Jenkins reheated


With the resurgence of interest in proportional representation in the UK, and calls for an electoral reform referendum at the 2010 general election, I tracked down and read the report of the Jenkins Commission, which proposed a PR system for the UK House of Commons in 1998.

The 1997 Labour manifesto promised a referendum on proportional representation, and upon election the Blair government appointed an independent commission headed by Lord Jenkins, former President of the European Commission, Deputy Leader of the Labour Party and founding figure in the Liberal Democrats.

It’s a fascinating proposal. It’s essentially a modification of the German MMP system, called Alternative Vote Plus. Essentially, 80-85% of MPs would still be elected by constituencies, although they would be elected using Australian-style preference voting. In addition, “top-up” seats would be elected in a large number of regions, with the Jenkins Commission proposing 80 regions for the entire UK, including 65 in England.

I’ve come around to the idea of an MMP system as a way of reducing the impact of PR on our culture of electorates, although I still would prefer a Hare-Clark/STV system. The main problem I see with the Jenkins model is that it really isn’t a proper proportional system. The combination of the low proportion of the MPs elected by top-up lists and the division of these lists into 80 regions means most regions only elect 1, 2 or 3 top-up MPs. This will mean that, in many places, one party will win more constituency seats than their total allocation, and any reallocation will largely be limited to the major parties. Any party smaller than the Liberal Democrats would be lucky to win any seats.

This is particularly bizarre when you consider that where MMP is already used in the UK, in Scotland, Wales and London, almost 50% of the representatives are elected from top-up lists. I tend to think that, if you increased the number of top-up MPs to about 30% of the Parliament, and reduced the number of regions used to elect these MPs to the same constituencies used to elect the European Parliament.

It is fascinating that the UK now has a flourishing electoral reform movement, led by the umbrella group Make My Vote Count, which includes the fantastic Electoral Reform Society. We have nothing like that in Australia. While we have a Proportional Representation Society, it is a tiny group that really is more of a society of interested people than a campaign group. In particular, they have adopted a model for the NSW Parliament which is bizarre and completely impossible to implement.

I had an interesting debate across Twitter with Possum on Friday regarding the possibility for PR in Australia. Some people tend to assume that, just because PR is not in the interests of the major parties, it cannot be implemented. This ignores the fact that major parties in New Zealand, the UK and various Canadian provinces have moved to various degrees towards implementing PR. The ALP has also shown a clear preference for PR in upper houses in Australia, which demonstrates some appreciation of the benefits of the system.

However, all of those countries saw PR became an issue on the agenda once a political campaign group began actively campaigning, lobbying, signing up members and getting media attention. We are a long way away from that here in Australia. Such a campaign group was always in place to be ready for a future political crisis. We can see this in the UK now, where decades of campaigning by the Electoral Reform Society has put them in a position to take advantage of the current political crisis.

This makes me think there is room to move on this issue in New South Wales, as a starting point for future campaigning. If an Electoral Reform campaign could be started over the next year to be pushed during the 2011 election, I believe it could gain traction with voters tired of the current government. Considering that Barry O’Farrell has already opened the door to constitutional change by questioning the current fixed-four-year-term arrangement, I believe there is an opening to pressure the Opposition on some sort of constitutional debate, such as a constitutional convention, royal commission or citizens’ assembly.

Liked it? Take a second to support the Tally Room on Patreon!


  1. Why is the Lesslie Model “bizarre and completely impossible to implement”? It seems a bit strange and unlikely to work, trying to make the seats in NSW/Vic/Qld cross sections of the state and getting rid of territorial senate representation, but I don’t think that’s the same as bizarre and impossible…

  2. I certainly think it is bizarre, as far as Single Transferable Vote goes.

    The PRSA supports Robson Rotation, which means you expect voters to rank candidates within each party according to preference. Yet how can you expect voters to know candidates when there are 19 seats available (in the NSW state model, but the federal model isn’t much better) and covering a massive area with little or no community of interest.

    When you have a dozen or more seats in a single electorate, you will have massive numbers of candidates running. In these cases, it can only work if you effectively turn STV into a party-list-preferences system as it is in NSW and South Australia and to a lesser extent in the other mainland upper houses and the Senate. In that case you lose the benefit of STV in maintaining local links for MPs and competition between MPs of different parties.

