全球智库信息集成服务系统

This report contains corrections.

The recent elections of Donald Trump and Barack Obama were influenced in no small measure by shifts in the nation’s underlying demographic structure—the rise of communities of color, the increase in the number of older Americans, the sharpening of education divisions—and the distinctive voting behavior of these demographic groups. This 2018 report of the States of Change project, the fourth in an annual series,¹ examines an array of future presidential election outcome scenarios—from 2020 through 2036—that could arise as the demography of the nation and its 50 states changes over the next 18 years.

These scenarios, developed by the authors, include outcomes that favor both Republican and Democratic candidates. They are not intended as predictions but are simulations based on assumptions about different demographic groups’ future voting patterns. Each of the alternative scenarios assumes the same projections for the nation’s underlying demographic structure of eligible voters (EVs) with respect to race, age, and education attainment. As such, the scenarios provide for a more in-depth understanding than national or state polling trends can supply about how emerging voting patterns may interact with changes in the demography of the nation’s electorate to affect future popular vote and Electoral College outcomes.

Many analysts suggest that if current voting patterns remain the same as in recent elections, the projected rise of communities of color—Hispanics, blacks, Asians, and others—will favor Democrats as the Republican-leaning white share of the electorate shrinks. However, the aging of the population and the continued substantial political clout of whites without college educations played a key role in electing Republican Donald Trump. Because the demography of these latter groups differs across states in ways that tend to benefit Republicans, this report finds that quite a few future scenarios could mimic the result of the 2016 election—a Democratic win in the popular vote with a Republican win in the Electoral College.

The 2020 through 2036 presidential election scenarios presented here are of four types:

1. Assuming 2012 and 2016 voting patterns in future elections: Not surprisingly, a scenario that attributes the voting patterns of all groups from Obama’s 2012 win to future more racially diverse populations, yields solid Democratic popular vote and Electoral College wins from 2020 through 2036.

More surprisingly, changing demography has a clear impact on future outcomes when Trump’s 2016 voting patterns are attributed to the 2020 population. Here the modest shift toward more racially diverse voting populations in several states is enough to provide Democratic wins in both the popular vote and Electoral College—not only in 2020 but in subsequent elections.

However, the 2016 election result was unusual because of a high level of third-party voters. When those extra third-party voters are allocated back to one of the two major parties, based on underlying partisan preferences, projections to 2020 show a dead heat in the Electoral College.

2. Voting assumptions about communities of color: Several scenarios assume that future voting patterns of racial groups will differ markedly from those observed in 2016. Taking note that whites tend to have the highest voting turnout rates of all racial groups, a scenario that assumes that all racial groups turn out at the same rate improves the voting clout of racial minorities, especially Hispanics and Asians. When this assumption is made—while leaving other aspects of 2016 voting constant—Democrats win both the popular and Electoral College votes in 2020 as well as subsequent elections. Additional scenarios for blacks assume their 2012 turnout rates, their 2012 Democratic voting preferences, and both together—which factored into Obama’s election victory—continue in the future. In all three of those black voter scenarios, Democrats win both the popular vote and Electoral College in future elections.

However, there is one scenario here that yields Republican Electoral College—though not popular vote—wins: a pro-GOP margin swing of 15 points among Hispanics, Asians, and other nonblack racial minorities. But if these groups shift their margins by the same amount toward Democrats, the latter party wins by large margins in 2020 and future elections.

3. White college graduate versus white noncollege-educated voting preferences: The 2016 election showed a sharp divide in Republican voting between white college graduates and whites without college educations. Among the simulations examined, the greatest opportunity for Republicans to extend their 2016 victory model assumes an expansion of the already-substantial voting margin that the GOP has gained among white noncollege-educated voters. When this margin is expanded by 10 points, Republicans win both the 2020 Electoral College and popular vote. They continue to win the Electoral College—though not the popular vote—through 2036, despite broadening diversity and other predicted changes across the country.

Several other scenarios are explored by shifting future voting patterns of white college graduates and whites without college educations in different ways. While strong Democratic wins occur when either group shows improved Democratic support, it is clear that the most persistent gains for Republicans come from those scenarios where they improve their appeal to noncollege-educated whites.

4. Voter group trade-offs: Several scenarios assume that voting swings toward one party on the part of one demographic group could precipitate a backlash voting swing toward the other party among another group. One scenario postulates that an increased Republican margin of 15 points among Hispanics, Asians, and other nonblack minority groups, perhaps due to increased outreach efforts, might trigger a swing toward greater Democratic support—back to relatively good 2012 levels—among noncollege-educated whites. In such a trade-off, Republicans would be disadvantaged and lose both the 2020 popular and Electoral College vote. In another simulation, a Republican swing of 10 points among noncollege-educated whites is countered by Democratic swings of 10 and 15 points, respectively, among white college graduates and among Hispanics, Asians, and other minorities. This simulation also yields 2020 Democratic popular vote and Electoral College victories and bigger ones thereafter. But in a scenario where increased Republican success among white noncollege-educated voters is traded for increased Democratic success among white college graduates, the GOP does gain an Electoral College victory in 2020, even while losing the popular vote.

