Canadian Parliamentary Review

Current Issue
Canadian Region CPA
Upcoming Issue
Editorial and Stylistic Guidelines

HomeContact UsFrançais

Reconsidering Constructive Non-Confidence for Canada: Experiences from Six European Countries
Elsa Piersig

Canada’s recent run of hung parliaments (2004-2011) gave rise to a number of proposals intended to stabilize minority government. One such proposal recommends fixing the confidence convention by adopting a constructive vote of non-confidence that requires non-confidence votes to simultaneously elect a new head of government. Aucoin, Jarvis and Turnbull suggest that constructive non-confidence will increase parliamentary stability, legitimize mid-term transitions and reduce executive dominance. Yet, a cursory investigation of research on the constructive non-confidence votes demonstrates a dearth of evidence on the rule’s effects. This article fills this gap by reviewing other jurisdictions’ experiences with constructive non-confidence in order to unpack how the rule might work within the Canadian context. The comparative research demonstrates that though constructive non-confidence will enhance parliamentary stability, it will do so at the cost of decreasing the legitimacy of mid-term transitions and bolstering executive dominance over parliament.

The election of a minority government in 2004 was seen as a positive result by commentators across Canada, who believed that minority government would provide a healthy check on executive dominance and prime ministerial power. However, as the period of minority government unfolded between 2004 and 2011, several controversial uses of the confidence convention gave rise to reform proposals, including replacing Canada’s tradition of “negative” non-confidence votes with more “constructive” votes. Whereas negative votes simply withdraw confidence and generally trigger new elections, constructive non-confidence votes not only deconstruct a government but designate who should form a new one from the existing parliament (i.e., without new elections).

Constructive non-confidence votes exist in a number of countries, including Germany, Spain, Belgium, Poland, Slovenia, and Hungary. Peter Russell mentions this model as a possibility for Canada, and Peter Aucoin, Mark Jarvis and Lori Turnbull incorporate it into their constitutional reform package, alongside fixed election dates and the transfer of the prerogative power of dissolution from the governor general to parliament.1 These commentators believe that this constructive non-confidence package will enhance parliamentary stability by reducing brinkmanship, re-legitimize the ability of parliament to make and break governments, and reduce executive dominance.2

But would a constructive non-confidence reform package live up to these expectations? Would it truly enhance parliamentary stability and the importance of parliament, and if so, at what cost? This paper addresses these questions by examining the history of the constructive non-confidence confidence and its use in six European countries. Experiences with constructive non-confidence in these countries have been overlooked by Canadian academics despite their relevance for understanding the potential effects of such a reform in Canada. The evidence from this comparative review will be used to analyze whether constructive non-confidence would live up to the expectations placed upon it by its Canadian proponents. Drawing primarily on evidence from Spain and Germany, I conclude that adopting constructive non-confidence in Canada will increase parliamentary stability at the cost of greater executive dominance and the furthered delegitimatization of mid-term government transitions.

Constructing Confidence by the Half Dozen: Experiences from Six Countries

The constructive vote of non-confidence has received very little scholarly attention from both Canadian and comparative scholars. Within literature on government formation and resignation, it is generally covered only in passing.3 One notable exception is Diermeier, Eraslan and Merlo’s study, which measures the impact of constructive non-confidence on government stability. Beyond this literature, the most substantial coverage is featured in the literature on the development of the German Federal Republic’s Basic Law4 and in the literature on “chancellor democracy”/German executive dominance.5

Constructive non-confidence is the German response to the extreme parliamentary instability that undermined the Weimar Republic. The Weimar constitution had adopted a proportional representation electoral system that resulted in highly fragmented parliaments. Extremist parties capitalized on this fragmentation by uniting as negative majorities within the system – regardless of whether they had any other common ground, they agreed to undermine the chancellor and his cabinet, and thus the regime. This led to successive negative non-confidence votes and a diet of dissolutions which the Weimar president eventually tried to control by bypassing the parliament and appointing emergency chancellors. Since they generally lacked parliamentary support, these chancellors proved ineffective.

To address these problems, the German Parliamentary Council tasked with drafting a new constitution in the late 1940s sought rules that would allow the democratic regime to defend itself against radical elements, anti-democratic ideas, and alternative regimes.6 Known as the principle of military democracy, the drafters wanted the new constitution to protect the executive from “irresponsible” parliamentary behaviour and presidential interference and guard against destabilizing dissolutions.7 The Parliamentary Council found the solution in Baden-Wurttemberg where politicians had designed a constructive vote of non-confidence in conjunction with their American military advisors to secure the parliamentary executive.8 Constructive non-confidence was seen as a promising solution because it ensures that parties unable to agree on a new government cannot force new elections by withdrawing confidence from the incumbent government. The Parliamentary Council chose to combine the constructive vote of non-confidence with a fixed parliamentary term of four years and a safety valve needed for an early election if confidence was unattainable for all government options.

