BC electoral reform options for referendum

[Correction made, re the MMP proposal]

The electoral-reform process in British Columbia has advanced another step with the Attorney General’s release of recommendations. The existence of this process is a product of the deal struck after the 2017 election, which resulted in a minority NDP government, backed by the Green Party.

In either October or November, there will be a province-wide vote consisting of two questions: First, do you want to keep the current FPTP system or replace it with a proportional system? Second, which of three PR options would you prefer if the province were to adopt PR? The second part will permit voters to rank-order their choices.

The choices being put to the voters are (using the names given in the recommendations):

Dual-Member PR (DMP)

Mixed-Member PR (MMP)

Rural-Urban PR

I will take the three in order of most familiar to least.

Readers of this blog are probably generally pretty familiar with what MMP is. The short version is: a system in which some percentage of seats (the proposal is for no more than 40%) are elected as “compensatory” list seats, while the rest continue to be elected as under FPTP, in single-seat districts (ridings) by plurality. The key feature is the compensatory nature of the process, so that any party’s number of seats in the provincial assembly are roughly proportional to its province-wide votes. Any seats the party wins in individual ridings are deducted from this total, and remaining seats a party is entitled to hold under the proportionality principle come from the list. It should be noted that the report says “List PR seats are to be allocated within defined regions, not on a province-wide basis.” This could limit proportionality somewhat, but this proposal is likely to be the most proportional of the three (unless perhaps if the compensation regions are quite small). However, it also says “The overall share of seats each party holds in the Legislative Assembly is determined by the party’s share of the province-wide vote it receives.” Thus the regional allocation refers only to which candidates get the seats and where, not to how many seats in the assembly each party wins.

Rural-Urban PR is mostly single transferable vote (STV), but in the rural parts of the province MMP would be used. This is an unusual system in that it actually is two electoral systems, depending in where you are in the province: STV in urban and “semi-urban” areas, but MMP elsewhere. Under STV, voters rank candidates, but under MMP it seems that it would remain plurality. The logic is to prevent rural ridings from being significantly larger than they are now–one of the concerns raised with “BC-STV” when it was proposed in the province’s previous electoral reform process. The provision for MMP compensation regions in rural areas is obviously an effort to allow for proportional representation even in the areas where the districts will be single-seat. The proposal suggests district magnitudes for the STV regions of 2-9, with a preference for the higher end of that range.

Both MMP proposals–the full province-wide one or the rural component of Rural-Urban PR–could have either a single vote or two votes. The proposal explicitly leaves this (important!) detail to the legislature after the referendum. If there were one vote, then votes for candidates in single-seat ridings would be aggregated by party for the purposes of carrying out proportional compensation (which, for Rural-Urban, would be taking place only over the regions in which the single-seat districts were located). With a two-vote MMP (as found in Germany and New Zealand), voters can vote directly for a list of their preferred party, and thus vote for a party that is different from that of their preferred local candidate. Yet another feature to be left to the legislature to decide would be whether lists would be closed (as is typical for MMP systems), open, or flexible (also known as semi-open, but misleadingly called in the recommendations “open list with party option”).

DMP is a system not actually used anywhere (but see the earlier P.E.I. proposal), and it is a bit complex. In general, each district would elect two members, although the recommendations allow that the “largest rural districts could remain unchanged as single-member districts.” Each party could nominate two candidates, and they would be ranked by the party. Thus we have a closed-list system. Voters would cast a single party vote. The first seat in each 2-seat district would be won by the first-ranked candidate on the party with the most votes in the district. The second seat would actually be allocated based on province-wide votes. It is thus a two-tier compensatory closed-list PR system. How would the assignment of the compensatory seats to districts–given that there is no separate list or compensation district–be done? The proposal says only, “The process for allocating the second seat in each district is fairly complicated.” (Perhaps it would be something like the provisions in Slovenia or as an option for parties in Denmark.)

