Extortion, Civic Action, and Political Participation among Guatemalan Deportees

Abstract How do deported migrants engage in civic and political life after being forcibly returned to their home countries? Do experiences during the migration journey impact how deportees (re)engage? We explore how extortion experienced during migration alters political and civic engagement preferences. We utilize a multi-method approach combining original survey data of Guatemalans deported from the United States and a series of qualitative deportee interviews. We find that extortion during migration has a significant direct effect on increased citizen engagement. Economic hardship exacerbated by extortion may mediate this effect. Overall, extortion experienced while migrating has long-term financial consequences for deportees, with implications for their reintegration and the broader health of civic institutions in their home countries.

Migrating is very hard.I'd tell others not to go, to keep striving here in our country.I'd ask them not to make the decision to migrate because [extortion] happened to me, I lost money, and now things are complicated for me because of the American Dream.
-Deportee 5, Group 2 Hundreds of thousands of people are deported each year from North America and the European Union; over 1.1 million deportations occurred between 2019 and 2021 alone. 1In some countries, deportees comprise a measurable percentage of the population.For example, 271,000 Guatemalans -1.5 per cent of the national populationhave been deported from the United States and Mexico in the last four years. 2 While there is extensive literature on migrants' political behaviour in receiving countries (for example, Dancygier 2017;Hainmueller, Hangartner, and Pietrantuono 2015;Kustov 2021), there is limited knowledge about deportees' political behaviour after returning to their country of origin.
Knowing if and how deportees reinsert themselves into the civic fabric of their home countries is valuable.When home countries have weak governance structures, highly contentious politics, or economic insecurity and violence, deportees' (dis)engagement has the potential to influence national politics.In countries with large migrant outflows and weak institutions, such as Guatemala, Venezuela, Haiti, or Niger, deportee civic (dis)engagement may impact community strength and institutional stability.
Unfortunately, extortion is one salient and common part of the migration journey (Heidbrink 2019;MSF, 2017;Vogt 2013).Extortion amplifies the economic and security hardships commonly experienced among deportees and migrants more generally.Consequently, extortion may have significant downstream consequences for engagement.Many existing studies on the effect of victimization of citizen engagement (for example, Bauer et al. 2016;Ley 2018) are complicated by the fact that adverse economic and security conditions are usually endogenous to an individual and their local circumstances.For example, violence and economic coercion are more likely to occur in places with high crime rates, and local criminal networks may also shape how and to what degree one is willing to take civic action.In contrast, studying the victimization of migrants, which occurs in a geographically distinct location from their civic and political engagement, allows the experience of extortion to be disentangled from potential confounders in home environments.
On the one hand, extortion is an economic shock that exacerbates migrants' financial hardship; such economic conditions may have catalyzed migration in the first place (Abuelafia, Del Carmen, and Ruiz-Arranz 2019;Cohn, Passel, and Gonzales-Barrera 2017;Meyer 2022).This economic shock generates grievances, thereby increasing engagement (for example, Aguilar and Pacek 2000; Rhodes-Purdy, Navarre, and Utych 2021).On the other hand, extortion is a form of victimization that increases fear of crime (for example, Dammert and Malone 2003; Perreira and Ornelas 2013), and fear is demobilizing (for example, Cohn et al. 2015;Druckman and McDermott 2008).Thus, there are divergent expectations about whether extortion during migration decreases or increases deportees' civic and political engagement.
Examining extortion during migration, which occurs quasi-randomly, provides the opportunity to understand how deportees re-engage in civic life and how migration experiences affect this involvement.This project draws on original survey data from Guatemalan deportees who were surveyed immediately upon returning to Guatemala as well as one and six months later.Qualitative interviews were also conducted to gain a more nuanced understanding of the mechanisms at work.Analysis of this data provides evidence of two competing mechanisms linking extortion and civic/political behaviours.The results suggest that extortion increases economic hardship, fear of crime, and citizen engagement.There is also some suggestive evidence that economic hardship mediates the relationship between extortion and increased civic action.These findings demonstrate that extortion experienced while migrating has long-term financial consequences for deportees, which may shape reintegration into their home countries.