    I think it’s impossible to implement because it is so far off the chart as to make it impossible to be achieved politically.

    As a related point, I actually think it is a fundamentally bad model. If you’re going to divide New South Wales into wedge electorates, then you might as well adopt the Israeli model and abolish all electorates. There is value in having no more than 7 MPs per seat, retaining some local community of interest and giving voters an opportunity to make a connection with their MPs.

    The Lesslie model seems aimed to deal with one problem, which is electoral stasis, the problem where, even in a quota system, the quotas are so large that no reasonable swing would see a seat change hands. Say there are four seats in a district and two parties each have 50%. If you assume that a 10% swing is too dramatic to be considered reasonable, then there is stasis, in that neither party will bother to put effort in, and all MPs become safe. Of course, this assumes no intra-party competition. It’s safe to say that the NT senate race is in stasis, in that neither major party has hope of winning the other seat or fear of losing their current one, and no minor party has a chance.

    But ultimately, no electoral system is perfect and I think the Lesslie model throws the baby out with the bathwater, creating a shockingly bad system to deal with a minor problem.

    I don’t particularly want to spend this debate talking about the Lesslie model. It’s a random model with no prospect of implementation and no precedent around the world.

  3. Hey Ben, I think you’re being a little harsh on the PRSA, they’ve done a fair amount of campaigning on electoral reform here in Victoria, which was definitely valuable when it came to getting upper house reform here in Victoria. Sure, the ALP was supportive of it, but the PRSA’s formal submission and their extensive participation in the public consultation meetings really helped, as they were seen as experts in this area, and non-party political. They have also been consistently making extensive submissions when local councils have come up for review here in Victoria, and more councils here have been moving towards multi-member wards with STV.
    I agree with you about Robson Rotation, btw. I can see the point of Robson rotation when the alternative was ballot papers ordered by a candidates name alphabetically, and when no party affiliation was listed – but that isn’t the case now, and the double-blind method for ordering ballot papers is a fair way to randomise the advantage a candidate may get from donkey votes.

  4. Ben, I haven’t been following UK politics as closely as I was a few years back, so you’re no doubt better informed about the current situation than I am, but I recall that after the 2005 election there was a big campaign launched by The Independent and the Lib Dems to push for PR. I would imagine that having the cause supported by a media outlet, and adopted as a major issue by a large party has been very helpful towards building their electoral reform movements. They would also seem to have a more widespread understanding of the shortcomings of the present system in the UK, as they have more familiar examples of PR in Scotland, Wales and Europe to point to, and the present first-past-the-post system is more manifestly unfair (especially for a multi-party democracy) than our preferential system.

    Whilst we have plenty of examples of PR in Australia, most people, even many who are pretty engaged with the political scene, don’t really understand them. The NSW upper house for instance is a complete mystery to most people, and few realise that the current government doesn’t actually have a majority in both houses. My personal favourite though was, back in 2004, having to explain to a House of Reps candidate that the Senator whose re-election he was supposed to be supporting actually represented all of NSW, not some mystery seat somewhere in Sydney.

    There would seem to be some pretty good arguments that can be made for reform, especially in NSW, where the fact that our current government won a majority in the lower house with only 39% of the primary vote might suggest to many people that there is a good case for change.

    Perhaps if the Greens were to make PR a major campaign issue the cause might gain some more traction, but yes, a non-aligned campaign with widespread grass-roots support is probably going to be vital if reform is to occur. The current state of NSW politics certainly presents an appealing opportunity to press for reform in many areas.

  5. Polly,

    I wasn’t aware of what the PRSA did in Victoria, and it’s probably true they’ve had success. But it’s still not a proper campaign group like I would like to see. My point regarding the electoral model was to do with the NSW PRSA.

  6. I don`t think that there is time for a referendum before the next general election in the UK.

    A hung parliament is the best hope for the PR in the UK (because of the LibDems and various smaller parties) and the best chance of a hung parliament is if Labour brings in either preferential voting or supplementary voting (a truncated version of preferential voting where voters only get 2 preferences and all the candidate elimination is done at once) (the latter is more likely) because the LibDems will get more sets because seats where they come second and there is no majority, they are likely to win.

