Whatever the reason for Canada being one of the world’s oldest and most stable democracies, it is not because Canadians understand exactly how it works. Since Confederation we have had 17 changes in the party controlling government at the federal level, and hundreds provincially – all of them peaceful, some of them consequential. But as the end of the 2019 federal campaign made clear, Canadians can be quite confused about how governments are chosen.

While citizens of France and the United States vote for who their president will be (in the U.S. case, if we ignore the Electoral College), Canadians vote for their prime minister only if he or she happens to be running in their riding. Instead, we elect a federal House of Commons or provincial legislative assembly. The partisan makeup of the house determines who gets to wield executive power. When one party wins a majority of seats, how this occurs is pretty straightforward: the leader of that party becomes the first minister – that is, prime minister or premier – and selects a cabinet.

But things are not so clear if there are more than two parties and none of them gets a majority of seats. If one party gets the most seats, does that party automatically get to form the government? Or is it legitimate for the other parties to agree among themselves and depose the government without an election? What happens when the incumbent party does not get the most seats, but the former opposition does not have a majority? Should the incumbent first minister give way to the leader of the opposition, or can he or she try to stick around and put together a working majority? When a government is defeated on a matter of confidence, when does that mean it must give the reins over to someone else and when can it “go to the people” in an election?

All of these issues have been matters of partisan debate in Canada in the last decade. In some cases, there is an expert consensus, but sometimes controversy arises among the coterie of constitutionalists – a group with no formal membership qualifications, no principle of accountability to anyone and no clear way of resolving disputes.

A Murky Area

Constitutional issues other than government formation are legal matters – which, while sometimes uncertain, can at least be authoritatively resolved by the courts. But the courts have refused to step into government formation – although even that principle took a beating in the United Kingdom when its Supreme Court held Boris Johnson’s request for prorogation to be unlawful and of no force and effect. Canada’s system is based on the U.K.’s, but if courts were to step in here, it would be revolutionary and inevitably controversial. And of course, since what is at stake is power, these disputes are not going to be conducted disinterestedly.

To be sure, most of the time, the system works whether it is universally understood or not. In practice, the “conventions of responsible government” – however mysterious they may be to the laity and even sometimes the clerisy itself – give a clear result about who is supposed to occupy 24 Sussex Drive (assuming it is ever made habitable). Contrary to semi-informed opinion, the representatives of the Crown – the governor general at the federal level and the lieutenant governors in the provinces (collectively, the governors) – rarely have any real choice in what to do.

But there are real question marks. In Canada, there is a particular question about whether an incumbent government that gets fewer seats than one of its rivals but can see a way to put together a working majority must give the party with the most seats a shot at governing. Since this eventuality almost happened in both of the last two federal elections – and in fact occurred in New Brunswick in September 2018 – we really should have some clarity about it.

Unfortunately, the answer requires some nuance, which partisan politics and media regard the way cats feel about baths. A governor would – and should! – let an incumbent first minister, no matter how many seats his or her party got, put a throne speech to the house. If the incumbent government lost a confidence vote at that time or shortly afterwards, then, and only then, the governor would call on the leader of the party with the most seats.

However, this does not mean the first minister who decides to do this is off the constitutional hook. First ministers are supposed to give governors the right advice. So even if the governor lets the first minister meet the house, that leaves open the question of whether the first minister ought to put the question to the governor in the first place.

In my view, an incumbent first minister whose party (or pre-election coalition) does not get a plurality should advise the governor to call on his or her more successful rival to take the first crack at governing. In this respect, Andrew Scheer was right in October 2019 to argue that there is a “modern convention” that the party with the most seats has a right to try to govern.

However, Conservative partisans were wrong to suggest – either in 2019 or in 2008 – that their opponents are obliged to leave them in office. On the contrary, if a parliamentary majority supports the old government, then the right thing to do would be to let the plurality party give a throne speech, but move an amendment that the house has no confidence in the new government. If that passes, the old government has every right to come back and govern as long as the new parliament lasts.

Put that in your 30-second ad buy.

Election 2019 and the Constitution

Election 2019 was not all about the Prime Minister’s more or less youthful ventures in racially insensitive costuming or the Leader of the Opposition’s inability to keep straight his qualifications to practise as an insurance broker or to get a U.S. passport. In addition to such substantive issues as climate change, tax policy and the prospects of a national pharmacare plan, the campaign briefly touched on the mysteries of the constitutional principles governing the formation of executive government.

