The NDP surge in the polls seems to have come to an end. At the very least the the NDP-PC gap has been shrinking over the last few days. At this point, unless there is a very late surge, I think it's fair to say a clear NDP win in the popular vote is unlikely (but not impossible). By clear win, I mean going over 40% and beating the PC by 5 points or more.
This means that projections models like mine are showing the Tories ahead and favourite since the popular vote is almost tied. Vote efficiency has always assumed to be helping the PC. What I'm a little bit surprised by is the magnitude of the inefficiency of the NDP vote. Earlier this campaign I had found that the NDP probably needed to win by 2 points in order to win the most seat. But as the Liberals have slowly but surely fallen to 20% and below, the same analysis now shows the NDP would need to win by almost 5 points! That's what my model shows and I obviously trust it. But I can't help but feel I'm doing something wrong.
1. The inefficiency of the NDP vote
So, why would the NDP vote be so inefficient? Let's look at the projections by regions.
Two regions offset each other: Central and Northern Ontario. The latter is going PC and all 11 ridings are projected to elect a PC MPP. On the other hand the North is currently projected to go all NDP. So that's 11 vs 11 (my "North" is slightly smaller than the definition of others, not going to change it now). With that said, the PC has a chance in some northern ridings such as Sault Ste. Marie or Kenora-Rainy River. In Central Ontario the NDP would really only have a shot in York-Simcoe.
The Southwest and Midwestern Ontario (Kitchener, etc) are two fairly competitive regions. The NDP is projected to win 7 out of 11 in the SW (London and Windsor help a lot). In the other region, the PC is winning 7 out of 11 (and it's not sure the NDP can really prevent the Green from taking Guelph). So again, it mostly cancels out.
In the Hamilton-Niagara region, the NDP is edging the PC 7 to 2, nothing surprising.
Alright, so far the NDP is actually ahead. It leaves us Toronto, the GTA and the East.
In the East, the PC wins 6 out of 7 (only Kingston and the Islands is going NDP). In most cases the margins of victory is comfortable (10 points and more). Even if all the OLP votes were to go NDP, the PC would still win those seats! So let's say the East compensates for the Hamilton-Niagara region.
In the Ottawa region, it's currently 7 PC, 1 OLP and 1 NDP with a couple of clsoe races in Ottawa itself.
Next Toronto. It's interesting that for once we have an election where the core urban center of the province is competitive, including for the right-wing party. Urban centers have become electorally boring in the 21st century with the Conservative party being unable to even compete (remember how hard Harper had to fight to finally win seats in Toronto and one in Vancouver proper in 2011?).
Polls show the NDP well ahead in Toronto, at the expense of the Liberals. if there is one place where strategic voting to avoid Ford has happened, this is it. Right now the projections have 1 seat going Liberals, 11 going PC and 13 going NDP. This is by far the most competitive region with multiple 3-way races. The 3 Don Valley ridings are pretty much toss-ups at this point (projections have the PC winning all 3 but by very small margins. Riding polls were even less favourable to the NDP's chances there, but I already talked about how riding polls seemed to underestimate the New Democrats). All the Etobicoke ridings are leaning PC but it's close in at least 2 of them. Scarborough is a giant mess currently leaning NDP. Only the core downtown is probably less uncertain (and even there, it depends if the Liberals hold better than expected).
The city of Toronto is clearly the one region where anything can happen. I expect the projections to potentially be right overall (total number of seats) but having a lot of cancelling mistakes.
Alright, so we are left to the culprit of the inefficiency of the NDP vote: the 905. I'm talking of York-Durham and Peel Halton. So ridings like Ajax, Markham etc in the east and Mississauga and Brampton in the west. The PC of Doug Ford is currently projected to win 22 seats there versus only 7 for the NDP! This, right there, is why the NDP isn't favourite.
Could my model be wrong? Well my model is already nicer to the NDP there than the riding polls. These show the PC crushing it in Mississauga for instance and even winning two of the Brampton ridings (and the other two are won by razor thin margins by the NDP!).
Polls and projections agree: the PC isn't crushing the 905 in terms of voting percentages. The Tories are winning by around 5-7 points. But it's winning most of the seats by far.
Can this be changed? Even if I somehow boost the NDP and decrease the Tories by a few points, I still have the PC winning 21 seats! This is because out of the 22 projected wins, the PC has 12 by a margin of 15 points and more (over the NDP, I'm leaving the Liberals behind here). On the other hand the NDP has 4 close wins out of the 7 it wins!
In other words: the GTA isn't about to break NDP anytime soon, at least not based on what we know. And this is giving the PC an advantage that is almost insurmountable. At this point the only way I could see the NDP having a path to victory is if they breakout in the city of Toronto (helped by a collapse of the Liberal vote).
2. The impact of the Liberal collapse.
When the OLP dropped from 25% to 20% (in average), it helped the PC a lot more than the NDP. This is particularly true in urban ridings like the ones in Toronto and Ottawa. Basically it just created more vote splitting between the OLP and the NDP. If the Liberals want to help the NDP (obviously the party itself doesn't...), then one of two things need to happen. Either the Liberals voters need to move away from the OLP and massively go to the NDP. Or they actually need to show up and vote for the OLP. From the point of the view of the NDP, every seat taken away from Ford is good news, even if it means the OLP getting more than 3 seats. Either it causes the PC to finish second overall or it at least forces a minority.
3. How can the NDP win?
There are mostly two paths to victory.
1) "Brute force" scenario where the NDP just wins the popular vote by 5+ points. It doesn't flip most of the 905 but it's enough to pick up seats elsewhere.
2) The surge in core urban areas scenario. In this one, the NDP needs to change the electoral map significantly. It would need to crush it in Toronto, possibly helped by a collapse of the Liberal vote. Out of 11 seats projected to go PC, 5 are by less than a 5 points margin and another 3 is by less than 10 points. Then you need the same in Ottawa. In this scenario, the NDP could go and grab an extra 5-10 seats from the PC. That brings the total to 60-60 pretty much. In this scenario, the NDP also needs luck in the sense that it better wins all the close ridings in the North, Guelph, etc.
So, that's pretty much it. As it stands there is no question the Progressive Conservatives of Ford are favourite. Scenario 1 here above is becoming less and less likely and would require a pretty important polling error. How likely is scenario 2? I can't really tell, or at least I can't quantify it.
