Trump Has No Room for Error in 2020

Changes in the electorate are putting the squeeze on the president.

President Trump speaks into a microphone at a campaign rally as supporters, one wearing a white cowboy hat, look on.
Andrew Harnik / AP

The risk in Donald Trump’s base-first electoral strategy is only rising—because the size of his base is shrinking.

Working-class whites are on track to continue declining as a share of eligible voters in 2020, according to a study released today by the liberal think tank Center for American Progress. In turn, two groups much more resistant to Trump will keep growing: Nonwhite voters will swell substantially, while college-educated white voters will modestly increase.

These shifts in the electorate’s composition may seem small, but they could have big implications next year. The report projects that these demographic changes alone could provide Democrats a slim Electoral College majority by reversing Trump’s narrow victories in the three blue-wall states that keyed his 2016 victory: Michigan, Wisconsin, and Pennsylvania. Indeed, the shifts could be enough to narrowly tip these states back toward the Democrats even if college- and non-college-educated whites and minorities behave exactly as they did in 2016—if their turnout rates stay the same and if they split their votes between Trump and the Democratic nominee in exactly the same proportions as they did then.

And, if all other voting patterns hold equal, these changes alone could add another percentage point to the Democratic nominee’s margin of victory in the national popular vote, giving that candidate an advantage over Trump of more than 3 points. All told, the study, provided exclusively to The Atlantic, underscores how narrow a pathway the president is following headed into 2020.

Ruy Teixeira, the veteran Democratic election analyst who co-wrote the study with CAP senior fellow John Halpin, cautions that relying on demographic change alone to recapture the three blue-wall states would leave Democrats with virtually no room for mistakes if Trump improved on his 2016 performance there. But, Teixeira notes, the continually changing electorate means that Trump, both nationally and in battleground states, will likely need to squeeze more advantage from working-class whites—through either higher turnout or a bigger margin of victory—just to offset his weakness with the groups that are growing.

If Democrats “are relying on demographic change to win, you are praying nothing else will change, and you are hoping to squeak out a victory by the tiniest margins,” Teixeira says. But it is a “bonus” for Democrats, he explains: “Even if [Trump] accomplished exactly what he did in 2016, he would probably lose the election. So he not only has to do that; he has to try to do more.”

Much remains unknown about the 2020 election, of course, including which Democrat will win the nomination, whether Trump will survive impeachment, and how it will affect his political standing if he does. But the study addresses one of the crucial baseline questions that will shape the coming election regardless of those other dynamics: How will the universe of Americans eligible to vote in 2020 differ from 2016?

To answer that question, the authors rely on demographic projections for the eligible voting population from the nonpartisan States of Change project, a joint effort by CAP, the Brookings Institution, the Bipartisan Policy Center, and the Democracy Fund Voter Study Group.

The composition of the population eligible to vote differs from the composition of those who actually vote, for two reasons: Eligible whites, especially those with college degrees, turn out at a higher rate than eligible minorities, and older people vote at a much higher rate than the young. That means eligible minorities and non-college-educated whites usually comprise a smaller share of actual voters, while college-educated whites, whose turnout levels exceed the other two groups, consistently represent many more actual than eligible voters.

But the eligible and actual measures have moved in tandem over recent decades, with non-college-educated whites consistently decreasing as a share of both groups, and both minorities and college-educated whites increasing. Even Trump’s strong appeal to working-class whites in 2016 only slightly altered these long-term trends. The new report concludes that those patterns will largely continue through 2020, complicating the electoral puzzle facing Trump, but also leaving Democrats on a slim ledge.

The study forecasts that whites without a college degree will decline as a share of eligible voters in 2020 by 2.3 percentage points. It anticipates increases for Latinos (1.3 percentage points), Asians and other minorities (0.6 percentage points), and African Americans and college-educated whites (both at 0.2 percentage points). That could translate into an actual electorate in which college-educated whites hold steady at about 30 percent of voters, non-college-educated whites slip to about 42 percent (from 44 percent in 2016), and minorities rise to make up the difference.

These changes are expected to unfold consistently across all the key battleground states. Not surprisingly, the study anticipates that the biggest decline for working-class whites, and the biggest growth for minorities, will come in Sun Belt states that both parties are targeting. It forecasts that the share of eligible working-class white voters will decline by more than the national average in Nevada (3.2 percentage points), Arizona (2.8 points), and Texas (2.5 points); and by about 2 percentage points each in Florida, North Carolina, and Georgia. In the former group—all southwestern states—the study forecasts that Latinos alone will increase by at least 2 percentage points. In the latter group—all southeastern states—it anticipates that Latinos and African Americans will combine for about a 1.5 percentage point increase in eligible voters. (Notably, in all three southeastern states, growth among Latinos exceeds that of African Americans.)

