Wednesday, March 13, 2013

Tasmanian Lower House: 25 or 35 Seats?

 Advance summary:

1.  The possibility of restoring the old 35-seat system in the Tasmanian House of Assembly is currently being discussed ahead of a motion to be moved by the Greens.

2. Looking at past election results and current polling, the 35-seat system is slightly more proportionally accurate, while the 25-seat system is slightly more prone to "over-represent" the major parties in comparison to vote share.

3. However, precise proportional representation in the Tasmanian context can easily be argued to be overrated anyway.

4. Of the elections considered (and a 2014 projection based on current polling), only in the case of 1998 did the choice of systems determine the election result.

5. Majority government is slightly more likely on average with 25 seats than with 35 seats, but in many scenarios the number of seats makes no real difference to its chances.

6. Strategic considerations favour the Greens supporting an increase in the number of seats and the Liberals opposing it, while for the Labor Party there are arguments on both sides.

7. The view that the Greens could plausibly win more seats than Labor at the next election if the 25-seat system is retained is not consistent with current polling.

8. It is not correct to blame too many problems in Tasmanian politics on the 25-seat system since politics under the 35-seat system was also very turbulent and crisis-prone during its last two decades.



--------------------------------------------------------------------------------------------------------

Admin note to readers:  My main computer is currently off being (hopefully) repaired after an accident on the weekend and therefore data-rich historical federal stuff is on hold for a while.  Not too much to see there anyway except that there were some signs of recovery for Labor in the Newspoll (52-48) presaged by the Coalition-friendly Essential and Galaxy both returning "only" 55s.  However, a reading of 55.5 to Coalition from the second Morgan Multi-Mode (based on preferences from the last election; the user-distributed result of 57.5 shouldn't be taken seriously) has pointed in the other direction.  I've updated my aggregation methods page here to detail what I am doing about MMM.

Meanwhile there's an increased danger that some of the more opinionated and less data-rich pieces lurking in the drafts section may come to fruition!  Those enviro-alarmists who have been living in fear of the release in the wild of the long-threatened Tasmanian Devil article (gee I hope you poor ninnies were watching that ABC-TV news item recently!) have increased reason to be nervous and had better hope my main computer is restored to full function fast ...

-----------------------------------------------------------------------------------------------------------------

A perennial subject of Tasmanian political debate is the comparative impact of having 25 seats in the Tasmanian House of Assembly or 35 seats.  This follows the extremely contentious 1998 reduction of the size of the House from five seven-member electorates to five five-member electorates.

By way of history, the Hare-Clark system was introduced in 1909.  Seventeen elections were held under the 30-seat system (five six-member electorates) with the following outcomes:

* Nine outright majorities.
* Four parliaments in which one party held exactly half the seats.
* Two parliaments in which no party held half the seats (one of these was a conservative-parties coalition, which collapsed.)
* Two deadlocked parliaments with 15 seats for each major party.

The two deadlocked parliaments, resulting from the 1955 election and then the 1956 election called to try to break the deadlock, were the final straw for the 30-seat system and five extra seats were added.

Eleven elections were held under the 35-seat system with the following outcomes:

* Seven outright majorities
* Four parliaments in which no party held a majority

Two of the last three elections produced under this system did not produce majorities, following the statewide rise of the Green Independents, who eventually became the Tasmanian Greens.

In the 1989-1992 parliament, Labor entered into a formal Accord with the Greens, which collapsed with the Greens then supporting the Government informally until that arrangement too collapsed and the Liberals won an outright majority at the 1992 election.  However the Liberals lost their majority in 1996.  Labor refused to take government in minority and the Liberals under Tony Rundle were left holding the baby.  With the Liberals governing with very limited Green support it was a turbulent term, and you can read more about its voting patterns (especially compared to the current parliament) here.

After lengthy debate about possible alternative models (driven both by cost-cutting following the 40% pay rise during the Groom administration, and by desires to increase the chance of majority government) the Liberals capitulated to Labor's proposed model and a 25-seat parliament was created.  There was a massive outcry from the Greens which won them very little sympathy at the ballot box, and they were reduced from four seats in the 35 seat parliament (11.4%) to just one in the 25-seat parliament (4%) although their vote fell only by one point. 

