19 October 2007
As everybody except the editorial writers at The Australian knows, Labor needs to win 16 seats to win the
election. (Memo to The Australian: Labor won 60 seats in 2004. A majority of 150 seats is 76 seats.
Seventy-six minus 60 is 16. It’s quite easy, really.)
According to the Mackerras pendulum, Labor needs a uniform national swing of 4.8% to win 16 seats. Of course,
swings are never uniform – at any given election, some states swing more than others, and within each state some
seats swing more than others. But the higher and lower swings tend to cancel each other out, so that if a party
needs a 5% swing to win an election, it will usually win if it gets that swing.
There are exceptions to this rule. Kim Beazley needed a uniform swing of 2.5% to win in 1998. He got a swing of
4.6%, but failed to win enough seats. This happened partly because the swing was wasted in big swings in seats
Labor already held, but mainly because of the distortion of the usual voting pattern by the
phenomenon. There were large apparent swings to Labor in a swag of rural seats where One Nation stole a chunk of
the Coalition’s base vote and gave it to Labor as preferences. Thus there were swings of over 10% in
and Fairfax, and above average swings in many similar seats – none of which Labor won – while the key suburban
La Trobe and
Deakin stuck with the Liberals.
Usually, however, when there is a big swing on, the suburban seats go down like dominoes, regardless of how hard
the local member has campaigned or how mediocre the opposition candidate is. In country seats, local and personal
factors count for more, and there are more seats which defy the trend – some, like
Riverina in 1969, producing
much bigger swings than the average, others less. But even country seats usually go with the flow, which is why
Leichhardt, for example, have elected government members at every election since 1972.
In the “big swing” years of 1949, 1969, 1975, 1983 and 1996, the two-party swing obtained by the winning party
and the number of seats they won were reasonably accurately predicted by the Mackerras pendulum. The same has
been true at most state elections. This is the “default pattern” for Australian elections. It requires
exceptional circumstances – like the One Nation insurgency in 1998 – to produce a significant deviation from that
This article will therefore assume that if Labor gets a national swing of around 5%, it will probably win the
16 seats which the Mackerras pendulum suggests that it should, give or take a few surprises in both directions.
I may be quite wrong about this. Like all election analysts and commentators, I am frequently wrong – though not
as frequently as some. Maybe Labor will win
Maranoa and lose
Grayndler. Maybe Labor
will get 10% swings its own seats and no swings in the Coalition marginals. But there’s no reason to suppose
those things will happen, and I don’t think they will.
The next thing to note is that Labor goes into the election campaign well ahead in the national polls, with a
comfortable lead which it has held all year. After an initial surge when
Kevin Rudd was elected Labor Leader
in December, peaking at well over 60% of the two-party vote in March, Labor support drifted back to the
low-to-mid 50s in June and July. Since August, however, Labor’s support in the polls has risen again, and has
been remarkably stable at around 56%. A Labor two-party vote of 56% represents a two-party swing of 8.5%, which
on the Mackerras pendulum means a Labor gain of well over 30 seats.
There has been a common assumption among media commentators that Labor’s lead must fall once the campaign period
begins, when the voters “start to take notice” and the government’s re-election campaign begins in earnest. This
is known in the commentariat as “The Narrowing.” Following the Prime Minister’s announcement of large tax cuts
should the Coalition be re-elected, the first round of campaign polls did indeed show some movement back to the
Liberals. We are still to discover, however, whether The Narrowing is a reality or a myth.
If there is to be a Narrowing, it will need to be a very substantial one, far bigger than has occurred in any
recent federal election – or indeed any election at all that I can think of. The swing to Labor shown in current
polls is more than twice as big as the swing Labor needs to win. Essentially the government needs a 5%
two-party swing towards it in the next five weeks, after three months in which the polls have hardly moved at
all. Well, stranger things have happened in politics, but I remain sceptical that the Coalition can pull this
Let us assume, however, that The Narrowing is indeed about to begin. In fact, there’s not much point in writing
articles like this unless we make this assumption, which is why some sections of the media (that means you,
The Australian) have been so busily promoting it. No-one is going to bother reading election commentary
if they believe that Labor is going to hold its current lead and coast to an easy victory. But there is no law
of electoral history which says that The Narrowing must happen.
