The Mackerras Pendulum ranks all the seats in two-party preferred terms from highest to lowest Labor percentage. Marginal seats, therefore, lie in the middle, at the bottom of a horse-shoe type of diagram. The party holding the seat is always shown. Next to it is shown the percentage swing required for that party to lose the seat.
For some fifty years, my pendulums have been published in newspapers and magazines. I would call it “electoral pendulum” but then, in the early seventies, came along the eminent British psephologist David Butler. He started to call it “Mackerras Pendulum”. I thought this to myself: “If David Butler thinks it should be called that why should I not also call it the eponymous ‘Mackerras Pendulum’?”
Regarding the order of presentation of my pendulums, my idea is that the first to appear on the blog should be that in respect of the election most proximate in the future. Consequently the first pendulum to appear here (note that it is two-sided) is the one being the guide to the October 2020 Queensland State election.
With the exception of Queensland, for the time being I do not plan to post state pendulums, though I may change my mind on that in the future.
Following each election people ask me this: how well did your pendulum work? In respect of the May 2019 federal election I can honestly say that it worked very well indeed. The election saw a swing to the Coalition of 1.1%. This is where my argument about the pendulum comes into play. I do not claim that swings are uniform. What I claim is that deviations from uniformity can be relied upon to cancel out. On that basis my pre-election pendulum worked well – and as expected.
That does not mean my personal prediction of the result was correct. Relying on the opinion polls I predicted that the distribution of the two-party preferred vote would be 51.5% for Labor and 48.5% for the Coalition. The actual vote turned out to be the reverse of that. So, Labor did 3% worse in the two-party preferred vote than the polls predicted.
Following the election disaster for me (and virtually every other pundit) my critics told me that it was “time to give the pendulum the flick”. Not so – it is time to give the opinion polls the flick, not the pendulum. If I had known the swing would be one per cent to the Coalition I would have predicted that the majority for the Coalition minority government would increase from 3 to 11 or 13. That is what happened. The reason is that the deviations from uniformity did cancel out – one seat wrong, to be precise.
Look at it this way. If there had been a uniform swing of 1.1% to the Coalition then the Coalition would have won Herbert, Corangamite, Cowan, Longman and Lindsay. As things turned out Labor won Corangamite and Cowan, but the Liberals won the Tasmanian seats of Bass and Braddon, cancelling out Corangamite and Cowan.
When I wrote above “one seat wrong, to be precise” the one seat wrong was Gilmore which Labor gained from the Liberal Party in peculiar local circumstances.
The most interesting seat in the election was Warringah. The actual result is shown on the reverse side of the pendulum. Why, then, does the pendulum place Warringah where it is? The answer is that the Australian Electoral Commission recounted votes to being between Tony Abbott and Labor. The result was a narrow win for Abbott over Labor. Hence, I show Warringah with the other very marginal Liberal seats, Bass, Chisholm and Boothby.
The next two seats on the pendulum above Warringah are Mayo (where the circumstances were rather like those of Warringah) and the median seat which is now Swan (WA). Labor needs a swing of 2.7% to win Swan. Therefore, I say that, to elect a Labor government, it will require a uniform swing of 2.7% in 2022. All that will change, however, when there are further redistributions of seats during the term of the 46th Parliament.
On the eve of the 2016 double dissolution, The Australian newspaper unexpectedly told me they were no longer interested in my services for federal elections. This prompted me to have them printed professionally by Digital Synergy, so I could continue to offer them to others. Consequently, the first pendulum below is that preceding the 2016 election. It was done by Digital Synergy Pty Ltd, a firm located at 26 Navigator Place, Hendra, a north Brisbane suburb.
Having been doing these pendulums for fifty years (more than a hundred of them is my guess) I think I am entitled to say that my pendulum is “the original and the best”. For that reason I do not intend to change the methods I developed fifty years ago. So I had best explain where I differ from those experts who are much younger than me. I do that by taking the most marginal seats on my federal 2019 pre-election pendulum, that being the first to appear here.
There were redistributions but the following statistics need to be remembered: with 151 seats to be contested it is important to note that 76 seats are exactly the same as in 2016. They are the 47 seats in New South Wales, the 16 seats in Western Australia, 12 of the 30 seats in Queensland and one of the 38 seats in Victoria, that seat being Goldstein. None of the above seats has seen its boundaries changed – nor even had a name change.
There are 75 seats that have seen their boundaries or name changed – or both. They are 37 of the 38 in Victoria, 18 of the 30 in Queensland, all ten in South Australia, all five in Tasmania, all three in the Australian Capital Territory and both in the Northern Territory. These changes (or lack thereof) explain my new pendulum.
In the 46th Parliament elected in May 2019 there are, for the House of Representatives, 47 seats in New South Wales, 38 in Victoria, 30 in Queensland, 16 in Western Australia, 10 in South Australia, five in Tasmania, three in the ACT and two in the Northern Territory, a total of 151. Compared with the 45th Parliament that is one more member in each of Victoria and the ACT and one fewer in South Australia.
