The America That Votes in 2020 Will Look Radically Different Than it Did in 2016

The day after Hillary Clinton announced in March that she would not make a third bid for the White House, President Trump expressed in a tweet his mock disappointment that 2020 would not be a rematch. “Aw shucks,” he wrote, “does that mean I won’t get to run against her again? She will be sorely missed!” (As something of an authority on the Mean Girls mythology, I can’t help but remind you that Clinton fired back with a GIF from Mean Girls in which she pondered, “Why are you so obsessed with me?”)

In fact, even if Trump had gotten his wish to “run against her again,” the contest would not have been a rematch for one simple reason: it would not have been played in the same stadium.

When gaming out possible 2020 election outcomes, particularly in states where there was a razor-thin margin of victory last time around, one tends to start with how a growth or decline in the turnout of various demographic groups could shift the scale given their partisan leanings. But equally important is the fact that four years can make a tremendous difference in the proportions of those demographics in any given state, as people relocate and new potential voters age into eligibility for the first time.

Indeed, a TIME analysis of the voting-eligible population (VEP) from the Census Bureau’s annual American Community Survey found that the demographics of many states crucial to both parties’ odds of victory in 2020 are evolving rapidly. This churn among eligible voters — many of whom, to be fair, are not yet registered — will be especially critical to the campaigns next year because, more than a year out, there remains a wide variety of paths to victory for both parties. As TIME’s Brian Bennett noted this week:

To win, Trump probably needs to come up with a different set of states than those that garnered 304 electoral college votes and carried him to the White House: public polls show his disapproval ratings swamp his approval numbers by at least 9 percentage points in his 2016 blue-to-red trifecta of Michigan, Wisconsin and Pennsylvania.

For our purposes, the voting-eligible population is defined as citizens, either native-born or naturalized, who will be at least 18 years old on Election Day and are not institutionalized. While the United States Election Project estimates this eligible population more precisely, accounting for modifiers such as different state laws on whether convicted felons can vote, our state-by-state figures match its numbers closely and allowed us to study this population in detail. The data was gathered from IPUMS USA, where one can freely download unit-level, weighted Census data for precise analysis of a trove of variables, like whether a person has recently moved or what language they speak at home.

While the most recent data is from 2017, even the one-year developments since 2016 show a significant change in the electorate between citizens born before the fall of 1998, who were largely eligible to vote in 2016, versus those born before 2002 and largely eligible next year. (Sorry, those of you with Dec. 2002 birthdays.)

Every state with a sizable Hispanic population, for example, will see a small but significant growth in the percentage of eligible voters in that demographic. In Texas, which will host the third Democratic debate this week, we estimate that about 5,800,000 Hispanics will be eligible to vote in 2020 compared to 5,200,000 in 2016. That’s a nearly 2 percentage point increase when accounting for overall population growth. (The 2016 figure closely matches the Census’ flat tables if one is willing to do a small amount of addition.) This may not seem like a tremendous figure, but given that the Texas Democratic Party just released an ambitious plan to turn the state blue next year, 600,000 new voters in a left-leaning demographic is far from negligible.

Of course, this across-the-board growth largely comes from individuals who have recently aged into the voting population — an age group that is historically the least likely to vote. But there is a second factor, one that’s more difficult to cast forward to 2020 but introduces even more uncertainty: our calculations show that in 2017 alone, 2.4 percent of eligible voters moved to a different state, which is close to national figures for all residents. That’s a fairly consistent figure from year to year, but it is not equally distributed in terms of where people leave and where they end up.

In North Carolina, for example — a state where Democrats would love to repeat Barack Obama’s 2008 victory — an average of about 234,000 new potential voters have moved into the state each year since 2010, while 188,000 have departed. The incoming population is significantly more educated — 32% have at least a Bachelor’s degree — compared to the state’s overall education levels (again, just when considering the 2020 voting-eligible population.) This is good news for Democrats, given that a variety of surveys and polls indicate that voters with a college education break in their favor by a 21-point spread, according to a 2018 Pew Research Center study. A high-stakes special election this week in North Carolina’s 9th Congressional District, which Trump carried by 12 percentage points, went to the Republican candidate by 2 points.

We’ll be unpacking these migration and eligibility trends in more detail, particularly after the Census Bureau releases its 2018 figures later this month, which will be available for this level of microanalysis by the end of the year.

Whether a state imports or exports more voters, and what we can surmise about those transients — plus, what we know about the newly-minted young voters — could very well be a deciding factor in the 2020 election. We all know a Democrat is theoretically capable of far outpacing Trump in the popular vote. Whether he or she can win the election itself will have a great deal to do with where America’s voters live a year from now.