Underestimate Trump’s Reelection Odds at Your Own Peril


This is part two of a back-and-forth between Crystal Ball Managing Editor Kyle Kondik and veteran reporter and presidential historian Paul Brandus assessing President Trump’s reelection odds in 2020. See the piece from Brandus here.

The Editors


— So long as President Trump is on the ballot in 2020, history suggests he will benefit from incumbency.

— While Trump’s approval rating is only in the low 40s, some election models suggest that he would still have 50-50 or better odds to win reelection if that’s his approval level in 2020.

Trump’s reelection odds may be better than many realize

One might have done better in predicting the 2016 presidential election, or at least in anticipating the very close eventual outcome, by basing a projection of the national popular vote on the findings of several political science models released prior to the election. These models, which were compiled by James Campbell of the University at Buffalo, SUNY and printed in both PS: Political Science and Politics and here at the Crystal Ball, generally pointed to a close election. These models mostly made their predictions several months in advance of the election and were based on the incumbent’s approval rating, the economy, and other “fundamental” factors.

Most of the models, accurately as it turned out, showed Hillary Clinton winning the national two-party popular vote. But the average of the 11 models showed Clinton winning just 50.8% of the two-party vote, with the median projection showing her winning 51.0%. Both the average and the median were basically spot-on, given that Clinton ended up getting 51.1% of the two-party vote. Donald Trump’s strength among white voters who do not hold a four-year college degree allowed him to win the Electoral College because of the overconcentration of these voters in several electorally important swing states in the Rust Belt. National polls showed a similarly small lead for Clinton on Election Day, although Clinton’s leads in these polls were generally larger than her eventual margin for much of the 2016 calendar year.

I bring this up just to note that, as we begin to assess Donald Trump’s reelection odds, it seems possible that the polls and the election models will again be at odds in 2020.

One of the models included in Campbell’s 2016 survey was the Time for Change model, created by Crystal Ball Senior Columnist Alan Abramowitz of Emory University. Abramowitz’s model was one of just two to project Trump winning the national popular vote in 2016 (by three points), in part because it emphasizes the electoral advantage that an incumbent running for reelection enjoys, and 2016 lacked an incumbent. Trump underperformed the model’s prediction by five points in terms of margin, as Abramowitz himself suggested Trump would prior to the election.

The model gives a bonus to an incumbent who is seeking his party’s second straight term in the White House, meaning it very well could smile on Trump in 2020 while being more bearish on someone like George H.W. Bush in 1992. That year, Bush was running for his second-consecutive term, but his party’s fourth-straight term (Bush lost to Bill Clinton). The Abramowitz model also incorporates the incumbent president’s approval rating in the Gallup national poll and the state of the economy, as measured by quarterly GDP growth.

The Abramowitz model will make its 2020 projection officially using 2020’s second quarter GDP growth and whatever Trump’s approval is at that time. Still, we can plug in current numbers to give a sense of what the model might project. Right now, Trump’s net approval rating is -16 points according to Gallup (39% approve/55% disapprove), and 2017’s fourth quarter GDP growth (the most recent quarter available) was 2.9%, according to the most recent revision from the Bureau of Economic Analysis. Using those figures in Abramowitz’s model projects Trump with 51.6% of the national two-party vote. Even if Trump were to underperform the model again, like he did in 2016, it would still make the election a Toss-up, especially because Trump could win again without winning the national popular vote given the demographic patterns of his support.

So the United States could reelect an incumbent president with an average approval in low 40s? Yes. And, actually, that’s perhaps what we should even expect given the performance of similarly-situated incumbents across many different countries.

In October 2015, Clifford Young and Julia Clark of the international polling firm Ipsos — with whom the University of Virginia Center for Politics has partnered on some recent polling — wrote a column for Reuters arguing that Democrats were unlikely to retain the White House in 2016. That’s in large part because the Democrats were trying to hold the presidency without having an incumbent on the ballot. They created a model using the results of more than 450 elections across 35 different countries since 1938 that suggested a president with relatively middling approval, as President Obama’s was at the time of their writing, had relatively low odds of being able to hand off power to a successor of the same party. Even with Obama at slightly over 50% approval on Election Day, according to the RealClearPolitics average, the Ipsos model suggested the Democrats would only have about a one in three chance to hold on to power. They didn’t, despite Trump’s problems.

Incumbents get a boost in the Ipsos model, though, just like in Abramowitz’s model. Young and Clark suggest that as long as an incumbent is at 40% approval or better, that person is probably at least a little likelier to win than lose. So Trump, at 41.9% approval in the RealClearPolitics average, would be capable of winning reelection, if not outright favored, if that’s where his approval rating is in fall 2020.

There’s obviously a long way to go before the 2020 election. We have a national election this fall in which Democrats are very likely to register at least some gains in the U.S. House and at the gubernatorial and state legislative level (though perhaps not in the Senate, for reasons we have previously outlined in detail). Robert Mueller’s investigation into Russian meddling in the 2020 election could have a bad outcome for Trump and could even hypothetically force him from office, although there’s really no telling where the investigation could ultimately lead, and Trump may very well survive it. There are tons of other uncertainties, including the eventual identity of the 2020 Democratic nominee, the possibility of a prominent third-party candidate, and, perhaps most importantly, the state of the economy two and a half years from now. We think the decent economy is helping Trump maintain a weak but passable approval rating despite all of the swirling controversies of his presidency. If the economy weakens, Trump may dip back into the 30s in approval and further complicate his reelection path.

But assuming Trump is on the ballot, and assuming his approval rating stays around the 40% mark, it would probably be wrong to assume he’s an underdog for reelection. That’s not to say he would be a sure winner, but he wouldn’t be a sure loser, either.

The biggest mistake analysts made in 2016 was believing that Trump was such a weak candidate that he would prove to be unelectable even though a close reading of historical results, both in the United States and elsewhere, suggested that any Republican would be a formidable contender for the White House in a year like 2016, when the Democrats were attempting the difficult task of winning the presidency a third consecutive time (and they had nominated a weak candidate to face Trump). That same history, a history that is built into models like the Ipsos and Abramowitz models, suggests an incumbent in Trump’s position will not be a pushover unless his approval and/or the economy significantly decline from their present levels.