Emigrants and the onset of civil war

Explaining civil war onset

Civil conflict is an interaction between a state government and an organized group of non-state actors that results in a number of battle deaths. In the UCDP/PRIO armed conflict dataset the threshold is 25 annual battle deaths for armed conflict and 1,000 for war (Gleditsch et al., 2002; Themnér & Wallensteen, 2013). It is a competition for resources between two actors with opposed preferences (cf. Grossman, 1991; Garfinkel & Skaperdas, 2007). Engaging in violent conflict is costly, such that the political outcome of the conflict is worth less after fighting than it would have been under an arrangement without violence. Scholars usually explain civil war onset by highlighting one of two processes, both of which are necessary for conflict to occur. First, there must be a reason for individuals to join a rebel movement, despite the ability to benefit from others’ costly efforts without contributing. Second, there must be a reason that a formed group and the state cannot reach a mutually beneficial agreement that avoids costly conflict. A discussion of the scholarly literature on both the collective action problem and the bargaining problem illuminates a general theory of civil conflict onset.

Conflict can only occur if a group of non-state actors is willing to work together to challenge the state, but individuals face incentives that discourage their participation. Unlike governments, rebel leaders rarely have a group of people on which they can rely to be ready to engage in conflict. Rebel leaders can coerce individuals into participating with pecuniary incentives or by threatening their lives or well-being (e.g. Beber & Blattman, 2013), but this is costly to the group, requiring organization and enforcement (Beardsley & McQuinn, 2009). More commonly, individuals join the cause in the interest of material incentives, either for private gain (Collier & Hoeffler, 2004; Ross, 2004; Fearon, 2008) or to alter a dissatisfying status quo (Gurr, 1970; Walter, 2004). Rebellion is more common, for instance, when there is inequality across groups within the state (Sambanis, 2005; Stewart, 2008), such that individuals desire improvements that they believe to be attainable. ² In other words, grievances and discriminatory treatment can be a motivating factor for dissent. Furthermore, experiencing state repression helps bind groups together in the cause of altering the status quo (Young, 2013).

Dissatisfaction with the status quo may incentivize individuals to join a rebel group, but it alone is insufficient for action. Lichbach (1995) argues that circumstances must enable actors to overcome the individual incentives to stay at home, expecting others to bear the costs of rebellion while they reap the non-excludable benefits. The ‘rebel’s dilemma’ can be solved when the cost of rebellion is sufficiently low (perhaps because others bear the burden of resource provision) that the individual will value the change enough to join the movement (Lichbach, 1995: 38–47).

Having overcome the collective action problem, a non-state group can credibly threaten the state with a challenge, but this still does not necessarily lead to war. Engaging in reciprocated violence costs resources, damages valuable infrastructure, and results in loss of life. The costs of fighting make conflict ex post inefficient, creating a range of possible settlements both actors would prefer to war. To explain onset, one must address why this bargaining range is not available to the disputants, and most scholarly explanations for civil war onset reduce to either asymmetric information or commitment problems. ³

Uncertainty about the limits of the bargaining range coupled with some reason why the information cannot be credibly revealed outside of war can lead a disputant to make demands outside of that range, causing conflict (Blattman & Miguel, 2010: 11–12). Governments seeking to maximize their own benefits would prefer to negotiate bargains only with those groups who can inflict heavy costs on the state in the context of violent conflict, but they do not know which groups have such capacity (Walter, 2004). Rebel groups may not know what resources they have at their disposal themselves, particularly if those resources are decentralized, difficult to obtain, and/or connected to individual households who may or may not participate in the collective action (e.g. Beardsley & McQuinn, 2009). Even if they do know their own strength, rebels (and states, for that matter) have incentives to misrepresent their true capabilities to extract more favorable deals (Walter, 2009). As a consequence of either true uncertainty or deliberate misrepresentation (or both), governments can underestimate a group’s resolve and offer insufficient concessions in the bargaining stage, leading to conflict. ⁴

To understand the conditions that influence the onset of civil war, we look for explanations that impact whether individuals will overcome the rebels’ collective action problem and why a rebel group or government would fight rather than accept a bargain. In the next section, we examine an entity that affects both of these aspects of domestic conflict: migrant networks.

