Cosmopolitan Cities? A Natural Experiment on Anti-Immigration Vote in Cities

• In Event: Comparative Politics is Local: Local Politics in the Comparative Perspective

Abstract

Immigration is a salient topic in the political agenda in the United States and Western Europe. However, the attention devoted to immigration seems far from being a temporary phenomenon. Attention has steadily increased since the mid-1970s, across the political spectrum and partly independently from the emergence of extreme right-wing parties (Alonso and Fonseca, 2012; Mudde, 2013; Dancygier and Margalit, 2018). This study aims at understanding the largely neglected effect of immigration on political behavior in urban contexts.

Specifically, I aim at understanding if immigration influences the political demand and supply of anti-immigration policy positions in cities: (i) does immigration cause larger anti-immigrant vote? (ii) Does immigration cause a larger supply of anti-immigrant candidates?

Most studies investigate the effect of immigration on voting behavior using aggregate municipality data and tend to find a negative correlation between municipality size and anti-immigrant sentiment (e.g. Dustmann et al. 2016). However, two main identification problems persist. First, the location of migrants is endogenous to other characteristics, that per se trigger patterns of political behavior. Second, the lack of fine-grained data does not allow to distinguish aggregate from dis-aggregate effects (Enos, 2016), inducing ecological inference. Both problems are exacerbated when studying cities.

I exploit a natural experiment across French municipalities, consisting of a legal population-based discontinuity in the provision of social housing, and study the 2012 French legislative elections. Through a regression discontinuity approach with mediation, I show that municipalities that increased social housing between 2000 and 2012 also experienced an increase in the immigrant population, inducing different political behavior. Mediation analysis allows disentangling the direct effect of social housing from the indirect one that goes through migration.

The policy has a number of important features for my research design. First, interested municipalities are located in large urban areas, with no geographic discontinuity from the main city (and are often qualified as the suburbs of large cities). Second, social housing is proportionally more used by immigrant households, who tend to have higher geographical mobility and accede to house ownership later in life. Third, social housing is not exclusively targeted to the poorest part of the population but is designed for extremely poor to middle-income households.

The empirical strategy is composed of three steps (combining the methodologies by Dippel et al. 2018 and Calonico et al. 2017): (i) showing the effect of social housing on immigration, (ii) showing the direct effect of social housing on political behavior, (iii) showing the indirect effect of social housing on political behavior, through immigration. The third step is the one that allows establishing a causal effect of immigration on political behavior, where I exploit the exogenous variation in immigration induced by the policy.

Preliminary findings suggest that immigration induces anti-immigration political behavior in cities. The effect exists for both political demand and supply. However, the result for political supply is more clearcut: when immigration increases, the supply of anti-immigrant candidates responds positively. Voting behavior seems to follow.

Author

• Gloria Gennaro, Bocconi