In the end, we had a fairly boring result from the perspective of an enthusiast of Westminster system arcana. As a result of the much greater efficiency of their vote compared to that of the slightly larger portion of the electorate that voted Conservative, Prime Minister Justin Trudeau’s Liberals, while denied a majority, received a strong plurality of seats in the October 21 election. Since Jagmeet Singh’s New Democratic Party made it abundantly clear in the campaign that it would never support a Conservative government, and since many of the NDP’s policy objectives overlap with those of the Liberals, no one doubts that Trudeau can continue as Prime Minister.

Pundits can of course speculate on how long a Trudeau government will last before another election, but the Liberals clearly have the authority to remain in government with the cooperation or acquiescence of the smaller parties. The Liberals have ruled out formal cooperation, but they will be able to get any measure passed as long as they have the support of one of the Conservative Party, Bloc Québécois or NDP. A premature end to this Parliament forced by the opposition seems unlikely.

But if things had turned out slightly differently, we might be in the midst of a constitutional crisis. Shortly before voting day, Conservative leader Andrew Scheer created a ruckus by claiming that if his party received the most seats, “modern convention” meant he should get the first chance at being prime minister after the election. He could point to a similar statement by Justin Trudeau before the 2015 election, when it appeared quite likely that the Liberals would get a plurality at the expense of the Harper Conservatives, but before their last-minute momentum delivered a majority.

Scheer’s advisers appear to have thought that raising hypotheticals about what might happen after an election was a strategic misstep and he quickly turned to emphasizing the benefits of a Conservative majority. The only party leader who directly engaged Scheer’s constitutional claim was Elizabeth May, leader of the Green Party, who claimed that Westminster tradition gives the first right to form a government to the incumbent party, regardless of how many seats it gets.

The unceasing electronic bar fight / seminar that makes up our contemporary public sphere briefly filled the gap. Constitutional law professors, media talking heads and partisan trolls with five Twitter followers debated questions of “convention”, “precedent” and “principles of responsible government.” And just as quickly, the election was decided and the bar fight / seminar turned to new entertainments.

It might be worth thinking about the fact that this leaves an unexploded landmine in our political garden party. Sooner or later, the scenario Scheer raised will happen. The controversy was reminiscent of the debate that followed on the non-Conservative parties’ brief attempt to replace the Harper government at the end of 2008. Lovers of Canadian party politics can point to numerous earlier examples of party conflicts over the rules of the Canadian political road, most memorably the King-Byng crisis of 1926.

In these situations, media speak loosely. Partisans argue partisanly. Once upon a time, perhaps, there were universally recognized constitutional experts such as the late Senator Eugene Forsey who could upbraid imprecise punditry and silence the hacks. But in today’s flattened opinion environment and general distrust of expertise, who will play that role when a future electorate steps on the landmine?

The Mysteries of Westminster Government

It would be nice to clarify beforehand how power would peacefully be transferred. So what can we say for sure? What are the fundamentals of how Canadian governments are chosen?

For most purposes, if we are interested in how the right to exercise executive power gets determined, we can fairly simply divide democracies up into those with a parliamentary system and those with a presidential system. In a Madisonian or presidential system like the United States, the legislature and executive are elected separately, and they frequently have different partisan alignments. In the United States, as I write, the Democratic House of Representatives is about to impeach Republican President Trump. Whether he is removed from office by the Republican-controlled Senate (which seems unlikely) or not, whether he is reelected or not, and whoever replaces him, we can expect American politics to be dominated by conflict and occasional compromise between the executive and legislative branches for the foreseeable future.

In parliamentary systems, this is not supposed to happen (although the Brexit imbroglio, touched on elsewhere in this issue of Inroads, shows it sometimes can). The executive is not separately elected. Instead, the right to exercise executive power depends on being able to get the support or acquiescence of the legislature. If the executive and legislative branch come into serious conflict, then one of them must go: either the executive by a change in government or the legislature by a new election.1

Parliamentary systems have arcane and technical distinctions in how it is decided who has the right to be in government when the will of the legislature is not clear – differences that do not matter when there is a clear majority for a party or coalition, but can make a big difference when there is not. Many parliamentary systems provide for an explicit “vesting vote”: after each election and a transition period, the legislature votes for who the new executive will be.

For example, section 46 of the Scotland Act gives the Scottish Parliament the power to nominate one of its members as first minister during a transition period after an election or the fall of an old government. While that person is technically appointed by the Queen, in effect the Scottish people elect the legislature and the legislature elects the executive. Most democracies in the world that have avoided the United States’s separation of executive and legislative authority provide for some similar process.

But this is not how it works in the United Kingdom as a whole or in countries, like Canada, that have adopted its specific form of parliamentary government. While executive power depends on the “confidence” of the legislature, this is not determined by the relatively straightforward method of a vote at the outset of a parliament, but through conventions about when first ministers are supposed to resign and when governors are supposed to dismiss them.

