Conventional wisdom is that an incumbent president has a huge, very hard to overcome advantage over a contender, and that anytime an incumbent loses it’s a major upset. But a recent thread prompted me to think about that, and it doesn’t appear to be true in modern times. I’ve picked 1968 as the cutoff since it’s the first election after the Republicans adopted the Southern Strategy, which pushed the Democrats and Republicans into their modern form. In that time, there have been nine elections where a President sought reelection. Four of them were won by the incumbet, four of them were won by the opposition, and the 2020 election has not finished but appears likely to go to Biden (non-incumbent) so I’m including it in that tally.
So in the modern era, it looks to me like Presidents who sought a second term don’t seem to have a significant advantage - they’ve lost more often than they’ve won. Is there any substantial support for the idea that incumbency provides more advantages that disadvantages with the current voting patterns in the US? The results seem to me say that the advantages and disadvantages cancel out - being president when things are going good is generally an advantage, but if things are bad being president gets you the blame.
Here are the elections I’m looking at for the totals:
1968 Nixon beat Humphrey, who beat Johnson (I) in the primary
1976 Carter beat Ford (I)
1980 Reagan beat Carter (I)
1984 Reagan (I) beat Mondale
1992 Clinton beat H. Bush (I)
1996 Clinton (I) beat Dole
2004 W. Bush (I) beat Kerry
2012 Obama (I) beat Romney
2020 Biden likey beats Trump (I)
This is probably the wrong way to analyze it, since there have been relatively few elections in the modern era with an incumbent running, the Law of Small Numbers is going to mislead you.
The right way to do it is to look at all elections with incumbents (not just Presidents), and to look not just at whether they won (which is distilling the rich data of elections down to a single bit) but to look at vote patterns.
If you do that, you get something like this study, that shows that the incumbency advantage is real, likely in the range of 4-8 percentage points of the vote, and increasing over time. Skip to the end to the tables to see the basics, or read the whole thing if you want (I didn’t).
No, I don’t think that’s the right way to look at it at all. How on earth is it valid to take a study using data from races that are based on a state or smaller region and doesn’t even include the presidential race, then declare that its results must apply to the presidential race? The more local a representative is, the more they do directly for their constituents, while presidential candidates are generally more aspirational or a referendum on the state of the economy, and clearly work differently from smaller races.
The fact that presidential races in the era of modern parties have gone against the incumbent more often than not is significant, and seems to get overlooked in discussions and articles. Also, I get extremely skeptical because all of the articles I’ve come across talking about the power of presidential incumbency choose to ignore 1968, where the incumbent couldn’t even win his own primary.
It’s not accurate to say Humphrey beat Johnson. Johnson was effectively hounded out of seeking another term. He’d lost control of the country in 1968 over the war in Vietnam and the civil riots at home. Johnson was on the ballot initially but his growing unpopularity, his health, and frankly his declining appetite for the job meant he announced from the Oval Office he would not seek another term as president. This was a man who craved power yet the view from the top depressed him by the end. As a result he put his weight behind Humphrey who was of course his vice president. Johnson wanted to end the war in Vietnam before the election which would have helped Humphrey.
Ford was the incumbent president in 1976 but as we know he was an unelected president. 1976 was the first time he was on the ballot nationally. The worst thing he did that everyone remembers was pardon Nixon but I guess in a lot of people’s mind that’s the only thing they remember him doing. Which isn’t that bad when you compare the worst things other presidents committed. Still the fact that he did that and the election came so close is probably due to incumbency. The Republicans got wiped out in the 1974 mid terms a couple of months into his presidency yet he almost won the 1976 election after Watergate. He didn’t really accomplish much in his two years nor was he an inspiring figure. If he was just another Republican nominee as opposed to the President of the United States he would have probably been defeated handily. Jimmy Carter was a nice man who preached morality and integrity in government but by the end of the election campaign a lot of people were thinking “well Jerry Ford’s a nice man too and he’s not done a bad job. No one knew who Carter was six months ago. Lets keep Ford on the job”.
Well at least now we know that discourse isn’t perfectly reliable. Maybe it’s the PDF, maybe it’s using a right-click instead of clicking through, maybe it’s my anti-ad extensions, but what you posted above is an untrue accusation. Let me quote the paper you say I didn’t read: “The paper documents that all executive and legislative offices from utility commissioner to Governor, from state legislator to Senator have experienced a similar electoral transformation since WorldWar II.”
Notice that it doesn’t even include presidential election data, just like I said? But yeah, I’m sure I somehow quoted from the paper without reading it.
It’s not. Having an advantage from X is different from winning because of X.
[/quote]
First you accuse me of not reading the paper, then you say that actual data doesn’t matter. If the supposed advantage can’t actually be seen in any data, it’s safe to conclude that it’s nonexistent, so small as to be meaningless, or that the advantage is canceled by the disadvantage. When the best argument you can provide is “Look, I’ve decided that this paper that doesn’t even include data on presidential elections proves there’s an incumbency advantage for president, also because discourse didn’t track that you read it if you dispute me by referencing the content of the paper I’m going to claim you didn’t read”. Not exactly a strong argument.
It is, however, accurate to say that the incumbent Johnson lost decisively. Had he simply chosen not to run I would accept that he wasn’t in the race, but the fact is that he ran and only withdrew after the primaries started and he didn’t get a decisive enough vote in New Hampshire. Getting trashed in your own party’s primary is losing the election even worse than making it to the main event as candidate.
