Constitutional Council

Just another WordPress site

Elections to Althingi

The Constitutional Council’s proposal for the election of members of Althingi includes a fairly radical, although not unexpected change, proposing a system of preferential voting.  The proposal is a little short on details, which will be provided in a legislation about parliamentary elections.  However, the proposal stipulates that 63 members will be elected from eight or fewer districts and that voters can either vote for a party list or individual candidates.  If the voters opt for voting for individual candidates they are allowed to distribute their votes across candidates of different parties and are free to vote candidate running in any district (if there is more than one district).   The electoral law can require up to 2/5 of the seats in parliament to be elected from district lists (rather than a national list).

It is a bit difficult to evaluate the proposal as it is rather vague.  I assume the intention is that the system is one of proportional representation but it is not explicitly mentioned.  Thus, the legislature may have considerable leeway in shaping the electoral law although the constitution fixes some of its parameters. Unfortunately, the Council no longer appears to include its rational for particular proposals (although they may be buried in meeting notes, etc. on the website).

There has been a push to adopt a form of preferential voting in Iceland for some years now so it is not surprising that it appears in the Council’s proposal.  The main reason preferential voting has been championed in Iceland is that the parties have frequently been seen to hold too tight a grip on its MPs.  The adoption of preferential voting is expected to loosen that grip and allow MPs greater freedom.  There are, at least, a couple of reasons to question this argument.  First, preferential voting essentially exists in Icelandic politics as most of the parties hold primaries most of the time.  Thus, the candidates don’t owe their position on the party list to the party but to the voters in the primaries.  Still, voting cohesion in parliament is very high.  It is not at all clear that using preferential voting in parliamentary elections will have any effect on the independence of MPs.  Second, in a previous post I argued that a high degree of voting cohesion was a function of the parliamentary system of government and that MPs are, more often than not, I suspect, motivated by the prospects of holding higher office.  That is, voting against the government is normally not good for ones ministerial career.  It is not an accident that the government MPs that we see vote against the government are precisely those that would never have become ministers – and they certainly won’t now.

So, perhaps introducing preferential voting is a harmless innovation – it only moves the preferential vote from the primaries into the general election and a potential benefit is that voters can now vote for candidates from different parties (which they can also do now by voting in multiple primaries).  But it is a missed opportunity.  Despite what Transparency International might lead one to believe, Icelandic politics are corrupt (the reason Iceland appears clean in TI’s ranking is that corruption in Iceland is not the type captured by TI’s methodology, e.g., bribes).  And there is a fair amount of research that suggests that preferential voting affects corruption and the importance of pork barrel politics.  Of course, if the electoral law reduces the number of districts to one it might seem that the incentive to engage in pork barrel politics might be removed but the absence of (small) districts doesn’t mean that politicians are unable to figure out carve out constituencies on their own.  At any rate, it seems extremely peculiar, given the political surroundings that led to the constitutional reform process, that there is no consideration (as far as I can tell) of how the electoral system may affect corruption in Icelandic politics.

Another part of the proposal requires changes to the electoral law to be approved by 2/3 of the members of Althingi.  While this doesn’t bother me, I do worry about whether this doesn’t invite a constitutional crisis as presumably it implies that the new electoral law will have to be adopted by 2/3 majority… and there is no guarantee that the political parties will find it easy to come to an agreement.  Unless of course the law is passed before the new constitution is ratified in anticipation of the new provisions in which case it could be adopted by a simple majority?

p.s.  David Farrell pointed out to me that the system is somewhat similar to system used to elect the Australian Senate.  However, there is no mention of STV in the proposal.  A member of the Constitutional Council provides some additional details about the ‘proposed’ (in quotes since it is not actually in the proposal) electoral systems: here (only in Icelandic).  In addition to panachage (allowing voting for candidates from different parties) it sounds like the system that the council has in mind also incorporates an element of cumulative voting as by voting for multiple candidates voters essentially split their votes (if you vote for two candidates, each gets half a vote – not clear whether you can decide to give two-thirds of a vote to one candidate and one-third to another).  The (semi-)cumulative vote element also suggest that the Council didn’t have STV in mind.


