What if Japan used MMP?

The following post is by Professor Michael Thies of UCLA. [Corrected since initial posting.]
One weirdness of MMP is what to do when a party wins more seats in single-seat districts than its PR vote share would have earned. A few “overhang seats” are easy enough to deal with, but I wondered how last month’s Japanese election results would have looked under MMP (with the dubious assumption that nothing else changes).

If we simplify and assume nationwide PR, and use the PR vote shares that each party actually earned in the 16 December election (1st column of the table below) for all 480 seats, the 2nd column shows the seats “earned.”

If this were Germany, with overhang seats, the LDP would get to keep all 237 SMD seats (not 294 combined total that it actually received, because it would get no PR seats), and the legislature would have to grow to 584 seats. Of course, if overhangs were not part of the rule, the LDP would have 27.6 percent of the seats instead of the 61.3% they do have. This way, LDP-Kom would be well short of a majority (133+57)/480 = 39.6% w/o overhangs, and with a slim majority with overhangs:  (237+57)/584 = 50.3%.

Japan 2012 MMP scenarios

11 thoughts on “What if Japan used MMP?

  1. I don’t see how you could (legitimately) remove overhang in MMP, as you would have to deprive MPs of constituency victories. What you can do if you have to keep the house at a certain number of seats, is use the British version of MMP, but this of course only strengthens the effect of overhang at the expense of other parties.


  2. It seems bizarre that without overhang seats the LDP/New Komeito coalition is without a majority, and with overhang seats, they have a bare majority.

    Does anyone have any history on why Japan ever used the Single Non Transferable Vote? What electoral system did the country use before it’s Pacifist constitution?

    When Japan did have electoral reform, why did it choose a MMM system? Was it possible for Japan to move from SNTV toward STV? It doesn’t seem logical to me for a country use toward semi-proportional system using multi-member districts moving toward a system that mixes plurality single member districts and regional list allocation with multi-member districts.


  3. Suaprazzodi @2,

    > “Does anyone have any history on why Japan ever used the Single Non Transferable Vote? What electoral system did the country use before its Pacifist constitution?”

    My recollection of reading Hoag & Hallett (pub 1922-23) is that Japan was using SNTV before WW2 and even, I think, since the Meiji era. Taiwan (and pre-Communist mainland China) have used SNTV, as did South Korea, I believe. The short-lived Republic of South Vietnam had a Senate where 30 seats were filled each election by the three 10-candidate slates with the highest vote totals. Thus intra-party SNTV seems to be the norm in East/ South Asian democracies.

    The Encylopedia Britannica observed in the 1970s that, in elections with free voting, SNTV gave results that were “very close to proportional”, but omitted to mention this was because of rigid vote-management by the political parties. As Afghanistan is shown, without party machines on the ground telling voters “surnames A-G vote for Candidate #1, North and east wards vote for Candidate #2…”, SNTV can elect candidates with very widely varying levels of support. Japanese electoral law in the SNTV era set a minimum threshold (I’m not sure what it was, just that it was substantially below a Droop quota of the votes) below which a runoff was required for the unfilled seats, so technically it wasn’t “pure” SNTV.


  4. I think Mike’s “MMP seats” column simply means what the full nationwide PR allocation would be.

    I agree with you, JD. And in addition to Scotland, Wales, and London, you also have uncompensated overhang in Bolivia and Venezuela.


  5. Another question to ask is; Is Japan’s MMM system working well or would Japan be wise to consider a different electoral system?

    Which system of PR would you recommend for Japan to use if it underwent another electoral system change?

    Would closed party list PR eliminate factionalism with in parties?


  6. I’ve recently been doing some investigating on electoral coalitions in PR countries (can’t say Parliamentarian because countries like Canada and the UK don’t use PR at the national level).

    I always wondered what if voters were given a ballot and on that ballot they were given a full list of possible outcomes, where voters can vote for a coalition of 2 or more parties or vote for a single party. I know more details are needed but I guess a system like this would be very interesting, especially since the voters would decide who gets in!


  7. Japan began using SNTV in 1925, at the same time as they introduced universal male suffrage. The usual story is that there were three big (all conservative) parties at the time, so they chose 3-seat districts (mostly) so that each could win a seat. The first postwar election, in 1946, used a different system, with much larger districts, and the Socialists finished in first place. Yoshida Shigeru then convined MacArthur to revert to SNTV.

    Lots of folks have written about the choice of MMM in 1994. The most important lesson is that it was nobody’s actual choice, but a compromise among parties with divergent interests. I think that the LDP preferred MMM over MMP because they were still by far the largest party, and figured that they’d benefit most from the SMD portion, while keeping the opposition divided in the PR tier. And, actually, that’s precisely what happened in 2012.


  8. @3, the South Vietnam model was not SNTV, but rather plurality at-large (voting for lists rather than individuals), at least according to Wikipedia and the IPU (http://www.ipu.org/parline-e/reports/arc/VIETNAM__REPUBLIC_OF__E.PDF).

    In 1967, voters had 6 votes each to cast for one of 48 10-member lists, and the 6 lists earning the most votes were elected. Apparently, many voters did not use all 6 votes (the average voter cast 4.5 votes). Over 82% of votes went to lists which were not elected.

    In 1970, half the senate was up, with voters each having up to 3 votes for 10-candidate lists, and the top 3 getting elected (no idea how they chose which 30 were up for election).

    In 1973, voters could cast up to 2 votes for 15-member lists, with the top two getting elected. Only 4 lists contested this election


  9. … And some older texts weren’t all that accurate. I distinctly recall one 1960s-era children’s encyclopaedia stating that cumulative voting was used to elect “several US city councils, along with the French National Assembly”. Descriptions of IRO-AV in Australia also tended to be a bit misleading.


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 )

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.