Economic voting has been examined for more than fifty years in Canada and the economy is probably the short-term factor of which the impact in Canadian elections has been studied most (Anderson, Reference Anderson2008, Reference Anderson and Stephenson2010; Clarke and Kornberg, Reference Clark and Kornberg1992; Gélineau and Bélanger, Reference Gélineau and Bélanger2005; Nadeau and Blais, Reference Nadeau and Blais1993, Reference Nadeau and Blais1995; Nadeau et al. Reference Nadeau, Blais, Nevitte and Gidengil2000). Canada, however, is not a homogeneous country and the presence of different regional party systems renders it possible that the economic vote varies from region to region. Previous research has indeed suggested that the impact of the economy on vote choice is different in Quebec than what holds in the other provinces. Guérin and Nadeau (Reference Guérin and Nadeau1998), for example, have argued that francophones in Quebec did not vote according to their judgment of the government's economic performance during the 1972–1980 period. Their explanation for this observation was that Quebeckers’ loyalty to the Liberal party inhibited them from voting economically. Consequently, during this time-period, voters in Quebec did not punish the Liberal party for worsening economic conditions. However, from 1982 onwards—when the Liberals repatriated the Constitution—Quebeckers’ votes were affected by their evaluations of the state of the economy. The implication of the findings of Guérin and Nadeau (Reference Guérin and Nadeau1998) is that the structure of the party system has a strong impact on whether or not the economy affects voting behaviour in Quebec. Furthermore, Godbout and Bélanger (Reference Godbout and Bélanger2002) have demonstrated that the rest of Canada is not homogenous either and that economic voting varies between regions depending on the economic structure of a province.
Guérin and Nadeau (Reference Guérin and Nadeau1998) link differences between Quebec and the rest of Canada as well as the change in the strength of the economic vote in Quebec over time to changes in the party system. In this respect, it is surprising that there is no systematic analysis of economic voting comparing Quebec and the rest of Canada after 1993.Footnote 1 This is an important limitation, because 1993 is one of the most important turning points in the history of Canadian party systems. The 1993 electoral earthquake exacerbated regionalism in Canadian politics. This high level of regionalism was somewhat reduced when different rightist parties merged into the Conservative Party of Canada in 2003, but the strong presence of the Bloc Québécois in Quebec implies that there still was a high level of regionalization in the Canadian party system after 2003 as well.
In this paper, we provide an extensive analysis of economic voting in Canada during the 1993–2008 period. We compare the strength of the economic vote in Quebec with its strength in the rest of Canada. Furthermore, we offer more insights into the individual-level mechanisms that could explain how regionalism is linked to economic voting as we examine how the intensity of preferences for sovereignty and their impact on the vote moderate the extent to which economic evaluations affect vote choice. We answer two important research questions: Does the dominance of the Bloc Québécois in Quebec inhibit economic voting in the same way as was the case during the dominance of the Liberal party? For voters in Quebec, do attitudes on the issue of independence moderate the relationship between economic considerations and vote choice?
Using the Canadian Election Studies (CES) data from 1993 to 2008, we demonstrate that the economy is not a significantly weaker determinant of vote choice in Quebec than what can be observed for the rest of Canada. In the absence of an overall weaker economic vote in Quebec, we additionally examine the heterogeneity of economic voting within Quebec. That is, we stipulate that only for voters who have a strong stand on independence will economic voting be weakened. The rationale is that those with a strong position on this issue will vote for or against the Bloc Québécois, no matter how they perceive the economic situation and no matter who is the ruling incumbent. However, our results do not offer support for this individual-level hypothesis either. We conclude by emphasizing that nationalism and high levels of regionalism do not limit economic accountability. The economic vote, it seems, is more stable and general than it is regularly argued to be.
These results are important for a number of reasons. First, they are an important nuance to the conventional wisdom that voters in Quebec are different from voters in the rest of Canada and that different factors explain the choices of Quebeckers. Our results show that voters in Quebec are not all that different from voters in the rest of the country when it comes to economic voting. The implication is that analyses on pooled samples of all Canadians will perform quite well and that researchers need not worry too much about different subgroups of voters. Second, our results have important implications beyond the Canadian case. In showing such small differences in the strength of economic voting between Quebec and the rest of Canada, our results speak to the salience and stability of the economic vote in general. Previous work has already suggested that institutional differences between political contexts do not strongly affect the strength of the economic vote (Dassonneville and Lewis-Beck, Reference Dassonneville and Lewis-Beck2017), and our results could be taken to suggest that even in countries where the political space is multi-dimensional and where regionalism is an important dimension, the economic vote is still present and about equally as strong as what holds for other contexts.