DMP ensures that all candidates, even those elected on province-wide votes, would represent a riding. It also ensures that one member in each 2-seat riding is elected based on the riding’s own votes. (Exception: it seems there would be a 5% threshold for any party to win seats under any of the list-based provisions being proposed; if that is correct, then it is conceivable that a party could win a plurality of votes in some riding but not be entitled to a seat.)

A puzzling aspect of DMP is the one on independents. If an independent places first in a district, that candidate is elected. That is straightforward enough. However, it is also the case that if an independent places second, that candidate is elected and “the district is removed from the remainder of the second-seat allocation process.” I don’t understand the logic of that provision.

Each proposal allows for the assembly to be increased from its present 87 members up to a maximum of 95, but does not require that the assembly size be increased.

An observation: Why not do DMP with open lists? Have the voter vote for a candidate, rather than a party. It would be a lesser break with FPTP while still being quite proportional. A potential answer: it could mean the candidate with the most personal votes is not elected (because that candidate is on a party that is overall less popular than another in the district). That could be addressed by making it more like MMP–the leading candidate wins the first seat, but that seat is deducted from the compensation entitlement. Otherwise, the DMP provisions would apply. It is interesting that the DMP proposal is explicitly closed list, whereas the list type in the MMP variants would be left up to the legislature.

Any of these systems would seem like a clear improvement for BC. Rural-Urban is an odd mix, but it could work. MMP is proven. DMP is unusual but not based on wholly unknown principles.

(For further reading, see On Elections or Sightline)

32 thoughts on “BC electoral reform options for referendum

  1. Do we know whether the MMP top-up regions would be the whole province taken as a unit or some subdivisions of the province? Also, would the top-up provision apply only to rural seats or not? I have seen bits of information in each direction.


    • MMP top-up regions, yes. “List PR seats are to be allocated within defined regions, not on a province-wide basis. No region will have fewer MLAs than it does currently.” So it would apply equally in all regions, similar to Scotland, 40% top-ups. Except I believe it will be open list, otherwise much like Scotland.

      Liked by 1 person

  2. I’ve read in several places that on the second part of the ballot voters will rank the alternatives in order of preference. Has it been explicitly stated whether the votes will be counted by IRV/alternative vote rules?


  3. I actually like “open list with party option” as a term. Fair Vote Canada’s Statement of Purpose says ” A democratic voting system must encourage citizens to exercise positive choice by voting for the candidate or party they prefer.” Why shouldn’t “open list” allow voters the right to choose the party’s list rankings?


    • Let’s keep “open list” as a term meaning strictly a list in which final ranks are determined exclusively by preference votes obtained. Many such systems have a “party option” in the sense that the voter can vote for the list without indicating a candidate preference, but the key thing is that such votes only help the list overall, while defaulting the order of candidates to other voters who cast preference votes on the list.

      If the “party option” is one that accepts a pre-election list ranking, and final ranks are determined via some combination of those ranks and preference votes, that is a flexible list. Also known, perhaps more sensibly, as semi-open. Even under flexible/semi-open lists, it is possible to require candidate preferences (as in the Netherlands) or to make the preference voting optional (as in Belgium).


      • “Flexible” and “semi-open” are terms geeks know, but they are far from self-explanatory. When you’re explaining, you’re losing, as many have said.

        Yes, “open list with party option” might mean the model used in Brazil and elsewhere, where the vote for a party does not affect the ranking. However, I have seen no one in Canada advocate such a model. When, for example, the Quebec Director-General or Elections described flexible list favourably in December 2007, he said it was intended to get the advantage of both voter choice and also allowing an alternation between female and male candidates and of placing candidates from minorities or the regions in a favourable position. So anyone in Canada seeing the term “open list with party option” would assume the “party option” included the party’s rankings.


      • “Semi-open” seems pretty straightforward.

        I completely sympathize with the broader point that what scholars use as precise terms for analysis (or at least what should be precise terms) may not be the same as what advocates need to use. But we do need to be sure here–in a blog mainly devoted to analysis and comparison–not to get terminology confused with terms that mean different things elsewhere.


