2012 Québec Election: Just the Numbers First
The Québec Liberals were tossed out of office last Tuesday, but in a surprisingly gentle manner. In fact, I think Québec voters made a wise decision on September 4th: they changed the governing party but put the PQ on a very short leach; they got rid of the negative figurehead Premier Jean Charest had become but didn’t convict his Liberals before knowing more from the Charbonneau Inquiry on corruption in the construction industry and collusion with said industry and political parties, and they told the new up-start CAQ that it needs some fine tuning before it can aspire for prime time. In short, instead of swinging wildly as they’re often known for, they wrote a sensible prescription for Québec:
- one (1) weak minority government for 18 to 24 months;
- one (1) party rendered leaderless to ensure an 18- to 24-month government since said party won’t want an election without a leader;
- one (1) party given a lesson in humility and left with little cash to jump too quickly into another election, thus also ensuring an 18- to 24-month governement, and
- one (1) party given a bit more representation in the National Assembly [a] to hear more about its ideas and [b] to make us feel good for voting for a progressive party.
Many polls prior to the election predicted that the Liberals would find themselves as the second opposition party in Québec City, with virtually no support among francophones (translation: outside Montréal). However, they did end up forming the official opposition, and not only that, a very strong one at that (alternate interpretation: The Parti Québécois has formed the weakest minority government in Québec history). Perhaps the high percentage of undecided voters that persisted until the last opinion poll at the eve of the election was “hiding” closet Liberals who didn’t want to openly say that they were leaning toward the party at the helm of a highly unpopular government.
Here’s a table of the raw data, which shows just how close the popular vote was. I drew these results from Le Directeur général des élections du Québec website, which I carefully entered into my personal database of election results.
Let me summarize the most striking points from this table. First, do consider just how close the results were among the top-three parties.
Closeness of the Popular Vote
- # of votes between the winning PQ and 2nd-place Liberals: 31,922
- % difference between the winning PQ and 2nd-place Liberals: 0.73
- # of votes between 2nd-place Liberals and 3rd-place CAQ: 180,860
- % difference between 2nd-place Liberals and 3rd-place CAQ: 4.15
- # of votes between 1st-place PQ and 3rd-place CAQ: 212,782
- % difference between 1st-place and 3rd-place CAQ: 4.88
In other words, 96.21% of the votes were shared by the four parties attaining more than 5% of the overall popular vote, while 90.18% of the vote was split among the top three parties.
These results, at least correctly on the surface, have led some analysts to claim that Québec has almost evenly split into three camps. However, as expected, many are also looking at how the seat count doesn’t reflect that reality:
- 0.73% more votes gave the PQ 4 more seats than the Liberals, thus the government;
- 4.15% fewer votes gave the CAQ 31 fewer seats than the Liberals, and
- 4.88% fewer votes gave the CAQ 35 fewer seats than the PQ.
So once again, the first-past-the-post electoral system did not correctly translate into seats the prescription voters attempted to write (see more below).
It could be argued that the Liberal vote is the strongest, especially on the Island of Montréal, since it has the largest number of “safe” ridings where it doesn’t need to grow. Out of 125 ridings, only 27 were won with a real plurality of votes (i.e., 50% + 1 votes), and of those 27, the Liberals took more than half:
- Liberals: 14 ridings
- PQ: 8 ridings
- CAQ: 5 ridings
On the other hand, if it can be argued that second-place finishes are a gauge of growth potential, then the CAQ is clearly in the lead, indicating that its support is more evenly distributed across Québec. However, in a first-past-the-post, single-member-plurality system like ours, seats are earned when support is concentrated within ridings (think Québec Solidaire’s Françoise David in Gouin), with the perverse effect that the excess votes for a party winning a true majority within a riding are lost to the whole (think the Liberals’ habitual landslides in D’Arcy-McGee and see also the discussion on unused votes below). Nonetheless, in this election, the second-place finishes were as follows:
- CAQ: 53 ridings
- PQ: 39 ridings
- Liberals: 30 ridings
- Québec Solidaire: 2 ridings
- Option Nationale: 1 riding
A more polished gauge of growth potential would subtract the number of ridings in which the winner earned a real plurality. Just a few examples:
- As previously alluded to, in the riding of D’Arcy-McGee, Liberal Lawrence Bergman won with 84.72% of the vote against the CAQ candidate’s 7.37%.
- In the less extreme case of LaSalle, Robert Poeti gave the Liberals the riding with 56.88% of the votes against the PQ candidate’s 19.76%.
