Group Working Paper . Jakiela, Pamela and Owen Ozier. 2019. “The Impact of Violence on Individual Risk Pref- erences: Evidence from a Natural Experiment.” The Review of Economics and Statistics 101(3):547–559. Justino, Patricia, Marinella Leone and Paola Salardi. 2014. “Short- and Long-Term Impact of Violence on Education: The Case of Timor Leste.” The World Bank Economic Review 28(2):320–353. Kling, Jeffrey R., Jeffrey B. Liebman and Lawrence F. Katz. 2007. “Experimental Analysis of Neighborhood Effects.” Econometrica 75:83–119. Kondylis, Florence. 2010. “Conflict displacement and labor market outcomes in post-war Bosnia and Herzegovina.” Journal of Development Economics 93(2):235–248. ´ Gianmarco. 2012. “Civil Conflict and Human Capital Accumulation: The Long- Leon, ´ The Journal of Human Resources 47(4):991– term Effects of Political Violence in Peru.” 1022. Lind, Jo Thori, Karl Ove Moene and Fredrik Willumsen. 2014. “Opium for the Masses? Conflict-Induced Narcotics Production in Afghanistan.” The Review of Economics and Statistics 96(5):949–966. an Casas. 2009. “The association between armed Londono, Alicia, Perla Romero and Germ´ 30 conflict, violence and mental health: A cross sectional study comparing two popula- tions in Cundinamarca department, Colombia.” Conflict and health 6. aki, Roope Uusitalo, Markus Jantti. 2009. “Long-Term Effects of Forced Mi- Matti Sarvim¨ gration.” Discussion Paper, IZA (4003). Mehran, Weeda. 2022. “The Evolution in the Taliban’s Media Strategy.” GW Program on Extremist . URL: https://extremism.gwu.edu/evolution-in-taliban-media-strategy Mesa-Vieira, Cristina, Andreas D Haas, Diana Buitrago-Garcia, Zayne M Roa-Diaz, Beat- rice Minder, Magda Gamba, Dante Salvador, Daniel Gomez, Meghann Lewis, Wendy C elie Pahud de Mortanges, Chepkoech Buttia, Taulant Muka, Gonzalez-Jaramillo, Aur´ Natalia Trujillo and Oscar H Franco. 2022. “Mental health of migrants with pre- migration exposure to armed conflict: a systematic review and meta-analysis.” The Lancet Public Health 7(5):469–481. Monitor, Internet. 2023. “Afghanistan.”. URL: https://thenetmonitor.org/country-profiles/afg es. 2018. “Violence, psychological trauma, and risk attitudes: Evidence from Moya, Andr´ victims of violence in Colombia.” Journal of Development Economics 131:15–27. Murthy, Srinivasa and Rashmi Lakshminarayana. 2006. “Mental health consequences of war: a brief review of research findings.” World Psychiatry 5:25–30. Narang, Neil and Jessica A. Stanton. 2017. “A Strategic Logic of Attacking Aid Workers: Evidence from Violence in Afghanistan.” International Studies Quarterly 61(1):38–51. Naylor, Sean. 2008. “Insurgents in Afghanistan Have Mastered Media Manipulation.” Armed ForcesJournal . 31 NCTC. 2023. “Afghan Taliban.”. URL: https://www.dni.gov/nctc/groups/afghant aliban.html PBS. 2006. “Flying Down to Kabul, Women in Afghanistan: Education.”. URL: https://www.pbs.org/wnet/wideangle/uncategorized/women-in-afghanistan- education/2200/ Pourghadiri, Aryan. 2020. “The Roots of Public Distrust to the Afghan Government.” East Studies: Institute for East Strategic Studies . Roggio, Bill. 2017. “SIGAR: ‘Afghan government district control deteriorated’.” FDD’s Long War Journal . URL: https://www.longwarjournal.org/archives/2017/10/sigar-afghan-government-district- control-deteriorated.php Ruiz, Isabel and Carlos Vargas-Silva. 2013. “The Economics of Forced Migration.” The Journal of Development Studies 49(6):772–784. Shemyakina, Olga. 2011. “The effect of armed conflict on accumulation of schooling: Results from Tajikistan.” Journal of Development Economics 95(2):186–200. Singh, Prakarsh and Olga N. Shemyakina. 2016. “Gender-differential effects of terrorism on education: The case of the 1981–1993 Punjab insurgency.” Economics of Education Review 54:185–210. Swee, Eik Leong. 2015. “On war intensity and schooling attainment: The case of Bosnia and Herzegovina.” European Journal of Political Economy 40:158–172. Trebbi, Francesco and Eric Weese. 2019. “Insurgency and Small Wars: Estimation of Un- observed Coalition Structures.” Econometrica 87(2):463–496. UNESCO. 2023. “Let girls and women in Afghanistan learn!”. URL: https://www.unesco.org/en/articles/let-girls-and-women-afghanistan-learn 32 UNHCR. 2021. “Information for asylum seekers on Refugee Status Determination Procedure in the Republic of Tajikistan.” The UN Refugee Agency . URL: https://help.unhcr.org/centralasia/wp-content/uploads/sites/81/2021/08/Brochure-for- Asylum-Seekers-English.pdf UNHCR. 2023. “Tajikistan.” The UN Refugee Agency . URL: https://reporting.unhcr.org/tajikistan-factsheet United Nations. 2023a. “Afghanistan: UN experts say 20 years of progress for women and girls’ rights erased since Taliban takeover.” United Nations Human Rights . URL: https://www.ohchr.org/en/press-releases/2023/03/afghanistan-un-experts-say-20-years- progress-women-and-girls-rights-erased United Nations. 2023b. “Security Council Emphasizes That Punitive Restrictions on Women’s Rights, Escalating Hunger, Insecurity Taking Devastating Toll in Afghanistan.” United Nations: Meetings Coverage and Press Releases . URL: https://press.un.org/en/2023/sc15222.doc.htm U.S Department of State. 2009. “Report on the Taliban’s War Against Women.”. URL: https://2001-2009.state.gov/g/drl/rls/6185.htm Verwimp, Philip and Jan Van Bavel. 2014. “Schooling, Violent Conflict, and Gender in Burundi.” The World Bank Economic Review 28(2):384–411. Voors, Maarten J., Eleonora E. M. Nillesen, Philip Verwimp, Erwin H. Bulte, Robert Lensink and Daan P. Van Soest. 2012. “Violent Conflict and Behavior: A Field Experi- ment in Burundi.” American Economic Review 102(2):941–64. Yehuda, Rachel. 2002. “Post-traumatic stress disorder.” New England journal of medicine 346(2):108–114. 