The Long-Term Consequences of Armed Conflict on International Migration
Total Page:16
File Type:pdf, Size:1020Kb
The Long-Term Consequences of Armed Conflict on International Migration Michelle L. O’Brien University of Washington [email protected] Abstract Demographers and refugee scholars have long argued that international migration increases under the threat of physical security that accompanies armed conflict. The long-term effects of conflict are more difficult to ascertain. In this paper, I develop an argument that community institutions, which can moderate migration, are the key mechanism linking conflict and migration long after the peace accord has been signed. Using individual level data from Tajikistan, I find evidence that in districts that had experienced intense fighting, the presence of community institutions moderate contemporary migration decision-making. The findings suggest that the impacts of armed conflict are not limited to individual experiences, and support the call for increased investment in local rebuilding after conflict. This study has implications for understanding processes of demographic change in former conflict zones. 1 Introduction This paper describes an empirical study of variation in international migration patterns in the aftermath of civil war in Tajikistan. The specific focus of this paper is to examine the long-term consequences of both destruction and fatalities on regional push and pull factors. The body of literature addressing increased migration during periods of conflict has expanded (Davenport, Moore, and Poe 2003; Melander and Öberg 2006; Moore and Shellman 2004; Schmeidl 1997, 2001; Stanley 1987), including studies modeling migration as complex household and individual decisions, even under conflict scenarios (Lundquist and Massey 2005; Morrison and May 1994; Williams 2013; Williams et al. 2012). Taking a broader sociological approach, Williams (2013) demonstrates the value of community organizations in moderating the effect of conflict events on migration. Yet the literature does not consider migration dynamics in the aftermath of conflict. Considering the resources required to rebuild damaged institutions, I argue that migration increases persist after the battles have come to an end. I develop a conceptual model that draws on existing migration theory and incorporates a distinction between short- and long-term outcomes. Where previous models of migration increases during armed conflict have been useful in estimating refugee flows and distributing aid, this new model provides insight into population dynamics during the rebuilding process. I use the armed conflict in post-Soviet Tajikistan to examine this model. My analysis leverages detailed panel data from the World Bank on migration, a dataset of conflict events, and Census data from this period. Theoretical Framework 2 Two major theoretical and methodological approaches characterize the scholarly developments around social conflict and migration. The first approach examines the macro-level characteristics that explain the variation in refugee ‘production,’ that is, why do some conflicts produce refugee movements across borders while others do not? These are typically large, cross-national studies with the state as the unit of analysis. The authors typically aim to examine why some conflicts in some countries produce refugee movements, while others do not. Schmiedl (1997) is the oft- cited standard for this area of research. Where previous studies of refugee populations typically followed refugee streams or described camps and conditions in a particular destination, Schmiedl focuses on the origin of refugee movements, seeking to identify structural conditions that produce refugees. She focuses on three groups of factors involved in refugee migration: root causes, such as economic underdevelopment and population pressure; proximate conditions, such as human rights violations and civic or ethnic conflict; and intervening factors, which either facilitate or obstruct migration actions. She finds that genocide and politicide, foreign military intervention in civil wars, civil wars even in the absence of military intervention, and interstate wars are the best political predictors of refugee stock. This evidence supports the general claim of refugee literature, that “political violence is the most important cause of refugee migration” (1997:302). Taking their cue from this seminal work, Davenport, Moore, and Poe (DMP, for short) (2003) critique other large-N studies for suffering from selection biases, from an emphasis on push factors at the sake of pull factors, and an overemphasis on structural forces in lieu of microfoundations. These studies typically have quite crude measures of political violence, as well (Davenport et al. 2003). DMP find confirmatory evidence that politicide and genocide are the primary factors in refugee migration. However, the authors provide much more nuance and complexity than previous work. The authors find evidence that the more distinct types of conflict 3 dissidents use, the more likely individuals will perceive threats and leave. In addition, they make a distinction between the perpetrators of conflict events, and find independent effects for state actions, dissident actions, and the interaction between state and dissident actions. They find that shifts toward democracy are associated with forced migration, and fail to find support for the concept of economic threats forcing refugee movements. However, the measure of GNP per capita as a proxy for economic threat is quite crude, and the dependent variable of identified refugees, by definition, excludes migrants who left the origin for economic reasons, failed to apply for refugee status, or were denied refugee status. Some studies in this area of research deviate slightly from the cross-national, global analysis that Davenport et al. and Schmiedl present. Iqbal (2007) narrows the focus to the African continent and finds evidence that refugees are more likely to seek destinations that are geographically close to home, free of conflict, and large states that may be more likely to accept refugees. The author provides some evidence that refugees fleeing political conflict may seek alternative political systems from that of their origin state. However, this focus on refugees’ seeking behavior is not consistent with the measurement at the state-level, and the findings might be more convincing if considered through the lens of state characteristics associated with the acceptance of refugees, rather than demonstrative of the preferences of refugee populations. One case study that situates itself in the refugee literature, but examines only one case is that of Czaika and Kis-Kastos (2009). The authors attempt to explain village-level displacement, and use methods that closely resemble cross-national analyses. The authors ask what characteristics of the village are associated with net emigration from the village. However, this study has several methodological issues, including the measurement of net emigration from the village, 4 which in actuality, is a measure of population change. This is problematic for a number of reasons, which I will further explore in the next section. The second approach examines the relationship between conflict and migration at the individual and household levels, often within a broader demographic framework of the determinants of migration. These are typically studies that utilize survey data and examine one, or a handful of cases, generally within the same geopolitical region. Within this micro-level approach, two main goals are typically the center of research: 1) is a particular migratory stream ‘forced’ or ‘voluntary’? or 2) how does the probability of migration respond to ‘threats’ such as conflict events? The first goal, that of attempting to delineate between forced and voluntary migration (or ‘economic’ and ‘political’ migrants) typically stems from policy decisions and borrows administrative definitions as analytical categories. This false dichotomy is problematic. It implies a lack of agency in the case of forced migrants, and conversely, a lack of political threat in the case of voluntary migrants. Nevertheless, the policy importance of this dichotomy is certainly relevant, and has generated a great deal of debate. This issue found political salience in the U.S. in the late 1980s as asylum seekers from conflict zones in Latin America fled uncertain circumstances wrought with political violence and extra-legal killings. Where Stanley (1987) found that political conflict was at least a contributing factor in Salvadoran migration to the United States, Jones (1989) argued that there had been no systematic relationship between political killings in El Salvador and migration to the U.S. However, while both studies remain important as the first systematic analyses of political violence as a ‘push’ factor, several key critiques of the two studies have emerged. First, both studies measure Salvadoran migration to the U.S. using the measure of apprehensions as the dependent variable. This dependent variable 5 is biased by changes in border enforcement spending, and subject to variation within different U.S. states and entry points. The measure may also suffer from selection bias, as more isolated migrants without kinship ties to secure passage across the border may be at a higher risk of apprehension. Further, while Jones analyzes conflict-related economic disruptions, his interpretation of the measures is that they are purely economic motivators. These measures of economic sabotage include attacks on the elite and business owners and