Rudd's double dissolution bluff Sunday, November 29, 2009

A double dissolution election, where twice the number of Senate seats are up for grabs, would be more trouble than it is worth to the Labor party.

While control of the Senate is a glittering prize for parties of all persuasions there are also several consolation prizes when it comes to the Upper House. In 2004 the Coalition won a fourth seat in Queensland which, combined with solid results everywhere else and in 2001, tipped them into a majority. Labor strategists are hoping that the next election will smash the Coalition's stranglehold on the Senate and either give Labor a majority of its own (unlikely), or at least reduce the power of the Coalition so that no single party has control. With Labor high in the polls and the Liberal's vote sinking lower and lower, the temptation is there for Labor to pull the double dissolution trigger and hope it translates into a radically improved situation for the government.

Unfortunately for Rudd a double dissolution would not improve Labor's practical position above that of a regular election. On the contrary, a double dissolution holds significant drawbacks for the government.

Running a wide ranging simulation based on the swings recorded in the latest Nielsen poll as applied to the 2007 vote in the Senate, the results under a double dissolution show that Labor would go sideways, the Coalition would go backwards and Greens would emerge as the big winner, leaping from five seats to around eleven, an unprecedented number for a third party. Independent Nick Xenophon could capture two seats in South Australia, but lose the balance of power, surely a negative outcome for the feisty independent.

The practical result of all this is that Coalition would lose its ability to block legislation and the Greens would gain the balance of power on the cross-benches.

However, under a regular half-senate election, which will occur naturally if the government serves out its full term, while the number of seats won by the parties varies in the simulation, the practical results are the same:- the Coalition loses its majority and the Greens gain control of the cross-bench.

To repeat, the practical result from the next Senate election is the same whether it is a half-Senate or full-Senate election.

So with nothing to gain from a double dissolution except breaking the Liberal's hold on the Senate a few months early, the risks of such a course of action are very high. A double dissolution would possibly split the upper house and lower house elections, meaning Australians would have to go to the polls effectively twice as often until they were re-synchronised, surely an unpalatable situation for any major party. A full-Senate election would also crack open the door for micro-parties of a right-wing nature to fluke one or two seats. The prime candidates for an upset being a Pauline Hanson-type candidate in Queensland and a Family First candidate in a state like Victoria or South Australia.

For the hassle, cost and future pains caused by a double dissolution election, the gains for Rudd and the Labor government are simply not worth it. Labor's double dissolution bluff to tame the Senate is just that, a bluff.

What is the most likely outcome at the next election? Sunday, November 8, 2009

I thought I would start proceedings by asking the most basic and frankly most boring Senate question possible: What is the most likely outcome at the next election?

Answer: The Australian Labor Party is in the box seat to secure half of the senate seats available at the next election. The Greens are set to gain the sole balance of power between the two majors, and the Coalition will lose its ability to block legislation in the senate.

Ironically for such a straight-forward question, working this out is fairly complicated. We have to massage data from the last election and the latest polls (we apply the swings from the polls to the 2007 senate vote), and then run simulations across all six states. Six senate seats are up for grabs in each state and two in each of the territories. With an average of 16 parties contesting each state you can see that simulating a nation-wide senate election is a big task (not many people do it!).

To assist with my idle speculation, I use my scenario designer cassandra to design a scenario, then I use a program called apollo to run 20,000 simulations with random preferences based on that scenario. So for example if the Australian Labor Party (ALP) wins the 6th seat in Tasmania in 10,000 simulations, I say they have a 50% chance of winning that seat. Apollo is a very powerful program, allowing me to run simulations around any combination of votes and/or preferences.

New South Wales

If we look at the 2007 election, the ALP had almost enough quota to win the fifth seat outright and only a smattering of preferences were required to push them over the line (fig 1-A). However, this left the Greens and the other minors without an ALP remainder for the 6th seat, and apollo dutifully notes the Coalition had a 67% chance of winning that final seat compared to the Greens on 24% (fig 1-B). The most likely result being a 3/3 split, which is what happened (fig 1-C).