Regionalization, Party Systems and Economic Voting
One of the main features of Canadian politics, if not the main one, is its level of regionalisation (Cairns, Reference Cairns1968), referring to the fact that some regions have their own micro partisan system within a macro national partisan system. The 1993 electoral earthquake election has strongly increased the level of regionalism in Canadian politics. The incumbent Progressive Conservative government, led by Brian Mulroney, was re-elected with a majority of seats in the 1988 election (for an analysis of this election, see Johnston et al., Reference Johnston, Blais, Brady and Crête1992). Five years later, however, after the Charlottetown referendum on constitutional change had failed, the party was nearly decimated. Indeed, in the 1993 elections, the Progressive Conservative coalition collapsed across Canada.
After the 1993 elections, Progressive Conservatives’ representation in Parliament went down from 169 to only two seats. The party's virtual disappearance gave rise to a high level of regional differences, with Ontario and the Atlantic provinces voting for the Liberals and the West supporting the new Reform party (that eventually became the Canadian Alliance). However, the most important source of change came from Quebec, where a new separatist party, the Bloc Québécois, won a majority of seats in the province and was the Official Opposition at the national level between 1993 and 1997. In fact, the Bloc Québécois maintained the strength it gained in the 1993 election until the rise of the NDP in 2011 (Fournier et al., Reference Fournier, Cutler, Soroka, Stolle and Bélanger2013; Laycock and Erickson, Reference Laycock and Erickson2015).
In a number of ways, the dominance of the Bloc Québécois in Quebec after the 1993 elections is similar to the position of the Liberals between 1972 and 1980, the period when economic voting was suppressed (Guérin and Nadeau, Reference Guérin and Nadeau1998). First, Liberals and “Bloquistes” both obtained an impressive plurality of the votes in the province of Quebec (a mean of 58% of the votes for the Liberals and 40% for the Bloc Québécois).Footnote 2 Second, and more importantly, both parties consistently obtained a majority of seats in each election during their period of dominance. Considering the very similar dominance of both parties during the two periods (that is, the period 1972–1980 for the Liberals and the period 1993–2011 for the Bloc Québécois), we expect that consequences on economic voting will be similar during both periods. More specifically, we suggest that the dominance of a single party weakens mechanisms of accountability and the extent to which the governing party is rewarded or punished for economic conditions. Under the period of Liberal dominance, Quebec voters mostly voted for the Liberal party. During the dominance of the Bloc Québécois on the other hand, they voted mostly for the sovereignist party.
Guérin and Nadeau (Reference Guérin and Nadeau1998) suggest that when competitiveness is lower in a party system—which undoubtedly is the case in when a single party dominates—the role of short-term factors such as economic evaluations will be reduced. However, their expectation is not solely based on this structural argument. In addition, Guérin and Nadeau refer to previous work on group loyalty. This strand of the literature has shown that identification with a particular group can lead voters to vote loyally for the party that is thought to defend the interests of this group. Such loyalty prevents others factors from affecting the vote choice, hence weakening the economic vote (Guérin and Nadeau, Reference Guérin and Nadeau1998). For this argument as well, the parallel with the period post-1988 in Quebec is evident. The group loyalty of voters in Quebec to their province and to the project of independence, and the fact that the Bloc Québécois claims to defend citizens’ interests on this issue, potentially has made for high levels of loyalty to the party—and a weaker economic vote compared to the rest of Canada.
The straightforward expectation that we test in this paper is that the dominance of a single party in the region of Quebec and strong loyalties to this party suppress the economic vote. The implication is that even if a Quebec voter's view about the Canadian economy is positive, she will still not vote for the incumbent, as there is a party entirely dedicated to the province's interests.
