Submission to Select Committeeinquiry Into the 2004 Local Authority Elections

1

Submission to Select CommitteeInquiry into the 2004 Local Authority Elections

by Keith Rankin

Senior Academic Staff MemberSchool of Accountancy Law and FinanceUnitec New ZealandCarrington Road, My AlbertPrivate Bag 92025Auckland

23 February 2005

Choosing the Best Method of Voting - Principles

New Zealanders currently experience five methods of voting:

1. Simple FPP, which appropriate only for 2-candidate elections where there is one winner and one loser. This is the system used to elect electorate MPs and most mayors. Political parties never choose FPP when selecting their leader, unless that is the contest has been reduced to 2 candidates.
2. Multiple FPP, where voters get to vote for as many candidates as there are persons to be elected, and each vote is considered to be of equal preference. This has been New Zealand's traditional method of voting for City/District/Regional Councils.
3. Simple STV, ("single transferable vote") – also known as Preferential Voting – which is used in a few mayoral elections in New Zealand. (It is most familiar as the method used to elect electorate MPs in Australia.)
4. Full STV, where more than one candidate is to be elected, and where voters cast only one vote which may be fully or partly transferred to a second or lower preference once the most preferred candidate is eliminated or is elected. This was used in 2004 for District Health Boards and some City/District Councils.
5. MMP ("Mixed Member Proportional") which is principally a contest between parties rather than between persons. In New Zealand, where it is used to elect the national parliament, it acts to correct for the inability of simple FPP to deliver a result which is even close to being representative. It may include Simple FPP (as it does in New Zealand) or Simple STV as a means of electing local representatives. Indeed MMP gives voters the opportunity to place a personal vote that is quite independent of their party vote. IT acts simultaneously as a national and a local election.

When choosing an electoral method, four broad principles need to be considered, in descending order of importance. (Accurate counting of votes is of course important in all methods of voting.)

1. Accuracy in translating voters' choices into actual results.
2. Transparency in allowing voters to see the results and believe in their validity.
3. Simplicity of casting one's vote; the minimisation of voter confusion.
4. A speedy initial counting of the vote that enables a substantial majority of candidates to know of their success or otherwise on election night. In other words, the pattern of party or factional support – the "guts" of an election result – should be known on election night.

While Principle 1 is clearly the most important, it is not the only consideration. Some advocates of unfamiliar electoral systems do not give sufficient weight to points 2, 3 and 4.

ACCURACY

By accuracy we mean that the election result accurately reflects voters’ preferences.

The most accurate voting method depends on more than mathematical considerations. It also depends on the extent to which voters know their candidates and what they stand for. Party (or ticket) affiliations become important for any election in which the majority of voters do not know many (or even any) of the candidates personally. Under any electoral system, knowing a candidate's affiliation is a near costless way of gaining significant information about a candidate who is otherwise unknown to voters.

In Parliamentary elections, local candidates are generally better known to voters than are candidates in local authority elections. However, candidates in power ("sitting" candidates) are almost always better known than their rivals.

Given the tradition of a strong party system (which New Zealand developed in the early 20th century) and the importance given by New Zealanders to party unity, an explicitly party-based system will give the most accurate results in a national parliamentary election. We have such a system in MMP, and it does yield a parliament that accurately represents voters' preferences on election day.

In local authority politics, different considerations prevail. At this level, government is much more about people than parties. However, the larger the population base, the more likely it is that voters will not know much about most of the candidates. Hence the need for a loose-fitting party system; a "ticket" or endorsement system.

It is in elections for local councils and community boards that STV is easily the most accurate method. First it enables people to vote for their preferred ticket, by ensuring that candidates on that ticket are their first, second and third preferences. In doing so, they give most weight for their preferred candidate within that ticket. That is, it gives them a way of choosing within a ticket as well as choosing a ticket, a very important consideration at the local level. STV, by splitting votes into fractions, also gives voters the opportunity to favour their preferred candidate from another party or ticket over the candidates they most dislike.

In Council and Community Board elections, multiple FPP is hugely inaccurate. In particular the system itself works on the "all or nothing" principle. If all voting was ticket voting, each ward and board would register a clean sweep for a single ticket. If an election where there are just 2 factions follows an election contested by 3 factions, then the sitting faction may well suffer an apparent “landslide” defeat, even if when its voter support increases. Such a case was Auckland in 1995, when Alliance councillors were defeated despite gaining more votes. The deciding factor was the coming together of the two non Alliance factions. Under STV, the Alliance could not have dominated councils as it did from 1992. It would have increased its representation in 1995, in line with its increased vote.