Four elections have thus far been held under the 25-seat system with the following outcomes:

* Three outright Labor majorities
* One parliament (the current one) in which no party holds a majority.

Proposals to restore the size of the House of Assembly have been on and off the drawing board throughout this term of parliament.  They have always been supported by the Greens, but the Labor government has tended to make support conditional on the state's economic performance, while the Liberals initially appeared to be supportive in principle, but later announced they would oppose an increase.

Concerns raised about the 25-seat system have included:

* The impact on proportionality of representation
* A perceived difficulty of removing incumbents
* Lack of voter choice among party candidates, especially in case of countbacks
* The lack of an effective backbench, providing limited replacement options if a minister has to resign
* Excessive accumulation of ministerial responsibilities under specific frontbenchers, sometimes leading to conflict of interest
* Increased importance and proliferation of "minders", which not only cancels out some savings of the size reduction but may also be seen as increasing the power of unelected figures
* Impact on the quality of committee work

Concerns raised about a return to the 35-seat system have included:

*Cost
*Cost
*Cost
*Cost
*Cost

Actually, that's not quite all there is to it.  Another objection sometimes heard is that the Tasmanian political talent pool is naturally shallow and that adding another ten seats will bring in a lot of new MPs who really don't belong in parliament, or save some who deserve to be thrown out.  (On Tasmanian Times this position became far more popular when Graeme Sturges got his seat back!)

For those interested, the 2010 election under the 35-seat system might have added Lisa Singh, David Llewellyn, Scott McLean and Shane Broad (Labor), Tony Mulder, Jane Howlett, Brett Whiteley and Michele McGinity (Liberal) and Helen Burnet and Adam Burling (Green).  The list includes, for instance, three then-incumbent MHAs, one candidate since elected to the Legislative Council and a former Hobart Deputy Mayor who nearly won the Lord Mayor position - so it's difficult to argue that most of these people would not have had any idea.  On the other hand, the Greens are much better off not having radicals like Burling (a very ill-advised selection as #2 candidate) in their parliamentary party room.

A few quick comments on some other points.  Firstly it probably is more difficult to get rid of party incumbents under the 25-seat system, because it encourages parties to strongly back their existing incumbents to avoid missing seats on leakage.  In Franklin in 2006, Labor ran a team of three high-profile incumbents and two relatively obscure candidates, Ross Butler and Daniel Hulme.  This choice worked as the vote for the party was concentrated in their incumbents, reducing leakage and resulting in them saving a seat (Paula Wriedt's) they may otherwise have lost.  However, the downside was seen when two of the three incumbents resigned during their term, both Butler and Hulme were elected on countbacks, and as a result Labor went into the 2010 election with a weak Franklin team and polled a poor vote there.

That said, the idea that incumbents are really hard to dismiss under the 25-seat system is an overrated one stemming mainly from the freak 2006 result, in which none of the 23 recontesting incumbents lost.  In 2010, six of Labor's 12 recontesting incumbents were shown the door, with their seats taken by three new Labor MHAs and three candidates from other parties.

25 vs 35 and proportionality

Proportionality (the closeness of fit between each party's vote and its share of seats in the parliament) is often seen as an important test of the current 25-seat system compared to the 35-seat system, and one that it is prone to failing.

In my view the proportionality argument, and the case for proportional representation in the Tasmanian lower house generally, is actually a bit overrated, because:

1. Proportional representation does not deliver proportional power. In a parliament like the present one, where one party has 40% of the seats, another also 40% and a third 20%, the party with 20% of the seats has great power because of its ability to determine who will govern, and also because while party lines are followed the numbers may as well be 33% of each.

2. Voters who are concerned about 1.) will rort around the system by deliberately voting for the major party they think could win a majority, rather than necessarily the one they prefer based on policies.

3.It's all very well to debate which system best delivers proportionality to the major parties and the Greens, but this does nothing for the chances of independents and fourth parties.  Independents might even be more likely to win than now if there were 25 single seats.

