Electoral Insight Logo

All Issues | Search | Complete List of Articles

Electoral Insight – January 2005

Henry Milner
Chair in Canadian Studies, Université Paris III Sorbonne Nouvelle

In this article, I look at certain aspects of the results of the June 2004 Canadian federal election and reflect on the possible effects if Canada were to adopt a different electoral system. Changes to the electoral system could have an impact, first on the representation of political parties, nationally and regionally; second, on the representation of women; and third, on voter turnout, especially that of young people. Given that the 2004 election was the most competitive for many years, yet saw a further drop in turnout, the last is an issue we cannot ignore.

The question of a new electoral system is also pertinent because electoral system reform is on the agenda of British Columbia, Ontario, Quebec, New Brunswick, Prince Edward Island and Yukon.1 Meanwhile, Parliament has taken a major step towards beginning a review of the federal electoral system. On October 18, 2004, it unanimously amended the Speech from the Throne to instruct the Standing Committee on Procedure and House Affairs "to recommend a process that engages citizens and parliamentarians in an examination of our electoral system with a review of all options."2 We can thus expect serious discussion in the House of Commons on the report tabled there in March 2004 by the Law Commission of Canada,3 which set out in detail a new electoral model for Canada, with elements of a proportional representation (PR) system.

In any such discussion, the very existence of a minority government will be seized upon by opponents of reform to remind us that PR electoral systems seldom result in majority governments. This paper starts from the premise that we have reached a point in this country when the burden of proof lies with those who claim that producing single-party majorities in Parliament is the sine qua non of a good electoral system. The fact is that most stable democratic countries use a form of PR and systematic evidence4 shows that such elections are no less likely to produce good government. Worry among Canadians that instability will result when no one party has a majority of seats stems from our experience of minority situations under the first-past-the-post (FPTP) system. Though most minority Canadian governments in memory, e.g. under Pearson and Trudeau, were in fact quite effective, they were comparatively short-lived. But minority governments under PR need not give rise to the same concerns. Unlike the situation under FPTP, provoking an election under PR is not likely to bring majority government: so nothing is to be gained by hastily causing an election.

It is true that PR puts governing parties to harder tests. Getting a legislative agenda through takes more effort and time. Without an automatic majority, they have to spend time between, and not just during, elections justifying what they are doing. Since they have to work harder, in narrow efficiency terms – output per unit of work – governments under PR are less efficient than governments under FPTP. But the democratic quality of the output is another matter. Overall, had the 2004 federal election been held under PR, it would have produced a minority or coalition government supported by parties representing more than half the voters, one no less efficient, accountable and transparent, and certainly more stable than the government produced under the current system.

The Law Commission proposal and the June election

So much for efficiency. What about democracy? The new Parliament, as noted, has before it a detailed proposal from the Law Commission of Canada for a mixed-member proportional (MMP) system to elect its 308 members. This is a compensatory two-ballot form of PR first developed in Germany and adapted for use in New Zealand (in 1996) and for the new Scottish Parliament and Welsh National Assembly (in 1999). In the system proposed for Canada, each province constitutes an electoral region, except Ontario and Quebec; Ontario is subdivided into four regions and Quebec, into two. Voters in each region would cast two ballots,5 one for a candidate in a single-member constituency (just under 2/3 of seats6), the other for a "flexible"7 party list (just over 1/3 of seats), with candidates permitted to run both in a constituency and on the list.

Table 1 displays the number of seats won by each party in the 2004 election, as well as what they would have won, in both list and constituency seats, under the Law Commission model. Note that, as in all such simulations, we can only assume that the voters would have voted the same way, an assumption we know to be inaccurate.8 We can see that application of the model results in outcomes quite faithful to voting intentions, i.e. full proportionality. This is in contrast to the actual outcome, in which the disparity between votes and seats won, though smaller than in previous elections due to the closeness of the vote, was still significant. For more on this topic, readers may wish to see the article by Lawrence LeDuc in this issue.

Only the Conservatives (who won 30% of the votes and 32% of the seats in June 2004) would win roughly the same share of seats under the Law Commission model (31%). The Liberals' advantage is reduced to 2% (37% of the vote netting 39% of seats, compared to the 44% of seats they actually won). The Bloc Québécois, which won 12% of the votes, would receive the same percentage of seats (compared to the 17.5% it actually won). The N.D.P., with 16% of the votes, would win 15% of the seats (compared to the 6.2% it actually won). Finally, rather than being shut out, the Green Party would win 3% of the seats with 4% of the votes.

