Lucky numbers

All that time redesigning the site has prevented me making a timely entry into debate over the report of the Joint Standing Committee on Electoral Matters. The main headline-grabbers have been voluntary voting and four-year terms, which are respectively dead-on-arrival and likely to land in the too-hard basket due to the need for a referendum (at least on the current Prime Minister’s watch). Of greater significance is recommendation 37, "that compulsory preferential voting above the line be introduced for Senate elections, while retaining the option of compulsory preferential voting below the line". Antony Green has had a fair bit to say on this one:

A stinker of an electoral reform idea … a recipe for an informal rate of 20% or more. Sure you wouldn’t have had to fill in preferences for all 78 candidates below the line on the 2004 NSW Senate ballot paper, but you will have to fill in 29 preferences above the line. Whoopee! … Above all, the Committee’s mindless insistence on refusing to countenance any form of optional preferential voting is breathtaking. In his speech to the Sydney Institute last week, Senator Abetz stated that apart from the first five preferences, and perhaps the last five, voters don’t really care about their preference sequence and probably end up filling in the numbers randomly. If that’s the case, what exactly is the point of making voters fill in all their preferences?

Writing in yesterday’s Crikey email, Charles Richardson broadly agreed but noted that the abolition of preference tickets would mean an end to the use of dummy parties for preference harvesting, thus reducing the number of boxes to be numbered. The estimable Dr Graeme Orr of Griffith University Law School went so far as to dispute Antony’s basic premise, suggesting a requirement to number every box would bring the two houses’ systems into line and reduce "’1′ only" informal voting for the lower house. Antony has responded to both in typically persuasive fashion with another article in Crikey.

Ultimately, this is a disagreement about the second best system – all concerned agree that optional preferential voting for both houses is the best option. It is the only solution to the preference lottery that will not inflate the informal vote, and it will deliver the last two seats in any given Senate race to the candidates who come nearest a quota on the primary vote, as natural justice demands. The Coalition and the Democrats, and hence JSCEM (the committee includes five Liberal, one Democrat and four Labor members), have concluded otherwise. The Greens are not represented on the committee but they evidently concur – Bob Brown introduced a bill to the Senate earlier this year to require that all above-the-line boxes be numbered, with no provision for optional preferential voting.

It is clear why the Liberals and Nationals are not keen on OPV as their votes could no longer be relied upon to reinforce each other through preferences. This would compel them to avoid three-cornered contests in the House and run joint tickets in the Senate, provoking an eruption of turf wars they would prefer to avoid. Presumably the Democrats favour full distribution of preferences as it boosts the prospects of parties who underperform on the primary vote, which is the regrettable position in which they find themselves. The Greens’ wariness about OPV for the Senate (the House is a very different matter) makes less sense, as I argued in April.

An interesting observation was made at Palmer’s Oz Politics by "Sceptic", bearing in mind that anonymous blog comments should be treated with due caution:

The interesting side issue will be the attitude of the party administrations. I don’t think either (Liberal federal director) Brian Loughnane or (ALP national secretary) Tim Gartrell will be too impressed with changes to the ATL Senate voting. More informal votes lead to less public funding for the major parties. It is an open secret that Mr Loughnane leaned on John Howard to junk the option of voluntary voting for this very reason. Imagine if these reforms lead to a financial crisis for the major parties in the future. You would assume that the PM, a loyal party man, would consider this very carefully.

For this and other reasons, my expectation is that the Government will invoke concerns over the informal vote to justify the abandonment of recommendation 37, which will join voluntary voting and four-year terms in the graveyard of major reform proposals resulting from the committee inquiry.

Do as you’re told

Even before the Prime Minister formally scotched the idea, it was clear that Liberal Party advocates for voluntary voting should not have been holding their breath. The implacable opposition of the non-government parties would have required united Coalition support to get it through the Senate, but the Nationals (not to mention many Liberals) are no more keen on the idea than Labor. No doubt the National Party knows its own business, but I am puzzled by their apparent conviction that voluntary voting would damage them. It is true that much of their support comes from rural and small-town areas where incomes are no higher than in urban Labor seats, but it’s also true that voluntary council elections attract far higher turnouts on the National Party’s turf than in the cities. Why would federal elections be any different?

