Sunday, February 4, 2018

“Local” STV (“Local PR”)

I do not generally comment on electoral reform models for Ontario other than the Mixed Member Proportional system. Some people falsely call it the NDP model, but it has been used in Germany since 1949 (after being invented in the British Zone with the help of British political scientists). It was adopted by New Zealand in 1993, and Scotland and Wales in 1998. It was recommended for Canada by the Law Commission of Canada in 2004, just as the NDP also decided to propose it and for similar reasons.     

However, a group in Guelph has designed an STV model for Ontario with an unusual twist, that they call “Local PR” (also described here) although I would call it “Local STV.” STV is the system used in Ireland.

So here is a simulation on how it would have worked on the votes cast in the 2014 Ontario election.

(Feb. 11 note: I am about to update this blog post.)

This model groups ridings into regions of 4-7 ridings, like any STV model. But the present ridings continue as nomination districts: each party can nominate one candidate in each former riding. And the counting rules will prevent more than one candidate nominated in a nomination district (former riding) from being elected, so each nomination district will find one of its candidates elected. That’s the unique feature of this model.

All of the ballots in a region contain the same candidates, organized by riding (the columns) and party (the rows).


I have divided Ontario into 19 regions which, on the 2014 total of 107 MPPs, each have an average of 5.63 MPPs. Although the rules call for regions of 4-7 ridings, Northwestern Ontario had only 3 ridings in 2014 although they will have four ridings in the 2018 election, so I have made it a separate region.

Province-wide outcome:

A fully proportional system would, on the votes cast in 2014, have elected 42 Liberal MPPs, 34 PCs, 26 New Democrats, and 5 Greens, rather than the actual outcome of 58 Liberal, 28 PCs and 21 New Democrats.

Because of the small regions, “Local PR” would have elected no Greens. I estimate the result would have been 47 Liberal MPPs, 33 PCs, and 27 New Democrats.

Regional MPPs or local MPPs?

In almost all of the 19 regions, voters for all three parties would have elected a representative. (No New Democrat in four-MPP Ottawa East-Cornwall, no PC in three-MPP Northwest Ontario.) Every vote counts, almost.

However, are these MPPs “regional MPPs?” Or do they claim to represent only the nomination district (riding) under whose name they were listed on the ballot?

Really, they are both at once, so they can claim to be local MPPs when it suits them, or regional MPPs when it suits them. They will need to hope “their” riding is not jealous of the time they spend across other ridings. Will they have offices in each riding? If that MPP is the only one representing his or her party in that region, they’re going to need offices across the region.

Local results and “wrong-winners”

Take the nine New Democrats who were defeated in 2014 but would have been elected in place of Liberals. In Toronto, Rosario Marchese would have been re-elected in Trinity—Spadina despite getting fewer votes in his riding than Liberal Han Dong. Critics of this model will say this is a “wrong-winner” outcome.

On the votes cast in 2014, Tom Rakocevic would have been elected in York West in place of Mario Sergio. Neethan Shan in Scarborough—Rouge River in place of Bas Balkissoon. In York Region, New Democrat Miles Krauter in Oak Ridges-Markham in place of Minister of Community and Social Services Helena Jaczek. In Brampton--Springdale, Gurpreet Dhillon in place of Status of Women Minister Harinder Malhi. In Halton, Nik Spohr in place of Minister of Education Indira Naidoo-Harris. In Ottawa Centre, Jennifer McKenzie in place of Attorney General Yasir Naqvi. In Kingston, Mary Rita Holland in place of Sophie Kiwala. In Barrie, David Bradbury in place of Ann Hoggarth.

Those Liberal star losses would not happen so much in MMP, where a strong candidate running in another party’s stronghold can be elected to a regional top-up seat.

Three current New Democrat MPPs would have lost instead of being elected: two in Northeast Ontario and one in Hamilton.

Counting the ballots:

As usual with STV, any candidate who has reached quota is elected. I will use the example of a six-MPP region, where “quota” is 14.29% of the votes cast in the region.

On the votes as cast in 2014, no MPP – not even France GĂ©linas, the NDP MPP for Nickel Belt who got 62.7% of the vote, the highest in Ontario – would reach quota on the first count. In a real election, no doubt Kathleen Wynne, for example, when running in the six-riding North York region with enough Liberal voters to elect 3.7 MPPs, would reach quota on the first count once every voter in the region could vote for her. 

(The original counting system proposed in "Local PR" is even more complicated than described below, which I will update shortly.) 

If a candidate is elected, no other candidate nominated in that riding can be elected, so those candidates are eliminated. Next, the elected candidate will often have more votes than the quota. His or her surplus votes are transferred to that voter’s next choice not yet eliminated (so that will be a candidate from a different nomination district). These transfers might put another candidate over quota, and again the surplus will be transferred. Finally, ballots for the candidates eliminated because a candidate from their nomination district has been elected will be transferred to that voter’s next choice not yet eliminated.

Once all candidates who have reached quota have been elected, the candidate in the region who has the fewest votes is eliminated, and those ballots are transferred to that voter’s next choice not yet eliminated. However, if that candidate is the only one nominated in that riding who remains in the count, that candidate will be elected even without having reached the quota for election. This would happen quite often on the votes cast in 2014. This elimination process continues until six MPPs have been elected from that region.

One important point about STV is that, on the final count, there will be seven candidates, six elected and the “final count loser.”

Note on Simulation:

For the purpose of this simulation, I used the over-simplified assumption that voters are 100% loyal to their party, so their ballots will transfer to another candidate of the same party. If all candidates from that party have either been elected or eliminated, I have no data on second choice parties in 2014, so I have used 2015 federal data.