Community Boards elected by multiple FPP can be quite unrepresentative of their communities. They act as little more than a mouthpiece for the faction who happened to clean sweep the most recent election. If another faction clean sweeps the next election, all sitting Board members will be defeated, removing any continuity of representation from one Board to the next.

Mayoral elections under STV are no more complex than are Australian electorate elections. They give more accurate results though, in that they prevent a candidate who is rejected by the majority from becoming (or staying on as) mayor. Examples of such mayors who won despite being disliked more than they were liked include John Banks (Auckland 2001) and Sir Barry Curtis (Manukau City 2004).

Mayoral elections under simple FPP are something of a lottery. In 2004, John Banks, who was soundly defeated, polled much better than did Sir Barry Curtis who retained his office due to the anti Curtis votes being split between two candidates. In an earlier example, Les Mills, when he first became Mayor of Auckland, "won" with about 25% of the vote. The remaining 75% was split between 8 other candidates.

TRANSPARENCY

Transparency is all about perceptions. An optimal voting method should be accurate. It should also be seen to be correctly counted, manually verifiable, and difficult for unscrupulous or corrupt persons in power to falsify the results.

It is here that FPP, which is simple to count and to tabulate the results, has an advantage over STV. Also, any familiar method of voting has an advantage over an unfamiliar method.

To satisfy the transparency criterion (criterion 2) without sacrificing the more important accuracy criterion (criterion 1), an alternative version of STV to that used in New Zealand in 2004 can be adopted.

To be called STV, the method must involve a single vote, even when more than one representative is being chosen. To be transferable, an opportunity to mark at least a second preference must be provided.

To be transparent, results must be able to be tabulated in a way that can easily be understood. And the votes should be able to be formally recounted by hand in the event of a disputed result. A simpler counting algorithm than Meek's Method would be preferred, therefore. Most STV counting sequences will result in the same people being elected. The more complex counting methods are subject to diminishing returns; the miniscule gains in accuracy are outweighed by substantial losses of transparency and public credibility.

SIMPLICITY

The third criterion for choosing a voting method is simplicity or user friendliness. Democracy requires that everyone has the opportunity to cast a vote, not just those who can read and carry out unfamiliar instructions.

Simplicity is an advantage of FPP voting, but is not a large advantage. Certainly the 1978 New Zealand general election led to a very high number of invalid votes despite being conducted under FPP.

MMP is slightly more complex, with voters having two votes. The biggest problem here is not knowing what to do, but in understanding the importance of each vote. Some voters in 1996, 1999 and 2002 may have given undue weight to the FPP electorate vote that perseveres as a part of the MMP system, treating it as a party vote when it is really a personal vote.

STV is said to be too difficult for many voters. Certainly, it can be complex. But it does not have to be. In the 2004 local body elections, it seemed difficult because in most cases STV ballots were mixed with FPP ballots. It is also potentially complex in that it seems to give voters too much choice; ie "vote for as many or as few candidates as you wish".

The remedy is to make it simple by making the instructions universal. As noted, STV is any system in which just one vote is cast, but which votes may be fully of partly transferred to a second or subsequent preference.

I recommend the use, in all local body elections, of what can be called Trifecta Voting, named after a popular type of bet in horse racing. Trifecta voting limits voter selection to just three preferences. It is the form of STV which is the simplest to instruct voters, while also being less complex to count.

The instruction to voters would be the same, whether one or many representatives will be elected. It would say:
"Vote for three (3) candidates in order of preference, using the numbers 1, 2 and 3."

(The only exception would be in a contest where there are only two candidates. Such a contest always reverts to an FPP vote, and the instruction would have to be:
"Vote for one (1) candidate, using the number 1". )

A vote for less than three candidates would not be invalid. Nor would a vote for more than 3 candidates be invalid. The remaining preferences would simply be ignored.

As race horse betting figures show, punters have little difficulty in ranking their three preferred horses, which they must do to make a trifecta bet. The word "trifecta" taps into working class culture, removing the common perception that STV is middle class and geeky.