The Australian Election Study has been asking respondents if they would have voted if not compelled to since the survey after 1996 election, providing a one-to-five ranking from "definitely would have voted" (chosen by about 70 per cent of respondents) to "definitely not". Combining results from the four surveys from 1996 to 2004 allows us to cross-tabulate these responses with voting intention for nearly 7000 respondents, including 2976 who voted for the Liberal Party, 2607 for Labor, 333 for the Nationals, 332 for the Democrats and 305 for the Greens. The following table combines responses of "probably would have voted" and "might, might not" into a "maybe" category, and "probably not" and "definitely not" (the latter accounting for only 2.6 per cent in 2004) into "unlikely". The Democrats and Greens vote has also been combined because patterns for the two were very similar.

This makes it clear enough why Labor is united in opposition to voluntary voting, and bolsters Liberals who support it on the basis of realpolitik. But it doesn’t explain why the Nationals are less keen, since their own pattern is no different from the Liberals?. It’s also interesting to note that the Democrats and Greens appear to do well out of reluctant voters. This doesn’t surprise me, since minor parties have traditionally absorbed protest votes from the politically disengaged. However, it runs contrary to Bob Brown’s argument that his support for compulsory voting must be founded purely on principle since his party has "the highest proportion of tertiary-educated voters who are most likely to vote without compulsion". I might have marked that down as self-serving spin, but Antony Green also reckons that "the only party certain to benefit from voluntary voting would be the Greens, who have by far the highest ratio of members to voters of any Australian political party". I hesitate to say that an argument of Antony’s does not persuade me, but I still need to be sold on this one. While it is true that the party has a large activist support base, this does not preclude the possibility that just as many of its voters are alienated and disengaged, and that it receives relatively little support from those who fall in between.

One thing past experience makes clear is that electoral reform is governed by the law of unintended consequences, as demonstrated by the parties’ repeated failures to skew the system in their favour. Malcolm Mackerras notes in Crikey that the last two substantial changes to the system – the expansion of Parliament in 1984 and the introduction of semi-proportional representation to the Senate in 1948 (note the eerie Orwellian inversion of those two dates) – were introduced by Labor governments acting out of perceived self-interest, but they have resulted in fairly regular Senate majorities for the Coalition and none for Labor. By the same token, the conventional wisdom that Labor would suffer under voluntary voting might well be misplaced. Antony Green observes that turnout at this year’s British election was especially low in Labour strongholds, suggesting compulsory voting would have boosted their share of the national vote without having much effect on seat totals. By extension, voluntary voting in Australia could cut margins in safe Labor seats without having a substantial impact in marginals.

There are countless other imponderables, such as the impact it would have on the way parties operate. The need to mobilise voters might even compel Labor to re-engage with the world beyond its internal culture and the attendant bastardry and skulduggery that we’ve been hearing so much about lately.

Poll Bludger Version 2.0

Welcome to the new-look, WordPress-powered Poll Bludger, which comes complete with comments and automatic archiving and sundry other delights that I wrongly thought I woudn’t need when I got the ball rolling. I have ironed out the site’s kinks to the extent that it looks okay on Internet Explorer 6.0, but past experience suggests there might still be problems with other browsers. Any issues that could be brought to my attention would be much appreciated. Comments are open on this post and the one previous, so please drop in and say hi.

Mind exercised

With the unstoppable Kim Beazley juggernaut powering on to certain victory at the next election, the Poll Bludger has been bombarded with emails demanding to know what sort of Senate an incoming Labor government might face. Well, one email anyway. For the benefit of anyone who has been suffering sleepless nights over this recently, I present for public enlightenment my humble reply.