33 Figure 1. Taliban’s Territorial Control in Afghanistan Post-U.S. Invasion - Early 2002 Pre-9/11 - 2000 ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( Kabul ! ( ! ( Kabul ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( Province Status ! ( Province Status ! ( ! ( None or Insignificant ! ( ! ( None or Insignificant ! ( ! ( Taliban Control ! ( Capital Status Minimal Taliban Influence (-25%) ( Taliban Control ! ( Moderate Taliban Influence (+/-50%) Capital Status ! ( ! ! ( ( Nothern Alliance ! ( ! ( ! ( ( Government ! ! Pre-Surge - Early 2009 Post-U.S. Troop Surge - 2012 ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( 34 ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( Kabul ! ( ! ( Kabul ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( Province Status ! ( ! ( ! ( ! ( None or Insignificant ! ( ! ( Minimal Taliban Influence (-25%) Province Status ! ( ! ( ! ( ! ( ! ( ! ( ! ( Moderate Taliban Influence (+/-50%) ! ( None or Insignificant High Taliban Influence (+75%) Minimal Taliban Influence (-25%) ! Capital Status ! Moderate Taliban Influence (+/-50%) ( Contested Capital Status ( ( ! ( ! ( ! ( ! ! ( ( Government ! ( Government ! Withdrawal - June 20, 2021 Endgame - Aug 15, 2021 - Sept 6, 2021 ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( Kabul ! ( ! ( Kabul ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( ! ( Province Status ! ( ! ( ! ( ! ( None or Insignificant ! ( ! ( Minimal Taliban Influence (-25%) Province Status ! ( ! ( ! ( ! ( Taliban Control ! ( ! ( ! ( Moderate Taliban Influence (+/-50%) ! ( High Taliban Influence (+75%) Capital Status ( Taliban Control Capital Status ! ! ! ( Contested ( ( ! ( ! ( ! ( ! ! ( ( Government ! Source: Data comes from the FDD’S Long War Journal Figure 2. Arrival Data of Afghan Refugees in Tajikistan in 2023 1200 Afghans Inflows to Tajikistan 1000 800 600 35 400 200 0 2017m1 2017m7 2018m1 2018m7 2019m1 2019m7 2020m1 2020m7 U.S Withdrawal 2021m1 2021m7 2022m1 2022m7 2023m1 Date Source: Afghanistan Welfare Monitoring Survey conducted for this study in the fall of 2021. Figure 3. Geographical Distribution of Taliban Exposure in Afghanistan Share of Taliban Presence in 2017 Share of Taliban Presence in 2018 0.000 0.000 0.001 - 0.142 0.001 - 0.142 0.143 - 0.190 0.143 - 0.190 0.191 - 0.335 0.191 - 0.335 0.336 - 0.571 0.336 - 0.571 36 Share of Taliban Presence in 2019 Share of Taliban Presence in 2020 0.000 0.000 0.001 - 0.142 0.001 - 0.142 0.143 - 0.190 0.143 - 0.190 0.191 - 0.335 0.191 - 0.335 0.336 - 0.571 0.336 - 0.571 Share of Taliban Presence in July 2021 0.000 0.001 - 0.142 0.143 - 0.190 0.191 - 0.335 Share of Taliban Presence in Aug. 2021 0.336 - 0.909 1 Notes: Taliban exposure is defined as the total number of districts with Taliban presence over the total number of districts in the province. Figure 4. Geographical Distribution of U.S. Allies Exposure in Afghanistan Share of Government Presence in 2017 Share of Government Presence in 2018 0.000 0.071 0.001 - 0.333 0.072 - 0.333 0.334 - 0.500 0.334 - 0.500 0.501 - 0.720 0.501 - 0.720 0.721 - 1.000 0.721 - 1.000 37 Share of Government Presence in 2019 Share of Government Presence in 2020 0.000 0.000 0.001 - 0.333 0.001 - 0.333 0.334 - 0.500 0.334 - 0.500 0.501 - 0.720 0.501 - 0.720 0.721 - 1.000 0.721 - 1.000 Share of Government Presence in July 2021 0.000 - 0.067 Share of Government Presence in Aug. 2021 0.068 - 0.200 0 0.201 - 0.308 0.309 - 0.571 0.572 - 1.000 Notes: Allies exposure is defined as the total number of districts with U.S. presence over the total number of districts in the province. Figure 5. Geographical Distribution of Contested Territories in Afghanistan Share of Contested Provinces in 2017 Share of Contested Provinces in 2018 0.000 0.000 0.001 - 0.150 0.001 - 0.150 0.151 - 0.285 0.151 - 0.285 0.286 - 0.375 0.286 - 0.375 0.376 - 1.000 0.376 - 0.857 38 Share of Contested Provinces in 2019 Share of Contested Provinces in 2020 0.000 0.000 0.001 - 0.150 0.001 - 0.150 0.151 - 0.285 0.151 - 0.285 0.286 - 0.375 0.286 - 0.375 0.376 - 0.867 0.376 - 0.867 Share of Contested Provinces in July 2021 Share of Contested Provinces in Aug. 2021 0.000 0 0.001 - 0.150 0.151 - 0.285 0.286 - 0.375 0.376 - 0.933 Notes: Contested territories is defined as the total number of districts with contested presence over the total number of districts in the province. Table 1. Descriptive Statistics: Treatment and Outcome Variables N Mean SD Min Max Panel A. Treatment Variables Taliban exposure 5,592 0.25 0.08 0.20 0.59 U.S. allies exposure 5,592 0.60 0.27 0.00 0.80 Contested territories exposure 5,592 0.16 0.21 0.00 0.65 Panel B. Outcomes for Integration Integration index (SD) 5,592 0.00 0.93 -3.52 4.36 Feel comfortable if your child socialize with host children (1-5) 5,285 3.97 0.73 1.00 5.00 Feel welcomed in this city (1-5) 5,535 4.17 0.61 1.00 5.00 Feel part of this village (1-5) 5,469 4.06 0.71 1.00 5.00 Feel that Tajikistan is your home (1-5) 5,592 7.53 2.56 1.00 10.00 Number of refugee friends 5,592 4.20 6.83 0.00 100.00 Number of Tajikistan friends 5,589 2.26 3.89 0.00 50.00 Panel C. Outcomes for Educational Attainment Education index 5,592 0.60 0.23 0.00 0.75 Literate [=1] 5,592 0.91 0.29 0.00 1.00 Some education [=1] 5,592 0.85 0.36 0.00 1.00 Primary [=1] 5,592 0.14 0.35 0.00 1.00 39 Secondary [=1] 5,592 0.49 0.50 0.00 1.00 Panel D. Outcomes for Mental Health Mental health issues index 5,590 0.16 0.19 0.00 1.00 Felt little interest in doing things [=1] 5,554 0.30 0.46 0.00 1.00 Had sleeping troubles [=1] 5,570 0.14 0.35 0.00 1.00 Had poor appetite or overeating [=1] 5,573 0.11 0.31 0.00 1.00 Suicidal thoughts [=1] 5,568 0.01 0.10 0.00 1.00 Mental health assistance [=1] 259 0.77 0.42 0.00 1.00 Panel E. Outcomes for Labor Market Access Labor index (SD) 5,592 0.12 1.10 -0.47 2.72 Employed [=1] 5,592 0.22 0.41 0.00 1.00 Monthly wage (somoni) 5,592 357 1,086 0.00 40,000 Weekly hours worked (somoni) 5,592 11.75 23.72 0.00 84.00 Panel F. Outcomes for Income and Consumption Income and consumption index (SD) 5,592 0.01 1.00 -7.03 8.29 Monthly house income (somoni) 5,587 5,159 3,412 0.00 30,850 Monthly consumption (somoni) 5,592 11,151 234,477 0.00 12,000,000 Table 2. Descriptive Statistics: Control Variables N Mean SD Min Max Panel A. Individual Controls (in Afghanistan) Age 5,592 33.65 15.66 14.00 98.00 Male [=1] 5,592 0.49 0.50 0.00 1.00 Student before migrating [=1] 5,592 0.31 0.46 0.00 1.00 Employed [=1] 5,592 0.34 0.47 0.00 1.00 Friends in Tajikistan (before moving) [=1] 5,592 0.44 0.50 0.00 1.00 Panel B. Household Variables (in Afghanistan) Household size 5,592 6.70 2.61 1.00 25.00 Own an enterprise in Afghanistan [=1] 5,592 0.05 0.22 0.00 1.00 Own a dwelling [=1] 5,592 0.53 0.50 0.00 1.00 Dwelling contract: own it [=1] 5,592 0.01 0.09 0.00 1.00 Dwelling contract: pay rent or lease [=1] 5,592 0.98 0.13 0.00 1.00 Dwelling contract: consent with the owner [=1] 5,592 0.01 0.10 0.00 1.00 Dwelling: individual apartment or house [=1] 5,592 0.92 0.27 0.00 1.00 Dwelling: shared apartment or house [=1] 5,592 0.08 0.27 0.00 1.00 Dwelling: temporary shelter, shack or tent [=1] 5,592 0.00 0.04 0.00 1.00 