Anyway, the following is a model of how results under the two systems might have differed since the rise of the Greens in 1989.  What I've done is extrapolated the 1989-1996 elections from the actual 35-seat results to the results I think would have been most likely under the 25-seat system, and done the reverse for the the 1998-2010 results.  I've also thrown in an extrapolation of the 2014 results, under both systems, based on the completely dubious assumption that nothing changes between the February 2013 EMRS poll (see my analysis of it at EMRS - Liberals locking in support) and the election.  (Click for a larger version that is easier to read.)


A number of points about this (and I can just imagine various of these being used as arguments by both sides of the debate):

* In terms of affecting the actual outcome, the big difference is seen in 1998, immediately after the size of the House was changed.  At that election Labor won majority government, but would not have done so with the same vote share under the old 35-seat system.  This election produces the sharpest difference between the proportion of Greens elected under one system and the other.

* At most elections, whether there are 25 or 35 seats would have made no difference to the overall outcome of the election, and would not have made that much proportional difference.  (I add a caveat that there is an outside chance Robin Gray would have held majority government under the 25-seat system in 1989, as his party would have been close to winning a 13th seat.)

* On average across the seven real elections and the one hypothetical election, the 25-seat system is slightly less proportional, favouring the major parties slightly more at the expense of the Greens.  However, its results are normally not grossly disproportionate - 1998 being the obvious exception.

* Neither system returns fourth-party candidates with the exception of former federal Liberal Bruce Goodluck in 1996 (he probably would have very narrowly won under the 25-seat system too).  I add that under the 35-seat system the Tasmania First party would very nearly have won a seat in 1998.  Andrew Wilkie nearly won a seat under the 25-seat system in 2010, but paradoxically would not have gone close under the 35-seat system (because the second Green would have been well ahead of him, depriving him of his major preference source.)

* While there are some cases where the Greens appear to do better under the 25-seat system, these are rather arguable.  For instance the four seats for the Greens in the 25-seat system based on the current 2014 projection is more like 3.7 seats as Tim Morris in Lyons is not safe.

* The high error figures for "others" are a little bit misleading since "others" is a catch-all for a very large number of political forces that have scored tiny percentages of the overall vote with just one of them winning a seat.  "Others" do not necessarily preference each other.  The most significant force included in "Others" was the Tasmania First Party, which accounts for 0.7% of the overall vote in my model, but does not win any seats in either system.

Is majority government easier with 25 seats?

On average it is slightly more likely majority government will occur with 25 seats rather than 35.  However, the difference isn't massive, and in a lot of cases it will make no difference.

In most cases, the challenge of winning majority government is to win more than half of the seats in each of three electorates - three threes in the 25-seat House, three fours in a 35-seat system.  Majority government is slightly more likely in general with 25 seats because:

1. It is easier for a major party to win three seats in an electorate under the 25 seat system if the Green vote in that electorate is low.  For instance, a vote distribution of 48-39-13 is a 3-3-1 under the 35-seat system, but under the 25-seat system it could well be a 3-2-0.

2. If the Green vote is high there are cases in which the Liberals can win three seats in the 25-seat system by taking advantage of vote-splitting.  For instance suppose an electorate returns figures of 47 Liberal, 28 Labor, 25 Green.  In the 25-seat system the Liberals have 2.82 quotas, Labor 1.68, Green 1.5.  Even if the Green voters are feeling pretty pro-Labor at such an election (which isn't always the case) a lot of Green votes will just exhaust when the end of the party ticket is reached, and 3-1-1 is fairly likely.  In the 35 seat system the same figures produce 3-2-2, with the Liberals close to a fourth quota but with nowhere to get it from.

However, the amount of difference the change makes is highly sensitive to the current vote levels of the parties.  Thus the change "worked" in terms of hobbling the Greens based on their polling numbers at the time the change was made, but hasn't "worked" in other elections since. 

Other strategic issues

These are my thoughts on strategic considerations for each party, irrespective of the proportionality issue.