Even those who believe that The Narrowing is coming accept that the Coalition cannot expect to hold its current
comfortable majority. Getting re-elected even with a majority of one would be quite an achievement for a
government which is starting from so far behind. Getting re-elected without any losses at all would be downright
My starting assumption is that, Narrowing or no Narrowing, the Coalition is going to lose at least ten seats.
Looking at the Mackerras pendulum, it’s not hard to nominate those ten. There are seven seats with Liberal
majorities of under 2%. These are
confident the Liberals will lose all seven. There may be some hesitation about Braddon in the wake of the Mersey
Hospital takeover, but I doubt it. This is a blue-collar seat and one which fear of
WorkChoices will deliver to
Labor’s Sid Sidebottom. There may be more substantial doubts about Hasluck, a Perth suburban seat. But the polls
in WA have improved for Labor lately and I believe
Sharryn Jackson will regain this seat, which she lost in
In the next tranche of seats (2 to 4%) I nominate
Eden-Monaro as certain government
losses. There has been one rather dubious poll suggesting that Moreton may be salvageable, but I doubt it.
I may be proved wrong about any or
all of these seats, but my current view is that Labor is going to win all of them.
That’s eleven likely Coalition losses, so Labor needs five more gains for a majority. There’s certainly plenty
of seats to choose from. In recent weeks we have seen reports of “internal polling” showing some extremely
unlikely Liberal losses, including such traditional bastions as
Ryan (10.4%) and
Casey (11.4%). It’s
impossible to know the truth of these reports. “Internal polling”
is always unverifiable, and is usually leaked for political reasons – if not actually fabricated. These huge
swings may be in the offing, driven by upper-middle-class concerns about
climate change, the
Iraq war and
refugee issues, but then again they may not.
I am not going to be led astray by such speculation. I’m going to nominate twelve seats which I believe will
decide the election – on top, that is, of the eleven I have already nominated as likely Labor gains. All are
well within the range of the swings currently being shown in the polls. If Labor wins all of them, they will
win the election comfortably. If the seats split about evenly, either Labor will win the election narrowly, or
we will have a hung parliament and
Tony Windsor and
Bob Katter will decide our fate. If the Coalition retains
all or most of them, they will have had a remarkable escape from apparently certain oblivion, and
will go down in history as the most brilliant politician Australia has ever produced.
The Twelve Seats of Doom
1. Stirling, WA (2.0%)
Perth’s northern beachside suburbs, has been a marginal seat ever since it was created in
1955 – every member for the seat has been defeated eventually. The inland Labor strongholds of Balcatta and
Balga are balanced by Liberal-voting beachside suburbs like Scarborough and Trigg.
Jann McFarlane won Stirling
for Labor in 1998, but was defeated by the Liberal
Michael Keenan in 2004. Labor has found a strong candidate
in former SAS Major
Peter Tinley, while Keenan seems to be no more than an averagely competent backbench member.
A local poll in June showed the Liberals narrowly ahead. In August another local poll suggested that Labor was
leading. The general view is that Stirling will be harder than Hasluck for Labor to win, but with Labor’s recent
improvement in polls in WA, Tinley must be rated a good chance.
Current assessment: A real cliffhanger, but I think Labor will win.
2. Wentworth, NSW (2.6%)
Second only to Bennelong,
in Sydney’s eastern harbourside suburbs, is the glamour contest of this
election. A seat which Labor has not won in 106 years and which includes some of the wealthiest suburbs in
Australia is now within Labor’s reach, thanks to social change and new boundaries. Wentworth has higher levels
of income and of people in professional occupations than almost any other seat. It has a large Jewish community
and also a large gay and lesbian presence at the western end.