Citizen Engagement of Migrants and Deportees
To build expectations for the citizen engagement of deportees, existing research on citizen engagement by migrants is briefly discussed, starting with the definition of a few key terms.'Citizen engagement' encompasses political and civic activities that citizens undertake to influence their government and the broader community (Boulding and Holzner 2021).A 'migrant' is a person who moves away from their residence, whether within or outside of their home country.3'Returnees' are migrants who return to their country of origin, and 'deportees' are a subset of returnees who have been forcibly returned; in this paper, the terms are used interchangeably to refer to deportees.'Extortion' involves using force or threats to obtain something of value.
Literature on migrants and citizen participation can be divided into three strands.First, a range of works consider whether migrants are politically engaged.There is robust evidence that international migrants tend to vote less in receiving countries when compared to natives (for example, Leal 2002;Lim 2023;Verba, Schlozman, and Brady 1995).The same may be true for internal migrants (Gaikwad and Nellis 2021;Rozo and Vargas 2021).Migrants may vote less as a result of rules restricting participation (Bhavnani and Lacina 2018; Dancygier 2017), policies that make registration harder (Gaikwad and Nellis 2021), country of origin policies limiting voter enfranchisement abroad (Wellman 2021), or difficulties surrounding naturalization (Hainmueller, Hangartner, and Pietrantuono 2015).Deportees, however, do not face these constraints, leaving the question of how deportees engage politically open.
Second, research on how settlement patterns affect migrant vote choice suggests that migrants may move to places where other migrants with similar political ideas live, leading to increased left-wing vote shares, cosmopolitanism, and over-representation of home country interests in areas with large migrant populations (Lim 2023;Lueders 2022;Maxwell 2019).Yet, this literature assumes that migrants choose where they live, whereas deportees are forcibly moved back to their country of origin.
Third, more limited work has been conducted on migrant civic engagement.Civic engagement of migrants in their receiving countries may depend on how assimilated they are (Ahmadov and Sasse 2016), whether they have assimilation plans (Leal 2002), or whether they face nativist policy threats (Zepeda-Millán 2017).In some cases, research finds little difference in citizen engagement between migrants and non-migrants (Leal 2002).
Theory: Catalysts of Deportee (Dis)Engagement Deportees are distinct from other migrants on several key dimensions.For example, they do not face host country constraints on voting, and self-selection into specific geographic areas via relocation has been expressly thwarted.For deportees, the location of possible citizen engagement is their home country rather than their receiving country.At the same time, deportees have unique experiences that make them distinct from non-migrants in their home country.
Here, the focus is on the experience of extortion during migration.In the survey of Guatemalan deportees, over 17 per cent of coyote-using deportees reported being extorted while migrating.Extortion amplifies the kinds of economic hardship or trauma experienced more broadly by migrants during their journey.The following sections discuss how the consequences of extortion may affect citizen engagement.Extortion should affect deportees' citizen engagement via two potential mediators: a victimization-caused fear of crime and a grievance-generating economic shock.

Extortion
Many migrants are extorted on their journey to the United States (for example, Heidbrink 2019; MSF, 2017; Vogt 2013), though extortion is widespread in much of Latin America (Dammert 2021; LAPOP 2019) and occurs in a wide range of contexts around the world (for example Chin 2000;Frye 2002;Gambetta 1996;Magaloni et al. 2020).Such criminal extortion can have consequences for citizen engagement.Several scholars have studied the relationship between criminal activity and political engagement (Córdova 2019; Ley 2018; Ley 2022).However, these studies do not focus specifically on extortion but examine the interactions between communities and groups that engage in extortion as one tactic among many.Moncada (2022) studies different strategies communities adopt to resist criminal victimization.By analyzing the repeated interaction between businesses and criminals, he suggests that extortion can lead to violent civic engagement.However, unlike Moncada's work, here the focus is on non-violent citizen engagement after a one-time, geographicallydistinct interaction between individuals and criminals.
How does the extortion experienceseparate from community conditionsshape citizen engagement?The question is particularly important in countries with high volumes of returned migrants who have experienced hardships such as extortion on the migration journey.Citizen (dis)engagement among returnees has implications for the strength of civic and political institutions.
To formulate expectations about extortion's impact on deportee behaviour, the next section turns to broader theories tying two elements of the extortion experience to citizen engagementvictimization and economic hardship.

Victimization, Fear, and Citizen Disengagement
Although there is little research on how migrant exposure to violence affects political participation, there is a wide range of research (with mixed findings) on the relationship between victimization and political participation more broadly.This literature considers violence or the threat of violence in the context of crime and armed conflict.Some research suggests that people exposed to violence will be more politically engaged (for example, Bateson 2012;Bauer et al. 2016;Sønderskov et al. 2022).Scholars have suggested a range of explanations, including individual growth and activation following trauma (Blattman 2009), anger (Ditton et al. 1999), and the social affirmation of in-group membership (Dorff 2017;Schuessler 2000).Conversely, other work suggests that victimized individuals lose faith in government institutions and withdraw from public life (Collier and Vicente 2014;Coupé and Obrizan 2016;Ley 2018).
Many mechanisms that may explain the relationship between victimization and political engagement generally do not seem to apply to deportees.For example, it is unclear why the victimization that deportees experience while migrating should cause them to lose faith in the government of the country that they had already left.Post-traumatic growth requires major life crises that rupture people's assumptions about their world, but extortion is only one trauma among many that migrants experience.In terms of arguments grounded in political participation as an affirmation of in-group identity, it is not clear what identity deportees could affirm unless they left their home country because of persecution based on their identity.Deportees do not typically know each other and often arrive in small groups.Thus, two other mechanisms tying extortion to citizen engagement are posited here: fear of crime and economic grievance.
A range of non-political outcomes evidences fear and lingering trauma from migration victimization.Experiencing abuse while migrating is predictive of depression and alcohol dependency (Altman et al. 2018).In one study, among those individuals victimized during the migration process, such as being robbed or attacked, 21 per cent are at risk of post-traumatic stress disorder (PTSD) (Perreira and Ornelas 2013).This is true not only for Central American migrants but also for global migrant populations.For example, one meta-analysis of 113 articles confirms that exposure to violence during migration affects mental health; the most frequent consequences include PTSD (Kirmayer et al. 2011).PTSD is closely tied to anxiety (Torres 2020).
It is theorized here that victimization from extortion correlates with increased fear of repeated crime.We follow Gabriel and Greve (2003) in defining fear of crime as a 'disposition (trait) [which] describes my tendency to experience fear of crime in certain situations' (p.601).As such, fear of crime varies across individuals.Studies in diverse contexts have shown a correlation between victimization and fear of crime (Dammert and Malone 2003;Singer et al. 2019).In a non-medical context, one way to evaluate respondents' fear of crime is to consider whether they avoid everyday activities out of fear.
According to appraisal models, emotion, judgement, decision-making, fear, and sadness correlate with pessimistic risk estimates and, thus, risk aversion (Keltner, Ellsworth, and Edwards 1993;Lerner et al. 2015).In particular, fear leads to lowered risk tolerance and behavioural avoidance among a wide range of individuals (for example Campos-Vazquez and Cuilty 2014; Druckman and McDermott 2008;Guiso, Sapienza, and Zingales 2018).Thus, we hypothesize that those deportees who were extorted during the migration process will suffer from higher levels of fear of crime than other deportees; this fear will predict citizen disengagement.

