i

In an article on visir.is, Þorvaldur Gylfason, a member of the constitutional council, argues that the size of parliament should be reduce from 63 to 53 seats.  Þorvaldur, a former colleague at the University of Iceland, had actually asked me to put together a short summary about research on the size of legislature before the Supreme Court invalidated the outcome of the election to the Constitutional Assembly – at which point I put the matter aside and then, apparently, forgot about it.  Þorvaldur’s article was a timely reminder so here are my two cents.

In short, not a whole has been written about the size of legislatures and much of it focuses on the effects legislature size on government spending.  Most research finds that larger legislatures spend more.  The argument is that individual legislatures are concerned with winning election and in order to win election are more likely to propose pork barrel projects to build a personal vote base.  In addition, larger legislative size makes pork barrel projects cheaper as the deliver target benefits while the costs are distributed more widely (greater number of districts).  A lot of this literature is focused on the U.S. where the conjecture appears to be supported by the empirical evidence (in unicameral legislatures). Mukherjee (2003) does find the same effect in a cross-national study.  As pork barrel politics appear to have been fairly important in Iceland historically and the preferential votes system is not likely to reduce those incentives, reducing the size of Althingi can be expected to reduce government spending (which may or may not be a good thing depending on one’s political leanings).  On the other hand, Primo and Snyder (2005) work suggests that the type of spending that is affected are, as the argument above suggests, inefficient pork barrel projects, which are less controversial.  However, that depends a lot on what the electoral system will look like – pork barrel projects may very well benefit the rural areas, which traditionally have been overrepresented in Althingi.  That said, even under a  preferential voting, it seems quite plausible that legislators/candidates will seek to carve out ‘strongholds’ even if the country is a one electoral district.

There is no question that Iceland’s legislature is large given the country’s population – there are roughly 5060 citizens for each legislator.  Only Tuvalu, Andorra, and Antigua and Barbuda have fewer citizen per legislator.   One implication of this is that under the preferential vote system shaking hands and kissing babies (hopefully they will stop short of kissing chickens as Irish politicians have been known to do) really looks like a feasible strategy.  Randomly looking up MPs, I found, e.g., that Helgi Hjorvar has 2407 friends on facebook – surely that is enough to get elected under the preferential system.  Reducing the number of MPs from 63 to 53 is, however, unlikely to make much of difference – it increases the number of citizens per legislator by slightly less than thousand.  Thus, under the preferential system, shaking hands, doing favors and providing pork would remain a feasible strategy unless a much larger reduction in the number of legislators was undertaken.

However, at 63 legislators, Althingi is already a small legislature.  A lot of the discourse about the reform has focused on strengthening Althingi – although I think that discourse is in large part based on rather strange ideas about parliamentary regimes.  Leaving my skepticism aside, the literature on legislature frequently cites strong committee systems as an important, if not the primary, source of legislatures’ power vis-a-vis the executive.  One strand of those arguments focuses on institutional powers (ability to amend bills or kill bills in committee – which often are absent or dominated by party discipline in parliaments).  Such institutional powers are not likely to be affected by the number of legislators.  Another strand of arguments about the importance of legislative committees focuses on the opportunity that permanent legislative committees offer MPs to specialize.  As it stands there are 12 permanent committees composed of nine MPs each (11 for the the budget committee).  Thus, on average, each MP serves on 1.74 committees.  That seems quite a lot to me and I am skeptical that this allows MPs to become specialists in any policy area.  Reducing the number of MPs by 10 would increase that to 2.07 committees per MP, which certainly won’t help matter.  All in all, if one considers the committees the way to increase parliament’s influence, I think there is a strong case for increasing the number of MPs.

Parliament’s role is not just to legislate – it is supposed to provide oversight.  One might argue that Althingi failed spectacularly in its oversight role in the years leading the the economic crash in 2008, which in no small part pushed the constitutional reform forward.  Despite being charged with protecting the interests of its citizens, not a single MP (save one minister) thought it necessary to resign.  The question is why Althingi didn’t take its oversight role seriously.  While I don’t know the answer to that question, a part of the problem has to do with Althingi’s resources – including the number of MPs for the reason cited above.  If Althingi is to have any real influence on policy or be able to exercise oversight it needs to have the resources to do so – meaning that either it needs to provide the right conditions for MPs to specialize in policy or vastly expand (establish?) Althingi’s research capabilities.  The problem with Althingi is not that it is too expensive but that it is too cheap.