Income from wage labor (Somoni) 5,592 29,628 73,163 0.00 1,555,000 40 Income from self-employment (Somoni) 5,592 16,779 47,019 0.00 800,000 Income from livestock or crops (Somoni) 5,592 1,593 16,205 0.00 300,000 Income from handcrafted products (Somoni) 5,592 317 2,452 0.00 60,000 Income from remittances (Somoni) 5,592 2,711 13,513 0.00 350,000 Income from a loan (Somoni) 5,592 20,353 516,832 0.00 20,000,000 Number of refrigerators in the HH. 5,592 1.06 0.57 0.00 11.00 Number of washing machine in the HH. 5,592 1.03 0.55 0.00 10.00 Number of AC in the HH. 5,592 0.62 1.20 0.00 40.00 Number of fan in the HH. 5,592 1.63 1.25 0.00 10.00 Number of stoves in the HH. 5,592 1.46 1.13 0.00 15.00 Number of bread oven in the HH. 5,592 0.78 0.64 0.00 8.00 Number of gas heaters in the HH. 5,592 0.78 0.93 0.00 19.00 Number of microwaves in the HH. 5,592 0.38 0.59 0.00 8.00 Number of radios in the HH. 5,592 0.48 0.80 0.00 22.00 Number of TVs in the HH. 5,592 1.66 1.00 0.00 15.00 Number of computers in the HH. 5,592 1.07 0.94 0.00 12.00 Number of bicycles in the HH. 5,592 0.80 0.91 0.00 6.00 Number of motorbikes in the HH. 5,592 0.12 0.36 0.00 2.00 Number of cars and other vehicles in the HH. 5,592 0.62 1.06 0.00 44.00 Number of mobile phone in the HH. 5,592 3.12 1.96 0.00 50.00 Sqm plot of land in the HH. 5,592 12.26 387 0.00 20,000.00 Number of livestock or farm equipment in the HH. 5,592 4.37 83.28 0.00 2,000.00 Number of place of business in the HH. 5,592 0.06 0.31 0.00 4.00 Number of business assets in the HH. 5,592 0.07 0.78 0.00 30.00 Table 3. Impacts of Extremism on Refugees’ Integration Feel Comfortable if Feel Welcomed Feel Part Feel that Number Number of Integration Variables in SD. your Child Socialize in this of this Tajikistan is of Refugee Tajikistan Index with Host Children City Village your Home Friends Friends (1) (2) (3) (4) (5) (6) (7) Panel A. Taliban Exposure ETi -0.566*** -0.500*** -0.189* -0.261* -0.366** -0.335** -0.451*** (0.160) (0.174) (0.160) (0.161) (0.164) (0.164) (0.138) FDR q-values [0.001] [0.007] [0.074] [0.056] [0.027] [0.034] [0.004] R-squared 0.027 0.026 0.030 0.024 0.046 0.060 0.037 Panel B. Contested Territories Exposure ECi -0.086 -0.090 -0.113 -0.091 0.034 -0.029 -0.051 (0.062) (0.067) -0.062 (0.063) (0.064) (0.063) (0.054) FDR q-values [0.468] [0.468] [0.468] [0.468] [0.468] [0.468] [0.468] R-squared 0.025 0.025 0.030 0.024 0.045 0.059 0.035 41 Panel C. U.S.- Allies Exposure EAi 0.103 0.098 0.084 0.079 0.012 0.047 0.071 (0.048) (0.052) (0.048) (0.048) (0.049) (0.049) (0.041) FDR q-values [0.167] [0.167] [0.167] [0.167] [0.299] [0.167] [0.167] R-squared 0.025 0.025 0.030 0.024 0.045 0.059 0.036 Mean Dependent Variable 0.002 -0.027 0.026 0.038 0.074 -0.007 -0.099 Observations 5,592 5,285 5,535 5,469 5,592 5,592 5,589 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Yes Notes: The Integration Index is constructed using the outcome variables of columns (ii) to (vii) using the methodology of Kling, Liebman and Katz (2007). It implies i) standardizing the variables, ii) averaging, and iii) standardizing the final average again. See Appendix B for details of variable’s constructions. Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. Table 4. Impacts of Extremism on Educational Attainment Education Some Literate [=1] Primary [=1] Secondary [=1] Index Education [=1] (1) (2) (3) (4) (5) Panel A. Taliban Exposure ETi -0.136*** -0.073* -0.185*** -0.019 -0.267*** (0.037) (0.045) (0.056) (0.059) (0.084) FDR q-values [0.001] [0.056] [0.002] [0.177] [0.002] R-squared 0.192 0.178 0.182 0.053 0.073 Panel B. Contested Territories Exposure ECi -0.058*** -0.034* -0.070*** -0.014 -0.115*** (0.014) (0.017) (0.022) (0.023) (0.032) FDR q-values [0.001] [0.027] [0.002] [0.121] [0.001] R-squared 0.193 0.178 0.182 0.053 0.073 Panel C. U.S. - Allies Exposure EAi 0.047*** 0.027** 0.058*** 0.010 0.092*** (0.011) (0.013) (0.017) (0.018) (0.025) FDR q-values [0.001] [0.023] [0.002] [0131] [0.001] R-squared 0.193 0.178 0.182 0.053 0.074 Mean Dependent Variable 0.599 0.911 0.848 0.141 0.495 Observations 5,592 5,592 5,592 5,592 5,592 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Notes: Dependent variables: Education Index is constructed using the average of the variables in columns (ii) to (v). See Appendix B for details of variable’s constructions. Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Dis- covery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 42 Table 5. Impacts of Extremism on Mental Health Felt Little Ask Mental Health Felt Had Interest Suicidal Mental Variables are indicators Issues Depressed Eating in Doing Thoughts Health Index or Hopeless Problems Things Assistance (1) (2) (3) (4) (5) (6) Panel A. Taliban Exposure ETi 0.075** 0.039 0.099 0.129** 0.009 1.750*** (0.032) (0.079) (0.068) (0.053) (0.018) (0.445) FDR q-values [0.035] [0.286] [0.125] [0.035] [0.286] [0.001] R-squared 0.031 0.025 0.033 0.039 0.031 0.406 Panel B. Contested Territories Exposure ECi 0.019 -0.000 -0.005 0.082*** 0.002 0.476*** (0.012) (0.031) (0.026) (0.020) (0.007) (0.137) FDR q-values [0.185] [0.977] [0.977] [0.001] [0.977] [0.003] R-squared 0.030 0.025 0.032 0.040 0.031 0.397 Panel C. U.S.-Allies Exposure EAi -0.017 -0.002 -0.005 -0.058*** -0.002 -0.399*** (0.010) (0.024) (0.020) (0.016) (0.005) (0.107) FDR q-values [0.102] [0.849] [0.849] [0.001] [0.849] [0.001] R-squared 0.030 0.025 0.032 0.040 0.031 0.402 Mean Dependent Variable 0.159 0.301 0.197 0.106 0.011 0.768 Observations 5,590 5,554 5,578 5,573 5,568 259 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Notes: Dependent variables: Mental Issues Index is constructed using the average of the variables in columns (ii) to (vi). The average is calculated with the number of variables available per individual, the index do not take into account the variables with missing values. See Appendix B for details of variable’s constructions. The Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 43 Table 6. Impacts of Extremism on Labor Market Access Weekly Labor Monthly Employed [=1] Hours Index Wage* Work* (1) (2) (3) (4) Panel A. Taliban Exposure ETi -0.161 -0.053 -0.598 -0.229 (0.175) (0.066) (0.517) (0.305) FDR q-values [0.822] [0.822] [0.822] [0.822] R-squared 0.165 0.156 0.169 0.163 Panel B. Contested Territories Exposure ECi -0.083 -0.027 0.077 0.060 (0.068) (0.026) (0.105) (0.059) FDR q-values [0.697] [0.697] [0.697] [0.697] R-squared 0.166 0.156 0.177 0.176 Panel C. U.S.