Website CV Paper
Fields
Political Economy, Urban & Regional
Job market paper
Where Do Protests Win or Lose Votes? Exposure and Participation in the George Floyd Movement
This paper asks how protests affect voting in the neighborhoods where they take place and in those where their participants live. In the 2020 George Floyd movement, I find that two effects coexist and pull in opposite directions: neighborhoods near protest sites shifted toward the Republican candidate—a backlash—while those where participants lived shifted toward the Democratic candidate—a mobilization. To separate them, I combine 7,920 geocoded protest events with mobile phone location data and measure for each block group in American metropolitan areas its exposure to nearby protests and its residents’ participation. I then exploit differences in exposure and participation across block groups within the same zip code to estimate the effect of each on the change in vote shares between the 2016 and 2020 presidential elections. The backlash decays with distance and its point estimates are larger near violent protests, while the mobilization comes mostly from participation in nonviolent protests. Both results hold in a difference-in-differences design that follows the same neighborhoods across the presidential elections from 2012 to 2024, with no detectable pre-trend before the protests. Taken at face value, my estimates imply that participation moved about 137,000 votes toward the Democrats across the sample, while exposure within three kilometers of protest sites cost them about 87,000; adding the effect of exposure farther away, the net is not distinguishable from zero.