Even more important is the effect on regional representation. Under the Law Commission model, in the combined Atlantic provinces, the Liberals would give up two seats to each of the Conservatives and N.D.P., the former breaking into the Liberal monopoly in P.E.I. and the latter gaining a foothold in Newfoundland and Labrador. In Quebec, the Conservatives and N.D.P., instead of being shut out, would win six and three seats respectively. In Ontario, the Liberals would lose 25 seats, including 7 to the Conservatives, 12 to the N.D.P. and (up to) 4 to the Greens.9 In Saskatchewan and Manitoba, the Conservatives would cede 6 of their 20 seats, 4 to the Liberals and 2 to the N.D.P. (thus giving the N.D.P. representation in Saskatchewan, where it was shut out despite receiving 23% of the vote). In Alberta, the Conservatives would cede 7 of their 26 seats, 4 to the Liberals, 2 to the N.D.P. and 1 to the Greens. In British Columbia, the main beneficiaries would be the N.D.P. with an additional 4 seats and the Greens with 2.

The representation of women under the Law Commission model

How would women candidates have fared if the 2004 election had been fought under the Law Commission's proposed version of MMP? It is, of course, impossible to know exactly who would have been on the party lists and, had the flexible list system proposed by the Commission been used, the extent to which voters would have taken advantage of it to move women candidates into or out of winning top-of-the-list positions. However, given the significant public attention paid to the proportion of women candidates in the recent election,10 the parties would have been under considerable pressure to place women candidates in high list positions.

If the parties alternated men and women on their lists (as is done in Scandinavia, for example), half the 106 list MPs under the Law Commission model would be women. Maintaining the same proportion of women in the district seats as in the election itself (21%), application of the Law Commission model would have resulted in 95 women elected in total: 53 (of 106) list MPs, plus 42 (of 202) constituency MPs. A perhaps more realistic simulation might be to apply the New Zealand ratios. In their version of MMP, with 43% of MPs coming from lists, women won an average of 38% of list seats in the three elections held under this system (compared to 22% of district seats).11 Applying the 38% ratio to the list seats for Canada under the Law Commission model, we would have elected 82 women MPs (40 list and 42 district), still a distinct improvement over the 65 actually elected.

The thorny question of turnout

According to Elections Canada, 60.9% of registered voters turned out on June 28, the lowest turnout ever, down from 64.1% in 2000.12 Yet 2000 was an election in which the outcome was widely anticipated in advance, while 2004 was expected to be a cliff-hanger right to the end. Part of the explanation lies in the election date, June 28, when many Canadians were getting into a vacation mood. But there is clearly something deeper at work: this was the fourth consecutive election in which turnout declined. We know that a key factor in the declines of 1993, 1997 and 2000 was abstention among young people.13 Initial reports14 suggest that this was also the case in 2004.

A highly useful guide to what is happening is found in a recently published analysis of turnout in 22 democracies that have held elections continuously since World War II.15 Mark Franklin shows that the most important factor explaining turnout is the "character of elections," consisting of the electoral system, the fractionalization of the party system, the time elapsed since the previous election, and the closeness of the outcome. Changes in the character of elections, he finds, largely account for the average 7% turnout drop in the past 30 years among the 22 countries.

What makes Franklin's analysis original is that he places at its core the fact that age groups (cohorts) are differently affected by the character of elections. Voting, we know, is largely habitual; hence, the crucial group is the young, who have not yet developed habits of voting or not voting. These habits, he shows, are developed especially as a response to the character of the first (and second and third) elections for which one is eligible. But the effect is a long-term one, since with each new cohort arriving at an election, another is leaving. Turnout is affected as cohorts newly eligible to vote become set in their non-voting ways, replacing other cohorts that were more likely to vote.

The force of Franklin's argument emerges in the connection shown between declining turnout and the time when countries, typically in the 1960s and 1970s, reduced the voting age. The effect was that a certain number of individuals failed to vote when first able, because they were at an age when they lacked the social networks conducive to voting and became socialized into non-voting behaviour. Part of the recent turnout decline is thus traced to the replacement by these cohorts of earlier cohorts whose first electoral experiences were more conducive to developing habits of voting. Note that such replacement is slow: it is only years after the fact that the statistical effect becomes clearly visible.

This highly sophisticated analysis could explain why turnout continued to decline in 2004, even if there was a higher turnout by first-time voters than in 2000 in response both to the more competitive character of the 2004 election and extra efforts to get young people to the ballot box. But such an effect would be small and slow to appear, and could not by itself have offset the replacement of cohorts more inclined to vote by those arriving on the political scene in the non-competitive 1990s.

Were the next election and the one after it to be competitive, the effect would be compounded and begin to make itself felt on turnout. But this is unlikely under FPTP. For almost 100 years, Canada has alternated between periods of single-party (Liberal) domination and two-party competition. Moreover, regional voting patterns have – even in competitive epochs – led to a large number of uncompetitive districts.