Peter King won the seat in 2001, but lost his
Liberal preselection in 2004 to
Malcolm Turnbull, now Minister for Environment and
Water Resources. The 2006
redistribution has cut the Liberal majority to only 2.6% – although it seems likely that the real Liberal
majority is larger than this, since the 2004 vote was distorted by King’s independent candidacy. Also, in 2004
the Liberals ran dead in the Darlinghurst booths which were then in the safe Labor seat of Sydney. The Labor
candidate is George Newhouse, a Jewish human rights
lawyer. Labor’s decision to support the
Tamar Valley pulp mill has caused Newhouse some embarrassment, and the preferences of the large field of candidates
are hard to predict. Turnbull is no fool and has buckets of money to spend. Most commentators seem to think
he will survive.
Current assessment: I agree: Turnbull to hang on.
3. Solomon, NT (2.8%)
Solomon is the
“mystery seat” of the election, because the national polls don’t bother with the remote and
sparsely populated Northern Territory, and no-one locally seems to care enough about politics to conduct one.
The seat was created in 2001 when the old seat of Northern Territory was divided in two. It covers the city of
Darwin and most of its suburbs. Darwin is a fairly wealthy city with a high proportion of government employees
and professionals. Only 9% of its population is Indigenous, while nearly 5% were born in South-East Asia. In
2001 it had a small notional Labor majority, but a weak Labor candidate allowed the Liberal
Dave Tollner to
scrape in with a tiny margin. In 2004 he substantially increased his majority. Tollner is a rough character of
the type popular in the Territory, but Labor has tried to match this by running
Damian Hale, a popular football
coach. In the absence of polls it is hard to know what is going on, but most local comment suggests that Hale is
expected to win.
Current assessment: Anyone’s guess, but I will take a punt on the coach and say Labor will win.
4. Bennelong, NSW (4.0%)
As everyone knows by now, no PM since
Stanley Bruce in 1929 has lost his own seat. Could it happen again?
Several local polls as well as the national polls suggest that it could, and brave commentators such as
Malcolm Mackerras say that it will.
Bennelong, on Sydney’s
North Shore, was created in 1949 and has had only
two members, both Liberals: John Howard has held it since 1974. It has usually been fairly safely Liberal, although
Labor came close in 1961 and 1972. Since 1977 Bennelong has been shifted westwards by successive redistributions,
and the 2006 redistribution added three Labor booths in Ermington, slightly weakening Howard’s position. More
important has been demographic change, as Ryde and Gladesville have become less reliably Liberal, and as migrants
have moved into the area. Bennelong has the unusual combination of high median family incomes and a high
proportion of people born in non English speaking countries: it now has the highest proportion of such people
of any Coalition-held seat – and this is usually strongly indicative of support for Labor. Many of these
migrants are people from Asian countries who are in professional occupations. We can be sure that Labor is
reminding them of Howard's anti-Asian immigration remarks of 1988. Labor's candidate is the high-profile former
Maxine McKew, and she has proved an
excellent candidate. An upset loss for the PM seems increasingly likely.
Current assessment: I think Howard’s concession that he will retire during the next term will be fatal for him
locally. Labor will win.
5. Dobell, NSW (4.8%)
Dobell, on NSW’s
Central Coast north of Sydney, is a mix of Sydney exurbia and tourism and retirement
centres such as Wyong, The Entrance and Bateau Bay, and has a high proportion of over-65s. It was a fairly
safe seat for the rising Labor star
Michael Lee until 2001, when he was unexpectedly beaten by the Liberals’
Ken Ticehurst. Despite being a very low-profile member, Ticehurst gained a solid swing in 2004, aided by Mark
Latham, the area’s changing demographics and a weak Labor candidate. This is why the seat is considered more
vulnerable than its paper majority would suggest. Labor has been helped by the addition at the 2006
redistribution of Labor-voting Toukley. Labor’s candidate this time is
Craig Thomson, national secretary of the
Health Services Union. Do Dobell voters care
that the Labor candidate is a union official? Apparently not, since
local opinion seems to be that Ticehurst is trailing.