Our Misleading Constitution

Canada’s written constitution, although it purports to declare “the Nature of Executive Government,” is in fact completely misleading on the subject. If you just read Canada’s constitution, you would think executive power belongs to the Queen (section 9), that she delegates it to a Governor General who serves at her pleasure (this went without saying), who in turn appoints members of the Privy Council (section 11), decides on judges (section 96) and has to agree to all legislation (section 91). The prime minister is not mentioned at all in the 1867 constitution – he or she is just an unnamed member of the Privy Council chosen and removed by the obviously more important governor general “from Time to Time.”

The prime minister only plays a cameo role in the rest of our written document: under section 35.1 of the Constitution Act, 1982, he or she gets the neocolonial power to decide what representatives of Aboriginal people will be invited to constitutional conferences about amendments that affect them and must attend such conferences. That’s it for textual attention to the most powerful official in the country. As a written document, the Canadian constitution is about as deceptive a guide to what is really going on as Stalin’s 1936 Constitution of the Soviet Union. Where the USSR was a personal dictatorship pretending to be a democracy, on textual evidence Canada is a democracy pretending to be a personal dictatorship.

One thing the constitution does make clear is that the executive power and legislative authority are legally distinct. This principle predated the loss of effective authority by the monarch personally and was inherited both in the United States (where the executive became directly elected) and in Westminster systems. The legislature is sovereign, except as limited by the constitution: the executive only has the powers the law gives it and in particular must always abide by statutes the legislature enacts. This legal superiority of the legislative branch can create a cynical contrast with the effective power of the executive in a system of party discipline – but it can also sometimes bite governments when they don’t expect it.

But despite the importance of this legal principle of legislative sovereignty, it would be unwise to rely very much on the text of the constitution to understand the link between “supreme executive power” and a “mandate from the masses” (in the words of Dennis the Peasant in Monty Python’s classic take on the British constitutional tradition). The unwritten “conventions of responsible government” provide the missing link between Canada’s monarchical written constitution and its representative reality. With very limited exceptions (the “reserve powers”), the governor must always do what he or she is told, whether by the first minister (for example, in appointing a cabinet), by cabinet (in passing an order in council) or by the legislature (in giving assent to legislation). Since the first minister decides who the cabinet is, the key question is who gets to be first minister.

The rule is not, as in Scotland, the positive one that the legislative body elects or nominates the first minister. Rather, the rule is a negative one. Canadian governments do not die of natural causes: they must be killed or commit suicide. A government continues until the first minister resigns or (much more rarely) is dismissed. The rules about when these things are supposed to happen are therefore all that keep Canada democratic. Governments are not elected: the provincial legislative assemblies and the federal House of Commons are the only elected bodies in our system.

Some of the conventional rules are clear. If a government loses the confidence of the elected house, then the first minister must either resign (in which case the governor will call on the leader of the opposition party with the most seats) – or ask for a new election. The governor will accede to a first minister’s request for a new election if the parliament has been around for a while – about six months – but not otherwise (unless it has been demonstrated that forming a stable government is impossible). In 1981, the Supreme Court of Canada added that there is a convention that if the opposition obtains a “majority” at the polls, the government must resign “forthwith” (in practice, whenever the new government is ready to be sworn in).

These bare-bones rules are enough to say what will happen with the current Parliament. Justin Trudeau has nether resigned nor been dismissed, so he remains Prime Minister. No opposition party obtained a majority at the polls, so there is no convention that requires him to resign “forthwith.” He will have to ask the Governor General to appoint a new cabinet. While it is conceivable that he might nominate cabinet members from other parties, he is under no obligation to do so.

The government will put forward a throne speech. The other parties will have the opportunity then or later to vote nonconfidence, but they will not do so unless they see the advantage of an alternative government or a new election. If the Liberals lost a confidence vote, they would not be entitled to an election for the first six months or so, but after that they could have one any time the Prime Minister decided it was in his political interests. (Canada has a fixed election law, but it has an exemption in these circumstances and the courts have already made it clear that they will not get in the way.)

How Minority Governments Govern

That does not mean it would make sense for the Liberals to try to rule as if they had a majority, as Joe Clark rashly promised to do when he had a minority government in 1979. In the last Parliament, the Liberal Party controlled every legislative committee and the government could prevent any legislation passing against its will if it was willing to whip its own caucus. This is no longer the case. The Liberals have said they will not try to get a coalition or even confidence and supply agreement with another party: they will have to rely on the desire of other parties to avoid elections to get budget measures and other matters of confidence through. But there is very little doubt about what is supposed to happen.

