Research

Regional exposure to Chinese import competition has often been linked to support for the Leave option in the 2016 UK EU membership referendum. Looking at 143 harmonised International Territorial Level 3 (ITL3) regions covering England and Wales, and using data on the density of online social ties between them, I show that regional support for leaving the EU was also associated with exposure in socially connected regions. I first delineate 18 commuting zones based on interregional flows over three Census years. For each region, I then construct a measure of own exposure to Chinese import competition and a measure of exposure in a set of social neighbours located outside its commuting zone. Exploiting variation within commuting zones, and using an instrumental variable approach, I find that the two measures have comparable positive effects on the regional share of the Leave vote. In a series of checks, I do not find evidence that the effect of social neighbours’ exposure is driven by an economic channel or a relationship between import competition and social ties. I also corroborate the regional results using survey data on vote choice. I interpret these findings as indicative of social spillovers between local labour markets: information flows from social neighbours are a likely channel behind the estimated spillover effects on voting outcomes.

Partisan segregation in the United States is often interpreted as evidence of limited social interaction among out-partisans, or partisan homophily. In this paper, I draw on 2020 US presidential election results and data on the pairwise density of social ties between the populations of 22,537 zip code tabulation areas (ZCTA) to examine how different areas are socially connected to politically similar others. Using the local Moran index, I first identify clusters of ZCTAs where there is evidence of partisan homophily or heterophily. In a series of multinomial logistic regressions, I then also examine differences in the probability of each cluster across different settlement types and regions, and across areas with differences in the relative connectedness and geographic distance to others. I find that partisan homophily is the norm across areas, broadly tracking partisan segregation along the urban-rural continuum. However, the populations of Democratic-leaning areas, which are most likely to be in cities and suburbs, are on average likely to have more of their co-partisan social ties in relatively distant areas when compared to the populations of Republican-leaning areas. This highlights the prospect of partisan differences in the role of non-local context in local political outcomes.

All-mail voting has been introduced in several jurisdictions in the United States since the COVID-19 pandemic, with the resulting need for processing larger volumes of mail ballots posing challenges to election administration. Leveraging data on more than 4.5 million eligible voters in the state of North Carolina and on online social ties between zip code tabulation areas (ZCTA) across the country, I show that local rollouts of all-mail voting also have spillover effects on choice of voting method in distant areas. Using a difference-in-differences research design, I find that an increase in the ZCTA-level share of social ties in counties that switched to all-mail voting between the 2016 and 2020 presidential elections by one standard deviation (1.77 percentage points) corresponds to an increase in the probability of casting a mail ballot by roughly 3.8 percentage points. These socio-spatial spillovers are stronger for older voters, non-Republicans, and those residing in metropolitan counties. The findings suggest that considering the aggregate social ties of a local jurisdiction in other areas may aid election officials in planning for changes in the demand for mail ballots.

Tolbert-Sizer commuting zones based on harmonised local authority district (LAD) commuting matrices from the 1991, 2001, and 2011 UK Census.

Harmonised classification of local authority districts (LAD) that is stable over time and consistent with official data on commuting and migration.