A Refugee shelter in my neighborhood!? Depends. Who, exactly, is inside? by Carlos Brito

Migration has become one of the main issues in elections worldwide and is often cited as one of the key drivers behind the rise of far-right populist candidates. In fact, an extensive literature has documented the causal impact of migration inflows on locals’ growing support for such political platforms across multiple settings. Yet, there is still an open debate about why locals react in such manner. Papers have highlighted the relevance of both economic mechanisms (such as competition in the labor market, welfare, public goods) and non-economic ones (including cultural clash, tradition preservation, and media).

However, the literature has mainly relied on aggregated data (municipality, provinces, or state level) which miss important within-city variation in migrant exposure. In addition, most consequences in receiving immigrants are inherently local (effects on neighborhood amenities and shared local public goods) and daily direct interactions can also shape locals’ attitudes. Therefore, in my job market paper, I propose to zoom in on a migrant hosting city to study, using granular data, how an inflow at the neighborhood level affected safety (crime) and local schools (composition and congestion), and whether locals changed their voting behavior.

The ideal setting

An ideal setting we would have: (1) Quasi-random assignment of areas within a city to receive the migrants. (2) If possible, quasi-random variation in cultural differences between migrants and locals to verify the role of cultural exposure. (3) Granular geo-located data not only on election results but also relevant mediators. (4) It would be great if the setting is in the developing world, which is far less studied by the literature, despite hosting 75% of the world’ s displaced migrants.

Fortunately, the Venezuelan refugee reception in Northern Brazil satisfies all the above. The Brazilian government in partnership with NGOs and international organizations set up refugee shelters in different areas of the main city (Boa Vista) close to the border with Venezuela in the amazon state of Roraima. In addition, Indigenous Venezuelan migrants (primarily agricultural and fishing communities whose main language was not even Spanish) were assigned to specific shelters. This generated additional variation in the cultural composition of locals’ exposure to migrants. Finally, Venezuelans were also granted access the labor market and public goods (health, education and cash transfers). This reception policy was considered by UNHCR and scholars a modernized version compared to the secluded rural camps and restrictive rights typically offered by other developing countries. So, this setting also provides an additional interesting feature: (5) It is a relevant policy to study!

Data and Method

In terms of data, I explore election results at the geolocated pooling station level. To evaluate the effects on crime, commonly present in political rhetoric regarding concerns about inflow of migrants, I use geo-referenced address level reports of robbery, homicide, and assaults. Finally, I explore yearly data on public schools’ students’ composition and congestion measures (classroom sizes and student per teacher ratio), to evaluate shelters impact on a relevant locally share public good.

I explore a difference-in-differences design comparing areas located closer to refugee shelters with those further away. Shelters’ location was mainly decided by the Brazilian military and selected sites with physical space to host migrants, such as public building parking lots or gymnasiums.  The identification assumption consists of the parallel trends in the studied outcomes and is supported by the event study estimates of pre-treatment effects.

Results

According to the results, locals living close to shelters reduced support for the incumbent governor (who participated in the reception efforts) while increasing votes for far-right presidential and governor candidates (Jair Bolsonaro and his local ally, Antônio Denarium) by 2 to 4 percentage points. In contrast, I find no evidence that shelters influenced crime. Similarly, although shelters raised the share of Venezuelan students in nearby schools, they had no discernible impact on classroom size, student–teacher ratios, or school infrastructure.

Importantly, these political effects were driven entirely by the shelters hosting Indigenous Venezuelans. They are an especially vulnerable group, facing higher adult illiteracy, lower vaccination rates, and lower levels of integration (social security registration and work permit possession). Moreover, despite Indigenous shelters housed relatively more children, surrounding schools did not experience a corresponding rise in Venezuelan enrollment. Together, these results indicate that the salience of refugee presence and locals' exposure to poverty and vulnerability, such as children out of school and child labor, may underlie the political response. Indeed, in conversations with residents of Boa Vista suggest that one of the most striking changes following the inflow was the visible increase in homelessness and the presence of out-of-school children begging in public spaces, all of which is more likely to be experience closer to Indigenous shelters.

Finally, estimates from a municipality-level synthetic control indicate that Boa Vista’s overall far-right voting closely matched those of non-affected comparable municipalities. This suggests that shelters effects were not large enough to shift aggregate electoral outcomes. This finding also underscore how aggregate data can hide important nuances in natives' attitudes towards migrants.

What we learned:

This paper reveals that political reactions to migrant reception can be highly localized and that establishing a refugee shelter might be less important for political backlash than who is housed. In other words, cultural differences, intrinsically connected with vulnerability, can dominate other economic channels in shaping local political responses to migration inflows. Finally, this paper's results also document a trade-off to tailored and culturally specific reception assistance: while they can address the particular needs of certain groups, depending on their logistics, they can also enhance perceived social distance and trigger stronger locals' reactions.

Job Market Paper URL: Click here to read the full paper

About the Author

Carlos Brito is a Ph.D. candidate at the University of California, Davis.

His research focuses on labor and development, migration, political economy, and education.  To learn more about his research, visit: https://www.carlos-brito.com/

Social Media Handles:

LinkedIn: https://www.linkedin.com/in/carlos-brito-28513b1b5/

Twitter/X: @Carlos_H_Brito