Emigrants and domestic conflict

Migrant networks (sometimes called diasporas) consist of people living outside of the state of their birth who remain connected in some way to the same home state. ⁵ Over two hundred million people are living outside of their place of birth, or about one in 33 people in the world. ⁶ Emigrants write letters home, communicate with friends and family on the phone and the internet, send money to family members, and stay involved in and informed about the political and economic life of their home state (Sheffer, 2003). Migrants transfer information and resources voluntarily and reciprocally, sending news and money from abroad to friends and family in their state of origin and receiving information about conditions in their homeland. Through these repeated, two-way interactions, migrant groups shape the expectations and incentives of those remaining at home – and thus their behavior.

While other organizations – NGOs, for instance – can also be tied to a particular state of origin and engage in global communication and resource transfers, the membership of migrant groups is critical to their influence. Emigrants are tightly connected to their home state by the nature of familial ties and similar cultural backgrounds. A common reason for migration is the hope of economic prosperity abroad to help one’s kin back home; family members remain in the state of origin while migrants work abroad and maintain their familial ties. The tight relationship of ‘familyhood’ (Bryceson & Vuorela, 2002: 3) even across national borders makes communication and transfers between groups more common than between international groups without such fundamental connections. Further, emigrants, even if not related by blood, are similar in background, culture, and experience to the groups remaining at home, making the ties between groups stronger, comparisons more valid and reliable, and resources more forthcoming than in other types of transnational organizations. To put a finer point on our argument, the connections between friends and family in the state of origin and emigrants abroad enable information and resource transfers, but these transfers are both more salient and more likely than in networks of less similar members because of the strong ties associated with kinship and familiarity.

We argue that emigrants have an important influence on the likelihood of civil war onset in their state of origin. The familiarity between migrants and individuals in their home state and the connections between them can affect both the likelihood that individuals will join rebel movements and the likelihood that a rebel group will be unable to reach a settlement to avoid war with the government in the state of origin. In this section, we posit three ways by which migrants can affect the likelihood of civil war onset in the home state, namely by highlighting differences in rights protection, transferring resources to enable conflict action, and increasing the amount of information available for bargaining.

Migrants form a type of social network through which information can be shared with members back home and influence those individuals considering collective action against the state. An actor can gather more information from sources in a social network than she could on her own, being otherwise limited by resource constraints. The more sources of information there are, and the more actors interact and communicate with one another, the more information becomes available for use in an individual’s decisionmaking, thereby facilitating political activity (McClurg, 2003) such as challenging the government (Lohmann, 1994). Emigrants frequently communicate with friends and family in the home state, sharing information about their experiences abroad. The personal communications include more than mere facts, as would be reported in the media, extending to testimony – ‘stories told by people whose lives have been affected’ (Keck & Sikkink, 1998: 19). In this way, individuals who remain at home are tied to sources of information abroad, allowing them to access more information than would be otherwise available to them about differences in policies and treatment elsewhere.

Information from migrants allows kin to make a more valid comparison of conditions abroad to their own situation than they would with information from, say, a newspaper or NGO report from the host state. Emigrants are very similar to those who remain behind, in most cases sharing a language, religion, ethnicity, experiences, etc. If people in the state of origin learn that others with the same characteristics make more money, receive better rights protection, or participate in governance more than themselves, this new and uniquely valid knowledge can emphasize grievances they may feel or draw attention to new ones. Information about better outcomes for individuals like themselves living elsewhere can make people in the state of origin (increasingly) dissatisfied with the status quo. The more dissatisfied individuals are, the more they can gain by joining groups aiming to challenge the status quo (Gurr, 1970). Thus, information from migrants living in states who experience more rights protections, democratic institutions, higher income, etc. can lead individuals in the state of origin to overcome the collective action problem that might otherwise prevent rebel group formation. ⁷

Here, we focus on rights gaps, such that communication with emigrants who experience more human rights protections than people living in the state of origin should increase individual citizens’ motivation to form or join groups and rebel. People living under a particular status quo may not be dissatisfied with their treatment until they learn that they could be living under more favorable conditions, perhaps ones they did not realize existed (Gurr, 1970). When actors who remain at home learn through normal interactions with emigrants that the rights of similar actors living abroad are better protected, they are likely to conclude the status quo of their own treatment is unsatisfactory, leading them to join rebel causes.

