In my previous post I looked at how multi-district list PR could improve regional representation. To illustrate this I examined how the 2015 voting patterns would have produced different results had they occurred under list-PR (while acknowledging that voting patterns would be significantly different under a list PR system). I argued that multi-district PR produced more accurate regional representation because it led to more regionally diverse parliamentary caucuses. It is worth examining the impact that different electoral systems have on the way that votes are translated into seats in a highly regionalized electorate. The emergence of both the Bloc Quebecois and the Reform party made regionalism central to Canadian politics throughout the 1990s. This post argues that while there were certainly important regional divides throughout the 1990s, that the first past the post system exaggerated the extent to which they mapped onto party conflict. First past the post led to an under-representation of the West in government and and under-representation of Ontario and federalist-Quebec in opposition. One of the ways that the West could have ensured that it was consistently in government (to borrow language from the Reform party) was to have pushed for multi-district PR.
The approach that I took to calculating how multi-district PR would have allocated seats from votes is outlined in detail in my previous post. Existing ridings are grouped together into larger districts in which multiple members are elected. The seats in each district are distributed proportionally based on the overall percentage of the vote each party won in the district. Like in the last post, I have done nothing to change alter the number of seats allocated to different provinces or regions in the country, and, consistent with the last post, I have used the D’Hondt counting mechanism for PR. There was, however, redistricting between the 1993 and 1997 elections as well as between the 1997 and the most recent elections. As a result the number seats allocated to each region differs slightly in each election that I looked at. A list of how seats are distributed by district is provided below, and a break down of which ridings fall into which districts is included on the data page. This should not be taken as a “what if” scenario examining what would have happened in 1993 and 1997 if the elections were contested under multi-district PR. The strategies of both voters and parties would have been vastly different under a different electoral system (the Reform party might not even have existed) and so it is impossible to know what would have happened in a particular election under a different electoral system. This is, rather, a demonstration of how different electoral systems can create substantially different degrees of regional representation from the same voting patterns. It is an examination of how counting mechanisms in different electoral systems change results if vote preferences are held constant.
An examination of the overall proportionality of seats yields the results one would expect. The graph below shows the share of Liberal seats under multiple district PR is smaller and closer to the Liberals’ overall vote share. In 1993 there is still some disproportionality, and the Liberals end up with 46% of seats with 41% of the vote. This is not, however, enough to take them from a minority to a majority government. In 1997 the Liberal’s seat share is quite close to their share of the popular vote, with the party taking just over 40% of seats on 38% of the vote. As expected, the Progressive Conservatives and NDP do substantially better when votes are counted using multi-district PR. Both parties suffered under first past the post from having a small proportion of the vote spread out over a large of number of ridings. Finally it is worth noting that the Reform party results are about the same, regardless of whether counting is done using first past the post or multi-district PR. The difference between the two systems in 1993 is about 2%, and in 1997 is 1%. Despite being exactly the kind of regionally based party that is theoretically supposed to benefit from first past the post, the Reform parties’ overall first past the post seat shares were pretty close to their overall share of the vote.
The differences in where parties would have won seats under a multi-district PR system are striking. The graph below shows the different regional seat distributions for each party in the 1993 election. Under multi-district PR the Liberal party still ends up being a largely Ontario based party, but it increases the share of it caucus that comes from Quebec and the prairies, and doubles the share of its caucus that comes from BC. Each of the three federal opposition parties sees a substantial change in their regional make-up. The increased number of seats the PC party would have won under multi-district PR also gives the party broad regional representation. It ends up with substantial numbers of members in Atlantic Canada, Quebec, and Ontario, as well as a handful of seats in Western Canada. This is compared to the 1 member received from Atlantic Canada and one they received from Quebec under first past the post. Unlike under first past the post, the Reform party and the NDP would not have been purely Western parties under multi-district PR. Instead of winning just one seat east of Manitoba, the Reform party would have had gotten almost a third of its caucus from Ontario and even would have a member from Atlantic Canada. The NDP would have still been primarily a Western party, but it would have a handful of caucus members from Ontario.
The changes in regional support are even more striking in 1997, as shown in the graph below. The extent to which the Liberal party was an Ontario based party would have been dramatically reduced. Instead of taking 65% of its seats from Ontario it would have only taken 45% of its seats. It would have a larger proportion of its caucus from Quebec, and it would have more than doubled the proportion of its caucus from the prairies and BC. The changes in the Reform party would have been similar to those it would have experienced in 1993, with the party gaining a substantial Ontario-based caucus. Both the PCs and the NDP would have gained significant Ontario caucuses, and the PCs would have increased the size of their caucus that came from Quebec as well. In the case of the PCs this would have changed them from a primarily Atlantic party to one that had a MPs from across Eastern Canada. In the case of the NDP, this would have made the party a broad English Canadian party instead of one dominated by MPs from Atlantic Canada, Manitoba, and Saskatchewan.
A multi-district PR system would have decreased the extent to which parliamentary debate was framed along regional lines. Under first past the post, in 1993 and 1997 a Liberal parliamentary caucus that had a majority of its members from Ontario (and in 1997 had almost 2/3rds of its members from Ontario) faced two main opposition parties that had all of their members from either Quebec, in the case of the Bloc Quebecois, or the West, in the case of the Reform party (though Reform took 1 seat in Ontario in 1993). This inevitably made debate in parliament regional as a majority government largely from Ontario was allowed to legislate its agenda against the criticism of an opposition from Western Canada and Quebec. Increasing the regional diversity in the both the government and the opposition would have changed these dynamics so that Western interests were more prevalent within the government and Ontario’s interests were better represented in opposition criticism of the government. The governing Liberals would have had a stronger contingent of Quebecois and Western MPs, and as a result, a greater incentive to take both interests into account when making policy. The PCs, a more regionally representative party, would have replaced the Bloc Quebecois as one of the stronger opposition parties in Canada. This would have added a more regionally diverse critique of the government to the BQ’s separatist one. Finally, as noted in the table below, the Reform party would have had 17 MPs from Ontario in 1997 and 18 in 1993 with incentives to consider Ontario’s interests when holding the government to account. Providing more accurate and diverse regional representation in both government and opposition party caucuses would have lessened the degree to which debate between government and opposition also ended up being a debate between Ontario and Quebec and Western Canada.
There is no escaping the fact that Canadian politics was highly regionalized through the 1990s. Even multi-district PR would have created parties that had strong bases in particular regions. First past the post, however, exacerbated these divisions by creating parties that were reliant on particular regions for almost all of their members. This meant very limited representation of Westerners in government, over-representation of non-Ontarians in opposition, and a parliament that made Canadian politics look like it was Ontario against everyone else. A system, like multi-district PR, that would provided not only more accurate regional representation, but it would have given a much stronger voice to significant numbers of Ontarians who had not supported Liberals and to Westerners that did. Such representation would have highly valuable at a time when Canada was experiencing high levels of regional tension as a result of Quebecois nationalism and Western alienation.