For my federal pre-2019 pendulum, I come to the three most marginal seats for the Coalition and the four most marginal for Labor. I begin with Capricornia in central Queensland. The redistribution added 1,092 electors from Dawson, giving Capricornia 99,939 electors. My calculation of the effect of such addition was that the Liberal National Party had 50.59 per cent of the two-party preferred vote on the adjusted boundaries on 2016 statistics. To reduce that to below 50 per cent needs a 0.6 reduction. Therefore, the swing required for Labor to take Capricornia was 0.6 per cent. The next seat is Forde (Qld), the median seat. It has not seen its boundaries changed. On the 2016 figures the Liberal National Party secured 50.63 per cent. Therefore, on my logic, Labor needed a swing of 0.7 per cent, not 0.6 because a 0.6 reduction would still leave the LNP with 50.03 per cent.
New South Wales and Western Australia did not have redistributions during the 45th Parliament. They had theirs during the 45th Parliament when WA gained a seat from NSW. So I now come to Gilmore where the Liberal Party secured 50.73 per cent during the 44th Parliament. Therefore, on my logic, the swing required for Labor to take Gilmore was 0.8 per cent, not 0.7 since such a reduction would still leave the Liberal Party with 50.03 per cent. Labor did win Gilmore in May 2019 and the swing was 3.3 per cent.
Here is where my fifty-year-old analysis differs from that of those men much younger than I am. They would say: “The margin in both Capricornia and Forde was 0.6 while in Gilmore it was 0.7.” I admit that “margin” is easier to say than “percentage swing required to lose” but I have a different idea of the word “margin”. Take the case of a seat having been 54 Labor and 46 Liberal last time. I would say the margin is 8 and the swing needed by the Liberal Party to take that seat is 4.1 per cent.
I come now to the Labor side, beginning at the other end of the first band of marginal seats. Where analysts agree is on Longman. It is the universal practice to ignore by-elections – unless the party holding the seat changes. Longman is unchanged in boundaries and Labor’s share in 2016 was 50.79 per cent. So the “margin” (on new-fangled language) is 0.8 and the percentage swing needed by the Liberal National Party to take Longman is also 0.8 per cent. Exactly this same kind of analysis applies to Cowan (WA) where the Labor percentage in 2016 was 50.68 and that for the Liberal Party was 49.32. So the Liberal Party needed a swing of 0.7 per cent to take Cowan. As things turned out, the Liberal Party gained Longman but Labor held Cowan.
The single most interesting seat in that whole round of redistributions was Corangamite (Vic). In my own calculations I had it as just Liberal, much weakened by boundary changes but still Liberal none-the-less. I did an interim draft pendulum showing it that way which I gave to a few people who wanted it. However, my uncertainty about this case was such I decided to wait until the “official” statistics from the AEC were published. They were published early in February 2019 in a document known as “National Seat Status”. That document showed Labor on 50.03 per cent in Corangamite and Liberal on 49.97 per cent.
So I junked my interim draft pendulum and published the one on this blog dated 12 February 2019. I deferred to the AEC in five seats after receiving that document, the biggest change being in Bass. On my interim draft pendulum I showed the Liberal Party as needing a swing of 6.4 per cent to take Bass. The AEC statistics show Labor on 55.42 per cent and the Liberal Party on 44.58 per cent. Therefore, I now show the Liberal Party as needing a swing of 5.5 per cent to take Bass from Labor. The Liberal Party did, in fact, take Bass on a swing of 5.8 per cent.
The closest result in 2016 was in Herbert, based on Townsville (Qld). The winner was Labor’s Cathy O’Toole with 44,187 votes (50.02 per cent) while the loser was then sitting member Ewen Jones of the Liberal National Party who secured 44,150 votes (49.98 per cent). Since Herbert is unchanged in boundaries I show Labor as needing a swing of 0.1 per cent to lose Herbert. Note, however, it was a weaker seat for Labor than notionally-Labor Corangamite. Labor lost Herbert but won Corangamite.
Labor suffered an overall swing of 1.2% against it in May 2019 (meaning the Coalition received 51.6% of the two-party preferred vote). Of the seats up to the two per cent marker, Labor lost Herbert (Qld), Longman (Qld), Lindsay (NSW) and Braddon (Tas). Labor won Corangamite (Victoria), Cowan (WA), Macnamara (Vic), Dunkley (Vic) and Griffith (Qld).
Back in the days when I was in favour at The Australian they would always publish post-election articles of mine (with pendulums) for all federal and state elections employing systems of single-member electoral districts. The last such federal article was published in The Weekend Australian on 4 and 5 of January 2014. The article on page 13 was titled “Parliamentary ups and downs: it’s all in the swing” to which there was my further comment on the September 2013 election. It read: “After its respectable loss, Labor will need a 4.1 per cent vote gain to win in 2016”.
Anyway, I did a post-2016 pendulum which is the third to appear here. Since no newspaper wanted it I would send it to people who wrote to me asking for it. Notice that it has two sides, the pendulum on one side (being entirely on a scale of Labor versus Liberal-National) while on the other side are those counts made by the Australian Electoral Commission not conventional for a two-party preferred vote.
The third pendulum shown here is the one devised prior to the 2016 election. Two newspapers took up my offer. Consequently, that pendulum was published with an article by me on page 49 in the 12-page Election Special of Melbourne’s Herald Sun newspaper on Friday 1 July 2016. The article was titled “A Tight Race to the Finish Line”. The following day that pendulum was published in Sydney’s Daily Telegraph newspaper on polling day itself, Saturday 2 July 2016. The two-page article on pages 44 and 65 was titled “Winners and Losers of the Race” to which there was this further description: “It’s Easy to Predict Malcolm Turnbull’s Victory, but the Details of the Win are much harder to Pin Down, writes Malcolm Mackerras”.