Ranked Ballots and Proportional Representation in Canada:
A Look at the Arguments and Options
November 26, 2019
Réal Lavergne, Ph.D.
For FAIR VOTE CANADA
Discussions of electoral reform in Canada today inescapably include some consideration of “ranked ballots” or “preferential voting.” This is intended to mean the option of continuing to elect our representatives the way we do now but using a ranked ballot instead of a single X-ballot. Internationally, this system is called the “Alternative Vote” (AV) and the way candidates would be elected is by using a runoff method called “Instant-Runoff Voting” (IRV).
Referring to AV as a “ranked ballot” or “preferential voting” system is misleading because ranked ballots can be used in both non-proportional and proportional systems. Fair Vote Canada thus recommends using the expression AV to refer to the use of ranked ballots in single-member ridings, and using the expression “ranked ballots” to refer to the type of ballot that is used, regardless of the electoral system in place.
IRV ensures that the candidate elected in each riding has majority support in some sense and is not elected based on a mere plurality of votes. Should the first round result not provide a majority of support for any single candidate, IRV eliminates candidates one at a time and transfers the votes for that candidate to each voter’s second-preference candidate. This continues until one of the candidates wins 50% of the vote.
AV is rarely used for legislative elections anywhere in the world. Indeed, the only industrial country that uses it for parliamentary elections in Australia. There is therefore not much comparative experience from which to draw lessons on how well it works.
When the Special Parliamentary Committee on Electoral Reform (ERRE) held its hearings in 2016, there was conspicuously little support for AV from experts. As a result, AV did not figure anywhere in the recommendations of the ERRE. However, Prime Minister Trudeau has declared AV to be his favoured model of electoral reform, and there is support for it in Liberal Party ranks.
Fair Vote Canada is of the view that if a national citizens’ assembly is organized on electoral reform, we should let that assembly assess for itself whether first-past-the-post, AV or some form of proportional representation (PR) is what they would recommend for Canada at this stage in its democratic history.
In Canada today, voters elect one representative per single-member district. This places our first-past-the-post system in the winner-take-all or single-winner family of electoral systems. The composition of government depends on how many ridings are won or lost across the country or province, and different parties may be more or less well represented compared to their share of the popular vote.
AV also elects only one winner per riding and is also a winner-take-all system. All such systems are “majoritarian” in the sense that they are likely to produce majority governments thanks to the built-in winner’s premium: the built-in bias that gives the winning party considerable more seats than its share of the popular vote would warrant.
The alternative approach is to organize elections to ensure proportionality of representation (PR). This requires the election of candidates in multi-member districts or some sort of compensatory top-up mechanism. Among the world’s 35 most robust democracies, only six (20%) use winner-take-all systems to elect their representatives. Twenty-five of them (71%) use PR and four use a parallel system falling somewhere in between.
The fundamental difference between winner-take-all and PR is the bias of the former in favour of major parties and against smaller parties. First-past-the-post produces distortions of the sort we saw in the 2019 federal election in Canada. In that election, the NDP and the Greens together took 22.4% of the votes but won only 8.0% of the seats. The corresponding winners’ bonus went primarily to the Liberals, who took 46.4% of the seats with 33.1% of the vote, and the Conservatives and the Bloc also enjoyed modest winners’ premiums in this case. Oddly the Conservative Party won more votes than the Liberals, but won 36 fewer seats, because Conservative votes were less efficient at converting votes into seats.
Distortions of the same sort manifest themselves with a vengeance at the regional level. It is such distortions that led to the Conservatives party sweep in Alberta and Saskatchewan and the near-complete Liberal sweep in the Atlantic provinces and major cities such as Toronto, Ottawa and Vancouver, making Canada appear far more divided than it is.
More often than not, winner-take-all models also produce false majorities: governments formed by a party with a majority of the seats based on a minority share of the vote. Quebec currently has a majority government based on 37% of the popular vote; Ontario’s current majority government is based on 41% of the popular vote.
The other fundamentally disturbing feature of all winner-take-all systems is that not all votes count to help elect anyone. To fully understand how this works, one has to imagine a system that does make all votes count: a proportional system. With PR, every possible vote is tallied and assigned to help elect a representative that matches each voter’s political preferences. That is not what happens in winner-take-all systems, where the only votes that matter are those that helped to push one candidate over the top in each riding.
By way of example, Liberal voters in Alberta and Saskatchewan did not elect anyone at all. Under a proportional system, they would have elected representatives in numbers corresponding to their share of the vote in those provinces (5 in Alberta; 2 in Saskatchewan). That’s what it means to say their votes would have counted.
Under an AV system, we are guaranteed that at least 50% of voters would help elect a representative they prefer compared to some others, but remaining voters are in the same situation as they are in other winner-take-all systems. There is no guarantee that the Liberals would have won any seats in Alberta and Saskatchewan under an AV system, and no guarantee that Conservatives would have done any better in the Atlantic provinces, Toronto, Ottawa and Vancouver. Based on experience in Australia and simulation work here in Canada, it appears that AV would give a different result than first-past-the-post in only 5-10% of the seats.
It is difficult to argue that any system that produces distortions of this scale can be considered a truly representative democracy or that voters can easily hold their politicians to account when vast swaths of voters are disempowered at the ballot box the way they are.
Winner-take-all systems also underperform PR countries on basically any criterion of governance that one could imagine, including the ability of parties to work together more collaboratively once elected. Readers should refer to Fair Vote Canada’s “A Look at the Evidence” for a review of this literature.