    The other hope for PR in the UK is if the government is taken to the European Court of Human Rights for enforcement of Protocol 1 article 3 of the European Convention on Human Rights but this may not work.

    The plan shown by the NSW PRSA is a bit nutty but the main and Victorian parts of it seen sensible.

  7. That proposal for NSW is just so wrong. Why would you create 19 member selectorates and then use Hare-Clark and fully optional preferential voting. It just distorts the whole idea of ‘proportional representation’. Fine if you are electing small numbers of members, but for 19? List PR with open candidate choice is the way to go if you are electing 19 and trying to overcome the problems of party control over candidate choice. But Hare-Clark for 19 members is just wrong wrong wrong.

  8. The Guardian and The Observer were the first to call for significant constitutional and electoral reform, but its since spread into virtually all the British media and the calls for change within parties seem to growing.

    What I find interesting is that whilst the expenses scandal was the catalyst for this discussion, most of the reforms proposed have very little to do with expenses. Brown’s plan to “overhaul the expenses system” has been completely swamped by much more radical ideas, being proposed by those in his own party, like PR, weakening the power of whips, updating parliamentary traditions, devolving some powers to local governments and giving individual MP’s more power when determining the make-up of select committees.

    So whilst you could say “The expenses scandal is unrelated to Australia” (which I would dispute, but let’s leave that proposition there for arguments sake), you can’t make that same claim about virtually every other reform proposed.

    All the problems currently being highlighted by MP’s of all stripes, the media, lobby groups, activists and high-profile individuals are inherent in the Westminster system and exist in Australia as well. In the UK they had the expenses scandal to galvanise support for reform, but if they’re are analysing their structures and discovering major flaws I don’t know how long we’ll have to wait before someone in a position of influence will say “Hey! We have those same problems too!” and attempt to fix them, as hopefully the UK will.

  9. Can I just make it clear I’m not against the PRSA. I think they are right in supporting PR-STV in general (although I think a PR group, while supporting STV, shouldn’t rule out types of PR like MMP – it’s not as good as STV, but it is massively better than single-member system), and they probably have been very active in helping Victoria change it’s Legislative Council electoral system and shifting council electoral systems towards PR.

    I just think that the NSW model is ridiculous, and that we need a much broader and bigger and more ambitious campaign group more like Fair Vote or the ERS.

  10. Ben, if you look around on the web, at the start of last year, Jack Straw issued a paper stating the government’s position on electoral reform. It looked at the experiments that had been tried, in Scotland, in Wales, in the London Assembly, and looked at options. It then sqibbed and came to the conclusion that reform of the Commons’ electoral system should a await a resolution of what to do with the House of Lords. Making one unresolved problem dependent on another.

    The original push for PR came about because the Labor Party was of the view in the 1990s that it would struggle to get more than a single term in office so intended to introduce PR in the hope of forming future Coalitions with the LibDems. After getting into office, it started to dawn on a few members of the government that they might be on office for while snd so the idea of giving up government in their own right for the principle of PR dimmed.

  11. If only Labour and the LibDems had taken a few more seats off the Tories in 1992 the UK would have STV and be a better place. The privatisation of the railways under Major would not have taken place and earlier peace in Northern Ireland to name a couple of examples.

  12. I understand what you’re saying, Antony, about the position of the government, but I think the current news suggests the government’s position is up in the air now. I’m sure Jack Straw and Gordon Brown don’t want PR, but they may not be in a position to make the decision.

  13. I don’t doubt it Ben. I know from personal experience there are some UK Labour MPs who have always wanted our preferential voting system, though Jenkins’s Report pointed out this would have inflated Labor’s victory in 1997 even more than first past the post. That inflation is exactly why some Labor activists now find it attractive.

  14. I should also say that there was a joke about the Jenkins Commission amongst UK political scientists that after examining all the other electoral systems, Jenkins went and invented a new one. The Jenkins model was an orphan at birth, the government having already decided on a non-preferential system for Scotland and Wales, and the long term advocates of the single transferable vote unhappy that Jenkins went for a mix of the alternate vote and list PR.