To test this expectation, we compare voting behaviour between Quebec and the rest of Canada. In line with Guérin and Nadeau's (Reference Guérin and Nadeau1998) thesis on the period of Liberal dominance in Quebec, we expect weaker effects of economic evaluations on vote choice among the electorate in Quebec compared to what holds in the rest of Canada. Therefore, we test the following hypothesis:
Hypothesis 1: Voters from Quebec are less affected by economic evaluations when voting than voters from the rest of Canada during the period of dominance by the Bloc Québécois (1993–2008).
It is possible that this rationale may also apply to provinces of Western Canada. Therefore, we ran additional tests re-estimating the models while replacing the Quebec dummy for a variable that distinguishes between regionalist and non-regionalist provinces. In this test, we treat Quebec and Alberta, Manitoba and Saskatchewan—in election years where there was a strong regionalist party—as “regionalist provinces” and contrast voting behaviour in these provinces to the behaviour in the rest of Canada. Doing so does not alter our conclusions (see the results in Table 6 of the appendix).Footnote 3
Our first hypothesis requires assessing differences between regions, as we expect differences in the political context—and in the party system—to affect individuals’ voting behaviour. The mechanism explaining these contextual differences, however, is to be found at the individual level. More specifically, it can be assumed that it is a focus on other, more salient, political issues that inhibits voters in Quebec from voting economically. This would be in line with a rich literature that has shown the salience of an issue to moderate the extent to which these issues affect vote choice (Bélanger and Meguid, Reference Bélanger and Meguid2008; Singer, Reference Singer2011). Furthermore, focusing more specifically on the impact of economic voting and performance evaluations, de Vries and Giger (Reference De Vries and Giger2014) as well have shown this to be strengthened by the salience of the issue on which the government is being evaluated. In the province of Quebec, separatism can be thought of as a more salient political issue, effectively trumping the impact of economic evaluations on the vote choice of the Quebec electorate. To test the validity of this mechanism, we examine whether within Quebec, during the period of dominance by the Bloc Québécois, the intensity of voters’ considerations about the separation of Quebec explains differences in the extent to which economic evaluations affect voters’ choices.Footnote 4 That is, voters with a strong and thus clear standpoint on the issue of independence are assumed to cast votes mainly based on the issue of independence, regardless of the state of the economy. In contrast, voters who do not have a strong opinion on the issue of independence can be thought to consider this issue less important, allowing them to vote economically. We thus derive a second hypothesis:
Hypothesis 2: In Quebec, voters with strong attitudes toward independence will be less affected by economic evaluations when voting than voters with moderate preferences on this issue.
Data and methods
To test our hypotheses, we use data from the Canadian Election Studies (CES). We include data from all election surveys between 1993 and 2008. Importantly, the study design and survey mode was the same for all election studies. Surveys were conducted using a rolling cross-section design and telephone interviews. (Kanji et al., Reference Kanji, Bilodeau and Scotto2012). The election studies consistently included a pre-election wave, with interviews during the campaign, as well as a post-election wave shortly after the election. For each election, data was collected from about 2,500 respondents with about 25 per cent of the respondents living in Quebec (see number in parentheses of Table 1). Importantly, given that our argument is based on the importance of the issue of independence, we restricted the sample of respondents in Quebec to francophone Quebeckers only, excluding allophones and anglophones. As such, our comparison consists of a contrast between francophone voters in Quebec and voters in the rest of Canada. Table 1 summarizes the number of observations in the pre-election wave for each election study included in our analyses. The re-interview rate for the post-election waves is about 70 per cent every year. While we explain the vote choice as reported in the post-election wave of each survey, the independent variables are taken from the pre-election survey waves. Doing so does not eliminate, but does somewhat diminish, the problem of potential endogeneity in our data.