- The Liberals took Saint-Laurent with Jean-Marc Fournier’s 65.69% of the vote compared to only 14.32% for the CAQ.
- For an outside-of-Montréal, non-Liberal example, consider how Alexandre Cloutier won Lac-Saint-Jean for the PQ with 53.15% against the CAQ’s 24.12%.
- Finally, for an example of a more traditional PQ/Liberal contest in the PQ stronghold of René Lévesque, Marjolain Dufour kept the riding for the PQ with 59.68% against the Liberal’s tiny 18.21%.
In short, as long as our current electoral and multi-party system persists, all three main parties need to put a bit of water in their wine when looking at their second-place finishes. However, for Québec Solidaire, by winning two seats but neither with a real plurality, its two second-place finishes, both in the east end of Montréal close to the “conquered” Gouin and Mercier ridings, could point to its true growth potential.
Correctly Translating the Voters’ Prescription
While I personally do not like the new centre-right upstart CAQ, if only on the basis that it IS centre-right (and that its proposed solutions to “clean up” government and bureaucracy seem unrealistic or inhumane or too anti-union or all of the above), I have to admit that an important number of my concitoyens québécois felt that the CAQ’s proposals had merit. Conversely, a much smaller but significant number also felt that Québec Solidaire had something to bring to the table. Yet both parties obtained a smaller proportion of seats in the National Assembly than indicated in the popular vote (-11.85% for the CAQ and -4.43% for QS), while the top-two parties got a larger proportion of seats compared to the popular vote (+11.27% for the PQ and +8.80% for the Liberals).
One reform of the electoral system would give voters two ballots: one for a local member and one for a regional member. Proponents of this format argue that greater proportionality can be achieved due to voters being able to split their support between two parties. In fact, more than ever before in this election, had I been given one vote for my local MNA and one for my regional MNA, I may have split my votes. For instance, I discovered that there’s an upstart left-of-centre but non-sovereignist party called the Quebec Citizens’ Union. Given that I live in the part of Montréal that has gone Liberal for decades, I could have voted “left” locally while ignoring that party’s sovereignist plank but QCU regionally since it currently holds little chance of winning a local seat yet could gradually grow to become a regional player.
Another form would maintain a single ballot but for fewer local ridings and grouping those locals into a small number of regions with roughly the same number of locals in each. If the total number of ridings in Québec was maintained at 125, two-thirds (or 83) would be local ridings in which first-past-the-post would remain, and one-third (or 42) would be filled based ONLY on the popular vote. In that case, knowing that voting for a fringe party might not be the equivalent of throwing away my vote, I might consider a party like the QCU.
To achieve near-perfect proportionality, the common denominator that would be used would be that of the entire territory, namely 42. However, deciding which region would get members of a given party could be tricky. Say the QCP mentioned above were to earn one of those 42 seats, it would make no sense to assign that seat in a region where the party earned little to no support. Distributing regional seats by regions could alleviate that problem, but reducing the common denominator from 42 to 10 or 11 could result in a far less perfect proportionality, though certainly much closer than FPTP can ever achieve.
This is strictly a mathematical problem, however. With computers making complex mathematical calculations quick and easy, the total number of regional seats could be calculated with the larger denominator of 42, but distributed by regions with at least 5% support. Say Québec Solidaire earned 5 regional seats overall but got less than 5% in one region, 6% in each of two other regions and 15% in the Montréal region, 3 regionals would likely end up representing the Montréal region while 1 would go to each of the other two “eligible” regions in order to bring the QS’s representation as close as possible to the regional popular vote. This idea may sound complicated, but remember: it’s only maths which a computer can tabulate in less than a second and the dust would settle as soon as all the results are in.
Caution must be taken when trying to “re-count” results yielded from a first-past-the-post election to project what the results may have looked like in mixed-member proportional (MMP) system.
- First, some political pundits don’t believe that strategic voting exists in the current FPTP system, but while I believe its existence may be overstated, I do think it exists. Had I believed the CAQ stood of a good chance of taking my riding of Outremont, I would have strategically voted Liberal — no doubt about it. But where Outremont has been Liberal since its creation in 1966 and it would take a heck of a lot more than the CAQ to end this dynasty, I voted with my heart.
- Second, these numbers were the result of 125 individual races. If there had only been 83 such races, each riding would have been considerably bigger and would have included a wider array of political views. For instance, if my Outremont riding went much further east, it would take in a lot more PQ or QS sympatizers, so it wouldn’t necessarily be a Liberal stronghold.
Still, using real election results is the soundest indicator of alternate results at our disposal.