Aucoin, Jarvis and Turnbull also provide for a “safety valve” in case of unworkable parliaments. In fact, all constructive non-confidence regimes provide some kind of safety valve, and such a provision was certainly included in the pioneering German case. The German safety valve is provided for in Article 68 of the Basic Law, which allows the chancellor to call a confidence vote. Unlike non-confidence votes triggered by the opposition, an Article 68 vote called by the chancellor is not “constructive.” If the chancellor’s government loses the vote, he can ask for and receive new (and early) elections, which happened in 1972, 1982 and 2005.

Since the Germans put constructive non-confidence on the constitutional map in 1949, it has been picked up in five other European countries. Spain incorporated it into its 1978 post-Franco constitution for reasons similar to Germany’s; it too had seen political extremism overturn parliamentary and executive instability – cabinet duration was even shorter in the Spanish Second Republic than it was in the Weimar Republic – and sought a constitution that promoted a strong and stable democratic regime. In the wake of the fall of communism, Hungary (1989), Slovenia (1991), and Poland (1992) all followed suit in order to protect their emerging democracies. Lastly, Belgium (1995) adopted the reform as a means to ensure that once a government was formed it could remain in office, thus injecting a little more stability into a highly divided parliament.

Like Germany, all of these regimes also have safety valve provision quite similar to Article 68 in order to break parliamentary gridlock. One country, Slovenia, has added an extra twist that more explicitly encourages the formation of a new government over dissolution: if the president of the government (prime minister) introduces a confidence vote, the National Assembly must attempt to respond within 30 days by either electing a new government or reaffirming confidence in the incumbent administration. If the National Assembly fails to do so, only then is an early dissolution possible. However, regardless of this grace period, the Slovenian safety valve still opens the door for a government to engineer its own defeat in order to secure new elections.

Table 1 summarizes the use of constructive non-confidence votes to date. Votes are infrequent in five of the six democracies and have yet to occur in Belgium. When the rule has been used – a total of 12 occasions – it has been employed in two different ways: first, to engineer a mid-term transition involving a change in the composition of the governing parties, and second, to change the head of government without any change in the governing parties. The first category can also be sub-divided into cases where the parties moving a constructive vote actually wish to replace the government and cases where parties clearly have no expectations of success and are using the rule for some other reason such as raising their profile and gaining earned media.

The first way of using constructive non-confidence – to achieve a mid-term change in government – is the expected outcome of the resignation rule given that it was created to structure opposition behaviour so that it would be more “responsible.” As such, it is unsurprising that 10 of the 12 cases fall within this category. However, only five of the constructive non-confidence votes (two in Germany and three in Slovenia) were introduced with any expectation of success. Even in this category (1.a), success was not always achieved. In Germany, constructive non-confidence was narrowly defeated in 1972 and then successfully passed in 1982. The successful 1982 vote, however, did not result in a stable mid-term transition; it was quickly followed by early elections under the German safety valve provision. In this category, only the three Slovenian uses (1992, 2000, and 2013) actually achieved successful mid-term transitions resulting in governments that lasted between six and 15 months.9

The five constructive non-confidence votes in category 1.b of Table 1 – two Spanish votes (1980 and 1987) and three Polish votes (1997, 2012, and 2013) – all failed because they were triggered by parties that had no realistic expectations of success. In these cases, the confidence votes were used mainly to raise the public profile of the parties that called them.

Although constructive non-confidence was not originally conceived as a tool for removing only the leader of the governing party without actually deposing that party, Polish and Hungarian parties have successfully used the rule for this purpose in 1995 and 2009 respectfully. In both cases the rule was used by the parties that formed government to formally remove one prime minister and replace him or her with another from within their ranks, meaning that success was guaranteed.

Clearly, constructive non-confidence votes occur infrequently and result in real mid-term transitions even more infrequently. Astonishingly, only one of 12 votes (Slovenia in 2013) has resulted in a government lasting over a year. The kind of stable government transition originally contemplated by the German originators is very rare indeed under European rules of constructive non-confidence. This does not mean, however, that early elections do not occur. As we have seen in Germany, they sometimes occur under the safety valve provision.

Table 2 provides additional information on non-confidence votes in Europe situation placing both non-confidence votes and early elections in the context of the number of elections and cabinets.

Table 2 reiterates the rarity of constructive non-confidence votes, especially successful ones, in the six European regimes. This rarity is to be expected considering the barrier it places on members of parliament to remove a government – it is much harder to agree on who should form a new government compared to agreeing that the government must simply fall.12 At the same time, it also shows the instances in which early dissolutions occurred despite the constructive non-confidence rules. These early dissolutions, which occur under the various safety valve provisions, generally outnumber successful mid-term transitions (except for Hungary). For instance, the Spanish prime minister’s right to advise the king to dissolve the Congreso has resulted in the early dissolution of more than half of the Spanish parliaments since 1978. A similar situation exists in Belgium, where the legislature has made use of the safety valve to end three of five parliaments early due to the complex political context and difficulty with government formation. It is perhaps for this reason that Aucoin, Jarvis and Turnbull advocate the comparatively more difficult safety valve of a two-thirds vote of the Canadian House of Commons.

