Having Cake and Eating it Too: How List PR Could Improve the Proportionality and Regional Representation of Parliament (Part I)

Every electoral system has benefits and costs. One of the trade-offs that is is often asserted between different electoral systems is one between regional and proportional representation. It is often argued that first past the post systems provide strong regional representation while proportional systems sacrifice regional representation in order to provide accurate partisan representation. The extent to which first past the post systems provide for regional representation and the extent to which proportional systems cannot include regional representation are often over-stated. This post will examine the 2015 federal election and show how a multi-district list proportional system could lead to stronger regional representation than the current first past the post system.

The major problem with regional representation in first past the post systems is that, like national representation in such systems, it is rarely an accurate reflection of the diversity of opinions within a particular region. The winner-take-all nature of first past the post means that parties that are strong in a region can dominate it, winning most or all of a region’s seats even if there is a substantial support for other parties. Because of this, parties in first past the post systems tend to be over-regionalized. The Conservatives, for example, because of their strength in Alberta and Saskatchewan, end up having proportionally more parliamentary members from the prairies than they do voters and fewer members from Ontario and Atlantic Canada than they have voters form those regions. The result is a parliament where regional divides fall along party lines to a greater degree than they do in reality. It is also a parliament in which important and substantial regional populations are under-represented or not represented at all. Regional representation under first past the post, like national representation, is highly distorted.

This is problematic for two reasons. The first is that it tends to polarize conflict between parties in parliament along regional lines. Because a large proportion of the Conservative parliamentary caucus is from the prairies they tend to see as their job to defend the interests the prairies. Conversely because the Liberals have weak parliamentary representation from the West there are fewer voices advocating Western issues in their caucus and they often are perceived as weak when it comes to Western issues. This can lead to a political discourse that over-emphasizes the divides between different regions. The second is that governments can often have difficulty building cabinets that are regionally representative. When a party has few or no representatives from a region they have few or no people to appoint to cabinet to act as representatives for those regions in discussions over government policy, and few people in caucus to challenge policies that do not reflect that region’s interest. The Harper Conservatives when they were in government struggled to find cabinet ministers that could serve as representatives for Montreal or Vancouver because they won few seats in either city. Given the amount of control that governing parties have over policy making, regional under-representation in governing parties can do significant damage to regional representation.

List proportional systems are often noted for their lack of regional representation. Such a system, however can include significant regional representation if a list system is designed for multiple districts. Rather than taking the entire country as single electoral district (as is done in Denmark or Israel) the country can be divided into several electoral districts in which multiple members are elected (these systems are used in countries such as Finland and Sweden). Each party submits a list of members running for each district and seats a apportioned to parties based on the percentage of vote that the party wins in that district. A city, such as Edmonton for example, could be made an electoral district with 10 seats. If the Liberals won 40% of the vote in Edmonton they would win 4 of Edmonton’s seats and the first 4 candidates on the Liberal’s Edmonton list would be elected into parliament. Such a system could not guarantee the riding by riding representation that exists in first past the post, but it would ensure that the different regions of the country are represented. Since parties rarely win 100% of the vote in a particular region, it would also ensure more regional representation within each party in parliament.

In order to illustrate this I look at the 2015 federal election. I combine different ridings into larger regional districts* and calculate the number of seats that a party would have won under a multi-district PR system in each region. The calculations are based on the sum of the votes for each party over all of the ridings grouped into each region (data is taken from Pundit’s Guide). Seats are apportioned using the D’Hondt method. This method of counting limits the number of seats that small parties win because it effectively rounds down when translating vote percentages into seats. This means that a party that wins 3% or 4% of the vote in a 10 seat district is unlikely to win any seats. Below is a table with the number of regions and number of seats each region would have under a multi-district PR system. Because different regions have different populations each will have different numbers of seats. Also, because I have simply grouped together existing ridings within districts, I have done nothing to change the number of seats each region has. Finally it should be noted that this is not a simulation of what would have happened if the 2015 system had been run under a multi-district PR system. Running an election under an electoral system that is so different from the current system would dramatically change the way that voters and and parties behaved (and might even change the number of parties contesting the election). This is, rather, a demonstration of how a different electoral system can produce broader regional representation within parties even if the vote choices of individuals does not change. The point of this piece is to show how multi-district PR can correct for the way that first past the post over-regionalizes party representation in parliament, it is not a what if scenario.

