Incentives Matter: A Case for Dual Member Proportional Brief prepared for the Parliament of Canada’s Special Committee on Electoral Reform

The following is my submission to the Parliament of Canada’s Special Committee on Electoral Reform. For those interested, the Committee is inviting participation until October 7, 2016, via this link.

(Download PDF)


The Committee should help Canada move responsibly toward proportional representation by recommending the smallest change that will accomplish this goal. Proportionality necessitates larger districts served by multiple MPs. It follows that the smallest necessary change is to have two MPs serving districts twice as large. Hardly any change is needed to how Canadians vote. A paper ballot marked with a single ‘X’ will work just fine. It is also possible to continue dedicating a seat to the most popular candidate in each district. In that case, the remaining seats would have to be awarded in a compensatory manner to ensure overall proportionality. The one system with all of these properties is called Dual Member Proportional, or DMP.

In this brief, I compare DMP to other options based on incentives created for parties, politicians, and voters. My predictions are that DMP will minimize tactical voting, provide reasonable protection against the diffusion of seats to regional and single-issue parties, promote engagement, reduce partisanship, and enable a substantial increase in the number of female MPs elected to Parliament. Most importantly, DMP will incentivize competition for every vote, giving Canadians everywhere a genuine influence over how they are governed.

1         Introduction

To understand the impact of changing Canada’s voting system, we need not rely solely on the experiences of other countries. A great deal can be predicted from the incentives each system produces for parties, politicians, and voters. To the extent that self-interest motivates behavior, the incentives inherent in each concrete option will largely determine the impact of changing the electoral system or keeping it the same. In this brief, I discuss the likely benefits of adopting Dual Member Proportional (DMP), and explain why this system represents the smallest change necessary to rectify the undesirable incentives inherent in First Past the Post (FPTP).

2         Incentives under FPTP

Much attention is paid to the distorted results produced by Canada’s current voting system. However, in my opinion, the main reason to move beyond FPTP is the fact that its incentives contradict the fundamental principle of electoral democracy, and will likely lead to increasingly undesirable behavior on the part of parties, politicians, and voters.

2.1        Parties’ Incentives under FPTP

Parties currently have an incentive to conduct extensive polling, determine which voters will have no effect on the final seat distribution, and ignore those voters during an election campaign. If polling methods become increasingly sophisticated, a growing percentage of Canadians will be ignored by the national campaign teams. We can also expect platform information to be released at progressively later stages in a campaign when the most reliable polling data is available; this will allow parties to choose policies that target the smallest possible subset of the electorate with the potential to influence election results.

2.2        Politicians’ Incentives under FPTP

Politicians currently have an incentive to connect with voters, but only if they are affiliated with one of the leading parties in a battleground riding. Many candidates are aware well in advance of an election that they will not be elected, or that they almost certainly will be elected. In either case, there is little benefit in dedicating time and energy to the campaign effort. If polls become more reliable, a smaller percentage of candidates will be motivated to pursue public support.

2.3        Voters’ Incentives under FPTP

Many voters in safe ridings become aware, either consciously or subconsciously, that their opinions carry little weight for those in power. Furthermore, many voters in battleground ridings realize they can only command the attention of politicians if they consider voting tactically for a 2nd or 3rd preference. Improved polling will place greater numbers of voters in the first category where they have no genuine influence on how they are governed, or in the second category where they can influence policy but not in the direction they truly want.

2.4        Incentivizing Competition for Every Vote

The fundamental principle of electoral democracy is that every citizen of voting age has some ability to influence how he or she is governed by casting a vote. This only truly occurs when as many voters as possible have a chance to affect who gets elected and who does not. If every voter is given a chance of affecting election results, parties and politicians will have an incentive to compete for every vote. Yet the current system leaves many Canadians with essentially zero probability of producing any effect, and rewards parties for identifying and ignoring these voters. Faith in Canada’s democracy will degrade if advances in polling allow campaign teams to categorize greater numbers of Canadians as irrelevant for the purpose of winning seats.

Canada is more than ready to replace FPTP with an electoral system that incentivizes competition for every vote. To achieve this goal, it is essential that the Special Committee help Canada move responsibly toward proportional representation. Under a well-designed proportional system, every voter has a chance of impacting elections results through his or her influence on the popular vote.

3         The Smallest Necessary Change

Although a transition to proportional representation creates an opportunity to make a number of significant changes to Canadian elections and the associated form of governance, it is worthwhile to reflect on which changes are actually necessary. Let us focus separately on the districts, the ballot, and the formula.

3.1        Districts

It is widely believed that proportionality would require larger districts served by multiple MPs. For all intents and purposes, this belief is correct. So what is the smallest change necessary to achieve proportionality? One option is to introduce large regions served by a second tier of representations, but this is a fairly significant change. Another option is to have 5 or more MPs serving districts 5 times larger than the current ridings. This would also constitute a major change to Canada’s electoral system. However, there is another option: each district could be served by 2 MPs in districts that would be twice as large, on average.

