The following comes from Wilfred Day, whom I thank for helping keep me up to date on developments in electoral reform.
Will the winds blow MMP’s seeds across the Tasman sea and the Indian Ocean to Sri Lanka?
Two years ago the Interim Report of Sri Lanka’s Select Committee on Electoral Reform recommended “a combination of both the FPTP and the PR system, resembling the German electoral system.”
Today the People’s Daily reports
that all parties represented in the parliament have agreed to the mixed system and a final decision would be made on May 29.
[The rest of this post is mine; the first paragraph has been revised since the original.]
Sri Lanka currently has a system of intra-list preference votes. I had thought that the system was a flexible list (defined as one in which a party-give rank order prevails except for candidates who obtain some stipulated quota of preference votes allowing them to move up ahead of higher-ranked candidates). However, in the comments below, Wilf says he believes the lists are open (i.e., preference votes alone determine the order in which candidates are elected from a list). The second link that Wilf supplied (above) notes the difficulties the battle for preference votes has caused:
Critics have pointed out that the people’s wishes had not been properly addressed in the proportional representation and it had also led to unhealthy intra-party competition among candidates of the same party.
The system of two-tier (and apparently parallel) PR is quite proportional currently. For instance, in 2004, the United People’s Freedom Alliance won 105 of the 225 seats (46.7%) on 45.6% of the vote, and the runner-up United National Party won 82 seats (36.4%) on 37.8% of the vote. The Lanka Tamil State Party won 22 seats (9.8%) on 6.8% of the votes, showing that the regional multi-seat PR districts can allow a regionally concentrated party to obtain some bonus in the context of overall proportionality (probably not a bad thing given the Sri Lanka context).1
Given the current seats distribution, it is hard to see how “the people’s wishes” might not be “properly addressed.” This is even more puzzling to me than the claim about “unhealthy intra-party competition.”
And, to add further confusion to the People’s Daily report, it says:
Another setback in the present system is that no party is able to gain decisive parliamentary majority in the assembly consisting of 225 legislators.
Normally, one does not expect “proportional” systems to do this. I wonder if what is actually being considered is MMM, which would be “mixed” but would also generate seat majorities (given a largest party in the 40-percent range) more reliably than either the current system or MMP. The first link above is also not clear on the nature of linkage (if any) between the FPTP and PR seats, though one of the parties clearly thinks Germany’s MMP is not “P” enough:
S. Subairdeen, Leader of the Ashraff Congress (AC): “…instead of a prototype of the German system, which is based on 50% PR and 50% FPTP, we want to suggest more percentage of representation under the PR so that the minorities will be benefited . If necessary the number of parliamentary seats should be increased for this purpose.
Of course, the German system could hardly be more proportional than it already is, for parties that clear the threshold. It seems the Sri Lankan party negotiators (and the Chinese People’s Daily reporters covering them) are not very well informed about key intricacies of electoral systems.
Sri Lanka–the scene of a long and especially brutal civil war–has shown an unusual propensity to undertake major institutional overhaul (though without evident success at resolving the polity’s most pressing problem thus far). It was a Westminister FPTP system at independence, but switched to the above-described “PR” system in 1978. Around the same time it adopted a semi-presidential system with a very strong presidency.2 In 2000 the powers of the presidency were sharply reduced, making it closer to a standard “premier-presidential” system.3 Various autonomy and federalizing arrangements have also been discussed as potential means to resolve the separatist conflict based in the Tamil-dominated regions in the northeast of the island.
1. The Tamil-dominated area includes the districts of Batticaloa (magnitude of 5), Jaffna (9), Vanni (6), and Trincomalee (4). The Lanka Tamil State Party won from 64 to over 90% of the votes in these districts and 19 of the 24 seats available in these districts.
2. Of the “president-parliamentary” hybrid type: The president could freely apppoint and dismiss the cabinet, which also could be dismissed by the parliamentary majority, and the president held extensive emergency powers. The most similar design of executive-legislative structure that I have seen to that of Sri Lanka during the 1982-2000 period would be that of the former Weimar Republic of Germany.
3. Many sources referred to the 1982 adoption of direct presidential elections as a move to something resembling the “French” model, but it was not much of a resemblance. However, since 2000, the family resemblance has been strong. The president still has authority to appoint a prime minister, but the PM forms the cabinet, which depends now on the exclusive confidence of the parliamentary majority–as in France. There are still emergency powers, though the inability of the president to maintain a cabinet of her own choosing limits them in practice–just as in France. The president also retains dissolution power, but the threat that new elections may force the president to accept a “cohabitation” cabinet often limits the gains she can expect from dissolving the assembly–again, as in France. Sri Lanka has already experienced cohabitation under the weakened presidency, until the 2004 legislative and 2005 presidential elections restored unified partisan control of the presidency and cabinet.
As to intra-list preference votes determining which of a party’s listed candidates will be elected, the only reference I can find says that the are decisive for the 196 normal seats (not counting the further 29 national-tier seats.) The Constitution of Sri Lanka says, in article 99, that votes are cast for a party, and a voter may also cast preferences for not more than three candidates of that party; and the candidates securing the highest numbers of preferences are elected. It seems that, as in Brazil, the order on the party list counts for nothing. Still, Brazil allows only one preference. Sri Lanka, by allowing three, is more flexible, perhaps encouraging voters to choose the top three, and reducing the premium for a single “strong man” candidate.