In many of the simulations that favor Democrats in the Electoral College, especially those beyond 2020, swing states that favored Republicans in the 2016 election turn Democratic. These include Florida, North Carolina, Georgia, and Arizona, in the Sun Belt, as well as Pennsylvania, Michigan, Wisconsin, Iowa, and Ohio in the North. But in scenarios where Republicans are Electoral College victors, they retain most of their 2016 swing states and often add new ones, including Nevada, Minnesota, New Hampshire, and Maine.

The scenarios in this report suggest that there are paths for both parties to win the Electoral College in 2020 and beyond. For Republicans, future success is tied to mobilizing their strength among whites without college educations—a still-substantial but shrinking portion of the electorate—while attaining gains among at least some growing demographic groups. A narrow Republican reliance on noncollege-educated whites would lead, at best, to continued popular vote losses and ever smaller Electoral College wins, which would eventually peter out.

While Democrats appear to have the advantage in future popular vote contests, their success in the Electoral College will likely require some combination of intensifying their support among voters of color and improving their margins among white, particularly white noncollege-educated, voters. This delicate balancing act will provide a challenge for the party that cannot be met by simply waiting for demographic change to reshape the electorate.

Demographics are not destiny, but steady and predictable changes to the electorate play an important role in defining the landscape of American elections. The authors have every reason to suspect that the composition of the American electorate will change dramatically over the next five presidential elections. These shifts—which vary considerably state by state—will force parties and candidates to recalibrate their strategies for success goinfg forward.

This report explores how these demographic changes could shape the next five presidential elections using national and state projections produced by the States of Change project. In a 2015 report,² this project presented long-term projections to 2060 of race and age profiles for the populations and eligible electorates of all 50 states. The authors have since supplemented these data with education projections, further segmenting the population into those with four-year degree and those without. This report focuses on what those projections imply for the presidential elections between 2020 and 2036 under different assumptions about future turnout and voter preference patterns.

How is the electorate changing?

The national story

First and foremost, the eligible voter (EV) population is becoming more racially diverse. Younger, incoming generations of Americans are more racially diverse than prior generations. As they come of voting age, they will slowly but surely alter the makeup of the electorate. Whites made up 69 percent of EVs in 2016—a figure expected to drop to 67 percent by 2020 and 59 percent by 2036. During this time period, the Hispanic population is expected to grow by 6 points—going from 12 percent in 2016 to 18 percent in 2036—while Asians and other racial groups grow by 3 points, or 7 percent to 10 percent. The share of EVs who are black will be mostly stable—rising less than 1 percent between 2016 and 2032.

Second, the population is aging. Those 65 years old and older will make up a larger share of EVs—going from 21 percent in 2016 to 22 percent in 2020 and to 27 percent in 2036—while those ages 18 through 64 will shrink. While seniors will continue to be less racially diverse than younger age groups over this period, white senior EVs as a share of all EVs will rise by only a single percentage point from 2020 to 2036. The overwhelming majority of growth in the 65 and older age category is going to come from people of color.

Finally, the electorate is becoming more educated. This change is particularly important among whites, where the political and behavioral differences between those with and without college degrees tend to be largest. While whites without a college degree made up 46 percent of EVs in 2016, this group is expected to drop to 44 percent by 2020 and 37 percent by 2036.

The state-by-state story

Like most trends, the changes occurring nationally are not evenly distributed. Some places are changing quickly and others much more slowly. This will naturally shape how quickly states respond politically to overall demographic trends.

Sharp increases in diversity can be highlighted by looking at those states where racial minorities should comprise more than 40 percent of the eligible electorate. In 2020, there should be six states over this threshold: four states—Hawaii, New Mexico, California, and Texas for the first time, as well as the District of Columbia—where minorities are more than half of EVs, plus Maryland and Georgia, where minorities will make up between 40 percent and 50 percent of EVs. By 2036, eight additional states should cross the 40 percent threshold: Arizona, Alaska, New Jersey, Nevada, Florida, Mississippi, New York, and Louisiana. Also four additional states should have majority-minority eligible electorates: Texas in 2019, Nevada in 2030, Maryland in 2031, and Georgia in 2036.

