Presidential Incumbency Isn’t Really That Valuable

Incumbent presidents’ record when running for reelection is good but not great.

T he volatile nature of 2020 — in which a pandemic and civil unrest have already rendered even the presidential impeachment a distant memory — makes it impossible to predict much about the November election with certainty. We know that Donald Trump is in trouble: Joe Biden has built a strong position in the polls, and the booming economy that was supposed to be the centerpiece of Trump’s reelection campaign is now a distant memory. But we’re still months from knowing how much that really tells us about the fall.

This much is certain: Against Biden’s early lead, Trump has the advantage of incumbency. He can raise gobs of money and dominate the news. He can take official actions, while Biden can only make speeches. The last three presidents have been reelected, each after a lot of premature obituaries were written for his reelection prospects. Americans haven’t thrown a sitting president out of office in 28 years, and only once in the past 120 years (in 1976 and 1980) have American voters changed party control of the White House in two consecutive elections.

But how much is incumbency actually worth in predicting who will win an election? Let’s walk through the record of American incumbent-reelection efforts compared with the president’s initial election, using four helpful measures: (1) shifts in the national two-party popular vote, (2) changes in voter turnout, (3) number of battleground-state flips, and (4) battleground-state electoral-vote shifts. (Note: To be clear, the term “reelection” excludes incumbents who succeeded to the presidency through the previous president’s death or resignation.)

The National Two-Party Vote
In previous studies of elections after an incumbent was reelected, I found a clear, overwhelmingly consistent historical trend favoring the party out of power. As I explained in my original 2014 analysis, the two-party vote (i.e., excluding third parties from the calculation) is typically more useful in analyzing American presidential elections (both nationally and state-by-state) than the overall vote. Trump, of course, does not need to win the national two-party vote to be reelected (he didn’t win it in 2016), but his margins of victory in the 2016 battleground states of Michigan, Pennsylvania, Wisconsin, and Florida were narrow, so he is unlikely to win reelection if he suffers more than a small overall reduction in his national two-party vote. How have elected incumbents fared in improving or retaining their position in the two-party vote?

There are some caveats here, mainly applicable to the earliest elections. Washington and, in his second run, Monroe were unopposed. Popular votes were not uniformly recorded until 1828, and at least one state (South Carolina) did not count popular votes until after the Civil War (its legislature chose the presidential electors). This point has particular pertinence for 1824, when John Quincy Adams won the election while trailing considerably in the recorded popular vote, in large part because he won most of the electoral votes of the largest state in the union (New York), which did not count popular votes. The 1836 and 1860 elections also both involved parties running different candidates in different parts of the country, and in 1864, the states where Abraham Lincoln had done the worst had seceded and did not vote.

Those issues aside, since 1872, only nine out of 16 incumbents have gained ground in the national two-party vote. That’s not a reassuring figure if you’re the Trump campaign. The rate improves slightly, to five out of eight, since the Second World War, but even in 2012, Barack Obama lost enough ground with the electorate that he would have lost if he’d had as small a margin for error as Trump does. Only one incumbent since 1872, Richard Nixon in 1972, did a lot better than he had four years earlier, but this year Trump would have needed a Bernie Sanders nomination combined with a roaring economy to even begin to replicate Nixon’s 1972 strategy.

Here are the average numbers, by party and by winning and losing campaigns:

On average, incumbents have lost ground in their reelection attempts. This is true for Republican incumbents and Democratic incumbents, overall or since the formation of the Republican party — every category but winning incumbents (which selects out the failures). The results partly reflect the fact that incumbents rarely lose close races — indeed, as I have detailed previously, incumbent elections in general are rarely close. More often, either the country makes a collective decision to stay the course or the bottom drops out. Only when an incumbent was elected the first time in unusual circumstances, against fragmented or listless opposition (Woodrow Wilson, James Madison), does the incumbent survive a strong challenge his second time around.

I’ve excluded from my study those cases where the incumbent either was a vice president who succeeded to the presidency or (in Franklin D. Roosevelt’s unique case) was running for a third or fourth term. Those elections, however, do not provide much reassurance:

Incumbency or no, four of these seven elections saw a decline in the incumbent’s share of the two-party vote, and there was an average decline overall.

Voter Turnout
What about turning out voters? As in my previous analyses, I use the historical “Voting-Eligible Population” (VEP) figures compiled by Michael McDonald at Elect Project to calculate turnout, which is then multiplied by each candidate’s overall vote share to compute a VEP percentage — that is, the percentage of all eligible voters who cast a ballot for each candidate. Over time, of course, turnout has risen and fallen, due to factors such as changes in eligibility, suppression of eligible black voters, etc. Turnout was highest between 1840 and the 1880s, with Ulysses S. Grant in 1868 garnering the highest share of all eligible voters of any candidate in American history. Turnout dropped off when the electorate was expanded in the 20th century to include women and, later, young voters.