Hypothesis 1: As a larger proportion of emigrants from a given state of origin live in states with higher rights protections than the state of origin, civil war will be more likely to occur in the state of origin.

Migrants transfer tangible resources to groups at home that can affect the bargaining aspect of civil conflict and enable political activity that would otherwise not be possible. Resources can include expertise, arms, money, etc. The most prominent resources migrants transfer back home are remittances: emigrants sent more than $440 billion dollars to their states of origin in 2010 ($325 billion to developing states), a low estimate as it does not include informal or unreported remittances (International Organization for Migration, 2011: 56).

Remittances can make it increasingly possible for individuals to mount a challenge against the government. This inflow of money makes it possible to fund a rebellion when household incomes alone would be insufficient or when income from work would have to be forfeited to participate in an organized challenge. Furthermore, remittances require little investment from recipients, such that the return on their investment is much higher than, for instance, their incomes from employment (Beardsley & McQuinn, 2009). Migrants have even been known to specifically send resources in support of violent challenges against the home government (Bercovitch, 2007: 26), as has been the case during the Troubles in Ireland or the civil war in Sri Lanka (Angoustures & Pascal, 1996). However, resources need not be targeted toward a rebel effort to be used in that way; the simple influx of additional resources into a group’s economy makes it easier to choose to fund a rebellion. These additionally available resources make war-fighting more attractive as compared to bargaining than it otherwise would be.

Remittances as a source of war-fighting resources can create uncertainty for the home government. Though there have been known instances of migrants funding rebel groups directly, most remittances are sent to households. Since the government cannot know which households are likely to contribute to the cause, resources like these that offer a high return on investment can be difficult to anticipate. Further, migrants can send resources to kin via informal channels that are particularly difficult to track and thus anticipate. For example, the hawala transaction system facilitates transfers of money between migrants abroad and their families at home through a clearing system of credits and debits without physically transferring money. The system operates outside of the purview of government institutions, using reputational mechanisms and the shared values of migrants and individuals in the home state to supplant formal transaction systems (Schaeffer, 2008). Schaeffer (2008: 4) notes that similar methods of remittances are presently in use around the world and are particularly valuable in states like Somalia with weak or corrupt economic institutions. Since these informal channels exist outside of government control, the extent of these transactions is difficult to gauge (Ratha & Shaw, 2007). ⁸

With more resources available from which to fund a rebellion, the state cannot accurately know what resources a rebel group has – and the group has incentives to misrepresent that information – and so the disputants become more likely to engage in civil conflict rather than bargaining. We posit that a group in the state of origin will be more likely to rebel as individuals receive more remittances from emigrants abroad.

Hypothesis 2: As remittance inflows from abroad increase, the likelihood of civil war onset in the state of origin will increase.

In contrast to the ways migrants can contribute to an increased likelihood of conflict, migrants can also share knowledge of their home state with outside actors and cultivate a decrease in the likelihood of civil war. Emigrants can convey information about conditions in their state of origin to actors in their host state, who can promote the interests of those back home. Leblang (2010) highlights this type of interaction: emigrants provide potential investors with information about markets, workers, and other economic conditions in their home state, increasing the likelihood that economic actors will invest in that state of origin. Such a transfer of information can lead to diplomatic interactions between host and home states as well. For example, a host state may apply political pressure on a home state to remedy grievances, or it may confer international legitimacy on a migrant group’s cause by recognizing the plight of citizens in the home state. Keck & Sikkink (1998) named this type of interaction the ‘Boomerang Pattern’ of international pressure. In this model, groups experiencing grievances in state A can send information to connected groups in state B, who put political pressure on state B to take international action against state A. Though the model describes the transfer of information through transnational advocacy networks consisting of domestic and international NGOs, its implications apply similarly to the connections between migrants and groups in their native country.