It’s definitely possible that the Coalition could lose power and retain control of the Senate. In 2001 the Tasmanian Senate election result was Liberal 3, Labor 2 and Greens 1, even though Labor won all five of the House of Representatives seats. At the 1993 election when John Hewson went down to Paul Keating, the Coalition won three of the six Senate seats in every state except Tasmania, where Brian Harradine won the sixth seat. So unless the Coalition suffers a big defeat at the next election there is some chance that they would maintain their absolute Senate majority in opposition, and a very high likelihood that they would only fall one or two seats short (with one seat still held by Family First).

Therefore, I think it all but certain that an incoming Labor government would call an early double dissolution election, as did Bob Menzies in 1951 and Gough Whitlam in 1974 (the Coalition’s aforementioned strong performance in the 1993 Senate election precluded such a necessity in John Howard’s case, and Bob Hawke came to power at a double dissolution election in 1983). Given the very different arithmetic that applies when the states elect 12 members rather than six, such an election would substantially cut the Coalition’s Senate numbers. A few more half-Senate elections hence, Labor might again face the problem of a Coalition with at or near half the Senate seats if they are still in power, given that the Coalition finds it easier than Labor to win three seats from six in any given state. But the existence of the double dissolution mechanism means that governments will only suffer an opposition majority in the Senate if they are in too weak a position to face an election, the obvious example being Whitlam in 1975.

The power of one

These are momentous times in the electoral history of the Poll Bludger’s newly readopted home state of Western Australia, which is about to become the last state to abandon rural vote weighting for the lower house and adopt what is known, more or less accurately, as "one-vote one-value". After numerous amendments made to secure the support of the Greens and ex-Liberal independent Alan Cadby in the upper house, the bill finally completed its passage through parliament on Tuesday, just days before members elected on February 26 were due to take their seats and put the necessary majority beyond the government’s reach. The legislation as passed abandons Labor’s campaign promise that the five remote electorates in the upper house region of Mining and Pastoral would be quarantined from its effects. Instead, concerns about servicing of remote areas have been accommodated through a measure similar to that which operates in Queensland, in which electorates with an area of more than 100,000 square kilometres will be deemed to have a bonus enrolment of "phantom" voters equal to 1.5 per cent of the electorate’s area in square kilometres. Other electorates will have roughly 21,000 voters, compared with the current average of about 26,000 for metropolitan and 14,000 for non-metropolitan seats.

The bill provides for two extra members in each house, with the metropolitan area gaining eight lower house seats and the non-metropolitan area losing six. This is unambiguously bad news for the Coalition in general and the National Party in particular, but few who are not directly affected would argue that it amounted to a violation of natural justice. The existing system has produced all manner of absurdities, like the existence of tiny non-Perth urban electorates such as Mandurah, Albany, Bunbury, Kalgoorlie, Dawesville and Leschenault. Furthermore, the Liberal leadership chose to deal itself out of the game by insisting that there could be no merit in any alternative to the status quo. This troubled some of the wiser heads in the Liberal Party, as Robert Taylor reported in The West Australian on May 7:

(Liberal upper house leader Norman) Moore knew that Independent Alan Cadby, who holds the balance of power on the legislation, might be interested in a Liberal version of electoral reform. Mr Moore confirmed in Parliament yesterday that he met Labor’s Electoral Affairs Minister Jim McGinty to discuss the possibility of another model emerging … Mr Moore’s plan increased the size of the Legislative Assembly by four MLAs, kept the status quo in the Legislative Council and guaranteed five Lower House seats in the remote Mining and Pastoral region but within their existing boundaries. The effect of his model was to save two Lower House country seats and reduce the number of country seats transferred to the city from six to four … Mr Moore took his plan to Tuesday’s Liberal party room meeting and by all accounts then listened in amazement as Mr Birney turned the debate into a test of his own leadership. Mr Birney refused to discuss the plan, arguing that the Liberals would lose all credibility in the bush if they voted for the legislation no matter what eventual form it took in the Parliament. An argument was also put that by voting for a one vote, one value model, the Liberals would merely provide the Nationals with ammunition at the next election.

