Finland’s election–and coalition bargaining (updated)

Revised and extended version of the earlier planting (most recently on 21 March)

Another squeaker

Or should I say photo Finnish?

The Centre Party won 23.1% of the vote, the National Coalition 22.2% and the Social Democrats 21.5%.

The current Prime Minister Matti Vanhanen is the head of the Centre Party, which will have 51 of 200 seats. The conservative National Coalition is not part of the current governing coalition, but is now one seat behind the Centre.

Compared to 2003, the Centre Party lost support, even though it remains the largest party. The Social Democrats fall from the 53 seats they had won in 2003 to only 45 now. Thus it looks like there could be a realignment of the Finland’s coalition government towards the right.

Analysis: Government possibilities

Given that a Centre-National coalition would have exactly one more than half the seats, a narrow center-right government could form. However, that does not mean it will form, and there is a fairly strong tendency of Finnish governments to be greater than minimal winning (and you can’t get more minimal than 101 seats out of 200). Therefore, it is not a given that a rightward shift of the coalition will result. See Michael’s more-detailed analysis, in which he notes that two other leftish parties combined for 32 seats and two others of the right for just 12.

Analysis: Impact of (not-quite fully) proportional representation

It’s worth noting that the proportionality in Finland is not calculated nationally, but rather in a series of regional multi-seat districts of varying district magnitude. Thus a party with an optimal geographic distribution does slightly better than one that may have a similar nationwide votes share but less optimal geographic distribution.* (See Alex’s comment for more.)

Look at the advantage ratios (% seats/% votes) for the top three parties:

    Centre, 1.104
    National, 1.126
    Social Dem, 1.047

These are not huge differences, and would be almost trivial compared to what we see in less proportional systems. However, in such a close election, they matter. In fact, the Centre-National minimal winning coalition is possible only because those two parties were somewhat over-represented, compared to the Social Democrats.

If all three of the leading parties had had the same advantage ratio as the Social Democrats, the seats would have been a bit different: Centre (48), National Coalition (46), Social Democrat (45), instead of 51-50-45. The Centre-National coalition would not be feasible.

I do not know Finnish politics well enough to predict the result of the bargaining that will now follow. However, the advantage that the National Coalition (and, to a lesser extent, Centre) obtained from the electoral system gives likely Prime Minister-designate Vanhanen and his Centre party leverage over the Social Democrats that it otherwise would not, even if Vanhanen ultimately reconstitutes his center-left coalition.

In very close elections, even very small deviations from proportionality matter. If the Social Democrats had won the most votes by the same narrow margin as they trailed in the actual results, but with the parties’ having their same respective advantage ratios, we would have had a plurality reversal in the seats relative to the votes. ** That did not happen here, but it was close, and the actual differential treatment of the parties by Finland’s PR system may yet affect the coalition result.

* The differential treatment of the Finnish parties appears to be another case of the bias introduced against predominantly urban parties by magnitude variance. The bias results when one set of parties (usually conservative) are especially strong in rural areas that have lower district magnitude, and thereby benefit from the lesser proportionalityof votes-to-seats translation in those districts. On the other hand, the (usually leftist) parties that are strongest in urban areas with very large district magnitudes do not get the same sort of bonus out of their own strongholds. Meanwhile, the “rural” party has a (minority) constituency within the large urban districts, it gets proportionally repersented there, thanks to the large magnitude. (The “urban” parties get under-represented in the smaller districts even if they have a comparable minority share of the vote in such districts, which they may not.)

Of course, this bias, while real, is nothing like we see in plurality systems.

(Added 21 March: See Alex’s comment on the geographic distribution. Unlike me, he actually bothered to look at the district-level data and saw that the National and Centre did quite well in several large districts. However, the higher advantage ratios for these parties can’t result only from large districts, given the inherently greater proportionality of such districts. So, there is clearly more to the story. As Alex also reminds us, the provisions in the Finnish electoral law for inter-party alliances (vote pooling across parties within a common list) are undoubtedly also part of the picture.)

** Excuse me for almost wishing something that interesting had happened.

0 thoughts on “Finland’s election–and coalition bargaining (updated)

  1. Finland sure has a very unusual party system where all three parties are neck, neck, and neck. It is quite possible for any of these three parties to lead a government. In most countries, there are always two large parties that are always neck and neck.

    Is Finland unique in having a three-party system like this? The Finnish electoral system is one that uses open party list proportional representation in multi-member districts. The smallest district elects 6 and the largest 33 depending on the constituency’s population. That means the electoral threshold is much higher by using multi-member districts than it would be if the whole country was treated as a single district like in Israel or the Netherlands.

