Saturday, September 6, 2014

New Zealand elections-towards a two-party system? (Updated 22/09)

New Zealand will be holding a parliamentary election on the 20th of September. National Party leader John Key is the incumbent prime minister, and has been leading in the polls with around 45-50% of the vote since the last election. Opposition Leader David Cunliffe, who is leader of the left-wing Labour Party, has been trailing, with his party only having about 25-30% of the vote.
Under New Zealand's previous single-member plurality system, which was used in all elections up to 1996, Key would win a landslide victory, with about 100 seats in the 120 House of Representatives. However, a series of referendums during the 1990s, the new mixed-member proportional system was introduced. This system made New Zealand elections much more proportional.

The Mixed-Member Proportional system

The Mixed-Member Proportional voting system originated in Denmark, where it was used for some elections between 1915 and 1920. However, it was introduced as a system for West Germany after the Second World War.
Under the system, New Zealand is divided into 70 single-member electorates. Voters have two votes: one vote for a candidate in their electorate, and one party, or list, vote. In the single-member electorates, the candidate with the most votes is elected as the electorate member. But after these members are elected, all the party votes are added up. All parties that have either won 5% of the party vote or one single-member electorate have their votes put into a Sainte-Lague proportional representation allocation (an explanation here). The electorate seats that a party has won are subtracted from the total allocation of seats, and the remainders are distributed to party lists, submitted by the party before the election. If a party wins more electorate seats than its proportional allocation, it keeps these seats, and parliament increases in size accordingly.
This system aims to produce an overall allocation of seats that is proportional to votes, and it has meant that there have been no single-party government, and a significant number of small parties have won seats that they would not have won under single-member plurality.
However, recent political developments have reduced the number of political parties in the New Zealand political system, to the point where New Zealand may become a two or three party system in the near future.

Increased Polarisation

The chart below shows how the New Zealand Parliament stood after the 2011 election.
There have been some changes since this election. ACT's single member of parliament, John Banks, resigned over corruption allegations. A New Zealand First MP, Brendan Horan, resigned from his party to become an independent.
While there appear to be many parties, most of them align behind either Labour or National, and the only 'independent' party appears to be NZ First.
Another notable feature of this is that there are not many medium-sized parties. The two largest parties have 77% of the seats, and adding the Greens to this brings it up to about 93%. Why has this happened? Well, it's all part of the movement towards a two-party system.

"Electorate Seat" Parties

About half of the parties in the New Zealand parliament are dependent on electorate seats to win any seats at all. The electorate seat threshold means that winning an electorate seat can get list members into parliament, a useful mechanism for small parties. The Mana Party was the only party to have serious competition from a party in the same ideological column for its electorate seat. No National candidates stood in any of the seats won by the Maori Party, and while National candidates ran in Ohairu-Belmont (United Future) and Epsom (ACT), they mostly ran to get electorate funding, and their opponents received implicit support from the National party leadership.

Why would National sacrifice these seats, some of which were winnable for the party, to get candidates from minor parties elected? The answer is simple. National needs extra list seats to get a majority, and therefore needs ACT and United Future to pass the threshold, so they can get these seats. If ACT and United Future had won no seats, National would need to get very close to a majority to win government.

The Decline of Medium-Sized Parties

Recently, New Zealand has reduced its number of parties significantly. A good measure of this is the 'effective number of political parties'.  In order to calculate this, all of the percentages of the party votes (or seats) are squared, and added together. The number 1 is than divided by this figure.


As we can see, the number peaked in 2002, and has declined since then. It reached a bottom in 2008, and hasn't moved much since then. Why has this happened? Well, one issue is the issue of coalitions. Small parties in coalitions tend to lose votes after being parts of coalitions. After a coalition with National, NZ First fell below the 5% threshold, and had to rely on leader Winston Peters' electorate seat to stay in parliament. The left-wing Alliance splintered after a coalition with Labour, United Future went from 8 to 3 to 1 after Labour government (2001 to 2005 to 2008), and NZ First was forced out of parliament after a Labour coalition.

If Key falls short of a majority, this record may mean that the Greens and Mana Party (in alliance with Kim Dotcom's Internet Party)  pause for thought before jumping to coalition with Labour. In fact, it may be the ability of the Green Party to win votes that prevents New Zealand from becoming a pure two-party system.

Post-election update-22/09/14

The election is over. The National Party have been re-elected, with an overall majority of 61 members in a 121 member Parliament (Peter Dunne is the overhang). The main surprise of the night, however, was the strong performance by New Zealand First, and the poor performance by Labour. NZ First have had the benefit of eight years out of government (the longest stretch since MMP was introduced). The result is their third-best since MMP was introduced.

The effective number of parties in terms of votes has increased (from 3.15 to 3.21)  There are no new parties in Parliament (and the Internet Mana alliance did not enter Parliament, as Hone Harawira lost his electorate race).The Conservatives (a new right-wing party founded in 2011 by Colin Craig) increased their vote to about 4%, but fell short of the 5% threshold. 