Birney must take a dim view of his country constituents if he imagines they would prefer the purity of the impotent to a sober display of pragmatism in the face of the inevitable. The result of Birney’s idealism in pursuit of a low principle is that lower house representation in the non-metropolitan south-west, currently home to 14 Coalition and four Labor members, will indeed fall from 18 seats to 12. Moore’s plan promised a better outcome for all concerned. Proposals to increase the number of politicians are always vulnerable to populist rabble-rousing, but adding all four new members to the lower house would have been easier to sell as a necessary boost to country representation (for some historical perspective on this matter, check out the remarkable list of Western Australian electoral facts assembled by shy Perth blogger "Ross of Rockingham" – among other things, it tells us that the original electorate of Murchison had 24 voters when it was created in 1890). As for Birney’s reported concern regarding the threat from the Nationals at the distant 2009 election, this seems an insignificant consideration at the best of times, especially now that the affected area will have fewer seats.

The Western Australian Electoral Commission has prepared an indicative map showing how non-metropolitan electoral boundaries might look under the new system, which projects some rather quirky outcomes. Since the Mining and Pastoral region accounts for roughly three-quarters of the Western Australian land mass, the Queensland-style model largely replicates Labor’s original proposal to maintain the strength of the region’s representation while providing for one-vote one-value elsewhere. However, there will now be very large discrepancies within these five seats, which will have to be dramatically redrawn. The WAEC projects the existence of a vast and sparsely populated new electorate called Eyre, covering desert emptiness from north of Kalgoorlie to the South Australian/Northern Territory border, which will have an estimated 9215 voters – substantially fewer than the smallest electorate under the current system, the absurdly compact south-west seat of Leschenault (12,104 voters). It is almost double the estimated 18,179 voters in the projected new Mining and Pastoral seat of Pilbara, which will take in relatively populous mining areas on the north-west coast.

Much has been made of the fact that Matt Birney’s own electorate of Kalgoorlie, the only one of the five Mining and Pastoral seats not held by Labor, will be abolished under the new model. Where the current electorate of Kalgoorlie is based entirely within the city that bears its name, the projected new electorate of Dundas will absorb the whole city and surrounding areas as far east as the South Australian border. While this might look impressive on the map, Dundas will add a mere 961 new voters to the current enrolment of Kalgoorlie for an increase of about 7 per cent. Birney’s margin at the February election was 2024 votes. The new legislation only slightly aggravates an already existing problem, namely the Liberal leader’s relatively insecure hold on what was traditionally a Labor seat. Other changes projected by the WAEC for the Mining and Pastoral region include the expansion of the state’s northernmost seat of Kimberley, which will boost Labor’s margin through the recovery of remote territory it lost at the previous redistribution, and the creation of Murchison, which is similar to the seat of Ningaloo that existed before the previous redistribution. This area has normally been considered Labor territory, but Ningaloo was won narrowly by the Liberals at both elections of its short life (1996 and 2001). Murchison joins Dundas as the other seat with a conspicuously small enrolment, in this case an estimated 10,019 voters.

As far as the major parties are concerned, these changes are a case of swings and roundabouts. The 12 seats that will replace the existing 18 in the non-metropolitan south-west are quite a different matter. The six seats tipped for abolition include four held by the National Party (Greenough, Merredin, Wagin and Stirling) and two by the Liberals (Leschenault and Dawesville). That leaves the following survivors:

Moore: The WAEC projects that this electorate will include all of the existing electorate of Moore except for the Shire of Toodyay, along with the northern half of the exisiting electorate of Merredin (which does not include Merredin proper) and most of the area of the electorate of Greenough. Moore looms as an interesting contest between the Liberals, who won the seat for the first time in 1986, and the Nationals, who had a morale-boosting win in Greenough at the expense of a sitting Liberal member at the February election and for whom Merredin is a party stronghold. The National Party members for Merredin (Brendon Grylls) and Greenough (Grant Woodhams) could be left contemplating a run against the newly elected Liberal member, Gary Snook.

Geraldton: Labor’s Shane Hill narrowly won the existing seat of Geraldton at the past two elections, but would have to do very well to hold it now that it is set to be augmented by more than 8000 new rural voters from the solid conservative seat of Greenough. The weak performance by the National Party in Geraldton at the recent election suggests that it will fall to the Liberals, but the newly elected Nationals member for Greenough, Grant Woodhams, could prove popular enough to buck the trend if obliged to seek refuge here in 2009.