Now however, the slight increase in the Green and ALP vote since the election has had a dramatic change on the outcome of the sixth seat. The ALP now has a complete lock on the fifth seat (Fig 2-A), winning in every simulation, and its small spill over into the sixth seat opens the door for all the

minor parties. In particular this helps the Greens, who go from a 24% chance of winning the seat in 2007 to a 67% chance of winning the seat (Fig 2-B). The Coalition switches positions from 2007, and will be extremely lucky to win the final seat in 2010. Having won the fifth seat in every simulation, the ALP has nothing left in the tank for the sixth seat. Of the others, the CDP is the most likely, with a very small 0.55% chance of winning the seat.

Victoria

There is a big swing to the ALP in Victoria. A lot of that seems to be coming from an increase in the "others" vote, with a 3.1% increase in the Greens polling. On a 2PP Labor is looking at a 5.1% increase.

Running apollo over the 2007 results, two things are interesting. That after the first four seats were distributed, the ALP, Greens and Coalition were all on a high quota (> 0.7) for the fifth and sixth seats. The ALP on 0.9 of a quota saw it winning the 5th seat in 72% of apollo simulations (Fig 2007-A). However, in reality, preferences flowed to the Coalition and despite apollo (with its random preference allocation) only rating them a 17% chance of nabbing the 5th seat, they edged ahead at the ballot box (Fig 2007-B). This meant the ALP had a huge surplus for the sixth seat and overwhelmed the Greens. People often say the Greens were unlucky to miss the 6th seat in 2007, but a "more true" statement would be that Labor were unlucky to miss the 5th seat, which pushed them down to "only" winning the 6th seat (Fig 2007-C).

The second item of interest was how contested the 6th looked to be, with Greens, Coalition and Labor all in with a good chance of winning it.

Now, with the higher ALP and Greens vote, a similar thing happens as in NSW, there is less scattering of the scenarios and the results appear stark. The ALP has a complete lock on the fifth seat now (up from 72% in 2007), winning in all 20,000 simulations (Fig 2010-A). In the sixth seat the Greens are a near certainty, up from a 34% chance in 2007 to a 97% chance in 2010. The Coalition is almost locked out, with apollo suggesting only a 3% chance of the last seat (Fig 2010-B).

Queensland

Queensland proposes a unique challenge for analysis. Generally, treating preference tickets as a random act of nature provides a very close guide to the likely outcomes. However, in the Sunshine State there was a conscious decision by a large number of parties to put one high scoring ticket last: Pauline Hanson's United Australia Party. Hanson received a phenomenal 4.19% of the vote. This is normally enough to put a candidate in serious contention for a seat. However, with all the preferences going against her, it was far short of actually winning the 6th seat. What will apollo say after 20,000 simulations using random preference allocations?

Apparently Hanson never really had a chance, even if the preference flows had not been directed against her ticket. The PHUAP wins the final seat in only 2% of simulations, with another three cornered contest between the ALP, Greens and Coalition, decided mainly on who gets the fifth seat (Fig 2007-B).

In 2010, we have a small wrinkle which requires some speculation. A 6% collapse in the coalition primary vote is not reflected in the 2PP vote, meaning that most of the votes are returning to the Liberals via minor parties other than the Greens. We'll let apollo inflate the minor parties votes in proportion to the vote they received at the last election.

The result is a most common outcome of 3 seats the ALP, 2 seats to the Coalition and 1 seat to the Greens (Fig 2010-C).

I'm sensing a theme here. A common boring theme. Most states are going to split 3 seats Labor, 2 seats Coalition and 1 seat Greens.

However it is a very close thing in QLD, with the Greens having only a 50% chance of getting the final seat with a tight battle with the Coalition (Fig 2010-B). It would be a rare event indeed for the Coalition to not get 3 seats in Queensland but there's a 50% chance of it happening in 2010!

So far this is the closest contest predicted by cassandra and apollo in 2010 and so one to watch.

Western Australia

Because of the high Liberal vote in WA, the results are almost a mirror of the regular pie chart, with the Liberals gaining 3 seats, ALP 2 and Greens 1. The 2007 Senate election was not much of a contest in WA, with the Greens the solid favourite ahead of Labor for the last seat (Fig 2007-B).
In 2010 the plummeting Liberal vote means we might expect to see the standard East coast result of 3 ALP, 2 Liberal, 1 Green repeated.