Current assessment: I think Labor will win.
6. Deakin, Vic (5.0%)
Deakin is a block of
Melbourne’s eastern suburbs centered on Mitcham and Ringwood. Deakin is remarkable for
its social and political homogeneity: in all but two booths in 2004, the major parties each polled between 45
and 55% of the two-party vote. Since 1969 Deakin has always been a marginal seat, but Labor has only won it
once, in 1983. The seat's stability is largely due to its solidly middle-class, home-owning character. The
Phil Barresi is a hard worker and will be hard to beat
unless there is a strong swing to Labor
across Melbourne. The Labor candidate,
Mike Symon, is an official of the
Electrical Trades Union, and the
unions are reportedly spending heavily on his campaign. Barresi will certainly try to exploit Symon’s link to the
ETU, whose leader Dean Mighell has been expelled from
the Labor Party. This will have some effect in a middle-class
seat, but will it be enough to offset the anti-Liberal swing which seems to be brewing in Melbourne? I really
can’t call this one, but I will err on the side of caution.
Current assessment: Liberals to hang on.
7. McMillan, Vic (5.0%)
McMillan is a
semi-rural seat in the West Gippsland region of Victoria. It was a safe Liberal seat until
the 1970s, and was not won by Labor until 1980. Since then it has been a marginal seat, changing hands several
times. Socially, the seat consists of several parts. Pakenham is a mortgage-belt outer suburb. The La Trobe
Valley has declining working-class towns like Moe. South Gippsland is a dairy farming area.
won McMillan for the Liberals in 1996, but was defeated in 1998 by Labor’s
Christian Zahra. Zahra’s career was
derailed by the 2004 redistribution, which made the seat notionally Liberal. Even so he might have hung on had
it not been for
Mark Latham's anti-logging forestry policy, but this was enough to return the seat to Broadbent.
The loss of Zahra’s personal vote probably means
that Broadbent is safer than the nominal margin suggests. His reputation as a backbench rebel on issues such
as refugees may also help him locally. The Labor candidate is
Christine Maxfield, a
Unionemployee whose husband
was the local state member until his defeat in 2006. Local reports suggest she is not a particularly brilliant
candidate. Rural seats don’t always reflect statewide swings, although McMillan did so in 1980 and 1993.
Current assessment: Liberals to hang on.
8. Corangamite, Vic (5.3%)
a semi-rural seat in south-west Victoria, based on the Colac area, the outer suburbs of
Geelong and the Surf Coast towns. The growth of these towns has made what was once a very safe Liberal seat
increasingly marginal, although Labor has not won it since 1929.
Stewart McArthur, who comes from an old Western
District and is well-known in the rural parts of the seat, has held Corangamite since 1984, and at 70 he is
seeking another term. His appeal in the urban areas and among the many new voters on the coast is uncertain.
Labor had high hopes of winning Corangamite in 2004, but the combination of Mark
Latham’s unpopular forestry
policy and a Labor candidate surrounded by controversy kept McArthur safe. This time the Labor candidate is
Darren Cheeseman, a young
Labor hopeful of no particular distinction, and thus probably not the best choice
against an entrenched sitting member. But most reports suggest that there is a big swing going on in urban
Victoria, and that will include the Geelong suburbs and the Surf Coast towns. The general view is that McArthur
has stayed on to fight one battle too many.
Current assessment: Labor to pinch it.
