Thursday, March 15, 2012

How to get to the fair voting system: consultation

(Updated Nov. 3, 2014)

Thomas Mulcair, like all NDP leadership candidates, said he will move forward with party policy: “New Democrats believe in:
a. Reforming Canada’s electoral system through mixed member proportional representation
b. Ensuring electoral reform is based on a transparent process with wide citizen involvement.”

The NDP motion in the House of Commons on March 3, 2011, proposed a Special Committee engage with Canadians, and make recommendations, on how best to achieve a House of Commons that more accurately reflects the votes of Canadians by combining direct election by electoral district and proportional representation, and report its recommendations in no more than one year.

That's called a Mixed Member Proportional system (MMP), or in French, le mode de scrutin mixte compensatoire (mixed compensatory voting system). The basic idea of MMP is simple and practical: We still elect local MPs. Voters unrepresented by the local results top them up by electing regional MPs. The total MPs match the vote share.

All MPs, both local and regional, will be accountable. Instead of having only a local MP -- whom you quite likely didn’t vote for -- you can also go to one of your diverse regional MPs, all of whom had to face the voters. An exciting prospect: voters have new power to elect who they like. Cooperation will have a higher value than vitriolic rhetoric. Governments will have to listen to MPs, and MPs will have to really listen to the people. MPs can begin to act as the public servants they are. All party caucuses will be more diverse. No party rolls the dice and wins an artificial majority. Every vote counts equally, whether it's Saskatchewan NDP voters, most Quebec Conservative voters, or Alberta Liberal voters, all of whom were robbed of their voice in 2011.

Design options

How will this Mixed Member Proportional (MMP) system work? What design options will the public be consulted on?

There are lots of working models, such as Germany, Scotland, New Zealand, and different variations in different German provinces. They have already been reviewed in detail by the Law Commission of Canada, after holding 15 public consultations across Canada. The Commission’s goal was to balance the benefits of introducing some element of proportionality into the existing system with the capacity to maintain accountable government, most notably as a direct link between elected politicians and their constituents.

But that was in 2004. Designing today’s model needs new consultations.

The model designed by the Ontario Citizens Assembly in 2007 was, in hindsight, not the best with its province-wide closed lists. As the Jenkins Commission in the UK said, “additional members locally anchored are, we believe, more easily assimilable into the British political culture and indeed the Parliamentary system than would be a flock of unattached birds clouding the sky and wheeling under central party directions.”

Six important decisions must be made, with consultation, about how to combine direct election by electoral district with proportional representation.

1. Will voters elect the regional “top-up” MPs by a closed list model, open list, flexible list, or no-list? Most MMP models use closed list. The German province of Bavaria uses open list; the regional candidate with the most votes gets any regional seat needed to top-up the local results to make every vote count equally. The German province of Baden-Wurttemberg uses no lists; the regional “top-up” MPs elected by voters for an under-represented party are the local candidates of that party who came closest to winning local seats (see note below). The Jenkins Commission in the UK proposed adapting Sweden’s list model, which lets voters vote either for the regional list or for one candidate on it (“flexible list”). Canada’s Law Commission also recommended flexible list. Quebec’s Director-General of Elections also discussed MMP design in Dec. 2007 after a one-year study, and said “The aim of the flexible list system is to achieve a balance between voter choice, which is usually associated with open lists, and better representation of women and minorities, usually attributed to closed lists.”

2. But even with open lists or flexible lists, how can we guarantee voters that the regional list will be nominated democratically in the region? Should we have German-style democratic nomination laws?

3. Will the total number of MPs change? Some residents of remote regions like Northern Ontario would prefer to keep all their present electoral districts and add new regional MPs. Some urban residents would prefer to keep the House of Commons at its present size (338 MPs in the 2015 election). The Law Commission recommended no change in the number of MPs except three more from the Territories. And what is the ratio of local MPs to regional top-up MPs? Germany generally uses 50/50. Scotland has 57% local MPs. New Zealand has 58% local MPs. Wales has 67% local MPs, the highest ratio found anywhere. Higher ratios risk not having enough regional MPs to compensate perfectly for disproportional local results. The Law Commission recommended almost 2/3 local MPs, perhaps 62%. With smaller "top-up" regions, the regional MPs feel more accountable, and the ratio of local MPs could be lower. If we keep 338 MPs, that might be from 213 to 203 local MPs, and 125 to 135 regional MPs. Often, three existing ridings would become two larger ridings, or five would become three, but the number from the region would not change.