Indeed, I believe that kiwis will take very well to trifecta betting, to the extent that it could become known world-wide as the New Zealand system of voting. Further, it would become an obvious system to use in the candidate/electorate vote in the MMP national elections.

Trifecta voting would achieve just about everything in terms of accuracy that complex STV (eg STV by Meeks method) would achieve. It gives voters a chance to vote for their preferred candidate, while also being able to vote against candidates they do not want by choosing two acceptable back-up selections.

COUNTING

FPP and MMP are both simple to count, as no votes are transferred, and ballot papers rarely need to be repaired. Invalid (or informal) votes are a very low proportion of total votes due to the basic simplicity of the voting instructions.

With Trifecta STV as the sole method of voting in local elections, the numbers of informal votes are unlikely to be much higher than in an FPP election. The opportunity to repair ballots by fractionating votes where more than one candidate is chosen but no preference is shown can reduce informal votes to almost zero, leaving as invalid only those papers that are intentionally spoiled.

Trifecta STV will be quicker to count by computer than by hand. However (and this is important for transparency purposes) it should always be possible to get the same result from a manual recount. By limiting to three the number of preferences allowed to voters, even a manual count would be able to be completed on the election evening. (We should note that, where just one candidate is to be elected, all forms of STV are equivalent to the preferential voting system that Australians use to elect their MPs.)

The steps for counting could be as follows:

1. Count the number of votes to be counted.
2. Determine the number of votes a candidate requires in order to be elected (the quota).
3. Count the Number 1s (ie the first preferences).
4. Declare elected any candidates whose votes exceed the quota.
5. Whenever a candidate is elected, transfer the excess fraction of each vote for the elected candidate to the voters' second preferences (or, in cases where the 2nd choice has already been elected, to the third preferences).
6. Whenever there are no remaining candidates reaching the quota, but there are still places to fill, then the lowest polling candidate will be eliminated and have his/her votes switched to their second preferences. (If the second preference of such a transferred vote has already been elected, then the vote will be counted for the third preference.)
7. Towards the end of each count, there will eventually be a situation where there is one place to be filled and two remaining candidates. The final place will be filled by the remaining candidate with the greater number of votes.

The published result could be a table much like an Olympic Medal Table. Each candidate would have 4 numbers after his/her name: number of 1s; number of 2s; number of 3s; ranking.

The first candidate to be elected will be ranked 1. If 7 candidates are required to be elected, the winning candidates will be those ranked from 1 to 7.

Conclusion – New Zealand Local Authority Elections

Traditionally all mayoral elections have been held in New Zealand under the Simple FPP system, while all other local authority elections have used the Multiple FPP system. This has created huge problems: dramatic swings, especially on Community Boards and Regional Councils; groups of candidates (“tickets”) being swept out of office despite receiving increased support (eg Alliance candidates in Auckland in 1995); and many successful Mayoral candidates receiving electoral endorsement from less than half of the voters (which generally means less than a quarter of eligible voters).

The problems with the FPP system are so great that we should not continue with it.

STV was introduced in some of the local elections in 2004. Sadly, the election was blighted by ballot papers that included both FPP and STV, and a by complex STV counting system that was too slow and not-at-all transparent.

On the most important accuracy criterion, all STV systems are vastly superior to the status quo FPP systems. FPP however has advantages on the other three criteria: transparency, simplicity and speed of counting. Indeed the advantages of FPP in these areas are seen to be sufficiently great to lead many media commentators and talkback hosts to prefer FPP over the more accurate STV.

The solution is to adopt a form of STV that is transparent, simple and quick to count.

Trifecta voting is a form of STV that meets these requirements with almost no loss of accuracy compared to other forms of STV. Indeed, if its simplicity reduces the number of spoiled ballots, then trifecta voting provides an even more accurate count of who voters prefer to represent them.

MMP is better than STV for nationwide elections where the principal contest is between parties and their contestable policy programmes. STV is much better than MMP for local elections, where it is the individual people elected that matter most, and where tickets do little more than give voters guidance as to where individuals are likely to stand on different issues.

The transition to Trifecta Voting ("TriV", pronounced "try-vee"?) would ideally be accompanied by a switch to Trifecta STV for electorate voting for the national parliament. However, to be politically realistic, electorate elections for Parliament could not switch to STV until some form of STV has already proved its worth in local elections.

In 2007, all local elections should be conducted using the simplified "Trifecta" form of single transferable voting.