Dual-member districts have a few obvious advantages:

  • Canadians would have their choice of which MP to approach with an issue.
  • With only 2 MPs, Canadians would still have a good grasp on who represents them.
  • With only 2 MPs, instead of 3, 4, or 5, there need not be too many MPs duplicating each other’s efforts by working on the same local issue.

3.2        Ballot

There seems to be a perception in Canada that proportionality would require more complex ballots with rankings or multiple votes. It is unfortunate that this false perception exists. Proportional representation works perfectly well with a single-vote paper ballot, and there are good reasons to keep the act of voting the same.

Here are the obvious reasons for Canadians to continue marking ballots with a single ‘X’:

  • Voting will remain as easy as under the current system.
  • Election workers will require little, if any, re-training.
  • Canadians who have only a single preferred party will not feel shortchanged by the introduction of the new voting system.

3.3        Formula

FPTP always elects the candidate with the most votes in each district. This need not change under a proportional system with dual-member districts. However, if one seat in each district is given to the locally most popular candidate, the second seat would have to be awarded in a compensatory manner to achieve overall proportionality. With a single-vote ballot, compensatory seats would be awarded according to the percentage of votes each candidate wins in his or her district.

This mixed compensatory formula has the following merits:

  • Voters in a district retain a sense of autonomy in that they continue to elect the locally most popular candidate.
  • The second seat is awarded in a manner that accounts for two performance metrics: a candidate’s local popularity, and also his or her popularity relative to other candidates affiliated with the same party.

3.4        Dual Member Proportional

The smallest necessary change is a proportional system with dual-member districts, a single-vote ballot, and a formula that elects each district’s locally most popular candidate. This exact system exists. It was invented in Canada, developed in detail, and thoroughly studied in a Canadian context using computer simulation. It is one of five voting systems to appear as an option in Prince Edward Island’s Fall 2016 plebiscite on electoral reform. The system is called Dual Member Proportional, or DMP1.

Table 1 compares a number of Canada’s most prominent options, including DMP:

System Districts Ballot Principle
First Past the Post 1-MP districts 1-Vote Majoritarian
Alternative Vote 1-MP districts Ranked Majoritarian
Single Transferable Vote multi-MP districts Ranked Proportional
Rural Urban Proportional 1-MP/multi-MP districts, multi-MP regions Ranked Proportional
Closed List MMP 1-MP districts, multi-MP regions 2-Vote Proportional
Open List MMP 1-MP districts, multi-MP regions 2-Vote Proportional
List Free MMP 1-MP districts, multi-MP regions 1-Vote Proportional
Dual Member Proportional 2-MP districts 1-Vote Proportional

Table 1. Comparison of voting systems based on district types, ballot type, and principle.

The option most similar to DMP is the List Free MMP system used in the German state of Baden-Württemberg. This system has a single-vote ballot, elects the locally most popular candidate, and awards the remaining seats in a compensatory fashion. However, these compensatory seats are associated with an encompassing region, introducing a second tier of elected representatives2. DMP keeps all MPs local.

4         The Benefits of DMP

There are limits to what can be achieved via electoral reform. In particular, it is difficult to support any claim that a new system will lead to economic benefits or higher voluntary voter turnout. Yet if we are willing to assume that parties, politicians, and voters will generally act in their own self-interests, we can make reasonable predictions based on the incentives that reside in the details of the electoral systems. From this perspective, DMP can be expected to yield a number of benefits over FPTP and other options in Table 1.

4.1        Tactical Voting

DMP will minimize tactical voting. As with most proportional systems, DMP reduces the incentive for a voter to favour a high-polling candidate over his or her preferred candidate. But unlike certain implementations of proportional representation, DMP avoids the creation of new incentives to vote in a dishonest fashion. Specifically, it avoids the 2-vote ballot of Closed/Open List MMP, where citizens can sometimes increase their likely impact by giving the district vote to a large party and the regional vote to a small party3.

4.2        Regional and Single-Issue Parties

DMP will provide reasonable protection against the diffusion of seats to regional and single-issue parties. The system incorporates a 5% district threshold, meaning that any candidate with less than 5% of the local vote becomes ineligible for a seat. Such candidates are unlikely to be elected anyway, since the formula favours those who win high percentages of the local vote. Yet the district threshold acts as a safeguard. Regional and single-issue parties must overcome the 5% barrier in a number of districts, as otherwise the seats allocated to them are likely to be forfeited for lack of an eligible candidate. The seats a small party loses due to the district threshold are re-allocated on a proportional basis.

The single-vote ballot provides an additional level of safety. Unlike Closed/Open List MMP, DMP would not allow an emerging party to ask Canadians for just one of their two votes. Similarly, unlike systems with a ranked ballot, a 1st preference could not be given to a small party based on the assumption that the vote would likely transfer to a 2nd– or 3rd-ranked major party. With DMP, giving any support to a regional or single-issue party would require conviction on the part of the voter, which is fair.

