Monday, September 2, 2019

What would a quick, simple “no-list” version of the Mixed Member Proportional system look like?


This year’s Canadian election is likely to produce a minority government. A new book says Trudeau was convinced by his caucus in 2015 to “leave the door open at least a crack for proportional representation” because he thought that he might be “willing to be convinced that (he was) wrong.” The Liberals will realize they made a mistake by letting Trudeau’s PMO break their promise to fix our broken voting system.

Let’s say they set up a Citizens Assembly on Electoral Reform to design a fair voting system that will make every vote count effectively in every region across Canada, leaving no voice unheard.

Suppose Citizens Assembly members say “a normal Mixed Member Proportional system, like the one recommended by the Electoral Reform Committee (ERRE), will need a new round of hearings by the Electoral Boundaries Commissions. Voters will need to understand how their two votes work (one for a local MP, one for a regional MP for top-up seats that counts as a vote for their party). Parties will have to adopt new systems to nominate regional candidates democratically. Hard to get all this done before the next election. Can we design a simple MMP model that uses the present boundaries, and the present ballot, one vote for your local MP?

Sure. No-list MMP.

It’s not my favourite model. But it makes every vote matter. It ends “strategic voting.” It’s perfectly proportional within reasonable limits, like a 5% threshold for parties to win seats in a province. It preserves the traditional link between voter and MP. And it’s practical, since the German province of Baden-Württemberg has used it for more than 60 years to elect their provincial parliament in Stuttgart.

Trade-offs:

MMP model design is about trade-offs: do you want larger ridings and close to perfect proportionality, or smaller ridings with more risk of a party getting a “winner’s bonus” in its stronghold.

My ideal MMP model would look like the regional open-list model for which the ERRE found consensus. (No, not the MPs -- they didn’t write the whole 245 pages, that was the staff seconded from the Library of Parliament who worked from the testimony of the 196 expert witnesses). That model would have 60% or 65% of MPs elected from local ridings. It needs new boundaries, though.

“No-list” MMP model

The “no-list” Stuttgart model has voters elect half of the MPs from local ridings. The other half, for top-up seats for parties under-represented by the local results, are simply the best runner-up candidates in a local region -- the defeated candidates of that party who got the highest % of the vote in that region. Ridings are twice the present size, so just pair up our present ones. In return, voters also elect regional MPs as ranked by voters in that region.

For example, take the region of Hamilton—Niagara—Waterloo—Halton, with 22 MPs. Now it will have 11 local MPs. Take Hamilton Mountain-West—Ancaster—Dundas (HMWAD). On the votes cast in 2015, HMWAD would have elected current Liberal MP Filomena Tassi. But the runner-up, current NDP MP Scott Duvall, would have been one of the three additional NDP MPs elected across the region, topping up the results so every vote counts equally.

Conservative voters in HMWAD would have helped elect Conservative MP Lisa Raitt, the runner-up next door in Milton—Burlington. Green voters in HMWAD would have helped elect Gord Miller, the Green candidate in Guelph—Wellington—Halton Hills, the top Green in the region.

Voters in this region voted 41% Liberal, 37% Conservative, 17% NDP and 3% Green. So Liberal voters elect 9 MPs, Conservative voters 8, NDP voters 4, and Green voters 1. (For this simulation I used only a 2% threshold, so as not to exclude the Greens. In a real election, once every vote counts the Greens would, even in 2015, surely have gotten over 5% almost everywhere.)

Results across Canada

On the votes cast in 2015, with this model (giving the Territories 2 MPs each) voters would have elected 341 MPs: 139 Liberals, 108 Conservatives, 70 NDP, 14 Bloc, and 10 Greens. That’s almost perfectly proportional. (When you divide Ontario into seven regions, rounding anomalies give the NDP an extra seat at the cost of the Conservatives. Making Alberta two regions gives the Liberals a seat at the cost of the Conservatives. And my Quebec four regions give the Liberals a seat at the cost of the Bloc. The Greens get their fair number.)

Voters get what they voted for. Most democratic countries don’t have one-party governments, nor one-man governments run by the PMO. Canada has seen ten coalition governments.

Now keep in mind that this is only a simulation. As Prof. Dennis Pilon says: "when you change the voting system, you also change the incentives that affect the kinds of decisions that voters might make. For instance, we know that, when every vote counts, voters won't have to worry about splitting the vote, or casting a strategic vote. Thus, we should expect that support for different parties might change."

Pros and Cons of “no-list” MMP

With my favourite regional open-list model, voters like me in Northumberland County would have an accountable local Liberal MP, and I would have voted for the regional NDP candidate I preferred in a small 9-MP region, so I would have an accountable regional NDP MP I helped elect.

With the no-list model, I am not limited to a 9-MP region, since a 20-MP region will not mean a large “bed-sheet ballot.” So I will now be able to point to more than one NDP MPs I helped elect: Dave Nickle in Peterborough—Northumberland and Mary Fowler in Oshawa—Whitby. My Green neighbour can point to Glen Hodgson in Simcoe North—Parry Sound—Muskoka. Local Liberals and Conservative will each have an MP they helped elect in Peterborough—Northumberland.