Table 1 Number of respondents (and % by group) included in CES datasets (1993 to 2008)
Note: Respondents included in the post-election wave of the Canadian Election Studies. Source: http://ces-eec.arts.ubc.ca/english-section/surveys/
We conduct analyses of the vote choice, which is a dichotomous variable that is constructed by means of information from the self-reported vote choice of a respondent in the post-election survey. For each party, the variable takes the value of 1 if a respondent voted for the incumbent party and the value of 0 if she did not vote for this party. Non-voters are coded as missing values, which is in line with how research on economic voting generally deals with non-voters (Duch and Stevenson, Reference Duch and Stevenson2008: 45).Footnote 5 We follow Godbout and Bélanger (Reference Godbout and Bélanger2002), as well as others by using logit models to estimate economic voting (De Vries and Giger, Reference De Vries and Giger2014; Nadeau et al., Reference Nadeau, Lewis-Beck and Bélanger2012). This implies we test the economic voting theory in its purest form, modelling voting for or against the ruling incumbent. While previous work has indicated that, in a context of coalition governments, the economic vote is directed primarily towards the party of the prime minister (Debus et al., Reference Debus, Stegmaier and Tosun2014; Duch and Stevenson, Reference Duch and Stevenson2008), it is important to point out that for the full period under consideration single-party governments were in office, further validating our choice of a binary dependent variable.
Our main independent variables are measures of citizens’ economic evaluations. The data include two types of measures that have been included in every election study between 1993 and 2008.Footnote 6 The first indicator is a sociotropic and retrospective measure, which is the standard and most-often used indicator in the economic voting literature (Lewis-Beck and Stegmaier, Reference Lewis-Beck and Stegmaier2013). The question wording is “Over the past year, has Canada's economy…,” followed by the choices “gotten better, gotten worse or stayed about the same.” Furthermore, we also include an egotropic and retrospective measure of respondents’ economic evaluation. The question wording for this item is “Over the past year, your personal finance has…,” followed by the same answer options. For both measures, the answer “gotten worse” is coded 0, “stayed about the same” takes the value of 0.5 and “gotten better” is coded 1.
The mean value of our two economic variables—across the whole time-period—is rather similar, and is .48 for the egotropic indicator and .53 for the sociotropic one. That is, people are not so optimistic, but also not so pessimistic. This is, of course, related to the fact that a non-negligible proportion of the respondents answered that the economy “stayed about the same,” which takes the value of .5 out of 1.
For testing our second hypothesis, which is restricted to analyzing the determinants of vote choice among voters in Quebec, we include a measure of attitudes towards sovereignty. The question asked is “Are you very favourable, somewhat favourable, somewhat opposed or very opposed to Quebec's sovereignty?” We expect voters who have strong attitudes toward independence to be less affected by economic evaluations when voting compared to those who have a moderate opinion on this issue. However, we take into account the possibility that effects are different for those opposed and those in favour of independence by creating two dummy variables: those strongly favourable and those strongly opposed to independence.Footnote 7 For both groups of voters, moderates (that is, somewhat favourable or somewhat opposed) are the reference category.Footnote 8
We add control variables to make sure the estimated effect of economic evaluations on vote choice is not spurious. First, we include age, which is coded in a continuous way and varies between 18 and 101. Gender is a dichotomous variable, with female as the reference category. Education is divided into three categories to capture potential non-linear effects in the model. The “low education” category includes those without a secondary (high school) diploma. The middle category consists of those who completed high school, technical studies, or CÉGEP in Quebec.Footnote 9 The “highly educated” have done at least some university up a doctorate level. Furthermore, we include a party identification variable. It is coded -2 if someone strongly identifies with an opposition party, -1 if she identifies, but not strongly, with an opposition party, 0 if there is no identification, 1 if she identifies (but not strongly) with the incumbent party and 2 if she strongly identifies with the incumbent. Finally, we control for the role of religion (coded 0 for no religion, 1 for Catholic and 2 for Protestant) on vote choice. Given that these sociodemographic factors are likely to affect voting for the incumbent differently depending on what party is in government, we include the main effects of these interactions as well as their interaction with a Conservative party dummy (the reference category is the Liberal party) when analyzing the pooled data from multiple elections. Descriptive statistics on all variables included in the analyses are reported in Table 2 in the appendix.
Table 2 Economic voting in Canada
Note: Entries are logit coefficient. Standard errors are in parentheses and are clustered by election (1993–2008). * p < 0.05, ** p < 0.01, *** p < 0.001
Before estimating the impact of economic evaluations, and differences in their effect between Quebec and the rest of Canada, we must ascertain that economic perceptions do not differ significantly between Quebec and the rest of Canada. In line with how Guérin and Nadeau (Reference Guérin and Nadeau1998) proceeded, we present means and trends in economic perception over time in both regions. Figure 1 includes the over-time trend in mean values for sociotropic as well as retrospective evaluations in Quebec and the rest of Canada. As is evident from eyeballing these graphs, the trends are strikingly similar over time and the mean values differ only slightly.