For references: See Required Readings

Some additional notes on related readings.  See, e.g., Chang and Golden (2007) for the relationship between electoral systems and corruption.  Cain, Ferejohn, and Fiorina (1987) on the personal vote and pork barrel politics.  Ames (1995) on pork barrel spending and how politicians carve subconstituencies in multimember districts.   Hallerberg and Marier (2004) on electoral systems and budget deficits.  The literature on the relationship between electoral systems and various political and policy outcomes is huge and some additional readings are listed in the required readings.



  1. Introduction.
    The main issue here appears to be electing the Icelandic parliament rather than executive positions. To help other readers understand, one can initially classify legislative/parliamentary systems into three categories:

    1. Single-Winner Districts. This uses inherently disproportional systems. Other problems include minority-majorities, gerrymandering, wasted votes, and noncompetitive races. On the plus side, constituents get a member to represent their individual district. Realistically, however, elected members will have a difficult time representing the people that voted against them.

    2. Multi-Winner Districts. This setup is quite proportional as a proportional representation (PR) system, and it has voters select candidates, not parties. Typically, the election threshold for these systems can be estimated using the Droop Formula: > (1/(Seats + 1)). These types of districts may use a preferential system such as Single Transferable Vote (STV), but it doesn’t have to. There are also proportional cardinal systems such as Reweighted Range Voting (RRV) and Proportional Approval Voting (PAP). An advantage of cardinal systems over ranked systems is that it’s easier for voters to rate or approve large sets of candidates than to rank them. Further, these two cardinal systems are monotonic, which STV is not.

    Asset voting is another option that is very easy for voters. This uses an innovative vote-for-one/most-votes-wins approach that has candidates negotiate their received votes. Asset Voting is a process similar to how Netherlands’ Parties negotiate to form lijstverbindings or kartels in order to have smaller parties reach the minimum party-list threshold. Technically, this could be used for Country-Wide Districts as well. Finally, there is Cumulative Voting—a bare bones PR option. As noted by the page’s author, it suffers from vote splitting as well as deficiencies dealing with surplus and deficit votes. I’ll not discuss disproportional systems such as Multi-Member Plurality because they contain countless debilitating flaws.

    3. Country-Wide District. These are your Party-List systems, which are highly proportional. The threshold for election can get very small and countries often set a minimum amount such as 4-5%. The most proportional of the common PR algorithms is Sainte-Lague, which uses sequential odd divisors to select the winners by party. Party-List systems can be open to allow voters to select candidates within the party they select. Typically, this actually nests Limited Voting (drawbacks similar to Cumulative Voting) for selecting candidates within the Party-List system. But this can just as easily nest another method such as a cardinal system or Asset Voting.

    There are some questions here. Iceland currently uses a unicameral 63-member parliament with a party-list system. This is great for proportionality purposes and is pretty good in its own right. Any kind of multi-winner district-based system will not be as proportional. One could create another parliamentary chamber for multi-winner district-based elections to keep a balance. Further, the chambers would actually be different–one with a more localized focus, another with a national representation focus. There are also options for mixed systems like Germany’s that use Party-List PR and single-member districts. There, the PR vote is really the main vote that matters. This mixed system addresses some of the main concerns with single-winner districts, but not all.

    Executive Positions.
    Finally, Iceland should consider how it elects its executive offices. Iceland does use a direct election to select its president, though it takes on mainly figure-head roles. Its prime minister takes more of the classic executive position. One consideration may be a direct election of this executive position by the people rather than the parliament. This would change the system from a parliamentary system to a legislative system.
    There is a very simple and fair system for electing executive offices. That’s Approval Voting. This system should be considered for local executive positions as well. A thorough article on Approval Voting can be found here:

    I hope this furthers the discussion.

  2. Einar Steingrímsson

    July 16, 2011 at 9:04 pm

    You say:

    “… preferential voting essentially exists in Icelandic politics as most of the parties hold primaries most of the time.”


    “And there is a fair amount of research that suggests that preferential voting affects corruption and the importance of pork barrel politics.”