-Allies Exposure EAi 0.061 0.020 0.238 0.083 (0.052) (0.020) (0.154) (0.091) FDR q-values [0.568] [0.568] [0.568] [0.568] R-squared 0.166 0.156 0.169 0.163 Mean Dependent Variable 0.117 0.219 1.722 1.004 Observations 5,592 5,592 5,592 5,592 Controls in All Panels Individual Covariates Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Notes: Dependent variables: The Labor Market Index is constructed using the outcome variables of columns (ii) to (iv) using the methodology of Kling, Liebman and Katz (2007). It implies i) standardizing the vari- ables, ii) averaging, and iii) standardizing the final average again. Wages and Hours Worked were trans- formed using the inverse hyperbolic sine transformation. See Appendix B for details of variable’s construc- tions. The Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 44 Table 7. Impacts of Extremism on Income and Consumption Income and Monthly Monthly Consumption House Income Consumption Index (log) (log) (1) (2) (3) Panel A. Taliban Exposure ETi -0.176 -0.026 -0.178 (0.165) (0.108) (0.110) FDR q-values [0.460] [0.756] [0.460] R-squared 0.091 0.095 0.073 Panel B. Contested Territories Exposure ECi -0.051 -0.035 -0.011 (0.064) (0.041) (0.043) FDR q-values [1.00] [1.00] [1.00] R-squared 0.090 0.095 0.073 Panel C. U.S.- Allies Exposure EAi 0.050 0.025 0.025 (0.049) (0.032) (0.033) FDR q-values [0.780] [0.780] [0.780] R-squared 0.091 0.095 0.073 Mean Dependent Variable 0.011 8.385 8.326 Observations 5,592 5,453 5,558 Controls in All Panels Individual Covariates Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Notes: The Income and Consumption Index is constructed using the outcome variables of columns (ii) to (iv) using the methodology of Kling, Liebman and Katz (2007). It implies i) standardizing the variables, ii) averaging, and iii) standardizing the final average again. See Appendix B for details of variable’s construc- tions. The Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 45 Contents A Data Description 47 A.A Characterizing Refugees in Tajikistan . . . . . . . . . . . . . . . . . . . . . . . 47 A.B Characterization of Treatment Variables . . . . . . . . . . . . . . . . . . . . . 51 B Description of Variable’s Construction 56 B.A Outcome Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 B.B Treatment Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 B.C Social Desirability Bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 C Robustness Tests 60 D Additional Exercises 65 D.A Impacts by Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 D.B Impacts by Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 46 A Data Description A.A Characterizing Refugees in Tajikistan Figure A.1. Age Distribution 80+ 75-79 70-74 Male 65-69 Female 60-64 55-59 50-54 45-49 40-44 35-39 30-34 25-29 20-24 15-19 10-14 5-9 1-4 <1 600 450 300 150 0 150 300 450 600 Total Number Notes: Source: Afghanistan Welfare Monitoring Survey conducted for this study in the fall of 2021. The Figure includes the 5,592 individuals using in the main regressions. 47 Figure A.2. Refugee Characterization - Individual Characteristics 27 7 26.5 6.60 Years of Education 26.28 6.5 Age 26 6 25.5 25.46 5.83 25 5.5 Female Male Female Male .96 .31 0.95 .94 .3 Student [=1] Literate [=1] .92 0.29 .29 0.29 .9 .88 .28 0.87 .86 .27 Female Male Female Male .49 .48 Married or Engaged [=1] 0.47 .47 .46 0.46 .45 .44 Female Male Notes: The data comes from the Afghanistan Welfare Monitoring Survey conducted for this study in the fall of 2021. The Figure includes the 5,592 individuals using in the main regressions. Red lines represent 95% confidence intervals for the mean. Student is restricted to the population aged between 6 to 24 years old. 48 Figure A.3. Refugee Characteristics before and after moving to Tajikistan .5 1 0.46 Labor Contract in Afghanistan [=1] 0.83 .4 0.38 .8 0.77 Employed [=1] .3 .6 0.55 .2 .4 0.16 0.29 0.10 .1 .2 Female Male Female Male Tajikistan Afghanistan Tajikistan Afghanistan 50 40000 Monthly Wage in Afghanistan (Somoni) Weekly Hours Work in Afghanistan 46.07 30843.08 45 30000 40.02 40 20000 15077.52 35.83 35 10000 31.43 79.70 496.90 30 0 Female Male Female Male Tajikistan Afghanistan Tajikistan Afghanistan .45 7 6.86 0.41 6.77 .4 0.38 Household Size Student [=1] 6.5 .35 .3 0.29 0.29 6.03 6 5.92 .25 Female Male Female Male Tajikistan Afghanistan Tajikistan Afghanistan Notes: The data comes from the Afghanistan Welfare Monitoring Survey conducted for this study in the fall of 2021. The Figure includes the 5,592 individuals using in the main regressions. Red lines represent 95% confidence intervals for the mean. Employed is [=1] if the individual has worked for remuneration during the past 7 days. Student is restricted to the population aged between 6 to 24 years old. 49 Figure A.4. Refugee Literacy Rate by Gender .8 % of Literacy Rate .6 .4 .2 0 79 85 91 97 try 03 09 15 al aw En 19 19 19 19 20 20 20 dr .S ith U .W .S U Year Adult literacy rate female Adult literacy rate male Adult literacy rate 15-24 female Adult literacy rate 15-24 male Source: The data comes from the The United Nations Educational, Scientific and Cultural Organization (UNESCO). 50 A.B Characterization of Treatment Variables Figure A.5. Geographical Distribution of Taliban, Allies, and Contested Districts - Surveyed Sample District Control in 2017 District Control in 2018 Out of the Sample Out of the Sample Taliban Presence Taliban Presence Government Control Government Control Contested Contested 51 District Control in 2019 District Control in 2020 Out of the Sample Out of the Sample Taliban Presence Taliban Presence Government Control Government Control Contested Contested District Control July 2021 Out of the Sample Taliban Presence District Control in Aug. 2021 Government Control Out of the Sample Contested Taliban Presence Notes: Source: FDD’s Long War Journal Figure A.6. Geographical Distribution of Taliban, Allies, and Contested Districts- All Sample District Presence in 2017 District Presence in 2018 Taliban Control Taliban Control Government Control Government Control Contested Contested District Presence in 2019 District Presence in 2020 Taliban Control Taliban Control 52 Government Control Government Control Contested Contested District Presence in 2021 Taliban Control Government Control District Presence in Aug. 2021 Contested Taliban Control Notes: Source: FDD’s Long War Journal Figure A.7. Treatment Variables Distribution Panel A. Taliban Exposure