  15. It’s also interesting that the UK is the only place to consider combining preference voting with MMP. Considering that MMP attempts to ensure that a minority party doesn’t win a majority on a plurality of the vote, it seems bizarre to tie that in with FPTP. I guess Germany and other such countries wouldn’t have had preference voting as an option to consider, but I’m surprised NZ didn’t consider including preference voting in their model. Do you know if this was ever considered?

  16. The Lesslie model is a package. It requires not only optional preferential voting but also the Robson Rotation. This combination means the vast majority will vote for a successful candidate and the informal vote will plummet. Even if every voter voted for a single candidate, there would be very few exhausted votes. Of course, most voters will not stop at just one candidate. (An analysis of ACT House of Assembly voting will confirm this.)

    In Australia, the geographical size of electorates is always going to be a problem. It should be noted that the two jurisdictions with excellent STV proportional representation systems (the ACT and Tasmania) are jurisdictions in which the population is reasonably uniform. Every other jurisdiction has to deal with the problem of massive population density within the capital city and vast underpopulated regional areas.

    Conventional PR systems returning five to seven members would require geographically huge electorates in regional NSW. Such electorates would be larger than any that I propose. Some critics argue against the combination of city and country electors – yet this is exactly what the Senate does.

    Regional electors are not disadvantaged by the Senate voting system, as major parties always include regional candidates on their tickets. Regional voters would be better represented if this system was applied to the House of Representatives – they are likely to single out their preferred (and local) candidates, whilst their city cousins let the Robson Rotation work. The net result is an increase in support for regional candidates.

    Ben Raue is concerned that large electorates will suffer from lack of community of interest, but a voter does not need to connect with all nineteen members. How well do you know your local member and the members of the surrounding 18 electorates? You don’t need to know all of them – you just need to choose one.

    Antony Green should know that the Legislative Council is a Hare Clark electorate returning 21 members. Demonstrably, large electorates do work and do provide genuine representation (just ask the Greens!) – the introduction of the Robson Rotation and optional preferential voting are all that is required to give voters genuine choice.

  17. Nice to hear from you Stephen. A few points regarding what you said:
    1) Of course, it would be very difficult to cast an informal vote under your model, but you would still have the problem of a tablecloth ballot paper and it would be very difficult for voters to cast informed votes that give their preferences value.
    2) Of course there will always be geographically massive country
    electorates in Australia, which will be even larger under any STV model. I’m willing to accept that it can make it harder for MPs to represent that area and to campaign. But I don’t think that’s a good reason to make all electorates massive electorates as well.
    3) I think the point of electorates is to have some sort of community of interest. If you have five electorates that all include part of inner Sydney and the fringe of the state, all MPs will effectively represent the whole state. If you’re going to do that you might as well do what Israel do and have no electorates with a 5% quota. When you bear in mind the inability for voters to make a connection with such a large group of candidates and MPs, the only point of electorates in this model is to impose a 5% quota (since electing 95 MPs as a single group would produce a quota of just over 1%).
    4) The NSW Legislative Council (or any mainland upper house or the Senate) is not a true Hare-Clark system. Hare-Clark is generally considered to include Robson Rotation and local electorates. The NSW Legislative Council is effectively a party list system with preferences.
    5) I’m not saying that the NSW Legislative Council and the Senate don’t provide good representatives to their broad electorates, but it can’t be considered to be ‘local’ representation. All of the models which have been seriously considered in the English-speaking world include some degree of local representation, be it BC-STV, the Jenkins AMS system, NZ MMP or the Scottish system.
    6) Apart from everything else, it’s so far out of left field that it fails the pragmatism test. Of course, there’s a degree of idealism in this whole debate in a country where the major parties are steadfastly opposed to electoral reform. But at least you need to consider other systems that have been implemented around the world and maintains some continuity.

  18. Noting some similarities between PR reformists and the republican movement.

    Everyone agrees on the principle but spend more time arguing amongst themselves for the best model then trying to get others on their side.

  19. I utterly disagree. Conventional PR systems do not return 5 or 7 members. Conventional Hare-Clark systems might, bot most PR systems return more members so that proportionality can be achieved.

