Today I’m going to take some time to write about the electoral system in Canada. We have an election coming up at some point this year, most likely in October, and one of the issues that at least two parties are supporting is proportional representation. Both the New Democratic Party and the Green Party support electoral reform, with the NDP favouring what is called Mixed Member Proportional or MMP.
In later posts I’ll go into detail about MMP, but for now it will suffice to show what I think the problems are with the existing system. Our system as it stands is called First Past the Post or FPP. Our country is divided into 308 (which will change to 338 this year), with each of these ridings being represented by a single Member of Parliament (MP). In a general election eligible voters cast a ballot in the riding in which they live for the candidate of their choice. At the end of the night the candidate with the most votes wins, regardless of the margin or of the actual proportion of the vote they received.
In a country with only two parties, such as the United States, this will usually lead to the ‘winner’ in the riding with at least half of the vote. The problems only become pronounced when a country has more than two parties, with the problems becoming more pronounced the more parties we have. In Canada we have three main, national parties, each with support greater than 20% of the electorate, but less than 40%. The Conservative Party is right of centre, the Liberal party is roughly centrist, and the NDP is slightly left of centre. In addition to the main three, we also have one main regional party, the Bloc Quebecois which only operates in Québec, and the Green party, which has support of around 5% of the electorate. Both parties are roughly centrist, with the Bloc a bit further left than the Liberals.
To illustrate the main issue with the FPP system I’ll take the example of a recent poll released by OKOS Research. In the province of Ontario they have the Conservatives at 37%, the Liberals at 35%, the NDP at 18%, and the Greens at 6%. If we were to treat the province as a single riding with one representative there would be no clear winner, but the Conservative candidate would still ‘win’ because her or she has a plurality of the vote.
This gets even more pronounced with Québec where, in the same poll, the Conservatives have 22%, the Liberals have 23%, the NDP have 24%, the Bloc have 23%, and the Greens have 6%. If we treated this province as a single riding the NDP candidate would ‘win’ with less than a quarter of the vote.
Of course, nothing I have written here is novel, especially to people who follow politics on a regular basis. There is an organization called Fair Vote Canada which has as its goal to reform the electoral system and replace it with MMP. A couple years ago they released the following video where they illustrate FPP with pizza (in their example two people vote for meat lovers, with the remaining four people splitting their vote):
What I would like to do for the rest of this post is build on their example because I think that it falls short in demonstrating just how unintuitive FPP can be. When FPP goes from a single riding and is expanded to the results of multiple ridings the results can look very different from what the popular vote was. Further, this isn’t a mere theoretical possibility, the unintuitive results are actually projected.
At threehundredeight.com they compile and combine the polling data from different sources and use their expertise to project a seat count. They project that 33.4% of people would vote Liberal, 32.6% Conservative, 20.8% NDP, 5.9% Green, and 4.8% for the Bloc. From those figures they project that the Liberals would get 124 seats, the Conservatives would get 142, the NDP would get 66, two for the Greens, and four for the Bloc. This translates into the following proportion of seats for the following parties, 32% for the Liberals, 37% for the Conservatives, 17% for the NDP, 0.5% for the Greens, and 1% for the Bloc (mind the rounding).
What is stunning about the forecasted results is that it would give the Conservatives more seats than any other party even though they are forecasted to come in second place in terms of the popular vote.
To illustrate just how wrong this could turn out, let’s expand on the pizza example from Fair Vote Canada. Let’s imagine that we are students in a school with eleven classes. Each class has thirty students, giving us 330 students. In order to decide what types of pizza to order for an upcoming school-wide pizza party the principal has asked each class to elect one representative from their class to come to a meeting. At this meeting the students must vote on what type of pizza will be ordered.
Now, to make this manageable let’s assume that there are only three kinds of pizza which they can vote for. These are: vegetarian, meat lovers, and cheese (represented by V, M, and C).
Here is one way in which the vote could turn out. In classes 1-6 V receives 9 votes, M receives 10, and C receives 11 (adding up to a total of 54 for V, 60 for M, and 66 for C). If we were using FPP the voters for cheese would elect six representatives (out of 11). As for the remaining five classes, let’s assume that there is unanimous support for V, giving a total of 150 votes for V (but only five representatives.
Our tally is now at 204 for V, 60 for M, and 66 for C. But since the students don’t vote for the type of pizza directly, and since the principal only asks for a majority vote in this meeting to decide which pizza to order, the cheese party will win even though they represent only 20% of the students, whereas the vegetarians, with 62% of the votes get effectively no say.
The chief issue with FPP is not the first step where a plurality of votes is sufficient to ‘win’ a vote, although this is a problem as well, it is the second level where a majority of representatives have absolute control over the agenda of parliament, even though there is no guarantee that, at this second level, they even have a plurality of the original vote. Under FPP obscure results like the one I have described become more likely as the number of parties increases. It also accentuates regional differences and under-represents parties with even support over the entire country.