Seats Per District Table

The graph below shows the increase in overall proportionality that is achieved under a multi-district PR system. Multi-district PR does not create perfect proportionality because in each district there is some rounding that goes on when translating votes into seats. The result is that the largest party in most districts, the Liberals, win slightly more seats than their share of the popular vote (44% of seats instead of 39%). Conversely the smallest party in most districts, the Greens, win slightly fewer seats than their share of of the vote with over 3% of the vote getting them less than 1% of the seats in parliament. The problem the Green Party has is that in every district other than Vancouver Island, their share of the vote is lower than the share needed to win a seat. In Edmonton, where 1 seat is given to a party for every 10% of the vote it wins, the Green Party is under 10% and in Toronto, where 1 seat is given to a party for every 4% of the vote a party wins, the Greens are under 4%. The results produced by a multi-district PR system, however are far closer to the share of the vote each party receives than the results produced by a first past the post system. Multiple district PR would not manufacture majorities from under the 40% of the vote the way that FPTP does.

FPTP PR Proportionality

The benefit to regional representation that results from multiple district PR comes through in the graph below. Under this system each of the major federal parties becomes less regionally homogenous. The Liberals see the share of their caucus that comes from Atlantic Canada, Quebec, and Ontario decrease, and the Western share of their caucus increase. The party goes from having only 15.8% of their caucus in the West to having just under a quarter of their members from the West. The Conservatives, conversely become less of a Western party, increasing the percent of their caucus that comes from Ontario and Atlantic Canada. Finally the NDP becomes less of Quebec/BC party, increasing the extent to which its caucus includes representatives from Ontario and Atlantic Canada. Under first past the post just under 70% of the NDP’s caucus is from Quebec or BC. Under multi-district PR that number drops to just under 50%.

Percentage of Party Caucus From Each Region

The results produced by a multi-district PR system would increase the Western voice within the Liberal government. This is important given the extent to which governing parties are more responsive to their caucuses than they are to parliament as a whole. The increased proportion of Westerners in the governing party would increase the incentive for the Liberals to take into account the interest of Western Canada when making policy. The same is true when it comes to opposition parties and their critiques of government. Opposition parties with broader regional representation are more likely to put forward critiques of the government that reflect the interests of all regions and not just the one or two where they have the majority of their members from. Parliamentary debate in a multi-district PR system would be less regionalized as all parties parliament would craft policies and critiques that take into account the interests of a broader number of regions.

A look at the number of seats that each party would win in each district underlines the extent to which multi-district PR would improve regional representation within parties. The Liberals get seats in each district while the Conservatives are shut out of only Newfoundland, Halifax, and Northern Canada. The Liberals would win one seat in both Northern Alberta and Southern/Central Alberta, regions where Liberals have rarely had representation under first past the post. The Conservatives would win seats in both Montreal and Vancouver, two cities that the Harper Conservatives struggled to find cabinet ministers to represent while they were in office. This has two important implications. The first is that it increases the likelihood both rural Alberta and all of Canada’s major cities will be represented in cabinet or at the very least in government caucus no matter which party is in government. This is important because both cabinet and caucus have significantly more influence over government policy than parliament as whole. Increasing regional representation in government parties increases the extent to which governments include voices from every region in the country, and as a result, the extent to which every region is able to influence policy. Second, increasing regional representation within parties creates a parliament that is a more accurate reflection of the diversity of Canadian opinion. While rural Alberta is indeed mostly Conservative, there are some Liberals in rural Alberta and their ideas an interests ought to be represented in parliament. The same is true for Conservatives in Montreal and Vancouver. Under the current system these groups of people are almost never represented in parliament.

Party Seats by District

None of this is to say that there are not still some advantages to first past the post and some drawbacks to multi-district PR. Under a multi-district PR system citizens would lose the link to an individual MP that they have under first past the post. Some of the regions created under a multi-district PR system would have to be very large, which carries with it the possibility that smaller regions would lose representation. There is no guarantee under this system that there would be an MP from Northern Saskatchewan because there is no district dedicated only to Northern Saskatchewan. To some extent the size of each district could be shrunk in order to create more specific regional representation, but that would come at the cost of proportionality. The smaller districts are and the fewer the number of MPs elected from each district, the less proportional each parties’ share of overall (and regional) seats is going to be. Finally this system will create either minority or coalition governments. The extent to which a majority government is good or bad is debatable (and so this is not going to be much of cost for those that prefer minorities), but for those that dislike minority situations this is a cost of the more accurate regional representation multi-district PR provides. Proportional representation systems, especially pure PR systems are often accused of hurting regional representation. The extent to which they do this, and the extent to which first past the post provides accurate regional representation are often overstated. A multi-district proportional system can indeed provide more accurate regional representation than first past the post does.