    It is too late at night for me to do this in detail so will return to the topic tomorrow. I think electing so many members under optional preferential voting and a Hare-Clark counting system is madness. The system would be wide open to distortion of proportionality caused by exhausted preferences. When you are electing so many people at once, the number of votes wasted with candidates who aren’t elected is so small, then what’s the point of all the preferences. The NSW LC system currently uses a Hare-Clark counting method, but it effectively operates as a List PR system with a highest remainder method to determine the final seats. When you are electing so many members, why use this sort of system when instead you could ditch the preferences and ditch the quotas and use a fairer divisor system and include open lists to allow voters to choose candidate from within a party list.

  20. At the 2007 NSW Legislative Council election, more than 80% of votes from excluded candidates exhausted rather than take part in the election of a candidate. 1.57 quotas worth of votes had exhausted by the end of the count, and that would have been closer to two quotas had the final Australian Democrat candidate been excluded. The last three vacancies were filled with the less than a full quota. 16 of the 21 quotas were basically filled on the first count, the second Green candidate was elected when the votes for the rest of the Green ticket finally trickled to the head of the ticket after a couple of hundred counts, and the last four candidates elected were exactly the same as would have been elected had a list PR system been in operation with a highest remainder method of filling the final vacancies. Since the new NSW election system has been in place, all the operation of Hare-Clark has not shifted the result one iota from what would have taken place under List PR with a highest remainder method to fill the final vacancies.

    I should also point out that because NSW, like the ACT, excludes exhausted preference before calculating transfer values, if the Tasmanian rules had applied, then even more votes would have exhausted in the count.

    Those statistics were produced by a system that had a minimum 15 preferences applied. It would have been even worse and ultimately distorted proportionality had full exhausted preferential voting applied. In conjunction with Robson Rotation, this would have produced exhaustion of preferences within party tickets, not just between party tickets. Under the proposed systems with 19 members and a low quota, a party would find its vote going backwards during the count because of exhausted preferences within the ticket.

    This has the perverse effect that a party would be disadvantaged if it stood more candidates than it had the chance of electing because the more candidates it stood, the higher the exhaustion rate and the more its vote would slip during the count. If a party only stood as many candidates as it thinks it could elect, then you have extreme difficulty introducing a countback system to fill casual vacancies.

    Any high profile candidate who gets beyond a quota in their own right but doesn’t have many preferences flowing to other candidates will also distort the proportionality of the system and create a lottery in the final vacancies being left to whoever remained at the end of the count.

    In summary, I don’t see that Hare-Clark is a sytem you would choose to elect 19 members. The use of preferences in the count distorts the proportionality that could easily be achieved with a highest average algorithm to allocate seats, and just as easily ensure that the minimum number of votes are wasted in the count. Second, if full optional preferential voting were used, the proportionality is further distorted by the final vacancies being filled by whoever remains at the end of the count, resulting in average representation by party being distorted in a way that would not be acceptable to most systems of proprtional representation.

  21. Another example. Campbelltown Council elected at last year’s NSW local government election. 15 councillors elected at large using a very simlar system to the Legislative Council. 8 preferences were required for a formal vote so all groups stood 8 candidates to qualify for an above the line group voting square. The first 11 candidates were all elected on the full quotas. After the distribution of preferences, two full quotas of votes had exhausted and the last four candidates were elected with less than a quota. These four candidates were the four that had the highest partial quota at count one. Hare Clark counting was used, but when electing 15 members with semi-optional preferential voting, the electoral system turned into List PR with a highest remainder system to fill the final seats.

  22. I know Campbelltown Council well, as I ran there in 2004 and 2008.

    I don’t like 3-councillor wards. I would prefer 3 wards of 5 councillors. One would be centred on Ambarvale, Rosemeadow and Campbelltown, another on the suburbs around Raby, Eagle Vale and St Andrews, and the third around Glenfield and Macquarie Fields, with the middle parts divided between those three wards.

  23. To Ben,

    I actually started with the perceived wisdom that five, seven and perhaps nine member electorates was the way to go. But I could not solve the challenge drawn up below. I think you are honour bound to produce a model that you believe will deliver a fair and proportional result.

    Can you draw up electoral boundaries for the STV election for the House of Representatives 48 NSW seats? Current electoral boundaries give a good idea of population densities in regional NSW.

