I want to pick up here on comments to two of my post-election analyses of Colombia.
Rici noted that what really mattered for the exclusion of paramilitary-affiliated candidates was not the change from single nontransferable vote (SNTV) to list PR, per se, but “list discipline.” I agree. The point is that the new electoral system made possible what was previously almost impossible: the exercise of control by party leaders over who could be elected–precisely what Rici means, I assume, by “list discpline.” The previous system, as with all SNTV systems, made the equivalent to list discipline nearly impossible. The threshold for election with SNTV is very low (often in practice well under 1/2M, where M is the district magnitude–and lower the more candidates there are in the running).
While parties could control who used their own labels somewhat under SNTV, if they wanted to (the major parties in Colombia tended not to do so), parties under SNTV cannot keep candidates with name recognition and resources from being elected. That is, such candidates often can get themselves elected, even if they do not have the formal endorsement of an established party.
Japanese parties, including the long-ruling LDP, knew this: often “independents” denied the party endorsement would win anyway. Under party lists, on the other hand (open or closed), parties must present lists with a finite number of candidates, and they have the legal authority to decide who may and may not bear the party label.*i.e. they can exercise “list discipline.” Moreover, given the threshold (2% in the Senate, and half a simple quota in a House district), and the d’Hondt allocation formula (which slightly favors larger parties), prospects for election without a recognized party label are reduced, compared to under SNTV (more so in the Senate than in the House).
It is true, as Rici notes, that parties had to have the organizational will to exert list dicipline–and in his comment, he makes some very interesting observations about how interparty competition within uribismo might have facilitated this. What I was highlighting is that, given the will, list PR made is feasible to exercise this discipline over who ran with the party label, in a way that SNTV did not.
Colombia’s list proliferation is just an extreme example of a standard SNTV problem: the priority the system puts on personal over party criteria in elections, and the extremely low threshold for winning a seat.
Steven Taylor addresses the latter point when he notes the rise in effective number of parties over time in Colombia–from under three (in seats) in the early 1990s to 3.5 in 1998 and over nine in 2002–and says that this shows there was more to list proliferation than SNTV. Of course, that is true. There was a tremendous proliferation of labels in 2002. But this demonstrates precisely the problem: With such a low threshold, candidates with local name recognition and resources effectively could get themselves elected, without any endorsement from a party. What changed leading up to 2002 was that fewer of these candidates chose to run under Liberal or Conservative labels, and instead adopted their own labels. The reason from that is outside the electoral system–principally, the disintegration of the Liberal party once Uribe left it and decided to run as an independent rather than seek his old party’s primary nomination.
I like the term “personal list” to describe the old Colombian system. In fact, I like it so much that, as far as I know, I invented it! However, if no list obtained enough votes to have both a simple quota (1/M) and be in the running for one of the largest remainders, the effect is identical to pure SNTV: the winners are the candidates heading the M lists with the M highest vote totals. The tendency for virtually every list to elect only one candidate–and therefore not to be a list in any meaningful electoral sense–was not new in 2002. It was very much the case in 1998 and 1994, too. The difference was that more lists in those years bore one of the main party labels, whereas in 2002, more bore new labels.
On candidate proliferation and narrow margins in SNTV more generally, aside from the large literature on Japan, see my post-election analysis of Afghanistan, which shared the feature of 2002 Colombia: weak (or virtually non-existant) parties, making it almost a pure personal electoral system.
*Brazil used to be a partial exception to this; there was a provision called “birthright candidate” by which a politician, once elected, was entitled to be renominated. This is no longer the case, and even when it was in effect, parties that wanted to exercise list discipline (the PT, for example) were able to do so. They key is that most parties chose not to. The threshold for election is very low in Brazil, so many of these candidates could have been elected on another party’s list even if their previous party declined to renominate them. In this regard, Brazil’s open-list PR was more like SNTV in its operation than other list-PR systems, including Colombia’s new system (or Brazil’s now).