We’ve taken your request to beef up our Victorian election coverage. Today, Crikey’s Philosopher-in-Residence and former Kennett staffer, Charles Richardson reports:

This month’s Victorian election is more than usually difficult in this respect. Commentators have been saying that Labor needs to win seats to retain government, but they offer conflicting (and confusing) explanations of why this is and how many seats are needed.

This article, then, is an attempt to fix the proper starting point – the numbers of seats with which the parties will start, and therefore on the targets that they need to win. In the just-dissolved Legislative Assembly, Labor held 44 seats, the Liberal Party 35, the National Party 6, and there were 3 independents, for a total of 88. So it would seem Labor needs to gain one seat for an absolute majority.
There are three reasons to doubt that simple analysis: the redistribution, by-elections, and the independents. Let’s take each of those in turn.

In the redistribution conducted last year, the total number of seats remained at 88. Of those, 67 have retained the same names – some of them have significantly changed boundaries (we’ll come back to that), but none of them have changed so much that it is unreasonable to think of them as still being the same seats. A further 18 of the new seats can be readily identified with old seats of a different name. Some of these have undergone major changes, but the boundary commissioners also went in for quite a lot of renaming even when there were only minor changes (the old Dromana, for example, was almost identical to the new Nepean). These 18 seats, with their old and new names, are as follows:

Berwick -> Narre Warren South
Coburg -> Brunswick
Dandenong North -> Mulgrave
Dromana -> Nepean
Geelong North -> Lara
Gippsland West -> Bass
Gisborne -> Macedon
Glen Waverley -> Mount Waverley
Knox -> Ferntree Gully
Mooroolbark -> Kilsyth
Pakenham -> Gembrook
Springvale -> Lyndhurst
Sunshine -> Derrimut
Tullamarine -> Yuroke
Wantirna -> Scoresby
Warrnambool -> South-West Coast
Werribee -> Tarneit
Wimmera -> Lowan

(Antony Green, in his analysis last year for the parliamentary library, describes
Pakenham and Warrnambool as abolitions, and their replacements as fresh creations, but in each case the new seat clearly corresponds to the old one and gets the large majority of its voters from it.)

That leaves only three seats that have disappeared altogether – Bennettswood, Frankston East and Portland – and three genuinely new seats: Hastings, Kororoit and Narre Warren South. Bennettswood and Portland were Liberal seats; the new Hastings and Narre Warren South are also Liberal seats (the latter very marginal). Frankston East was a Labor seat, and so is the new Kororoit. Those changes therefore involve no net gain or loss in seats for either party.

A number of existing seats, however, have changed in political colouring as a result of the new boundaries. The exact magnitude of those changes is uncertain, because projecting old results onto new boundaries is not an exact science. It involves making a series of assumptions and educated guesses, which leave a fairly substantial margin of error.

I have made my own calculations; Antony Green made some last year for the parliamentary library, and a revised set that appears on the ABC’s website and also appears in newspapers. The three sets of numbers are in substantial agreement, which is encouraging. I shall use Green’s latest figures, since they are the most widely available (the differences from my calculations are mainly due to a different way of allowing for postal and absentee votes; Green’s method should in principle be more accurate, but both are speculative). On these figures, there are four seats that were won by Labor in 1999 but would now be Liberal seats: Macedon (formerly Gisborne), Geelong, Yan Yean and Narracan. All of them, however, are extremely close; the biggest margin of the four (Narracan) is still only 0.6% (about 200 votes). Moreover, none of them have actually changed very much; they were all marginal seats to start with, and they still are.

In my view it is counter-intuitive to now start thinking of these four as Liberal
seats. Labor will have the advantage of incumbency in all four, and as parties have become more and more unscrupulous about using taxpayers’ resources for their own purposes, incumbency has become more and more of an asset. (A development that was pioneered by the United States Congress, whose re-election rate once exceeded that of the Supreme Soviet.)

It seems more logical to continue counting these four seats as Labor-held, albeit
ones that the opposition needs the minimum possible swing to win. Yan Yean might seem at first to be something of an exception (its sitting Labor member is moving to a safer seat, and its territory is largely metropolitan, where incumbency is less important anyway), but since demographic change in that area is running in Labor’s favour, and since the calculation of the redistribution’s effects there is particularly tricky, I propose to count it on the Labor side as well.

One seat, Cranbourne, has moved far enough the other way to become notionally
Labor-held on the new boundaries. Both the new margin and the change involved are
greater than they were with the four seats that moved in the Liberals’ favour (new margin 1.4%, previously 5.7%). Even so, the change is not great enough to prevent us treating it the same way and continuing to regard it as a Liberal seat – their most marginal. In summary, the redistribution, while it has been favourable to the opposition, has not handed it a big swag of seats. It will still have to work to win the seats that have notionally moved its way, just as Labor will have to work to win Cranbourne. Governments are made and unmade by the voters, not the boundary commissioners. As far as the redistribution is concerned, I feel it makes sense to regard Labor as being still only one seat away from majority government.

The second complicating factor is that two of Labor’s seats, Burwood and Benalla,
were not won by it at the 1999 election but were gained in subsequent by-elections
following the retirement of Jeff Kennett and Pat McNamara respectively. Those victories lifted Labor’s tally from 42 to 44. (The Frankston East supplementary election, which gave Labor its 42nd seat, was sometimes referred to as a by-election, but it was really just a deferred part of the 1999 general election.)

When calculating the swing required for a seat to change hands, it is usual to ignore
by-elections and use just the figures from the previous general election. An exception is often made, however, when (as in these two cases) the seat changed hands at the by-election. Malcolm Mackerras, for example, put Ryan on the Labor side of his pendulum before the 2001 federal election, just as he had put Namadgi on the Liberal side prior to the 1996 election.

This practice makes sense, again because of the power of incumbency. It is by no
means certain that the party winning by-elections will hold them next time (they didn’t in Ryan or Namadgi), but they will have a much improved chance of doing so. Labor will defend Burwood and Benalla as Labor seats, vulnerable but nonetheless theirs to lose, rather than seeing them as the relatively safe non-Labor seats that they were at the 1999 election.

The third factor to consider is the presence of independents. The three independents
have stuck by their promise to maintain Labor in power, and it does not appear to have hurt them in their electorates. Only one of them, Susan Davies in Bass (formerly Gippsland West), seems to be in serious danger of not being returned, and that is more a result of unfavourable boundary change. Unfortunately for Labor, she is also the most pro-Labor of the three.

There is no certainty about how the independents, and others who may join them,
would vote if they should again hold the balance of power. It would be unreasonable to take them for granted by adding their numbers to the Labor total, or for that matter to the opposition’s. In this they should be contrasted with the Greens, who have a chance of winning some inner Melbourne seats. There is no real doubt that Green MPs, if their votes are needed, would vote to keep Labor in power, just as we know the National Party would vote to put them out.

My conclusion therefore is that none of these complicating factors fundamentally
changes the arithmetic of the election. Labor goes into the election with 44 seats; it needs to make a net gain of one for majority government. The opposition starts with 41 seats; it needs a net gain of four. Failing either of these results, independents will continue to hold the balance of power.

Simple, really. But sometimes it’s worth taking the time to establish how simple things are.

Charles Richardson