There are also some other issues that need to be considered.  Ministers have, nearly without exception, been drawn from the legislature.  That is, the legislature is effectively the talent pool from which ministers are drawn.  What are the effects of shrinking the talent pool?  Supposing there are 10 ministers (as proposed by the constitutional council) then nearly two-fifths of the MPs of the governing parties (assuming a narrow majority) are become ministers.  Frankly, I find that a little worrying.  I’ll quote (a great quote) from Dewan and Myatt (2008) about ministerial selection in the UK:

“Paxman (2003, p. 209) recounted that Tristan Garel-Jones, a whip in a United Kingdom government of the 1980s and a close confidante of Prime Minister John Major, recalled scanning a list of fifteen candidates for a junior ministerial post and thinking

“I wouldn’t employ a single one of them. The problem was that, if you include all the various ranks of ministers, you have to find maybe ninety people to form a government. You have perhaps 350 or so people to choose from. Once you’ve eliminated the bad, mad, drunk and over-the hill, you’ve got rid of a hundred. You then have to pick ninety people out of a pool of 250. Is it any wonder that the calibre is so low?” “

Changing the number of MPs is also likely to affect the number of parties (thought how that pans out depends on the electoral system) – most likely reduce the number of parties and raising the barrier of entry for new parties.  That brings us back to the size of government.  Mukherjee (2003) finds that the number of parties in the legislature has a positive effective on government spending and others have shown that the number of parties in government also increases government spending (Bawn and Rosenbluth) but greater legislative fragmentation should lead to larger coalition governments (i.e., more parties).

In sum, I can see arguments for and against reducing the size of Althingi – in the end I think it depends on the role one wants Althingi to play.  However, I think a lot more research is required to offer a conclusion answer.  As it stands, we actually know fairly little about what members of Althingi do.  For example, how important are the legislative committees in the current system?  How much does Althingi amend legislative bills?  Are these amendments substantive or are they primarily cosmetic?  How important is pork barrel in Icelandic politics?  Making an informed decision actually requires answering these questions – and ideally not just rely on our hunches about what the answers might be.

For references: See Required Readings

Some additional notes on related readings.  Stigler (1976) is one of the early studies of legislature size and its consequences.  Lowenberg (2007) discusses how different conceptions of representations have implication for the choice of the size of the legislature.  Gilligan and Matsusaka (2007), Primo and Snyder (2005) and Chen and Malhotra (2007) provide evidence for the positive relationship between legislature size and government spending in the U.S.  Mukherjee (2003) shows that legislature size and the number of parties increase government spending.  Ricciuti (2004) also examines government spending cross-nationally and fails to find evidence for a relationship. Nadia Fiorino and Roberto Ricciuti (2007) find a large effect of legislature size on per capita regional government spending in the Italian regions. Rasmusen and Ramseyer (1992) examine the relationship between size of legislature and bribes.

The ability of the Icelandic president to refer legislation to a referendum has been source of controversy every since the president used his veto for the first time in 2004 to veto a bill on the ownership of the media.  The episode lead to much debate that was primarily focused on whether the president’s actions were actually constitutional – a different clause in the constitution states that the president’s powers are exercised by the government’s ministers.  The debate was legalistic and basically focused on competing interpretations of the constitutions.  Regardless of what interpretation is ‘correct’, the episode highlights the limits of the legalistic discourse that has surrounded the Icelandic constitution – and still appears to surround the making of the new constitution.