Economic Hardship, Grievance, and Citizen Engagement
While extortion has psychological effects, it also has economic consequences.Like the literature on victimization, research into the relationship between economic shocks, limited economic resources, and political engagement is inconclusive.On the one hand, people with fewer resources are less able to engage in politics.Yet, negative income shocks may prompt grievances against the government, leading to increased political engagement.
Those who face economic insecurity or grew up economically disadvantaged are less likely to participate in politics than those who are more socioeconomically prosperous (Blais 2006;Ojeda 2018;Schlozman, Verba, and Brady 2013;Smets and van Ham 2013). 4At the individual level, the resource model of civic engagement provides one explanation: time, money, and civic skills provide the resources required to engage in politics (Brady, Verba, and Schlozman 1995).Because deportees who have been extorted are likely to be less socioeconomically prosperous than deportees who have not, this model suggests that victims of extortion will be less likely to participate in politics.
However, work focused on grievance suggests that people experiencing negative economic shocks may be more motivated to participate politically.For example, increased unemployment rates in the United States are correlated with higher turnout (Burden and Wichowsky 2014;Cebula 2017).These effects may vary across race and ethnic groups (Huyser et al. 2018) and regions (Boulding and Holzner 2021).Aguilar and Pacek (2000) similarly argue that macroeconomic downturns in developing nations increase turnout, particularly among the most affected: lower-status voters.In a test of the grievance mechanism, Rhodes-Purdy, Navarre, and Utych (2021) find that economic crises prompt anger. 5urer et al. summarizes the distinction between these two literatures thus: 'structural economic disadvantage unambiguously demobilises individuals, [whereas] the deterioration of economic prospects instead increases political activity ' (2019, 866).Given this distinction, the literature on negative economic shocks better captures the situations extorted deportees face.Deportees are, on average, socioeconomically disadvantaged, regardless of whether or not they have been extorted.They frequently left their home countries for better jobs and then were forcibly deported.Extorted deportees, however, have suffered from an additional and unexpected deterioration of their economic prospects.This economic shock will increase citizen engagement.

Study Expectations
In summary, extortion should affect deportees' political and civic engagement.This effect may work via two distinctand counteractingmediators.Extortion is a form of victimization that depresses engagement by increasing fear of crime.At the same time, extortion is a grievance-generating economic shock that increases citizen engagement.These arguments can be formally hypothesized as follows: • Hypothesis 1a (Extortion to Increased Fear of Crime): Deportees extorted while migrating are more likely to fear crime compared to deportees who were not extorted.
• Hypothesis 1b (Extortion to Increased Economic Hardship): Deportees extorted while migrating are more likely to experience economic hardship than those who were not extorted.
• Hypothesis 2a (Fear of Crime Mediator): Extortion's positive effect on fear of crime will lead to lower citizen engagement.
• Hypothesis 2b (Economic Hardship Mediator): Extortion's positive effect on economic hardship will lead to higher citizen engagement.
If fear and economic hardship mediate the relationship between extortion and political engagement, a final hypothesis should concern which mechanism plays a larger role.However, there is no prior reason to believe that one mechanism is more or less important than another.Thus, this issue is explored empirically.

