The skewed allocation of Turkey’s parliamentary seats was highlighted recently in a comment to a post I had written about the prospects for the proposed new constitution. The specific example given was the mismatch between Bayburt (with a voter-to-seat ratio of 26,700:1 in the 2007 elections) and Istanbul’s first district (where the equivalent ratio was 111,700:1).
To try to flesh this out a bit, I’ve put together the table below, which lists each of Turkey’s provinces. What I’ve done is to calculate a proportional allocation of parliament’s 550 seats to all 81 provinces, so that we can compare it to the actual allocation used in the 2007 general election. For any electoral geeks reading, I’ve used the Sainte-Laguë method, drawing on a recent report on electoral systems by Simon Hix for the British Academy.
The key columns to look at are the ‘SL’ and ‘2007’ ones, which illustrate, respectively, the Sainte-Laguë proportional allocation and the allocation that was actually used at the last election. But before getting to the discrepancies between those columns, let me run through the contents of the other columns.
Each row in the table corresponds to an electoral province and starts with the number of registered voters in that province. The next cell (TE) indicates the province’s theoretical entitlement in terms of seats, on a purely proportional basis. (If we divide the total electorate by 550, we see that there are 77,330 voters per seat. The TE column shows how many blocks of 77,300 voters each province contains.)
The problem with the TE figures is that they’re not whole numbers. By contrast, of course, the seats to be allocated aren’t divisible. The next column (RTE) therefore rounds up or down to the nearest integer. Which is all well and good, but as you’ll see from the ‘Total’ rows at the top and bottom of the table, this rounding process leaves us with 553 seats, meaning we’ve allocated three too many seats.
This is where Sainte-Laguë comes in. Under Sainte-Laguë we need to find a number such that when we divide it into all the provincial electorate figures, the results, when rounded up or down, sum to the number of seats to be allocated. In our case, a bit of trial and error establishes that the number needed is 77,650. If you divide that number into a province’s electorate, you get the number in the SLE column. If you then round those SLE figures up or down, you get the numbers in the SL column, which in turn sum to the total that we’re looking for, 550.
So in the SL column, we have a picture of what a proportional allocation of Turkey’s 550 seats would look like. And in the column labelled ‘2007’ we have a picture of the actual allocation that was used at the last general election. The final column in the table indicates the difference in each province between these two allocation methods. There are some glaring discrepancies.
For example, if we look at the two districts that I mentioned at the beginning, we see that under Sainte-Laguë Bayburt would have its representation halved from two seats to one. And Istanbul’s first district would have 35 seats rather than its actual 24. In fact, when Istanbul’s three districts are combined, the cumulative shortfall amounts to 26. There’s a similar pattern in Izmir (which is eight seats short) and Ankara (nine seats). In total, the allocation that was used in the 2007 elections ‘redistributed’ 61 seats to provinces that weren’t entitled to them in terms of the size of their electorates.
This is a problem in abstract terms, insofar as some individuals’ votes clearly carry much more parliamentary weight than others. But it also has potentially concrete political implications, insofar as the ‘missing’ seats aren’t randomly distributed around the country. As we have seen, more than half of them are accounted for by Istanbul and Izmir alone—these two major cities are grossly under-represented. Or if we look at the recent constitutional referendum results and divide the country into ‘yes’ and ‘no’ camps, then the ‘no’ provinces are under-represented in parliament by 13 seats.
Things get more complicated once we start to try to take the analysis forward to look at how seats end up getting divided between parties at elections. This is because all the other elements of the electoral system start coming into play at that stage, such as the 10 per cent threshold and so on. But the way in which Turkey allocates its 550 seats to its 81 provinces makes for a poor starting point. And it surely plays a role in the imbalance that emerges later when votes across the country are translated into party seats in parliament. (In 2007, the CHP needed 64,500 votes for every seat it won. The MHP needed 70,500. The governing AKP, by contrast, needed a significantly lower 48,000.)
I’ll try to make time in the days ahead to work out what difference it would have made to the 2007 results if the Sainte-Laguë allocation had been used.
|Table 1: Allocating parliamentary seats to Turkey’s 81 provinces|
Source: 2007 voter and seat numbers are from Psephos [http://psephos.adam-carr.net/]. Note: Electorate—registered voters in 2007; TE—theoretical entitlement; RTE—rounded theoretical entitlement; SLE—Sainte-Laguë exact entitlement; SL—Sainte-Laguë allocation; 2007—number of seats allocated for the 2007 general election; Diff—discrepancy between SL and 2007 figures.