In 2008, when a coalition of the Liberals and NDP supported by a confidence and supply agreement by the Bloc Québécois declared its readiness to unseat Stephen Harper’s minority Conservative government, the Conservatives argued that this arrangement was illegitimate. Their argument was as constitutionally unfounded as it was politically effective.

In the current Parliament, the opposition parties could, in principle, vote nonconfidence in the government – so long as they do so early in the Parliament – and the Governor General would call on Andrew Scheer to be Prime Minister. This is what occurred in Ontario in 1985 (when Bob Rae’s third-place NDP supported David Peterson’s second-place Liberals) and in British Columbia in 2017 (when the Green Party supported the NDP). However, as a matter of political reality, it seems next to impossible to imagine the federal Liberals facing any similar effort by the opposition parties this time, even if the NDP had not explicitly ruled out cooperation with the Conservatives.

There is a Modern Convention …

But what if the 2019 election had been slightly different? What if the mysteries of voting efficiency and strategic voting had resulted in the Conservatives obtaining more seats than the Liberals? What would convention call for then? Was Scheer correct that “modern convention” implies that he would have had the first chance at governing?

If we go back far enough in Westminster history, the answer would have been a clear no. Trudeau would remain Prime Minister unless and until defeated on a confidence vote, and so the situation would not be materially different from the one that actually unfolded. This is because responsible government emerged out of a system of dual confidence: in the 18th century, for example, the government of the day required both the confidence of the monarch personally and the confidence of the House of Commons, so that the monarch could continue to tax and spend (“enjoy supply”). Just as a minister could continue on so long as the monarch had not announced that this was no longer his or her pleasure, so too he could continue so long as there was no affirmative denial of confidence or supply by the lower house. The Hanoverians eventually lost the practical ability to dismiss governments for policy reasons, so the principle of responsible government became in effect that the Crown hired governments and the House fired them.

But it would be wrong to end the development of the principles of responsible government at the accession of Queen Victoria. In Canada, the aftermath of the 1896 election created a new principle. The Conservatives, led by Charles Tupper, lost that election to Wilfrid Laurier’s Liberals, who obtained a majority of seats in the new Parliament. Tupper took the perfectly orthodox view that he remained Prime Minister until the House met. He hoped, no doubt, to do some kind of deal with some Liberal MPs – a greater possibility in the late 19th century than it has since become. But the Governor General, Lord Aberdeen, refused to take instructions from Tupper on appointments, forcing Tupper to resign. Aberdeen asked Laurier to take office as prime minister.

Tupper complained about this breach of the principles of responsible government for the rest of his political life, but Aberdeen’s actions seem obviously correct to us now. Two conventions come out of this event: first, if another party obtains a majority, an incumbent first minister must resign effective as soon as the other party’s leader desires, and second, in such circumstances, the government must act as a “caretaker” – a role that has now been expanded to the entire period from when the election is called until it is affirmatively established who has the right to govern (in a minority, by the acceptance of the throne speech).

In the early-19th-century model, while the House decided when a government came to an end, the Crown had a great deal of discretion about whom to call on and whether or not to give a defeated government the option of going to the electorate. Scholars like Forsey and, more importantly, actual practice have constrained that discretion to a number of rules.

In particular, it is now widely (if not universally) accepted that when a government falls, the Crown must call on the leader of the party with the next-most seats and that it is wrong for the governor to use his or her own sense of who could command a working majority. It is also now accepted, on the basis of both scholarship and practice, that an election request will be denied early in a Parliament (with exceptions where it is beyond reasonable dispute that no government can function) but will be granted after six months or so. These are all additions to the original rule that governments continue until they lose the confidence of the house. They make sense based on practice and on the principles that the Crown should avoid controversial partisan decisions and parties should be treated with symmetry.

Viewed in this light, Scheer’s claim (earlier made by Justin Trudeau) – that an opposition party that obtains the most seats in an election should get the first opportunity to meet the house – would appear to have merit as a “modern convention.” It is what has in fact happened federally in every minority parliament after 1925.

In 1925, Mackenzie King’s Liberals initially held on with the support of Progressive and Labour MPs despite getting fewer seats than the Conservatives – this constitutional fact has long been overshadowed by the more famous decision of Lord Byng to deny King an election when the Progressives pulled their support. After the 1926, 1957, 1963, 1979 and 2006 federal elections, incumbent governments stepped down when they received fewer seats, while no incumbent government at the federal level has ever stepped down when it received a plurality of seats in a minority parliament.