Avon: The WAEC projects that this electorate will be augmented by one shire from Moore, four from Merredin and two from Wagin. None of this suggests the seat will become any less safe for the National Party, particularly in light of the stature of the sitting member, party leader Max Trenorden.

Murray: Located on Perth’s expanding southern fringe, the WAEC envisions this electorate making up the numbers by absorbing the southern half of the metropolitan electorate of Peel. This would make a fairly safe Labor seat out of one in which Liberal newcomer Murray Cowper prevailed by 198 votes at the recent election.

Mandurah: The WAEC projects the abolition of the existing electorate of Dawesville to boost the numbers in Collie-Wellington and Mandurah, with Mandurah set to take in the populous area north of the Dawesville Channel. Since the Liberals won Dawesville by 4.1 per cent and Labor won Mandurah by 12.3 per cent, this will still be a Labor seat but with a softer margin. Given that Mandurah was held by the Liberals before 2001, it looms as one to watch for the next election.

Collie-Wellington: The existing seat is over quota so this will not need to expand too dramatically to meet the new enrolment requirements. It will absorb the southern part of Dawesville while losing the Shire of Dardanup to Capel. Labor’s Mick Murray did well to blow out his margin from 2.6 per cent to 9.3 per cent at the February election, but the new additions will probably make life harder for him.

Capel: Capel will lose its ungainly appendage south of Busselton along with a small area in the north to accommodate the expansion of Bunbury, while gaining the Shire of Dardanup from Collie-Wellington in the north and most of the Shire of Bridgetown-Greenbushes from Warren-Blackwood in the south-east. This will do nothing to change its status as a safe Liberal seat in which the National Party would need to field an exceptionally strong candidate to be competitive.

Bunbury: Bunbury will expand to incorporate those areas of the City of Bunbury it does not currently include, from Leschenault in the east and Capel in the south, plus a further coastal strip west of Bussell Highway in the south. These changes will strengthen Liberals newcomer John Castrilli’s precarious hold on the seat.

Vasse: Currently hugging the Cape Geographe coast, this electorate will dramatically expand in area to accommodate all of the Shire of Augusta-Margaret River (currently in Warren-Blackwood) and most of the Shire of Busselton (taking in an area currently in Capel). This will strengthen the Liberal hold on a seat that should be a safe seat for them, although they have survived strong challenges from the National Party and an independent at the past two elections.

Warren: This electorate is less closely related to Warren-Blackwood than the name suggests, as it will include most of the abolished seats of Wagin and Stirling. It looms as a tense struggle between the Nationals and the Liberals. Affected members are Liberal veteran Paul Omodei, member for Warren-Blackwood; the newly-elected Nationals member for Stirling, Terry Redman; and Terry Waldron, Nationals member for Wagin since 2001.

Albany: Labor have done well to win Albany at the last two elections, enjoying narrow wins in a seat that was held by the Liberals from 1974 to 2001. Albany will now expand to take in an area from the National Party seat of Stirling, and while many of the new voters will be from outer Albany suburbs where Labor performs well relative to the remainder of Stirling, Labor successes will probably become even rarer here in future.

Roe: Roe will absorb the south-eastern corner of Merredin, including Merredin itself, and the Shire of Dumbleyung from Wagin. This is all National Party territory, which makes the seat doubly interesting in light of Graham Jacobs’ success in winning the seat for the Liberals in February after the retirement of the sitting Nationals member.

If the results from the last election were replicated under these boundaries, the Coalition would have won nine of the 12 seats, of which between one and four would have been won by the Nationals (who won five seats on February 26). The WAEC has not prepared an indicative map for the metropolitan area, but the seats would presumably have been won in proportions similar to the actual result – about 29 from 42, compared with 24 from 34. Throw in a status quo result in Mining and Pastoral and assume that the two independents would have held their seats, and you have Labor on 36 seats and the Coalition on 21, compared with 32 and 23 in the current parliament. Given that the Nationals would have borne the brunt of the cut in Coalition numbers, it is hard to see why Birney was terrified at the thought of losing votes to the Nationals, but relaxed about losing seats to Labor.

