The small island nation of Vanuatu held an election for the Parliament on 22 January. The election was called after fourteen MPs, including the Deputy Prime Minister and a number of other cabinet ministers, were sentenced to prison terms between 3 and 4 years for bribery. Despite an attempt by the Speaker (who is acting president when the President is out of the country, and is also one of those found guilty of bribery) to pardon himself and the other MPs accused, the sentences stood. Following this, President Baldwin Lonsdale (on the advice of Prime Minister Sato Kilman, who knew that he would likely face defeat in Parliament when it was recalled) called an early election.
I have written more extensively on Vanuatu's political history here, but that's quite a long story. It was initially a colony, governed by both the United Kingdom and France (with two entirely seperate systems of government, justice and law enforcement). At independence in 1980, Vanuatu's politics was dominated by two parties; the anglophone, socialist Vanuaku Party, and the more conservative, Francophone Union of Moderate Parties.
However, over the years, this two-party system has totally disintegrated. In 1991, two new parties split from the Vanuaku Party, and this has started a trend of both party splits and new parties. At the first post-independence election, the effective number of parties in terms of seats was 2.1; at this election, it was 11.36 (and even that number is an understatement, as it treats independents as one party). This has led to very unstable governments; in the 2012-2016 parliament, there were four Prime Ministers, and in the 2008-2012 term, there were six.
While personality clashes within parties have lead to some of this instability, the electoral system is also a factor. Vanuatu is one of the few countries to still use the single non-transferable vote (introduced before the 1975 colonial elections by colonial officials who were not provided with any information on alternatives). As many of you will know, this system is linked with factionalised parties, and Vanuatu is no exception. However, unlike Japan and Taiwan, the factionalised parties at the start have not stayed together, and have split. Under SNTV, there is relatively little incentive for parties not to split, at least from an electoral perspective.
I was in the country up until polling day, for unrelated reasons, and had a look at some of the campaigning (for images of posters, see my Twitter account). In terms of political reform (and with the caveat that I don't speak the language, though Bislama is not too hard to understand) I noticed little discussion, other than from the Vanuatu Presidential Party (whose position should be obvious). There was vague talk about ending corruption and supporting decentralisation, but nothing, as far as I could tell, about the electoral system.
Vote distribution is an important part of the single non-transferable vote. If a party nominates too many candidates, their vote will be too split to elect many MPs; if they nominate too few they risk winning not enough seats. The vote also needs to be spread equally amongst the candidates, so as to ensure the maximum number of candidates are elected. You would think that a nation like Vanuatu, which has used the single non-transferable vote for a long time, would be doing this well. This is not the case.
For example, in the seven-member district of Santo, the Union of Moderate Parties was the largest party in terms of votes, winning 14.34% (yes, that makes them the largest). However, they ran five candidates. Each one got no more than 4%, and they won no seats. In the two-member district of Epi, the Vanuatu National Development Party was the largest party, with 20.78% of the vote. However, they ran two candidates, and each one got about 10%. The two seats were won by one independent and one UMP candidate, who was the only UMP candidate.
These sort of errors were widespread in Vanuatu, and the incompetent attempts for the larger parties to divide their vote only compounded the fragmentation, as they allowed small parties with only one candidate to win seats easier. The below chart shows how the seats were distributed, and how they would have been distributed under the D'Hondt system (applied only in areas where it would have made a difference).
As you can see, the new system would be a boost to the largest parties (the VP and UMP), while it would remove some of the smaller parties and independents. This would likely be an advantage for government stability.
One other thing I am interested in comparing is the equation outlined in Shugart, Bergman and Watt (2013), for estimating the vote share for the first candidate under the single non-transferable vote to the results in Vanuatu (this equation being 'P1=C-.75, where P is the vote share within the party of the highest polling candidate and C is the number of candidates).
The below chart shows the correlation between the expected and the actual vote shares within parties is fairly strong.
One final interesting point; the below graph plots the difference from the expected vote share for the highest polling candidate against the advantage ratio for the party in that district.
It's worth noting that the linear trendline (as well as the correlation coefficient of 0.288) in this graph is relatively meaningless; the graph is dramatically skewed by the single data point in the top right. If removed, the coefficient drops dramatically, to -0.03. However, I don't have anything to compare this against, and I am curious as to whether the high number of independents and single-candidate parties means that parties where the highest polling candidate polls higher than expected do better. I intend to look into this further.