Over the past week a few psephological clues have come to light that assessed together could give us an important indication to the outcome of the next federal election.
Collectively, the clues, which we will list and assess below point to the possibility that Labor may need 51.5% of the national TPP vote to have a better than even chance of winning a House of Representatives majority. Conversely this figure for the Coalition may be 49.2% of national TPP vote.
While such a vote is not out of the realms of possibility, if Labor needs to win a TPP of 51.5% to win a majority, this makes their task of coming from behind more difficult. It may also foretell an election, where Labor wins a national TPP swing to it but still loses government.
We shall go through each clue below and analyse the size of the TPP hurdle they collectively create for Labor.
Last week, an article appeared on Fairfax news websites which indicated that Labor internal polling was showing a poor position for the party in New South Wales, with seat losses concentrated in Western Sydney. This article by Phillip Coorey can be found here. Peter Van Onselen also kicked along the topic in his Sunday Telegraph column, last weekend.
The Coorey article asserted that Labor internal polling was showing the following NSW Labor held seats were in play,
“The seats that could fall include McMahon, held by Immigration Minister Chris Bowen, and Watson, held by Environment Minister Tony Burke.
Other seats include Parramatta, Barton, Reid, Werriwa, Fowler, Banks and Greenway, which, one source said, would be ”gone by 10 past six on election night”.
There are also concerns for the central coast seats of Dobell and Robertson, while Blaxland and Chifley, both in western Sydney, would be close calls.”
This information has been sorted into a table of Labor held seats in NSW (ex- Sydney and Grayndler) ordered by margin, which is set out below.
Based on the above article, 9 Labor seats are believed to be falling. The 9 seats speculated to be falling to the Liberals are all in Western Sydney and a further 2 seats are of “concern” on the Central Coast while a further 2 Western Sydney seats would be retained by Labor with close margins.
What is more stark is the possibility that most seats with TPP margins of up to 9% may be lost when the state-wide TPP swing to the Coalition is currently 3.2%. Is a swing in Western Sydney of 9 % possible with a much lower state-wide swing?
If we take a very broad view of Western Sydney to include 16 seats including Barton, Banks, Parramatta, Barton, Reid, Werriwa, Fowler, Banks, Greenway, Macquarie, Macarthur , Hughes, Lindsay, McMahon, Watson and Mitchell, a Western Sydney TPP swing of 9% would account for the entirety of the current swing in NSW and leave all other seats with no swing to either major party.
Of course the flip side for Labor is that NSW marginal seats outside of Western Sydney will probably remain Labor held with no swing. While the article does mention that “concerns” are held for the Central Coast seats of Robertson and Dobell, this is no surprise. Robertson has a thin margin of 1%, which could fall if there is no overall swing in NSW outside of Western Sydney, while Dobell is exposed to the Craig Thomson scandal and has a stimulus to stray far from the mean.
Consequently, seats in the rest of the state are unlikely to change hands in such a scenario, Eden-Monaro, Page and Richmond are not identified as Labor losses in the Coorey article. There is also no mention of disastrous swings away from Labor in the Hunter or Illawarra. An intriguing omission from the Coorey story is the seat of Lindsay which is not indicated as a seat of concern even though it is a Labor held Western Sydney seat with a very small margin of 1.1%
If this internal polling is an accurate reflection of the intra-state picture in NSW then this is a negative result for Labor and not just for the obvious seat losses. A swing to the Coalition in NSW concentrated in Western Sydney is optimal for the conservative party as such a swing is concentrated in the Labor seats with TPP margins of under 10%. Of the 16 NSW seats held by Labor with TPP margins of 10% or less, 10 are in Western Sydney. The Coalition does not waste the swing in any of its own seats or on seats with higher Labor margins, in the Hunter and Illawarra. Consequently if the Western Sydney “black hole” persists until election day this will allow the Coalition to win more seats with less votes increasing its chances to win a parliamentary majority with minority TPP vote.
Another quarterly ERMS poll of state voting intentions was released last week by the Tasmanian market research company. The poll returned results of Liberal 55%, Labor 27%, Greens 13%, Others 3%. Such a result would translate into a solid Liberal majority under Tasmania’s Hare-Clark voting system.
However are there any federal implications in these results? Possibly, we cannot read too much into one state poll for federal results. However his poll is one of a string of ERMS polls since early 2011 that have placed the Liberals in a position to win a majority of seats under Tasmania’s Hare-Clark voting system and thus win government in their own right. Consequently Labor is on the nose at a state level and this can only hurt Federal Labor’s vote in Tasmania.
