Elections of late May, 2014

The next few days have a lot of elections!

The European Parliament elections have been ongoing and conclude Sunday, which is also the day for the election of:

President of Colombia (first round)
President of Ukraine (first round)
Parliament of Belgium
President of Lithuania (second round)

Then we have the two-day election of the president of Egypt. (I could have said the ratification of the coup led by Abdel Fatah el-Sisi).

In the Colombian election, this will be the second time an incumbent is seeking reelection, given that the no-reelection clause was only lifted during the first term of Alvaro Uribe, who won a second consecutive term in 2006. Of particular interest in this election is that it pits incumbent Juan Manuel Santos, Uribe’s preferred successor in 2010, against a new candidate aligned with Uribe (as well as other candidates). The two had a falling out, mainly over policy towards the insurgency. But the Partido de la U, originally founded by and in support of Uribe, remains the party of the incumbent.

The Ukrainian election represents the attempt to return to electoral legitimacy following the ouster in February of Viktor Yanukovych in the popular uprising. One of the biggest questions in this election is whether it can be effectively conducted in parts of the southeast where separatist insurgents are threatening government control.

Belgium’s federal election is concurrent with the regional and European elections, which I believe is a very rare combination.

How many losers do you need?

For a project I am currently working on, regarding district-level vote fragmentation, I am toting up the number of losing parties (with any vote share) in electoral districts in various countries.

Here are a few cases that I have completed. This first list of countries gives the number of losing parties, averaging across districts and (usually) more than one election per country.

Albania, 36.2
Australia, 2.5
Barbados, 1.4
Canada, 4.2
Czech Republic, 15.5
Israel, 11.6
New Zealand, 2.3
Spain, 14.7
UK, 4.6

The figure for New Zealand refers to several pre-MMP elections.

This is a small set of countries, and I am currently expanding it. But what really jumps out here is how many partisan choices Albanian voters have. This refers to the 2009 and 2013 pure-PR elections only. The mean district in these Albanian elections has a magnitude of 12, a mean number of winning parties of only 3… and thirty-six losing parties!

Some of the other PR cases also have a lot of losers, but none like Albania.

It is interesting to see how many losing parties are typical of the two big FPTP countries, Canada and UK, with both averaging more than four per district.

Another interesting summary statistic is what vote share the losing parties average. In Albania it is 1.05% per party. That sums to a LOT of wasted votes! Spain, too, has a high average vote share for losing parties for a PR system, at 2.8%. In the Czech Republic it is 1.2% and in Israel (with nationwide PR, rather than districted) it is only 0.4%.

Thus we should also consider the average percentage of the first losing party in a district:

Albania, 4.6
Australia, 35.8
Barbados, 39.1
Canada, 28.5
Czech Republic, 4.7
Israel, 0.94
New Zealand, 32.9
Spain, 6.4
UK, 28.6

Of course, in FPTP systems the average shares for first losers and thus for all losers are especially high, but the larger losing parties in any given district tend to win some seats elsewhere.* In the PR cases, a lot of these losers are not winning anywhere (or might win in just one to two high-magnitude districts); they are just small parties that have no chance at all. I wonder what it is about Albanian party law, or other features of the country’s politics, that contribute to so many micro-parties running.

_______
* The third party (second loser) in the FPTP cases of Canada, UK and (formerly) New Zealand averages 14-15%. The average winner in Canada and the UK, by the way, is at almost precisely 50% (in NZ it was 52.8%).

Most of the raw data from which I calculated the above numbers come from the Constituency Level Electoral Archive, although I am augmenting it from various other sources. The list of countries shown here is the subset on which I am currently working: parliamentary democracies with “simple” electoral systems by Taagepera (2007) criteria, meaning no second round or upper tier. Thanks to Cory Belden for her research assistance; some day she will get a much better acknowledgement than one on a blog post…

April 2014: lots of elections

April is a good month for election-watchers.

Today Hungary votes for the first time since the constitutional and electoral reforms imposed by Fidesz following the two-thirds majority granted it by the country’s mixed-member very-unproportional electoral system.

Monday is the general provincial (or “national”?) election in Quebec. The final ThreeHundredEight.com projection shows the Liberals most likely will win a majority, although the estimated range includes the possibility of a minority government. The Parti Quebecois (PQ) can be said confidently to be in second place, according to the projection, while the Coalition Avenir Quebec (CAQ) has closed in on the PQ in votes. The PQ’s lead over the the third place CAQ is now only around five percentage points, but it is projected to get only one fifth as many seats as the PQ. With projected votes breakdown of 39.0-27.6-22.7, but seats of 69-45-9, it is quite a “non-Duvergerian” result in the votes, but fairly Duvergerian in seats. The mechanical effect of FPTP will make the legislative party system much more two-party than the votes, by significantly over-representing both of the larger parties. The Liberals and PQ have been targeting ridings (districts) held by the CAQ, but it may not be working. There are indications that the CAQ is the party with the momentum, though getting into second place in votes is both unlikely and would not get them out of third place in seats. The fourth party, Quebec Solidaire, appears to have two safe seats despite only 8.4% of the projected votes; it is great to be small party under FPTP if your supporters conveniently concentrate their places of residence! (There were two earlier F&V posts about this election campaign: on by JD on the pro-PQ bias, and one by me on the surprising mid-campaign swing against the PQ.)