  2. Actually, the SDP is the party best represented nationally, but doesn’t “star” anywhere. The big cities were won by the Coalition, not SDP, in this election. Running the data, what I gather is that Center (KESK) and especially the Coalition (KOK) are favored by d’Hondt in the large constituencies that they won by big margins.

    For example, in Uusimaa (34 seats), KOK won 1.3 extra seats (were quotients being used, one of their seats would have gone to the Christian Democrats). In Helsinki (21 seats), KOK won 1.7 extra seats, taking 8 of 21 (38%) on 30% of the vote. Again, were quotas being used they would lose a seat. However, the Conservatives are not nearly as popular in the north.

    KESK (being the agrarians) does benefit a little from the “small-constituency” equation, but also won big in large northern constituency Oulu–43% to 16% for the second place Left Alliance–and benefitted disproportionately.

    It’s worth noting, too, that Finnish parties can run joint lists, and particularly in the smaller constituencies, they do form alliances. SDP actually lost a seat by allying with the Greens in the small Etela-Savo constituency (but, conversely, could have switched a seat from Center to Green or SDP by doing the same in North Karelia). In the same Etela-Savo constituency, KOK managed 2/6 seats with only 18% of the vote by allying with the Christian Democrats and True Finns.

    In short, the divergent ratio of the “big three” is more an issue of the Center and Coalition being relatively dominant in certain areas, and perhaps some bad SDP strategy.

    As for the government, KESK+KOK is only 101 seats, but SFP is pretty much always in government and would bring the total to 110. Alternatively, the current coalition could bring in the Greens.

  3. The Finnish Greens and Social Democrats spell it out for you:
    “The number of Members of Parliament elected from each electoral district is determined by the size of the population in the area.
    The differences in the size of the districts creates so-called hidden vote thresholds which vary considerably.
    The system particularly favours the Centre Party, which is strongest in rural areas.
    A possibility would be to allocate a certain number of Parliamentary seats to parties whose share of the national vote exceeds a certain threshold. In one possible application of this model, 189 seats would be allocated according to electoral districts as is the case now, and ten extra seats would be distributed among parties proportionally on the basis of the national result.”
    As in next-door Sweden.
    The Finns are so good at PR, having used it for so long. What’s holding them back from a national tier?

  4. It seems the Finns have found that even with an average dozen seats per district, you can still get a Tullimantta – – you don’t need to have very small PR districts (3-5), like Ireland.

    But then, even with a uniform number of seats for all districts (and PR), you can have perverse results: Malta in 1981 (all districts 5-seaters). But at least justice is seen to be done, even if it’s not actually done.

    And adding top-up seats can help, but then if qualifying for a top-up seat requires a party to win “one district-level seat” or “15% of the votes in one district”, then the same problem arises again, of unevenly distributed support.

    Does any list-PR system use uniform district magnitudes? I don’t know of any, probably because list-PR districts tend to be very big in magnitude (a dozen or more seats, on average), which means (a) you only need a bit over 53% to win a majority of seats (unlike the 60% needed to win 3 seats out of 4 in Ireland), and (b) more parties win seats, you are likely to get a coalition govt anyway than a two-party race. So making districts uniform in magnitude would gain almost nothing, and would stop you using provinces, counties, Cantons, etc as convenient non-gerrymanderable divisions.

    Now that Tasmania has reverted from 7-seater to 5-seater STV-PR districts (and with Western Australia and the Aust Capital Territory using a mixture of 5- and 7-seaters), I believe the largest uniform DM for STV is 6 seats for all districts, shared between the Northern Ireland Assembly and the Australian Senate. In other words, no assembly has uniform district magnitudes if it has more than 6 times as many districts as it has seats. (Correct me if I’m wrong!)

  5. Uniform magnitude in a list system: Chile since 1989. Of course, it is not really “PR” because the districts all have a magnitude of 2. That over-represents the second list (more so than the first), and obviously under-represents everyone else.

  6. Regarding the Malta case with STV and a vote-seat reversal: It was a “reversal” only because total votes are reported based on first preferences. As Alan noted in the discussion of Ireland, “STV is proportional to the final allocation of preferences, not first preferences.”

    Did Malta have a reversal based on final distribution of preferences? Almost certainly not, though I have never seen that detailed an analysis of the 1981 election.

  7. MSS: In 1981 the PN had 50.9% of the first preference vote, against 49.1% for Labour (only 29 first preference votes were cast for other candidates, also, transfers between the two main parties are rather uncommon in Malta). The election nevertheless returned 34 Labour and 31 PN members. There was a boundary adjustment prior to the 1981 election, and allegations of gerrymandering were made, but not really proven, I think. Anyway, after protests and parliamentary boycotts, an additional-seat-mechanism was introduced, which prevented a repeat in 1987, and prevented the opposite from happening in 1996.