Emigrants receive information about conditions in their homeland from relations remaining there and use it to promote the interests of the home country while living in the host state (e.g. Leblang, 2010; Bercovitch, 2007). Migrant groups are frequently willing to pressure their host governments to impose sanctions or lend diplomatic or military support to their kin in the state of origin, as the Jewish diaspora has on behalf of kin in Israel (Sheffer, 2007), the Armenian diaspora has to influence outcomes in the conflict between Armenia and Azerbaijan (Tölölyan, 2007), and Cuban emigrants have to maintain US pressure on Castro (Grugel & Kippin, 2007). Emigrants lobby their host governments, use transnational or intergovernmental groups to pressure the home government for change, raise media attention to the situation in their home state, and attempt to create or structure the portrayal of the home group in the public eye (Tölölyan, 2007: 107–108). Such activity is particularly likely in states where organization and lobbying are easier, such as in democratic societies (Shain, 2002: 120) or in states where civil society is active.

Even absent pressure from a host country, the information accumulated from migrants living abroad is likely to impact decisionmaking in the home state. Testimonial accounts shared by emigrants living around the globe can increase what government actors in the home state know about potential rebel groups. Migrants can signal their kinsmen’s discontent through political or social outlets available to them in their host countries and direct international attention toward particular grievances. As more migrants share their stories abroad and generate an international interest in homeland problems, actors within the home state learn once-private information that is valuable in the bargaining process. As Walter (2004) notes, civil wars will only occur when individuals are dissatisfied with the status quo and perceive violence as the only way to change it. If individuals can find alternative means of addressing their grievances, they will prefer these options and avoid costly conflict. However, when government actors lack knowledge of these grievances, conflict may occur despite the existence of peaceful alternatives. The information migrants can provide to civil society and the media while abroad can alleviate this uncertainty. Importantly, because learning this information from media or NGOs or even diplomatic interactions impacts the state’s international reputation, the government is likely to see this information as more credible when compared to similar information of dissatisfaction from domestic groups alone.

Emigrants’ testimonies abroad serve to mitigate incomplete information problems, thereby increasing the likelihood that a mutually beneficial bargain will be reached among actors in the home state. Of course, this transfer of information can only occur if migrants have a way of spreading the information about circumstances in their state of origin while abroad. We argue that the presence of a vibrant civil society (i.e. international nongovernmental organizations or INGOs) in the host state approximates the opportunities migrants will have to spread information to others. We thus posit the following hypothesis:

Hypothesis 3: As increasing numbers of migrants have access to INGOs in their host states, the likelihood of civil war onset in the state of origin will decrease.

It is possible that all these characteristics affecting civil war outcomes are actually the results of emigrant selection processes. If emigrants seek host states with more rights protections or higher INGO membership than their state of origin, then our assumptions that the processes of transfer are similar or equal across host states can bias the likelihood of finding that these effects are taking place. We proceed under the assumption that migrants have some, but actually quite limited, options when it comes to their choice of host state. The vast majority of migrants are economic migrants, following the possibility of employment and economic gain rather than political concerns (International Organization for Migration, 2011). More important than their preferences are the opportunities available to them: migrants frequently have limited resources for travel and relocation, keeping them within a short distance of their state of origin. Most migrants tend to move no farther than a bordering state (Moore & Shellman, 2007) or region (cf. International Organization for Migration, 2011: 62). Thus, while some migrants have the resources to choose to move to Switzerland or Finland, have been lucky enough to be from a state already near such a state, and receive visas from such places, most do not. ⁹

In combination, these ideas suggest that migrants have an important and multifaceted influence on the likelihood of civil war onset. The likeness within their membership makes information and resource transfers easier and more credible between groups at home and migrants abroad than is possible with other organizations, raising grievances and resources that increase the likelihood of rebellion. Transnational connections can also provide access to alternative means of communicating valuable information that can alleviate information asymmetries for actors within the home state, thereby increasing the chance that a nonviolent bargain will be reached. To be clear, we do not mean to suggest that emigrants are necessarily sending specific positive or negative incentives to friends and family that they should or should not rebel against the home state. Emigrants do all of these things – they reveal information about relatively lower repression abroad, they send money that would otherwise not be available in the state of origin, and they spread information about conditions in their home state to actors in their host state. The presence of these effects then increases (or decreases) the underlying propensity of actors in the state of origin to engage in a rebellion. Thus, we suggest that migrants represent an important addition to our understanding of civil conflict, in that they impact both the collective action and the bargaining processes that contribute to the likelihood of civil conflict onset.

