Full Citation

Title: A data-driven computational model on the effects of immigration policies

Citation Type: Journal Article

Publication Year: 2018

ISBN:

ISSN:

DOI: 10.1073/pnas.1800373115

NSFID:

PMCID:

PMID:

Abstract: Many scholars suggest that visa restrictions push individuals who would have otherwise migrated legally toward illegal channels. This expectation is difficult to test empirically for three reasons. First, unauthorized migration is clandestine and often unobservable. Second, interpersonal ties between migrants and would-be migrants form a self-perpetuating system, which adapts in ways that are difficult to observe or predict. Third, empirical evaluations of immigration policy are vulnerable to endogeneity and other issues of causal inference. In this paper, we pair tailor-made empirical designs with an agent-based computational model (ABM) to capture the dynamics of a migration system that often elude empirical analysis, while grounding agent rules and characteristics with primary data collected in Jamaica, an origin country. We find that some government-imposed restrictions on migrants can deter total migration, but others are ineffective. Relative to a system of free movement, the minimal eligibility conditions required to classify migrants into visa categories alone make migration inaccessible for many. Restrictive policies imposed on student and high-skilled visa categories have little added effect because eligible individuals are likely able to migrate through alternative legal categories. Meanwhile, restrictions on family-based visas result in significant reductions in total migration. However, they also produce the largest reorientation toward unauthorized channels-an unintended consequence that even the highest rates of apprehension do not effectively eliminate. immigration policy | migration | unauthorized migration | computational modeling P olitical leaders in many Western countries have called for increased visa restrictions to control immigration. In the aftermath of the November 2015 attacks in Paris, Marine Le Pen declared, "It is essential that France recover the control of its national borders, once and for all" (1). Similarly, one of the main tenets of the Brexit campaign was to "take back control of [UK] borders" (2). In the United States, Donald Trump was propelled to victory with a campaign focused on border control and "extreme vetting" of Muslim migrants. During his early days in office, he moved to change the composition of incoming migrants and reduce flows from family-based and high-skilled channels (3). But will more restrictive immigration policies stop individuals from migrating? Many scholars suggest that visa restrictions have counterproductive effects, leading individuals to reorient to unauthorized channels (4, 5). While this expectation is prominent in theoretical literature, scholars have struggled to demonstrate it empirically. There are three fundamental empirical challenges. First, unauthorized migration is often unobservable, due to its clandestine and sensitive nature. Even the best estimates of unauthorized migration are extremely limited, are vulnerable to bias, and are often reported only in the aggregate. Second, migration flows do not result from the sum of individual decisions to migrate-they are part of a dynamic and social process. The rich literature on migrant networks holds that interpersonal ties between migrants and would-be migrants form an adaptive self-perpetuating system. As individual preferences are modulated by the nonlinear effects introduced by social interactions, networks make migration difficult to measure and predict (4, 6, 7). To date, existing research has, generally, been unable to connect decisions and social processes occurring at the micro-and mesolevels to macrolevel trends in migration (8) (however, see ref. 9). Third, drawing causal inferences in empirical evaluations of immigration policy is problematic: Policies are not exoge-nous and we, generally, cannot observe counterfactual scenarios. Taken together, empirical challenges such as these have led the International Organization for Migration to conclude that, "dis-regarding the uncertainty and complexity of migration leads to an illusion of control on the part of the decision makers.. . [and] this is why attempts at managing migration often lead to unintended consequences" (ref. 10, p. 2). We present a data-driven agent-based computational model (ABM) to examine migration for an origin-destination corridor, which is tailor-made to address these unique empirical challenges. To be clear, our paper does not model all migration into a particular destination. This would require a cross-national data collection strategy. We focus on a single origin country. Significance Would more restrictive immigration policies stop individuals from migrating? We present an agent-based computational model, calibrated using original survey and experimental data, which represents an important step in estimating the "substitution effect" whereby migrants reorient toward unauthorized channels due to changes in policy. We find that government-imposed restrictions on migrants can decrease total migration. However, some restrictions are highly ineffective while others decrease legal migration only at the cost of driving migrants into unauthorized channels. Restrictions on students and high-skilled workers are least effective in reducing migration, and restrictions on family-based visas are especially counterproductive in diverting migrants to back channels. We also find that increasing enforcement would not effectively eliminate the diversion to unauthorized channels.

Url: https://www.pnas.org/content/115/34/E7914

User Submitted?: No

Authors: Simon, Miranda; Schwartz, Cassilde; Hudson, David; Johnson, Shane D

Periodical (Full): PNAS

Issue: 34

Volume: 115

Pages: E7914-E7923

Data Collections: IPUMS USA

Topics: Migration and Immigration

Countries: United States