Wednesday, December 3, 2014

The seat implications of electoral reform

In my article for the CBC today, I've taken a look at the various proposals for electoral reform and what they would mean in terms of seats. The Conservatives want to stick with first-past-the-post, the system currently in place, while the NDP is today putting forward a motion in favour of mixed-member proportional representation. The Liberal Party's membership has supported moving to a preferential ballot, though it remains to be seen whether or not the Liberals will add that to their platform for 2015. It seems likely, considering that Justin Trudeau has often mentioned it.

Please go take a look at the article at CBC.ca before reading any further. I've explained what I've done there, and also discussed some of the practical implications. But I thought I'd provide a little more detail here at the regional level for these seat estimates.

First, let's just start with the current polling averages as they were what all of my calculations were based on.

Current polling averages
The Liberals have 35% support, followed by the Conservatives at 31% and the New Democrats at 21%.

Now, here is what these numbers would likely give in terms of seats.

Note that, despite trailing by four points, the Conservatives narrowly come out in front in the seat count nationwide. Also, despite having a combined 11.4% support, the Greens and Bloc win just 0.9% of the seats.

The Conservatives win the most seats in British Columbia, despite trailing by three points there. The New Democrats take the bulk of Quebec's seats, despite being two points back of the Liberals.

Next, mixed-member proportional representation which, for the purposes of this analysis and easier comparison, I've simplified to proportional distribution of the 338 seats in the House (which is, in the end, the net result of the proposed system).

Note here how some of the discrepancies in FPTP are rectified, notably in British Columbia and Quebec. Also note that the opposition parties have a fairer shake in Alberta and the Prairies, while the Conservatives get their due in Quebec and Atlantic Canada.

And finally, my estimates of the outcome of a preferential ballot, all else being equal (with MMP and especially with preferential balloting, the parties would change their behaviour - but let's assume they didn't).

Now there are new discrepancies between first-ranked support and the eventual outcome. The Liberals fall just short of a majority. The NDP comes in second despite trailing the Conservatives by 10 points on first-ranked balloting.

Regionally, the Conservatives finish third in B.C. despite being second in support and the Liberals finish third in the Prairies despite being 13 points up on the NDP. The Liberals again finish second, but now by an even greater margin, in Quebec. And the Conservatives are shut out of Atlantic Canada though they are virtually even with the NDP.

But while preferential voting does not mimic first-ranked support, that does not necessarily make it non-representative.

Look at the EKOS polling I based these second-choice calculations upon. Combining first and second choice gives the Liberals a "theoretical ceiling" of 57%, against 54% for the NDP and 35% for the Conservatives. This is not all that different from Nanos Research's numbers on which parties Canadians would consider voting for: they were 56% for the Liberals and 43% apiece for the NDP and Conservatives in today's update from the firm. (This is actually an uptick for the Tories - over the last year the Liberals have averaged 53%, the NDP 43%, and the Conservatives 40% in Nanos's polling).

Food for thought, if this discussion ever becomes a serious proposal. I think a majority government by either the Liberals or the NDP would be required to get either of these changes implemented (though I wonder about the legitimacy of one party pushing through drastic electoral change). For the moment, at least, that does not seem like a probable outcome for 2015.

Monday, December 1, 2014

147 years of elections

Polling has been a little thin lately, so let's take a look at the "only poll that counts." (As an aside, I loathe that phrase - it is the equivalent of saying "the only time your opinion counts is when my job is at stake.") And that means every federal election that has been held since 1867.

One of the things that struck me as I constructed the chart below was the degree of stability in Canadian politics. When you see the peaks and valleys, you may wonder what I'm on about. But political parties have often stuck to a fairly steady level of support.

This was particularly so early on in Canada's political history. Until the First World War disrupted the old order, parties shifted by only a handful of points. Between 1867 and 1911, the Conservatives took between 45.4% and 53.4% of the vote without fail. The Liberals captured between 45.1% and 53.8%. It was shifts in the margins that changed governments.

But that was in the era of a two-party system. There were periods of stability after the CCF arrived on the scene as well. From 1935 to 1953, the Conservatives (Progressive Conservatives after 1942) took between 27.7% and 31% of the vote in every election. Between 1963 and 1980 the Tories took between 32.4% and 35.9%. And even in the current period of turmoil, the Conservatives have only gone from 36.3% in 2006 to 39.6% in 2011.

The Liberals, too, had long periods of stability, capturing between 40.1% and 44.4% of the vote in every election between 1917 and 1935. Between 1993 and 2004, the party took between 36.7% and 41.3%.

And then there are the historically smaller parties, who have a much more reliable clientele. After the NDP's first two elections, the party took between 15.4% and 20.4% of the vote in every vote between 1965 and 1988. In fact, with the exceptions of the 1993, 1997, 2000, and 2011 elections, the party has fallen in that range in every election since 1965.

In their times, the CCF and Social Credit (with the Ralliement des créditistes) had a solid chunk of the vote as well, hardly shifting from one election to the next.

