Thursday, May 7, 2009

STV vs MMP (Thoughts On STV 7)

Ironically, there are some electoral reform supporters who actually campaign against BC-STV. They believe that if the STV proposal is voted down, there will be another Citizens' Assembly summoned, and this time, the assembly will choose MMP. Obviously, the chances for a new Citizens' Assembly if BC-STV fails, are slim to nil, if not sub-zero. But the question of whether or not a Mixed Member Proportional system would do better in BC than the Single Transferable Vote, is worth looking at.

Proponents of the Mixed-Member Proportional system would be quick to remind us that it was MMP, rather than STV, which was first proposed for BC by the Citizens' Assembly. But they simply didn't have the time to consider all the details, so they had to go for the second choice - which was STV. What if they did have enough time to design the MMP proposal? Let's try to put ourselves in their place; let's try to finish their original project and see what obstacles were they facing.

We know the system - Open-List MMP. We know the number of MLAs: 34 local and 34 list. [According to Wilf Day's comment, the final proposal called for 47 local and 32 list MLAs (60:40 ratio). The proposal of which I heard back then, recommended 50:50 ratio, with local constituencies matching then 34 (now 36) Federal ridings.] And here's our first catch: if we are to elect 34 (or 32 or 36) MLAs using open party lists - that means, all their names should be on the ballot, so that voters could cherry-pick their preferred candidates from the list. Can we fit that many candidates for each party on a single ballot?

To implement an open-list system with 32-36 list seats, we'd have to divide the province into 5-8 self-contained regions. Each region would then have its own lists of 5-7 candidates and 5-7 directly elected MLAs. But here's another catch: if a region only contains 10-14 seats - will that be enough to establish proportionality? And what do we do with fractional seats? How do you handle 6% of the vote in a region that elects 10 members?

STV doesn't have that problem. If there are several candidates with only half of the quota - the one with the lowest number of votes is eliminated, his votes are redistributed according to voters' subsequent choices and in the end - one of the remaining candidates will have enough votes to get elected. With MMP, there are no second choices; vote fractions must be rounded either up or down and final results vary depending on the mathematical formula you use.

But what if we try adding more local seats to the regions without increasing the number of list seats? Can't we have a self-contained region of 17 local seats and 8 list seats? 70:30 ratio between local and list seats (as opposed to 50:50) was proposed for Ontario. But here's another catch: when you cut back on list seats, it becomes likely that one of the leading parties elects so many MLAs locally, that there just won't be enough list seats to offset the distortion. How should we handle a situation when a certain party wins 12 or 13 local seats, when a regional list vote only entitles it to 10? Not to mention that having 25-seat regions would still require rounding the results up or down to the nearest 4%.

So maybe we should go with closed province-wide lists? That's what the Citizens' Assembly of Ontario ended up choosing because they too, just didn't have the time to work out all the details for the open list model. But then - Ontarians voted their proposal down. PEI government too chose to propose a closed-list system, rather than designing an open-list model for 17 local and 10 list seats. The result was the same. Because, unlike Europe, we don't have the tradition of voting for party lists.

And, let's face it - a closed party list model has its disadvantages. It does give a lot more power to the party executives who compile the lists - even if a formal nomination process actually takes place. It does reduce voters' ability to directly decide on the consists of some 30-40% of the Legislature. It does create a possibility for all sorts of ethnic and gender quotas for list candidates - even if such quotas are not included in the original proposal. The simplicity of merely selecting a party list comes at a price. So when it comes to a question of what's more important - pure proportionality or ability to directly elect every single MLA, two thirds of the voters in both Ontario and PEI chose the latter.

Compare these results to 57% "Yes" vote for BC-STV. Compared to MMP, it has several advantages. First: all the candidates run locally. Thus - no easy ride for small parties that concentrate all their efforts on a list vote, without even bothering to run candidates locally. Second - it doesn't use party lists; it's each candidate for himself. Third - it uses preferential ballot - unlike the MMP, where voters are still compelled to vote strategically when electing a local candidate. And of course - STV doesn't have the problem with seat fractions, rounding and so on. Fractional seats (if any) are determined based on voters' second and third choices; votes for a party that doesn't have enough support to elect an MLA are not discarded, but transferred, according to voters' preferences.

So, no wonder that in the end, the Citizens assembly found a simple solution to all the issues I've mentioned in my article - they chose to go with the Single Transferable Vote.


Wilf Day said...

A few corrections are due here.

MMP was first proposed for BC by the Citizens' Assembly? No, they designed both as planned (STV first), but then voted for STV by 123 to 31 for MMP.

34 local and 34 list? No, 40% list was their model: 47 local, 32 list.

How many regions? They ran out of time before deciding. I'd say five. The largest would have been 22 (4 local, 8 regional), enough for a 5% party to win a seat. Likely they would have used "highest remainder" to deal with the fractions. They decided on open list.

A good MMP model, much better than Ontario's. But their mandate was to choose one model, and the consensus for STV was quite strong; it was even better than their MMP model.

Leonard said...

Thanks for the corrections. I heard about the 34/34 option, with the local constituencies matching the federal ones.

Obviously, if you have 32 list MLAs on open list, 5 self-contained regions is the minimum. But the largest reminder system would probably be denounced by larger parties, which would argue that such method disproportionally benefits smaller lists.

I agree that this could have been a better model than the one proposed for Ontario and PEI (not to mention the one proposed for New Brunswick,) but STV appears to be even better solution.

Anonymous said...

The problem of having to round a party's vote share to the next whole number of seats, but at the same time not having too large regions, could be solved liked this:

At first you count all party votes country-wide and determine the number of seats that each party will get. So the rounding problem is reduced. In a second stage you sub-distribute those seats to the regions according to the number of votes the party received in those regions. Then you allocate seats to individual people.

By the way, I would recommend open list, but not the traditional model where you have only one vote for one candidate, but a within-list STV. That means, voters can rank-order candidates from the list they chose, and the winners in each list would be determined by an STV count. So even on the candidate level barely any votes are wasted.