Should we have Confidence in Constructive Non-Confidence?

The evidence presented above indicates that constructive non-confidence has some powerful effects. Certainly, its infrequent use demonstrates that it is difficult to hold the government accountable by withdrawing confidence. There are high transaction costs for parties and their deputies as they need to find an absolute majority within the legislature to support a new government, not to mention come to an agreement about what party or parties will form the potential new government. In addition, the potential new government is an unknown factor, meaning that individual deputies might prefer to remain with the status quo – especially for government backbenchers.13 As a result, the ultimate accountability “weapon” is effectively tamed in the name of executive stability. Even when the constructive non-confidence regime includes a more relaxed safety valve, such as in Spain, constructive non-confidence delivers the increased parliamentary stability desired by its Canadian proponents.

For the Canadian reformers who advocate fixed parliamentary terms to enhance stability, constructive non-confidence and the safety valves of these European countries are not sufficient to combat the brinkmanship that has poisoned government-opposition relations in Canada. Aucoin, Jarvis and Turnbull propose a much tougher safety valve, requiring a two-thirds majority in the House of Commons. No longer would it be possible for the government to threaten an early election if defeated on legislation; nor could the opposition threaten to vote down legislation in hope of triggering an election. And, an early dissolution would most likely require the agreement of members from both the government and opposition parties.

The European evidence supports Aucoin, Jarvis and Turnbull’s assertion of diminished brinkmanship. De Winter finds that between 1945 and 1990, governments in Western European countries with negative confidence rules (of the kind that exist in Canada) were more than twice as likely to be defeated on a non-confidence motion as in countries with positive rules, that is, non-confidence votes that follow either constructive or absolute (requires a majority of the total number of parliamentary deputies) rules.14 Using De Winter’s data to single out constructive non-confidence regimes, I found that only 6.45 per cent of governments fell early under constructive non-confidence regimes (including dissolutions under safety valve provisions) compared to 18.35 per cent of governments removed by negative non-confidence votes. Table 2 confirms that stability of the parliamentary term and the resulting rarity of early elections. If Spain and Belgium are ignored – both have very weak safety valves – only 10.24 per cent of German, Hungarian, Polish, and Slovenian cabinets ended as a result of early elections, no mean feat considering all but one of the cabinets was either a majority coalition, minority coalition, or a single-party minority government. Including Spain and Belgium only increases this number to 21.47 per cent. There is little doubt, then, that constructive non-confidence does enhance stability during hung parliaments.

Germany illustrates and confirms the pattern found by De Winter: out of 16 parliamentary terms since 1949, only three have been dissolved early despite the fact that all but one of the country’s elections has resulted in a hung parliament (see Table 2). Diermeier, Eraslan and Merlo’s counterfactual analysis finds that if Germany had not included constructive non-confidence in the Basic Law, there would have been a 12 per cent reduction in the average length of the parliamentary term from 727 days to 637 days.15 Spain has had a less successful track record because its constructive non-confidence package does not bind the monarch from dissolving parliament early. Thus, out of ten parliamentary terms – seven of which were hung parliaments – six have been dissolved early. Gunther, Montero and Botella find that Spain scored well in terms of the length of the parliamentary term with the average being 42 months, the longest in Europe.16 Of a total of 40 Spanish and German governments, only 10 per cent fell due to a loss of confidence.

Clearly, Aucoin, Jarvis and Turnbull are correct in expecting enhanced parliamentary stability from constructive non-confidence. Moreover, their more difficult safety valve provision would ensure even more stability in the Canadian context than we have seen in the European regimes. But they expect more than just parliamentary stability from their constructive non-confidence proposal. In particular, they think constructive non-confidence would 1) legitimize mid-term governmental transitions, and 2) reduce executive dominance by empowering the House of Commons vis-à-vis the executive. Yet evidence from Germany and Spain, the most useful comparator cases for Canada because they are the most similar cases with federal and bicameral institutions and have the longest experience with the reform,17 indicates that the parliamentary stability Aucoin, Jarvis and Turnbull covet might actually exacerbate executive dominance and further delegitimize mid-term transitions. Canadian proponents of constructive non-confidence have overlooked how rare mid-term transitions are in constructive non-confidence regimes and the consequences of this infrequency. This finding should not be surprising when considering that Germany originally designed the resignation rule not to stabilize parliaments, but to enhance executive stability.