9. Page, NSW (5.5%)
Page is based on the
North Coast regional centres of Grafton and Lismore. Its economic base is farming and
low-wage tourism jobs, and it has a large population of retired over-65s. These factors make Page more
politically marginal than most rural seats are these days. In 1990 Page unexpectedly elected a Labor member,
Harry Woods. Woods hung on
in 1993, but was predictably defeated in 1996 by the Nationals candidate, former state minister
Causley, who was quite popular locally, is now retiring. The Nationals candidate is Clarence Valley councillor
Chris Gulaptis. National Party seats are always vulnerable when
sitting members retire, and Labor has come up
with a credible candidate in former local state member Janelle
Saffin. The Nationals are in decline everywhere,
but particularly on the North Coast, and this is also an area where many low-wage workers will be very worried
about WorkChoices. The loss of Causley’s vote may be just
enough to tip the seat.
Current assessment: I think Labor will win.
10. La Trobe, Vic (5.8%)
La Trobe is
a typical mortgage belt seat in Melbourne’s outer eastern suburbs. It takes in the Dandenong
Ranges townships, which generally vote Labor, as well as more middle-class areas like Narre Warren and Berwick,
which are full of aspirationals and usually vote Liberal. In the 1970s La Trobe was a key marginal seat, but
since being reclaimed by the Liberals in 1990, it has grown increasingly secure, mainly because of the
increasing conservatism of the Berwick - Narre Warren area. Nevertheless it remains within reach for Labor in
a good year, as the polls are suggesting that 2007 is.
Jason Wood easily won the seat in 2004, which wasn’t
really a test given Mark Latham’s failure to attract
suburban voters. Earlier this year Labor suddenly replaced
its candidate, a union official, with
Rodney Cocks, a former soldier and
Victorian of the Year. This
suggested to many that Labor polling was showing the seat to be winnable, and subsequent rumour seems to
confirm this. Despite the solid Liberal majority, the swing seems to be on.
Current assessment: I think Labor will win.
11. Blair, Qld (5.7%)
started out in 1998 as a largely rural seat. It attracted national attention at the 1998 election
because Pauline Hanson, the extreme right-wing member
for Oxley, elected as a
disendorsed Liberal in 1996,
stood in Blair for her new One Nation
party. She topped the poll on primaries, but was defeated on preferences
by the Liberal Cameron Thompson, who has held the seat since without much difficulty. The 2006 redistribution
has moved the seat eastwards, removing many of its rural areas and putting the city of Ipswich, which is
strongly Labor, into the seat. This has brought Blair within reach of the swing currently being shown by
opinion polls in Queensland. The Labor candidate is
Shayne Neumann, an Ipswich lawyer.
Thompson is a fairly
inconspicuous backbencher and I doubt he has much personal vote even in the parts of the seat he currently
represents. Since 1961 Ipswich has usually delivered whatever seat it is in to Labor.
Current assessment: I think Labor will win.
12. Herbert, Qld (6.1%)
covers Townsville, the regional administrative centre of north Queensland and a booming tourist
town. It was once a safe Labor seat, but in recent times, as Labor’s regional base has declined, Herbert has
trended Liberal, a process hastened by the large military vote.
Peter Lindsay has held it for Liberals since
1996 without too much difficulty, although he was run close in 2001. Lindsay is a hard worker and is popular
locally, but the same was true of Ted Lindsay (no relation), the Labor member he defeated in 1996. If there
is a big swing, Herbert behaves like an urban seat and goes with the flow. Labor now has a strong and atypical
candidate in well-known Townsville businessman
George Colbran. With polls currently
showing Labor gaining a
big swing across Queensland, Herbert is at serious risk, and a local poll in September showed Labor ahead.
Current assessment: I think Labor will win fairly easily.
So there we are. I have tipped Labor to win nine of the twelve deciding seats, on top of the eleven I named
earlier as likely Labor gains. That makes 20, and whatever new figure The Australian may come up with
tomorrow, that will be enough to make Kevin Rudd Prime Minister. Of course there are many other seats well
within reach of the swings currently being seen in the polls or suggested by "internal polling." These include
Robertson in NSW,
McEwen in Victoria,
Petrie in Queensland,
Canning in WA, and
Sturt in SA. It's quite
possible that the Liberals will hang on to some of the twelve seats I listed earlier, yet lose some of these