The 2019 election produced an interesting result. Despite winning a plurality of the popular vote, the Conservatives ended up with just 121 seats compared to the Liberals’ 157. The NDP also had a disappointing result, finishing a couple of points below the 18 percent that polls had projected for them on the eve of the election. With just 24 seats, the party won fewer seats than it had in any election since 2004. Both the Conservatives and NDP had trouble winning seats because they did particularly poorly in competitive ridings. While the Liberals lost vote share in most ridings, they were able to avoid suffering particularly large losses in swing ridings.
In this post I look at the different changes in vote share across ridings in which parties were either safe, competitive or well behind. To determine competitiveness I estimated parties’ vote shares in ridings using on a universal swing model based on regional polling averages from the CBC poll tracker from the day before the election. I use the polls and not the actual results in order to get a sense of what parties could have expected based on the knowledge that would have been available to them prior to the election. I count any riding in which a party is within 10 points of either winning or losing as competitive. Any riding in which the party is up by 10 is counted as a safe riding and any riding in which the party was predicted to be down by 10 was considered to be a lost cause riding. While my model was not perfect in predicting which ridings were safe for parties, it did pretty well. The Conservatives won all but one of their 84 predicted safe ridings (99%), the Liberals won every one of their 89 predicted safe ridings, the NDP won 18 of their 21 safe ridings (86%), and the Bloc Quebecois won 16 of their 18 predicted safe ridings (89%).
The graph below shows the change in vote share since 2019 for different types of ridings for each party, with the dots showing the average and the lines showing the ranges for a 95% confidence level. It is striking that most of the Conservatives’ gains came in ridings that were safe for them, while the party made only marginal gains in competitive seats. The party averaged vote shares almost 7 percentage points above their 2015 result in safe seats but only less than a point above their 2015 result in competitive seats. This meant that the Conservatives could only win competitive seats where Liberal or NDP losses were sufficiently large to make up the difference between the Conservatives and their strongest competitor in 2015. Almost all of the Conservative growth since 2015 came in ridings that were safe, and therefore had only a limited impact on the number of seats that the party won. Because both the Liberals and NDP averaged losses in ridings that were competitive for them, the Conservatives were able to pick up competitive seats without substantially growing their vote share. They would have been able to make more gains, however, if they had managed to increase their vote share in competitive ridings.
The distribution of gains and losses is even worse for the NDP. The party averaged losses in safe, competitive, and lost cause ridings, but its worst losses came in competitive ridings. In such ridings, the NDP average vote share was over 7 points below the party’s 2015 results. Compared to this, the party managed to do relatively well in safe seats, losing less than a couple of percentage points. The NDP’s failure to do well in competitive ridings was particularly notable in Ottawa and Toronto. The party lost Ottawa Centre, Parkdale High Park, and Toronto Danforth by over 14 percentage points even though these were all ridings that the party had done well in prior to 2015 and which the party was likely targeting.
The Liberals, meanwhile, managed to do slightly better in competitive ridings than in safe or lost cause ridings. The party managed to keep its losses in competitive ridings to just over 5 percentage points compared to almost 7 percentage points in safe ridings and almost 8 in lost cause ridings. As a result, the Liberal vote in 2019 was more efficient with respect to turning votes into seats than the vote for any other party. Had the Liberals not been able to limit their losses in competitive ridings, they may not have been able to hold on to a minority government despite not winning a plurality of the popular vote.
It is finally worth noting that the Bloc Quebecois managed large gains across Quebec, but also that those gains were strongest in safe and competitive ridings for the party. It is not surprising that the Bloc Quebecois made significantly smaller gains in lost cause ridings. These ridings are disproportionately Anglophone and so would be less likely to be affected by the surge in Bloc Quebecois support that occurred over the course of the campaign.
Both the Conservatives and the NDP can take lessons from the way that their vote was distributed in this past election. For the Conservatives the lesson must be that they need to do more to reach out to voters beyond their base. This is particularly true in the GTA, where the Conservatives failed to get close to winning many of the ridings that Harper picked up when he won a majority in 2011. Playing up concerns over Western alienation is only likely to make their vote more inefficient and increase the likelihood that the Liberals will be able to win a plurality of seats without a plurality of the vote. The worst thing that the Conservatives could do, at least in so far as their future electoral success is concerned, is to spend the coming minority parliament focused on issues that are most important in the Interior BC, Alberta, and Saskatchewan ridings that the party already wins by large margins. It may be more fruitful for the party to focus on issues relating to affordability, particularly with respect to housing, then to attack the Liberals over their handling of the Trans Mountain pipeline or their decision to implement a carbon tax.
The NDP, meanwhile, needs to think carefully about what it needs to do to win the urban ridings that it is competition with the Liberals for. It is notable that the Liberals won 17 of the 18 ridings that my model was showing as Liberal vs. NDP races on the eve of the election, and that the other one (Kenora) was won by the Conservatives. In addition to the aforementioned Ottawa and Toronto ridings, the party failed to compete with the Liberals in Halifax, lost the Burnaby North Seymour riding they had been hoping to pick up, and failed to make inroads in either the Brampton area ridings that Singh had represented provincially or the Newton and Surrey ridings that the party won in 2011. The NDP’s struggles may be a result of strategic voting, or they may reflect the Liberal party’s ability to appeal to urban social progressives. The NDP will have a difficult task ahead of them in this minority parliament. They will have to exert enough influence over the government that urban voters start to see the party as one that can influence policy (and thus feel less of a need to strategic vote). At the same time, they will have to find ways to oppose the Liberals on issues popular with the urban social progressive voters that live in the swing ridings that they lost.
The Liberals’ ability to hold on to government after the 2019 election came about in part because of both major federal opposition parties’ inability to make gains in swing ridings. As the Conservatives and NDP move forward, they will have to think carefully about how they can move beyond their regional bases of support to appeal to the voters that live in the ridings that regularly determine the outcomes of federal elections.