The Legitimacy of Mid-Term Government Transitions

Mid-term transitions are a traditional part of responsible government. Yet, the fact that they have been so infrequent in Canada has undermined their democratic legitimacy. Most Canadians believe they choose the next prime minister and government and that a government can only be displaced by new elections.18 Polling even shows that a majority of Canadians think the prime minister is directly elected.19 According to this logic, mid-term transitions are democratically illegitimate and a loss of confidence in the government should always result in an early election that allows the people, not parliament, to choose the next prime minister.

Constructive non-confidence is supposed to reverse this trend in public opinion by insisting that the consequence of a non-confidence vote is a government transition without new elections. However, the effect of the resignation rule in Europe has been somewhat different: mid-term transitions are as rare in Germany and Spain as they are in Canada. Of these two countries, only Germany has had a successful mid-term change of government, and a careful examination of this case highlights the difference between the entrenched constitution law and how it operates in practice.

Evidence from Germany illustrates that when mid-term transitions occur, the new government must seek an early election to gain full democratic legitimacy, that is though constructive non-confidence creates a constitutionally legitimate government, the demos views the mid-term transition as an usurpation of its ability to select the government. For example, in 1982, Helmut Kohl managed to win a constructive vote of non-confidence for the Christian Democrats (CDU) by convincing the Free Democrats (FDP) to leave their coalition with the Social Democrats (SPD). Despite this show of support, Kohl’s government contrived to lose confidence and new elections were called for March 6, 1983. In the ensuing constitutional debate, Kohl argued that his government was only temporary and a new election was necessary to sanction the change.20 All of the parties – even the SPD which had lost the 1982 constructive non-confidence vote – supported the early dissolution, meaning President Carstens had little issue acquiescing to the chancellor’s request.21 The President revealed in a public statement on January 7, 1983 that the FDP, as a condition for its support in the constructive non-confidence vote, expected that the mid-term transition would be followed by new elections. Accordingly, Kohl did not have a true majority but was chancellor “subject to proviso.”22

On the public opinion side, there was strong evidence that the people viewed the early election as necessary. Polls during the 1982 confidence crisis showed that a majority of Germans felt that any change of government should be accompanied by new elections.23 For example, a survey of 1,622 voters interviewed in November 1982 found that two-thirds of respondents were dissatisfied with the way that the government had changed hands and 58 percent thought that the FDP had committed treason by withdrawing support for the SPD, with which it had campaigned during the 1980 election.24 Since German voters had come to expect parties to publicly indicate what their coalition preferences are before the election, the majority of Germans believed they had the ability to choose the government.25 In fact, over the last few decades, federal elections in Germany have often been viewed as a Kanzlerwahl or “chancellor election,” meaning that the Bundestag only had to “ratify the decision of the electorate.”26 This has given German democracy a plebiscitary element that undermines the constitution’s parliamentary representative principles.27

These opinions were drawn upon by the Constitutional Court in the 1983 Bundestag Dissolution Case.28 The Court decided 6-2 against the position that premature dissolution was unconstitutional and should be overturned.29 It refused to nullify the President’s decision to dissolve the Bundestag because it determined that it had to take at face value the good faith of the President, Chancellor, and Bundestag.30 That did not stop the Court from disapproving of the executive’s use of Article 68 to circumvent the fixed parliamentary term.31 In a concurring opinion, Justice Zeidler went even further. He pointed out that the FDP’s behaviour had precipitated a crisis of legitimacy: the FDP had pledged to work in coalition with the SPD in the 1980 election so by entering into a coalition with the CDU-CSU, the FDP broke their pledge. Thus, Zeidler argued that the new coalition government had no choice but to return to the electorate for the sake of its own legitimacy.32

The Court also suggested that the infrequency of elections under the in Germany since 1949 had created a new convention. They found that the Basic Law created a representative democracy marked by general elections held at regular intervals. Therefore, the electorate expected that the government they “elected” would last for the duration of the parliamentary term and would not be replaced without new elections.33 With this argument, the Constitutional Court set the precedent for an early election after a constructive non-confidence vote.34

The convention identified by the Court removes parliament’s role in making government and primes voters to believe they have taken on that government-making responsibility. As a result, any mid-term transition must – despite their constitutional legality – be popularly approved or it is otherwise democratically illegitimate. Clearly, this convention modifies the Basic Law which legally allows mid-term transitions and does not require an early election. This has important implications for constructive non-confidence in Canada. The 1982 German case demonstrates that constitutional legality does not equate to political legitimacy and parallels the 2008 prorogation and coalition crisis in Canada. Because constructive non-confidence is so effective at stabilizing the executive against the opposition, even in a hung legislature, it actually supports the expectation that the people are electing a government (rather than a parliament that then chooses – and perhaps replaces – a government). That is, constructive non-confidence in Germany seems to have contributed to the very elections-based sense of democratic legitimacy that Canadian proponents hope it will counteract. The German experience provides little support for the hope that constructive non-confidence will alter the Canadian expectation that mid-term governmental transitions should be sanctioned by elections.

