In light of the Austrian presidential runoff election this past Sunday, the question arises: how common is it for the first-round leader to be defeated in the runoff? I can’t claim to have a specific quantitative answer to this question. However, I do have a graph that I prepared some years ago that helps contextualize the Austrian case.
The original graph has only Latin American cases, and ends around 2006. In the version posted here, I add the Austrian election and another prominent example from Europe: Portugal (1986). The x-axis shows the vote percentage of the candidate who came in first in the first round, while the y-axis is the second candidate’s vote percentage. A triangle indicates cases in which the second-place candidate came back to win the runoff.
//embedr.flickr.com/assets/client-code.js
The vertical grey line is at 50%. Obviously, if the first round result is to the right of this line, there is no comeback possible because the threshold for victory was met in the first round. The diagonal grey line marks equality between the top two candidates.
We should expect runoff comebacks (those marked with triangles) to be decreasingly common as we move farther down from the diagonal, and also as we get closer to the 50% line. In this sense, the Portugal 1986 case is truly extraordinary, as the first-round leader was quite close to 50% and faced a runoff challenger who had finished far back. The Austrian case is quite striking, too, with the second candidate having a shortfall to make up that was almost as big as in the Portuguese case. On the other hand, a leading candidate on only 35% is quite vulnerable. Most of the earlier cases depicted in the graph are of front-runners who were under 35%, and there are two others (aside from Portugal 1986) in which the first-round leader was over 40%. But all of these other cases had with a runner-up much closer than was the case in Austria.
Some day I should update the graph with more recent Latin American cases, as well as the full range of cases from other regions. This graph, however, makes clear that a comeback like we just saw in Austria is unusual.
_____________
(The original version of the graph appears in Matthew Søberg Shugart. 2007. “Mayoría relativa vs. segunda vuelta: la elección presidencial mexicana de 2006 en perspectiva comparada,” Política y gobiernoXIV, 1, Primer semestre.)
Fruits and Votes 
As it happens, the moment I learned about the outcome of this year’s presidential election in Austria, the first thing that came to mind was Portugal’s 1986 presidential poll, which shared other similarities with the Austrian vote. Specifically, the first round winner, Diogo Freitas do Amaral of the CDS (nowadays the CDS-PP) was widely perceived as a staunch conservative, well to the right of the country’s political center of gravity at the time; nevertheless, he appeared set to prevail in the runoff election after topping the poll in the first round by a large margin. Meanwhile, Mario Soares of the Socialist Party had just barely managed to arrive second, but his chances of defeating Freitas do Amaral in the runoff appeared slim-to-none, in large measure due to the bad blood between Soares and Francisco Salgado Zenha, who finished third, behind Soares and Freitas do Amaral; Salgado Zenha had once been a leading figure in the Socialist Party, but he and Soares had had a nasty falling out a few years earlier, and the presidential race did little the heal the old wound. Moreover, Salgado Zenha had secured the backing of the Communists, who loathed Soares and the Socialists almost as much as they hated the center-right parties (some would say even more so). On top of that, the Socialists weren’t in great shape either, having just been trounced in legislative elections held the preceding year.
However, in the end Soares scored a stunning upset in the runoff election, largely because most of the non-Socialist left-wing electorate decided that having Freitas do Amaral as head of state was too much of a risk, and that as such Soares was preferable, if only as the lesser of two evils.
LikeLike
Very interesting!
However, in the European context runoff comebacks are not that unusual.
In Austria, the runoffs in 1951, 1992 and 2016 were eventually won by the runner-up of the first round; only in 1986 the first round winner (Kurt Waldheim) managed to maintain his lead.
In all presidential elections in Europe (i.e. all current EU member states that hold two-round popular presidential elections) since 1950, there were 18 cases of runoff comebacks (out of 48 runoffs and 89 elections overall).
This is the full list:
Austria 1951, 1992, 2016
Croatia 2014
Cyprus 1988, 1993, 2008
France 1974, 1981, 1995
Poland 2005
Portugal 1986
Romania 1996, 2004, 2014
Slovakia 2004, 2014
Slovenia 2007
LikeLike
Sorry, I posted this to Facebook, but I had a similar list to Philipp: Croatia 2015; France 1974, 1981, and 1995; Lithuania 1998, 2003, Moldova 1996; Poland 2005; Romania 1996, 2004, 2014; Slovakia 2004, 2014; Slovenia 2007; Ukraine 1994; Senegal 2000 and 2012; Guinea-Bissau 2005; Niger 1993; Benin 1996; Timor-Leste 2007. I know some of those countries aren’t in Europe!
LikeLike
I somehow missed the two runoff comebacks in Lithuania (although I included it in overall count of elections) – so that makes it 20 out of 48 runoffs in the EU 28.
And another addition to non-EU list: Macedonia 1999.
LikeLike
Serbia also had a couple: 2004 and 2008.
LikeLike
Argentina 2015 would probably count as an example of a candidate trailing on the second round winning, although it wouldn’t fit neatly into that graph, given the lowered threshold for first-round victory. Perhaps also Argentina 2003, though that seems to me to be more of a case of a poorly written electoral law than anything else.
LikeLike
You would expect second round comebacks to happen more often in three circumstances:
1. A relatively low first round percentage for the first round front runner
2. A narrow first round margin for the first round front runner
3. A multi-party system where two of the three main parties are seen to occupy the same electoral space, while a third, larger party is seen as an outlier.
Any of these could produce a first round comeback.
I would add some event during the second round campaign that really negatively impacts the front runner, causing a shift in support, but I don’t think this has ever happened historically. All the other cases have.
One reason for majoritarian systems is in fact to prevent large, united outlier parties from “coming up the middle” and prevailing over two more similar parties, that between them probably have the support of a majority of the electorate.
LikeLike
A number of US presidential elections have resulted in no popular vote majority. However, since the Civil War, these have all tended to coincide with low popular margins between the first and second place candidates, so all would have been candidates for second round comebacks for that reason alone. The one case with a relatively large margin for the first round candidate, in 1912, was also a case where the majority of the electorate was inclined to support one of the two Republican candidates, Roosevelt or Taft, so Wilson “came up the middle”. In 1948 Truman, with a 3% margin, would have had momentum and a situation where the supporters of the minor candidates were more likely to vote for him over Dewey. All the other cases could well have led to a second round comeback.
LikeLike
Peru might be joining the list as a triangle too (if PPK’s narrow lead holds).
LikeLike
I’m reading the subtext here as: rare comebacks, especially from low absolute percentages and/or from far behind, provide support for the Double Complement variant of runoff — ie, that the first round is decisive if either (a) the highest candidate reaches 50% or (b) the second-highest candidate is more than twice as far below 50% as the highest is.
If you can divide runoff results that actually occurred, in hindsight, into those where the second ballot did and did not change the plurality order on the first ballot, and if this correlates closely, or at least fairly closely, with a robust and objective numerical threshold (for which the Double Complement formula is the leading candidate, as it were), then polities can save a lot of money on unnecessary second ballots, while still avoiding the risk of a Thatcher ’83/ Roh ’88/ Clinton ’92 defeat for the party a majority (or at least a larger plurality) could well have preferred on a two-way runoff contest.
LikeLike
“then polities can save a lot of money on unnecessary second ballots”
I coundn’t help thinking of Haiti…
LikeLike
Pingback: Peru’s narrow presidential win–and unusual divided government | Fruits and Votes
Pingback: Democratic primary: Don’t understate Biden’s dominance | Fruits and Votes
Pingback: Turkish presidential and assembly votes | Fruits and Votes