For the Liberals, the change is easy to oppose.  At the moment they are likely to win a majority no matter what the system, and opposing the change is an easy fit with a strategy based around cutting spending and attacking the Greens.  An election under the 35-seat system could more than double their numbers, but that could be too much of a good thing, leading to an unruly and inexperienced backbench (just ask Campbell Newman!) They would also be hopeful of evicting some of Labor's younger stars or forcing some Labor veterans to retire, and perhaps even of leaving the government with only five or six seats and great problems plotting a path back to government in 2018.  All this is much more likely in the 25 seat system.

For the Greens, the change is easy to support.  It's a longstanding historical issue that allows them to keep maintaining their rage over 1998 at very little electoral risk.  Their supporters are not generally concerned about the cost issue, and while the 25-seat system will give them better results in some cases, they would be more concerned about being confident of keeping their representation at a reasonable level.  Being wiped down to 2/25 seats at the next election is still possible but under the 35 seat system they would have little trouble holding at least four seats.

For Labor, there are arguments cutting both ways and it is no wonder their position has been tentative and conditional so far.  The party wants to be seen as promoting spending restraint, but at the same time it wants to retain a competitive number of MHAs and it wants to avoid losses that will impact on its future in 2014.  On present polling the party is not assured of holding two seats in any electorate, and would be likely to lose most of its second seats if an election was held now.  This problem is seen as especially acute in Lyons, where the party could lose either its veteran Speaker Michael Polley or new recruit Bec White and Bass, where the party would very likely lose either Michelle O'Byrne or Brian Wightman.  In Franklin, the loss of either Premier Giddings or David O'Byrne would be disastrous, but in my view Franklin is the seat in which Labor is most likely to retain two anyway.  (Incidentally a scurrilous Twitter rumour suggesting David O'Byrne would negate this risk by switching to Denison was vigorously denied by its target.)

"Brutally simple"?

An article by Barry Prismall in today's Examiner argues that this is a complete no-brainer for Labor, which must agree with the Greens' push to avoid disaster.  Prismall argues:

"Based on recent polling, which is becoming remarkably consistent in pointing to a Labor wipeout, the Labor caucus would lose about five seats and be reduced to a rump next March.  The Greens could win even more seats and become the official opposition."

There is actually nothing in recent published polling that suggests this as remotely likely.  Every EMRS poll since the last election has shown Labor with a greater vote than the Greens on the figure EMRS now uses as a headline, save for one which showed them equal.  The one poll which showed them equal, in May 2012, had the highest Greens vote at any time since Feb 2011, and their last two results have been much lower.  Furthermore there is that very strong evidence from previous elections that the EMRS headline vote tends to overestimate the Green vote and underestimate Labor's.  Let's not forget that in 2010 the last two EMRS polls both underpredicted the margin between Labor and the Greens by eleven points.  It's not likely to be that much again next time, but even assuming it is only, say, half that, the idea of the Greens winning a vote share very close to Labor's is not looking likely.

While it is possible on current polling that Labor would win only five seats, there is no sign in the last two of the Greens winning more than four.  At no stage since the blowout to the Libs started in late 2010-early 2011 have the Greens looked like they even might win six seats (holding five would be remarkable enough), and nor at any stage has it been credible to project that Labor will fall well below 16.7% in any seat and suffer the ultimate indignity of recording a zero-seat result anywhere.

The most that can be said is that by loading every assumption in town against Labor and in favour of the Greens, from the current situation there is a very remote chance of 15-5-5, which would result in Labor and the Greens having co-opposition status.  But that's not new if so; the Liberals and the Greens held co-opposition status after the 2002 election, when they were called the "Liberal Opposition" and the "Greens Opposition".

There is nothing in current polling to suggest that an outcome in which Labor wins fewer seats than the Greens is even possible. Of course, polling can change, but I think such an outcome would be correctly placed at very long odds indeed.

Prismall also states:

"The difference for Labor is, that with a restored 35-seat House, the current caucus would all keep their seats."