4. How large are the regions? Scotland’s have 16 MPs, Wales has 12. The Jenkins Commission in the UK recommended medium-small regions of only 8 MPs. A small-region model for Quebec (three or five MNAs per region) was rejected in 2006 after its Citizens’ Committee reported the effective threshold “could be between 13 and 17%. Given that one of the objectives of reform is to ensure effective representation of the electorate in terms of equality of votes, this threshold is far from being acceptable.” For Canada’s four smallest provinces, the “region” is likely the whole province. If we use the Law Commission’s 64% local MPs, we would see nine local MPs in Saskatchewan and five provincial MPs; Manitoba would be the same. In larger provinces, regions with an average range of 14 MPs (9 local, 5 regional) would reflect experience in Scotland and Wales. In BC we might have four regions: two in the Lower Mainland (maybe both 8 local MPs, 5 regional), the Interior (6 local, 3 regional), and Vancouver Island (4 or 5 local, 2 or 3 regional). In Alberta we might also have three regions: Calgary (6 local MPs, 4 regional), Edmonton (7 local, 4 regional), and the rest of Alberta (8 local, 5 regional). In Ontario we might see nine regions: Central Toronto—Scarborough (8 local MPs, 5 regional), Northern Toronto—Etobicoke (7 local, 5 regional), Peel—Halton (8 local, 6 regional), York—Durham (9 local, 6 regional), Hamilton—Niagara—Brant (7 local, 5 regional), Southwest (Waterloo—London—Windsor) (9 local, 7 regional), Central (Barrie—Guelph—Bruce—Peterborough) (9 local, 5 regional), Eastern Ontario (Ottawa—Kingston) (10 local, 6 regional), and Northern Ontario (6 local, 3 regional). In Quebec we might see five regions with a total of 47 local MPs, 31 regional. But advocates of "moderate proportionality" would suggest regions of only six or eight MPs, such as regions averaging eight MPs. And regions of 12 MPs would work well.

5. Should a party have to reach a threshold, such as 5% of the vote in that province, to qualify for regional top-up MPs in that province? Or 4%? Both are common thresholds in Europe. A threshold of 3% is unusual, and when the Ontario Citizens’ Assembly proposed 3% it was an unpopular feature. New Zealand’s Royal Commission recommended 4%, their government changed that to 5% when the system was designed, but the latest review of their model again recommends 4%. But with regions of only 14 or eight MPs, a threshold might look unnecessary. Still, suppose 11% of the voters in Northern Ontario voted for a Northern Ontario Party, enough to elect one MP, but that would be less than 1% of the total Ontario vote. Should it have to meet a 4% threshold? Next, should a party be exempt from the vote threshold if it wins one or two local seats in a province? Germany exempts a party that wins three local seats nationally. New Zealand has a local-seat threshold of one seat, but this made some small parties captive to a larger ally who let them win a local seat, so its latest review recommends abolishing that.

6. Should voters elect local MPs as we do today, by First-Past-The-Post, or should they use a preferential ballot? No working MMP model does this. However, some supporters of proportional representation for Canada admire the Jenkins Report in the UK, which did propose this option, so it should be considered.

Other issues are pretty easy:

1. Should voters have two votes, one for local MP, one regional vote for a regional list or a regional candidate? All normal MMP models give voters two votes, making the local MP more accountable. As many as 35% of voters in New Zealand feel free to vote for a local candidate of a different party than they want in government, since only their second vote determines the party make-up of the House.

2. Can a party nominate a candidate to run for a local seat and also run regionally? If the candidate wins a local seat, their name drops from the regional list. “Dual candidacy” is permitted in almost every MMP model, and with any flexible list or open list model it gives voters more choice. (However, with closed lists a dual candidate can look immune to voter rejection.)

