One of the most persistent claims about Donald Trump’s rise is that it’s a response to economic anxiety and struggles among white working-class Americans.
This is a comforting notion, particularly for those on the left. It suggests that large numbers of Americans are not being drawn to a racist demagogue because he’s a racist demagogue, but because of the failures of modern capitalism. It also implies that these voters could be won over by a robust left-wing economic agenda that addressed their plight.
But there’s also plenty of data suggesting that this isn’t a satisfying explanation. Trump’s rise has always seemed more closely related to prejudice than economics. Analysis of surveys has shown consistently that racial resentment correlates more strongly with Trump support than one’s income or degree or pessimism about the economy.
Now, Gallup’s Jonathan Rothwell (via the Washington Post) has offered a particularly detailed argument that there’s a lot more than pure economic anxiety at work here. Gallup's regular surveys offered Rothwell a large dataset of 87,428 Americans who told pollsters whether they held a favorable or unfavorable view of Donald Trump. That sample size let him drill down geographically, analyzing support for Trump at the regional and local level — even, on some questions, down to individual zip codes.
He then linked that geographic information to data about the decline in manufacturing, about how affected each area was by the rise of Chinese imports, about intergenerational mobility, about racial segregation, and about white mortality rates.
That let him test each of those factors as explanations for Trump. Is Trump support correlated with areas affected by globalization, as many commentators have suggested? Check the Chinese import data. Does Trump support coincide with increased death rates for white middle-aged women, an increase chronicled by Princeton economists Anne Case and Angus Deaton? Check the health data.
Trump's base is not poor whites — it's way more complicated than that
What Rothwell found was revelatory, to say the least. He finds that individuals who are struggling economically are not more likely to support Trump, nor are people living in areas that have suffered a loss of manufacturing jobs, an influx of immigration, or competition from China. By contrast, people in areas where whites are struggling health-wise, and in terms of intergenerational mobility (and in areas that are very racially segregated), do seem more likely to back Trump.
Trump supporters are richer, not poorer, than average: For one thing, Rothwell found that both across the overall population and among whites, support for Trump is correlated with higher income, not lower. That’s not surprising; low-income people have always preferred Democrats. But it definitely contradicts the image of Trump as spokesman for the economically struggling.
Rothwell also found that Trump supporters are no likelier to be unemployed or to have left the workforce. The problem of men dropping out of the labor force doesn’t seem to be a factor behind Trump’s rise.
"The individual data do not suggest that those who view Trump favorably are confronting abnormally high economic distress, by conventional measures of employment and income," he concludes.
Nonetheless, Trump supporters tend to be blue-collar and less educated: On the other hand, Rothwell also finds that Trump supporters are more likely to work in blue-collar fields and to have less education. This fact, however, sits uneasily with Trump’s greater support among the wealthy and lower support among the poor, and suggests that his sweet spot is less-educated people in blue-collar fields who are nonetheless doing pretty well economically.
Trump does well in racially segregated areas: Turning to the geographic data, Rothwell finds that segregated, homogenous white areas are Trump's base of support. "People living in zip codes with disproportionately high shares of white residents are significantly and robustly more likely to view Trump favorably," he writes. "Those living in zip codes with overall diversity that is low relative to their commuting zone are also far more likely to view Trump favorably." Put another way: If you're in the whitest suburb in your area, you're likelier to back Trump.
Trump doesn’t do well in areas affected by trade or immigration: This is perhaps the most surprising finding. Contact with immigrants seems to reduce one's likelihood of supporting Trump, as areas that are farther from Mexico and with smaller Hispanic populations saw more Trump support.
Areas with more manufacturing are significantly less likely to support Trump. An increase in the level of manufacturing employment from 2000 to 2007 predicted higher Trump support — which is the opposite of what you'd expect, given the narrative around this campaign. While the finding isn't statistically significant, greater exposure to Chinese imports predicts lower support for Trump, despite his agitation for higher tariffs on the country.
Trump-friendly areas are struggling in other ways: While individual Trump supporters appear to be doing pretty solid economically, they tend to live in areas that are struggling on two important dimensions.
Rothwell finds that Trump support increases mildly in areas with lower intergenerational mobility, as measured by data from economist Raj Chetty and his team. "This is not meant to suggest that with undue certainty that growing up in a place that causes lower social mobility causes Trump support," Rothwell clarifies. "This analysis only identifies the correlation."
Much stronger is the relationship between Trump support and higher regional white mortality. Overall mortality is also predictive, but nowhere near as much as white mortality, and particularly middle-aged white mortality.
The findings on mobility and white middle-aged mortality are commuting zone level, connected to the wider region in which the respondent lived rather than their specific municipality or neighborhood. That makes it hard to draw too many fine conclusions about how well that regional data reflects the circumstances of the respondent’s own life.
Some commentators have speculated that this result implies parents are worried for their children based on how their area as a whole is doing (even if they personally are doing fine). There’s nothing in the study suggesting this is the mechanism, but it’s potentially plausible.
4. Parents worried about their unemployed kids. Grandparents raising their grandkids bc the nuclear family fell part. Etc.— Ross Douthat (@DouthatNYT) August 12, 2016
Also plausible is basically the opposite conclusion: These are relatively well-off less-educated blue collar workers who see poorer blue collar whites who are suffering (as indicated by low mobility and poor health in the region) and view them as undeserving recipients of government aid. That’s less a story about personal anxiety and more one about class politics between the petit bourgeois and the proletariat.
But the basic point is the straightforward story of Trump supporters as poor whites abandoned by the loss of manufacturing to China is not the case. The story is, at least, much more complicated.
"I find only limited support that the political views of US nationalists—as manifest in a favorable view towards Donald Trump—are related to economic self-interest," Rothwell concludes. "If so, the self-interest calculation must go beyond conventional economic measures to include one’s physical health and inter-generational concerns."