Migration from and Deportation to Guatemala
Given the prevalence of deportees in the population and the nature of the country's institutions, Guatemala is an important context to study these expectations.Between 2012 and 2021, Guatemala received approximately 379,000 deportees from the United States, which is 8 per cent of the total deportations (DHS 2021).Hundreds of thousands more were deported from Mexico. 6This large number of deportations makes it crucial to understand how and when deported migrants are able to reintegrate into their 'home' societies and political systems, often after years abroad.To that end, it is valuable to provide an overview of the conditions in Guatemala that contribute to migration, the experience of extortion during migration, and the experiences of deportees upon their return to Guatemala.
Qualitative and policy studies suggest that a range of macro-level conditions in Guatemala have contributed to recent emigration to the United States: socioeconomic difficulties (for example, Abuelafia, Del Carmen, and Ruiz-Arranz 2019; Cheatham 2019; Meyer 2022); violence associated with transnational organized crime (for example, Bermeo 2018; Cheatham 2019; MSF 2017); and corruption (for example, Cheatham 2019; Meyer 2022).As a result of these push factors, between 2012 and 2021, the United States Border Patrol apprehended over 1 million Guatemalan migrants (DHS 2021).
Many migrants from Central America suffer extortion, assault, kidnapping, and rape during the journey (for example, Abuelafia, Del Carmen, and Ruiz-Arranz 2019; Infante et al. 2012;Slack, Martínez, and Whiteford 2018).A cross-sectional study of over twelve thousand migrants in transit from Mexico to the United States suggests that nearly a third of migrants from Central America report experiencing physical, psychological, and sexual violence during the journey (Leyva-Flores et al. 2019).
The original data, described in the Research Design section below, suggests that the median deportee spent a year and a half in the United States before their forced return to Guatemala.Over 80 per cent of deportees left at least one family member in the United States, and many parted with significant savings. 7Having left behind family and assets, deportees arrive via plane to Guatemala City, where only 5 per cent of the deportees are originally from.Most migrants then return to towns of their birthplace; 80 per cent of deportees report being in locations with five or more family members one month after their arrival (see Fig. A4 for departments of origin).Deportees face a range of challenges in the locations where they settle.Over 80 per cent of respondents indicate that their community has no employment opportunities, and more than 50 per cent say there is either some or a lot of gang activity.Deportees also suffer from police harassment (Fig. A1).When deportees were asked about the degree to which the need to pay outstanding debts posed a challenge to reintegration, the average response was 7 (on a 10-point scale).8Thirty-seven per cent of deportees in the original sample said they intended to return to the United States in the next year.See Dow et al. (2021) and Table A3 for more details.
It also is important to briefly describe the strength and nature of the political institutions deportees encounter upon their return to their country of originparticularly given that weak institutions often contribute to the conditions that drive migration in the first place.Guatemala's party system has a low level of institutionalization (Mainwaring 2018;Sánchez 2008), and the military plays a prominent role in politics (Isaacs and Schwartz 2020).There is evidence of vote-buying, violence against voters (Gonzalez-Ocantos et al. 2020), and corruption (CICIG 2019;Trejo and Nieto-Matiz 2022).In the 2019 presidential election, only 42 per cent of the country's registered voters turned out to vote (Cuffe 2019).
Even if they do not participate in organized national politics, many Guatemalans engage in the civic life of their community.For example, civil watches and community patrols play an influential, if often harmful, role in the country (Bateson 2017;CICIG 2019).Indigenous governance structures are important, too (Hawkins, McDonald, and Adams 2013;Sieder 2020).
In terms of generalizability, Guatemala has one of the highest volumes of deportees globally.Guatemala's Electoral Democracy Index (EDI) is very close to the average for all countries receiving deportees from the United States and the European Union.Thus, in terms of democratic engagement, the results could plausibly travel to other similar cases.At the same time, Guatemala's Party Institutionalization Index (PII) is below average for countries that receive the most deportees, but Guatemala is not an extreme case.Countries like Afghanistan, Haiti, the Philippines, and Egypt also have similar or worse levels of party institutionalization (see Fig. A3 and related discussion). 9Thus, the results for political engagement specifically may generalize more directly to other states with weak party systems.