It is reasonable that a party with a low percentage of the popular vote receive less than its exact proportional share of the seats. However, a party should receive enough seats that its share of the power rises and falls with its overall vote count. In a DMP simulation I conducted by merging pairs of ridings using 2015 federal election data, a party with 3.4% of the popular vote ended up with 4 seats: fewer than its proportional share but enough that it would have something to gain and something to lose4. Majoritarian systems suppress small parties to such an extent that the decisions they make, whether popular or unpopular, yield neither reward nor consequence. All parties, both large and small, require incentives to act on behalf of the people.

4.3        Engagement

DMP will promote engagement. Similar to any other proportional system, DMP will intensify competition at the local level within districts currently occupied by safe ridings. Canadians who receive little attention under FPTP will become more engaged in politics as a result of parties’ increased efforts to win their support. Also, when compared with a ranked or multi-vote ballot, the one vote permitted by DMP may lead Canadians to feel more invested in the choices they make. A decision on whether to rank Party A before Party B or vice versa may well seem less consequential than a decision on which single party to reward with 100% of one’s support.

4.4        Partisanship

DMP will reduce partisanship. The elimination of safe ridings means that small advantages become valuable to all parties in every district. Nowhere in Canada will parties be content to nominate a candidate who merely totes the party line, knowing that a more honest nominee would attract greater support and raise the party’s share of the popular vote. Also, with two MPs per riding, many candidates would emphasize their willingness and ability to work across party lines at the local level.

When discussing proportional representation and partisanship, it is important to recognize that closed lists create an incentive for politicians to exhibit greater loyalty to their parties in the hopes of receiving a higher list position. Fortunately, Canada has many proportional options that exclude closed lists, and DMP is one of them.

4.5        Women

DMP will enable a substantial increase in the number of female MPs elected to Parliament. The increase may not be satisfying after the first or even the second election under DMP. Yet over a 10-15 year time frame, a significant improvement in gender balance is likely to take place as a result of adopting DMP.

The key to improving women’s representation is two-fold. First, the ballot structure must be altered such that every party can nominate more than one candidate. Second, the electoral system must be changed to a form of proportional representation so that the small advantage a party gains from including a female nominee is of value to the party in any district in Canada. By giving parties (a) the option of displaying gender diversity at the district level, and (b) a small but widespread incentive to do so, electoral reform can have a substantial long-term positive impact on the number of women in the House of Commons.

Nevertheless, proportional systems with multi-candidate ballots differ in the opportunities and challenges they provide for women. Under DMP, an aspiring politician of either gender would initially compete for a secondary position on the ballot. All else being equal, a candidate whose gender complements that of the more experienced primary candidate will have an advantage in obtaining the secondary position, as there would be a small incentive for every party to display gender diversity on campaign signs and pamphlets. In the first few elections, most of the primary candidates would be men, so female candidates would be favoured to win the secondary nominations. This post-reform advantage for women over men seeking first-time nominations is inherent in the system. Affirmative action is neither needed nor advised.

Although women would initially acquire secondary nominations more easily than men, most would have to advance into the primary ballot position before being elected. This is fair, since no incumbent should lose his position solely on account of his gender. Over 10-15 years, many incumbents will either retire or be defeated. Having gained name recognition in their local communities, secondary candidates—a fair number of whom would be women—will be well situated to take over the vacated primary positions. By producing a short-term incentive for parties to nominate more women, DMP will lead to a long-term increase the number of female MPs.

5         Conclusion

Parties, politicians, and voters generally act in their own self-interests. This is not a criticism, but rather an acknowledgement that desirable behavior will follow from a system that introduces the right incentives. Above all else, Canada should implement a system that incentivizes competition for every vote. This narrows the choice to a form of proportional representation, and DMP stands out as the proportional system requiring the least change. The adoption of Dual Member Proportional will produce a number of benefits, not least of which is a long-term transition toward gender balance in the House of Commons. I ask the members of the Committee to apply the knowledge they have gained for the benefit of the Canadian voter, and make the most of this opportunity to modernize Canada’s democracy.


  1. DMP is defined in a report by Sean Graham, titled “Dual-Member Mixed Proportional: A New Electoral System for Canada”, available at
  2. The Baden-Württemberg system has a second drawback in that the ballot features only one meaningful candidate per party. If Canada adopts a List Free MMP system, half the winner’s votes should be transferred to a secondary candidate, similar to DMP. This enhancement would help elect women, as explained in Section 4.5.
  3. Certain specific MMP implementations have been carefully designed to discourage tactical voting. I strongly recommend that a Canadian 2-vote MMP model be based on the Bavarian system, with modifications proposed by Wilfred Day (see
  4. The figures pertain to the Green Party of Canada, which is not a regional or single-issue party. Simulating the Green Party’s 2015 performance under DMP provides insights into the effect of a 5% district threshold on small parties in general.

1 Comment

Leave a Reply

Your email address will not be published. Required fields are marked *