    Certain conditions must be met:

    1. Exactly 48 seats (Constitutional requirement)
    2. No 2,3,4or 6 member electorates.
    3. No electorate to be in electoral stasis.
    4. Regional electorates should be of reasonable geographic size to maintain a local sense of identity (Note: I think this is a furphy but you seem to think it important.)
    5. Quotas need to be large enough to prevent a raft of bigots and racists being elected.
    6. Quotas need to be small enough to enable genuine minor parties such as the Greens to be elected.
    7. There should be no artificial devices, such as thresholds, forcing voters to consider whether or not to vote tactically.

    To Antony,

    It is a good scientific principle to compare like with like.

    All the examples that you give for exhausted votes are from ballots that have separate party voting boxes above the line. These party voting boxes effectively discourage voters from continuing their preferences beyond their initial choice. The number of candidates standing in the party list also gives the impression of “don’t worry; your vote will count through to number twenty one.”

    My model does not have separate party boxes and the number of candidates is self limiting. With the Robson Rotation it is an advantage to concentrate the votes on the number you expect to be elected (plus a couple more, hoping for a favourable outcome). This in itself greatly reduces the exhausted rate as there are no exhausted votes from successful candidates. A high electoral deposit on unsuccessful candidates, not just unsuccessful parties, will concentrate the mind.

    You should use the ACT House of Assembly elections for valid comparisons.

    In the 2008 ACT elections there were no party voting boxes. The Robson Rotation and a fully optional preferential voting system were also in place. The figures clearly demonstrate that when there are continuing candidates within a party group the exhausted rate is miniscule.

    Furthermore, even when there were no more candidates left in a party, many voters still continued on to other candidates and parties. For example, a third of voters for the Australian Motorists Party (AMP) continued on to other parties and candidates when the last of the AMP candidates was excluded in the three ACT electorates This occurred despite the fully optional preferential voting provision and the parties choosing to run a full slate of five or seven candidates.

    To Ben and Antony,

    Table cloth ballot papers are a concern that can be easily solved. If all the makeweight candidates were removed from the last Legislative Council elections the number of candidates would be reduced from 333 to about 50. The last election should have consisted of 12 Labor; 12 Liberal/National; 4 Greens; 2 CDP; 2 Shooters; 2 Democrats and no more than 10 others.

    How can this be done?

    1. Remove the requirement to run 15 for a party group.
    2. Allow optional preferential voting and groups do not need to run excess candidates to ensure a formal vote. The groups in Campbelltown had to run at least eight candidates for that reason even though most groups would have been pleased to just have their lead candidate elected. In my local government experience most makeweight candidates need to be assured that they won’t be elected!
    3. Robson Rotation encourages groups not to scatter their support.
    4. High electoral deposits on candidates and not just parties. Voting is an important duty for Australians and should not be undermined by frivolous candidates. Reduce the threshold required for a return of the deposit if you think it undemocratic but it will make no difference these groups never even reach 1%, let alone 4%, of the vote.
    5. The ballot paper would be substantially reduced by the elimination of above the line voting with the removal of the boxes and the big black line.

  24. I meant to say 5 councillor wards but I made a brain to page error.

    The 48 seats are only constitutionally mandated for the current parliament size. If the Senate was put back to the constitutional minimum then NSW would have 24 and there is no upper limit on the size of parliament.

    The Greens seem to do quite well with 7-member electorates. 7-member electorates should almost always have their tipping points within reach of the 6% either way swing.

  25. Sorry to clog up your blog Ben, but I’m not letting go on this one.

    Stephen, I’m quite happy to compare like with like. But, an ACT electorate of 7 members, around 100,000 electors, quota of around 11,000 votes and in a single city with city wide media, is not in my opinion a like with like comparison to an electorate covering a quarter of NSW, containing around 900,000 voters and electing 19 members with a quota of 45,000. The comparison is even worse with the Legislative Council.

    The justification for Hare-Clark is that it is a system that avoids the vote wastage problem of single member electorates but at the same time maintains a link between members and constituents. As a by-product it is a semi-proportional system, but its operation is about personal rather than party voting. Once you get into large electorates, such as those you now propose, that personal link breaks down and you end up having to sell Hare-Clark as a proportional system when it is less efficient at producing party proportionality than other PR systems. I’m happy to talk like with like, but I am completely unaware of anywhere using Hare-Clark in such large electorates.

