As Turkey prepares for a parliamentary election, triggered by the deadlock over the attempted election of a president by the current parliament, the parties turn their attention to the mobilization of voters. I have noted before that the current AKP majority is based on barely over a third of the votes. It counts as one of the largest manufactured majorities anywhere in recent decades. Not only does the ruling party enjoy a majority on far less than half the votes, it also is barely short of two thirds of the seats. Many electoral systems distort the votes-to-seats conversion process in favor of the plurality. However, most of those that do so are FPTP systems. For instance, the British Labour Party currently enjoys a comfortable majority of seats on only around 35% of the votes.
Whatever one’s position on the question of the desirability of close approximation of partisan vote and seat shares (and even the most casual glance at F&V makes clear my own position on that question), such overrepresentation is much more easily justified on democratic grounds when the electoral system is nominal in character, as is FPTP. In a nominal system, voters are choosing a legislator in a local district and, no matter how strong the normal party discipline in Westminster-type parliamentary systems, the representation of parties in parliament remains fundamentally centered around locally accountable individual members of parliament.
In Turkey, however, as we shall see, even local accountability is limited. Not only is the electoral system one of (apparently closed) party lists, rather than nominal votes for locally elected and accountable members. More significantly, it has a very high national threshold, notwithstanding the existence of numerous regional multiseat districts.
Additionally, a nationwide threshold can introduce disproportionality, but it has the potential advantage of ensuring that parliamentary blocs meet a minimum size, while ensuring that all parties that cross the threshold are treated fairly. Most list-based electoral systems with nationwide thresholds result in each represented party having about the same advantage ratio (% seats/% votes) as every other. We can expect voters and politicians alike to adapt to such thresholds, with the result that strategic entry (of parties) and strategic voting kick in, and few parties just fail to meet the threshold. Instead of strategic voting district-by-district, as occurs in FPTP systems, it is strategic voting based on expected national party vote shares. That’s perfectly consistent with the competing-parties logic of list systems.
Turkey’s electoral system really has the worst of all of these provisions. On the one hand, manufactured majorities, but without the local and individual accountability of members. On the other hand, a nationwide threshold, but in the context of exclusively district-based allocation of seats. Unlike many two-tier proportional systems, in which seats are allocated first at a local multiseat district, and the threshold is applied to national (or regional) compensation seats after the local results are known, in Turkey the nationwide threshold is applied first.
The Turkish system has local competition in self-contained districts–79 of them in all (for an average magnitude of around 7)–but the seats in these races are allocated only after the nationwide votes are tabulated and it is determined which parties have cleared the nationwide threshold. And that threshold is high, at 10%. This is the highest threshold I have ever seen, but I want to keep the stress here not on the absolute magnitude of the threshold (significant though that is), but on its unusual application in a way that overrides the local accountability otherwise implied by a system based on medium-sized local districts.1
This is a genuinely perverse system. It often results in districts being represented only by the parties that placed third and fourth, simply because the leading parties in the district did not have 10% of the national vote. For instance, in Afyon, the AKP won 6 of the 7 seats on 42.6% of the vote. The seventh seat was won by a party with only the fourth highest local vote total. With a more typical d’Hondt PR allocation of the seats, the AK would have won four seats, and each of three other parties would have won one each.
Or take Agri, where the AKP won three of five seats on a mere 17.6% of the vote. The leading party in the district, the DHP, had 35% of the vote, yet won no seats. Two seats were won by the CHP, despite its having only 9.6% of the vote. The DHP was similarly shortchanged in Van, where it won 40.9% of the votes but no seats, while the AKP won 6 of the 7 seats on only 25.8% and the fourth-place CHP the other seat on just 5.2%.
In Ardahan, the AKP had the fourth highest vote total (11.7%), yet won one of the two seats, while parties with 18% and 15.8% went without representation.
The Turkish electoral law allows legislators to be elected on local strength if the candidate runs as an independent. The district of Sirnak offers a good demonstration of the impact of this provision. There, one independent was elected with just over 10,000 votes (9.7%), while a party with just over 1,500 more votes than the independent won no seats. Oh, and incidentally, another party had over 47,000 votes and elected no one. (In this district, the AKP won two seats on 14,512 votes.) The Sirnak example gave rise to a case before the European Court of Human Rights.
Earlier this year, the Court ruled that the Turkish electoral law was not in violation of Article 3 of Protocol No. 1 to the European Convention on Human Rights.2 Two candidates had alleged that the 10% threshold’s exclusion of parties who won more votes locally than other parties or independents who won seats “interfered with the free expression of the opinion of the people in their choice of the legislature.”
Whether it is a human rights violation or not, the exclusion of all votes cast for parties that fail to cross the nationwide threshold obviously lowers significantly the threshold at the district level for non-party candidates. Another example, the district of Sanliufra, is perhaps even more egregious. There an independent won with only 4.4% of the district vote while three parties with more than 10% (including one with more than 19%) went unrepresented. (Three seats in Sanliufra were won by the CHP on 9.9%.)