At the other end of the continuum, the number of states where whites exceed 80 percent of EVs should be reduced from 23 states in 2020 to just 11 states in 2036. In 2036, the 11 still-heavily white states should include the three upper New England states of Maine, Vermont, and New Hampshire; the Southern states of West Virginia and Kentucky; the Midwestern states of North Dakota, Iowa, and Wisconsin; and the Western states of Montana, Idaho, and Wyoming. Notably, by 2036, the traditionally heavily white state of Utah should no longer be part of this group, as racial minorities will comprise 25 percent of its EV population due to the projected dispersion of Hispanics and other racial groups throughout the state.

Looking more closely at the demographic projections of selected Southern and Western swing states, it is clear that the Hispanic presence in particular should become quite a bit stronger in several of these states, including Nevada and Florida, where Hispanics are projected to become more than one-quarter of the eligible electorate in 2036. Due to additional substantial gains by Asians and other races, Nevada’s white share of EVs should plummet from 58 percent in 2020 to just 45 percent in 2036. North Carolina and Virginia should maintain their sizable black electorates and also show significantly increased shares of other voters of color. This is also the case for Georgia—a swing state in waiting—whose 2036 eligible electorate will be more than one-third black and 15 percent Hispanic, Asian, and other nonwhite groups. And two other potential swing states, Texas and Arizona, display sharp drops in their white EV profiles, with substantial gains among Hispanics and other nonwhite races.

Several Northern swing states show smaller gains in diversity. By 2036, Iowa’s EVs will still be 84 percent white; Wisconsin’s will be 80 percent white; and Ohio’s will be 77 percent white. The populations of all three states, as well as several of their Northern counterparts, will also be older than their counterparts in the South and West. This could help the GOP consolidate their gains in these states, especially if white seniors maintain their voting preferences despite generational turnover in this group.

Potential influence of demographic changes on future elections

To assess how demographic change might affect future elections, the authors performed a number of different simulations, each of which assumes that the nation’s underlying EV population will change according to race, age, and education projections in every state. The simulations differ only in what voter turnout and Democratic and Republican preferences are assumed for race, age, and education groups in the various states. However, before diving in, it is worth considering the value of such an exercise and what can reasonably be learned from it.

First and foremost, these are simulations—not predictions. When talking about results under a given set of assumptions, the authors are not expressing the belief that this is what will happen in a given presidential election. At heart, these are thought experiments—revealing tomorrow’s contours under a certain set of assumptions.

Additionally, those assumptions will almost certainly be wrong in ways both big and small. While the expectations the authors have about the underlying race, age, and education makeup of the electorate are probably the soundest of their assumptions, those pertaining to voter behavior are more likely to stray from reality. This is not a problem unique to this endeavor—the future itself is inherently uncertain—but it is one worth keeping in mind. As such, the results presented here are best thought of as baselines rather than ironclad truths. 

This report investigates four types of scenarios to see how the electoral future might look under a variety of assumptions. These are:

1. If future turnout and party preferences by group match those of the most recent presidential elections
2. If turnout or support levels shift significantly among communities of color: blacks, Hispanics, Asians, and those of other races
3. If support levels shift significantly among whites with and/or without a college education
4. If there are trade-offs between different voting blocs with gains among one group resulting in losses in another 

The results of each simulation are discussed in detail below, but the top-line results for 2020 are displayed in Figure 2. These 2020 results may be summarized as follows:

Baselines

• 2012 baseline: Democrats would win the popular vote by 6 points, and the electoral vote would go to the Democrats 332-206—same as the actual 2012 election.
• 2016 baseline: Democrats would win the popular vote by 3 points, and they would take back Michigan, Pennsylvania, and Wisconsin to carry the Electoral College 279-259.
• 2016 baseline, coming home: Democrats would take the popular vote by a little under 3 points, but there would be a 269-269 tie in the Electoral College.

Communities of color

• Equal turnout by race: Democrats would carry the popular vote by 4.5 points and the Electoral College by 288-250.
• Black turnout or support like 2012: If black turnout or support or both are like that of the 2012 election, Democrats would win the popular vote by 4 to 5 points and the Electoral College with 294 to 338 electoral votes.
• Hispanics, Asians, and other races swing to GOP: Democrats would narrowly win the popular vote by .8 points, but the GOP would win the Electoral College easily by 315-223.
• Hispanic, Asians, and other races swing to Democrats: Democrats would carry the popular vote by 6 points and the Electoral College by 319-219, including flipping Arizona.