This explanation assumes that an incumbency advantage exists and that he benefitted from it, so I don’t see how it’s an argument that the advantage exists. What I see is that generally incumbent presidents who have popular accomplishments, a good economy, high charisma, and/or weak opposition win, while if they have bad events (especially ones they didn’t handle well), economic troubles, low charisma, and/or strong opposition, they lose. That is, the incumbency is only an advantage if their term has achievements, and is a disadvantage if they have failures or unpopularly handled issues.
I do think it’s an advantage, but clearly not a decisive one. I view it as sort of like “home field advantage” in the NFL, which is generally believed to be worth about 3 points.
You’re able to run on your own record and accomplishments in the position. In addition, in most cases, you don’t face significant opposition in your own party during the primaries, which means that you don’t have top members of your own party criticizing you during the primaries, and you don’t need to spend much of your campaign war chest before the general election.
But, if the economy is in the toilet (Ford, Carter, GHWBush), or the country is seen as going in the wrong direction (Johnson), those factors are going to more than outweigh that “home field advantage.”
Of course the data matter. The outcome, less so. Which is why we look at other executive elections to perform some actual statistical analysis instead of just a counting exercise.
Yeah, that’s the best answer. Another good one is that we’re coming off three straight presidents who had two terms and that hadn’t happened since Jefferson, Madison, Monroe. How’s that for incumbent advantage? Look at what happened before 1968 as well: Kennedy/Johnson two terms, Eisenhower two terms, Roosevelt/Truman five terms, Harding/Coolidge two terms, Wilson two terms, McKinley/Roosevelt two terms.
An incumbent president losing an election is an anomaly. Look at the losers from the start of the 20th century, when what we can call modern politics began.
Taft. Lost in four way race because the most popular person in the country ran and split the Republican vote.
Hoover. Worst Depression in American history.
Ford. Interim president who pardoned Nixon, who won two terms.
Carter. Won in a heavily Republican country over Ford and would have needed an extraordinary performance to retain office, which he failed to achieve.
Bush. Third-term Reagan - another two term President - who reaped the horrors of the Reagan failed budget program when the country went into recession.
Trump. Worst president ever.
George W. Bush somehow got re-elected in the middle of a totally unnecessary war that drained a trillion dollars out of the country and destroyed the actual budget surplus he inherited from Clinton. How’s that for incumbent advantage?
Counting Johnson as a defeated incumbent is ludicrous. As someone around in 1968, I can state confidently that LBJ would be beat Nixon if he stayed in the race. That race was ridiculously close. Even a small number of people who defer to the presidency and won’t kick one out would have swung it. Besides, Humphrey was one-tenth the politician LBJ was.
It’s not very complicated: incumbency is an advantage in a normal economy and a disadvantage in a bad economy. All the four Presidents who lost their re-election bids in the last 100 years were contending with a bad economy in one way or another: Hoover obviously, Carter facing a deadly combination of high inflation and unemployment in 1980, and HW running after a recession in 90-91 and a still sluggish economy in 92. 2020 was somewhat ambiguous because while the economy was in poor shape in 2020 because of Covid, perhaps voters might not blame Trump for it and remember the relatively decent economy in 2019. Probably some voters did do just that which is perhaps why Trump did not lose as badly as many expected.
You forgot to mention that Bush 43 had to contend with Ross Perot and Pat Buchanan as candidate. In a head to head against Clinton he would surely have won. Also I think that without Covid, Trump would have won.
That certainly was a weird election, but I’m basing my conclusion mostly in the fact that his approval rating went from 89% in 1991 to 29% in October 1992. He was not going to beat Clinton.
Two things happened this year, COVID and the protests. Every sane person would conclude that COVID hurt Trump’s chances. Yet he won 93% of the counties with the highest per capita COVID rates. Yes, COVID may have hurt him with, say, suburban women, but it cuts both ways.
The protests brought out huge numbers of activists who starting grassroots efforts for candidates and registered voters. But they also frightened huge numbers of people away from the evil libs.
I think it will take years to untangle the threads, if that’s even possible.
I mean, it’s not perfect, but I’d bet it’s going to generate better results than your plan of looking at very few results and also ignoring vote percentages and taking only a single bit of information from each one.
What is your theory for why Presidential elections are so different from every other election we have data for that there’s no advantage to incumbency?
I don’t think that completely ignoring presidential elections is going to result in a good data about presidential elections. It might do so by chance, but it’s not reasonable to completely ignore the presidential race and just decide that other, non-national races are the only data needed. Again, you posted a study that makes no attempt to look at the national vote as a whole, and doesn’t even look at presidential elections!
I think that sticking with the null hypothesis in the absence of data saying that incumbency is an advantage or disadvantage for presidential elections is a reasonable thing to do, and the only studies that I’ve seen are about non-national elections and generally (like yours) don’t even include presidential data.
It’s a national election and people often vote for president based on ‘state of the country’ rather than ‘things the president directly controls’ - as people point out, presidential races for many people act as a referendum on the state of the economy, even though congress and large-scale events have more to do with that than presidential actions. State and local elections involve a lot more practical concerns, and the people being voted for are close enough that it’s a lot easier to actually know someone who’s worked directly with the person in office. For example, veterans usually lean right, but Bernie Sanders is extremely popular with veterans in Vermont because he provides them a lot of help in dealing with the VA.
And while I’m sure people will nitpick that theory, even if you destroy my theory completely it doesn’t actually offer any support for the assertion that incumbency is an advantage for presidential candidates. Like I said before, in the absence of good data, one should tend towards the null hypothesis, and I’m not seeing an outpouring of good data.