Even more concerning for Trump, Rust Belt states, which are less diverse, older, and growing more slowly than others, will not be immune to these trends. The report projects that non-college-educated whites will shrink as a share of eligible voters in all of the key midwestern battlegrounds, with Pennsylvania and Wisconsin recording the biggest declines (each at 2.3 percentage points), followed by Michigan, Minnesota, and Iowa (all around 2 points). In those states, growth among minorities is not as large as in the Sun Belt states; instead, the study anticipates significant increases for college-educated whites, a group that has moved away from Trump during his presidency.

Across the Rust Belt, “the non-college whites are getting older, and the younger cohorts tend to be more educated and more diverse,” Teixeira says. “That’s enough to push the white non-college share down inexorably in these states, despite the fact that they are not particularly dynamic in a demographic sense.”

Beyond the three battlegrounds Trump dislodged from the blue wall, the study does not project that demographic changes alone would tip any other Trump-won state back to the Democrats. But it does forecast that these shifts would bring Democrats closer in Arizona and North Carolina (reducing Trump’s advantage to 2.5 and 2 percentage points, respectively). Shifting demography would also cut his margin in Florida in half, to less than 1 point. By contrast, Trump’s advantage would remain larger in Georgia and Texas, underscoring the need for Democrats to also make gains in those states with white voters.

These results indicate that Democrats can’t solely rely on demographic change to carry almost any of the states both sides are contesting in 2020. (Even in the three blue-wall states, Democrats would have only minuscule advantages of about half a percentage point in both Michigan and Pennsylvania and only one-tenth of a point in Wisconsin.) The study runs several scenarios for how Democrats could improve on their 2016 performance, from expanding their margins among college-educated whites; to expanding both turnout and the Democratic advantage among African Americans, both of which dipped in 2016; to eroding Trump’s big lead among whites without a college degree.

Teixeira says the study wasn’t designed to pick one of those alternatives as the best course for Democrats, and that the precise recipe for victory will vary from state to state. But, on balance, Teixeira believes that the party’s shortest path back to the White House would come from focusing on recapturing Michigan, Pennsylvania, and Wisconsin by slightly eroding Trump’s margin with working-class whites there, while treating Arizona as the principal backup plan if Trump holds Wisconsin.

“It does seem like the right path,” he says, given how close those states were in 2016, “looking at the 2018 results, [and] looking at how small the change would have to be among white non-college voters to undercut Trump’s chance of winning.”

Given the resistance Trump has faced from minorities and college-educated whites, especially women, the report’s findings highlight how much he will depend next year on maximizing both turnout and his margins among white voters without a college degree. Yet both goals could prove challenging. Though Trump remains extremely popular with non-college-educated white men, an array of polls shows him suffering erosion among white women without a college degree, a group that notably moved away from the GOP in 2018.

And even if Trump, as he’s shown the capacity to do, can further juice turnout among non-college-educated whites, especially outside of urban areas, those gains could be offset by higher engagement among the groups that are growing in the electorate. That’s what happened last year: Working-class whites turned out at elevated levels for the midterms, but that increase was swamped by a spike in participation among college-educated whites and minorities. As a result, blue-collar whites plummeted by 4 percentage points as a share of the total vote compared with the 2014 midterms, according to calculations by Michael McDonald, a University of Florida political scientist who specializes in voting behavior. That’s an unusually large decline.

Something like that could happen again in 2020, Teixeira notes. “They can shrink [more] if everyone else comes out—that is the nature of the beast,” Teixeira says. “This is part and parcel of the challenge that Trump will face. He is utterly dependent on this group, but he can’t really stop their ongoing decline as a share of voters … In a high-turnout election, things may turn out even worse” for them, given the growth among other demographic groups.

The report makes clear that Trump could again manage a narrow Electoral College win, even if he loses the popular vote. (Teixeira believes that the demographic trends virtually ensure he won’t win the most voters overall.) Non-college-educated whites, even after their expected decline, will still represent a much larger share of the total vote in the key Rust Belt battlegrounds than they do nationally. And in many of the Sun Belt battlegrounds, Democrats have struggled to turn out growing minority populations, failed to match their gains elsewhere among college-educated whites, or both. The strong economy might also allow Trump to suppress defections among college-educated white men and possibly make some gains among black and Latino men.

Taken together, these factors could mean, as I wrote on Election Day in 2016, that the Democratic coalition in the Rust Belt again crumbles faster than it coalesces in the Sun Belt, allowing Trump to break through to another win. But the study also pinpoints the inexorable math pressing on Trump’s exclusionary vision for the GOP: His most likely path to a second term will require him to squeeze even bigger advantages out of dwindling groups.

Trump’s incendiary populism has been unable to reverse the tides of demographic change that, like an ocean encroaching on a beach, are diminishing the constituencies most drawn to him. And that unstinting current of change leaves him with even less margin for error in 2020 than in 2016.