Of course, there were moments of great change: the first appearance of a three-party system in 1921, when the Progressives and United Farmers took 21.9% of the vote. The depression, followed by the Second World War, heralded a collapse in the Conservatives that was only briefly undone by John Diefenbaker between the 1957 and 1962 elections. Then there was the smashing victories by Brian Mulroney in the 1980s, followed by the final collapse of the Progressive Conservatives and the inauguration of a five party system. Most recently, there has been the swift decline of the Liberals and the breakthrough of the NDP in 2011. These again may be reversed in 2015.

(Note: The charts above and below were based primarily on the Library of Parliament figures on election results. Where there were gaps, I relied on the Wikipedia entries for older elections, which in turn are based on the Library of Parliament results as well. There are some notable differences in some of the numbers, however. This is likely due to the Library of Parliament records including some other candidates in the totals of the major parties, as in many elections there were Independent This and Labour/Progressive/Nationalist-That. I have deferred to the Library of Parliament when there have been discrepancies).

The chart I put together below shows the average results for each of the parties and their various iterations over select periods of time. In the brackets above each of the years I've put the reason why I selected these periods (i.e., a two-party system, a three-party system, the formation of the NDP, the collapse of Social Credit, etc.).

Overall, the Liberals have had the best historical performance with an average of 42.3% support, followed by the various classifications of the Tories (between 40% and 41.5%).

The NDP has averaged 16.5% support since 1962, putting them just up on the Reform and Canadian Alliance parties (who, recall, did field candidates in 1988).

The CCF averaged 11.2% during its history, while the Bloc Québécois has averaged 10.6% support. These are the only parties to have averaged over 10% support over their entire history.

The Progressives and United Farmers averaged 6.7% support, followed by Social Credit at 5% and the Greens at 2.4% (the Greens first fielded candidates in 1984, and took less than 1% of the vote in every election until 2004). Labour candidates averaged 0.9% in contested elections.

If we look at the era of the two-party system, from 1867 to 1917, the Conservatives were just a few points ahead of the Liberals. But since the three-party system started in earnest in 1935, the Liberals have averaged 39.2%, against 32.9% to 36.1% for Conservative parties.

Since the NDP was formed, Liberal support was roughly even with the Conservatives at about 36.7%, and since Social Credit and the Créditistes petered out in 1980, the Liberals have averaged 33.7% to 28.8% for the PCs or 38.1% for the combined totals of the PCs/Reform/Conservatives. The NDP upticks to 16.7% over that time.

During the era of the five party system, the Liberals reigned supreme averaging 40.2% between 1993 and 2000. Reform and the Canadian Alliance came second with an average of just 21.2%, followed by the PCs at 15.7%, the Bloc at 11.6%, and the NDP at just 8.8%.

But since 2004, in what we would consider the current system, the Conservatives have been ahead with an average of 35.8% support, followed by the Liberals at 28%, the NDP at 20.5%, the Bloc at 9.8%, and the Greens at 4.9%.

I also identified each party's 'peak period', which I have defined as three consecutive elections starting in 1935 (and for the Progressives and Labour, prior to that). 

It can be difficult for parties to put together three good elections. Consider Mulroney's victories in 1984 and 1988, in which he averaged 46.5%. That was the Tories' best performance following the inauguration of a three-party system. But those two elections were book-ended by defeat in 1980 and catastrophe in 1993.

The Liberals had their highest peak between 1940 and 1949, not coincidentally straddling World War II. The Liberals averaged 48.8% over that time, the PCs averaging just 29.3% and the CCF taking 12.5%.

The Conservative peak occurred under Diefenbaker between 1957 and 1962, when the party averaged 43.3% support. The Liberals took 37.8% over that time, the NDP/CCF taking 11.2%, and Social Credit averaging 7%.

The New Democrats are currently in the midst of their peak, and it seems likely that it will be extended into 2015. They have averaged 22.1% support since 2006, with the Conservatives at 37.9%, the Liberals at 25.1%, the Bloc at 8.9%, and the Greens at 5.1%.

Continuing through the list, the CCF peaked between 1945 and 1953, when it averaged 13.5% of the vote. The Bloc Québécois was at its height between 1993 and 2000, when it averaged 11.6%. And the Greens peaked between 2004 and 2008, with 5.2%.

Social Credit and the Ralliement peaked between 1962 and 1965, when together it averaged 10.7% support. That still put it behind the NDP, a party it never displaced for third spot.

The Progressives and the United Farmers peaked between 1921 and 1926, when they together averaged 12.2%. This overlapped somewhat with the peak of Labour candidates, who took 2.3% of the vote between 1917 and 1925. Though the three-party system only consistently took its place starting in 1935 with the formation of the CCF, from 1921 to 1930 the Liberals and Conservatives were deprived of an average of almost 15 points.

And that ends our little trip down memory lane as the on-going history of Canadian politics continues.