Enhanced Executive Dominance

The reform agenda advanced by Aucoin, Jarvis and Turnbull targets prime ministerial power. They seek to reduce executive dominance by removing the prerogative powers from the governor general, and thus, from the prime minister. Constructive non-confidence contributes to this goal by empowering the House of Commons to address government legislation, since non-confidence votes would be limited exclusively to constructive non-confidence votes.35 Although Aucoin, Jarvis and Turnbull pair their constitutional reforms with other institutional reforms, the European evidence provides little support for the hope that constructive non-confidence could decrease executive dominance. A close study of the resignation rule’s origins in Spain and Germany demonstrates the rule’s potential for enhancing executive power and indicates that executive dominance would likely continue in Canada after the reform program’s adoption.

The Spanish prime minister is among the most powerful in Europe. Spanish executive dominance comes from a number of factors, including a reliance on one-party governments, the d’Hondt electoral formula, which, like SMP systems, over-rewards the winning party, power over ministerial appointments, and a high degree of party discipline.36 As a result, the prime minister is much less reliant on coalition partners or support parties and has significant control over his or her own party. Constructive non-confidence adds another layer to the Spanish prime minister’s power because it insulates the office by making it harder for parliament to remove the government from office. It is these extra layers of protection that provide the Spanish prime minister with the strongest institutional powers in Western Europe.37 As a result of these institutional and party system powers, Lijphart assigns Spain one of the highest scores (higher than Canada’s) on his scale of executive dominance (see Table 3).38

The German situation is more complicated than Spain’s because the Bundestag was constructed as a working parliament with a far greater role in the legislative process.39 This is complemented by the institutionalized consensual nature of modern German politics. Since 1949, there has been hostility towards party conflict among both German elites and voters. Parties are hesitant to push political differences beyond a certain point and tend to seek inter-party accord.40 The separation of the office of party chairman and chancellor candidate is an example of the consensus-building and power-sharing tendencies. While it is possible for one person to hold both offices, this does not always occur and when a chancellor does not concurrently hold the party chairmanship, his or her position is relatively weaker as a result of being unable to control his parliamentary party.41

A German chancellor who does manage to hold both positions attains a very strong position that is only really checked by the political strength of the coalition partner. If the coalition partner is in a weak position, then there are even fewer checks. For example, Konrad Adenauer was able to chair the CDU and lead governments with weak coalition partners. His power was such that it led to the Federal Republic being characterized as a deviant form of parliamentary democracy entitled Kanzlerdemokratie (“chancellor democracy”).42 His chancellorship was the most powerful in Europe since 1945 – more powerful than the Spanish and British prime ministers.43 However, since then, no chancellor has been able to exert the same level of control and instead functions as the chief executive of policy and must oversee policy coordination between ministries and determine the general direction of government policy.44 This coordination role remains important for the chancellor who takes substantial public credit for making government work.45

All of this combines to give Germany a lower rank in Lijphart’s index of executive dominance (Table 3). Lijphart finds that between 1945 and 2010, the average cabinet duration in Germany was 3.80 years, significantly lower than Spain’s ranking.46

In Canada, constructive non-confidence is more likely to operate like the Spanish model rather than the German model. This is because Canada lacks Germany’s proportional electoral system and consensus-based politics, and shares Spain’s tendency towards single-party majoritarian governments. In fact, it is possible that the Canadian prime minister could become more powerful than his/her Spanish counterpart under constructive non-confidence. The Canadian prime minister can draw more authority from the Canada’s party system and high level of party control and discipline. According to Bergman et al., the British prime minister ranks higher than the Spanish prime minister in their party system and party cohesion ranking.47 Since the Canadian prime minister is in a similar position to the British prime minister adding constructive non-confidence would likely give the Canadian prime minister powers similar to the institutional powers of the Spanish prime minister. Under the Aucoin, Jarvis and Turnbull proposal, the Canadian prime minister would have slightly less institutional power because of the two-thirds dissolution rule. In Spain, the prime minister does have the virtual right of dissolution.

As noted above, coalition partners serve as a check on the German chancellor’s power. This is true to a lesser degree in Spain. However, when it comes to Canada, the fact that Canada has fewer competitive/effective parliamentary parties than Germany and Spain means that there are fewer possible coalition options. Between 1945 and 2010, Canada had on average 2.52 parties while Germany had 3.09 and Spain 2.66.48 Furthermore, the fact that coalition governments are more likely when parties are ideologically connected further reduces the practical coalition options in Canada and the possibility of replacing the incumbent government through a constructive vote of non-confidence.