Figure 1 Economic evaluations in Quebec and the rest of Canada (1993 to 2008)
Previous work that has investigated differences in economic voting between regions in Canada has done so by estimating separate models for different regions (Godbout and Bélanger, Reference Godbout and Bélanger2002; Guérin and Nadeau, Reference Guérin and Nadeau1998). In this paper, we take a different approach and estimate interactive models. Doing so is not only a more efficient estimation approach (Kam and Franzese, Reference Kam and Franzese2009), a pooled interactive model also allows moving beyond description and estimating whether differences in the economic vote are significant or not (Dassonneville and Lewis-Beck, Reference Dassonneville and Lewis-Beck2017). This interactive approach is even more important in the context of the logit models that we estimate. As indicated by Williams, “differences in the degree of residual variation across groups can produce apparent differences in slope coefficients that are not indicative of true differences” (Reference Williams2009: 523). As a result, we cannot straightforwardly compare the coefficients of different logistic regression models. By estimating a full interactive model, in contrast, we obtain precise estimates of the differences in slope coefficients for different groups: voters in Quebec and voters in the rest of Canada in our paper.
Results
In a first step, we estimate the same vote choice models for explaining the probability that voters in the rest of Canada and in Quebec respectively vote for the incumbent. We focus on the estimates of the two variables that capture economic evaluations. Findings are displayed in Table 2. The models presented are based on a pooled dataset that combines data from 1993 to 2008, clustered by election. First, the results of Model 1 indicate that both economic indicators are positive and significant at p < 0.001. Simply put, voters with more positive economic evaluations are more likely to vote for the incumbent. However, it is worth noting that the sociotropic variable has a substantially stronger impact than the egotropic one, which is in line with the literature on economic voting (Lewis-Beck and Stegmaier, Reference Lewis-Beck and Stegmaier2013). In fact, the impact of the sociotropic evaluation is about twice the size of the coefficient of the egotropic evaluation.
The results of Model 1 confirm the presence of economic voting in Canadian elections. More importantly, in Model 2 and Model 3 we examine whether economic voting is weaker in Quebec than what holds for the rest of Canada. For doing so, we add to the base model two interaction terms: In Model 2, we add an interaction between the sociotropic economic evaluations and the dummy variable for Quebec. In Model 3 we add an interaction between egotropic economic evaluations and the Quebec dummy.Footnote 10 If voters in Quebec are affected less by economic considerations, we expect these interaction effects to be negative. Further, the p-value of this interaction term will give insights on whether we can reject the null hypothesis of “no difference between Quebec and the rest of Canada.” As can be read from the results in Table 2, while the interaction term is correctly signed (it is negative), it does not reach a conventional level of statistical significance. The p-value of the interaction term is .985 in the case of the egotropic evaluation and .154 for the sociotropic evaluation, even though this pooled dataset includes more than 14000 observations. As p-values tend to be low when sample sizes are this large (Lin et al., Reference Lin, Lucas and Shmueli2013), this pooled approach is biased towards rejecting the null. Despite this large sample, our conclusion from the results in Table 2 has to be that there is not sufficient evidence to reject the null hypothesis in this case.Footnote 11
The magnitude of economic voting does not differ significantly between Quebec and the rest of Canada, even though for the period under observation a regional party dominated Quebec politics. With regard to the control variables, the results in Table 2 do not contain any surprises. The largest impact comes from party identification. That is, if a voter identifies with the incumbent party, she is more likely to vote for this party, and quite strongly so.
As a way to illustrate these findings, Figure 2 displays the marginal effects of respondents’ sociotropic (left panel) and egotropic (right panel) economic evaluations on voting for the incumbent. Marginal effects are presented for both regions and are based on the estimates of Models 2 and 3 in Table 2. It can be observed that the marginal effect of sociotropic evaluations and egotropic evaluations is significant. Furthermore, and importantly, the confidence intervals indicate that the impact of economic evaluations on voting for the incumbent in Quebec is not distinct from the impact we observe in the rest of Canada.