The fact remains that the president did use his veto.  He was able to do so because the constitution was sufficiently ambiguous and the political context was right – the government forced a bill, seen as targeting particular individuals, through parliament without much consultation and it appeared unpopular with the public.  It can be argued that the president’s veto fundamentally changed the political system – previously it had been semi-presidential in name only.  Since then the president has used the veto twice.  In short, all the legal arguments about the constitutionality of the presidential veto didn’t matter at all.  The veto came to be because it served the political ends of the president (whether those were opposition to the legislation or a desire to strengthen the presidency).  On one hand, this suggest that the law is important, i.e., if we want the constitution to be effect it needs to be clear and unambiguous.   On the other hand, it suggests the limitation of legalistic approaches as politicians (and people in general) react strategically to the institutions that constrain their actions.  In other words, politicians generally don’t get involved in politics out of desire to follow the law but rather to achieve a variety of other goals, e.g., changes in policy or a desire to hold a public office.  The constitutions sets the rules of the game but politicians will, of course, try to use those institutions to their advantage – much like football players occasionally take a dive or pull a shirt.  The legal approach to constitution making largely ignores how political institutions create strategic incentives and, therefore, remains silent on what the unintended consequences of those institutions might be.  As I have mentioned before, the fact that most of the writings that the Constitutional Council appears to be consulting (judging by its website) is written by lawyers (not to mention primarily being case studies focused on Iceland).  I think Posner once wrote that there was few things as dangerous as a lawyer that didn’t know economics.  These days I worry that there are few things as dangerous in Iceland as lawyers (or constitutional writers) that don’t know political science.

In some ways the controversy surrounding the use of the veto in 2004 can be seen as having sparked the constitutional reform debate so it is not surprising that the Constitutional Council has offered a proposal.  The Council actually offers two options.  The first one is to retain the current veto + referendum combination while an alternative proposal adopts a more traditional form of the veto where a veto sends the bill back to the parliament which can override the veto by simply passing the legislation again with a majority (i.e., no supermajority is required).

The first option, retaining the veto in its current form, might sound like sticking with the status quo but that is not entirely clear.  The veto has been used by a single president that was elected as a figurehead rather than a political actor with the ability to shape policy outcomes.  If the president retains his veto, chances are that the next presidential election in which there is no incumbent will be quite different – the candidates are likely to distinguish themselves more on policy grounds than simply, for the lack of better words, their ‘presidential-ness’.  In other words, the actions of the current president are not necessarily a good indicator of how future presidents will behave.  I would not be surprised that the use of the veto would increase, especially under co-habitation, as the role of the president becomes more political.  Alternatively, the government will increasingly consult and bargain with the president over major legislation in order to avoid the exercise of the veto.  Whether that is a good thing or not, I don’t know.  It doesn’t strike me as a terrible idea, but it has the potential to change Icelandic politics quite radically.

I am less enamored with the second option, combining a veto with a simple majority override provision.  Essentially it makes for a very weak veto power – parliament can always override the president’s veto.  In essence, the power of the veto only amounts to being able to delay legislation for a little while.  The proposal does require the president to explain why he vetoes a bill, which some might argue does allow the president to shape the agenda a bit but it is not clear to me that the president doesn’t have that same ability to persuade without the veto.  The proposal also has a built in incentive for parliament to ignore the president’s suggestions as any amendments to the legislation would offer the president another opportunity to veto the bill.  Thus, in circumstances in which parliament is uncertain about what compromises are required to obtain the president’s signature, it may consider itself better of by making potential improvements to the legislation in order to avoid further delay.

Somewhat strangely, then, the more substantial proposal for changing the constitution is actual the less radical one – the relationship between the president and the government will remain similar to what it has been for the past 65 years.  The seemingly less radical proposal of retaining the current veto powers, on the other hand, actually has a far greater potential for changing politics in Iceland because the use of those powers has been reinterpreted in the past few years to provide for an active political role for the president.  Thus, the proposal retaining the status quo relationship between government and president is the more radical one as it seeks to codify the president’s powers.

The constitutional council has begun publishing its proposals for sections of a new constitutions. So far these sections include human rights, the work of parliament, and the judiciary. With this post, I start reviewing some of the proposals concerning the organization and work of parliament. I will focus on the proposals that may be considered consequential or controversial. In translating the proposals I focus on what are the key issues rather than provide a full translation.

2. Greater role of the speaker of parliament. Alþingi elects the Speaker with 2/3 of the vote at the beginning of each electoral term. The Speaker organizes the work of parliament. The Speaker vacates his seat in Alþingi (replaced by another member from his/her party).

The goal is to increase trust in parliament by requiring broad support for the Speaker in parliament and, as a consequence, to strengthen Alþingi as an institution. This strikes me as an odd proposal for several reasons. First, I’m not aware of the Speaker having caused much controversy so I’m not sure this will help much even if it works out the way that the Council thinks it will.