Panel B. Taliban Exposure 3000 600 2500 500 2000 400 Frequency Frequency 1500 300 1000 200 500 100 0 0 .2 .3 .4 .5 .6 .2 .3 .4 .5 .6 Taliban Exposure Taliban Exposure Panel A. Allies Exposure Panel B. Allies Exposure 3000 500 450 2500 400 350 2000 Frequency Frequency 300 1500 250 200 1000 150 100 500 50 0 0 0 .1 .2 .3 .4 .5 .6 .7 .8 0 .1 .2 .3 .4 .5 .6 .7 Allies Exposure Allies Exposure Panel A. Contested Territories Panel B. Contested Territories 3500 350 325 3000 300 275 2500 250 225 Frequency Frequency 2000 200 175 1500 150 125 1000 100 75 500 50 25 0 0 0 .1 .2 .3 .4 .5 .6 .7 0 .1 .2 .3 .4 .5 .6 .7 Contested Territories Contested Territories Notes: Source: FDD’s Long War Journal 53 Figure A.8. Treatment Variables correlation with Conflict Indicators 1 1 U.S. Allies Exposure by Province Taliban Exposure by Province .8 .8 .6 .6 Taliban Exposure U.S. Allies Exposure Fitted Values Fitted Values .4 .4 .2 .2 0 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0 500 1000 1500 2000 2500 3000 3500 Number of Affected and Killed People in Afghanistan Number of Affected and Killed People in Afghanistan Note: Correlation=0.28*** Note: Correlation=-0.33*** 1 1 54 .8 .8 IDPs Province of Origin IDPs Province of Origin U.S. Allies Exposure Taliban Exposure .6 .6 Taliban Exposure U.S. Allies Exposure Fitted Values Fitted Values .4 .4 .2 .2 0 0 0 10 20 30 40 50 60 70 80 90 100 110 120 130 0 10 20 30 40 50 60 70 80 90 100 110 120 130 Number of Individuals Displaced (Thousands) Number of Individuals Displaced (Thousands) Note: Correlation=0.23*** Note: Correlation=-0.26*** Source: United Nations Office for the Coordination of Humanitarian Affairs and United Nations Assistance Mission conflict data Figure A.9. Share of Surveyed Individuals by Province Uzbekistan Tajikistan Turkmenistan Badakhshan Jawzjan Kunduz Balkh Takhar Faryab Samangan Baghlan Sar-e-Pul Panjsher Nooristan Badghis Parwan Kapisa Kunarha Bamyan Laghman Kabul Wardak Nangarhar Herat Ghor Logar Daykundi Paktya 55 Ghazni Khost Urozgan Farah Paktika Pakistan Zabul Helmand Kandahar Iran Nimroz Individuals Surveyed (% Tot. Sample) 0.000 - 0.100 0.101 - 0.300 0.301 - 1.500 1.501 - 3.500 3.501 - 51.788 Notes: Source: Afghanistan Welfare Monitoring Survey conducted for this study in the fall of 2021. The Figure includes the 5,592 individuals using in the main regressions. B Description of Variable’s Construction This section describes how we constructed each of the variables in our main analysis. B.A Outcome Variables Outcome variables were constructed using the data from the Afghanistan Welfare Moni- toring Survey conducted for this study in the fall of 2021. • Integration: we measured the integration dimension with three different types of questions. The first included the individual reported answer from a 1 to 5 scale, where 1 means strongly disagree, and 5 strongly agree of the following questions: (i) ”Would you feel comfortable if your child or grandchild were to socialize or be friends with children of Refugee community people?”, (ii) ”Do you feel welcomed in this village/town/city?”, (iii) ”Do you think that everyone living in this village/ town/ city feels like they are a part of this village?”. The second, included the answer to the following question ”On a scale of 1 to 10, with 1 being not very much to 10 very much, how much do you feel that Tajikistan is your home?”, and the third, the total number of refugee, and Tajik friends among all of their friends, respectively. The Integration Index is constructed using the methodology of Kling, Liebman and Katz (2007) with the variables explained above. It implies i) standardizing the vari- ables, ii) averaging, and iii) standardizing the final average again. • Educational Attainment: we measured the educational attainment dimension using the following questions: (i) ”Can ¡name¿ read and write in any language?”, and (ii) ”What is the highest level of formal school ¡name¿ completed?”. With the answer to the question (ii), we generate an indicator variable corresponding to each level of education: Some Education [=1] if the refugee has completed any level under the primary education level, Primary Education [=1] if the refugee has completed basic primary education level, and Secondary Education [=1] if the refugee has completed secondary education level. The Education index is constructed by calculating the average of the four indicator variables explained above. • Mental Health: we measured the mental health dimension by constructing an in- dicator variable, equal to 1 if the respondent answers ”More than half of the days or Nearly every day” or 0 if the answer is ”Several Days or Not at All”, for each of the following questions: (i) ”Over the last 2 weeks, how often have you felt little interest or pleasure in doing things?” (ii) ”Over the last 2 weeks, how often have 56 you felt down, depressed, or hopeless?” (iii) ”Over the last 2 weeks, how often have you had poor appetite or overeating?” (iv) ”Over the last 2 weeks, how often have you had thoughts that you would be better off dead or of hurting yourself in some way?” and (v) ”Did you seek mental health assistance during the last three months? (1=Yes or 0=No)” The Mental Health index is constructed by calculated the average of the five out- come variables explained above. The average is calculated just with the information available for each individual, it excluded from the average the missing answers. • Labor Market Access: we measured the labor market access dimension using the following variables: (i) Employed is an indicator equal to 1 if the respondent re- ported to have worked for remuneration for at least one hour in the last 7 days, (ii) Weekly Hours Worked: is the total number of reported hours the respondent spent working in the last 7 days, and (iii) Monthly Wage: is the total amount in somoni that the respondent earned from salary or self-employed profits after taxes in the last 30 days. The Labor Market Access Index is constructed using the methodology of Kling, Liebman and Katz (2007) with the variables explained above. It implies i) stan- dardizing the variables, ii) averaging, and iii) standardizing the final average again. • Income and Consumption: we measured the income and consumption dimension using the answer to the following questions: (i) ”Approximately, what was the to- tal household income in the last month? (amount in Somoni)”, and (ii) ”Approxi- mately, how much did your