Also coming right up are general elections in two of the world’s largest democracies. Indonesia’s legislative elections are this Wednesday, 9 April. Indonesia uses open-list PR,* and is the world’s largest democracy to use any kind of PR (just ahead of Brazil, which also uses open lists, and votes later this year). Indonesia uses a counterhoneymmon electoral cycle, with the presidential election coming on 9 July.

Also this week marks the beginning of the biggest voting exercise of them all, India. The general election will take place in nine phases,** starting 7 April and ending 12 May (see map with schedule). Results will be declared in mid-May. Indications from polls are that the opposition Bharatiya Janata Party and its National Democratic Alliance (NDA) of various state-based parties will win, and that it will be the worst outcome ever for the Indian National Congress and its ruling United Progressive Alliance. The NDA likely will be short of a majority, however, and will need outside support from other state-based parties that have joined neither alliance for the election. BJP leader Narendra Modi, current Gujarat state Chief Minister, is the likely Prime Minister. However, if the NDA is short of a majority, it can’t be ruled out that the support parties could demand a different PM, given what a lightning rod Modi is for communal tension.

________
* And SNTV in the largely powerless second chamber, also being elected Wednesday.

** Or “stages”. Please don’t call them rounds.

Costa Rica’s record fragmentation

I already said this in a comment at an earlier thread, but it is the sort of thing that is at the very core of Fruits & Votes, and hence deserves its own space…

By my calculation, the effective number of vote-winning parties in Costa Rica’s 2 Feb. election was 6.21. The effective number of presidential candidates was 4.36. Both easily break the record for the country’s elections back to 1953. The old averages and maximums were 3.21 and 4.84 for vote-winning legislative parties and 2.37 and 3.30 for presidential candidates. (Historical figures based on Bormann and Golder’s dataset.)

Costa Rica’s party system sure is unrecognizable from what it was for so long!

___________
The effective number of parties or candidates is by now the most well established measure of electoral or legislative fragmentation. It is simply a weighted count of the number of components (parties, candidates, or anything else) where the components are weighted by their own size through squaring them. Operationally: square each component’s share (out of 1), sum the squares, and take the reciprocal of the sum. Originally proposed by Laakso and Taagepera in a 1979 article.

How unusual is South Africa?

[Update: I added in a comment a list of cases with streaks of four or more elections with largest party having over half the vote.]

Given the news that a proposed merger of two opposition parties in South Africa was called off, I was wondering just how unusual is South Africa in having had a string of elections in which one party won a substantial majority of the vote.

Querying a dataset of over 1000 elections in countries that can be considered democracies (including many developing countries and young democratic regimes), the answer is: quite unusual.

The mean vote share of the largest party in these 1,148 elections is .42. In South Africa’s four democratic elections to date, the ANC has averaged .64. Such a high figure is just beyond the 95th percentile (.63) of the worldwide distribution.

Of course, the failure of opposition parties to offer a clear alternative does not help the cause of “normalizing” South Africa relative to other democracies. On the other hand, the extreme PR system also is an impediment. While normally, the more extreme the PR, the less likely is a party with over half the votes, if your starting point is a hegemonic party, then extreme PR presumably inhibits the development of an opposition that could challenge the leader. That is, numerous very small parties can all win their share without unifying, and the benefits from unifying are less clear for parties than they would be in a country with either a less proportional system or direct presidential elections (or both).

Speaking of direct presidential elections, the distribution of the worldwide data does not change much if we focus only on parliamentary systems, although there is small (and totally insignificant) tendency for bigger shares for the largest party under presidential systems (mean .43, although 95th percentile .61).

Democracy ranking

Democracy Ranking has the USA ranked at 15th in the world, just behind Australia, Canada and the UK, and ahead of France.


The Democracy Ranking of the Quality of Democracy 2013

Explore more infographics like this one on the web’s largest information design community – Visually.

One might note that this little cluster of long-term democracies at ranks 12-16 includes those that retain majoritarian electoral systems for at least their first (or sole elected) chamber. I will leave it to others to determine whether that is causal or coincidence, while noting that among long-term democracies, two PR (or mostly so) countries rank a good deal lower: Israel and Italy.* These just happen to be every anti-PR advocate’s favorite cases, notwithstanding that the top eleven are all long-term democracies using PR (including New Zealand, PR since 1996, at rank 8).

Japan, the other long-term, developed-country democracy with a non-PR electoral system, ranks at number 20. I know, correlation is not causation. But the correlation is noteworthy!

The full dataset can be downloaded from the linked page.



* Italy has not used a fully proportional system since 1992. After a period of mixed-member majoritarian from 1994 to 2001, it switched to a bonus-adjusted PR system that guarantees a majority of seats in the first chamber to the party or alliance of parties that gets a nationwide plurality. So any arguments about PR and flaws in Italian democracy are seriously out of date. Yet it remains true, especially since the 2005 reform, that Italy’s electoral system makes it easy for very small parties to gain representation.