Incumbents have been modestly successful in recent years in increasing their share of voters. Six of the last nine incumbents have expanded their share of eligible voters, while Obama joined McKinley and Grant as presidents who got reelected while failing to do so. But incumbents also drive more turnout among their opponents, with an average increase of 1.2 percent since 1832, and 3.7 percent in losing campaigns. George W. Bush, Ronald Reagan, and Woodrow Wilson all got reelected despite significant growth in opposing-party turnout (in Wilson’s case, it was mostly Republicans coming home after bolting to Teddy Roosevelt in 1912). Only four presidents — Bill Clinton, Richard Nixon, Dwight Eisenhower, and Franklin Roosevelt — managed the trick of increasing their own turnout while driving down the other side’s. I suppose Trump can take some comfort in seeing Clinton and Nixon on that list.

If we extend that analysis to include incumbents who had succeeded to the presidency through death or resignation (plus FDR in his third and fourth campaigns), the picture gets grimmer:

All these candidates except Lyndon Johnson saw turnout decline. That’s of limited use in projecting 2020, however, since these were not candidates reassembling their own coalitions for a second try.

Battleground States
Donald Trump didn’t get elected by winning the national popular vote; he won by flipping strategically crucial battleground states that had voted for Obama to build an Electoral College majority. How have previous incumbents fared in battleground states?

As in my prior studies, I define “battleground” states as (1) states that the party in power in the White House won with 60 percent or less of the two-party vote in the last election, (2) states that the party in power lost with 45 percent or more of the two-party vote in the last election, or (3) states outside those parameters that appear as battlegrounds in retrospect because they actually flipped parties. I used the 60 and 45 percent thresholds because we are comparing to a previous election that the party holding the White House won, so it is a bigger stretch to win states in the 40–45 percent range than to lose states in the 55–60 percent range.

Of the 545 battleground-state contests in 19 previous incumbent-reelection efforts (an average of 29 such states per election), incumbents have won 50 states that they lost the first time around, and their challengers have flipped 133 states. The average incumbent suffers a net loss of 15.2 percent of battleground states (for an average of four or five). Republican and Democratic incumbents follow historically similar patterns. However, there is a lot of variance between winning candidates (who average a 7.1 percent net gain) and losing ones (who average an eye-popping 47.5 percent loss).

The trend of the last three elections — if that’s enough to be a trend — has been towards fewer battlegrounds and fewer flips. Bill Clinton and Barack Obama both suffered a net loss of battleground states but were reelected nonetheless. Trump, however, was elected in the first place with narrow margins in a handful of large battleground states, and he cannot win reelection if he suffers a net loss of more than two battleground states.

Without consistent popular-vote totals, it’s not possible to define “battleground states” the same way for the period before 1828, but for completeness, the record of that period shows very few states defecting from an incumbent overall:

Large states, even then, were especially decisive; John Adams was done in by losing New York, and the same was true for his son.

Alongside flips, we can also look at popular-vote shifts within each state:

Overall, the general trend is negative, but driven by the disproportionately large number of states that broke against losing incumbents. Only two incumbents — Woodrow Wilson and Barack Obama — were reelected while seeing more states shift away from them than in their favor. The size of the shifts is not that important to assessing Trump’s chances; even a 1.2-point shift across the board against him would cost him the election. But it is relevant to the broader national trends that will show up further down the ticket.

Battleground-State Electoral Votes
What about the size of these battleground states? Trump in 2016 flipped a modest number of battlegrounds, but they included large states such as Florida, Pennsylvania, Ohio, and Michigan. Let’s examine the historical record:

The “Batt-EV Net Avg” column is the same as “Flip percent” in the prior set of charts, but for electoral votes: net electoral-vote gains or losses, divided by total battlegrounds. The overall pattern since 1832 is an average net loss of 2.2 electoral votes per battleground: +0.4 among winning candidates, –6.2 among losers. The average incumbent wins 17 out of 29 battleground states: an average record of 22–5 among winners, 8–23 among losers. Again, both in win/loss records and the size of the Electoral College shift, the average conceals the scale of the split between success and failure (most winning incumbents can’t gain much ground because most of the battlegrounds are states they won last time). Nixon in 1972 aside, most of the really large electoral-vote-per-battleground shifts have come in incumbent losses or (in Woodrow Wilson’s case) returns to earth from electoral-vote victories that were far out of proportion to their popular support. Wilson and Obama were the only presidents reelected with a net loss of Electoral College support; Wilson was also the only president reelected while losing more battleground states than he won.

Conclusion
In 2016 Donald Trump ran against the party of a president who had won two elections in a row. Historically, this provides a strong wind at the challenger’s back. Trump ran well behind that historic trend but made it count by winning important states by narrow margins. Incumbents seeking reelection, such as Trump in 2020, have been much more of a mixed bag, with a number of large-scale collapses of support. It is very possible that Trump will end up as one of those, although his opponent is not running the kind of campaign that has ever defeated an incumbent.

Public perception of the state of the economy this fall will doubtless be a major factor, and an unprecedented one: Just as no president has previously run for reelection after being impeached, no incumbent seeking reelection has ever faced an economic collapse and, possibly, recovery within a single election year. We’re in uncharted territory. Even if Trump avoids collapse, however, there is no unambiguous historical trend in his favor. Very few incumbent races have been close, and incumbents have lost as many of the close ones as they have won. In short, it is not true that incumbency by itself provides a historically significant advantage in a presidential reelection campaign.

For those interested in an election-by-election breakdown of battleground-states, see here.