What is the size of a swing to the Coalition in Tasmania? We can only speculate with some fragmentary evidence. We previously took a look at the federal polling position in Tasmania when Roy Morgan released its quarterly aggregates of its face to face polls in late August. We estimated based on these aggregates and the sample size that the swing to the Coalition was between 0.9% to 13.1% higher in Tasmania than the national TPP swing to the Coalition. Given the current national TPP swing to the Coalition of 2.9%, this could indicate a swing to the Coalition of anywhere between 3.8% and 16%. This is a far too large band of uncertainty and thus we haven’t made any adjustment to the seat projection.
Another tidbit came from Peter Brent at Mumble, a few months back in a post on Tasmania that Newspoll’s unpublished Tasmanian sub-samples were indicating the Tasmanian TPP was the same as national TPP. Given that the Tasmanian Labor TPP was running 10.5% ahead of the national figures, this would indicate a swing to the Liberals of 10.5% if there was no change nationally.
If the TPP swing to the Liberals in Tasmania sits at 10% even if the rest of the nation repeats a 50-50 result, such a swing would again represent an efficient seat winning swing for the Coalition, with its largest swings in Labor seats with margins small enough to allow the seats to change hands. Such seats are in plentiful supply in Tasmania (Bass 6.8%, Braddon 7.5%, Franklin 10.8%, Lyons 12.3%). If such a large swing to the Coalition eventuates in Tasmania, like Western Sydney, it will allow the Coalition to win more seats with less votes increasing its chances to win a parliamentary majority with a TPP under 50%.
At the next election 30 MPs elected to their seats for the first time in 2010 will have the opportunity to recontest their seats. Such incumbents have been referred to as sophomores and they get a increase in their vote as they earn a personal following. This increase is small but can make a difference in marginal seats. If all sophomores recontest their seats in 2013, 11 will defend marginal seats for the LNP while 4 will recontest marginal seats for Labor. Therefore the sophomore effect will advantage the Coalition overall.
The bulk of the LNP marginal seat sophomores will be in Queensland, with 6 MP’s recontesting marginal seats in Brisbane, Longman, Forde, Dawson, Flynn and Herbert . As the swing in Queensland is relatively favouring Labor, this is the state where Labor will be looking to pick up seats, however they will have to overcome the personal vote that each of these sophomores will attract in 2013. There are also Coalition sophomores in other states and territories that recontest marginal seats, Bennelong and Macarthur in NSW, Aston in Victoria, Hasluck in WA and Solomon in NT.
Labor has its own sophomores are present in 4 marginal seats, La Trobe in Victoria, Greenway & Robertson in NSW and Bass in Tasmania.
Quantifying the shift in the vote needed to win
Now that we have identified the factors that could raise the national TPP vote needed by Labor to win a majority and conversely lower the vote required by the Coalition to do the same task, we need the quantify this shift.
We will do this by using the poll average and seat projection mechanism to provide a range of seat probabilities under different assumptions using the current poll average and seat projection data as a baseline (i.e the current state swings). We will add to the baseline the following shifts to represent the factors set out above:
William Bowe at Poll Bludger for his “Bludgertrack” has quantified the sophomore effect as a TPP shift in favour of the incumbent of 1% for metropolitan seats and 2% for provincial and rural seats. We will adopt this benchmark for today’s exercise.
Based on the above information from Roy Morgan, Peter Brent’s insight and the general state polling, we will model a TPP swing to the Liberals representing the national swing plus 10%.
Given that the internal polling suggests a TPP swing to the Coalition of 9% in Western Sydney whilst the state-wide swing is only 3%, we will model a state-wide swing where Western Sydney seats exhibit a state wide swing with a 6% shift to the Coalition. To maintain the state-wide swing all other seats in NSW will have a swing that reflects the state-wide swing with a 3% shift to the Labor.
As noted above Western Sydney seats are defined in this exercise as the following seats: Barton, Banks, Parramatta, Barton, Reid, Werriwa, Fowler, Banks, Greenway, Macquarie, Macarthur, Hughes, Lindsay, McMahon, Watson and Mitchell.
Once the above shift adjustments are made to the seat projection, the probabilities of all seats are obtained for a spread of Labor national TPP votes between 48 – 52% by applying the required uniform national swing after the above shift adjustments are made.
The seat probabilities are then used in a Monte Carlo simulation to create a distribution of probabilities of the final outcome (a majority for either major party, whether a major party wins 75 seats, or a true hung parliament (74-74-2) situation, based on 20,000 simulations.
If this sounds like a lot to take, help is at hand as the above outcome probabilities can be plotted on an area graph which should make intuitive. Below is a representation of outcome probabilities plotted across possible ALP National TPP votes of between 48 & 52% only taking into account only the sophomore effect.