Not necessarily.  The threshhold for two seats in an electorate in a 35-seat House is 25%, although that may be more than needed or not quite enough depending on leakage and the results of other parties.  With the Labor vote around the mid-20s even after adjusting for EMRS house effects, it is no gimme that Labor would get there in every electorate.  Indeed on my own current model, they would be likely to still lose one in Bass, in which the Liberals could conceivably get five (as they did in Braddon in 1992 and Labor did in Braddon in 1972.) 

Prismall also writes:

"It doesn't augur well for new Labor candidates, but bugger them.  Survival is survival."

Actually the chances for new Labor candidates under the 35-seat system would be better than under the 25-seat one.  Under the 35-seat system it is likely that vulnerable incumbents like Graeme Sturges in Denison would be bumped off by new talent in their own parties, rather than just losing to Liberals.  So that's one argument against the change that I'd ignore.  The 35-seat system is Labor's best prospect for more new blood. 

The Big Question

Never mind the cost.  As Prismall points out, counter-savings can be made if the size of Parliament is again increased, and we should be willing to pay for quality government instead of expecting that cutting corners will really save us money.  Nor should this issue ideally be decided based on the pragmatic concerns of particular politicians, although of course it will be, as it has been in the past.  Rather, the crucial question for me is this:

To what extent is the "current mediocrity and crisis in state politics" really driven by the small size of the House - and to what extent is it just because state politics - and Tasmanian state politics especially - has long been like that?

After all, if you want "crisis", how about a government ripping itself apart then rolling its own Premier and ending its days in minority over a proposal to build a hydro-electric dam?  How about a majority government damaging the state's finances without this being known until it was evicted from office by the voters, at which point one of its media allies attempted to bribe a reversal of that result?  How about two governments that spent months at a time on tenterhooks as to whether they would even survive, with one having one of its own ministers sacked by the Greens and another getting voted down so often on the floor of the House that the Opposition was running the state almost as often as it was?  And that's all just from 1980 to 1998. Then look at the awfulness of some of the recent Labor state governments on the Australian mainland, that had long years in clear majority in large parliaments and still ended up being deservedly kicked out in 60-40ish results. 

In my view, there are ways (basically those itemised above, except where I've debunked them) in which the smaller house is not helping things and that do provide evidence in favour of a change back to 35 seats.  But those imagining that the change will be a magic cure for even a fraction of the things that ail Tasmanian politics - and that often get dragged into the seat-numbers issue - are dreaming.

==============================================
2014 Update: added May 2014

The 2014 state election saw the Liberals win 15 seats, Labor 7 and the Greens 3 off 51.1%, 27.3% and 13.8% of the vote respectively. While the Labor and Green shares were very close to proportional, the Liberals came in with 60% of the seats off a vote nine points lower than that.  The Palmer United Party, with 5% of the vote, could have expected at least one seat on a statewide proportional basis but got nothing.

The Liberals were very lucky in winning 15 seats, because they managed to win a slate of four in Braddon off below 60% of the vote. They did this as a result of a lot of the vote going to minor parties that did not get elected, and as a result of a lucky split between their third and fourth candidates that excluded Labor's Brenton Best.

Under the 35-seat system results based on the vote shares recorded would have been:

Bass: 4 Liberal 2 Labor 1 Green
Braddon: 5 Liberal 2 Labor or 4 Liberal 2 Labor 1 PUP
Denison: 3 Liberal 3 Labor 1 Green
Franklin: 4 Liberal 2 Labor 1 Green
Lyons: 4 Liberal 2 Labor 1 Green

In Denison in this simulation Labor poll 2.7 quotas to the Greens' 1.7 but the even vote split between the minor Labor candidates would mean the second Green would have no chance.  In Braddon, the increased Liberal quota would mean that PUP outlasted the Greens in the cutup but it is very unclear whether they would win even so.

Total: 19-20 (54-57%) Liberal, 11 (31%) Labor, 4 (11%) Green, 0-1 (0-3%) PUP

Thus the 35-seat system would have seen the Liberals closer to a proportional result, Labor with a favourable one for their vote share, and the Greens do no better proportionally at the election.