There is effectively only one way to ensure that new voters confront more competitive elections, and that is to change the electoral system. Adopting a system like the one recommended by the Law Commission would make virtually every vote count toward electing MPs and affecting the position of the parties. As such, at the very least, it would facilitate efforts in and out of civics classes16 to bring the young to the ballot box. Of course, any results would be slow to appear. Moreover, they would be limited, for there appear to be wider cultural forces, entirely unrelated to the character of elections, driving down turnout. However competitive, mainstream electoral politics is increasingly facing a different form of competition for the attention of younger generations, one from the world of electronic celebrities, which can be entered at the push of a button on a TV remote controller, computer mouse or Playstation joystick.

While we thus cannot hope to return to the turnout levels of the 1970s, our hands are not tied. A proportional electoral system, by establishing a context in which the composition of our representative institutions better reflects the views of electors, could make it easier to develop effective forms of civics education to address potential political dropouts, bringing the positions (and spokespersons) of the different parties into the classroom. At the very least, such combined efforts should stem turnout decline.

As demonstrated here, adopting an MMP system as proposed by the Law Commission of Canada would have produced election results that are arguably fairer in a number of ways: representation of parties, regions and women. For these reasons, we can only hope that the current minority government situation proves to be an opportunity to consider electoral reform, and not a stick with which to beat it.


  1. Information on electoral reform in the five provinces (identified by the relevant body) is available at British Columbia's Citizens' Assembly on Electoral Reform: www.citizensassembly.bc.ca/public, Prince Edward Island's Electoral Reform Commission: www.gov.pe.ca/electoralreform/index.php3, New Brunswick's Commission on Legislative Democracy: www.gnb.ca/0100/index-e.asp, Quebec's Secrétariat à la réforme des institutions démocratiques: www.institutions-democratiques.gouv.qc.ca/reforme_institutions/coup_oeil_en.htm, Ontario's Democratic Renewal Secretariat: www.democraticrenewal.gov.on.ca/english/. Information about Yukon's electoral reform agenda is available at www.gov.yk.ca/news/2004/04-007.html.
  2. Canada, Parliament, House of Commons, Debates of the House of Commons of Canada (Hansard), 38th Parliament, 1st Session, Number 012, October 20, 2004, www.parl.gc.ca/38/1/parlbus/chambus/house/debates/012_2004-10-20/han012-E.htm.
  3. Law Commission of Canada, Voting Counts: Electoral Reform For Canada (Ottawa: Department of Public Works and Government Services, 2004) at www.lcc.gc.ca/en/themes/gr/er/er_main.asp.
  4. See, for example, Arend Lijphart, Patterns of democracy: Government forms and performance in thirty-six countries (New Haven: Yale University Press, 1999).
  5. The calculations used to distribute seats are set out in Law Commission of Canada, Voting Counts, pp. 97–99.
  6. The odd fraction results because the formula used is 60:40 for the Atlantic provinces, rather than 67:33, given the small population of these provinces.
  7. The Law Commission proposes using the Swedish system, in which a locally popular candidate low on the party's regional list is moved to the top if he or she receives the "personal vote" of more than 8% of the party's supporters.
  8. This is because far fewer voters than under FPTP are in a situation in which acting on their first preferences amounts to wasting their votes.
  9. The simulation does not divide Ontario into the proposed regions, something beyond my capacity to do, which would probably reduce to three the number of seats for the Greens.
  10. "A non-partisan network of influential women" publicly took the parties to task – most notably the Conservatives with only 11% women candidates – for failing to place more women in winnable ridings. (Cheryl Cornacchia, "Fewer women likely in next Parliament," The Gazette [Montréal], June 27, 2004, p. D3.)
  11. Based on data from The New Zealand Electoral Compendium, 3rd ed. (Wellington: Electoral Commission, 2002), pp. 176–177.
  12. Note: the 64.1% figure for 2000 is a correction announced by Elections Canada in March 2003 of the 61.2% official turnout figure tabled in the House of Commons.
  13. While in most generations, people vote more as they get older, this does not explain the phenomenon. See Elisabeth Gidengil, André Blais, Neil Nevitte and Richard Nadeau, "Turned Off or Tuned Out?: Youth Participation in Canada," Electoral Insight Vol. 5, No. 2 (July 2003), pp. 9–14, at www.elections.ca/abo/eim/ insight0703_e.pdf.
  14. "When asked during the campaign's final 10 days how likely they were to vote in the upcoming election, respondents in their 20s were much less likely to commit to voting." Elisabeth Gidengil, Neil Nevitte, André Blais, Patrick Fournier and Joanna Everitt, "Why Johnny Won't Vote," The Globe and Mail, August 4, 2004, p. A15.
  15. Mark N. Franklin, Voter Turnout and the Dynamics of Electoral Competition in Established Democracies Since 1945 (Cambridge: Cambridge University Press, 2004).
  16. Fixed – and thus known in advance – election dates would help as well.


The opinions expressed are those of the authors; they do not necessarily reflect those of the Chief Electoral Officer of Canada.