Home » Research Blog :

The Majoritarian Bias in the Georgian Parliamentary Elections: Prospects for 2012

When discussing election results there is a fairly considerable tendency to talk about the party list and national polls. But Georgia’s Mixed Member Majoritarian system (MMM) means that 73 out of the 150 seats are majoritarian, and so are allocated on the basis of whoever gets the most votes in a given geographic region.

This system can give the largest party a ‘majoritarin advantage’. In the 2008 Georgian Parliamentary election, it helped United National Movement (UNM) to win 71 out of 75 majoritarian seats (there were two more majoritarian seats in 2008). As a result, with 59% of the national vote they got 79% of the seats in parliament. The opposition, by comparison, got 41% of the vote and only got 21% of the seats.

This ‘majoritarian advantage’ is a common characteristic of MMM voting systems but the scale of the advantage is not fixed and how it emerges is complex. There are four main reasons why the national movement did so much better in the seat count in 2008 than they did in the national vote, and the extent to which those characteristics are repeated will have a significant impact on the number of seats in parliament they secure this time.

First, majoritarian seats tend to be naturally inclined towards large parties. The size of this effect is actually determined by the lead that the main party has over the next largest party and by the national variation in the vote.

A low variation in the national vote means that small differences between parties become amplified. For example, if a party only has a 2% advantage over its opponent (say in a simple 49%/51%) and the support is absolutely consistent nationally, then they would expect to win all of the seats. However, if the national vote is varied (with some regions more supportive than others) then the electoral lead has to be larger in order to secure the same number of seats.

In 2008, the main reason why UNM did so well in terms of majoritarian seats was that there was a HUGE difference between them and the next largest party. Facing a hugely fragmented opposition, UNM got 59% and the ‘Joint Opposition’ got 17%, a difference of 42%.

Because there was such a huge gap, the variation of voting was rendered largely irrelevant, but national variation in voting was actually pretty large. 10 seats gave UNM less than 50% of their vote (with Kazbegi, at 34% the most pro-opposition) and ten seats gave them over 80% of the vote (with Ninotsminda the most pro-government with 97%). Between those extremes the districts were fairly evenly spread out.

However, in this election the wide geographic difference in voting patterns means that if the overall difference between UNM and Georgian Dream (GD) is fairly narrow, the large party advantage will also be fairly small.

Second, the bias towards the larger party is enhanced if those opposing the leading party are in competition with each other. This is often understood in terms of the opposition ‘splitting the vote’. The logic of vote splitting is simple. If the government has 40% of the vote and the ‘anti-government’ has 60%, then the ‘anti-government’ vote should win. But if this vote is split between two or more opposition parties and each of them gets less than 40%, then the government wins.

This is made possible because the threshold for gaining a seat is only 30%. So, the seat will usually, simply go to the party with the most votes. In this situation, if the race is close, one would routinely expect the winner of a majoritarian competition to win with less than 50% of the vote.

Again, in 2008, the UNM victory was so extreme numerically that vote splitting was rendered almost irrelevant. UNM won more than 50% of the vote in 63 (or 86%) of the majroitarian districts. So, even if the opposition had operated as a single block, they only would have picked up another 6 majoritarian seats (totaling 10).

Oddly, vote splitting might be more important in this election, even though the opposition vote will be more concentrated with GD. If the vote is, say, 45% government and 55% opposition – but 10 to 15% of the opposition vote goes to opposition parties other than GD, then UNM could have a 5% edge in individual districts and might be expected to win 60% of the majoritarian seats.

Winning 45% of the national vote and 60% of the seats would give them just over 50% of the seats in parliament. Note, that under these circumstances they could credibly claim to be the winner, in that they would have more votes than any other party, but the system would allow them to avoid being forced to form a coalition. This dynamic is very common in majoritarian systems.

The third advantage for UNM, allowed by the majoritarian system in 2008, was that it let a ‘local personality dimension’ into the race. When people vote for a majoritarian, they are not just voting for a party, but for an individual who may have a local reputation, network, or perhaps a more nefarious local power structure. Because UNM has history of influence and victory, it is sensible for local power-brokers to side with them, so one would expect their majoritarian candidates to do somewhat better than their party list.

The fact that voters vote separately for the majoritarian candidate and for the party list, gives us a neat way to mathematically look at the overall effect of this ‘local leader’ bias. In 2008, UNM gained an average of a 3% higher vote for its majoritarian candidates than it gained for its party list. The opposition gained a corresponding average of 3% lower for their majoritarian candidates.

This average of ‘3%’ does suggest that, generally speaking, list-voting numbers and majoritarian-voting numbers track fairly well. This, in turn, suggests that in most cases people are voting for the party more than they are voting for the person. Therefore, it would be strange for this division to grow dramatically. However, this 3% average hides considerable variation. In 22 districts the variation between majoritarian and party list was more than 5% and in 5 of those, the difference was over 10%. So, the personality angle could give UNM the edge, if voting is close.

The fourth bias in the Georgian Majoritarian system is created because opposition supporters are more concentrated in the cities, particularly Tbilisi, where votes are less valuable. Georgia’s majoritarian districts have wildly divergent populations. The average majoritarian district has 49, 505 registered voters (based on the Central Election Commission numbers) but the largest district of Kuitaisi has 163 thousand and the 10 districts in Tbilisi average over 100,000. The smallest district, of Kazbegi has 5,810. However, no matter how many people they contain, each district only gets one MP. Therefore, a vote is worth a lot less in big cities than it is in other districts. To put this in practical terms, the number of votes required to win Kutaisi would win you all 13 of smallest districts.

In addition, a massive win for GD in Georgia’s biggest cities would involve essentially ‘wasted’ votes because, in, a majoritarian election, any vote over 51% gets one nothing. So, one could imagine a scenario where GD might win big in Tbilisi, Kuitaisi, Rustavi and Batumi (where the opposition did relatively well in 2008), pushing up their party-list average above that of UNM, but still leaving them trailing in the other majoritarian districts, where votes are more valuable.

Altogether, what is one to conclude about the likelihood and size of a majoritarian bias in the upcoming elections? The biggest factor determining the bias is the gap between UNM and GD. If the gap is small, then the bias should be relatively small, but if the gap is big for either party, so will be the bias. So, the bias is very unlikely to be as big this time as it was in 2008.

However, if the election is close, while the ‘majoritarian bias’ may not be large, it could tip the balance so that a party that loses the party list could just about win a majority in parliament. This is perhaps the biggest problem with the mixed majoritarian system, particularly in a new democracy like Georgia. Because, while perfectly legal, such a result would be difficult to explain to the electorate.

Special thanks to Transparency International, GYLA and ISFED for supporting www.electionportal.ge which has detailed historic data on Georgia’s elections. That, along with the Georgian CEC, were the two sources for this article


Recent Blogs

Sunday, 07 October 2012
The Political Make-Up of Georgian Dream
Tuesday, 02 October 2012
The power of the Prime Minister
Sunday, 30 September 2012
The Majoritarian Bias in the Georgian Parliamentary Elections
Monday, 17 September 2012
Making Sense of the NDI Poll
Tuesday, 31 July 2012
Georgian Agriculture: Recovery After a Decade of Neglect
Tuesday, 03 July 2012
Employment needs in Tourism, Apparel, ICT and Food Processing