4 comments:

  1. As you say, increasing house numbers won't fix the general problems with the lower house, except possibly the staffing shortage re ministers.

    For me the discussion re numbers comes back to the capacity of the house to provide representation. Hare Clarke in this is superb; having a member who approximates your own beliefs in your electorate regardless of election outcome is great. And if you support one of the major parties, the choice of members.

    35 arranged as 5 areas with 7 members doesn't change that much. 7 areas with 5 members however may well change the capacity to provide representation to the less populated but still distinct areas. Or given our population distribution, it might not.

    There is probably more to consider in the upper house in order to further distinguish it from the lower house.

    ReplyDelete
  2. I ended up in an extended discussion at the Tallyroom regarding the point of an upper house and how it is best structured to achieve that goal.

    Is this something you have thought about / developed a position on?

    ReplyDelete
  3. A few years ago I would have just said abolish it. The government in the Lower House is accountable to the people for its mistakes and the fear of loss of office should act as a deterrent against really stupid politics; furthermore if a state government really goes too far there are many situations in which the federal government or the courts can intervene.

    I'm less sure now because of what we have seen in the past few years: state governments that are "walking dead" years in advance of the election, so that the fear of eviction from office no longer serves as any deterrent against terrible policy. Governments dependent on the support of minor parties for their very survival can also have imperatives to support bad policy immediately even at future electoral cost. Also I think there's a degree of turn-taking mentality among the major parties, such that they're willing to appease their bases even at risk of defeat because if they get booted then their turn will come again. So all up I just don't believe the fear of loss of office is working as a deterrent to bad policy.

    I therefore generally like the idea that the function of an Upper House is to make bills run the gauntlet of two different electoral systems - as seen in interstate systems where the lower house is single-seat and the upper house is some version or other of PR. This should make it more difficult for bad legislation to get through. It does concern me though that this power can be abused to obstruct the repeal of existing bad legislation, so I wonder if there should be some difference in the treatment of repeal (allow delay power only) as opposed to new legislation (allow outright blocking) and if so how this would be codified.

    However upper houses able to block legislation must be accountable to the electorate for that decision. And this is where the problems in Tasmania start - the Upper House is not effectively accountable because of long terms, rotating elections, lack of dissolution provision and low public awareness.

    Generally I don't see much point in trying to dream up systems for the Tasmanian Upper House because the core problem is that only the Legislative Council can reform the Legislative Council, so trying to devise an ideal system for them that they will not accept is just pragmatically a waste of time. The Legislative Council has to develop an interest in reforming itself, and because it would be very difficult for many of the incumbents to survive under any other system (at least without joining a party) I don't know if that's going to happen any time soon. When it happens then I will devote more energy to the idea of what sort of system would best accomplish it.

    ReplyDelete
  4. Some good thoughts in there.

    An upper house is inherently a house of restraint. The value of an upper house is fully contained in its difference to the lower house. I've also come to the view that the lower house is the house of democratic representation, and fulfils this function completely; i.e democratic representation is not a primary requirement of the upper house. Democratic representation as per the lower house remains a consideration, but only in the sense that it is the point of zero value.

    I don't mind the rolling nature of Upper House elections. It provides gradual opportunity for the actions of the lower house to be eased or restrained further throughout the electoral cycle, mitigating some of the effects of the 'walking dead'.

    As you say, the primary constraint of upper house reform is that it has to be reform the upper house wishes to conduct. This is not a problem as such, just a limiting condition. It's only a problem if people want something outside of that limiting bound.

    Generally, when people think about Upper House reform, it is on the basis of removing restraint out of a frustration that the Upper House has not moved with the Lower House. It can't be approached that way; that moment of difference is the specific and intended contribution of the Upper House.

    The only framework you can approach the Upper House with is defining the nature of that difference - the bias that it has. This structural bias is intended to maintain certain characteristics of the state against the various fluctuations of governance. So the question to be asked is what characteristics do we wish to either retain or develop in the long term - beyond the cycle of Labor / Liberal switching. Only when you have the answer to that question, you can approach reform in the upper house.

    ReplyDelete