  4. The provision on independents in DMP does sound troubling and could be open for abuse, though probably less doable than in a two-vote MMP system.

    I would be surprised if a party that wins a plurality of votes in a district but falls below the province-wide threshold is not allowed to keep the seat. Such is one of many pitfalls of designing a new electoral system: you neglect to address some potential complications and exceptions that might arise.


  5. The order of allocation of DMP list seats is based on the percentage of district vote received by a party’s unsuccessful primary candidates(*). It might be viewed as a closed list where the order of selection from the list was decided by a candidate’s performance against district opponents, rather than decided by the party. Each candidate would be in indirect competition with all the other losing candidates of his/her party (in contrast to open list MMP where the competition is direct).
    Proponents of DMP do tend to describe the allocation process in a way that is complicated (more complicated than is necessary, in my opinion).

    Giving a seat to an independent who finished second indeed is puzzling. Also, it might tempt a major party to run an “independent” candidate in all districts where polls indicate the party is running second.

    (*) Plus the running mates of successful candidates, through a provision for transfer of excess votes. They are unlikely to beat out losing primary candidates for a seat unless the party carried the district by a huge majority.


  6. Looks like the creator of DMP has a website which sets out his version of the system: https://dmpforcanada.com/

    As described there, the allocation of second seats is
    1) calculate the number of second seats per party (subtracting from overall allocation calculated through quotient/remainders)
    2) create a “reserve” of some amount (e.g. 10%) of each party’s seats
    3) in each riding, halve the vote percentage for the party winning the first seat and carry over the other parties’ percentages (i.e., use the d’Hondt quotient applied to vote percentage)
    4) allocate second seats (other than “reserve”) to parties based on their top quotients; resolve conflicts by allocating the riding to the party with highest quotient
    5) allocate “reserve” seats in the same manner, in the remaining ridings

    The “reserve” seat concept is designed to address the problem created by the allocation method in which smaller parties could get shoved into ridings where they performed poorly. However, it seems like a better solution to this might just be to allocate the second seats to parties in the order of their province-wide d’Hondt quotients and/or add a regional component (as in Slovenia).

    The independents clause appears to follow similar logic: an independent wins the second seat if (s)he 1) places second *and* 2) didn’t get doubled up–i.e. would be elected under d’Hondt. (Presumably, two independents could be elected by doubling the second-place party, though I don’t think that’s explicitly stated). Not sure why the AG’s report omitted the quotient component.


    • Would it really help to award seats in province-wide d’Hondt quotient order? A party with a 6% of the vote could still easily end up winning the very last seat, which would be in a random district.


      • I believe that assuming all (qualifying) parties receive seats in the second-seat distribution (i.e., no overhang), a party’s chance of receiving the very last seat distributed (in order of provincial d’Hondt quotient) should be equal to its percentage of the vote among qualifying parties. A party receiving 6% of the total vote for qualifying parties would have a 6% chance of winning the very last seat, which would most likely be awarded to a large party. (Additionally, if there is a 4 or 5% provincial threshold, no party would win just one seat.)

        Using the overall d’Hondt order (or some other divisor) after accounting for the district seats won would generally help the smaller parties by pushing them toward the front of the line, assuming they didn’t win the first district seats and thus haven’t used up their largest quotients.


  7. I actually favor something like DMP for the United Kingdom, because before 1885 most boroughs elected two MPs. So there is a tradition of electing two MPs and you can use that to import proportionality into the tradition-minded English politics, along with adopting the county MPs.

    It also helps in the sense that most people probably prefer having a representative or two in the legislature that they can call their own, not half a dozen, something often forgotten by proportional representation advocates. But STV with low district magnitude is OK in that regard. And MMP provides for district representation of course, but with the ugly provision of appointed legislators who represent nothing.