Deportee Survey
This project uses data from an original survey of migrants recently deported from the United States and returned to Guatemala. 10Beginning in October 2019, RTI International and Te Conecta, a Guatemalan NGO, partnered to implement a face-to-face survey of newly arrived deportees at the Guatemalan Air Force base in Guatemala City.This airport is the main arrival point for deportees sent to Guatemala and typically receives three to five planes of deportees four to five days per week.The first stage of the survey was implemented upon returnee arrival and was conducted from October 2019 to March 2020.After that, COVID-19 made in-person data collection impossible; follow-up surveys were conducted by phone.To recruit respondents initially, the survey team greeted deportees after they had been processed, as they were leaving the airport.Enumerators were instructed to randomly select the people to approach with information about the study: their choice was not to be based on observable characteristics.In practice, this meant selecting every third deportee (approximately) and providing them with a voucher to take to a nearby office to participate in the survey.Respondents were offered fifty Quetzales, equivalent to (about) $6.50 US, to participate in the first survey round.Out of the returnees that received vouchers in a given week, approximately 12 per cent completed a survey.In total, 1,357 deportees were interviewed upon their arrival to the country.
This recruitment procedure produced data with average demographics similar to official U.S. Immigration and Customs Enforcement (ICE) data on deportees tracked by Syracuse University's Transactional Records Access Clearing House (TRAC, 2023).For example, in Financial Year (FY) 2019, roughly 11 per cent of the Guatemalans deported were women: 8 per cent of the sample were women.Further, most deportations of Guatemalans departed from Texas (50 per cent) and Arizona (26 per cent); the sample reported these states as the ones they spent the most 9 The data used in this figure comes from the V-Dem project (Bizzarro et al., 2017;Coppedge et al., 2017), the U.S. Government (ICE, 2020), and the European Union (Eurostat, 2023).time in while in the United States (50 and 19 per cent, respectively).Many other characteristics of deportees to Guatemala are withheld by ICE, such as age, border versus interior apprehension, and previous deportations.In aggregate, however, ICE does report that 33.5 per cent of removals were from the interior of the United States in FY 2020 (see ICE [2022]).This corresponds closely to the sample, of which 34.7 per cent reported being deported from the interior of the United States.The systematic recruitment method, combined with these similarities with official ICE deportation data, gives confidence that the sample is relatively representative of Guatemalan deportees more generally.
Respondents who provided contact information were contacted for one-month and six-month follow ups.The follow-up surveys were conducted by phone, and the respondents who completed these surveys received a phone balance credit of at least fifty Quetzales for each survey.Phone surveys continued through October 2020, and 645 follow-up surveys were collected across the two waves, with 210 respondents interviewed in both follow-up waves.Questions relevant to the analysis were primarily asked in rounds 2 (one-month follow up) and 3 (six-month follow up).The results from rounds 2 and 3 were pooled, and all regressions utilized robust standard errors clustered by respondent.The multi-wave survey contained various questions covering topics ranging from demographics to experiences in the United States and Guatemala; specific wording for the relevant survey questions can be found in Appendix A.2.
Additionally, we conducted eighteen semi-structured phone interviews with deported migrants from the survey sample.These interviews covered the deported migrants' experiences after their return to Guatemala and the migration experience itself.These interviews aimed to provide a more detailed process tracing for mechanisms linking extortion (or lack thereof) and downstream behaviours in Guatemala.To recruit interview participants, all survey respondents for whom contact information was available and who reported using a coyote were divided into four groups, varying along two theoretically important dimensions: (a) whether they experienced extortion and (b) their intention to return to the United States.A random sample of respondents was selected from each of these four groups, and four to five respondents per group were interviewed.Interviews lasted 30 minutes on average, and participants were compensated with fifty Quetzales of phone credit.See Appendix A.3 for more information about each group's final number of interviews and a complete list of interview questions.
It is important to briefly discuss a few ethical considerations, given the vulnerability of the population this research focuses on and the sensitive nature of some of the questions.First, it was essential to safeguard the confidentiality of respondents.Thus, upon completion of the surveys, non-identifiable data was stored in encrypted form on an Amazon Web Services S3 server; only principal investigators on the project could download the data for decryption and analysis.Furthermore, all identifiable contact information for respondents was collected offline using paper and pencil and then stored in an encrypted database separate from the survey answers.Once transferred to the encrypted database, the paper and pencil versions of the contact information were destroyed.All qualitative interview recordings were made on devices without internet connectivity and deleted once the transcripts were completed, and all identifiable information was removed from the transcripts.Second, measures were taken to ensure the respondents were not coerced into completing the survey.Participants could skip questions and stop the surveys/ interviews at any point, though they only received compensation if they completed the survey round/interview.Given literacy rates, enumerators provided written copies of consent forms and read the consent script out loud.According to field notes, the compensation provided to respondents was reasonable and appreciated.It was not so large that it put participants at undue risk by carrying large volumes of money in Guatemala.Finally, COVID-19 posed ethical issues to continuing in-person surveys.Thus, once COVID-19 became a threat, all in-person surveys ceased, and all remaining surveys were conducted exclusively by phone.
Our key independent variable, 'Extortion', measures whether respondents (or their families) were forced (via threats to them or their families) to pay coyotes additional smuggling fees beyond what they had originally agreed to pay.For respondents who travelled to the United States multiple times, this question was specifically asked regarding their most recent journey.However, this question was only asked of the 87 per cent of respondents who used a coyote at some point in their migration journey. 11While those migrants who did and did not use coyotes to enter the United States differ in systematic ways, such as indigeneity and age (Appendix Table A7), the analysis focuses exclusively on the majority of respondents who used a coyote, of whom 17 per cent experienced extortion.Ten per cent of coyote-utilizing migrants were extorted en route to the United States, while the remaining 7 per cent were extorted after crossing the border.This extortion question was asked during the baseline survey upon arrival.It is unlikely to have primed respondents when answering questions about their economic circumstances or citizen engagement in follow-up surveys months later.
We draw on several survey measures to build a summary index measuring deportees' degree of economic hardship.Following Anderson (2008), a standardized inverse covariance index was constructed.This 'Economic Hardship Index' uses information about respondents' monthly income ('Monthly Income'), unemployment status ('Currently Unemployed'), how bad their current economic situation is ('Econ Situation [Bad]'), and financial difficulties since returning to Guatemala ('Economic Difficulties').Each of these was collected at one-and six-month post-arrival survey waves.For all measures, higher values indicate more negative or difficult economic situations.Indices help to avoid problems stemming from comparisons of multiple outcome variables.
Analyses drew on questions asked in the second and third survey waves about actions taken by respondents since their deportation out of 'fear of being a crime victim'.The behaviours include: avoiding leaving their homes by themselves, avoiding using public transit, preventing children from leaving the house, feeling the need to move to a different neighbourhood, changing their job or place of study, and obtaining a weapon for personal security.For the main analysis, the variable 'Fear of Crime (Count)' (ranging from 0-6) was used to count the number of fear items the respondent selected.
Civic action and political engagement were measured using two separate indices.Index items measured respondent likelihood (on a 5-point scale) of taking different types of action in the coming year.Three behaviours were tracked for the civic action index: participating in community meetings, volunteering, and mentoring youth.Civic action was thus analyzed as a simple three-item average 'Civic Action Index'.The 'Political Action Index' is the average response on three key types of behaviour indicative of political engagement: protest, affiliating with a political party, and voting. 12he regressions included a range of control variables, such as a binary measure of whether respondents left assets in the United States, a count of the number of children the respondent has in the United States or Guatemala, a measure of the highest level of education completed, a binary variable indicating whether the respondent was last apprehended at the border, a log of the number of years in the United States, and overall employment status since returning to Guatemala.The degree of migrants' social integration in the locality to which they returned was controlled for using a variable measuring the number of family and friends living nearby at the time of the follow-up survey.Various demographic variables were included.'Indigenous' refers to whether the respondents' mother tongue is other than Spanish.Another variable indicated whether the respondents had visible tattoos because of affiliations between gangs and tattoos in Guatemala.Finally, the survey round was controlled for.The one-and six-month follow up surveys straddled the onset of the COVID-19 pandemic, so this control variable also captures any changes in the dependent variable that may be linked to the pandemic, such as an overall lower interest in (or expectation of) civic/political engagement.Summary statistics can be found in Table A3.

11
It cannot be assumed that migrants who did not use a coyote did not experience extortion from some other criminal actor.Still, their experience is sufficiently distinct that we focus on coyote-using migrants only here.
Regarding social desirability bias, the United States and Guatemalan governments already knew that the respondents had crossed borders without the required documentation and thus had been deported.As a result, respondents had little to hide when discussing their migration experience.It is possible, however, that respondents were hesitant to admit that they had been victimized.This would be likely if the perpetrators had ties to Guatemala and could threaten deportees for speaking about the extortion.However, concerns about this are limited, given the openness of interviewees in discussing their victimization experiences during migration and mentioning victimization experienced by people they know.