However, while after 20,000 simulations from apollo we see that the Liberals lose their absolute lock on their third seat (the 5th seat), going from winning in all circumstances to winning in only 82% of chances (Fig 2010-A), they should still comfortably get 3 seats.

Interestingly, the Greens win the 5th seat in 14% of scenarios compared to the ALP's 5%, reflecting that at this point in the count the Greens are on a higher vote than the remaining ALP vote after the latter's vote has been expended on winning the 2nd and 4th seats (Fig 2010-A). In the event of the Libs gaining the 5th seat, the Greens are outright favourites for the 6th seat, but with the proviso they face a strong challenge from Labor.

South Australia

South Australia saw the popular Independent politician Nick Xenophon make a run for the Senate. His spectacular run (all the more impressive as it was from the brandless end-of-paper group S) is reflected by his 100% lock on seat 5 in all 20,000 apollo simulations of the SA Senate race (Fig 2007-A). That made the sixth seat an almost even split between Liberals, Greens and Labor on 32%, 32% and 36% respectively (Fig 2007-B). At this finely balanced point the real preference flows come into play, and in the actual 2007 election, they favoured the Greens, giving them seat 6.

At the moment, polls are showing a substantial collapse in the Liberal primary vote in South Australia, possibly the largest in all the states. And even though Xenophon has been elected, the "other" vote remains high. Trying to discern trends in South Australia at the moment is a bit like trying to predict the next rate rise from a particularly muddy cup of tea leaves. Of possible note is that Xenophon's 2007 vote was about 6% lower than his South Australian Legislative Council vote in 2006, but what does that mean for a 2010 federal election where Xenophon is not even a candidate? We do not know at this stage if Xenophon will run like-minded candidates on a "Friends of Xenophon" ticket. The March 2010 South Australian election may illuminate us as to whether he as the organisational support to stand people under his banner or if Xenophon only runs candidates when he himself is on the ticket.

So, plenty of idle speculation is required for South Australia 2010.

Let's assume he runs a "Friend of Xenophon" candidate and campaigns heavily for her (cough)Di Bell(cough) but that without his head being on the chopping block there is a softening of the vote, from 14% down to 10%.

Team Xenophon lose their lock on seat 5, with it becoming a likely ALP seat (Fig 2010-A). However, the potentially high Xenophon vote squeezes out the Greens in the sixth seat, with Team Xenophon winning in 51% of scenarios (Fig 2010-B). It might be worth running a few different scenarios on SA without a Xenophon candidate or when more information comes to hand.

Tasmania

Last but not least, we come to the Apple Isle. Tasmania in 2007 was interesting because a third party (in this case the Greens) polled above a quota. This gives the Greens a lock on the fifth seat (Fig 2007-A), where they win the seat in all 20,000 simulations, leaving the Liberals and Labor to fight over the last seat, with an easy win for Labor (Fig 2007-B). Interestingly, despite still having a large excess vote (over 4%) the Greens are never in contention for a second seat in any simulation.

On to 2010 and no polls are available for Tasmania, so our normal avenues of updating the 2007 figures are not open to us. Since Bob Brown is not up for election, let's assume a slight softening in the Green vote to just on a quota. I have no reason for this assumption, just IDLE SPECULATION. Alternatively, the Green vote may hold up, and that would be a worthwhile scenario in the future to test.

The only difference is that the Liberals have even less chance of winning the final seat. Dropping from a 30% chance if preferences are distributed randomly to a 13% chance (Fig 2010-B).

Tasmania seems fairly solidly 3 Labor, 2 Liberal, 1 Green.

Conclusions

So what are the interesting contests in 2010? If I had to pick three, it would be what Xenophon does in South Australia, the sixth seat in Queensland and the sixth seat contest in New South Wales.

Based on the current polls and a bit of speculation, the ALP can expect to win 20 seats, taking their total to 36 (+4 over the current senate), the Coalition can expect to win 16 seats for a total of 32 (-5), the Greens can expect to win 3 seats for a total of 6 (+1), and Friends of Xenophon, if they run can expect to win 1 seat, taking their total to 2.