If the constructive non-confidence package was adopted in Canada, it is possible that the Canadian party system will adapt in response to the dynamics of new institutional rules.49 This makes it difficult to predict what coalition options are possible, as does the fact that the Canadian parties – when compared to the international political spectrum – are in fact quite close on many issues. However, if the Canadian party system remains the same, Flanagan’s analysis of potential minimum connected winning coalitions helps illuminate why constructive non-confidence will restrict the possibility of replacing the incumbent government. He disregards the Bloc Québécois (BQ) because its anti-system agenda likely makes it an unacceptable coalition partner to other parties. Leaving aside the BQ, Flanagan finds that in the current Canadian party system there are only two possible ideologically connected coalitions: a Conservative-Liberal coalition and a Liberal-New Democratic Party (NDP) coalition. However, he discounts the first idea of a Liberal-Conservative coalition as both parties have led past governments and would have little incentive to enter into a grand coalition.50 Moreover, a grand coalition between major parties violates the minimum winning coalition principle, which states that parties will seek to form the smallest possible coalition needed to hold confidence, 51 and Canada, unlike Germany and its consensus-based politics, has no recent grand-coalition tradition. According to Flanagan, the second possible connected coalition of the Liberals and the NDP satisfies the minimum winning coalition principle necessary for forming a functioning government and is thus more likely.

In the present party system, the fact that the NDP tends to be ideologically to the left of the two traditionally larger parties limits the coalition options. As a result, the Conservatives have no practical coalition partner, which skews the political power struggle in favour of the Liberals during hung parliaments. Even if the Conservatives received the most seats and formed a minority government, the Liberals and the NDP could join forces and trigger a constructive non-confidence vote to remove the government from office. However, if the Liberals held a minority government, it would be virtually unimaginable that it would be removed by a constructive non-confidence vote compared to the current negative non-confidence rule. The Conservatives and the NDP might well form a “minimum winning coalition” in purely numerical terms, but they could not form a “minimum winning connected coalition.”52 True, under the Aucoin, Jarvis and Turnbull proposal the Conservatives and the NDP could defeat a Liberal government and trigger “early” elections if they could muster two-thirds support in the Commons, but it is unlikely that the two parties would hold the two-thirds of the seats needed.

This could leave the NDP as the perennial junior partner in Liberal-led alliances or coalitions. Since the NDP is to the left of the Liberals, the NDP would be severely disadvantaged because it would be unable to defect to join with the Conservatives (except in rare circumstances to trigger new elections under the two-thirds safety valve). As a result, the Liberals would have a significant advantage over the other parties and that constructive non-confidence initiated mid-term transitions would become even more difficult to achieve in Canada than in Germany or Spain where there are more coalition options or where the third party is a hinge party and located ideologically between the major parties (i.e. the German FDP).

In sum, while constructive non-confidence does rein in the prime minister’s control of the crown’s reserve powers, it is debatable whether it can reduce the overall phenomenon of executive dominance. In Germany, coalition government is a major check on the chancellor’s power and yet, as we have just seen, the development of coalition governments in Canada seems unlikely. It seems more likely that combining constructive non-confidence with the current incarnation of the Canadian party system will serve to limit the responsiveness of the government to parliament and thus to public opinion. According to Smith, anything that disconnects “[t]he non-confidence convention … from public opinion or the appraisal of public opinion by the political actors”53 should be avoided. In her view, such a disconnect would result if the non-confidence convention “were to be … completely formalized,” as it certainly would be under a constructive non-confidence requirement. In that case,

the system as a whole would lose a major inter-election link with the electorate. Specifically, the electorate would lose the energy and efforts of an ambitious opposition seeking government-defeating opportunities, and the energy and efforts of a tenacious government seeking parliamentary strategies and procedures designed to fit a fractured public opinion.54

For Smith, maintaining the fluidity of the convention is necessary to maintain the essential inter-election link between the government and the electorate. From this perspective, European-style constructive non-confidence curbs the flexibility of responsible government and makes the system less responsive for the opposition will be less able to remove the government from office.

The potential for constructive non-confidence to increase executive stability, delegitimize mid-term transitions, and increase executive dominance should cause significant concern for Canadians. It could fundamentally alter the practice of responsible government in Canada by codifying constitutional convention so that it is no longer incumbent upon citizens, but rather the constitution, to enforce the rules of the game. This is the reality of militant democracy: government is responsible to the people only when there is no threat to the regime. This is a perversion of responsible government; an overpowering of it by militant democracy.


Constructive non-confidence is a result of the desire to democratize the unwritten constitution by setting down the rules of the game in the formal constitution. However, adopting a reform package that includes constructive non-confidence could lead to institutional contradictions in the future. Constructive non-confidence comes from a very different type of parliamentary system that is based on consensus politics and a philosophy of militant democracy in response to past historical events. It is a poor fit for the Canadian system of responsible government that is premised on different principles. As such, constructive non-confidence will not operate in the same way as it does in Germany. It is thus both curious and regrettable that the Canadian proponents of constructive non-confidence have paid so little attention to how it has worked abroad and why it might work differently here. I have sought to cover these oversights by examining European uses of constructive non-confidence and how it might function in Canada. It is clear that constructive non-confidence does not deliver on all of its promises and would be a poor fit for Canada.

