The electoral system in Canada is known as first-past-the-post. The system is really simple: you finish first in a riding, you win. No need to get 50%+1 of the votes or anything, just finish first. Another system is the Instant Run-Off (IRO) where, if no candidate gets 50% of the votes, we then elimate all but the top-2 candidates and we run them against each other. This is the "instant" version as opposed to what they do in France for instance where they actually re-vote, having only the two remaining candidates. For the IRO, people give their first and second choices, so that we can redistribute their votes in case their first choice is eliminated.
Here I'm not really interested in the precise mechanic (Instant or not). I wanted to see (as suggested by someone on this blog) what would happen if we had such a system in Canada. This is particularly interesting since the Conservatives are pretty much alone in their part of the political spectrum. By that, I mean they are the only right-wing party and don't really have partners, as opposed to the Liberals and NDP for instance (and almost 50% of the CPC voters don't have a second choice, much more than any other party's supporter). Therefore, this simulation is like having an informal coalition All against Harper. I'm sure Harper would love that, as it fits perfectly his message.
So I took the current projections. If the party currently projected to lead in a riding has more than 50%, this party wins and that's the end of the story. If the leader is below 50%, I then redistribute the votes of all the parties, except the top-2 ones. I do the redistribution according to the latest Ekos poll providing us with second choices. Here are the results. There are 188 ridings where the redistribution would be necessary.
seats | CPC | LPC | NDP | Green | Bloc |
actual | 146 | 78 | 37 | 0 | 47 |
I-RO | 140 | 88 | 38 | 0 | 42 |
The Conservatives would lose 6 seats overall. Interestingly enough, it's very similar to the loss they would suffer in case of a pre-lection coalition between the Liberals and NDP, where this coalition leaves only one running candidate in any riding where LPC+NDP>CPC. I did such a simulation here. So what it means is that the Tories win 6 seats because the opposition is divided. It's not actually a big number and probably way less than when the Liberals could win 100+ seats in Ontario because the right was divided.
Another interesting thing is the Bloc. In Quebec, it's not so much ALL versus Harper, but federalist parties versus the Bloc. So the Bloc gets 5 seats more because the federalist parties split the votes between them. Here is the pdf. Please note than when I redistributed the votes, I didn't scale the results such that the votes sum to 100%. It doesn't change who would be eliminated or who would win at the end, it's a simple scaling and it was saving me some time.
By the way, I did the same exercise for the 2008 election and the results would have been:
CPC: 133
LPC: 90
NDP: 41
Green: 0
Bloc: 44
Of course, the limitation with such a simulation is that we need to assume that people would vote the same way they do under the new and current systems. It's possible that some voters could vote Green for instance with their first choice, knowing that their vote is not completely wasted since it will likely be redistributed. But I still think it's interesting to see these results, or at least entertaining. I might do it again later in this campaign if the seats projections change a lot.
[update: in order to have correct calculations for the Bloc, I re-did the simulation. More details are given in the next post]