3. If a regional seat becomes vacant, should it be filled by the party's regional candidate who stood next at the last election? Or should there be a regional by-election which would be rather costly? No working MMP model does this. Besides, as the Jenkins Commission stated, if a region-wide contest were to take place "it would almost by definition result in the victory of the predominant party in the area, thus negating the essential purpose of the Top-up seats."

Then there are some technical issues.

1. Within larger provinces, should seats be calculated by the regional count alone, or should the province-wide count set the number of top-up MPs, which are then allocated to each region in some way? Germany does this, but does not guarantee each region will have a fixed number of seats. Sweden and Norway set aside seats (11% in Sweden, 5% in Norway) to allocate to parties who were short-changed by the regional calculations, but again some regions would get extra seats. Using the regional count has the virtues of simplicity and transparency, but with only 38% top-up MPs they might not compensate perfectly for a regional sweep.

2. With “flexible list” a regional candidate will move up the regional list if he or she gets enough personal votes, and win election ahead of another candidate whom the party’s regional nomination process had ranked higher. How many is enough? Sweden and Belgium use different calculation techniques which need study.

3. What rounding formula should be used to round fractions up and down in the proportional calculation? “Highest remainder” helps small parties, “highest average” helps large parties, and the “Ste.-Lague” calculation now used in Germany is more neutral but harder to explain. With smaller regions such as eight MPs, "highest remainder" might be essential to offset the small regions.

4.  What qualification is required for regional candidates? The Law Commission of Canada recommended a party be eligible for compensatory regional seats only if it presents local candidates in at least 33 percent of the ridings in the region. Jenkins recommended 50%. The purpose of the compensatory regional seats is to correct disproportional local results, not to provide a parallel system of getting elected. These rules prevent the Berlusconi trick of running twin parties, one party with local candidates, the other with regional candidates. But these rules also prevent regional independent candidates, which Scotland permits but no other jurisdiction does.

One way to reach these decisions is to have a discussion draft. The Law Commission recommended: “The federal government should prepare draft legislation on a mixed member proportional electoral system as proposed in this Report. After drafting the legislation, a Parliamentary committee should initiate a public consultation process on the proposed new electoral system.” Quebec prepared a draft bill in this way in 2005, but it took a long time to get through their government caucus, and it produced a poor model. Another way is to have a new expert Commission hold the consultations, perhaps jointly with a Parliamentary Committtee and a Citizens’ Committee. These details are not easy to debate in a referendum; about 70% of Canadians support proportional representation, but in the Ontario referendum the image of unrepresentative party hacks from Toronto getting in through the back door was fatal.

Note: for problems with the "no-list" method used in Baden-Wurttemberg, see Chapters 9 and 14 of Massicotte's Report. In regions that tend to give the same party massive support, that party may have no losing candidates! Since 1956, such an outcome has occurred five times, always in the region of Tübingen, which very strongly backs the CDU. So any "top-up" seats go to substitute candidates nominated before the election at the same time as the primary candidate, in the event of such an outcome.


Bill said...

3. Should voters elect local MPs as we do today, by First-Past-The-Post, or should they use a preferential ballot? No working MMP model does this, since the ballot is already complicated enough.

Many who who can't see through complications suspect cheating or corruption. Consider the Income Tax Act: many think that almost all other tax filers are cheating the system. The result follows a species of Gresham's Law: Tax Cheating drives out Tax Honesty.

Wilf Day said...

Yes, a new voting system has to be as transparent as possible. That’s why I’ve given up on province-wide proportionality being used to adjust the regional results; it’s simple and practical if each region’s votes elect each region’s MPs.

Raymond said...

Excellent information & summary, Wilf! Here is my take on your design points:

Five Important Design Decisions:

1. Closed List is the most realistic. Parties will never agree to intra-party competition and the associated costs. New Zealand will likely stay with closed lists. “Closed Lists” should be renamed “publicly regulated” lists (according to publicly accepted democratic principles). Flexible List is too complicated for electing one or two regional MPs per party. No party is proposing a citizens’ assembly to promote open lists or STV to the public.

2. The number of MPs should stay the same or be reduced very slightly. Increasing the number of MPs would be politically unpopular.