Constructive non-confidence would generate real change in three areas of concern to the reformers, just not always in the predicted manner. In fact, it is more likely to dash some of the reformers’ hopes than fulfill them. While parliamentary terms have certainly been stabilized in Europe, this has come at the cost of insulating the executive and increasing its dominance. Moreover, while mid-term transitions are clearly constitutionally legitimate, they have over time suffered a loss of democratic legitimacy. Thus, true mid-term transitions from an incumbent government to a lasting alternative administration are rare. The few mid-term transitions that do occur tend to be part of a strategy to engineer an early election. Constructive non-confidence’s ability to reduce prime ministerial power is, after considering the cases of Spain and Germany, unlikely in the long run. Indeed, greater executive stability and the democratic illegitimacy of mid-term transitions actually bolster executive dominance. Consequently, constructive non-confidence seems counterproductive to the hopes pinned upon it by its Canadian proponents and is likely to undermine key features of Canada’s system of responsible government.


  1. Peter Aucoin, Mark Jarvis and Lori Turnbull, Democratizing the Constitution: Reforming Responsible Government, (Toronto: Emond Montgomery Press, 2011), p. 119.
  2. It is worth noting that the constructive non-confidence package proposed by Aucoin, Jarvis and Turnbull makes the confidence relationship between the House of Commons and the government asymmetrical because only those seeking to remove the prime minister can initiate a test of confidence. It abolishes the ability for the government to ask the House of a vote of confidence, an option that remains in the German and Spanish constitutions, because of the concern that the government can use such a vote in “bad faith” as the German government did in 1972, 1982, and 2005 to engineer an early election. Such asymmetry is a unique constitutional arrangement compared to all of the current European constructive non-confidence regimes.
  3. For example, see the following works: Torbjörn Bergman, “Constitutional Design and Government Formation: The Expected Consequences of Negative Parliamentarism,” Scandinavian Political Studies (1993), Vol. 16, No. 4, pp. 285-304; Lieven De Winter, “The Role of Parliament in Government Formation and Resignation,” in Parliaments and Governments in Western Europe, ed. Herberts Döring (New York: St. Martin Press, 1995), pp. 115-151; John D. Huber, “The Vote of Confidence in Parliamentary Democracies,” The American Political Science Review (1996), Vol. 90, No. 2, pp. 269-282; Jonathan Boston, Governing Under Proportional Representation: Lessons from Europe (Wellington: Institute of Policy Studies, University of Wellington, 1998); Daniel Diermeier, Hulya Eraslan and Antonio Merlo, “Coalition Governments and Comparative Constitutional Design,” European Economic Review (2002), Vol. 46, No. 4, pp. 893-907; Torbjörn Bergman, Wolfgang C. Müller, Kaare Strøm and Magnus Blomgren, “Democratic Delegation and Accountability: Cross-National Patterns,” in Delegation and Accountability in Parliamentary Democracies, eds. Kaare Strøm, Wolfgang C. Müller and Torbjörn Bergman (Oxford: Oxford University Press, 2003), pp. 109-220; Björn Erik Rasch, Shane Martin and Jose Antonio Cheibub, eds., Parliaments and Government Formation: Unpacking Investiture Rules (Oxford: Oxford University Press, 2015.
  4. Studies in this area include: John Ford Golay, The Founding of the Federal Republic of Germany (Chicago: University of Chicago Press, 1958); Karlheinz Niclauss, Der Weg zum Grundgesetz: Demokratiegründung in Westdeutschland 1945-1949 (Paderborn: Uni-Taschenbuch, 1998); Michael Bernhard, Institutions and the Fate of Democracy: Germany and Poland in the Twentieth Century (Pittsburgh: University of Pittsburgh Press, 2005).
  5. Good examples of this area are the following: Karlheinz Niclauss, Kanzlerdemokratie: Regierungsführung von Konrad Adenauer bis Gerhard Schröder (Stuttgart: Uni-Taschenbuch, 2004); David Southern, “The Chancellor and the Constitution,” in Adenauer to Kohl: The Development of the German Chancellorship, ed. Stephen Padgett (Washington: Georgetown University Press, 1994), pp. 20-43.
  6. Markus Theil, “Germany,” in The “Militant Democracy” Principle in Modern Democracies, ed. Markus Theil (London: Ashgate, 2009), pp. 110-115.
  7. Golay, p. 128.
  8. Peter H. Merkl, The Origin of the West German Republic (New York: Oxford University Press, 1963), pp. 81-82.
  9. The new governments formed in 1992 and 2000 were cut short by constitutionally required elections
  10. A new cabinet is counted for every general election, change of party membership in cabinet, or the appointment of a new prime minister.
