Canada’s Provincial Politics: Underhill, Revisited

I am following up on this recent analysis on the history of the federal party system by diving into the history of provincial politics. To begin, here is a diagram of all the provincial governments since Confederation and their partisanships, as well as (in the narrower bars below) the main party of opposition:

Governments of Yukon are included (all the way at the top, above BC!) as they are partisan; governments of the other two territories are not. Governments of New Brunswick were not formally partisan until about 1920, although there were clear informal Liberal and Conservative camps (hence the tildes in those cells).

There’s a lot of potential uses for this sort of diagram, but the most interesting for me was to see if it really is true that provincial governments systematically swing away from the federal party in power toward parties of the opposition: it very much feels like a general trend that provincial Liberals win when federal Liberals are out of power, and vice versa. It is known as the Underhill Theory when referring to Ontario specifically, and the reasoning presented is that voters might prefer to “balance” provincial and federal governments against each other, or that opposition partisans might be more motivated to turn out in provincial elections than government partisans, who might be disappointed in their party and want to send a message. (I covered this topic in a previous academic essay, but using other methods.)

Here’s the partisanship of provincial governments at each moment when the federal government flipped (except Arthur Meighen and Joe Clark, whose premierships were too short):

Year	Upon arrival of:	liberal	right
1874	Mackenzie	2	3
1878	Macdonald (2nd time)	2	3
1896	Laurier	5	1
1911	Borden	4	5
1921	Mackenzie King (1st time)	7	0
1930	Bennett	2	5
1935	Mackenzie King (3rd time)	8	1
1957	Diefenbaker	3	6
1963	Pearson	3	6
1984	Mulroney	0	9
1993	Chrétien	5	3
2006	Harper	3	7
2015	Trudeau fils	7	3
2023	[present]	2	8

Note that the categories are “Liberal” and “Right”: Social Credit and Union Nationale are Right; NDP, United Farmers, and Parti Québécois are neither; BC Liberals are both. A provincial government newly elected within six months of a federal government is counted as “upon arrival of”, as “honeymoon effects” are well-documented.

This is almost certainly a trend: not a single federal government left office with more provincial allies than when it arrived. While provincial Liberals were already in a rough spot at the conclusion of the Mackenzie King/Saint-Laurent years (1957), by the end of the Trudeau years (1984) they were decimated (or, indeed, worse) — not a single one was left in office. Not a single provincial Liberal party went from opposition to government between 1970 and 1985.

But then the rise of Mulroney quickly had the opposite effect: the long-running PC dynasty in Ontario was not just kicked out but was relegated to third place twice in a row (“third place twice in a row” is, of course, a phrase that perks up Ontario Liberal ears.) Social Credit in BC not just lost but ceased to exist; all four Atlantic provinces flipped to the Liberals.

(Side note: one possible evidence for a reverse effect, where the provincial government drives federal outcomes, is how the provincial NDP government in BC coincided with the collapse of the federal NDP there in 1993, when most of the its voter support moved to Reform. But once the provincial NDP was kicked out in 2001, federal NDP support recovered at the next election, in 2004.)

One possibility is that this is merely coincidental: as a rule, governments tend to tire out after 10-15 years. Voters might be throwing out both federal and provincial governments every 10-15 years without one causing the other, and it would yield a similar result. For example, 2003 saw Liberals take power in Ontario and Québec, even though a federal Liberal government had been in place for 10 years, because the local PC/PQ governments had worn themselves out after 8/9 years as well.

It’s very hard to conclusively prove or disprove this, given the small sample sizes. It is, however, worth noting that even exceptionally long federal governments, like Mackenzie King/Saint-Laurent and Pearson/Trudeau, saw provincial governments swing away on net, even though they lasted 1.5-2x as long as most governments; the “independent swings” theory would suggest that’s enough time for provinces to swing away and then back. Meanwhile, the Bennett government lasted only one term but still saw a collapse among provincial Conservatives — although he had a rougher tenure than most, having overseen the Great Depression. On aggregate, there probably is an effect of federal shifts causing provincial shifts, rather than the two operating independently.

Back to the present: the chart suggests that neither the bleak figure of two Liberal governments still in power (one provincial and one territorial), nor the net 5-government swing away from the Liberals over the last 7 years, is a historical aberration. Although this is small comfort for those who (let’s say, hypothetically) work for the Ontario Liberal Party.