Distribution of Regional Seats by “Unused Votes”
The basic idea is that some parties need fewer votes to win a seat while others need many more votes. If one party’s support is more concentrated in a set of ridings, its vote is more efficient than that of a party whose vote is more evenly distributed (i.e., not as concentrated). The most dramatic example of this notion is the 1993 federal election, when 2,189,067 votes (15.97% of the popular vote) spread nationwide gave the Progressive Conservatives only 2 seats, while 1,851,835 votes (13.51% of the popular vote) coming only from Québec gave the Bloc Québécois 54 seats and the official opposition. Thus arises the notion of “unused votes” — votes that are virtually “thrown out” and don’t earn a seat for a given party. Therefore, this number of unused votes should be brought down to compensate for a party’s inefficient vote.
Independant candidates and parties that don’t achieve a certain percentage in the whole territory would be immediately excluded from this compensatory redistribution. I chose 5% as the threshhold, but it could be as little as 2% or as high as 10%. However, my experiments with dozens of election results has shown me that 5% achieves a good proportionality while excluding truly marginal or dying parties.
The first table in this post already shows how many seats each party won, but here’s the simplication.
Understandably, the number of unused votes for the winning party is always the lowest. In this election, the PQ’s number of unused votes was 25,337, calculated as:
Total number of seats won + 1 (54 + 1)
Using this formula:
- the Liberals’ unused vote number was close at 26,698;
- the CAQ’s unused votes number was 59,038 (2.3 times higher than the PQ’s);
- Québec Solidaire’s was 87,744 (almost 3.5 times higher than the PQ’s), and
- Option Nationale’s, having earned no seat, was its whole take, or 82,857 (almost 3.2 times the PQ’s), but its poor overall showing in the popular vote would exclude it from receiving regional seats.
Invariably, as this table shows, parties whose vote is more efficient took more than their fair share of the seats in the National Assembly. Note that the “Votes/Seats” row should really be labelled “Unused Votes.”
I suppose the good news in this election is that, unlike in other elections, no party achieving at least 5% of the overall popular vote was shut out (i.e., didn’t earn at least one seat), and with only 1.9% of the popular vote, Option Nationale wouldn’t qualify for regional representation. Therefore, whether by FPTP or MMP, 165,576 votes (or 3.79%) still wouldn’t serve to elect a particular party. That wasn’t the case in New Brunswick’s last provincial election on September 27, 2010, where the NDP was shut out despite earning 10.4% of the popular vote. In fact, when one thinks about it as much as I do, it seems that the FPTP system almost always gets it wrong: in Newfoundland and Labrador’s last election on October 11, 2011, FPTP gave the official opposition to the third party in the popular vote despite it getting 7% fewer votes than the second-place finisher.
For the sake of argument, let’s assume that each party received the number of votes they did on September 4, but that there had only been 83 local races. Thus, the next table shows how many local seats each party would have won, proportionally, then how many regional seats each party would have received, the percentage of over- or under-presentation of each party, and each party’s final, more even number of unused votes.
Although it didn’t happen this time, independant candidates or parties not getting 5% of the popular vote could win one or more local seat, and they would retain such seats. In fact, in the previous Québec election on December 8, 2008, Québec Solidaire did win a seat through FPTP, but by virtue of only getting 3,78% of the overall vote, it would not have received any of the 42 regional seats.
Had the 83 local seats been distributed as shown above, the disparity in the number of unused votes by party would have been striking, particularly for Québec Solidaire.
In other words, the gap in the number of unused votes between the PQ and QS would have been a whopping 93,954. Thus, with its meager 1 seat and the highest number of unused votes, QS would get Regional Seat #1, and then its number of unused votes would be recalculated by taking into account its new seat count, as follows:
Total number of seats won + 1 (2 + 1)
This exercise would be repeated 41 more times, each time assigning the next regional seat to the party with the highest number of unused votes.
Thus the Liberals would only get their first regional seat at #22 and the PQ at #27, with the CAQ earning the most regional seats (22 in all). By the end of this exercise, the gap between the highest and lowest number of unused votes would be reduced to a mere 3,768 from 93,954. And the only reason that perfect proportionality wouldn’t be achieved is because 3.79% of the votes went to parties not meeting the 5% overall popular vote requirement, so truly “popular” parties would share the spoils.
I know some might gasp upon considering a such a weak minority government:
But wouldn’t that be the equivalent of saying that the citizens of a healthy democracy like Québec’s can’t write the right prescription for itself? For surely, if you ask the third of voters who supported either the CAQ and QS, something got lost in translation under the current FPTP system.