Empirical analysis

In this section, we analyze the effects of relative levels of repression, remittances, and access to INGOs in states hosting emigrants on the likelihood of civil war onset at the origin state-year level of analysis, using data from 1981 to (at least) 2003 on all states for which data are available, which varies slightly depending on the independent variable of interest – the years and number of states under study for each estimated model are listed in Table I. We describe our data and empirical assumptions before proceeding to an analysis of the results.

Operationalization

The dependent variable for all of our empirical models is the onset of civil conflict as indicated in the UCDP/PRIO Armed Conflict Dataset (UCDP/PRIO, 2011). To be considered a civil conflict, a case must involve the use of armed force between at least two parties (one of which is the state government) that results in at least 25 battle-related deaths per year (Gleditch et al., 2002; Themnér & Wallensteen, 2013). We code Civil conflict onset as 1 in the year in which the civil war first reached 25 battle deaths and as 0 in all other. Because this indicator is dichotomous, we use a probit likelihood estimator, with the errors clustered by state of origin, for the analyses reported below.

We first argue that emigrants provide information about the existence of better rights protections abroad than in their state of origin; individuals in the home state use this information to identify new or intensify existing grievances related to their treatment and join rebel movements against the state, making civil war more likely to occur. To test this proposition, we created a variable of Relative repression. The World Bank recently released the new Global Bilateral Migration Database (GBMD), which publishes the number of migrants from a given state of origin living in each possible host state around the world in each decade from 1960 to 2000 for 232 states (World Bank, 2011). We paired this database with the Cingranelli & Richards (2010) index of physical integrity rights protection (CIRI), in that each host state and each state of origin has a rating on the CIRI scale from 0 to 8, with higher scores indicating better rights protection (or fewer physical integrity violations) for the years 1981 to 2008. We coded a dichotomous indicator equalling 1 if a given host state has a higher rights protections score than the state of origin and 0 otherwise. ¹⁰ Multiplying this indicator by the proportion of the origin state’s total emigrant stock residing in that host state tells us the relative weight the host state’s rights practices are likely to have on decisionmaking by groups back home – the more migrants living in a host state, the more information should be received in the origin state about the relatively better rights protections there. ¹¹ Finally, these proportions are summed across all host states for each state of origin in each year, yielding the proportion of all migrants from a given home state living in host states with better human rights practices than the home state from 1981 to 2008. ¹² This represents the amount of information actors in the state of origin are likely to receive that many (or few) other states have better (or worse) rights protections than those they experience at home.

We next contend that migrant networks provide resources that groups in the state of origin may use to engage in rebellion, making civil conflict more likely (H2). Data on annual remittance inflows from migrants

OPEN IN VIEWER

Table I.

