Monday, July 26, 2010

Weekly Hurriyet Column: The referendum countdown

Below is the unedited version of my column for this week. You can read the final version at the Daily News website, but since I have been editing my columns myself on the Daily News website since March, you won't see much of a difference between the two. No cheesy title this time around.

As for the column, I don't have much more to say, except that I will be getting  (or was supposed to get, since it is almost 11pm, I and still haven't from them) the August KONDA survey tonight (August 10, Tuesday), so I might be able to share, either in the blog or in next week's Hurriyet column, some new/updated results.

As for my election calculator, it is quite simple to do: You just have to know about the mathematics of the Turkish electoral distribution system. Then, you take one (or the average of several) election poll, use the past election(s) to project the nation-wide poll results at the province level and build an Excel sheet that will distribute the seats according to your projection and the d'Hondt method.  Then, you just wait for the PM to try to get you fired. By the way, I actually stumbled upon a website that does that: You just enter the poll results on the left and it spits out projected electoral distribution on the right.
Anyway, on to the column:

Since polling firm KONDA asked me to write on the upcoming referendum for their August Barometer report, I have been immersed in their last two surveys, trying to find the magical determinants of an aye or nay on Sept. 12.

I am bound my KONDA’s commitment to their subscribers to keep the numbers to myself, but I can reveal that while the race seems to be quite close at first glance, you encounter a lot of voters who are unlikely to switch votes when you look at voting behavior by party affiliation. Go out and find me a Republican People’s Party, or CHP, supporter who will vote yes in the referendum, and I will be impressed. Find me a Justice and Development Party, or AKP, supporter who will vote no, and I will put your name down for my quest for the Holy Grail.

The same results hold true if you look at voting behavior by ethnic and religious identity as well as stance on current political issues. In fact, referendum voting choices are just another reflection of the political polarization I had demonstrated when I had first started working for KONDA in March. One surprise is Kurdish voters, who do not seem to share Peace and Democracy Party’s, or BDP, distaste for the constitutional amendments and are more likely to vote yes than any other ethnic group.

So does this mean that the referendum is already decided (remember that I am not telling you who is ahead)? Hardly, once you notice that neither the ayers nor the nayers hold the majority. The first place goes to the undecided and those who have no idea, what I call undecideas. Then, it is my civic duty to tell a bit more about these folks, in case any of the campaigners out there are reading my columns.

The undecideas are more likely to be women than men. The probability of being undecidea decreases with education. Finally, someone from Thrace is more likely to be undecidea, even after controlling for other factors. The latter result is a bit surprising, but digging a bit deeper (but not as deep as Bilica) reveals that polarization is considerably less in this region. It may then as well be that Thracians, known for their secularism, are torn between their distaste for AKP and merits of the constitutional changes.

This last result hints on the possibility of last-minute swing votes. Regardless of all the white noise in the academedia, worsening economic conditions or more terrorist attacks are unlikely to pull votes to the nay camp a lot. The anti-AKP camp’s only viable strategy at this stage would be to try to reach out to the less polarized among the undecideas and convince by reason.

By the way, I had nothing better to do on a Sunday morning, so I reran my election modeling toolkit from my days as a bank economist, which was dusting in my hard drive. As CHP has been arguing for the national election threshold to be reduced to 7 percent, I was wondering what would be the magical number that would deny AKP majority in the Parliament.

According to my calculations from the last two KONDA surveys, a 7 percent threshold is not much different from a 10 percent one, as both result in a three-party Parliament and an AKP majority, not much different from the status quo. But things change when the threshold goes down to 5 percent, letting BDP in. Suddenly, AKP has lost around 40 seats, all to BDP, as well as its majority in the Parliament.

I showed last week that AKP advisers were well-versed in Excel. It could be that they are working with different numbers than me, but their colleagues in CHP just do not seem to be as deft. Or maybe they are just making a conscious tradeoff between the lesser of two evils, as they see it, which is, when you think about it, is actually worse than being a skxawng.

Emre Deliveli is a freelance consultant and columnist for Hurriyet Daily News & Economic Review and Forbes as well as a contributor to Roubini Global Economics. Read his economics blog at

No comments: