The Constitutional Paradox in Quebec: A Research Note.∗ Jean-Franc¸ois Godbout Center for the Study of Democratic Politics Princeton University D´epartement de science politique Universit´e de Montr´eal March 15, 2014

1

Introduction

In the following research note, I demonstrate that the preferences of the population of Quebec over the different constitutional alternatives debated in the last thirty years constitute a Condorcet voting paradox. Since the 1980 referendum on the Sovereignty of Quebec, there has been three principal constitutional options in this province: sovereignty, renewed federalism, and the status quo. Each of these views has been considered in a province wide vote of some kind: the 1980 referendum on sovereignty under Ren´e L´evesque, the 1992 referendum on the Charlottetown Accord under Robert Bourassa, and the 1995 referendum on ∗ Paper

presented for the “Atelier de la Soci´et´e qu´eb´ecoise de science politique: Retour sur l’incertitude dans les d´ecisions constitutionnelles”, Ottawa, May 23, 2012.

1

sovereignty under Jacques Parizeau. In this study, I offer some evidence to show that the preferences of Quebec voters over these three options constitute a Condorcet Paradox. I also present a theoretical and empirical demonstration to support this claim. The finding presented in this study is associated with a fundamental dimension of social choice theory. Indeed, the Condorcet voting paradox challenges the traditional theory of democracy by saying that no voting rule will lead to ‘the will of the majority’ (Buchanan and Tullock, 1965, p.334). This conclusion carries important implications in Canada because, since the 1992 referendum, it seems highly likely that any future attempt to modify the Canadian Constitution will have to be ratified by the population in a nationwide referendum (Heard and Swartz, 1997). The research note is organized as follow. In the first section, I provide a short historical background to the constitutional saga in Canada. In the next section, I outline the broad theoretical argument. In the third section, I offer some empirical evidence. In the final section, I conclude.

2

Historical Background

Although Canada adopted a new constitution with an amending formula and a Charter of Rights and Freedoms in 1982, this document was never formally ratified by the National Assembly of the province of Quebec. Because the Canadian Constitution was originally an Act of the British Parliament, several attempts have been made over the years to “canadianize” the constitution to allow the fed-

2

eral government and the provinces to modify its content. However, prior to 1982, all of the previous attempts to repatriate the constitution failed, mostly over the objections of Quebec, who held by convention a veto to prevent any unilateral modifications of Constitutional Act of 1867. Quebec’s main opposition with most constitutional changes can be explained by federalism and the separation of powers between the provinces and the central government. Quebec’s position has always been to secure more autonomy inside the federation, but also to conserve a formal veto power to amend certain parts of the constitution. It was the election of the Parti Qu´eb´ecois (PQ) in 1976, and the subsequent referendum on sovereignty in 1980, that precipitated the repatriation of the Constitution from England in 1981. The then Liberal Prime Minister Pierre Trudeau, had indicated during the referendum campaign that a “No” vote would not be a vote for the “status quo”, but rather a vote in favor of renewed federalism. Thus, changes were promised to Quebec, but also for the other provinces of Canada. The referendum results gave the federalist camp a comfortable majority of 60%, however the Liberal government of Pierre Trudeau had to deliver on its promise. The results of the next round of negotiations was the 1981 constitutional agreement between the federal government and the canadian provinces. However, the ratification of the new constitution occurred without the support of Quebec, who for the first time lost its veto power in constitutional matters (?). Subsequent attempts were made by the Conservative government of Brian Mulroney to bring Quebec back into the constitutional fold. The Meech Lake and later Charlottetown Accords saw the recognition of Quebec as a distinct so3

ciety, and the acknowledgement of the province’s veto power on certain constitutional issues. However, because of the newly adopted amending procedure, both of these proposals required the unanimous support of the Canadian provinces to be ratified. The Meech Lake Accord failed to be ratified over the objections of Manitoba, whereas the Charlottetown Accord was submitted to a pan-canadian referendum, and failed to be approved in seven of the ten provinces, including Quebec. The last chapter of Quebec’s constitutional saga occurred three years later, in the 1995 referendum on sovereignty. In this ultimate attempt to modify unilaterally the Canadian constitution, Quebec voters rejected the Parti Quebecois’ sovereignty-association proposition in a referendum by less than 1%, in a 51%-49% vote. I will argue in the next section that the referendum outcome was the results of strategic voting on the part of pro-independence voters in the 1992 referendum. Unlike Imbeau (1991a,b); ?); ?, 1986), I do not focus on the preferences of constitutional actors—such as provincial premiers and the Prime Minister—during negotiations to explain the outcome of the Meech Lake or the 1981 agreements. Rather, I concentrate on the preferences of the Quebec population of voters to explain the outcome of constitutional referendums.

3

The Claim

I begin by providing a definition of a Condorcet voting paradox. A Condorcet paradox—also known as the paradox of voting—can be explained as follow: Sup-

4

Table 1: Condorcet Paradox of Voting. Actors 1 2 3

Preferences x>y>z y>z>x z>x>y

pose three actors 1, 2, 3 have transitive preferences over three alternatives, x, y, z. That is, if actor 1 prefers alternative x to y, and y to z, then it must also be true that actor 1 prefers alternative x to z. Suppose also that the actors have to choose among these three alternatives using simple majority rule. Given the preference ordering found in table 3, we can see that by arranging the different social choices: x is preferred to y in a 2-1 vote; z is preferred to x in a 2-1 vote; and y is preferred to z in a 2-1 vote. As Riker (1982) explain, “If one tried to arrange the outcome of voting in a sequence of ‘social preference’ one would not be able to do so because one could not say wether x or y or z stood first. Any one of [the previous] arrangements would be possible [...]. (p.18)” Arrow (1951) demonstrates that this cyclical majoritarian result can be generalized to any electoral system (e.g., plurality, majority, proportional) and for any population size. In the remaining of this note, I will argue that this finding applies to the preferences of the population of Quebec in the debates over the constitutional future of the province.

5

3.1

Quebec Paradox

The following demonstration rests on several assumptions. We will begin by considering the preferences of the French speaking population of Quebec. We will relax this assumption later in the analysis. The second assumption is that the actors are rational, and vote for their most preferred policy outcomes in a referendum. However, actors can vote strategically as well. Finally, the third assumption is that the sequence of votes over the different constitutional options occurred in the following order: first, in 1980 there was a referendum between Quebec sovereignty and the 1867 Constitution; second in 1992, there was a referendum between renewed federalism (the Charlottetown Accord) and the 1982 Constitution; third in 1995, there was a referendum between Quebec sovereignty and the 1982 Constitution. I begin my demonstration by identifying three different constitutional options which have been offered (or imposed) at one point or another to Quebec since 1982. The first option, which I call the Status quo, is the Constitutional Act of 1982 supported by the federal government and all Canadian provinces, except Quebec. The second option corresponds to the Meech Lake-Charlottetown proposals that recognized Quebec as a distinct society and provided the province with additional constitutional powers. I call this alternative the Quebec clause. Finally, the third option is associated with Quebec sovereignty, or a proposal to make Quebec an independent country. I call this alternative Sovereignty. To demonstrate the paradox, we need to assume further that the population of Quebec has transitive preferences over these three alternatives. In other words, 6

Table 2: Preferences over Constitutional Alternatives in Quebec. Actors 1 2 3 4 5 6

Preferences Sov > SQ > Clause SQ > Clause > Sov Clause > Sov > SQ SQ > Sov > Clause Sov > Clause > SQ Clause > SQ > Sov

a citizen who prefers the Status quo to the Quebec clause, will also prefer the Status quo to Sovereignty. There are of course several other possible ordering of preferences over these three options. In fact, we count 3! potential orderings. The different alternatives and the preferences outlined above are summarized in table 3.1 The reader will notice that the preceding ordering of preferences do not constitutes a Condorcet paradox as defined in the preceding section. In order to observe such an outcome, the preferences over the three constitutional options would need to be limited to sequences 1-2-3 or 4-5-6. Let us assume for now that there are six equally divided groups of francophone citizens in Quebec with the ranking of preferences reported in table 3. In this context, if the population had to vote between Sovereignty and the Status quo, the election would result in a tie, with 50% of the population preferring Sovereignty against the remaining 50% in favor of the Status quo. The reader can observe from the table that a vote between the two remaining options, Sovereignty versus Quebec clause or Quebec clause versus the 7

Status quo would also result in a tie. Of course, observing an equal distribution of voters across these six ranking of preferences is an unrealistic assumption. It is unlikely that 17% of the population of Quebec prefers the Status quo over Sovereignty, and Sovereignty over the Quebec clause (ranking 4). This last ranking may seem plausible in other parts of the country, like in the province of Alberta. Indeed, this is the position of Preston Manning and the Reform Party in the 1992 Charlottetown referendum. However, finding this ranking in Quebec seems highly unlikely—especially among the French speaking population. On the other hand, the other remaining sets of preferences seem plausible in the Quebec population. For instance, we can think that for some people, the Status quo is preferred to the Quebec clause, and that the Quebec clause is preferred to Sovereignty (ranking 2). Similarly, for others, the Quebec clause might be preferred to Sovereignty, and Sovereignty might be preferred to the Status quo (ranking 3). Now that we have identified the potential ranking of preferences over the three constitutional options in Quebec, it is necessary to have a better set of assumptions about the distribution these preferences in the population. This is necessary to understand the different constitutional outcomes that could be reached in a referendum. I begin with a very simple case. I identify three groups of voters (which I treat as unitary actors for now) who have preferences over alternatives 1-2-3. I return to alternatives 4-5-6 below. The first group corresponds to federalist voters. I postulate that federalist voters prefer the Status quo to the Quebec clause. If federalist voters had to decide be8

tween both alternatives in a referendum, they would support the Status quo or the 1982 Constitution. I also assume that federalist voters prefer the Quebec clause to Sovereignty. In this context, if the federalists had to decide between these two constitutional options, they would support the Quebec clause. For example, this group could be associated with the preferences of former Liberal Prime Minister Jean Chr/’etien. The second group corresponds to nationalist voters. Nationalist voters prefer the Quebec Clause to Sovereignty. They also prefer Sovereignty to the Status quo. It makes sense to assume that for an important proportion of voters in Quebec, independence was preferable to the 1982 Constitution—especially after the Quebec veto was rejected by the rest of Canada. Indeed, following the failure of the Meech Lake Accord in 1989, more than 65% of respondents polled in the province declared being in favor of Quebec’s independence ?. I include in this group the Liberal Premier Robert Bourassa. This was the position he outlined in a speech after the failure of Meech. The third group of voters represents sovereigntists who prefer sovereignty to the Status quo. However, and here lies the critical point of my argument, sovereigntists also prefer the Status quo to the Quebec clause alternative. This preference ordering is supported by two facts. First, a majority of Quebecois voted against the 1992 Charlottetown referendum—which implicitly recognized Quebec as a distinct society. And second, the Parti Quebecois publicly opposed this agreement in the 1992 referendum campaign. In fact, since this agreement was rejected by 54% of the population of Quebec, an important proportion of sovereigntists had to vote against the proposed constitutional modifi9

Table 3: Condorcet Paradox in Quebec I. Actors Sovereigntists Federalists Nationalists

Preferences Sov > SQ > Clause SQ > Clause > Sov Clause > Sov > SQ

cations. It is quite possible that sovereigntists reasoned that if they accepted the 1992 Accord and the Quebec clause, they would have to forfeit their objective of forming a new nation. This is precisely the strategy of the No side in the 1992 referendum. Parizeau, the leader of the Parti Quebecois, even went on record by stating that a No vote in 1992 was just the first step for accelerating plans to quit the Canadian federation (i.e., the Lobster trap strategy). He declared that a subsequent referendum on sovereignty would be held if the Parti Quebecois won the next election (New York Times, October 27, 1992). As we can see in table 3.1, given the preceding preferences, voting over the three options could constitute a Condorcet paradox. Each pair of alternatives has a winner and a loser. However, the winner of any pair of alternative will be defeated by the remaining third alternative. For example, if the population of Quebec had to vote in a referendum between two options—the Quebec clause and the Status quo—the federalists and the sovereigntists would support the Status quo, while the nationalists would vote for the Quebec clause. This is what occurred in the 1992 Charlottetown referendum. On the other hand, if there was a referendum on Quebec independence, like in 1995, the sovereignty option would win because 10

both nationalists and sovereigntists voters would support this option. Of course, the 1995 results do not confirm this claim since the No side (or the Status quo) won. However, the outcome of this referendum was virtually tied (49.42% vs. 50.58%) with a 94% level of participation. We could further argue that given the closeness of the results, this outcome could just have been as easily reversed. Furthermore, since more than 60% of the Quebec francophone population voted in favor of Independence in the 1995 referendum, we can conclude that the predicted outcome applies to this group at least. It is important to understand here that I am not arguing that the French population have more political rights, just that the paradox works within this subsample of the population. Lastly, when we compare the preferences of the Quebec population over Sovereignty or the Quebec clause, the table shows that federalists and nationalists would have voted in favor of the second option. Hence, in each of these three referendum scenarios there is no clear winner. The cycle of voting shows no equilibrium outcome. Of course, the previous sequence of preferences must necessarily exclude rankings 4-5-6 from the preferences of the population of Quebec. Recall that these three ordering could also constitute a Condorcet paradox. However, I have explained that it is highly unlikely that we will find voters in Quebec who prefer the Status quo over Sovereignty, but who also prefer Sovereignty over the Quebec clause (ranking 4). Thus, we can remove this sequence from table3.1. That being said, the preferences observed in rankings 5 and 6 are a definite possibility among Quebec voters. In ranking 5, we can think that some nationalists voters would prefer Sovereignty over the Quebec clause, and if the choice was between the Quebec 11

clause and the Status Quo, they would choose the former. Similarly, some provincialists (Imbeau, 1986) might prefer the Quebec clause over the Status quo, and the Status quo to Sovereignty (ranking 6). Now, if we assume that the population of Quebec is equally distributed among options 1-2-3-5-6 (20% of the voters in each categories), then the Condorcet paradox disappears. Under this scenario, a referendum between the Quebec clause and the Status quo would make the Quebec clause a winner (60%-40%). An alternative vote between the Quebec clause and Sovereignty would also make the same option a winner (60%-40%). Finally, a vote between the Status quo and Sovereignty would make this last option the winner (60%-40%). Knowing this, a vote between the Quebec clause and the Status quo—like in the 1992 referendum–should have seen the Status quo or the 1982 Constitution defeated. However, this is not what happened. Indeed, we can think in this context that a lot of sovereigntists voters who preferred the Quebec clause (ranking 5) heard their leader Jacques Parizeau and voted strategically for the Status quo option in this referendum (thus shifting their preference to ranking 1 (Sov > SQ > Clause). Given the high support over Quebec independence following the failure of the Meech Lake Accord in public opinion polls, Parizeau knew that this option could win. In fact, the Liberal party’s own Allaire Report and the National Assembly’s B´elanger-Campeau Commission both recommended a referendum on sovereignty after the failure of Meech in 1990. However, this options was never seriously considered. Bourassa chose instead to hold a referendum on the Charlottetown agreement, which would have given Quebec a “distinct” soci12

Table 4: Condorcet Paradox in Quebec II. Actors Sovereigntists Sovereigntists (strategic) Federalists Nationalists Provincialists

Preferences Ind > SQ > Clause Ind > SQ > Clause SQ > Clause > Ind Clause > Ind > SQ Clause > SQ > Ind

ety status in the federation and a veto power. Finally, what about the voters who preferred ranking number 6 (Clause > SQ > Ind)? This group should have voted yes in the 1992 referendum and no in the 1995 referendum. Hence, there was no need for these voters to be strategic. Assuming once again an equal distribution of preference of Quebec voters, but now by switching the ordering of group 5 (Sov > SQ > Clause) to the ordering of members of group 1 (Sov > Clause > SQ), we can see from table 3.1 that the Status quo would have won in the 1992 referendum (60%-40%), and Sovereignty would have won in the 1995 referendum (60%-40%). This last results corresponds roughly to the distribution of francophone votes in the 1995 referendum (Drouilly, 1997). In addition, the 60%-40% No results observed on the SQ vs. Clause vote also corresponds to the distribution of francophone preferences at the time of the Charlottetown referendum (Leduc and Pammett, 1995; Johnston et al., 1996). In short, we find that we are once again presented with a Condorcet voting paradox. Given this ordering of preferences, no option is a clear winner.