Analysis of Hypotheses 1–3

	(1) Hypothesis 1	(2) Hypothesis 1, 2-year lags	(3) Hypothesis 2	(4) Hypothesis 2, 2-year lags	(5) Hypothesis 3	(6) Hypothesis 3, 2-year lags	(7) Hypotheses 1–3	(8) Hypotheses 1–3, 2-year lags
Relative rights	0.773* (0.493,1.053)						0.759* (0.403,1.116)
Relative rights		0.351*						0.354*
(lagged 2 years)		(0.076,0.627)						(0.039,0.668)
Remittances			0.0000402* (0.000,0.000)				0.0000961* (0.000,0.000)
Remittances				0.0000597*				0.0000957*
(lagged 2 years)				(0.000,0.000)				(0.000,0.000)
INGO membership					–0.149* (–0.267,–0.031)		–0.298* (–0.440,–0.156)
INGO membership						–0.132*		–0.202*
(lagged 2 years)						(–0.239,–0.025)		(–0.329,–0.075)
Former colony	–0.287*	–0.208	–0.192	–0.194	–0.169	–0.241*	–0.350*	–0.293*
	(–0.484,–0.089)	(–0.423,0.008)	(–0.435,0.050)	(–0.426,0.038)	(–0.385,0.046)	(–0.463,–0.020)	(–0.618,–0.083)	(–0.552,–0.034)
Prior civil war	0.525*	0.611*	0.646*	0.576*	0.647*	0.673*	0.489*	0.463*
	(0.296,0.755)	(0.380,0.841)	(0.407,0.885)	(0.337,0.814)	(0.415,0.879)	(0.442,0.904)	(0.239,0.740)	(0.226,0.700)
GDP per capita (ln)	–0.182*	–0.181*	–0.202*	–0.220*	–0.162*	–0.164*	–0.191*	–0.190***
	(–0.248,–0.115)	(–0.245,–0.116)	(–0.283,–0.122)	(–0.303,–0.137)	(–0.220,–0.104)	(–0.221,–0.106)	(–0.285,–0.097)	(–0.272,–0.108)
Constant	–0.934*	–0.789*	–0.528	–0.361	–0.187	–0.208	0.290	0.163
	(–1.467,–0.402)	(–1.316,–0.262)	(–1.173,0.117)	(–1.010,0.288)	(–0.724,0.349)	(–0.748,0.332)	(–0.526,1.106)	(–0.599,0.924)
N	3,887	3,560	3,280	3,164	3,661	3,728	2,250	2,252
Years	1981–2008	1983–2008	1981–2008	1983–2008	1981–2003	1983–2003	1981–2003	1983–2003
No. of states	174	173	158	158	173	174	150	149
Log likelihood	–545.909	–506.448	–445.381	–439.178	–524.816	–531.445	–318.990	–342.466
χ– squared	117.304	89.919	65.017	62.865	74.664	88.287	88.515	74.018

95% confidence intervals in parentheses below estimated coefficients: *p ≤ 0.05. Dependent variable is Civil war onset.

abroad to their state of origin are available for 213 states from 1970 to 2012 from the World Bank (2013); for consistency across our analyses, all of our estimated models utilize data beginning in 1981. Remittances are reported in millions of current US dollars. We expect that an increase in monetary resources available to the population in the home state will increase the likelihood that a group may use those resources to rebel against the state.

Finally, we argue that migrants can provide information to audiences abroad about the grievances and likelihood of rebellion in their home state, making this information public so that the home state will be more likely to reach a bargain with a dissatisfied group. Hypothesis 3 predicts that migrants with greater access to outlets that can spread information internationally – namely, INGOs – will decrease the likelihood of civil conflict onset in the home country. To approximate the opportunities that migrants would be able to spread information about grievances and conditions in their state of origin while abroad, we create a weighted indicator of INGO membership, which accounts for the possibility that emigrants will be able to access INGOs and publicize conditions of their home state while living in the host state. Smith & Wiest (2012) provide data on the number of human rights INGOs that reported membership in a given state in the Yearbook of International Organizations from 1953 to 2003. The data were coded at 2- to 3-year intervals; like Murdie & Bhasin (2011), we use linear interpolation for the intervening years not coded. The greater the number of INGOs that have members in a given state, the more likely it is that diaspora members will have opportunities to speak on behalf of their kin while residing there. As we do for our indicator for Relative repression, we weight the (logged) ¹³ number of INGOs with citizens in a given host state by the proportion of migrants from a given state of origin living in that host state, collapsing this number to give us an indicator as to how much information there is about the home state’s grievances in the international community. ¹⁴

Our theory could suggest a dyadic (host–home) data structure, tying each state of origin with almost 200 possible host states for each year. However, we believe the effects of migrants on their home state work in the aggregate. For instance, some migrants from a given state of origin A will live in a host state B and experience more rights protections than their kin in state A do, but other migrants living in a host state C may experience more rights violations. To pair the migrant groups from each of these host states with the home state for individual analysis would cloud the fact that the groups in the state of origin receive information about experiences in both B and C. Indeed, treating each host–home pair as a separate entity is to suggest that information and resource transfers between each can function differently in their impacts on civil war. Instead, we argue that these efforts operate in the aggregate, such that citizens receive information from all host states, weighted by the number of migrants living in different places, and determine what to do with it once received. Thus, we use the origin state-year unit of analysis.