* For a list of existing ridings in each district see the data page where I have the calculations for the seats won in each district in a spreadsheet.


How Should We be Voting: How to Run a Fair Referendum on Electoral Reform

In my last post I made the case for holding a referendum on electoral reform. This was in response to a commitment made by the Liberal government to change the electoral system after holding public consultations, but not after a referendum. Referendums can be a lot trickier than they seem. Run poorly referendums can be end up being contests over the extent to which voters’ trust the current government or over whether voters understand the proposed policy changes instead of over which policy option voters prefer. A referendum on electoral reform needs to be careful designed. The aim should be to maximize the likelihood that voters see the referendum as a choice between different electoral systems, separate from concerns over which party is in power on how they feel about the general state of Canadian politics. It is likely impossible to prevent such factors from influencing public opinion at all, but some things can be done to weaken their influence. The electoral system put to a referendum should be chosen through a process independent from partisan politics, the referendum should not be held during an election, and it should be phrased as a choice between electoral systems not as a yes/no question over the status quo. Each of these three measures will increase the quality of debate in a referendum over electoral system change.

Ensure People Trust the Process for Selecting Alternative Systems

There is a danger that referendums become votes over how people feel about the current government instead of about the policy options being put to a referendum. Most recently, the Vancouver area referendum on increases taxes to fund public transit and infrastructure expansion ran into difficulty because many Vancouverites did not trust Translink, the organization that runs Vancouver’s public transit system. While this was not the only reason that the referendum failed, it certainly was a focal point of the “no” side’s campaign. The problem when citizens vote like this is that the value of the policy options being presented is often not related to trustworthiness of the government in power. An electoral system will affect the way that elections are run long-after the politicians who introduce and campaign against electoral reform have left politics. Their trustworthiness has little to do with the quality of an electoral system.

The danger that a referendum becomes about voters’ trust of politicians is particularly likely to influence referendums on electoral reform. Different electoral systems will benefit different parties. Because of this, it will be easy for opponents of electoral reform to make a case that electoral reform proposed by a particular party or government is being proposed only to increase the likelihood of that parties’ future success. The more removed the choice of electoral system for a referendum is from partisan politics the greater the likelihood that voters will trust that the system can serve their interests. It is thus imperative that any referendum be preceded by an independent examination of different options for electoral reform.

A citizens’ assembly, instead of parties, could decide what kind of system will be included in a referendum as an alternative to first past the post. A citizens’ assembly is a decision making body that contains randomly selected individuals from across the country. In the case of BC one man and one women was selected at random from each of the provinces ridings. These individuals then listen to presentations from experts on electoral systems, hold public consultations, and deliberate amongst themselves in order to come up for a recommendation for electoral reform that can be put to a referendum. By removing the decision over what kind of reform is put to a referendum from politicians, voters can be more confident that the electoral system being proposed is a product of deliberation that broadly considers that interests of the public and not those of politicians. If voters trust the citizens assembly (and the high level of support for the assembly’s recommendation in BC suggests that in that case they did) they are more likely to be receptive to its recommendations and see an electoral reform proposal as a reasonably alternative to the current system and not a cynical attempt on the part of politicians to implement a reform that suits only their interests. When citizens trust the process though an alternative electoral system is chosen, they are much more likely to give it fair consideration in a referendum.

Keep the Referendum Away from Elections

One of the other ways to try to limit politicians’ influence over a referendum is to hold the referendum in between elections. There are two problems with holding a referendum simultaneously with an election. The first is that it makes it easier for parties to mobilize to influence voters. Parties usually experience large increases in membership in election years, obtain more donations, and have an easier time recruiting volunteers. This means that parties fighting a referendum campaign while also fighting an election will have more resources to try to influence voters than they would competing in a referendum held separately from an election. Holding the referendum in between elections will not prevent parties from participating in a referendum campaign, and they will likely still be able to hold a great deal of influence over their hard core supporters, but it will help to level the playing field in the campaign between parties (who are likely to support different electoral systems out of self-interest) and other electoral advocacy groups (who are more likely to represent the interests of a segment of the population).  Holding a referendum apart from an election means holding a referendum when parties are relatively weak (compared to elections).

The second danger that comes with holding a referendum simultaneously with an election is that the debate over electoral reform can get drowned out by the election. Electoral reform debates are highly technical affairs and, as a result, it can be difficult for both advocates and opponents of electoral reform to hold voters’ attention. This is especially the case when those debating electoral reform have to compete with an election for voters’ attention. The kinds of issues that are raised during an election are often more interesting to voters. The result is that voters will tune out the electoral reform debate an instead follow the major issues that divide the political parties competing for government. At its worst, a referendum held at the same time as an election will lead voters to simply follow the cues that parties give them when it comes to voting on electoral reform. When a referendum is held separately from an election, it is more likely that the debate over electoral reform will be the most prominent political debate that voters engage with. This increases the likelihood that voters will hear from advocates and opponents of electoral reform without links to parties, and increases likelihood that the debate over electoral reform will be about the interests of different voters instead of the interests of different parties. This will, of course, not remove parties from the referendum debate completely, but holding a referendum independently of an election should decrease their influence over the referendum campaign and increase the extent to which the electoral reform debate becomes about the interests of different groups of voters.

Keep the Words “Yes” and “No” Off the Ballot

One of the challenges in designing a referendum question is creating a question that gets voters to focus on a discussion of the merits of both the current and the proposed electoral system. Part of this is affected by the way the ballot question is worded. There is a danger that a question phrased as a yes/no decision on whether to change from the status quo to another system becomes about whether individuals understand the proposed system (regardless of whether they understand the current one) and about whether they are satisfied with the status quo. A yes/no referendum question frames the choice that voters make, and the resultant debate that they have, as a choice over whether they want to change the current electoral system. As an alternative, voters could be asked whether they prefer a first past the post system to a proposed alternative such as proportional representation or alternative vote.  Rather than voting “yes” or “no” voters would choose between “first past the post” and an alternative such as “mixed member proportional” or “single transferable vote” This reframes that decision that voters make and can lead to a subtle but important change in the way that voters approach the referendum question.

Framing the debate as a choice between two electoral systems as opposed to a question about whether to change the electoral system increases the likelihood that voters will carefully consider the costs and benefits of different systems. Setting up the question as one between first past the post as an alternative makes it more difficult for advocates to make the case that voters should vote for change because of dissatisfaction with the current system. It also makes it more difficult for opponents to argue that voters should simply vote no either because they do not understand the proposed alternative or because they are satisfied with the current system. When advocates and opponents have to campaign for two particular systems instead of for or yes or a no, the likelihood that voters will ask them to explain both systems (and that voters will make a serious attempt to understand the costs and benefits of each system) increases. A fair question should encourage voters to raise questions about how both first past the post and its alternative work. When opponents of electoral reform have to run a campaign that says “vote first past the post” instead of “vote no,” they are more likely to be asked about how first past the post works and what its advantages are. Setting up the question as a choice between two alternative systems on a ballot instead of as a yes/no question can nudge the debate in a way that will produce more careful consideration of the costs and benefits of each system.

Ensuring that the choice of systems to put to a referendum is not done by politicians, that the referendum is held apart from an election, and that the referendum question is not a yes/no question are all valuable steps that can be taken to ensure that the debate over electoral reform is focused on costs and benefits of different electoral systems. This will not make the debate perfectly rational, it is unlikely that such a debate exists in politics. Different political parties, as well as other political actors, have strong interests in either maintaining the current electoral system or in seeing electoral reform enacted. These interests mean that any debate over changing the electoral system is likely to see its fair share of emotional appeals and misinformation. There are, however, ways in which the referendum can be designed in order to maximize the likelihood that voters will weigh the costs and benefits of each system, and which can reduce the likelihood that referendum becomes about something other than the costs and benefits of different electoral systems. A carefully crafted referendum question with its roots in a process that voters’ can trust and held apart from election can increase the quality of the debate over electoral reform that comes with a referendum.