    There is no reason to do something like DMP for Canada, which has no tradition of two representatives per riding. Canada does have the issue of geographically vast areas with small populations. Though I wish they could have a better name, it seems that “STV except for vast areas with low population densities”, or the “rural-urban” proposal, would work best. I’m glad they are giving voters a choice between four (including the current single member plurality system) workable systems.


      • My district of Vancouver-Point-Grey elected 2 members as recently as the election of 1986
        However, it was an exception, a relic preserved because its block voting helped the government of the day. So not a helpful example when explaining DMP.

        P.E.I. was a much different case. Dual member districts were the norm until the 1990’s.


    • DMP made sense only in PEI, where they had two-MLA districts until 1996. But it lost to MMP. So I consider it off the table, notwithstanding its apparent resurrection in BC.


      • I’m not sure that DMP has a lot going for it. Fructovotantes may be over-excited because it appears to be am entirely new system. A system that entitles most electors to cast a fully preferential vote but restricts remote electors to a non-preferential vote, and which then elects the second member by a slightly counterintuitive method is an invitation to FUD and a nightmare for electoral education.

        Liked by 1 person

      • On the other hand, above Dave says that some BC ridings elected two as of 1986. I had forgotten that. Still a full decade farther back than in PEI, and it was not the case throughout the province, so maybe not too relevant to current debates.

        Regarding Alan’s comment, I think all the “DMP” has going for it is its NOT being “entirely new” (despite being packaged as such). That is, it seems to me that it is based on some existing principles–it is two-tier, and the compensation seats get assigned on the basic of district results (or, relative to province-wide, they are really “sub-districts”).

        Note that there is not actually a preferential vote: the vote is for party, with no choice among the two candidates a party might have nominated in the district. It may be that it would work de-facto as a candidate vote, inasmuch as the district is so small and it is the plurality party winner that gets the first seat. But that is the case under standard closed-list PR with low magnitude anyway (a key point of Carey and Shugart, 1995).

        So it is a clever way to combine two basic principles: low-magnitude and full compensation. Note that “clever” is not an endorsement. But it is not something wholly new.

        Liked by 1 person

      • If DMP is viewed as a clever way to have full compensation with district magnitude M=2, should we view Balinski’s Fair Majority Voting, full compensation with M=1, as super-clever?

        Liked by 1 person

  8. I wonder if the fabled “biproportional apportionment” could be relevant in a method by which the second seats in the DMP system are allocated, thus providing a solid theoretical foundation to the system, to quash any doubts that could be raised (whether disingenuously or otherwise) about its fairness in the referendum campaign.


  9. Not to rain on the electoral reform parade, but we should not lose sight of the fact that in the fairly recent past Canadian politicians have shown time and again their willingness to go to great lengths in order to frustrate electoral reform attempts, so far quite successfully. From that perspective, it seems to me that in order for this later effort to be successful and long-lasting, reform backers should strive for three goals:

    1) A high voter turnout rate in the referendum: no less than 50%+1 of the electorate, but preferably a larger figure.
    2) A substantial majority for reform in the first part of the ballot, of at least 10% of the vote.
    3) A majority of all votes (not just valid votes) for the chosen alternative system in the second part of the ballot.

    While failure to achieve any of these goals might not prevent the enactment of electoral reform – particularly in light of the fact that the vote will be binding – it could set the stage for a subsequent attempt at repealing reform in the near future. In fact, we all know from B.C.’s previous attempt at electoral reform that turning around public opinion on the issue at hand is not that difficult.

    Incidentally, it should be noted that New Zealand’s 1992, two-part voting system referendum achieved each and every one of the stated goals. On the other hand, low turnout in P.E.I.’s 2016 electoral reform referendum was seized upon by authorities there as a pretext to ignore its outcome. Beyond Canada, that was also the case with Puerto Rico’s 2005 non-binding vote on switching to an unicameral legislature: the pro-statehood New Progressive Party (PNP), which controlled the Commonwealth House of Representatives (and for the most part had boycotted the event), refused to enact legislation for a subsequent binding referendum on the grounds that a 23% turnout was too low to be acted upon.