Figure 2 Economic voting in the rest of Canada (ROC) and in Quebec (QC), sociotropic (left panel) and egotropic evaluations (right panel), 1993–2008
Estimating these models for each election separately, the samples are substantially smaller than what holds for the pooled model, though the number of observations is still well over 2000 in each election year. As evident from the results in Table 3 in the online appendix, we find that only one of the interaction coefficients (egotropic evaluations in 2006) reaches statistical significance. Furthermore, note that we are engaging in multiple testing here—estimating twelve interaction terms in six different models—which increases the probability of finding significance by mere chance (Schafer, Reference Schafer, Lewis-Beck, Bryman and Liao2004). What is more, of all twelve interaction terms, only seven are in the expected negative direction. It is thus fair to conclude that the evidence in favour of the hypothesis that the economic vote is weaker in Quebec than what holds in the rest of Canada is underwhelming.
Table 3 Attitudes toward independence and economic voting in Quebec
Note : Entries are logit coefficient. Standard errors are in parentheses and are clustered by election (1993–2008). * p < 0.05, ** p < 0.01, *** p < 0.001.
After 2008, we can no longer claim that the party system in Quebec is different from what holds in the rest of Canada. The 2011 and 2015 elections were thus organized in a context in which there no longer was a dominant party in Quebec. This change in electoral context allows for an additional test to verify whether the period of dominance of the Bloc Québécois indeed was not particular or different from time periods in which there is no dominant party in Quebec. If the presence of a dominant regionalist party indeed does not strongly weaken the economic vote—as our analyses of elections in the period 1993–2008 seem to suggest—then the economic vote should have a similar effect in the 2011 and 2015 elections, two elections that are marked by the absence of a strong regionalist party. We analyzed the strength of the economic vote and differences between Quebec and the rest of Canada in the context of the 2011 and 2015 elections. As evident from the results in Table 4 in the online appendix, doing so indeed results in essentially the same picture: economic evaluations affect the vote choice of Canadians —in particular sociotropic evaluations—but voters in Quebec are not different from voters in the rest of Canada.
While our results do not offer indications of a regional difference in the economic vote between Quebec and the rest of Canada, it is still possible that the mechanism explains differences in economic voting within the province of Quebec. That is, at the aggregate level, we do not find weaker effects of economic evaluations on the vote of Quebeckers than what holds for voters in the rest of Canada, but Quebec voters with a strong opinion on the issue of independence of their province could be focused on this issue when casting a vote. If so, sovereignty could guide their vote, regardless of the economic situation. We thus hypothesized that in Quebec, voters with strong attitudes toward independence would be less affected by economic evaluations when deciding who to vote for than voters with a moderate position on this issue (Hypothesis 2). In line with how we proceeded for testing the first hypothesis, we pooled all data, this time limited to francophone respondents living in Quebec, and present a single model explaining voting for the incumbent (see Table 3). Standard errors are clustered by election.
First, the results of Model 1 of Table 3 indicate that the main effect of having strong preferences in favour of independence is significantly associated with voting against the incumbent (which is always a federalist party). To the contrary, voters having strong preferences against independence are more likely to vote for the incumbent party.
More importantly, Model 2 and 3 include interaction terms between strong attitudes towards independence and respondents’ economic evaluations. We expected that all voters who have strong views on the issue of independence, either in favour or in opposition to sovereignty, would show weaker levels of economic voting compared to voters who have moderate positions on independence. The results in Table 3 do not confirm this expectation. We find an interesting contrast between those who are strongly in favour of independence on the one hand and those who are strongly opposed to Quebec sovereignty on the other. That is, the sign of the coefficients of the interactions with the sociotropic measure of respondents’ evaluation of the state of the economy indicate that the group of voters who are strongly in favour of independence behave in accordance to our hypothesis; their evaluation of the state of the national economy affects their vote choice less than what holds for voters who have moderate views on independence. The same does not hold for voters who are strongly opposed to independence. For these voters, we even find that economic evaluations affect their vote choice more strongly than what holds for voters with moderate points of view. However, these interaction effects fail to reach a conventional level of statistical significance. When testing these interaction effects in each election separately (see Table 3 in the appendix) it furthermore becomes clear that we do not find consistent evidence of those strongly in favour and those strongly opposed to independence reacting in opposite ways.