It’s also interesting that Sri Lanka was going to have a local threshold of 12.5% in each district, but quickly came to its senses and dropped to the normal 5%.
Seed planted by Wilf Day — 27 May 2006 @ 00:52
On Sri Lanka, I stand corrected: Lists are open, not flexible. As for allowing more than one preference vote per voter, that was the case in the former Italian PR system, and is still the case in Peru. In Italy, the number of preference votes varied by district magnitude; in Peru, it is always two (I believe).
The distinction between single or multiple preference votes may matter in some important ways, but it does not make the list more ‘flexible’ as that term is commonly used. ‘Flexibility’ contrasts with the ‘rigidity’ of closed lists in that flexible lists allow voters to change the party-provided list order; in other words, when compared to open lists (defined as those in which preference votes are the sole determinant of final list order), flexible lists are always less flexible than open lists, regardless of how many preference votes the voters is permitted to cast.
Presumably, multiple preference votes reduce the zero-sum nature of intra-party competition, which might have the indirect effect of making party reputations more important. Nonetheless, the first target of Italian reformers in the run-up to the abolition of PR was the multiple preference votes: A 1991 referendum reduced the preference votes to one per voter, and the stated reason was that it would reduce corruption. It is not clear to me, however, that it would. In fact, I might expect the opposite, because a corrupt candidate would have to buy fewer voters to ensure a seat when every voter is casting, at most, one preference vote.
With the 1993 electoral reform, preference voting in Italy was abolished once and for all, and has not returned since. Even the 2006 election–an all-list, although not PR, system–had all closed lists.
Seed planted by MShugart — 28 May 2006 @ 09:39
The Expert Panel would propose electoral reforms on the German model with half from the first past- the-post system and the other half from the existing proportional representation system. So if the newspaper report is accurate, Ceylon may indeed get MMP.
Seed planted by Wilf Day — 19 November 2006 @ 22:19
How often do you see the chief Elections Commissioner of a country blame the electoral system for “canvassers badgering you with preference numbers of candidates outside polling booths” as well as many other woes?
In this case, Sri Lanka’s official is complaining of their open list system, not proportional representation per se. Voters can cast three personal (”preference”) votes for candidates, who are identified by number as in some South American countries.
His complaint about “virulent intra-party rivalry” was echoed by the EU Election Observation Mission to Sri Lanka 2004. The hardline leftist and anti-Tamil JVP ran in an alliance with the governing SLFP, but had its own candidates on the UPFA coalition list. “JVP’s policy was to present only three or fewer candidates on the UPFA district lists. As a result, in almost all districts, JVP preference votes were concentrated on their candidates. On the contrary, preferences expressed for the SLFP candidates were much more dispersed among them, and were therefore less “efficientâ€.” The result was that 35.5% of the coalition’s “preference votes” translated into 40% of its MPs in those districts. Meanwhile in the other alliance the SLMC and UCPF ran alone in several districts and supported the UNF coalition in the others.
The 5% thresold is applied district by district, not nationally, “to accommodate insurgent parties which had by then opted out of the electoral process.” In the result two parties won a single seat each.
Still, Sri Lanka has a slightly impure PR system: in each district the largest party gets a “bonus” seat before the others are distributed on a “highest average” calculation. Of the 196 MPs returned from the 22 districts, then, 174 are proportional to the votes, and a further 29 MPs are elected from the National Lists in a parallel proportional calculation.
Seed planted by Wilf Day — 18 July 2008 @ 14:12
I have seen Sri Lanka’s current electoral formula identified as “mixed” in some literature, but of course it is not.
It is simply a divisor sequence of 1,1,2,3,4,…
In other words, the first party gets the first two seats, and then gets the third only if it has twice the votes of the second party (the same criterion as D’Hondt for the first and second seats).
(I think Turkey uses the same, no?)
On the preference voting through candidate numbers, is this feature specifically South American? That is, do none of the European open-list (or ‘flexible-list’) systems have voting by candidate numbers? I wonder about Indonesia, another developing-country OLPR system.
Seed planted by MSS — 21 July 2008 @ 17:13
Didn’t Finland traditionally use writing-in the ID number of your preferred candidate?
Australia used that, but only once - the 1997 postal ballot for the 76 elected delegates (2 to 20 at large in each State/ Territory who formed half of the Constitutional Convention (”ConCon”), called by the Commonwealth Parliament as an advisory body to draw up a proposed Republic model to be put to referendum. (The other 76 were notables appointed by the federal and State cabinets). The system was very similar to Senate-style STV with ticket-voting, except that only names of teams were shown on the ballot: if you wanted to number individual candidates, by contrast, (a) you couldn’t give more preferences than the number of seats in your State/ Territory, and (b) you had to write in their 3-digit ID numbers, as shown in an accompanyng booklet.
Ironically, this “truncated” STV system led to a result that has never occured in 60 years of Senate PR elections (with candidates’ names all listed on the ballot paper itself). A
candidate won a seat, on personal votes, from a slot “too far down” the ticket to benefit from the list-votes. Hazel Hawke, ex-wife of Bob (Labor Prime Minister, 1983-91) was elected as one of the Australian Republican Movement’s 7 delegates, out of 20 from NSW, even though she was “only” Number 12 on the ARM’s ticket. (The parallels between Bob Hawke and Bill Clinton have always been rather eerily exact.)
Seed planted by Tom Round — 22 July 2008 @ 07:29