Second, the electoral rule is problematic because it is not resolute – meaning that it doesn’t always produce an outcome. What happens if no candidate wins two-thirds of the vote? Try again? From the Council’s explanations that accompany the proposals, I gather that these are not final proposals as the do address this possibility – the longest serving MP serves as Speaker during the election of the Speaker and the council suggest that s/he becomes the Speaker in the event Alþingi fails to elect a Speaker by two-thirds majority but what constitutes ‘failure’ here is not defined. Is it failing to elect Speaker on a single vote? Or just when everyone has had enough? The Speaker of the U.S. House used to be elected by majority rule (also not resolute if there are more than two candidates). In 1849 and 1855 it took, respectively, 62 and 132 rounds of voting to elect the Speaker – and only after adopting plurality rule (Stewart 1999). Perhaps the Council should read up on papal elections, which also use two-thirds majority. Between 1878 and 1978, from three to fourteen rounds were needed to elect the pope (Toman 2004). That doesn’t seem to excessive but it is probably aided by the fact that the cardinals are essentially placed under house arrest for the duration of the election. Before the ‘lock up’ procedure was instituted, the election of Pope Gregory X took over two years and nine months. John Paul II changed the rules for papal elections so that if a certain number of rounds of voting is reached, the only plurality instead of two-thirds majority is required. Perhaps the Council should consider adopting a similar rule? Or, at the very least, burn the ballots after each vote to make things a little more exciting. The fact that there is a default Speaker in the event of a failure of electing a Speaker by two-thirds majority also affects the government’s (and opposition’s) incentives. If the longest sitting member of parliament is a member of a government party, the government has no incentive to find a ‘unifying’ candidate that a supermajority can agree on. The same argument holds with respect to the opposition. Thus, the ‘default’ option always favors one side or the other, it appears likely that Alþingi will rarely elect a Speaker by two-thirds majority, it will be a source of conflict after each election, and will, therefore, be counterproductive in terms of increasing trust in parliament – which brings me to my third point.

Third, the proposal fails to recognize that its goals are, sort of, conflicting. The idea is to make the Speaker more powerful (although it is not clear what about the proposal achieves that) and for the Speaker to have broader support within Alþingi. Well, the more powers the Speaker has, the more sought after his position will be by both the parties and the individual candidates. Thus, the more powerful the Speaker, the less likely it will be that government and opposition parties will be able, or willing, to support the same candidate. Especially when at least one of these actors doesn’t have to (see above). The opposite is true – if the position of the Speaker is utterly irrelevant, then government and opposition parties will find it relatively easy to settle on a candidate and the potential for conflict at the beginning of each term is removed.

Fourth, if the Speaker holds important agenda setting powers it may have more wide ranging effects than just surrounding his/her election. The types of coalitions that form may be affected. There may, e.g., be an impetus to form surplus coalitions that don’t need to rely on the opposition to elect a Speaker. Is that a bad thing? It depends on what you want. The evidence suggest, e.g., that larger governments are more likely to spend more (Bawn and Rosenbluth 2006). The Speakership election also provides the opposition with an additional bargaining chip – it may only agree to vote for the government’s candidate (or to elect a Speaker at all) if the government agrees to some policy concessions. It is possible to think of these outcomes as signs of more consensual politics but it is equally easy to see them as providing a minority with a potentially important veto power.

Fifth, the introduction of a seniority rule (assuming that the parties will generally ‘fail’ to elect a Speaker) could even affect the parties’ decisions about how to set up their party lists for elections. If the role of the Speaker is an important one, the parties have an incentive to give senior members a secure spot on the party list – although I suspect other considerations are likely to trump these. Thus, it may serve to increase the incumbency advantage and reduce renewal in the legislature.

For references: See Required Readings

Some additional notes on related readings.  Caplin and Nalebuff (1994) demonstrate that a supermajority requirement of 64% is sufficient to guarantee that the  preferred alternative of the mean voter can not be beaten.  Indridason (2008) demonstrates that in the context of run-off elections with no threshold or participation requirements (similar to what is being proposed here – I think), the outcome is close to the median voter when only a majority is required.  The presence of ‘default’ Speaker selected on the basis of seniority would alter these results.