Randomness of Extortion
Evidence suggests that extortion suffered while migrating is a quasi-random experience.Both qualitative evidence concerning migration through Mexico and a quantitative analysis of balance within the sample supports this argument.
Violence, including extortion, can happen to any migrant.Put simply, 'there is no subgroup that seems particularly at risk among deportees . . .kidnapping occurs simply because one is a migrant' (Slack, Martínez, and Whiteford 2018, 196).Another scholar suggests that individuals of any income can be kidnapped to extort money from their families; the kidnappers 'know that their families will send money even if they cannot afford to' (Vogt 2013, 764).Indeed, violence along the migration route is viewed by many migrants as a necessary evil; it is, in a sense, expected (Vogt 2013).Statistically, years of schooling, having children, and having entered the United States previously are not correlated with the likelihood of experiencing violence (Leyva-Flores et al. 2019).Thus, random chance plays a significant role in who experiences victimization during migration. 13 Migrants could theoretically reduce the likelihood of victimization by selecting 'good' coyotes before they begin their journey.However, migrants cannot always gauge the trustworthiness of coyotes because of the networked structure of the coyote business; initial coyotes frequently transfer migrants to others for different route stages.Additionally, many migrants travel to border towns independently and contract coyotes there; such migrants are 'in effect, giving themselves over to fate' (Spener 2009, 179).Even as the interviewees recommend that new migrants should know the coyotes with whom they leave Guatemala, their broader advice for avoiding extortion is limited and underscores extortion's frequency and randomness.Two interviewees say that putting oneself in God's hands is the best suggestion they have to avoid extortion.Others note: 'All the routes are the same . . .dangerous' (Group 1 Interview 2) 'It's common for you to be extorted, robbed . . .You can't prevent it' (Group 4, Interview 2) 'I haven't heard of anything [to avoid extortion]' (Group 4 Interview 1) 'There is no good recommendation' (Group 2 Interview 5) Table 1 below provides support for the randomness of extortion by showing that the 'extortion' and 'non-extortion' samples of survey respondents are well-balanced upon re-entry to Guatemala along a range of theoretically relevant variables.
There are two exceptions.First, women were slightly more likely to report extortion during migration.However, this difference is substantively small and only significant at the 90 per cent confidence level.It is also based on a very small sample size, as only 8 per cent of the sample were women.Second, the extortion group has a slightly higher mean number of past migration trips (1.77) than the non-extortion sample (1.62).The sample is balanced when the respondents were restricted to those who completed the follow-up sample, with the exception of small differences in gender, the number of children in Guatemala, and whether the individual was detained 13 Extortion during migration may extract contact information for family and include threats to family members.Yet, these community ties occur due to the extortion act rather than causing it, allowing us to maintain our expectation of quasi-randomness, orthogonal to sending community conditions.at the border.This analysis suggests that attrition between the survey waves did not lead to an imbalance in extortion experiences (Table A6). 14 Demographically, the panel sample is largely similar to the initial arrival sample, as shown in Table A4.The follow-up sample, however, has fewer indigenous language speakers, more women, higher average education levels, and slightly fewer past migration trips compared to the group that dropped out of the sample. 15Yet, the samples are similar across all other dimensions, including extortion rates.Finally, although personality characteristics might predict both attrition and future engagement, the sample attrition across follow-up waves does not appear to be driven by personality (Table A5). 16To help account for non-random attrition, controls for each variable (and other factors) are included in the later regression analysis.

Results
The first set of results explore extortion's effect on the two conditions theorized to affect civic and political engagement: economic hardship and fear of crime.OLS regressions test the expectations that extortion will cause both increased fear of further crime victimization (H1a) and greater economic hardship (H1b).The effects of extortion on these outcomes are shown in Fig. 1.Full numerical regression results can be found in Table A8. 17s Fig. 1 demonstrates, there is a statistically significant positive relationship between extortion and both fear of crime and economic hardship.Respondents who experienced extortion report levels of fear 0.30 standard deviations higher than respondents that were not extorted.This amounts to taking roughly 0.45 more avoidant behaviours due to a fear of crime (on a scale from 0-6).Qualitative evidence supports this finding.For example, one respondent said of a friend who had been extorted, 'the fear stayed with him' (Group 1, Interview 4).Another returnee who had been extorted said one reason they are so scared is that 'things that happen during the trip . . .like threats . . .wake you up to the fact that this is something common' (Group 2, Interview 5).We, however, acknowledge that this quasi-randomness of extortion by coyotes in this context may not travel across time and space.Other processes could affect the use of extortion, such as collaboration with drug cartels (Izcara Palacios, 2015;Slack and Martínez, 2018) and the type of routes that migrants and coyotes choose (Farfán Méndez, 2019).15 Though a concerted effort was made to limit it, this kind of attrition across demographic characteristics is common in many survey settings.See, for example, Alderman et al. (2001).
Respondents who experienced extortion also reported levels of economic hardship 0.29 standard deviations greater than those who did not suffer from extortion.The qualitative interviews echo this; respondents frequently and openly expressed concerns about debt repayment and Guatemala's lack of economic opportunities.When asked about what has been hardest for them upon re-entry to Guatemala, interviewees almost unanimously cited a lack of work as the biggest challenge.Debts from the migration journeyexacerbated by extortioncompounded this stress.For example, Interviewee 5 from Group 2 was detained and threatened by coyotes until they were paid more money (and sooner) than initially agreed upon.This deportee had to take out a loan at 10 per cent monthly interest to pay off the debt.The individual explained, 'sometimes one can't even sleep, thinking about how there's a payment due tomorrow but there's no money'.
A series of robustness checks are included in the Appendix.There remains a positive impact of extortion on fear of crime using a negative binomial model for the unstandardized count variable and a logistic regression model for a binary dependent variable (Table A9).Models are shown for each dependent variable used to construct the economic hardship index (Table A9).In addition to the economic hardship index, Column 7 in Table A9 shows that respondents who suffered extortion expect debt to be a larger barrier to reintegration. 18

