The Georgia 2020-21 Senate runoffs

The runoffs for the two US Senate seats in the state of Georgia are now just over three weeks away (5 Jan.). The first rounds, which were concurrent with the presidential election, were about six weeks ago. To my surprise, limited polling shows both races very close, but with the Democrats holding slight leads.

Some readers might be asking, why surprise? After all, the Democratic presidential candidate, Joe Biden, just won Georgia. While I do not pretend to know enough about Georgia politics to prognosticate these races, I want to do what F&V always does–focus on the rules, and how they make some outcomes likelier than others.

At its most basic, the challenge for Democrats is that the Senate contests in Georgia require a majority–that’s why there are runoffs, after all–whereas winning the state’s presidential electors requires only a plurality. And a plurality is what Biden won: 49.495% to 49.260%. The Libertarian ticket won the remaining 1.245%. That is pretty close to a majority, but not quite there.

Now, look at the first-round results for the full six-year Senate term. The incumbent, David Perudue (R), won 49.73% to 47.95% for Jon Ossoff (D). That is pretty close to a majority, too! But because the requirement to win is that a candidate must obtain more than half the votes cast, it is not sufficient. (The remainder was won by a Libertarian candidate, with 2.32%.)

The basic premise is that it is nearly impossible to lose a runoff when you are that short of the winning threshold in the first round. Not impossible, just nearly so. If we look at presidential runoffs around the world–which is the most comparable dataset I have available–candidates that close to a first-round win generally do not lose the runoff. The graph I posted in 2017 of the cases for which I had data show only one such example. In Ghana 2000, the eventual winner had trailed in the first round, 47.92–49.13, and then won the runoff, 50.23–49.77. Note: 49.1%, not 49.7%, as Perdue won in this first round; however, the eventual winner in the Ghana case was at about the same level as Osoff in the first round, 47.9%.

Of course, a senate race is not the same as a presidential contest. Moreover, I am looking at only 36 presidential runoff elections that went to a second round. (It is not an exhaustive dataset, but it was pretty close to being so, for reasonably stable democracies, as of around 2015 when I collected it.) It is noteworthy that in the dataset, no candidate in a two-round majority system reached 49% without clearing 50%.

So, there are two ways to read this. If you get close to a majority in the first round, you are pretty likely to get there in the second. Or, if you fail to get there in one round despite being so close, maybe there really is a majority that doesn’t like you. Which one will the Perdue–Osoff race be? Beats me! But I certainly would caution against a bet on Osoff.

The other contest is a bit different, in that the rules are different. It is a special election, and the first round did not feature one candidate per party. Instead, the first round was a non-partisan election (not a primary!) in which there were multiple candidates of each major party. Had one of them earned more than half the votes, that would have sufficient, but when none did, the top two advance (and it would have been the top two even if both had been of the same party).

As it happened, neither was near 50% in the first round. In fact the leader was a Democrat, Raphael Warnock, with just 32.9%. The runner-up was a Republican, the appointed incumbent Kelly Loeffler, with only 25.9%.

Turning again to the presidential runoffs, a case where the top two were around a third and a quarter of the vote is more open to going either way in the runoff. Of course, in this case, that’s bad news for the Democrats, who have the leading candidate.

The other key consideration here is the partisanship of the also-rans. The third candidate was another Republican, with 19.95%. (No other candidate reached 7%.) So the top two Republicans combine for 45.9%, which is well ahead of Warnock’s total, but also still well short of 50%. Overall, there were six Republicans in the first round and eight Democrats. The Democrats combined for 48.39%, the Republicans for 49.37%. These are remarkably close to the combined totals in the full-term Senate race (47.95% D, 49.73 R).

Thus if partisans all voted for their party’s runoff candidate, the two races put the Republican within a hair’s breadth of a majority before any runoff ballots are cast. On the other hand, the close divide between two Republican candidates in the first round could indicate divisions sufficient to reduce turnout and same-party voting in the runoff for those who supported the main defeated candidate.

Taking it all together, in both races it really does come down to turnout, whether one or the other party has a harder time getting its voters back to the polls again. It is likely turnout will be lower without the presidential candidates on the ballot. But will it be lower on one side or the other? Democrats need to keep their motivation up, and count on the other side’s presidential candidate’s loss being demobilizing. I don’t have the knowledge of the situation to offer even a hunch as to which is more likely.