The lack of any significant change in terms of Parliament, other than the slight drop for Labour and the increase for NZ First, means that the effective number of parties in terms of seats has fallen from 2.98 to 2.9, which is not a significant decrease, but is part of a trend.

The election may provoke debate over the level of the 5% threshold. A committee set up to review MMP has recommended that the threshold be lowered to 4%.  My table below shows how the results would have differed with a 4% threshold.

With a 4% threshold, the centre-right's total would increase by 3. National would lose two, and thus not have an overall majority, but the Conservative Party's five would cancel that out. Even if Key wanted to exclude the Conservatives from his government, the current combination of National-Maori-United Future-ACT would still have a majority, with 63 seats. Key could probably count on the Conservatives to support him in confidence votes, so this would really not matter.

The change would have a negative impact on the centre-left. Internet Mana only won 1.26% of the party vote, and thus would not be saved by the change. The Greens and Labour would lose two and one seats respectively, weakening any basis for a centre-left government.

If the National party's support drops significantly during the next 3 years, and the Conservatives stay above 4%, there may be some temptation for Key to accept that recommendation and reduce the threshold to 4%, especially if Peter Dunne resigns or loses his seat. Unless a left-wing party appears on the horizon with support above 4% (unlikely), this decision would make tactical sense for the Prime Minister, and would strengthen National's political position in the near future.

Tuesday, September 2, 2014

The new Italian electoral system: a real solution to instability?

Italian Prime Minister Matteo Renzi has recently introduced a proposal to change Italy's electoral system after the Italian Constitutional Court declared that the current system was illegal. The system is important, as Italy's future political and economic stability may depend on whether this system can produce stable government, or whether, like so many other grand plans for Italy's elections, it falls.
But before we look at this system, it is important to look at how Italy's previous systems have worked, and why they failed to produce the holy grail of Italian elections: strong, stable majority government.

Proportional Representation (1946-1992)

The system used for the Constituent Assembly, and for all Lower House elections until 1992, was a system of two-tier proportional representation. Italy was divided into multi-member constituencies, electing an average of 20 MPs each. Lists were fully open, and voters had three or four preference votes for candidates within party lists.  In each constituency, the Imperiali quota was used (votes/seats+2), and seats were awarded to parties for each full quota. All remaining seats and votes were transferred to an upper level, where seats were awarded using the Hare quota and the largest remainder method. Until 1953, these seats were allocated to specific closed party lists at the upper level; after this date, however, seats were allocated to the best losers at the constituency level.

The system used for the Senate was, at least on paper, very different. The country was divided into 237 single-seat constituencies, but the Senate as a whole elected 315 members. To win in a single seat constituency, a candidate needed 65% of the votes. Only a few senators (members of the South Tyrol People's Party) won in this method. All votes that did not elect a senator in the single-seat districts were transferred to the upper level regional district, where seats were distributed using the D'Hondt method.

The main drawback was the high degree of proportionality. No majority government was elected from 1948 onwards, and as a result Italy was mostly governed by a loose alliance called the Pentapartio or 'five parties'. These parties were the Christian Democrats (centrist), the Socialist Party (centre to centre-left), the Republican Party (centre to centre-right), the Liberal Party (centre-right), and the Social Democratic Party (centre to centre-left). They were mostly opposed by the Communist Party and the neo-facist Italian Social Movement, but there were a few minor opposition parties such as the Radical Party (socially liberal) and the Proletarian Democracy (far-left communist). The low threshold for entering the parliament (one regional seat and 300,000 votes nationwide for the lower house) meant that a new party could easily get started up.
The system meant that Italy had a long string of unstable governments mostly led by the Christian Democrats for the entire period of this system. 

Another problem concerned the system of preference votes. While it seemed like a democratic system on paper, the fact is that it was used, especially in the South, as a means of multiplying and trading blocks of loyal followers in order to get certain people into parliament.

During the late 1980's, Italy's political system experienced a number of significant changes. The Communists dissolved themselves, and formed the new Democratic Party of the Left, a politically traumatic event that split the left vote between the new party and the Communist Refoundation Party (a group made up of radical Communist hardliners opposed to the name change). New political forces sprung up, such as the Northern League, a populist right-wing party, and the Green Lists, a green political party opposed to nuclear power.

A number of corruption scandals that involved senior politicians in the Pentapartio parties led to a poor result for the major parties at the 1992 election, and an increase in support for the Northern League and other populist, anti-corruption parties. There was a significant change in the electoral system at this election; voters had only one preference vote, no matter how many seats the constituency elected. 

The parliament that resulted was a very split one (results here-much more detailed than Wikipedia, but in Italian, so Google Translate may be useful). No one party, or group of two parties, had a majority; neither did the Pentapartio, a situation which paralysed government.

Following the election, a referendum was called to change the system of electing the Senate. As Italian referendums can only repeal laws, the referendum was on the topic of removing the 65% threshold. This would create a more majoritarian system, as all single-member districts would elect a senator based on plurality, but there would still be a proportional tier that would compensate for the disproportionality.

