It’s two weeks tomorrow since the Tasmanian state election, and the state remains in limbo as it waits on the result of the final seat in Bass. Counting is well advanced – you can follow the official figures here, and Kevin Bonham’s commentary here – but it’s still extremely close and is not expected to be decided until tomorrow. It looks to me as if Labor is to be slightly favored, which (as discussed last week) would count for something towards its prospects of forming government, if perhaps not much.
While we’re waiting on that, you can read an interesting post this week from Ben Raue, alerting us to the fact that it’s redistribution time again. Three states – Queensland, South Australia and Tasmania – plus the Australian Capital Territory are due to have their federal boundaries redrawn during the current term of parliament. The process will probably begin this year for all of them, although it’s possible (as Raue explains) that Queensland might be deferred for a year in case it turns out to be entitled to an extra seat.
Apart from that slim chance for Queensland, there is no prospect of any state gaining or losing a seat; the only reason for holding these redistributions now is that it has been seven years since the last ones. I remain sceptical about the need for this to be done so frequently, just as I’m sceptical about the need for reapportionment to be done every term.
We can’t say much about what will happen in these redistributions because we don’t yet have projected enrolment figures. These are the estimates from the Australian Bureau of Statistics that the commissioners use to equalise enrolments, to within 3.5% of average, at a date roughly three and a half years into the future. That’s the main constraint on how boundaries are drawn; since the primary criterion, that seats be within 10% of actual average enrolment, is much easier to satisfy, it’s the 3.5% rule that does most of the work.
Which is a great pity, because the projections are hopelessly unreliable. Consider, for example, Tasmania, which is not just topical but also significant, since its federal boundaries are used for state electorates as well. You can read the report of its last redistribution (finalised in 2017) here, and compare the table of projected enrolments then (it’s on page 30) with the actual enrolment figures for the projection date, namely May 2021.
I’ve compiled them into the following table, showing for each of the three sets of numbers the deviations from the average enrolment at that date.
| Division | Sep. 2016 (actual) | Deviation from avge | May 2021 (projected) | Deviation from avge | May 2021 (actual) | Deviation from avge |
| Bass | 74,467 | -0.7% | 75,653 | -1.7% | 78,434 | -1.0% |
| Braddon | 77,739 | +3.6% | 77,992 | +1.3% | 81,533 | +2.9% |
| Clark | 74,054 | -1.3% | 76,080 | -1.2% | 74,224 | -6.3% |
| Franklin | 73,304 | -2.3% | 76,877 | -0.1% | 78,576 | -0.8% |
| Lyons | 75,508 | +0.7% | 78,313 | +1.7% | 83,479 | +5.3% |
As you can see, there’s not much relationship between what was projected and what actually happened to enrolments. In three of the five, the deviation ended up being closer to what it started out at than what it was projected to; the commissioners would have been better off assuming there would be no relative change between the seats at all. In one of those (Clark) the change was in the wrong direction: the projections said it would grow (slightly) relative to the others, but in fact it went backwards quite substantially. In Lyons they got the direction right but the magnitude badly wrong.
You might think that this is unfair, and note that the Covid pandemic happened during that period, upsetting a lot of calculations. But that’s the point; unexpected things always happen, making the concentration on projections fraught with risk. As it happens, Tasmania is probably the most stable of the states: if the projections can’t get that right, they’re going to have trouble across the board.
And sure enough, if you look at the corresponding figures for Queensland (here’s the original report (refer to page 26) and here’s what actually happened), they’re even worse. Fully half of the thirty seats had relative enrolment move in the opposite direction to what had been projected: the commissioners could have saved much time and effort by just tossing a coin. Eighteen seats were outside the 3.5% tolerance; four of them were out by more than twice that.
Some of the differences are huge. Fisher, for example, started out at 0.5% above average enrolment, and was projected to decline modestly to 1.2% below average. Instead, by the projection date it was 7.0% above. Moreton started at 3.0% below and was supposed to grow to only 1.5% below; instead it kept falling, reaching 7.5% below. And this all means that some voters are being short changed to the unfair benefit of others.
It’s important to stress that none of this is the commissioners’ fault; they’re doing the best they can with the rules and the data that they’re given. What’s required is for parliament to address the problem by focusing the process more on actual voters and less on hypothetical ones – who so often turn out not to exist or to live somewhere quite unexpected.
The World is Not Enough 