13

Table 5: Condorcet Paradox in Quebec III. Actors Sovereigntists Federalists Nationalists Provincialists

Preferences Sov > SQ > Clause SQ > Clause > Sov Clause > Sov > SQ Clause > SQ > Sov

% of Voters 40% 20% 20% 20%

It is important to note that several other distribution of preferences over these five rankings could also generate a Condorcet voting cycle. For example, the reader can verify that a distribution of 40% of sovereigntists voters with 20% of federalists, nationalists, and provincialists could also generate a voting cycle (see table 3.1).

3.2

Status Quo

The implications of the preceding analysis is that we should be in a cycle of constitutional instability. Yet, it seems that we have reached in recent years a stable equilibrium in Canadian politics. The independence movement appears to be in decline in Quebec, the Bloc Quebecois has lost more than 90% of its seats in the 2011 federal election, and the Parti Quebecois remains in the opposition. However, this state of affair could very well prove to be temporary, especially given the renewed interest in reforming the Senate, and the proposed modification in the representation of the provinces in the House of Commons. Both of these changes would require constitutional amendments. Thus, the likelihood of another round 14

of constitutional negotiations between the provinces and the federal government remains a definite possibility in the near future. However, we know, based on the previous analysis, that the outcome of any future negotiations would in no way produce a stable resolution to the constitutional conflict in Quebec. This is explained by the fact that there is no clear equilibrium outcome in the debate surrounding the modification or the ratification of the 1982 Constitution. The same is true with regards to the sovereignty option. There will always be a majority of voters in Quebec who will prefer another constitutional option over sovereignty (in this case the Quebec clause which would recognize that the province is composed of a distinct society). Thus the paradox of voting over constitutional issues continues. Perhaps the fact that the 1982 Constitutional agreement was never ratified by the Quebec assembly help explain this paradox. The reason is simple, a majority of the population of Quebec preferred the 1867 Constitutional status quo to the proposed reforms of the 1982 package. Although the independence question was off the table at the time (following the 1980 referendum), the alternatives during this period were very similar to the one described above. The only exception being that the sovereigntists preferred to remain under the 1867 agreement instead of accepting a renewed federal proposal. This important change is described in table 3.2 below: From this table we can see that although it was not preferred by everyone, the 1867 Constitution was in fact a Condorcet winner and an equilibrium outcome for

15

Table 6: Condorcet Winner in Quebec. Actors Preferences Sovereigntists Sov > 1867 > 1982 Federalists 1982 > 1867 > Sov Nationalists 1867 > Sov > 1982

Quebec. As long as this province had a veto power over any modifications of the constitution (as was the case before the Supreme court’s decision on the question of unilateral patriation), it was impossible to depart from this position. In fact, it took more than 50 years for the federal government and the provinces to reach an agreement, and this had to be done without the consent of Quebec. Furthermore, the outcome of the 1980 referendum clearly shows that the Status quo of 1867 was preferred by a majority of voters in Quebec. Still, if a referendum over the 1982 proposal had been held in the province, I believe that a majority of the Quebec population would have voted against it.

4

The Empirical Evidence

The preceding arguments rest on several assumptions made about the preferences of the electorate in Quebec. A recent CROP survey on political attitudes in Quebec serves to illustrate this paradox (?). More than 30% of the French speaking respondents indicated in this survey that the best option for Quebec was to withdraw from the Canadian federation, 37% declared that the 1982 constitution should be

16

Table 7: Distribution of Constitutional Preferences in Quebec, 1992 CES.

1992 Referendum Yes No

1980 Referendum Yes No 10% (12) 60% (53) 90% (104) 40% (36)

1992 Referendum Yes No

1989 Meech Favour Oppose 59% (100) 26% (72) 42% (69) 74% (208)

1989 Meech Favour Oppose

1980 Referendum Yes No 32% (68) 63% (114) 68% (140) 37% (66)

Note: The numbers are drawn from the 1992 Canadian Election Studies survey.

amended to become acceptable to the Quebec government, and 34% declared that Quebec should remain in Canada, regardless of a constitutional change or not. Of course, in order to directly test whether a Condorcet voting paradox exists over these options, we would need to have a ranking of preferences over the three options. To my knowledge, no public opinion survey has ever asked this question. However, we can partially verify the claim that the sovereigntists voters were strategic in the 1992 Charlottetown Referendum by looking at the 1992 Canadian Election Study survey. Table 4 provides clear evidence that voters who voted Yes in the 1980 referendum also voted No in the 1992 referendum. This is true for over 90% of the sovereigntists. On the other hand, we also find that the same Yes 17

voters opposed the Meech Lake Constitutional Accord in 1989. However, this is true for only 68% of the sovereigntists. In this context, Meech was preferred over Charlottetown by sovereigntists. However, recall that the prospect for a new referendum was not as clear in the event that Meech failed. Thus, the incentive to be strategic may not have been as high then.