This spurred the government on, and during late 1993, the government introduced a bill for a new system, which I describe below.

Scorporo (1994-2001)

The government, as one of its last actions in office, introduced the scorporo (Italian for subtraction) system. I have already explained the system for the Senate, but the scorporo applied only the house.

The system worked like this: Italy was divided into 475 single-member constituencies, but there were 630 seats (the other 155 were elected by proportional representation). Voters had two votes: one was for a candidate in their single-member constituencies, and one was a vote for a party in the proportional seats. In the single-member constituencies, the candidate with the most votes won. There was a more complicated procedure for the proportional tier, however. For a start, all the ballots were counted up, and any parties that did not win 4% were removed from the count. Following this, all votes for runner-up candidates in a constituency that a party had won were removed from that party's list total. Seats were then allocated to parties using the Droop quota (votes/seats+1) and largest remainders, and this seat allocation was then allocated back to regions, also using the Droop quota. (Here's a more detailed explanation)

This system did indeed provide relatively stabler government than the previous system. It also reduced Italy's political system to more of a two-party system. However, the system had some serious problems.

The main problem was the blackmail power that the smaller parties received. A candidate from a small left-wing party in a marginal constituency could fatally split the vote, and elect a right-wing candidate or vice versa. Small parties demanded safe constituencies from the larger parties, which led to parties that could not elect members through proportional representation electing members in the constituencies, a process known as the 'proportionalisation' of the single-member constituencies.

It also did not mean complete stability. Silvio Berlusconi's first government ended abruptly after the Northern League left over corruption, and the centre-left government of Romano Prodi fell after communist senators voted against the government on certain issues of foreign policy. 

Another more technical problem was the possibility of 'decoy lists'. During the 2001 election, both parties registered some of their candidates to decoy lists. These lists did not campaign, received tiny amounts of votes, but had all the votes for the large parties subtracted from their total. This removed the connection between the constituencies and the party list tiers, and created a system of 'parallel voting".

Because of these problems, and perhaps for partisan advantage, the Berlusconi government replaced the scorporo with a system that is described below.

Fixed Ratio-Proportional Representation (2005-2013) 

The Berlusconi government introduced a system of proportional representation, but with a very large bonus for the largest group. For the lower house, the party or coalition that won the most votes nationwide automatically won 55% of the seats, unless they won 55% of the seats without the bonus, in which case the system became pure proportional representation. There was a threshold of 10% for coalitions, 2% for parties in coalitions with 10%, and 4% for parties outside coalitions. If a party won over 4%, but was in a coalition that did not win 10%, the party could still win seats.

For the Senate, a similar system applied, except that it was applied on a regional basis, and the thresholds were different. A party or coalition winning the most votes in the region would win 55% of seats, and there was a threshold of 20% for coalitions, 8% for individual parties, and 3% for parties in coalition.

This system, despite providing guaranteed majority government in the lower house, had a serious problem. In every election, the Senate has had a narrower majority than the lower house, and as every Italian government must have the confidence of both houses, this means that if a government has the 55% in the lower house, but a minority in the Senate, the lower house majority is useless.

This is the situation in the current parliament. The governing Democratic Party-Left Ecology Freedom coalition has a comfortable majority in the lower house, but needed the support of Mario Monti and Silvio Berlusconi's parties. After a split in Berlusconi's Forza Italia party, and the formation of the moderate New Centre-Right party, the government no longer needs Berlusconi's support, but still has a precarious majority in the Senate.

The system is also very disproportional. It is this disproportionality that has caused the Constitutional Court challenge, and which has meant that new Prime Minister Matteo Renzi has had to put together a new system

The new system

The new system has multiple changes. For a start, the lower house will have electorates, which will elect 3-6 members each. These electorates will have shorter closed lists. There will be a higher threshold of 5% for parties in coalition, 12% for coalitions and 8% for independent parties. There will also be a changed majority bonus. Instead of the largest party or coalition getting 55% of the votes, the largest party or coalition will get a bonus of 18%, as long as that party gets 35% of the vote. A second round is held if no party gets 35%.

Admittedly, the details are rather sketchy. I have been unable to find more information. But from what I have found, the system appears to be a rather draconian measure to produce a government. The thresholds will be one of the world's highest (as far as I know, only Turkey's threshold is higher). The possibility for coalitions to be created for purely opportunistic reasons remains. Closed lists stay, despite being shorter. A regionalist party can win a significant number of votes in a certain area, and still win no seats, while a party with a small number of votes in this area that crosses the threshold can win significant representation in that area.

Another problem is the size of parties in coalitions. If a coalition wins 35% of the vote and the 18% majority bonus, any party within that coalition may leave, an action that would certainly bring the government down. The new system does not apply to the Senate, either, though the Renzi government apparently intends to reduce the powers of the Senate.

Italy has many long-term problems with its electoral system. Removing the power of the Senate to vote confidence in the government will go a long way, but the new electoral system will not improve Italy's democracy, and will probably go down the same path as previous Italian systems.