The parts of the graph that are shaded in solid red and blue represent the probabilities of Labor or the Coalition winning a majority of seats. As you can see where the ALP TPP is at 48% on the left of the graph, all of the graph is solid blue, the Coalition has a 99.2% chance of winning 76 seats or more at a 48% ALP TPP.
On the right of the graph, where the ALP TPP is 52%, the graph is mostly red, Labor has 99.9% chance of winning 76 seats or more.
The “Colgate Triple Stripe” section the chart represents the three other possible outcomes. The light red stripe indicates the probability that Labor wins 75 seats (75-73-2). The light blue stripe indicates the probability that the Coalition wins 75 seats (75-73-2). The light gray stripe indicates the probability of a 74-74-2 outcome which would be a true hung parliament, where there are only 2 crossbenchers.
As you can see in the above graph the transition of a probable Coalition majority to a probable Labor majority takes place at approximately a 50-50 TPP. Now compare this graph which only uses the sophomore effect to shift the seat projection, with a graph that takes into account large relative swings in Tasmania and Western Sydney in addition to the sophomore effect. The second graph is below.
This is a far more favourable field of play for the Coalition than the former chart. The solid blue of the Coalition majority takes up most of the chart. The shift from a probable Coalition majority to a probable Labor majority takes place at between 51-51.5 ALP TPP.
The “Colgate Triple Stripe” section of the chart which indicates the probability of various hung parliament combinations is thinner, indicating a smaller probability of no party winning 76 seats or more.
For those of you that want numbers in a table below sets out the TPP votes where a 50% probability is reached on whether a major party will win 75 or 76 seats or more. Additional columns for a 95% probability for a major party winning 76 seats or more. The table is designed to show a continuum of outcomes (Coalition 76+, Coalition 75+, Labor 75+, Labor 76+) for each scenario.
In short if a major party outperforms the national TPP in the above table, they will have a higher probability than the threshold in the chart of obtaining that outcome.
In a scenario where only the sophomore effect is taken into account, Labor would need at least 50.3% TPP to have a greater than 50% probability of winning 76 seats or more.
Once the large relative swings in Western Sydney and Tasmania are taken into account this figure is pushed up to 51.5% for Labor to have a greater than 50% probability of winning 76 seats or more.
Conversely, the Coalition benefits from the above shift. Without the relative swings in Western Sydney and Tasmania, (but with Sophomore Effects) the Coalition would need 50.6% of the TPP to have a greater than 50% chance of winning a majority. With the relative swings in Western Sydney and Tasmania included, this figure falls to 49.2%.
It is important to keep in mind that even if a party has a greater than 50% probability that they will win a majority of seats, this is not a certain event. This is why the 95% probability columns are included the above table to give an idea of the zone of uncertainty where a hung parliament or an opposing party majority is still a possibility.
Another interesting facet about the above figures is the small effect of the Sophomores pushing the TPP figures towards to Coalition by 0.1-0.2% across all scenarios. The effect is dwarfed by the combined effects of relative swings in Western Sydney and Tasmania which shift the TPP figures towards the Coalition by 1.2-1.6% across all scenarios.
One last point about the above analysis is that it is predicated on only two crossbenchers being returned at the 2013 election; Bob Katter in Kennedy and Andrew Wilkie in Denison. This assumption is driven by our current seat projection treatment of crossbenchers. A larger or small crossbench will increase or shrink to probabilities of a hung parliament respectively. It will also increase or reduce the TPP required by a major party to reach the seat outcomes defined above.
For example if the Liberals preference the Greens ahead of Labor in Melbourne, this will send the seat to the Greens away from Labor, and the latter will need to win a another seat from elsewhere the make up the loss. This would push the TPP vote needed for a better than even chance of winning a majority up to 51.6-51.7%.
If Labor need 51.5% to win a majority of seats at the next election, this is a harder task for Labor but not impossible for a party coming from behind in opinion polls. Labor won the 1993 election with 51.4% of the vote after trailing badly in the polls for the most of the preceding term of office.
The above scenario is predicated on possible regional swings which we have no public polling evidence. With little transparent or substantive evidence they may be a mirage. But if such swings are true, given the normal course of polling until the election campaign where mainland state samples are the only sub-national polling we can expect to observe, the above regional swings may not become apparent in public polling until the election campaign itself.
Consequently this post should not be seen as a prediction but a warning of what may come to pass if these swings in Tasmania and Western Sydney eventuate at the 2013 election. The swings in Tasmania and Western Sydney may not occur, but we should not be surprised at the outcome of the election if they do.