household spend in the last month in total? (amount in Somoni)”. With these two questions, we created the Income and Consumption Index using the methodology of Kling, Liebman and Katz (2007). It implies i) stan- dardizing the variables, ii) averaging, and iii) standardizing the final average again. B.B Treatment Variables Treatment variables were constructed using the online data from the Foundation for De- fense of Democracies’ Long War Journal (LWJ). They map the districts controlled by Tal- iban, by U.S. allies, and contested territories between 2014 to 2021, when the Taliban’ took control over the total territory. The classifications are based on open-source information, such as press reports and information provided by government agencies and the Taliban. With this information, they defined the following: • Taliban and U.S allies controlled districts: places where the Taliban or the U.S. allies 57 are openly administering the territory, providing services and security, and running the local courts, respectively. LWJ may assess a district controlled by the Taliban or by the U.S. government if the district center frequently exchanges hands, and the Taliban or the government only controls a few buildings or villages in the district. • Contested districts: places where the government is in control of the district center or buildings within the district center, or a base, but little else, while the Taliban controls large areas or all of the areas outside of the district center. Or, the Taliban may control several villages, mines and other resources, runs prisons in the district, or administers areas of the district. With the classification of the LWJ, we construct the share of districts controlled by the Tal- iban, the U.S-Allies and the Contested territories in each of 31 provinces in Afghanistan. B.C Social Desirability Bias Individuals usually tend to answer according to how their responses will be viewed by others instead of answering what they really believe, and this phenomenon is known as social desirability bias. For this purpose, we measure social desirability bias by us- ing four questions from Marlone and Crowe’s social desirability scale (see Crowne (1964) for details). The questions assess whether or not respondents are concerned with social approval. A high number of socially desirable responses suggests the respondent is con- cerned with social approval. Each question on the scale includes a statement to which the respondent has to answer true or false. The four questions we included are: ”It is sometimes hard for me to go on with my work if I am not encouraged (false corresponds to higher social desirability)”, ”There have been times when I was quite jealous of the good fortune of others (false corresponds to higher social desirability)”, ”I am always willing to admit when I make a mistake (true corresponds to higher social desirability)”, and ”I have never intensely disliked anyone” (false is associated with higher social desirability)”. Each statement gets a score of zero or one, and the total level is calculated by adding up the scores of the four questions. To facilitate its interpretation, the total value is standardized. 58 59 C Robustness Tests Table C.1. Extremism Impact on Integration: controlling for Social Desirability Index Feel Comfortable Feel Welcomed Feel Part Feel that Number Number of Integration if your Child Variables in SD. in this of this Tajikistan is of Refugee Tajikistan Index Socialize with Host City Village your Home Friends Friends Children (1) (2) (3) (4) (5) (6) (7) Panel A. Taliban Exposure ETi -0.578*** -0.509*** -0.207 -0.277* -0.356** -0.343** -0.450*** (0.160) (0.173) (0.160) (0.160) (0.164) (0.163) (0.138) FDR q-values [0.001] [0.006] [0.060] [0.054] [0.030] [0.030] [0.004] R-squared 0.031 0.032 0.039 0.031 0.049 0.061 0.037 Panel B. Contested Territories Exposure ECi -0.093 -0.096 -0.123 -0.099 0.040 -0.033 -0.050 (0.062) (0.067) (0.062) (0.063) (0.063) (0.063) (0.054) 60 FDR q-values [0.360] [0.360] [0.360] [0.360] [0.360] [0.360] [0.360] R-squared 0.029 0.031 0.039 0.030 0.048 0.060 0.035 Panel C. U.S.- Allies Exposure EAi 0.107 0.103 0.092 0.085 0.008 0.051 0.071 (0.048) (0.052) (0.048) (0.048) (0.049) (0.049) (0.041) FDR q-values [0.137] [0.137] [0.137] [0.137] [0.331] [0.137] [0.137] R-squared 0.030 0.031 0.039 0.031 0.048 0.061 0.036 Mean Dependent Variable 0.002 -0.027 0.026 0.038 0.074 -0.007 -0.099 Observations 5,592 5,285 5,535 5,469 5,592 5,592 5,589 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Yes Notes: See Appendix B for details of variable’s constructions. We used the same controls listed in Table 3. We measure social desirability bias by using four questions from Marlone and Crowe’s social desirability scale. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. Table C.2. Impacts of Extremism on Educational Attainment Excluding Student and Employment Status from Control Variables Education Index Literate [=1] Some Education [=1] Primary [=1] Secondary [=1] (1) (2) (3) (4) (5) Panel A. Taliban Exposure ETi -0.142*** -0.069* -0.182*** -0.033 -0.284*** (0.037) (0.046) (0.058) (0.059) (0.084) FDR q-values [0.001] [0.071] [0.003] [0.196] [0.003] R-squared 0.180 0.150 0.148 0.043 0.068 Panel B. Contested Territories Exposure ECi -0.063*** -0.035** -0.074*** -0.021* -0.124*** (0.014) (0.018) (0.022) (0.023) (0.033) FDR q-values [0.001] [0.024] [0.002] [0.081] [0.001] R-squared 0.181 0.150 0.148 0.043 0.068 Panel C. U.S. - Allies Exposure 61 EAi 0.050*** 0.027** 0.060*** 0.015* 0.099*** (0.011) (0.014) (0.017) (0.018) (0.025) FDR q-values [0.001] [0.024] [0.001] [0.088] [0.001] R-squared 0.181 0.150 0.148 0.043 0.068 Mean Dependent Variable 0.599 0.911 0.848 0.141 0.495 Observations 5,592 5,592 5,592 5,592 5,592 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Notes: Dependent variables: Education Index is constructed using the average of the variables in columns (ii) to (v). Individual controls include: age, gender, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. Table C.3. Impacts of Extremism on Integration Excluding Provinces with more than 50% of Taliban Control in 2017 Feel Comfortable if Feel Welcomed Feel Part Feel that Number Number of Integration Variables in SD. your Child Socialize in this of this Tajikistan is of Refugee Tajikistan Index with Host Children City Village your Home Friends Friends (1) (2) (3) (4) (5) (6) (7) Panel A. Taliban Exposure ETi -0.557*** -0.508** -0.098 -0.281** -0.357** -0.353** -0.457*** (0.169) (0.184) (0.168) (0.169) (0.172) (0.173) (0.145) FDR q-values [0.008] [0.011] [0.127] [0.061] [0.034] [0.034] [0.008] R-squared 0.028 0.027 0.031 0.026 0.046 0.059 0.038 Panel B. Contested Territories Exposure ECi -0.076 -0.089 -0.084 -0.108 0.054 -0.023 -0.042 (0.066) (0.072) (0.066) (0.067) (0.068) (0.068) (0.057) FDR q-values [0.801] [0.801] [0.801] [0.801] [0.801] [0.801] [0.801] 62 R-squared 0.026 0.026 0.032 0.026 0.046 0.058 0.036 Panel C. Allies Exposure EAi 0.098 0.100 0.059 0.092 0.000 0.047 0.068 (0.051) (0.055) (0.051) (0.051) (0.052) (0.053) (0.044) FDR q-values [0.213] [0.213] [0.239] [0.213] [0.529] [0.239] [0.213] R-squared 0.026 0.026 0.031 0.027 0.046 0.058 0.036 Mean Dep. Variable 0.004 -0.026 0.030 0.038 0.075 -0.006 -0.097 Observations 5,383 5,087 5,327 5,262 5,383 5,383 5,380 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Yes Notes: The Integration Index is constructed using the outcome variables of columns (ii) to (vii) using the methodology of Kling, Liebman and Katz (2007). It implies i) standardizing the variables, ii) averaging, and iii) standardizing the final average again. Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. Table C.4. Impacts of Extremism on Educational Attainment Excluding Provinces with more than 50% of Taliban Control in 2017 Education Some Literate [=1] Primary [=1] Secondary [=1] Index Education [=1] (1) (2) (3) (4) (5) Panel A. Taliban Exposure ETi -0.105** -0.070 -0.109* -0.050 -0.192* (0.038) (0.047) (0.058) (0.062) (0.088) FDR q-values [0.031] [0.110] [0.079] [0.198] [0.062] R-squared 0.185 0.172 0.174 0.053 0.071 Panel B. Contested Territories Exposure ECi -0.045** -0.034* -0.033 -0.032 -0.080** (0.015) (0.018) (0.023) (0.024) (0.035) FDR q-values [0.016] [0.068] [0.118] [0.118] [0.042] R-squared 0.185 0.172 0.174 0.053 0.071 Panel C. U.S. - Allies Exposure EAi 0.037** 0.027* 0.030* 0.023 0.066** (0.012) (0.014) (0.018) (0.019) (0.027) FDR q-values [0.011] [0.059] [0.074] [0.101] [0.027] R-squared 0.185 0.172 0.174 0.053 0.072 Mean Dep. Variable 0.602 0.913 0.855 0.140 0.500 Observations 5,383 5,383 5,383 5,383 5,383 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Notes: Dependent variables: Education Index is constructed using the average of the variables in columns (ii) to (v). Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 63 Table C.5. Impacts of Extremism on Mental Health Excluding Provinces with more than 50% of Taliban Control Felt Little Ask Mental Health Felt Had Interest Suicidal Mental Variables are indicators Issues Depressed Eating in Doing Thoughts Health Index or Hopeless Problems Things Assistance (1) (2) (3) (4) (5) (6) Panel A. Taliban Exposure ETi 0.115*** 0.066 0.147** 0.187*** 0.018 1.750*** (0.034) (0.083) (0.072) (0.055) (0.019) (0.445) FDR q-values [0.002] [0.165] [0.031] [0.002] [0.161] [0.001] R-squared 0.034 0.027 0.035 0.041 0.032 0.406 Panel B. Contested Territories Exposure ECi 0.037*** 0.009 0.014 0.112*** 0.006 0.476*** (0.013) (0.032) (0.028) (0.022) (0.007) (0.137) FDR q-values [0.007] [0.655] [0.587] [0.001] [0.490] [0.003] R-squared 0.033 0.027 0.034 0.044 0.032 0.397 Panel C. U.S. - Allies Exposure EAi -0.032*** -0.010 -0.021 -0.082*** -0.005 -0.399*** (0.010) (0.025) (0.022) (0.017) (0.006) (0.107) FDR q-values [0.003] [0.513] [0.300] [0.001] [0.300] [0.001] R-squared 0.033 0.027 0.034 0.043 0.032 0.402 Mean Dep. Variable 0.161 0.301 0.199 0.108 0.011 0.768 Observations 5,381 5,346 5,369 5,364 5,359 259 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Notes: Dependent variables: Mental Issues Index is constructed using the average of the variables in columns (ii) to (vi). The average is calculated with the number of variables available per individual, the index do not take into account the variables with missing values. The Individual controls include: age, gen- der, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. House- hold controls include the variables listed in Table 2. Standard errors are reported in parentheses and False Discovery Rate (FDR) q-values are reported in brackets. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 64 D Additional Exercises D.A Impacts by Gender Table D.1. Heterogeneous Impacts on Educational Attainment Education Some Literate [=1] Primary [=1] Secondary [=1] Index Education [=1] (1) (2) (3) (4) (5) Panel A. Taliban Exposure β1 =ETi × I[Female] -0.205*** -0.359*** -0.420*** 0.074 -0.117 (0.072) (0.088) (0.111) (0.116) (0.165) β2 =ETi -0.029 0.115* 0.034 -0.058 -0.206* (0.053) (0.064) (0.081) (0.085) (0.120) β3 =I[Female] 0.012 0.043* 0.054* -0.031 -0.019 (0.019) (0.023) (0.029) (0.030) (0.043) Diff. Effect=β1 +β3 -0.194*** -0.316*** -0.366*** 0.043 -0.136 (0.055) (0.067) (0.084) (0.088) (0.125) R-squared 0.194 0.180 0.184 0.053 0.073 Panel B. Contested Territories Exposure β1 =ECi × I[Female] -0.059** -0.103*** -0.124*** 0.057 -0.068 (0.028) (0.034) (0.042) (0.044) (0.063) β2 =ECi -0.028 0.018 -0.007 -0.043 -0.080* (0.020) (0.024) (0.031) (0.032) (0.046) β3 =I[Female] -0.030*** -0.029*** -0.030*** -0.022* -0.038** (0.008) (0.009) (0.012) (0.012) (0.017) Diff. Effect=β1 +β3 -0.089*** -0.132*** -0.154*** 0.035 -0.106* (0.024) (0.029) (0.037) (0.039) (0.055) R-squared 0.193 0.179 0.183 0.053 0.074 Panel C. U.S.- Allies Exposure β1 =EAi × I[Female] 0.053** 0.092*** 0.110*** -0.042 0.053 (0.021) (0.026) (0.033) (0.034) (0.049) β2 =EAi 0.019 -0.020 0.002 0.031 0.065* (0.015) (0.019) (0.024) (0.025) (0.035) β3 =I[Female] -0.071*** -0.101*** -0.116*** 0.012 -0.080** (0.014) (0.017) (0.022) (0.023) (0.032) Diff. Effect=β1 +β3 -0.018* -0.008 -0.006 -0.030* -0.027 (0.011) (0.013) (0.016) (0.017) (0.024) R-squared 0.194 0.180 0.184 0.053 0.074 Mean Dep. Variable 0.695 0.981 0.922 0.227 0.651 Observations 5,592 5,592 5,592 5,592 5,592 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Notes: Dependent variables: (i) Education Index is constructed using the average of the variables in columns (ii) to (v). Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 65 Table D.2. Heterogeneous Impacts on Integration Feel Comfortable Feel Feel Feel Number Number if your Child that Integration Welcomed Part of of Variables in SD. Socialize Tajikistan Index in this of this Refugee Tajikistan with Host is your City Village Friends Friends Children Home (1) (2) (3) (4) (5) (6) (7) Panel A. Taliban Exposure β1 =ETi × I[Female] -0.323 -0.487 -0.072 -0.209 -0.014 -0.409 0.050 (0.314) (0.342) (0.315) (0.316) (0.322) (0.321) (0.272) β2 =ETi -0.398* -0.246 -0.151 -0.152 -0.359 -0.121 -0.477** (0.229) (0.249) (0.229) (0.231) (0.235) (0.234) (0.198) β3 =I[Female] 0.060 0.134 -0.040 0.018 -0.045 0.157* -0.013 (0.082) (0.089) (0.082) (0.083) (0.084) (0.084) (0.071) Diff. Effect=β1 +β3 -0.263 -0.354 -0.111 -0.192 -0.059 -0.252 0.037 (0.239) (0.259) (0.239) (0.240) (0.245) (0.244) (0.206) R-squared 0.027 0.027 0.030 0.024 0.046 0.060 0.037 Panel B. Contested Territories Exposure β1 =ECi × I[Female] -0.026 -0.039 -0.011 -0.014 0.127 -0.234* 0.052 (0.120) (0.130) (0.121) (0.122) (0.123) (0.123) (0.104) β2 =ECi -0.073 -0.070 -0.108 -0.084 -0.030 0.091 -0.077 (0.087) (0.095) (0.087) (0.088) (0.089) (0.089) (0.075) β3 =I[Female] -0.016 0.020 -0.056* -0.032 -0.069** 0.092*** -0.009 (0.033) (0.035) (0.033) (0.033) (0.034) (0.034) (0.029) Diff. Effect=β1 +β3 -0.042 -0.019 -0.067 -0.046 0.058 -0.142 0.043 (0.105) (0.114) (0.105) (0.106) (0.108) (0.107) (0.091) R-squared 0.025 0.025 0.030 0.024 0.045 0.059 0.035 Panel C. U.S.-Allies Exposure β1 =EAi × I[Female] 0.044 0.066 0.012 0.025 -0.072 0.176* -0.033 (0.093) (0.100) (0.093) (0.094) (0.095) (0.095) (0.080) β2 =EAi 0.080 0.064 0.078 0.066 0.049 -0.043 0.088 (0.067) (0.073) (0.067) (0.068) (0.069) (0.069) (0.058) β3 =I[Female] -0.047 -0.026 -0.065 -0.049 -0.006 -0.049 0.018 (0.062) (0.067) (0.062) (0.062) (0.063) (0.063) (0.053) Diff. Effect=β1 +β3 -0.002 0.040 -0.053 -0.024 -0.078* 0.126*** -0.014 (0.046) (0.050) (0.046) (0.047) (0.047) (0.047) (0.040) R-squared 0.025 0.025 0.030 0.024 0.045 0.060 0.036 Mean Dep. Variable 0.004 -0.020 0.008 0.000 0.007 0.002 0.016 Observations 5,592 5,285 5,535 5,469 5,592 5,592 5,589 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Yes Household Size in Afgh. Yes Yes Yes Yes Yes Yes Yes Enterprise Property in Afgh. Yes Yes Yes Yes Yes Yes Yes Dwelling Contract in Afgh. Yes Yes Yes Yes Yes Yes Yes Dwelling Type in Afgh. Yes Yes Yes Yes Yes Yes Yes Income Type in Afgh. Yes Yes Yes Yes Yes Yes Yes Number of Assets in Afgh. Yes Yes Yes Yes Yes Yes Yes Notes: The Social Capital Index is constructed using the outcome variables of columns (ii) to (vii) using the methodology of Kling, Liebman and Katz (2007). Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 66 Table D.3. Heterogeneous Impacts on Mental Health Felt Little Ask Mental Health Felt Had Interest Suicidal Mental Variables are indicators Issues Depressed Eating in Doing Thoughts Health Index or Hopeless Problems Things Assistance (1) (2) (3) (4) (5) (6) Panel A. Taliban Exposure β1 =ETi × I[Female] 0.037 0.092 0.069 0.008 -0.018 -0.452 (0.063) (0.155) (0.134) (0.103) (0.035) (0.612) β2 =ETi 0.055 -0.009 0.063 0.125* 0.018 2.036*** (0.046) (0.113) (0.098) (0.075) (0.026) (0.590) β3 =I[Female] 0.005 -0.009 0.009 0.005 0.010 0.179 (0.016) (0.040) (0.035) (0.027) (0.009) (0.155) Diff. Effect=β1 +β3 0.042 0.083 0.079 0.013 -0.008 -0.273 (0.048) (0.118) (0.102) (0.078) (0.027) (0.466) R-squared 0.031 0.025 0.033 0.039 0.031 0.407 Panel B. Contested Territories Exposure β1 =ECi × I[Female] 0.008 0.084 -0.031 0.002 -0.017 -0.136 (0.024) (0.059) (0.051) (0.039) (0.013) (0.233) β2 =ECi 0.015 -0.043 0.010 0.081*** 0.011 0.549*** (0.017) (0.043) (0.037) (0.029) (0.010) (0.186) β3 =I[Female] 0.013* 0.001 0.031** 0.007 0.008** 0.085 (0.007) (0.016) (0.014) (0.011) (0.004) (0.056) Diff. Effect=β1 +β3 0.021 0.085 0.001 0.009 -0.009 -0.050 (0.021) (0.052) (0.045) (0.035) (0.012) (0.209) R-squared 0.030 0.025 0.032 0.040 0.031 0.398 Panel C. U.S.- Allies Exposure β1 =EAi × I[Female] -0.009 -0.057 0.011 -0.003 0.012 0.108 (0.019) (0.046) (0.040) (0.030) (0.010) (0.177) β2 =EAi -0.013 0.027 -0.011 -0.057** -0.008 -0.459*** (0.014) (0.033) (0.029) (0.022) (0.008) (0.146) β3 =I[Female] 0.019 0.048 0.020 0.009 -0.001 0.000 (0.012) (0.031) (0.026) (0.020) (0.007) (0.121) Diff. Effect=β1 +β3 0.011 -0.009 0.031 0.006 0.010** 0.109 (0.009) (0.023) (0.020) (0.015) (0.005) (0.081) R-squared 0.031 0.025 0.032 0.040 0.031 0.403 Mean Dep. Variable 0.165 0.294 0.215 0.112 0.011 0.781 Observations 5,590 5,554 5,578 5,573 5,568 259 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Notes: Dependent variables: (i) Mental Issues Index is constructed using the average of the variables in columns (ii) to (vi). The Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%. 67 D.B Impacts by Age Table D.4. Heterogeneous Impacts on Educational Attainment Education Index Literate [=1] Some Education [=1] Primary [=1] Secondary [=1] Age in 2017 Age≤ 25 Age>25 Age≤ 25 Age>25 Age≤ 25 Age>25 Age≤ 25 Age>25 Age≤ 25 Age>25 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Panel A. Taliban Exposure ETi -0.011 -0.240*** 0.058 -0.193** -0.027 -0.326*** 0.120 -0.121* -0.196 -0.321*** (0.040) (0.059) (0.037) (0.078) (0.062) (0.089) (0.087) (0.072) (0.122) (0.111) R-squared 0.060 0.143 0.043 0.182 0.075 0.216 0.226 0.029 0.070 0.050 Panel B. Contested Territories Exposure ECi -0.012 -0.103*** 0.017 -0.087*** -0.011 -0.140*** 0.062* -0.052* -0.117** -0.134*** (0.015) (0.023) (0.014) (0.031) (0.024) (0.035) (0.033) (0.028) (0.046) (0.044) R-squared 0.060 0.144 0.043 0.182 0.075 0.217 0.227 0.029 0.071 0.051 Panel C. U.S-Allies Exposure 68 EAi 0.009 0.084*** -0.014 0.070*** 0.008 0.114*** -0.046* 0.042* 0.087** 0.108*** (0.012) (0.018) (0.011) (0.024) (0.018) (0.027) (0.025) (0.022) (0.036) (0.034) R-squared 0.060 0.145 0.043 0.183 0.075 0.217 0.227 0.029 0.071 0.051 Observations 2,801 2,791 2,801 2,791 2,801 2,791 2,801 2,791 2,801 2,791 Controls in All Panels Individual Covariates Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Household Size in Afghanistan Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Enterprise Property in Afghanistan Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Dwelling Contract in Afghanistan Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Dwelling Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Income Type in Afghanistan Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Number of Assets in Afghanistan Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Notes: Dependent variables are described in the Appendix D. Individual controls include: age, gender, employed, student status in Afghanistan, and number of friends before moving to Tajikistan. Household controls include the variables listed in Table 2. ∗∗∗ significant at the 1%, ∗∗ significant at the 5%, ∗ significant at the 10%.