We control for country-specific variables that may affect the relationship between emigrants and civil conflict in the home state. Former colony equals 1 if the country was a former colony of a Western power and 0 if not (Hadenius & Teorell, 2005). Moore & Shellman (2007) argue that former colonies of Western powers are more likely to experience significant emigration, and civil war is also more likely in these states due to the relative youth of political institutions (Fearon & Laitin, 2003). An indicator of Prior civil war equals 1 if the state of origin experienced ongoing civil war in any of the previous five years, approximating an increased propensity for relapse (Collier & Hoeffler, 2004; Fearon, 2005) and emigration. The natural log of GDP per capita accounts for the impact of socio-economic factors on both migration and conflict onset, using data from the World Bank’s (2012) World Development Indicators.

Conclusion

Our theory and attendant findings advance the study of both migrant groups and civil conflict in important ways. Migrants interact with family, friends, and acquaintances remaining at home in ways made possible through advancing technology and the rise of global organizations, requiring new analysis of how they impact individual behavior, group dynamics, and political outcomes (see e.g. Leblang, 2010). We add to the new understandings of the impacts of migrants by examining the possibility that they can affect the likelihood of civil conflict occurring in their state of origin. We propose that migrant groups offer a solution to both pieces of the civil war puzzle: they impact the willingness of citizens to join rebel movements, and they affect the probability that state and rebel groups will fight rather than bargain nonviolently.

Though there have been both scientific and policy-centered studies of how conflict causes migration (see e.g. Moore & Shellman, 2007; Rubin & Moore, 2007; Davenport, Moore & Poe, 2003), less attention has been paid to how migrants might impact the onset of conflict, particularly in the emigrants’ state of origin. Salehyan & Gleditsch (2006) and Salehyan (2007) assess how refugees and other migrants can spread the likelihood of rebellion to host states, arguing that sudden inflows of refugees can destabilize host states and migrant flows can bring rebellious actors with them to the new state. We argue that migrants of all stripes – economic and political – can have an impact not only on their host states but also on their states of origin.

To precisely identify the processes by which emigrants affect civil war in their state of origin, we disaggregate three possible mechanisms of influence and assess each of them empirically in a way that allows us to distinguish among them. Migrants provide information, particularly as a source of comparison for those at home to realize that their situation could improve. We find that as a larger proportion of emigrants experience better human rights practices than their kin back home, civil war is more likely to occur. This finding extends the premise that relative deprivation prompts individuals to rebel (Gurr, 1970), identifying a specific and reliable source of comparative information.

Evidence suggests that the more resources migrants send home – specifically money – the more likely the state of origin will be to experience war. Remittance transfers, conducted inside or outside of legal channels, are difficult for governments to monitor. Furthermore, it is not always clear whether the recipients of remittances will use them to feed their families or rebel against the state. Thus, remittances create uncertainty for state leaders as to the strength of the rebel movement. This uncertainty can cause state leaders to underestimate a rebel group’s capacity and offer unsatisfactory concessions to the group, thus leading to conflict. The identification of emigrants as a likely source of funding for rebellions has primarily been studied in single cases (cf. Angoustures & Pascal, 1996; Beardsley & McQuinn, 2009) rather than using a large-N analysis as we have here.

We further find support for the hypothesis that emigrants can prevent civil war at home by circulating information to actors in their host state who are in a position to make noise. By working through nongovernmental organizations in the host state to generate publicity for their fellow groups’ causes, migrant groups can reduce the incentive to rebel as home governments gain information about the credibility of threats. Migrants can use their access to host and international audiences to provide a political voice for disaffected groups in the home state, as well as an option for realizing change other than through violent means.

This study contributes to a growing body of literature on migration for home and host states, and to our understanding of civil war onset. We contend that migrant groups have an important and unique role to play in the possibility of civil conflict by the nature of their connections and likeness to those they left in the state of origin. Because members are often connected by blood, marriage, or cultural ties, they are more likely to interact and share information and resources than members of other types of transnational organizations. The similarity of these individuals makes comparisons more likely and more valid than comparisons with other groups. These groups, then, have an influence on the likelihood of civil conflict that is quite different from that of institutions with a more diverse membership. Migrant networks harness the power of their connections and familiarity to transmit credible and valuable information between members, and in doing so, affect individual and group behavior among members in the homeland in ways that uniquely influence the onset of civil conflict.