Turkish presidential and assembly votes

It looks like Turkey’s presidential election will go to a runoff, while the ruling AKP and allies have won a majority of assembly seats.

In the official reporting of results from Sunday’s elections, something looks implausible. The ballots for presidential candidates and (closed) party lists were separate. Yet the votes for incumbent President Recep Tayyip Erdogan and the alliance backing him are almost identical. In fact, for a good time yesterday the NTV site was showing the percentages as precisely identical to two decimal places. Today, at last check, they are 49.50% for the president and 49.49% for the assembly lists. So they are diverging. Kind of.

I have spent a lot of time over many years analyzing the interplay of presidential and assembly votes, and I would say this almost never happens. There is basically always enough ticket splitting for the percentages to diverge, with the leading presidential candidate usually out-polling the supporting party or alliance. In this case, Erdogan is far ahead of his own party, which has 35.6%, but has this essentially identical percentage when the alliance partners are included.

I am not alleging fraud. There are plenty of folks closer to the events doing that (for instance), and I certainly can’t claim evidence. This just looks strange. If they were going to inflate their vote one might think they’d look for a way to push the presidential vote over 50%, so perhaps it is the assembly vote that deserves more scrutiny than it is getting. Or maybe it is just one of those strange but true results. The reported seat total is not a narrow majority–the AKP, MHP, and a third partner together have 322 seats out of 600.

For the opposition, the divergence of presidential and assembly vote percentages is more normal-looking. The main opposition candidate, Kemal Kilcdaroglu, has 44.89% while the alliance backing him has 35.04%. The alliance of two left-wing parties (including the re-organized HDP) has 10.55%. If you add those together, however, their total of 45.59% is only slight higher than Kilcdaroglu’s, but this looks rather more normal.

The third presidential candidate in the race, Sinan Ogan, is on 5.17% while the ATA alliance backing him won only 2.44%. This is also odd, in that usually smaller political forces do better in assembly than in presidential elections. However, consider the unusual nature of the rules for these elections. The presidential election, first round, is arguably more “permissive” than the assembly electoral system, given the latter has a 10%* nationwide threshold. Obviously ATA was far short of this; its voters may have felt much more free to vote for their presidential candidate (who then could have leverage in the contest for support in a likely second round) than for assembly (where it would be a wasted vote).

Note that the 10%* threshold is applied to alliances, not to individual parties. If parties register an alliance, then the individual parties’ votes can contribute to seat winning even if the party itself fails to break 10% (as long as the alliance as a whole clears). Thus three alliances have won seats, but lists of seven different parties will be represented. *CORRECTION: That threshold is now 7% (see comments), although this does not change the argument I am making.

A ballot image shows how the parties that are allied are grouped on the ballot.

Thanks to Henry for finding and sharing this photo. Also note the very unusual split vote cast by this voter: Erdogan and the leftist TIP. Also, once I noticed the shoes of the voter, that was kind of all I could see.

As for the presidential votes, obviously a first-round percentage as high as 49.5%, taking it at face value even though there are allegations it is not “real,” makes a runoff comeback almost impossible. Not only is the leading candidate barely short of the required majority already, but also he has more than a four-and-a-half percentage-point lead over the runner up. Even if we knew nothing about the political leanings of voters who supported candidates other than the top two, we might be tempted to conclude the runoff is a foregone conclusion. In this case, the ATA alliance is right-wing and nationalist, so it seems likely its votes would go to Erdogan. (Perhaps those with actual knowledge of Turkish politics can disabuse me of this assumption.)

Erdogan made an interesting claim on election night, saying that if he did not win a majority in the first round, the majority his alliance won in parliament will encourage voters to support him in a second round (via journalist Ayla Jean Yackley on Twitter). Independent of the validity of his claim, no other case comes to mind of a presidential candidate claiming that a concurrent assembly election outcome will shape a presidential runoff. I suppose it is possible, although I am skeptical. On the other hand, I also would not expect voters to specifically vote to check the AKP and allies with their presidential vote. As I have said before in different contexts, elections in presidential (and semi-presidential) systems do not really work that way. They are more more like referendums on the president or presidential candidate–in favor when happening in close succession, and against at midterm and later.

Finally, it is notable that the AKP’s own vote percentage this time, at around 35%, is almost identical to what it won in 2002, when it and Erdogan first came to power. However, in that election, the party was able to turn that into 66% of the seats, due to so many parties falling below the 10% threshold.

From Erdogan’s perspective, it looks like the decision to move to presidentialism was a good call. The party is evidently not as popular as he is, and this time various other parties have grabbed on to his coattails in the form of an alliance. Presidentialization at work. On the other hand, of course, it could still turn out badly for him, if he loses the runoff. (I wonder if the allies would stay with him then. Maybe not, given the very same logic of presidentialization.)

The runoff, assuming it is indeed required per the final first-round results, will be on 28 May.