Thanks in large part to Donald Trump’s successes on the campaign trail and to the tighter-than-expected race between Hillary Clinton and Bernie Sanders on the other side of the aisle, there’s been no shortage of commentary lately on demographic factors in American politics. A recent report from the States of Change project—a joint venture of the American Enterprise Institute, Center for American Progress, and Brookings Institution—backs away from the specific dynamics of this year’s primaries and takes the long view on how demographics shape political outcomes.

As the report documents, race has been linked to certain partisan preferences in presidential elections since the mid-twentieth century. In fact, coauthors William H. Frey, Ruy Teixeira, and Robert Griffin note, “Whites have voted Republican in every presidential election since 1968; blacks have voted Democratic in every presidential election since the second term of Franklin Delano Roosevelt in 1936—and especially strongly after 1960.” Hispanics have also favored Democrats for as long as data on their voting patterns have been available, and Asians have come to vote Democratic since 2000. The margins by which each racial/ethnic group supported the Democratic candidate for president is depicted in the figure below.

race vote margins

Age, which in 2000 and 2004 did not appear to play a strong role in how Americans voted, has lately been linked to partisan preferences as well, though to a lesser extent. In 2012, for instance, 18- to 29-year-olds voted for President Obama over John McCain by a margin of more than 20 percentage points, and 30- to 44-year-olds by a margin of about 15. Forty-five- to 64-year olds split their vote pretty evenly, and those over 65 preferred McCain by an almost 20-point margin, as the below figure illustrates. Importantly, however, turnout rates are far higher among older voters than young ones.

age vote margins

It’s not hard to see why different party platforms attract different groups of Americans, and how a party’s voting profile will then affect their platform. That Republicans rely on the backing of older voters may make them reluctant to cut spending on Social Security and Medicare in the interest of balancing the budget, for instance. Strong support for Democrats among African Americans and Latinos could incline the party to devote more federal spending to, say, creating jobs (in light of the higher unemployment rates among these minorities) or improving public schools.

The roots of different individuals’ voting patterns certainly vary, but few doubt that these demographic tendencies will influence election outcomes to some extent as the U.S. population ages and diversifies. In the new States of Change report, using projections that they generated last year on how each state’s race and age profile will change in the next few decades, Frey, Teixeira, and Griffin explain how the next five presidential elections could play out under six scenarios with different assumptions about each race/age group’s voter turnout and party preference.

Three of the scenarios combine the actual voting patterns of different groups in the 2004, 2008, and 2012 presidential elections with the projections of changing demography among eligible voters in each state. One scenario combines Americans’ 2012 candidate preferences with the assumption that the turnout of Asians, Hispanics, and other racial groups matches whites’ turnout in each age bracket and that African Americans’ turnout will continue to exceed that of whites, as it did in 2012. The fifth scenario, using actual 2012 levels of turnout, assumes Republicans will tighten their (losing) margin among Hispanics, Asians, and other non-black racial minorities by 15 percentage points in each age bracket and in every state. And the final scenario takes 2012 levels of turnout and assumes Republicans will widen their (positive) margin among whites by 10 percentage points.

In general, demographic changes in the American population will benefit Democrats through at least 2032, the last year covered in the report. Their advantage among racial minorities will increasingly outweigh Republicans’ advantage among the growing elderly population, even if aging Baby Boomers replicate the pro-Republican leaning of today’s seniors (many of whom are from the Silent Generation). For this reason, Democrats would win the 2016 election in four of the six scenarios outlined above. By 2024, they would win the presidency in five of six; in 2032, they would win it in all six.

Frey, Teixeira, and Griffin emphasize that these projections are not predictions, and as political analyst Sean Trende has argued, chance and basic economic factors may be more relevant to election outcomes than most people appreciate. By a modest reading, then, the bottom line of the States of Change report is this: To win presidential elections, Republicans probably can’t keep losing minorities by such large margins, and for at least the next decade, Democrats probably can’t afford to lose many more whites. Building a platform that appeals to an electorate diverse in racial background, age, financial status, personal convictions, and political priorities may be difficult. But since the alternative could be losing elections, perhaps our two major parties will at least try.