Extortion's Effect on Engagement
The next set of analyses examine both the direct and indirect effects of extortion on citizen engagement.First, the results suggest that there is a large direct effect of extortion on civic action (Figs 2 and 3; see Tables A10 and A11 for corresponding full regression tables with controls).For civic action (community meetings, volunteering, and mentoring), extortion during migration correlates with a desire to take civic action that is 0.21 points higher on a 5-point scale.This direct effect is equivalent to an increase of about 0.25 standard deviations on the civic action index.The direct effect of extortion on political participation is of a similar magnitude but estimated with less precision: 0.23 units on a 5-point scale, or about 0.20 standard deviations (significant at the 90 per cent level).
The direct effect of extortion on political participation is almost exclusively driven by the index's 'protest' component.Extortion has a positive and statistically significant effect on protest but virtually no effect (in magnitude or significance) on political parties or voting (see Table A12).Protest is a form of political action outside the existing political system while voting and joining parties are actions within the existing system.This raises the question of whether extortion drives interest in institutional change but disillusionment with the existing political process.
The qualitative interviews support this argument, providing insight into the coefficient strength difference between political and civic engagement.Some extorted deportees have less desire to remigrate due to financial constraints or trauma.Consequently, they talk about what they want to do to improve their home communities now that they are staying.Yet, even those invested in their communities (helping neighbours, attending community meetings, etc.) are generally apathetic toward local politicians.For example, one returnee indicated that s/he was an active community member and explained, 'I am a taxi driver and I always help people at any hour' (Group 1, Interview 2), but noted that 'from politicians, you can never get any help'.
Mediation analysis tests whether extortion mobilizes via economic hardship or fear of crime.This analysis uses the approach described in Imai, Keele, and Tingley (2010) to estimate the Average Causal Mediation Effect (ACME) of extortion via the hypothesized mediators.Estimating causal effects from mediation analysis requires a strong assumption of sequential ignorability.First, sequential ignorability requires that the 'treatment' variable (extortion) is independent of the potential outcomes of both mediating variables (fear and economic hardship) and the political/civic engagement outcomes (conditional on observed pre-treatment covariates).Given the quasi-random nature of extortion, this is a reasonable assumption.Second, sequential ignorability also requires the observed mediators to be independent of all potential outcomes conditional on the treatment and pre-treatment covariates.This is a more difficult assumption to meet since both fears of crime and economic hardship are likely to be affected by factors other than migration journey victimization.The estimates of mediation effects are interpreted as more suggestive than causal.
There are separate mediation analyses conducted for the two mediators: fear and economic hardship.Figure 2 presents estimates of the Average Causal Mediation Effect (ACME) and the Average Direct Effect (ADE) for extortion and the 'fear of crime' mediator.The left panel shows the effects on the civic action index outcome, while the right panel displays the effects on the political action index.There is little evidence of any causal mediation effect for extortion via fear on civic or political action.The point estimates for the ACME are near zero and not significant when looking at either action index.This challenges the expectation that fear acts as a demobilizing forcein this case, fear of crime appears to have little relationship with citizen engagement.
By comparison, Fig. 3 plots the mediation analysis results using economic hardship as the mediator.A suggestive positive ACME of 0.02 is found for the civic action outcome, but it is very close to zero and only significant at the 85 per cent level.As a result, this finding is only interpreted as suggestive.Further, the mediated effect is relatively small compared to the ADE of extortion on civic action, representing about 9 per cent of the total effect.Finally, the ACME is estimated to be near zero for political participation.
Broadly, the results in Figs 2 and 3 indicate that, for Guatemalan migrants, economic shocks experienced hundreds of miles/kilometres away predict increased interest in civic action and protest after deportation back to their home country.This effect appears robust across the main outcomes, including when controlling for personality characteristics and department-level fixed effects to account for the location in which returnees live (see Table A13). 19Interestingly, this positive correlation between extortion and engagement does not appear to generalize to another common type of victimizationassaultwhich does not have the same direct economic consequences.Regressing the civic and political indices against a dummy variable for migration journey assault shows no correlation (see Table A14).This is another indication that an economic-grievance mechanism may be more powerful than a fear-based one.
Our qualitative evidence further suggests the importance of the link between extortion, economic hardship, and civic engagement.Despite widespread disaffection with political parties and the national political system, respondents talked about civic engagement outside of the traditional political system, such as volunteering labour for a potable water project (Group 2, Interview 5) or helping single mothers or the elderly (Group 2, Interview 3).For some, a sense of frustration related to migration victimization, lingering debt, and an ongoing lack of employment fuel a determination to pursue change.Multiple respondents use the word 'luchar' to describe this focus, which translates to striving, struggling, and fighting for change.For example, one extorted individual said, 'there is no work, and we have very low economic status and, for this reason, it is difficult to continue onward.But we have to fight, to hope' (Group 1, Interview 1).Another migrant said, 'particularly because of the debt, [it would be good] if there were some institution or something that could provide me help' (Group 3, Interview 4).This implies that there are a limited number of formal or governmental institutions to provide support.Therefore, returnees and their communities must help themselves.Indeed, the sense that change comes not from politicians but from the community may help explain why extortion predicts a higher interest in both civic engagement and protest but not other forms of political participation.
The study demonstrates that extortion directly affects civic action, economic hardship, and fear of crime.The data provides some evidence that economic hardship partially mediates the relationship between extortion and increased civic engagement.However, there is no evidence that fear of crime is a mediator, perhaps because the violence occurred in a distinct location.Further research should explore the extent to which victimization contributes to fear of additional similar victimization (that is, extortion on the migration journey) versus creating a more generalized sense of fear.While the pathway through economic hardship to civic action is most supported in the analysis, the direct effect of extortion on civic action is much larger than the mediated effect.Thus, alternative mechanisms should be explored, including cognitive or emotional ones.For instance, threats made during extortion attempts against migrants' family members in Guatemala might generate a psycho-social response, prompting deportees to engage in participatory actions and community work once they return to those same local/family contexts.20