Conclusion and Implications
Theoretically, Canada provides propitious conditions for voters to integrate the economy into their considerations of which party they should vote for. First, the winning party benefits from a disproportional allocation of seats due to the winner-takes-all system. Second, in case the first party only obtains a minority of the seats in parliament, its leader will not try to form a coalition with another party (Russell, Reference Russell2008). The Canadian political culture thus favours single-party governments over coalitions. As a result, unless external conditions—such as an international economic crisis—weaken the extent to which the incumbent is held accountable for economic conditions, Canadians can straightforwardly hold the incumbent government accountable for the state of the economy.
However, the literature on economic voting in Canada suggests that things are not that straightforward and that it might be difficult to hold the incumbent party accountable for economic conditions. Previous work has argued that the federal political structure of the country makes it harder to attribute responsibility to the “correct” level of government (Cutler, Reference Cutler2004). In this paper, we focus on another aspect that is argued to render the relationship between economic performance and vote choice somewhat fuzzy in a Canadian context. Indeed, the regionalization of party systems in Canadian politics and the importance of the issue of independence in the province of Quebec, in particular, have been argued to lead some voters not to prioritize the economy as a consideration of their vote choice. The first evidence for such an impact of regionalization was found by Guérin and Nadeau (Reference Guérin and Nadeau1998), who argued that francophones in Quebec did not vote according to their economic evaluations during the 1972–1980 period. The reason, they argued, was that the Liberals were considered the best party to represent Quebeckers’ interests. After the 1980 election, this was no longer the case, allowing for economic evaluations to affect the choices of voters in Quebec.
In this paper, we have argued that the dominance of the Liberals is comparable to the dominance of the Bloc Québécois in the post-1988 era. Both parties won a majority of seats in Quebec during their period of dominance and the rationale for voting for the dominant party (that is, it is the only one to manage the province's interest) is similar as well. Accordingly, we hypothesized that Quebec's voters would be guided less by economic considerations during the 1993–2008 period than voters in the rest of Canada. However, both pooled analyses as well as separate models do not offer strong indications of differences in the strength of the economic vote between both regions. We further suggested that even if there is not a difference overall, some Quebec voters—those with quite strong preferences on the issue of independence—would be guided less by economic evaluations when choosing a party. Again, the results of our analyses do not offer evidence that is in line with our hypothesis.
We fail to find strong evidence of a pattern in which voters who are focusing mainly on the issue of independence, either Quebec voters in general or voters who feel strongly about independence, are guided less by economic evaluations. Instead, we find that economic considerations have about the same effect on the probability of voting for the incumbent among voters in Quebec as what holds in the rest of Canada. In addition, the effect of economic evaluations is about the same regardless of whether or not Quebec voters have a strong opinion on the issue of independence. Overall, our results mainly highlight the stability of economic voting in Canada, despite the presence of regional party systems or a focus on independence.
Our results show that in Quebec, in a period in which the regionalist issue quite powerfully shaped politics and voting behaviour, the economic vote is still present. What is more, we find that the presence of a strong alternative—regionalist—dimension does not significantly weaken the weight of economic evaluations on the vote choice. These results speak to the salience of economic voting more generally, but it is clear that further research in other contexts is needed to verify whether Quebec is exceptional in this regard or whether a similar level of stability in the strength of economic voting across regions can be found in other contexts, such as Catalonia, Flanders or Scotland.
While rewarding and punishing incumbents according to the state of the economy might not be the best or most efficacious way to hold them accountable (Achen and Bartels, Reference Achen and Bartels2016), the presence of economic voting indicates that performance evaluations matter when citizens decide whom to vote for. As such, economic voting can be thought to contribute to the well-functioning of representative democracy (Przeworski et al., Reference Przeworski, Manin and Stokes1999). From this point of view, it can be considered reassuring that the presence of non-economic issues in the political debate, or having strong opinions about a non-economic issue, does not significantly weaken the extent to which economic evaluations affect the vote. Incumbents, we find, were held equally accountable for the state of the economy in Quebec and the rest of Canada throughout the 1993–2008 period.
Supplementary materials
To view supplementary material for this article, please visit https://doi.org/10.1017/S000842391800001X.