    I deliberately said party proportionality in the previous paragraph because in large electorates, that’s what voting becomes as the electorates are too large for most people to have real knowledge of candidates. In those circumstances, the only way to avoid Hare-Clark turning into Party List PR with a highest quota remainder is to advocate an obscure mechanism like Robson Rotation.

    Looking at your paper, my view is that you need to be more honest about the real impact of Robson Rotation and why it has been introduced elsewhere, and also be more up-front about how much your examples rely on parties running exactly the right number of candidates.

    For all the talk of Robson Rotation being about randomising the donkey vote, its introduction in Tasmania came about because the Lowe Labor government backed Liberal MP Neil Robson’s opposition bill in a deliberate attempt to undermine a decision by the Labor Party state executive to introduce a how-to-vote card with an official list of preferences. When rotation was first introduced in the ACT, it was an opposition backed measure partly aimed at undermining the Labor Party’s ability to get more booth workers out distributing how-to-votes and controlling preferences. In both jurisdictions, Robson Rotation was implemented before bans on how-to-vote cards. Sure it randomises the donkey vote, but the real justification has always been about restricting the ability of party groups to determine which of their candidates get elected, either by preventing the party ordering the list or by making it very much harder to use how-to-vote cards. It has never been implemented as a mechanism to average party vote across candidates, which is an implicit assumption of your paper.

    In the ACT, the first version of Robson Rotation produced a strange election result in 1998 where all seven Labor candidates in Molonglo were relatively unknown to the electorate, so they all ended up with roughly the same vote. People voted for the party not the candidate, and the effect of Robson Rotation was to randomise the vote across the candidates. As there were only 7 different rotations of the ballot paper, the small number of rotations played a part in determining which candidates were elected.

    So the argument shifted from ‘donkey’ votes to ‘linear’ votes, increasing the number of rotations to randomise all preferences rather than just donkey vote preferences because these pesky people were voting for the party without concern for candidate. Arithmetocracy was implemented because the people weren’t behaving in the right democratic way to suit the electoral system.

    In the ACT, the 5 member divisions require 60 different versions of the ballot paper, the 7-member district requires 420 versions. To get the rotations in every column equal, the number of rotations is equal to the first number divisible by every number less than the number of vacancies. But if you have more than 7 vacancies, the number of rotations required to get Robson Rotation working sky rockets, 2,520 for 9 person tickets, 27,220 for 11 and 360,360 for 13 or 15 person tickets, over a million when you get to 17. So I presume the number of candidates that can be nominated on a single ticket will be limited to overcome this problem.

    Which brings me back to your responses about parties only standing as many candidates as they hope they can elect. Yes, the system you propose will work best if parties get this matching of candidates to vote right. But I think all the tables of analysis in your paper fall apart if this condition isn’t met. Yet nowhere in your paper is it mentioned how important an assumption this is.

    All of the tables in your NSW proposal have three underlying assumptions that are not mentioned at all. These are that (1) parties should only stand as many candidates as they expect to elect, that (2) there will not be exhausted preferences between candidates on a ticket, and (3) that Robson Rotation will magically equalise the vote for a party across all the candidates.

    All the analysis tables in your paper are using a ‘highest average’ method of producing estimates of how many members each party elects. But you are assuming that Robson Rotation will do the averaging across candidates in each ticket for you. And all the analysis falls apart if the three conditions in my previous paragraph are not met.

    A much better approach would be to use a real PR averaging method to allocate seats to parties, and then use an open list ticket system to allocate seats within party. Your model relies on Robson Rotation to do both of these functions and I reckon it would be lucky to do so.

    This model does not have a hope of getting anywhere not only because the major parties would lose control over ordering their ticket, but because its operation would be unfair to parties whose vote gets averaged over lots of candidates, but advantageous to minor parties with one high profile candidate. Once parties stand the wrong number of candidates, individual candidates get more than a quota or preferences start to exhaust, the system starts to behave like List PR with a highest remainder algorithm and Robson Rotation starts to punish the larger parties.