Voters and politicians who are disturbed by vote-splitting and the need for citizens to vote strategically rather than with their heart have often looked to the use of preferential ballots as a solution. Preferential ballots would allow citizens to vote for the candidate they prefer, knowing that in a tight race one’s vote could count for either one’s first or second choice. AV would also ensure that no-one is elected with less than a majority of the final count in each riding. On the surface, this type of mechanism seems quite appealing.
The ranked ballot, combined with IRV counting, is likely to be a very good approach for elections involving a single position: election of party leaders or mayoral elections for example. Such mechanisms help to avoid electing a candidate who is not the preferred choice of the majority due to vote splitting by that candidate’s opponents.
However, for legislative elections, AV maintains the winners’ bonus that is typical of all winner-take-all systems and introduces a particular bias in favour of centrist parties. At the federal level in Canada today, that means the Liberal Party.
Simulations of Canadian federal elections based on existing shares of the vote and polls of second preferences systematically show the Liberals gaining more seats under AV than under FPTP. Based on the voting pattern and second preferences from polls in 2019, simulations show that Liberals would have won a majority government with 55% of the seats under AV compared to 46% of the seats under first-past-the-post. The NDP would have taken three more seats.
The big losers would have been the Conservatives who would have lost 30 of the 121 seats they won under FPTP. Remarkably, despite winning only 33% of first preference votes to the Conservatives’ 34%, the Liberals could have ended up with 186 seats to the Conservatives’ 91. Small wonder that the motives of those advocating for AV are often viewed with suspicion as a way for the Liberal Party to gain unfair political advantage at the expense of other parties.
An important consideration is what might happen to first preference votes when strategic voters are free to vote with their hearts. However, in most cases, that would not change anything and second preferences would systematically kick in to favour the Liberals. In multiple scenarios graphed by Byron Weber-Becker relating to simulations using 2015 election results, one sees that the Liberal Party comes out the winner under pretty much every conceivable voting scenario (the thin red line corresponding to the Liberal share of seats being disproportionately higher than the thick red line corresponding to their share of the vote).
That said, one of the features of AV is its unpredictability. When the BC Liberals introduced AV in BC for the 1952 provincial election, they expected to avoid splitting the vote with the Conservatives, but the result was a Social Credit minority in 1952, followed by a Social Credit majority in 1953, which the government used to bring back first-past-the-post!
A lot depends on what might happen in the face of sudden shifts in voter preferences. We’ve recently seen a shift in PEI in which the Green Party became the official opposition under first-past-the-post. Under an AV system, they might have won a majority.
As Steven Hurdle has noted in a FVC discussion stream (Dec. 8, 2019), “we're seeing a rise of voter bravery in voting for small parties (Greens in BC, Ontario, New Brunswick, and PEI; Quebec Solidaire; People's Alliance of New Brunswick). This trend isn't restricted to Canada, it's being shown across Europe too.” This sort of shift in voter preferences away from the mainstream parties could lead to all sorts of unpredictable results under an AV system, the way it did in BC in 1952.
AV would do little to resolve the distortions of winner-take-all models in elections to legislative assemblies such as parliament, provincial legislatures or city councils. Among the other critiques that can be levied against AV are the following:
What results is a proposal for electoral reform that is appealing to a lot of people, but which appears to many as a marginal or cosmetic improvement or even a step backwards for elections to legislative assemblies. AV could further reduce the diversity of representation in legislative assemblies and might further exaggerate regional differences. It would change nothing in safe ridings and parties would continue to focus on swing ridings. It would still leave vast numbers of voters and communities with no representation in the governing caucus.
Although AV is sometimes presented as a way to get the ball of electoral reform rolling, AV could well have the opposite effect were it to further lock us into a two-party system and make reforms oriented towards proportionality harder to achieve than ever. There are no known examples of AV paving the route to PR.
Although AV does little to solve the problems of first-past-the-post, ranked ballots could very well have a place in any proportional system that might be put forward in Canada or the provinces. As noted above, ranked ballots allow voters to express their preferences more fully, and help address problems of vote-spilling and strategic voting, while encouraging candidates to be more civil when running for office.
Discussions of PR in Canada have thus focused around regionally-based models of two types:
STV uses ranked ballots (preferential voting) just like AV, but does so in multi-member districts. This makes it possible for candidates with different political outlooks to be elected in each multi-member district. The larger the district, the more voters will succeed in electing their first-choice or second-choice candidate to office.
Under MMP, local ridings continue to exist as they do now, but in smaller numbers, to make room for compensatory seats to be used as top-ups and ensure proportionality on a regional basis. Here too, it would be possible to use ranked ballots. There is no necessary conflict between PR and preferential voting. [a][b][c]
In summary, there is a strong political case to be made that PR is a more democratic, fairer way to elect our representatives than winner-take-all models. Preferential voting using AV offers some advantages over first-past-the post but also some disadvantages for elections to legislative assemblies. It would achieve little by itself and could do more harm than good. There are reasons to fear that AV implemented in the absence of PR could derail future moves towards PR. As happened in BC in 1952, AV could yield totally unpredictable results.
However, there are many ways in which PR systems can accommodate preferential voting. Preferential voting in combination with PR would satisfy voters’ desire to express themselves more fully while ensuring a representative voice for all voters.
[a], but simulations have shown that IRV would increase the Liberal bonus to the extent that the Gallagher Index would be higher than the target of 5.
[b]_Marked as resolved_
[c]_Re-opened_
Isn't this worth adding?