One of the expectations of the reformers who advocated replacing Colombia’s former de-facto SNTV (single nontransferable vote) system with list-PR using D’Hondt divisors (and of the foreign political scientists who advised them) was that party-system fragmentation would be reduced.
In the first use of the new system in 2006, there was change, but not much. The effective number of seat-winning parties (Ns) in the single 100-seat district for the Senate fell from 9.33 in 2002 to 7.18 in 2006. That’s not a trivial drop, but 7.18 would still be “too high” by the normative standards likely to be used by almost any electoral reformer.
Now, in the just-concluded second run of the new system, the effective number of seat-winning parties in the Senate was 5.43.
It seems the reform is working.
Additional notes:
1. Already in 2006, we saw expected effects of the reform, even if they did not show up in this common comparative measure of electoral fragmentation. There were many fewer one-seat parties in 2006 (2) than in 2002 (42!). (I guess “many less” is an understatement!) The reason the effective number did not change much was that the size of the largest party fell from 28 in 2002 (Liberal) to 20 in 2006 (La U).
2. The effective number of seat-winners in 2002 was actually greater than the effective number of vote-winning parties (Nv). The latter fell from 8.93 in 2002 to 8.07 in 2006. It is very unusual for Ns>Nv, and that it was so in 2002 was a product of how easy it is to win a seat under high-magnitude SNTV with very small vote shares.
3. In a surprising way, the inability of Uribe to run for a third term might have helped produce greater concentration on a few large parties in 2010. In 2006, as in 2002, Uribe ran as an independent, which complicated the coordination of legislative candidacies around a smaller number of parties–especially given that legislative elections occur before presidential (by eleven weeks). In 2010, the various partisan options for succession to Uribe are clearer, even if the two main ones (La U and Conservative) are competing in the presidential election this time rather than endorsing the same one, as in 2006. (The main opposition party, Liberal, was the same size in 2010 as in 2006.)
4. Here are the preliminary seat totals, presented as 2010 seats, 2006 seats, party name (corrected):
-
28, 20, La U
23, 18, Conservative
17, 17, Liberal
9, 0, PIN
8, 15, Cambio Radical
8, 11, Polo Democratico
5, 0, Verde
2, 0, MIRA
0, 7, Convergencia Popular Civica
0, 5, Alas Equipo Colombia
0, 3, Colombia Democratica
0, 2, Colombia Viva
0, 2, others
In regards to point #3, 2010 is going to be the first time since 1994 and the Samper (PL) win that a president is elected as the officially endorsed candidate of only one party. Pastrana (1998) ran a very consciously multi-party candidacy not dissimilar to watch Uribe did in 2002 and especially 2006.
The only candidate making a somewhat vague run in terms of party is Sergio Fajardo, and at this point his chances strike me as practically zero.
It will be interesting to watch how this develops–especially given the PC-la U coalition in the Congress, yet the clear competition between Sanin (PC) and Santos (la U) for the presidency.
Seed planted by Steven Taylor — 21 March 2010 @ 18:39
I wonder if the de-facto SNTV system used in Colombia encouraged fragmentation even more than the actual SNTV systems used in Japan and Taiwan. In Taiwan, party nominees often campaigned together and asked voters to ration votes equally among them (increasingly common in the last few SNTV elections). In essence, they were telling voters that to think of them as being interchangeable, but that the electoral system prevented them from presenting themselves in this way. In Colombia, nominees did not have this option. They could either actually campaign on the same list and pay a heavy penalty in terms of how many votes were needed to elect each member, or they had to campaign separately and presumably justify this separate campaign by emphasizing differences.
Seed planted by Nathan Batto — 26 March 2010 @ 03:22
Nathan, such an effect might have been plausible in the immediate aftermath of the National Front (which ended in 1974), when many lists continued to elect more than one candidate. However, by the 1990s the system had largely lost its “quasi-” attribute.*
The declining “error” rate throughout the 1974-1990 period (Cox and Shugart, 1995) also supports the notion that the major parties adapted to SNTV much as theory predicts: they informally divided their electorates up to ensure election of several co-partisans.
I have not looked closely at the 1991, 1994, and 1998 elections (yet), but in 2002 (the last SNTV election), the parties were in general pretty good about equalizing their votes across ‘s’ candidates, where ‘s’ is the number actually elected. Many had >s candidates running, but votes for also-rans trailed far behind. This suggests, though does not prove, strategic vote-division. The smaller parties, of course, typically ran only one candidate (technically one list).
_______
* Aside from the fact that the elected member could leave his seat and hand it off to the no. 2 on his list.
Seed planted by MSS — 26 March 2010 @ 16:23
I have wondered if SNTV (and other versions of Limited Vote) would work better (or less crudely) if combined with Tasmanian/ ACT-style rotation of candidates’ order on the ballot (whether in one big column, or in separate party/ team columns). True, the party would need to estimate its total vote fairly closely, but once it had guessed its likely number of whole quotas, and run that number of candidates, this would simplify campaigning greatly. “These are our [four] candidates. Please vote for the [one/ two] highest on your ballot-paper. Unless, of course, you have a particularly strong preference otherwise”.
Filling casual vacancies with SNTV/ Ltd V is, agreed, a headache. No lists, no preferences, but by-elections are too majoritarian… Maybe have vacancies filled by that chamber itself, but authorise an MP to name a certain number of fellow MPs who then get weighted votes when the chamber votes (ie, if replacing an MP from a five-seat district, those named by the vacating MP get 5 votes each, and all other MPs get 1 vote each). I note that some US States permit the State legislators from the same party as the vcacationg delegate/ Senator to name his or her replacement, so this wouldn’t be totally unprecedented but would accommodate Independents. (It would also be useful if the legislature has to replace a vacating member in a collective Presidency – a1 la Bosnia, Iraq, Switzerland – that was elected by PR).
Seed planted by Tom Round — 27 March 2010 @ 05:20
It’s funny that you say that Colombia had lost the “quasi” by the 1990s. When I finally looked at some data several years ago, I was expecting to a world almost exactly like the Taiwanese and Japanese elections that I am more familiar with. I was instead struck by the number of lists electing more than one candidate. After reading some of the literature, I had assumed these creatures would not exist. If you were expecting to see List PR, it looked a whole lot like SNTV. If you were expecting to see SNTV, it looked a bit like List PR. At any rate, I’m going to have to take another look at Colombia before I start my book on SNTV.
Seed planted by Nathan Batto — 27 March 2010 @ 15:44