This “no-list MMP” is a quick and simple model. The majority of ridings across Canada will manage fine even doubled in size. It gives my Green neighbour an MP. No one has to worry about how parties nominate their regional list: no list.

But no-list MMP will miss some advantages. My county will no longer have an MP accountable mainly to us. And unlike normal MMP, I will not vote for a specific regional candidate and hold them accountable. Normal MMP gives voters the best of both worlds. Voters are guaranteed two things which equal better local representation:
1.         A local MP who will champion their area.
2.         An MP whose views best reflect their values, someone they helped elect in their local riding or local region.

The normal two-vote MMP model lets you vote for the local candidate you prefer without hurting your party, since your second vote determines the party outcome. About 30% of New Zealand voters split their ballots this way. The “no-list” model loses this advantage.

The normal two-vote MMP model also lets you vote for the regional candidate you prefer, so regional MPs are very attached to voters in all parts of the region. No-list MMP, however, means only voters in their local riding rank them. A good candidate who would have gotten strong support across the region will lose if she is running in a weak riding for her party. I can’t vote for a good MP from my region to be re-elected if I don’t live in her local riding.

Normal MMP with regional lists lets parties nominate “zippered” lists, alternating women and men. No-list MMP loses that advantage.  

However, no-list MMP has nice regions. The four large provinces have regions in my simulation averaging 18 MPs each, big enough to make almost every vote count. Still, no one outside Toronto will complain that Toronto voters helped elect their MP. Northern Ontario voters will see Northern votes elect Northern MPs. Francophone voters in the bilingual Ottawa—Cornwall region will see their votes stay in the region. The four medium-small provinces average 12 MPs each, not too dissimilar levels of proportionality.

The regional MPs are the “best runners-up;” the defeated candidates of the under-represented party who got the highest % of the vote in that region. The majority of the twinned local ridings will see two of their local candidates elected as either local MPs or regional MPs. This does not always happen, since the defeated candidates are ranked by local voters across the region. So in my simulation I find 34 local ridings where none of the defeated candidates are elected as regional MPs, while in another 32 ridings two defeated candidates win regional seats, and one lucky riding sees three defeated candidates elected as regional MPs. (Suburban Pierre-Boucher—Verchères—Beloeil—Chambly elects a local Liberal MP, while candidates from the NDP, Bloc and Green Parties are elected as regional MPs.) The Stuttgart model in Germany works just like this too.

How many indigenous MPs would have been elected? That would depend on whether parties who nominated an indigenous candidate in one of the present ridings in 2015 would have nominated that candidate in the larger twinned riding. Since indigenous people are under-represented in Parliament, I expect most would. I project 17 indigenous MPs, compared to the ten actually elected.

Even 50% regional MPs is not always quite enough. Both Southern Alberta and Northern Alberta would see every paired riding elect, on the 2015 votes, a local Conservative MP, and with 56% or 63% of the votes, they would also deserve a regional MP. They will need a regional party “standby” list after all, just in case they run out of local candidates. Similarly, Nova Scotia Liberal voters would elect all the local MPs and one more, as would Liberal voters in Newfoundland and Labrador. And my simulation from 2015 shows every local Liberal candidate being elected to either a local or regional seat in Toronto, and in York—Peel, in New Brunswick, and in Northern Ontario. Same for every local Conservative candidate in Saskatchewan. And the Quebec region of Montérégie—Estrie—Centre-du-Québec shows, despite having 20 MPs, such a Liberal sweep that they get an “overhang” (bonus) of one MP.

Large urban areas will manage ridings twice the present size. A few rural ridings will look difficult, notably the Pacific Coast riding of Skeena—Bulkley Valley—North Island—Powell River and the Northern Ontario riding of Nipissing—Timiskaming—Timmins—James Bay.

Here’s one more disadvantage: with only one local Liberal candidate for each pair of present ridings, 68 current Liberal MPs would find themselves squeezed out, with nothing but the very faint hope of being elected from a provincial “standby” list. So would 34 current Conservative MPs, 9 NDP, and 1 Bloc. As I said, this is a model a Citizens Assembly might design, not one current MPs would like. If MPs want a normal MMP model with smaller ridings and new boundaries hearings, they had better move fast in setting up that Citizens Assembly.


Since three provinces have uneven numbers of MPs, my simulation assumes Kenora, Labrador, and South Shore—St. Margaret's remain exceptional single-MP ridings. My seven Ontario regions are: Toronto (26 MPs including Thornhill); York—Peel (20), Central East (Kingston—Durham Region—Simcoe, 20), South Central (Hamilton—Niagara—Halton—Waterloo, 22), Southwestern (London—Windsor—Owen Sound, 14), Ottawa—Cornwall (10), Northern (9).