Precedents can be read in multiple ways. If a course of action has made political sense in the past, this does not mean it is a convention. A convention requires both that the course of action be viewed as binding and not merely strategic and that it make sense in principle. Here too, Scheer’s claim seems vindicated. The prime ministers who gave way all those times no doubt believed, without exception, that their own policies were better for the public interest than those of their opponents; they gave over power because they thought they were obliged to. In many cases, it is easy to imagine means they could have used to avoid a nonconfidence vote.

Moreover, Scheer’s approach is supported by principle. Having the first-mover advantage in a minority parliament carries the significant benefit that the other parties must affirmatively displace you – after six months or so, at the risk of an election. Therefore, if the system is to be symmetrical, this benefit should be allocated in a way that minimizes Crown discretion and is allocated as much as possible on the basis of how the people voted. That is why it is now more or less universally accepted that after a defeat, the governor should go to the party with the next most seats to form a government. The number of seats, while not determinative of the ability to command the confidence of the house, is an objective fact based on how people voted, and not a subjective decision of the governor or existing first minister.

… But it Has a Caveat

While the Conservatives were right this time, there was some suggestion that they were going to take the position they took in 2008 as well: that if they got the most seats, they not only had first crack at government in the new Parliament, but also last crack – that any arrangement between the Liberals and one or more of the third parties would be illegitimate. I was unable to track any example of Scheer himself making this claim, but there was some suggestion of this by some of his surrogates.

This would of course turn a reasonable position into the entirely unreasonable notion that governments, like parliaments, are chosen on a first-past-the-post basis. This is unreasonable because the whole point of responsible government is confidence of the majority of the legislature that can pass laws and approve taxes. If the Conservatives had received the most seats, Scheer would have the right to meet the House, but the House would have the right to vote him out and put Trudeau right back in office.

For this reason, there is undoubtedly a caveat to the modern convention. If an incumbent first minister could arrange for an agreement that guaranteed confidence and supply very quickly (say, in the first few days after the election), then calling on the leader of the party with the most seats would be pointless, since it would just lead to a nonconfidence vote and restoration. So, if the election had resulted in a situation where the Liberals came second, but there was a more or less instant promise of support from, say, the NDP, and the Liberals and the NDP combined could constitute a majority, then it would be legitimate for Trudeau to stay on. If it included this exception, Scheer’s “modern convention” could arguably even encompass the 1925 election, after which the Progressives and J.S. Woodsworth’s Labour group quickly supported the incumbent King government.

Many of those in the opinion sphere who argued that Trudeau could simply continue on if he failed to get the most seats cited Philippe Lagassé, a Carleton professor and expert on the Westminster system of government formation. In fact, Lagassé recognizes that, in Canada, incumbent governments that receive fewer seats in minority parliaments have not tried to stay in power since 1925 and that there is a norm that supports Scheer’s claim. He insists on calling this norm a “custom” rather than a “convention.” Since conventions are not laws and derive from the accepted norms of political actors, it is hard to see how this line can be successfully maintained.

To be fair, I would agree with Lagassé that a governor would probably not actually dismiss an incumbent first minister who tried to continue after getting fewer seats in a minority parliament. This is what Brian Gallant tried to do after the 2018 New Brunswick election, for example.

But where I disagree is that this is because there is no convention. A governor should only dismiss a first minister in the clearest of circumstances, where there is absolutely no doubt that the first minister ought to resign. For that reason, as long as everything is running correctly, a dismissal should never be necessary. A first minister should resign when convention dictates, and if first ministers regularly do resign – and feel themselves obligated to resign – in certain circumstances, then that is sufficient for there to be a convention. It is the first ministers themselves who are the first line of defence of conventions, although governors sometimes have to stand up for them independently.

If Trudeau had tried to continue with fewer seats and no agreement to get an effective majority – something he never said he would actually try to do – the Governor General might not have dismissed him on grounds of lack of clarity, but that would not itself mean he was acting appropriately. Of course conventions, like all norms, can ultimately cease to have force if they are violated enough, although they can sometimes get greater strength if they are violated but the violator is punished. This is true of Scheer’s modern convention – and also of all the other norms that are essential to the operation of the system.


1 Part of the root of the Brexit mess is that the Parliament of the United Kingdom recently eliminated this rule, while retaining a system that depended on it. While the U.K. House of Commons can still get rid of the U.K. government, the converse is no longer possible and Brexit shows that a majority of the House may have serious differences with the government without having a realistic alternative. The impasse was finally resolved when the opposition Labour Party finally agreed to Prime Minister Johnson’s request for an election, scheduled for December 12.