The low threshold for independents makes me wonder why more candidates do not run as independents, clandestinely backed by locally strong parties that might not pass the threshold. (I assume there are privileges in the electoral law for registered parties that discourage the practice.)3
Turkey’s electoral results clearly show a pattern of significant regional variation in the vote. As in many developing countries, national fragmentation masks considerable local bailiwicks of strong support for nationally minor parties. Yet the electoral system completely fails to represent this reality of Turkish regionalism.
Is the Turkish electoral system democratic? I think a strong case can be made that it is not.
1. It is actually a good deal more complex. From the IPU description, we find that there are several restrictions on parties:
Party-list proportional representation system using the d’Hondt method, with restricted options and a double barrier (at the local and national level). Accordingly, a candidate from a political party can only be elected if the party (a) is fully organized in at least half of the provinces and one- third of the districts within these provinces; (b) has nominated two candidates for each parliamentary seat in at least half of the provinces; (c) has obtained at least 10% of the valid votes cast nationwide; and (d) has received, in the constituency in question, valid votes at least equal to the applicable simple electoral quotient. Subject to certain conditions and exceptions, vacancies arising between general elections are filled through by-elections. Voting is compulsory, abstention being punishable by a fine. [Emphasis mine]
2. Thanks to Jonathan and Vasi in a comment thread at The Head Heeb for this tip.
3. In addition, it appears that it may simply be difficult for independents to obtain votes. A story in the IHT sent to me by one of my students, notes that independent candidates must pass out their own ballots. Apparently, the official state-provided ballots list only the parties. The story actually gives the impression that this makes it easier for independents. I don’t think so (and neither does my student). Maybe for a very well organized campaign, but otherwise, it has to be easier for a voter to obtain the official ballot and select one of the options already depicted on it. Besides, if the provision made it easier, we would expect to see many more independents, because running as an independent bypasses the risk that one’s party might not clear 10% nationally. (And, again, there is also the possibility that there are other legal provisions that give incentives to run as a party, such as campaign finance or the allocation of parliamentary rights.)
Independents in Turkey have the option of appearing on the general ballot *or* of handing out pre-filled ballots. The 2002 election, in which many parties fell below the 10% threshold, was unusual in Turkish history, where at least 4 generally make it through. That being said, most years it leads to a relatively better representation of the national vote than first-past-the-post.
Other than that, I agree with the general thrust of your argument that Turkey’s system can lead to lower-than-satisfactory representation. But Turks have a (I think justified) deep-seated fear of splintered parliamentary representation leading to unworkable governments as in the 1970s. The present electoral system was not handed down from heaven but is a response to Turkish political history. Over time, and as Turkish parties re-establish their roots, one would expect to see a decline in the cut-off to 5%.
Thank you very much, Darcy, for the corrections and amplifications. What would explain the preference of an individual candidate for one type of ballot or the other?
Yes, like many contemporary electoral systems, Turkey’s is an “engineered” response to previous crises. (I am not aware of any electoral systems having been handed down from on high!) I suspect there are better ways to engineer the desired outcome than what Turkey currently has, however.
I guess we will see how well running as independents works, since the DTP (Kurds) will run its candidates as independents this time around, assuming the establishment still doesn’t find a way to block them.
It seems increasingly unlikely that there will be a manufactured AKP majority this time around, given all the maneuvering taking place by the “secular” parties.
Interestingly, the system in use in Turkey before they adopted D’Hondt PR within constituencies (albeit with that bizarre threshold requirement) is very similar to the system in use in Singapore: more and more of single-seat constituencies there are being merged into Group Representation Constituencies, with a winner-takes-all system.
Opposition parties tend to poll above 30%, but currently only have one voting and one non-voting members in the 83-seat Parliament.
“What would explain the preference of an individual candidate for one type of ballot or the other?”
I would guess that too many parties spoil the broth. Turkish ballot papers can be very unwieldy. Presumably, it was much nicer to hand out a single ballot.
However, I understand that this system has been scrapped. All independents will now be on the proper ballot.
Pingback: Fruits and Votes
Pingback: Fruits and Votes
Given the new Italian electoral law will set the threshold for a single party at 8%, lists will still be closed, and electoral regions will be small, do you think that Italy is coming close to the Turkish-style system? Sure, Italy does not have an ethnic minority in the vein of the Kurds, but it does still seem to be a system designed to stifle new parties in the name of ‘stability’.
That is quite a restrictive system for Italy. Is it adopted, or still a proposal?
Still a proposal. I kind of shortened the idea, but a good outline is here:http://goo.gl/M7ktfU.
It’s a somewhat higher threshold for Italy, but if we consider that the effective number for a coalition was 10%, it’s not that different (it would favor larger parties within a coalition in the place of their smaller allies, as coalition members now need at least 5% of the total to win seats rather than 2%). For a stand-alone party it doubles from 4% to 8%, which is a high standard, but the Grillini were new at the last election and won over 25%, so it’s achievable.
Pingback: Turkish election, 2015 | Fruits and Votes
Pingback: Turkey, 2018: Unusual alliance behavior | Fruits and Votes