    (On an aside, that decision came back to haunt PNP leaders twelve years later, when an overwhelming majority of voters backed U.S. statehood in a plebiscite, but an opposition boycott resulted in an identical 23% turnout; I can only comment that evidently karma is something that rhymes with rich.)

    But back to Canada, my concern about the third goal – as noted on a previous comment on a related thread – lies in the possibility that a large number of FPTP backers might refuse to choose an alternative to the status quo (much in the same way as with MMP backers in New Zealand’s 2011 referendum or Commonwealth supporters in Puerto Rico’s two-part status plebiscite of 2012), thus resulting in a large number of blank and invalid ballots, which in turn might deprive the winning reform option of a majority of all votes, even after the distribution of preferences. Once again, this turn of events might not be in and of itself an obstacle to enacting electoral reform, but it could be seized upon by its opponents as justification for a subsequent repeal attempt.

    Now, just to be clear I’m not bringing this up to discourage the discussion on the proposed alternatives to FPTP, but to caution that it may come down to which sandcastle is the prettiest, while ignoring the incoming wave that may wipe out all of them.


      • A good argument for having an explicit support quorum (as distinct from a turnout quorum*) in referenda.

        * “Support quorum” = eg, “more than [50]% of persons actually casting valid ballots, as well as at least [33.33]% of all persons entitled to vote, must cast valid ballots in favour.” Whereas “turnout quorum” = eg, “more than [50]% of persons actually casting valid ballots must cast valid ballots in favour, and also at least [50]% of all persons entitled to vote must cast valid ballots either for or against.” The latter is non-monotonic and therefore very flawed: if a proposal would lose if the electorate votes 49% yes, 40% no, 11% abstain, you could help it win if you persuade 1 in 11 abstainers to turn out and vote no instead. Topsy-turvy incentives = bad institutional design.


      • … and having it clearly specified in advance, and making it clear (behind a Rawlsian veil of ignorance of the actual voting results) that exceeding both hurdles is necessary and sufficient to pass the proposal.


  10. I prefer MMP to all of these options, it seems as if electoral reform in candidate is like the miracle of the unicorn, as hip and progressive as Canada, Canada is a very conservative country when it comes to Constitutional and Electoral issues, something that is not likely to happen unless it passes by accident like what happened in NZ and the Brexit referendum, they were to make the issue go away and it backfired. It is a bit sad that STV is not being considered, why not STV with a top up, with the first preference vote for a party for the adjustment seats using best loser.

    If FPTP is retained, does this mean that that this is the end of the NDP minority government?

    Is PR being considered for local government as well?


    • I have not heard of any active discussion of local government electoral reform. But Vancouver is having elections soon, and the MNTV system for city council is a fairly extreme case: Citywide, vote for up to 10.

      Liked by 1 person

  11. At the 2017 general election combined spending by the BC Liberals, New Democrats and Greens was $22 409 932. The proposed referendum campaign limits, which also apply to political parties, are $200 000 per campaign. So a campaign group is limited to spending approximately 1/5 of what the Greens were able to spend on the general election. There will be two officially designated campaigns with $500 000 to spend. These are extraordinarily tight limits. They replicate the campaign limits that applied to the second STV referendum in 2009.

    Canadian political elites appear to have mastered the art of Bourbon electoral reform, where they forget nothing, they learn nothing, and somehow the electoral system never, ever changes.

    Liked by 1 person

  12. Pingback: What could we expect from electoral reform in BC? | Fruits and Votes

  13. Pingback: Open lists in MMP: An option for BC and the experience in Bavaria | Fruits and Votes

  14. Pingback: Correction on BC’s MMP proposal | Fruits and Votes

  15. Pingback: Is rural-urban PR gaining on MMP? | Fruits and Votes

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.