After every national election, including the most recent presidential election, it’s tempting to explain it with stories about local places. Many of the 2024 stories presume that there was something unusual or unique about that specific place that led Trump to make gains.
Trump’s gains in large urban areas have drawn a lot of attention, especially because those cities are reliably Democratic. A prominent theory emerged: More voters in those areas backed Trump because they were dissatisfied with local governance – high housing costs, too much crime and disorder, etc.
What actually hurt Democrats in big cities
But, in fact, that theory looks to be mostly untrue. In an extensive analysis published last week in the New York Times, Emily Badger and Alicia Parlapiano looked at dozens of urban areas and could not find much correlation between indicators of governance and shifts in presidential voting between 2020 and 2024. Here are their findings, in a nutshell:
To test whether those ills were related to the rightward shifts, we compiled data on the local cost of living; growth in rents and home prices; new housing permits; and homelessness. We looked at where migrants have settled, using court records from federal deportation cases. We considered long pandemic school closures. To identify where residents may have been disaffected enough to pack up and leave, we measured which urban counties had lost population share since 2019. We also considered the places hit hardest by Covid deaths.
Then we modeled whether these factors were able to predict how an urban county voted in the election.
Some data was incomplete, and many of these variables overlap, making it difficult to disentangle their effects statistically. When we sized them up individually, only some of these factors correlated with election results. After we controlled for race, those effects faded.
That last sentence is crucial. It turns out that the best predictor of Trump’s gains in urban areas was simply how many racial and ethnic minorities lived there. Trump’s gains with those voters helped him more in urban areas than other places. Here’s a graph from Badger and Parlapiano:
Now, you might be thinking that poor governance is the reason that Democrats lost vote share among racial and ethnic minorities. But because Democrats lost vote share with these groups in lots of different areas around the country – not just big cities – that seems unlikely.
In short, the 2020 to 2024 swings were more about people than places.
Other analyses confirm these findings
The findings in this New York Times analysis are consistent with other analyses of local voting patterns in 2024.
For one, Jed Kolko found that local economic conditions didn’t matter much:
In all, local economic conditions played at most a minor supporting role in explaining why some counties swung more or less toward Trump in this election relative to the previous one. Places with higher unemployment and a higher cost of living swung more toward Trump, but the racial/ethnic mix, educational attainment, and density were all more important factors.
This is consistent with the academic literature as well. Some studies have found relationships between local economic conditions and presidential voting. But as Andrew Healy and Neil Malhotra note in their review of this literature, the effects tend to be small.
Second, Kolko also found that migration flows were not strongly correlated with vote swings between 2020 and 2024. You might think that places that experienced a large uptick in new immigrants would swing toward an anti-immigrant candidate like Trump, but that didn’t really happen.
Again, this is consistent with previous research. The swings between the 2012 and 2016 presidential elections – which propelled Trump to victory the first time – were also not correlated with local trends in immigration. In general, as this research and other scholarly commentary make clear, it is not that the arrival of new immigrants has zero implications for people’s political attitudes or voting behavior. But there is no simple relationship between local immigration patterns and recent U.S. election outcomes.
Why local factors may not matter much in national elections
Of course, the point isn’t that what’s happening at the local level doesn’t matter at all. These changes and challenges certainly matter to the people experiencing them. Rather, the point is that local changes may not affect how people vote in national elections. Why?
In the New York Times article, I’m quoted talking about how elections are nationalized. Increasingly, U.S. politics hinges on national issues and trends, as opposed to the unique circumstances of localities. Political scientist Dan Hopkins wrote a nice book about this.
In 2024, the central concern for voters had to do with the national economy due to the long shadow of the 2021-2022 inflation. The politics of immigration are also pretty nationalized.
I think that’s why, as I noted the day after the election, Trump’s gains were so widespread.
There is a second factor beyond nationalization: Voters may not attribute local problems to national officials. Imagine that I live in Manhattan and I’m furious because, say, the subway sucks. Who should I blame for that? The mayor, Eric Adams? The governor, Kathy Hochul? Joe Biden? For local conditions to matter for incumbent officeholders, voters need to attribute those conditions to those officeholders. Attribution is a complicated task, especially given the layers of governance in our federal system. In their review, Healy and Malhotra have a good discussion of the importance, and challenges, of attribution.
The process of attribution ultimately depends on how credit and blame are constructed, which has to do with politics and political narratives. If somehow news stories about violence on the New York subway directly implicated Joe Biden and the national Democratic Party, then I could see how those conditions could have worked to Harris’ detriment, costing her votes in Democratic-run cities.
But did that process of attribution happen? It’s not clear that it did. And that makes sense. A lot of the measures that would most directly address local problems are local policies, not federal policies.
And so, if local officials try to fix those problems, they may benefit at the polls. But it’s not clear that a president hundreds or thousands of miles away in Washington, DC, would also benefit.