5

Conclusion

I have argued in this research note that the constitutional preferences of French speaking voters in Quebec have a high likelihood of constituting a Condorcet voting paradox. I have provided empirical evidence to show that in the 1992 referendum, sovereigntists voters who may have preferred the content of the Charlottetown accord, voted strategically against this constitutional agreement because they anticipated a subsequent referendum on secession. Indeed, a leading newspaper reported that on referendum night, “Jacques Parizeau, who leads the separatist Parti Quebecois, told cheering followers in Montreal: ‘We have tonight said what we don’t want. Next time we will say what we want.’ (New York Times, October 27, 1992)” Of course, the occurrence of a Condorcet paradox is conditional on the distribution of voter preferences in the electorate. As we have seen, several scenarios are possible, but we have yet to find a survey that precisely identify the preference ordering of the three main options (sovereignty, renewed federalism, status quo) in Quebec. Nonetheless, we have shown some strong evidence— especially with

18

regards to the Francophone voting population of Quebec—that there was a 60-40 split in the electorate over the options in the 1992 and 1995 referendums. Since there has not been a referendum between Renewed federalism and Independence to date, it is impossible to determine wether this outcome would have generated a 60-40 split in favor of federalism. This last point brings us to the potential strategic manipulation of the referendum agenda that could be used to favor one option over another. As we have seen, the Parti Quebecois has urged voters to oppose the 1992 Charlottetown agreement even if it represented a better option for most of the French speaking population of Quebec. In an era where major constitutional modifications need to be ratified by the Canadian electorate, identifying this type of strategic behavior is important. The individual may not find it “advantageous to vote for [constitutional] rules that may promote sectional, class, or group interests because [...] he is unable to predict the role that he will be playing in the actual collective decision-making process at any particular time in the future. (Buchanan and Tullock, 1965, p.78).” However, he or she can be easily manipulated into doing so.

References Arrow, K. J. 1951. Social Choice and Individual Values. Second edition ed. United States of America: Yale University Press. Buchanan, J. M. and G. Tullock. 1965. The Calculus of Consent: Logical Foundations of Constitutional Democracy. 2001 ed. United States of America: The University of Michigan Press.

19

Drouilly, P. 1997. Le Referendum du 30 octobre 1995: une analyse des resultats. In LAnn´ee politique au Qu´ebec, 1995-1996, ed. R. Boily. Montr´eal: Fides pp. 119–143. Heard, A. and T. Swartz. 1997. “The Regional Veto Formula and its Effects on Canada’s Constitutional Amendment Process.” Canadian Journal of Political Science 30(2):339–356. Imbeau, L. 1991a. Le compromis est-il encore possible? La n´egociation de l’apr`es Meech a` la lumi`ere de la th´orie des jeux. In Le Qu´ebec et la restructuration du Canada: 1980-1992, ed. L. Balthazar ans G. Laforest and V. Lemieux. Sillery, Qubec: Les e´ ditions du Septentrion. Imbeau, L.-M. 1986. “Why didn’t all Dissentig Provinces Dissent un 1981?: A Qualitative Voting Game Analysis of the 1981 Constitutional Decision in Canada.”. Imbeau, L.-M. 1991b. “Proc´edures et prise de d´ecision collective: une tude de la d´ecision constitutionnelle de novembre 1981 au Canada.”. Johnston, R., A. Blais, E. Gidengil and N. Nevitte. 1996. The Challenge of Direct Democracy: the 1992 Canadian Referendum. Quebec: McGill-Queen’s Press. Leduc, L. and J.H. Pammett. 1995. “Referendum Voting: Attitudes and Behaviour in the 1992 Constitutional Referendum.” Canadian Journal of Political Science 28(1):3–33. Riker, W. H. 1982. Liberalism Against Populism: A Confrontation between the Theory of Democracy and the Theory of Social Choice. 1988 ed. United States of America: Waveland Press.

20