    See for detailed results and other material (I still do not see how to insert hyperlinks when using this comment form, they do not seem to “take”, so apologies for that).

    Tom Round: For some reason the Basque Country uses three 25-seat (list PR) districts, corresponding to the provinces, despite Vizcaya having almost four times as many people as Alava. And at the Australian double dissolution election of 1987, the district magnitude for the Senate was 12 (excluding the territories). Nothing else comes to mind now.

    As you write, one advantage of list (or a Finnish-style) PR is that boundary adjustments are unnecessary, altering the seat distribution now and then is enough. Also, having an equal number of seats per district is not necessarily more likely to promote proportional overall results; too many other factors play a role. Of course, limited gerrymandering is nearly always possible, and intended or not, foreseeable or not, a chosen distribution of districts and seats is nearly always somewhat more to the advantage of certain parties rather than others, unless e.g. the district magnitudes are very large (and not malapportioned), or there exists an effective form of national compensation.

  8. Interesting case with the Basques. Forgot about 1987 double dissolution. Of course, NSW Legislative Council (State Upper House) has 21 seats each election by STV-PR statewide, but no districts. Ditto South Australia, 11 seats statewide every election.

  9. I retract my words about no List-PR systems having uniform D-Mag (assuming, as you say, that we count Chile as “PR”! But then, we Australians would say that the two Senators each for Canberra and the Northern Territory – elected at once by Territory-wide STV – are “PR”).

    I seem to recall some proposals for a “Triple-E” Canadian Senate having six per Province elected by List-PR.

  10. Suaprazzodi, a pedantic point: the Wikipedia page you linked to says 7-33 per constituency, not 6-33.

    In fact, to be ultra-pedantic, it should say “1-33” (although that would be misleading. Isn’t 1-33 the Spanish lower house’s range too?) since I see the Aland constituency elects only 1 single MP.

    Do the Alanders use first-past-the-post (like those Swiss cantons that have only 1 Nationalrat seat), or do they hold a runoff (like, IIRC, those French departments that have only 1 Senator)?

    What does list-PR usually “translate” to, when you reduce the D-Mag from average 13 seats to only 1?

    (I suppose if I were designing a list-PR system – despite being a known shill for STV, so I would have to recuse myself for bias – I would go for “approval voting within the list” for 2 or more seats, and “approval voting among all candidates” for a single seat [or a referendum]… assuming there were some constitutional clause forbidding my imaginary Rawlsian Solon from introducing preferential voting, ie Smith-Condorcet for majority decisions and STV for multi-seat allocations).

  11. List PR always reduces to FPTP if the magnitude =1, because all list allocation formulas give the first seat to the list with the most votes. If there is only one seat, the process ends there.

    So, Aland Islands, like the one-seat districts in Spain and Switzerland (and in all other list-PR systems of which I am aware) have their sole representative elected by plurality.

  12. IIRC, though, that in 1986, when the rest of France elected the National Assembly using List-PR, those French departments with only one Deputy (some of the outremers?) still used runoff. I think the same applies to elected councillors choosing French Senators.

    Likewise, most of the countries that use List-PR use runoff when they have to elect a President (Brazil, France, Timor, Austria, Weimar Germany [?]…). Very few still use plurality (Taiwan? South Korea?) and some that do, “filter” plurality through some mediating device that gives the candidate with most votes some kind of “absolute majority” at some stage (the US Electoral College, say, or the rule in Chile and Georgia – Carter’s Georgia, that is, not Stalin’s – that the legislature chooses among the top two if none has 50%).

    Okay, electing a President is not exactly the same as filling a legislative vacancy through by-election (the US doesn’t “appoint Electors” who then fill a Congressional vacancy), but usually a polity will tend to apply the same principle (or at least meta-principle) to all its elections.

    One could unify runoff and List-PR at the level of principle by saying that “If more than 50% of votes are cast for lists [including single-name lists] that win no seats, a runoff must be held”. Of course, if the first ballot allocates seats using PR, it would be almost inconceivable that more than 10-15% of votes could be wasted.

    PS: Yes, I know I should Google my facts, but I’m at work on my coffee break…

  13. PS: That should be an exclamation (not question) mark after “Weimar Germany”, and I know the USA and [Carter’s] Georgia don’t “use List-PR”.

Leave a Reply

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

You are commenting using your 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 )

Google+ photo

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

Connecting to %s