  11. This number includes the 1992 early election which was provided for in Slovenia’s post-independence constitution.
  12. Bergman, Müller, Strøm and Blomgren, p. 156.
  13. Ibid, p. 157.
  14. De Winter, p. 140.
  15. Diermeier, Eraslan and Merlo, p. 903.
  16. Richard Gunther, Jose Ramon Montero and Joan Botella, Democracy in Modern Spain (New Have: Yale University Press, 2005), p. 117.
  17. I focus on the German and Spanish experiences with constructive non-confidence based on most-similar case selection. Both are federal and bicameral states with strong regional parties. As well, Spain shares majoritarianism with Canada as both countries have electoral systems that over-reward the winning party.
  18. C. E. S. Franks, “To Prorogue or Not to Prorogue: Did the Governor General make the Right Decision?” in Parliamentary Democracy in Crisis, eds. Peter H. Russell and Lorne Sossin (Toronto: University of Toronto Press, 2009), p. 39.
  19. Peter H. Russell, “The Need for Agreement on Fundamental Conventions of Parliamentary Democracy,” National Journal of Constitutional Law (2009), Vol. 27, p. 207.
  20. R. E. M. Irving and W. E. Paterson, The Machtwechsel of 192-83: A Significant Landmark in the Political and Constitutional History of West Germany, Parliamentary Affairs (1983), Vol. 36, No. 4, p. 417.
  21. Donald P. Kommers, The Constitutional Jurisprudence of the Federal Republic of Germany (Durham: Duke University Press, 1989), p. 125.
  22. Southern, 1994: 30.
  23. Geoffrey Pridham, “Party Politics and Coalitions in Bonn,” The World Today (1983), Vol. 39, No. 1, p. 25.
  24. Max Kaase, “The West German General Election of 6 March 1983,” Electoral Studies (1983), Vol. 2, No. 2, p. 159.
  25. Kaase, p. 165.
  26. Southern, p. 27.
  27. Kaase, p. 165.
  28. The case was brought to the Constitutional Court by a number of former members of the dissolved Bundestag. They chose to exercise their constitutional right to challenge the President’s power to dissolve the Bundestag in an Organstreit proceeding in which the Federal Constitutional Court settles disagreements between different organs of government (Kommers, p. 124). The members argued that the Chancellor and President violated the constitution by dissolving the Bundestag because the successful constructive vote of non-confidence had clearly demonstrated that Kohl had the support of a parliamentary majority. Therefore, “the premature dissolution arose out of deception, and was thus unconstitutional” (Geoffrey Roberts, German Politics Today (Manchester: Manchester University Press, 2009), p. 222).
  29. Irving and Paterson, p. 418.
  30. Roberts, p. 222.
  31. The Court found that due to the presence of Article 39 in the Basic Law there was in fact no general power to dissolve the Bundestag in the mid-term, even with the use of constructive non-confidence (Southern, p. 31).
  32. Kommers, p. 128.
  33. Kommers, p. 128.
  34. Roberts, p. 222.
  35. Aucoin, Jarvis and Turnbull, p. 223.
  36. Gunther, Montero and Botella, pp. 116-118.
  37. Bergman, Müller, Strøm and Blomgren, p. 191.
  38. Arend Lijphart, Patterns of Democracy; Government Forms and Performance in Thirty-Six Countries, 2nd ed (New Haven: Yale University Press, 2012), pp. 120-121.
  39. Thomas Saalfeld, “The German Bundestag: Influence and Accountability in a Complex Environment,” in Parliaments and Governments in Western Europe, ed. Philip Norton (London: Frank Cass & Co., 1998), p. 53.
  40. Southern, p. 38.
  41. Padgett, pp. 49-50.
  42. Padgett, p. 18.
  43. Bergman, Müller, Strøm and Blomgren, p. 191.
  44. Southern, p. 79.
  45. Niclauss, 2000, pp. 70-71.
  46. Lijphart, p. 120.
  47. Bergman, Müller, Strøm and Blomgren, pp. 191-192.
  48. Lijphart, pp. 74-76.
  49. It is worth noting that there are other possible institutional changes that could greatly affect the Canadian party system, particularly electoral reform. The currently Liberal government is in the process of considering the possibility of adopting a form of proportional representation. It is likely that such a change would benefit smaller parties that are unable to win as many seats in the House of Commons as their share of the popular vote. An example would be the Green Party of Canada, which currently only holds 1 seat and does not qualify for official party status. Proportional representation may also encourage the growth of new parties.
  50. Thomas Flanagan, “A Canadian Approach to Power Sharing,” Policy Options (2010), Vol. 31, No. 8, p. 33.
  51. Flanagan, p. 33.
  52. Flanagan, p. 33.
  53. Jennifer Smith, “Responsible Government and Democracy,” In Taking Stock of 150 Years of Responsible Government in Canada, eds. F. Leslie Seidle and Louis Massicotte (Ottawa: Canadian Study of Parliament Group, 1999), p. 42.
  54. Smith, p. 42.

Canadian Parliamentary Review Cover
Vol 39 no 3

Last Updated: 2020-09-14