Whites by education

• White noncollege-educated voters swing to GOP: Republicans would carry the popular vote by 1 point and win the Electoral College easily by 329-209, flipping states such as Nevada and Minnesota.
• White noncollege-educated voters swing to Democrats: Democrats would win the popular vote by more than 7 points and dominate the Electoral College by 350-188, including flipping Arizona and Georgia.
• White noncollege-educated voter support like 2012: Democrats would carry the popular vote by almost 6 points and win the Electoral College 347-191, including taking back Ohio and Iowa.
• White college-educated voters swing to Democrats: Democrats would win the popular vote by more than 6 points and win the Electoral College by 334-204, including a flip of Arizona.
• White college-educated voters swing to GOP: Democrats would still carry the popular vote very narrowly by .2 points, but Republicans would win the Electoral College by 323-215, flipping states such as Maine and Minnesota.
• White college-educated voter support like 2012: Democrats would win the popular vote by only a little more than 1 point and the GOP would narrowly carry the Electoral College 273-265.

Voter group trade-offs

• Hispanics, Asians, and other races for white noncollege-educated voters trade-off: In a scenario where Republicans trade more new voter of color support for less white noncollege-educated voter support, the Democrats would carry the popular vote by more than 3 points, and the GOP would lose the electoral college 279-259, as Rust Belt states flip back to the Democrats.
• White college graduates for white noncollege-educated voters trade-off: In a scenario where Republicans trade more white noncollege-educated voter support for less white college support, the GOP loses the popular vote by a little under 3 points but carries the Electoral College 309-229.
• Hispanics, Asians, and other races and white college graduates for white noncollege-educated voters trade-off: In a scenario where Republicans trade more white noncollege-educated voter support for less support among Hispanics, Asians, and other races as well as white college graduates, Democrats would win the popular vote by 5 points and the Electoral College 309-229, including a flip of Arizona.

The wide range of scenarios considered here mostly have Democrats in 2020 maintaining and, in many cases, strengthening their popular vote victory from 2016. Indeed, in only two cases do the authors actually see a Republican popular vote victory in 2020: a 10-point pro-GOP margin swing white noncollege-educated voters and a 10-point pro-GOP margin swing among white college graduates—and, in the latter case, only if the third-party vote is reallocated.

Since Democrats registered popular vote advantages in almost all scenarios in 2020, it should be no surprise that they do so for later elections as well. In the projections that show a Democrat popular vote advantage in 2020, Democrats achieve even greater margins in each subsequent election as the projected demographic makeup of the eligible electorate continues to shift in a direction generally favorable to Democrats.

But, critically, it is electoral votes based on state outcomes, not the nationwide popular vote, that determine the winner in presidential elections. As this discussion details, many Democratic popular vote victories in these simulations do not translate into Democratic electoral vote victories.

In the 2020 election, these simulations include a scenario where Republicans gain a 15-point margin swing in their favor among Latinos, Asians, and those of other races, and a number of scenarios where the education gap among whites plays a key role. The following scenarios result in a GOP Electoral College victory but a popular vote loss: The GOP gets a 5-point margin swing from white noncollege-educated voters twinned with an equal swing toward the Democrats among white college-educated voters; a 10-point swing in Republicans’ favor among white college graduates; and a reversion to 2012 support margins among white college-educated voters. The exception to this pattern is the scenario in which Republicans gain a 10-point margin swing from white noncollege-educated voters, where the GOP carries both the Electoral College and the popular vote. Finally, simply leaving turnout and voter preferences as they were in 2016 while demographic change continues, yields a probable Republican Electoral College victory—though popular vote loss—if the third-party vote reverts to 2012 levels.

Thus, the GOP has many roads to the presidency in 2020 even though demographic shifts appear to make a Democratic popular vote victory easier than ever to obtain. Even more interesting, some of these fruitful scenarios continue to produce Republican electoral vote triumphs in 2024 and beyond, despite mounting popular vote losses. Their strongest future prospects are linked to a widening education gap among whites in their favor. Conversely, the brightest prospects for the Democrats are linked to an education gap among whites that widens in their favor, though they would also benefit greatly from stronger turnout and/or support among voters of color.

Detailed discussion of all scenarios is provided below.

Baselines

2012

The 2012 presidential election was a good year for Democrats. President Obama won the national popular vote by about 4 points and handily carried the Electoral College at 332-206. Notably, it marked a high point for black turnout, which surpassed that of whites for the first time in the modern political era.

Using this election as a baseline—that is, replicating the turnout and vote preference patterns of 2012 by demographic group but using the 2020 distributions of these demographic groups—the authors find that in 2020, a Democratic candidate would replicate the 332-206 Obama electoral vote victory of 2012. Here the exact same distribution of states is seen between Democrats and Republicans as in 2012, but with larger margins for the Democrats in the states they carry—reflecting the changing populations in these states. Nationally, Democrats would increase their popular vote margin³ by about 2 points compared with their margin in 2012—going from a 3.9-point to a 5.8-point win.