Paul's Internet Landfill/ lj/ MMP 102: MMP vs. Parallel Systems

MMP 102: MMP vs. Parallel Systems

In a recent comment, cincinnatus_c_ noted his confusion as to how the party vote is used in Ontario's proposed MMP system. I made the same error myself for several months after getting involved with Fair Vote Canada, so maybe it is worth clearing up the distinction now.

MMP is a compensatory system: the party vote is used to determine the overall share of seats in parliament. There is a related system (usually called "parallel" but sometimes called "non-compensatory MMP") (Cody 2007) which is different: the party vote is used to determine the list seat allocations only, independent of the riding outcomes.

A numerical example might be helpful here. Consider a hypothetical system which has 100 riding seats and 100 list seats, using a two- checkmark ballot where people choose their local representatives and their preferred parties separately. Let's say that there are four parties involved, and calculate seat totals under both systems. Say that the initial number of ridings and party vote work out to the following:

Party # Ridings % Party Vote
Avocado Party 60 40

Banana Party | 25 | 30

Celery Party | 15 | 20

Damson Party | 0 | 10

Under a parallel system, riding results are ignored in allocating list seats. So we would get the following results:

Party | # Ridings | % Party Vote | # List Seats | # Total Seats | % Total Seats
Avocado Party | 60 | 40 | 40 | 100 | 50.0

Banana Party | 25 | 30 | 30 | 55 | 27.5

Celery Party | 15 | 20 | 20 | 35 | 17.5

Damson Party | 0 | 10 | 10 | 10 | 5.0

Now let's redo the calculation using a MMP system, which takes the ridings into account when determining number of list seats:

Party | # Ridings | % Party Vote | # List Seats | # Total Seats | % Total Seats
Avocado Party | 60 | 40 | 20 | 80 | 40.0

Banana Party | 25 | 30 | 35 | 60 | 30.0

Celery Party | 15 | 20 | 25 | 40 | 20.0

Damson Party | 0 | 10 | 20 | 20 | 10.0

See the difference? Under MMP we use the party vote to determine the total number of seats the party deserves, and subtract off ridings from that total. In parallel systems we don't care, so even parties that win more than their fair share of ridings get lots of list seats.


Because the proportion of list seats is so high, this example makes the two systems look fairly close. But we can still see that parallel systems don't need to be proportional -- in fact, they continue to give the biggest party a boost in seats. Because they don't lock out small parties entirely, some people argue that this is a reasonable compromise between full proportionality and the distortions of FPTP. (The blood pact Fair Vote Canada made me take means I am not one of those people; any deviation from the dogma that Proportionality Is Sacred is punishable by dismemberment, excommunication and/or being fed to Xenu.)

Aside from the disproportionality issue, parallel and MMP systems differ in a subtle but crucial way: parallel systems pretty much guarantee quite a lot of list seats for bigger parties, while MMP does not.

Consider a situation like the 1993 Federal election, when the voting system crushed Kim Campbell's PC party. The party received 2 riding seats, but 16% of the popular vote. Under a parallel system the party would have won 16% of the list seats. That is fine, and similar to the results under MMP. But when the voting system gives the PCs a lot of ridings as they did in 1988 (43% vote, 169 seats) a parallel system would win the PCs even more list seats. Under MMP, the increased number of ridings could wipe out the list seats entirely. Unless a party collapses in the party vote, under a parallel system it can guarantee that it will win a certain number of list seats in each and every election.

This has big implications for accountability. It means that so long as a politician can stay near the top of the party list, he or she does not have to worry much about being accountable to voters. They don't even have to run in local ridings if they so choose. This can create an "elite class" of politicians, and this has been a problem in countries that use pure-list and parallel systems. I have read complaints about Japan and Spain, but there are probably others.

MMP does not avoid this problem entirely. Medium-sized parties will tend to win a lot of their seats through party lists, and those seats will be more-or-less safe. But bigger parties are volatile; perhaps the top couple of seats in every party will be safe, but the better parties do in ridings the fewer list seats that are available, so if politicians want to stay in power they had better run both on the list and in their ridings. Meanwhile, small parties are very volatile, because they run the risk of falling below the 3% threshold and losing all of their list seats. Overall, there are a lot fewer list seats that are "safe". That does not prevent career politicians from bouncing between ridings and list seats, but in practice I do not see this as being a lot worse than the situation we have now under FPTP, which uses gerrymandering and big spending to keep the big names in power. Neither of the voting systems up for grabs in the election does a lot to prevent career politicians.

Another big difference has to do with how list seats are allocated. In parallel systems, lots of list seats will usually go to members of the ruling coalition (in particular, to the biggest coalition partner). Being members of the biggest coalition party gives these list people opportunities for government power (for example, cabinet seats) that members of other parties don't have. That does not mean that these people get all the cushiest positions to themselves -- in Japan the perpetually-ruling Liberal Democratic Party tends to save the cushiest jobs for potential leaders and those dual-candidates whose ridings are at risk (Pekkanen Nyblade Krauss 2006) -- but members of the ruling party certainly get more opportunities than those in other parties.

In MMP very few of the list seats will go to members of the biggest coalition partner. Members of the smaller coalition partners will often come from lists, as will many of the seats for parties not participating in the coalition. Certainly the leaders of smaller coalition partners will be in a position to hold considerable power (as Winston Peters does in New Zealand) but most other MPPs will be limited in the power they can amass. Those list politicians will tend to spend more time on committee work than constituency work, but they tend to do some constituency work to "hedge their bets" in the hopes of getting elected to riding seats (Heitshusen Young Wood 2005). Incidentally, the people who hold riding seats hate this, to the point where the Labour party in Wales (which wins about 75% of the riding seats) introduced a proposal to eliminate dual-candidacy to dissuade list members from challenging riding members for power (Lundberg 2006).

Having said all this, it is important to remember that the choice in this referendum is not between a parallel system and MMP (or your favourite voting system vs. the Ontario proposal). The choice is between first-past-the-post and the Ontario proposal. My hope in this entry was to explain the difference between parallel and MMP systems so that you can evaluate MMP on its own merits and faults, and you will be able to evaluate the criticisms people make of countries that use parallel systems (and list systems in general) in an appropriate context.


Canadian federal election results are from Wikipedia: Canadian_feder al_election,_1988, Canadian_federal_election,_1993.

(Cody 2007): Howard Cody. "Recent New England and Canadian Elections and Electoral Reform in Canada", New England Journal of Political Science, vol 2, no. 2, pp. 122-152.

(Pekkanen Nyblade Krauss 2006): Robert Pekkanen, Benjamin Nyblade, Ellis S. Krauss. "Electoral Incentives in Mixed-Member Systems: Party, Posts, and Zombie Politicians in Japan", Amercian Political Science Review, vol 100, no 2, May 2006, pp. 183-194.

(Heitshusen Young Wood 2005): Valerie Heitschusen, Garry Young, David M. Wood. "Electoral Context and MP Constituency Focus in Australia, Canada, Ireland, New Zealand, and the United Kingdom", American Journal of Political Science, vol 49, no 1, January 2005, p. 32-45.

(Lundberg 2006): Thomas Carl Lundberg. "Competition between Members of the Scottish Parliament and the Welsh Assembly: Problem or Virtue?", The Political Quarterly, vol 77, no 1, January-March 2006. p. 107-117.

Livejournal URL:

Mood: hurting