The other half of the bargain is that the Legislative Council will gain two extra members. The state will still be divided into three metropolitan and three non-metropolitan regions, but these will have six members each instead of five or seven. The principle of rural vote weighting will thus endure in the upper house, which is the only place where it belongs. This change was insisted upon by the Greens despite the Poll Bludger’s conviction (as argued here) that it was not in the party’s own interest. Labor correctly concluded that they would be little affected and were happy to oblige.

Getting the upper house in order: part two

As outlined in the earlier instalment, Greens leader Bob Brown has a bill before parliament that will require above-the-line Senate voters to determine their order of preferences by numbering every box, rather than allowing their one chosen party to make their decisions for them. The Greens presume, no doubt correctly, that Labor voters required to engage their minds will baulk at favouring right-wing over left-wing minor parties regardless of what the how-to-vote card says. The Bob Brown model is similar to that introduced for the New South Wales Legislative Council at the 2003 election, with the significant difference that voters will need to number every box to record a formal vote. The New South Wales system of optional preferential voting requires voters to number only as many boxes as they desire, after which their vote exhausts. This would deal with the main problem of the Bob Brown model identified in my earlier post, namely the spike in the informal vote when voters are required to number too many boxes. Perhaps the Greens admire the mathematical elegance of the existing system, but it is just as likely that they suffer an aversion to optional preferential voting born of lower house state elections where they have lost the preferences that indifferent major party voters were once obliged to give them.

If this is so, their fears with respect to multi-member upper house elections might be misplaced. A paper by Antony Green for the New South Wales parliament informs us that 78.6 per cent of voters carried on their habit of marking one above-the-line box at the 2003 Legislative Council election, so that most votes exhausted rather than carry on to other parties. After nine Labor, seven Coalition and one Greens candidate were elected with full quotas, four seats still remained to be won. At that point, second Greens candidate Sylvia Hale led the field with 0.89 of a quota followed by Gordon Moyes of the CDP with 0.67, Labor’s tenth candidate Tony Catanzariti with 0.60 and John Tingle of the Shooters Party with 0.45. The high exhaustion rate meant that few preferences were distributed after this point and the four aforementioned candidates won their seats in the order listed. This underscores the point that unless a result is exceptionally close, optional preferential voting will deliver the final seats to whichever candidates get closest to a full quota on their own primary vote, in place of the current lottery created by monolithic preference transfers.

Had this happened in last year’s Senate election, the Greens would easily have defeated Family First in Victoria and Tasmania, where they respectively suffered defeat and an unnervingly narrow victory. The outcome would otherwise have been the same, with the Coalition still winning its four seats in Queensland, the major parties still monopolising New South Wales and South Australia, and the Greens still winning a second seat in Western Australia. In light of the primary vote figures, these were not unreasonable outcomes. It is harder to say what the outcome would have been under Brown’s proposed model of full above-the-line preferences, but it almost certainly would not have favoured the Greens more than optional preferential. Now that the party can no longer rely on tight preference arrangements with the Australian Democrats, all the Greens have to gain from compulsory preferential voting is continuing access to preferences from the major parties who invariably favour the Greens over each other. From a purely self-interested perspective, it is a debatable point as to whether this justifies the system for the Greens. Apart from the small surplus that major parties pass on when they exceed three quotas, the Greens can only access major party preferences if they perform well enough to overtake their third candidates, and performances of such strength would usually win them the seat under optional preferential voting. Nor should they overlook the possibility that Brown’s amendment will not produce hugely different outcomes to the current system, given the notoriously obedient attitude of major party voters towards how-to-vote cards.

Looking beyond the interests of the Greens, it becomes even harder to maintain an argument for compulsory over optional preferential voting, which would keep the informal vote to a minimum and give voters the widest range of options in directing their vote.