3. The ratio of top-up MPs should be enough to create realistic proportionality, consistent with the 4% - 5% thresholds. (At least 33%)

4. Regions should separate the three most urban areas (Toronto, Montreal, Vancouver) from northern or interior regions of those provinces. Also, 20- to 25-seat regions would make sense with thresholds of 4% - 5%. For large provinces this translates to several regions per province (about 6 for Ontario, 4 for Quebec, 3 for BC, and 2 for Alberta). “Regions of interest” should also be a factor, as opposed to perfect symmetry.

5. Provincial thresholds of 4% - 5%, are similar to the discussions taking place in New Zealand. A compromise might be 5% threshold in general, but a 4% threshold if a party wins a seat.

Other Issues:
1. Closed Lists require a local candidate vote & a party vote.

2. Dual candidacy with democratic nominations are the most practical.

3. First-Past- The-Post is the most practical.

4. Next on the list, is the most practical for filling regional seat vacancies.

Technical Issues:

1. Provincial calculations with regional distributions are manageable and practical with closed lists and reasonably proportional.

2. Closed lists: Publicly-regulated, according to publicly accepted democratic principles.

3. Rounding formula: Ste. Lague is probably the best for political consensus.

Wilf Day said...

Raymond, those are all arguable options. But I think the realities faced during the 2007 referendum tell us the Law Commission of Canada was right to nix closed lists. I think it is now clear that Ontario voters will not accept closed lists. This was a principal point of criticism in the 2007 referendum campaign: the “list MPPs” would not be elected and accountable since they had not faced the voters; they would be “appointed” said the critics. Opponents hammered away on the claim that the list MPPs would be beholden to parties instead of voters.

In the 2007 Ontario referendum, 63.1% voted against MMP. About 31% were simply against proportional representation. Many more were voters who wanted all MPPs to be personally elected, not on closed lists.

As the Law Commission said, “Based on the feedback received during our consultation process, many Canadian voters would also most likely desire the flexibility of open lists in a mixed member proportional system. Allowing voters to choose a candidate from the list provides voters with the ability to select a specific individual and hold them accountable for their actions should they be elected.”

In May 2008 the CA’s Chair, George Thomson, spoke to the Annual General Meeting of Fair Vote Canada. He said that, if those 103 Citizens had had another six or eight weeks to deliberate, he felt some elements might have been different, like regional lists and open lists.

As Fair Vote Canada’s Statement of Purposes says, “A democratic voting system must encourage citizens to exercise positive choice by voting for the candidate or party they prefer.”

So 20- to 25-seat regions are less likely. With open lists, the regions cannot be so large as to have the “bed-sheet ballot” found in some countries. List candidates need to be able to become known, with no unfair advantage to incumbents or “big names.” But we likely want at least five list candidates in each region, to help parties nominate and elect more women and minorities. That may mean manageable mid-sized regions like the nine the PC party uses.

Raymond said...


I agree with all your points. The difference is that the next change in the federal voting system will likely be a political decision (ie. with no citizens' assembly), possibly through a multi-party consensus.

This creates a different dynamic. Parties will have to sell any proposal to their volunteers & active party members. Open lists can be internally divisive to party unity.

The Ontario Citizens' Assembly proposal basically gave parties a free hand in nominating their list candidates. Public regulations would help change the image of "party leadership control" over the closed lists.

Wilf Day said...

Raymond, in Germany that works, and I wish we had their laws requiring democratic nominations. But I doubt Canadian voters would be satisfied.

As the Law Commission said, closed lists give “an enormous advantage to party elites, who can place themselves at the top of the lists. At the same time, however, closed lists can allow party officials to place members of under-represented groups, such as women and ethnic or religious minorities, at the top of the lists. Despite the advantages of closed lists as vehicles for electoral affirmative action, they have been unpopular with voters. . . . the goal should be to balance the voter choice commonly associated with open lists with the goal of promoting women, minority group members and Aboriginal people.”

That means flexible lists, said the Commission, where voters have the option of either endorsing the party slate on the regional portion of the ballot or of indicating a preference for one regional candidate; I agree.