A final consideration about the rules to bring in is: Could this be seen as a honeymoon election? We know that such election timing favors the newly elected president’s party. Recent cases in point would be France 2017 and Ukraine 2019. The timing–an early January runoff following an early November presidential election–certainly fits.

I would not consider these contests to follow the honeymoon-election logic for a couple of reasons. First, these Senate contests are already underway, with a concurrent first round. By definition, a honeymoon election is one that occurs in its entirety soon after a presidential election. That does not mean it could not follow the same logic. It just means we have no reason to expect it to do so, as voters have already registered their Senate candidate preferences once, concurrent with the presidential election.

Second, the actual cases of honeymoon surges for the winner of the presidential election all have taken place in multiparty systems. That means that some of the gain for the president’s party surely comes from those who voted for candidates of parties that finished third or lower in the presidential contest, not from the main loser. As noted above, the third-party vote in the presidential contest in Georgia was only 1.25%, and we hardly can count on Libertarians to be the key to Democrats winning these runoffs. (On the other hand, that the Libertarian senate candidate in the Perdue–Osoff race won 2.32%, more than a percentage point over what their presidential candidate got, could mean real resistance by this small party’s voters to Perdue. As an aside, in the special election, the Libertarian managed only 0.7%.)

The truth is that we do not know, in actual honeymoon elections, how much of the surge is voters swinging between the presidential election and the legislative (and how much such swing is from the first loser’s party) versus how much is turnout changes. In this case, importantly, any party swings or turnout drops (or increases) have to occur with respect to candidates for the legislative races who already contested a concurrent election.

The bottom line is that I am surprised the Democrats might actually be leading. I think the Republicans should be favored, Perdue somewhat more than Loeffler. The two senate contests could go different ways, but given the strength of partisan voting these days, that seems unlikely. I would not actually bet on any given outcome. Apparently it will be close. If the Democrats somehow pull off wins, it will be a pretty remarkable outcome!

5 thoughts on “The Georgia 2020-21 Senate runoffs

  1. Why doesn’t Georgia have the runoff election at the same time as the Presidential election? This is also a good case for preferential/ranked choice voting here as well.


    • My understanding is that federal elections which allow a candidate to be elected are required by federal law to be held on a certain date, the same as the election for President. Since a candidate who wins 50%+1 in a first round in Georgia wins the election, the first round has to be held on the same day as the Presidential election. The rule is the same in Louisiana. California gets around this by designating the first round as a primary and requiring runoffs in all districts, even if one candidate gets 50% or is unopposed.


      • There was a time when Louisiana had the first round earlier and the runoff in November, which meant some seats were filled before the presidential election. (Unlike the silly California system, a second round didn’t need to be held if one candidate had already won a majority for the seat.) This was changed, possibly due to a court challenge, but I do not recall.

        I guess Rob’s question is also why they don’t use AV. And the answer surely is just inertia. Georgia has had this conventional majority-runoff system, using two rounds, for a long time. Many southern states aside from Georgia still use two-round systems for partisan primaries, but I think Georgia is the only one that has two-round primaries followed by a two-round general. Someone please correct me if I am wrong. And, obviously, I am referring here to regular elections, not special ones like the Loeffler-held seat.


  2. Charles Blow at the NYT (2021.01.08) notes that “the Warnock win was the first time in American history that a Black senator was popularly elected by a majority-Black coalition. It was a momentous flex of Black power”, and says “We Need a Second Great Migration. Georgia illuminates the path to Black power. It lies in the South. Follow me there”.
    (No, Hiram Revels doesn’t count because he wasn’t directly elected).
    I wonder if this might take some of the momentum out of the US Left supporting PR. With removal of voter suppression and higher turnout (and, possibly, large scale migration as Mr Blow urges) there could be African American Senators even with single-seat FPTP. Likewise, if DC and Puerto Rico are admitted as States (or at least given voting representation in Congress), there will be Hispanic Senators even without PR. (Crubio don’t count – they mean Democrats who speak Spanish as their first language). I had read-ed somewhere that left-wingers in Scotland supported PR back in the pre-SNP days when FPTP meant that Thatcher was dominating, but once the SNP became the dominant left party in Scotland, many of them decided FPTP wasn’t so bad after all. (Although the SNP’s draft constitution for an independent Scotland still mandates some form of PR for the national parliament).


  3. Pingback: Will Macron lose his assembly majority? | Fruits and Votes

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.