Conclusion and Implications
We find that extortion has a significant direct effect on increased citizen engagement, especially when such action occurs outside strictly political spaces.Specifically, extortion is correlated with an increased desire to attend community meetings, volunteer in the community, and protest.Nevertheless, regardless of extortion status, deportees remain relatively disengaged from national 19 Since we are testing multiple hypotheses related to the extortion variable, we also report Romano-Wolf corrected p-values for the main outcomes in Table A15.The Romano-Wolf correction helps to control the familywise error rate (FWER).More details on this procedure can be found in the Appendix.Overall, while the adjusted p-values are slightly larger, as expected, the results on fear of crime, economic hardship, and civic action remain statistically significant at 95 per cent, and political action remains significant at 90 per cent.politics and explicitly political spaces.Qualitative evidence supports the argument that extortion increases a desire to build community, even amid dissatisfaction with political institutions.The analysis also explores extortion's potentially oppositional psychological and economic effects.There is evidence that extortion increases both fear of crime and economic hardship.Economic hardship exacerbated by extortion may mediate some of the relationship between extortion and increased citizen engagement.
Given the large volume of migration to the United States from Central America and deportation from the United States back to Central America, it is crucial to understand the impact that forcibly relocating migrants has on deportees' interactions with their 'home' communities and the prospects for stable governance there.This is especially true given the rise of nationalist parties and anti-immigrant voter sentiment (Brader, Valentino, and Suhay 2008;Choi, Poertner, and Sambanis 2023;Gaikwad and Nellis 2020;Hooghe and Dassonneville 2018;Ivarsflaten 2008) driving deportation as a policy tool.This project focuses on Central American/Northern Triangle migration to the U.S., but migration from North Africa and the Middle East to Europe is also an important consideration.Indeed, in terms of generalizability, Guatemala is about average in democratic strength compared to other deportee-receiving countries (Fig. A3).Thus, on average, deportees' perceptions of and interactions with civic spaces or institutions may extend to other contexts.In contrast, Guatemala's party system is weaker than many other countries, so the weaker relationship found in political engagement may be due, in part, to lower interest in engaging with weak political institutions. 21 The results provide insights for policymakers and programmes supporting displaced persons and migrants suffering from the negative psychological and economic effects of migration-related trauma.Information campaigns in sending communities might educate potential migrants and their networks about the true financial cost of migration, which, with extortion, can be higher than expected or initially quoted by the coyotes.Such information would enable households to make more informed decisions, more accurately incorporating the risk and cost of extortion into their migration decision.In addition, the results highlight a vicious cycle: economic hardship is a main driver of migration through Mexico to the United States, and extortionparticularly for deported migrantscompounds and extends economic need.
However, in another sense, the results are promising for the community reintegration of deported migrants: economic shocks and grievances seem to motivate individuals to become more civically active.This presents an opportunity for initiatives seeking to strengthen democratic norms and institutionsparticularly if the newly engaged returnees feel efficacious in their heightened community engagement.On the other hand, initial higher levels of engagement may lead to lingering resentment and discontent if underlying economic stressors are not addressed.Programmes that promote social cohesion and civic engagement among migrants and returnees would benefit from considering underlying motivations for participation; if economic hardship is reduced, programming may need to include more outreach, education, and alternative motivations to achieve higher levels of engagement.The results also suggest that one particular challenge for such policies and programmes is to build trust in politicians and increase formal political engagement.Thus, programming or reforms that increase the accountability of political officials and the Guatemalan government is essential to channel this increased engagement of returnees into formal institutions.Increasing state-led reintegration programming and support would likely better incorporate deportees in the future.
Supplementary material.The supplementary material for this article can be found at https://doi.org/10.1017/S0007123423000418 21 It is interesting to note, as Table A16 illustrates, that when looking at differences between Guatemalan migrants and non-migrants, the two groups are not significantly different in most citizen engagement measures.The main difference arises in voting during the last elections, which could result from migrants being more likely to be outside the country at the time.

Figure 1 .
Figure 1.Extortion predicts poor outcomes: fear of crime (count) and economic hardship index.Note: Y-axis displays dependent variables.Dependent variables (DVs) are on a standardized scale.Estimates are OLS regression coefficients for the extortion variable on each DV (95 per cent and 90 per cent confidence intervals).

Figure 2 .
Figure 2. Mediation analysis: extortion and fear of crime (count) on citizen engagement.

Figure 3 .
Figure 3. Mediation analysis: extortion and economic hardship index on citizen engagement.Notes: Mediation effects computed over 2000 simulations using the 'mediation' package in Stata.Models are OLS regressions.Plots show 90 per cent and 95 per cent confidence intervals.