The power of political parties has become, perhaps increasingly, a common complaint of observers and participants in Icelandic politics and many of the proposed reforms of the political system have focused on reducing the power of political parties.  For example, by introducing an element of a `personal vote’ into the electoral system (usually by adopting an open list proportional representation system) and measures to strengthen parliament vis-à-vis the executive.

The election of the constitutional assembly is a useful reminder that parties do serve an important function in politics.  The constitutional assembly election saw over 500 candidates running for 25 seats – the great majority of which was unknown to most voters.  As I have discussed before, learning about what these candidates stood for was an enormous and an overwhelming task for voters.  Contrast that with a regular parliamentary election in Iceland where the number of candidates is probably of a similar magnitude.  Yet voters don’t appear to have much difficult (most of the time) figuring out how to cast their votes in parliamentary elections.  The reason is that the candidates stand for parties and the voters cast their votes for parties rather than candidates.  And importantly, the parties usually have reasonably clearly formulated party platforms – choosing among 4-6 party platforms is a far easier task than choosing among 500 candidates.  This is one of the primary roles of parties, i.e., to aggregate policy preferences and to formulate policy alternatives so that on election day voters simply face a choice among a handful of party platforms.

The downside, of course, is if the parties do a poor job of presenting voters with platforms that they would like to see.  Overall, I think Icelandic political parties do a decent job of that – at the very least, most voters see the parties as representing different interests.  Of course, it is possible to argue that while offering different policy alternatives, the political parties still fail to represent the wishes of a majority of voters – and opinions are likely to differ on that point.  Regardless of what position one takes on that issue, it seems sensible to ask what the alternative might look like if one adopts institutional reforms aimed dismantling the party system in its current form.  While  extending voters’ ability to cast a personal vote arguably increases voters’ choice, those choices may ultimately be less meaningful.  An open-list proportional representation system would increase the independence of MPs (and candidates) from their parties, allowing them to carve out their own policy platforms.  The question then is, if I vote candidate X who runs for party Y but that candidate is not elected I may end up having supported a party whose MPs support policies that I don’t like that much.  In the worst case scenario, party labels would a worthless signal about candidates policy preferences.  A more likely scenario is that the party label will still provide a rough guide to its candidates’ ideological orientation.  But it is not clear to me that it is desirable to introduce greater uncertainty into the mix.  As it stands, voters have a difficulty predicting how their votes will influence policy – I may like a certain party but it may not always be in my best interest to vote for that party because I’m unsure which coalition will form after the election.  Increasing the scope for casting personal vote would likely increase that uncertainty – now I would both be uncertain about which coalitions may form and also about who the people at the bargaining table will be.

Of course, we cannot overlook the fact that, essentially, we already have a system based on a personal vote.  All the major parties regularly hold party primaries where voters, usually, determine the order of the candidates on the party lists.  This fact points to an interesting question.  First, if an electoral system based on a personal vote is seen as conducive to producing more independent MPs and a stronger parliament, why don’t the party primaries achieve this?  Or turning things around, what is it about making the personal vote important in parliamentary elections that makes MPs more independent?  My intuition is that MPs’ independence has less to do with the electoral system than the nature of parliamentary systems – MPs are forced to support the government lest they want to risk becoming a part of the opposition, which plays to the advantage of the cabinet.  Moreover, most MPs dream of becoming ministers one day and defying the party leaders is rarely conducive to realizing that dream.

At any rate, it is not clear to me that the case for weak political parties has been made – it is easy to point at the Icelandic parties and argue that their strength limits the independence of MPs and the power of Althingi.  The solution to those problems is usually treated as obvious – make the fortunes of MPs more independent of their parties – but I think it is also important to think more carefully about what the alternative might look like, i.e., what would politics with weak parties look like?  I have already suggested that it may make voters’ choice more difficult but one might also wonder about how it would affect legislative politics.  What happens when we don’t have strong parties to usher legislation through parliament?  If I had to guess I would say it would be less efficient and more time consuming without necessarily making for better public policy – in order to cut through legislative stale mate, policy makers will be more likely to cut through legislative deadlock by catering to special interests.  But that is just a guess – and there is no need to rely on guesses, there is some evidence, and possibly a lot of data, that could be used to answer questions such as these.