    You argue the ACT doesn’t have a problem with exhausted preferences. True, but its ballot papers do tell voters to fill in either five or seven preferences. How many will be recommended on NSW ballot papers, and with the smaller quotas in electing 19 members, how many quotas will ended up being wasted in exhausted preference?

    Finally, your comments on implementation for NSW are not correct. You cannot implement any part of your proposal for either house without a referendum. The use of single member electorates in the Legislative Assembly is mandated by Section 27(2) of the Constitution, counting in single member electorates by Schedule 7 of the Constitution and both provisions are entrenched by Section 7B of the Constitution Act and can only be amended by referendum. You cannot implement your proposal for the lower house by Act of Parliament as suggested by your proposal on Page 13.

  26. One other point in addition to yours Antony: unlike in party list systems, I doubt parties would only run the number of candidates that can win.

    In the ACT (maybe Tasmania too? not sure) the Liberal and Labor parties generally run a full ticket and campaign for all of them, as Robson Rotation individualises the campaigns and means that the party sometimes avoids making a choice by leaving that up to the voters. You could have a situation where Liberal and Labor both run 19 candidates in one of those seats.

  27. It also raises the problem that who is going to agree to count back as a method to fill casual vacancies when the system used at general elections would punish a party that ran extra candidates.

    Tasmania requires a minumum number of preferences equal to the number of vacancies. That is why the Greens run full lists of candidates in Tasmania, but in the ACT with fully optional preferential voting, only run 2 or 3 candidates because to run would more could put them at risk of losing votes by exhaustion. Having a minimum number of preferences requirement is one method of trying to protect the proportional nature of Hare-Clark.

    If the vote on a ticket clumps to known candidates, then some of the examples in Stephen’s paper just cannot happen. He predicts (p15) that Labor could win 9 seats from 7.83 quota while the Shooters with 0.56 would not elect anyone. This only works if Labor ran 9 candidates and all get 0.87 quotas through the operation of Robson Rotation, or if more candidates nominated, then all their preferences would flow equally with no exhaustion to the nine highest.

    All the examples seem to be built on a party nominating just enough candidates to match their quota, and Robson Rotation randomising the party’s vote so that each candidate gets about one quota. All the examples in the paper use this ‘averaging’ method to work out how many seats each party wins. I say if you want to do that in a 19 member electorate, methods like the variants of D’Hondt and Sainte-Lague are much more direct and reliable methods of working out seats. Robson Rotation can only do it if you match your candidates exactly to vote, and also if you assume there is not immensly popular candidates on any individual ticket. Well why use a Hare method based on vote for candidate if you think all the vote is so much for party it just randomises across all candidates?

  28. Robson rotation is good because it allows those care about which candidate with in a party to have a effect by not being overwhelmed by the those who don`t care and just vote for a party.

    Maybe candidate photos should be on ballot papers or are there too many issues with that?

    Allowing booth volunteers is not quite such a bad idea.

  29. It seems to me from all this discussion that the opinion of all, or at least some, is that the larger the electorate (more representatives) the more proportional the system is and thus the more fairer. However when electorates get too big, such as 19 members then you get the:
    A. “problem of a tablecloth ballot paper and it would be very difficult for voters to cast informed votes that give their preferences value.”
    B. A situation where “it can’t be considered to be ‘local’ representation.”; or there it does not “maintain… a link between members and constituents”

    I personally think the solution is quite simple. Have a one zone electorate with only the natural quota of whatever it would be, remove voter preferences (above the line voting for party and individual candidate), while maintaining candidates’ preferences.

    A. The tablecloth might still be rather big but only one box would have to be filled in rather than 19 or whatever.
    B. I would make a guess and say that not more than 1000 constituents in any current single member electorate care about having a link with their local member and make a point of communicating with him/her during the period in office. That leaves approximately 98% who only care that they are represented by a particular political party. The areas where a local member is important such as where there is a very high commonality of interests, such as a dairy farming area or a high unemployment area, the locals would then probably elect a local anyway if the voting population was at least approximately 130,000.
    C. As Antony said “When you are electing so many people at once, the number of votes wasted with candidates who aren’t elected is so small, then what’s the point of all the preferences.” Also when your options are from a range of 100 or so candidates then your choice would be so defined as to your own values that that candidate’s preferences would also be quite similar to yours.

Comments are closed.