    The latter one of these comments you make, as far as I can see, to point out problems with preferential voting. But, if we already have that in Iceland, as you claim in the first quote, that seems to have no relevance for the argument against instituting preferential voting in the constitution.

    • Indriði Indriðason

      July 16, 2011 at 10:59 pm

      Absolutely correct. What I am suggesting is that the primaries in Iceland help sustain corruption (it can’t explain it entirely since corruption in Icelandic politics precedes the adoption of the primaries). So while things may not get worse, I don’t see the point of taking bad aspect of politics in Iceland and making it a part of the constitution – if only for the reason that these effects may become clear one day and we may want to change the electoral system. If it is in the constitution it is going to be more difficult to change. I guess I would instead urge the constitutional council to think about what kind of (non-preferential) electoral systems would discourage the use of party primaries.

  3. Einar Steingrímsson

    July 17, 2011 at 11:58 pm

    I have to disagree with you that it would be preferable to avoid preferential voting altogether. It seems to me that this would strengthen the monopoly on power by the party elites in Iceland. This is one of the worst aspects of Icelandic politics, in particular because Icelandic political parties tend not to have a clear agenda, or any real debates leading to such a common agenda. The leadership (and even individual ministers) is therefore usually free to do whatever it pleases. A recent example is the Icelandic fisheries system. Both government parties claimed to want to change it in favor of one where the fishing quotas would be taken out of the hands of the current private “owners”. Instead, it now seems clear that the government will not do this, caving in to pressure from the powerful lobby of the fisheries owners, even though a vast and stable majority of the electorate has been for a drastic change for a long time.

    In short, the corruption in Icelandic politics has been created by a joint effort of the four major parties over a long time. This corruption will never be quashed while these parties dominate power in Iceland. To reduce, or abolish, preferential voting will only play into the hands of these parties.

    • Indriði Indriðason

      July 18, 2011 at 1:09 am

      I guess my response will have to cite your previous comment – preferential voting in the primaries has not prevented the parties to behave in this way so why would adopting preferential voting for legislative elections make a difference? So, yes, it is possible that MPs would be even more tied to party elites without any form of preferential voting but, then, voting in parliament indicates that voting cohesion is almost perfectly along party lines, i.e., things can’t really get much `worse’. Voting cohesion in Finland, which uses a form of preferential voting, is, e.g., still high: see, e.g., Pajala (2007). Depauw and Martin (2008) provide comparative data that highlight that this is true about parliamentary systems in general – Finland does a little bit ‘better’ than the other countries they look at – but almost all MPs vote with their party almost all of the type.

      As I hinted at in an earlier post, I think the key to weaken the parties requires more radical solutions, e.g., to abandon the parliamentary form of government. At the end of the day, MPs care much more about their party being in government and the possibility of becoming a ministers (or an ambassador later on), which provides strong incentives to vote with the government. Martin (2011), e.g., argues that electoral systems don’t do a terribly good job of explaining legislators’ behavior while the allocation of important positions in government and parliament does matter.

      While I am not going to defend the Icelandic parties (ok, just a little bit, I do think that whether an MP belongs to the Left-Greens or the Independence party actually tells me something – I may not like their policies but I have some idea what they stand for and what they might do in office) or particular policies, it is not clear to me at all why parliament with weak parties would be a good thing. I think parties have an important function as I argued here: Many other countries have strong parties that seem to do a decent job – why can’t we? I just think that we might want to think a little bit harder about, e.g., what institutions might make the parties perform their functions better instead of taking the risk of throwing the baby out with the bath water.

      One final point. If you see the problem being the collusion of the four parties, I think there are is a simple way to reduce their influence – make the country a single constituency and/or increase the number of MPs. That would reduce the electoral threshold, making smaller parties more viable. Thus, more parties would win seats in parliament, reducing the influence of the major parties.

      Thanks for the comments.

      Pajala (2007):
      Depauw and Martin (2008):
      Martin (2011):

  4. Einar Steingrímsson

    July 18, 2011 at 1:57 am

    Thanks for your replies!

Leave a Reply

Your email address will not be published.