The WA Government has announced its reform proposals for the Legislative Council. The proposals are based on a report by the government appointed Ministerial Expert Committee on Electoral Reform.
You can find the Committee’s report at this link. The government has made some minor tweaks to the recommendations and will introduce legislation to Parliament tomorrow.
In summary the major features of the proposal are –
- Regions to be abolished and Legislative Councillors will be elected from a state-wide electorate.
- The number of MLCs will increase from 36 to 37.
- The current ballot paper with above and below the line voting will be retained.
- Group voting tickets will be abolished, ending party control over between-party preferences.
- Optional preferential voting as used for the NSW and SA Legislative Councils will be adopted. A voter is required to mark one square above-the-line with all further preferences optional. This compares with the Senate’s rules where six preferences are recommended.
- Below the line votes must mark
2220 preferences for the 37 vacancies, down from the current region voting rules where all squares must be numbered. There are no provisions to save below-the-line votes with insufficient or incorrect preferences. As a comparison, NSW requires 15 BTL preferences for 21 vacancies, South Australia 12 for 11 vacancies, and the Senate 12 for 6 vacancies at a half-Senate election, 12 for 12 at a double dissolution. - The quota for election will be 2.63%. Because of optional preferential voting and exhausted preferences, the last vacancies are likely to be filled by candidates with less than a quota of votes.
- There are a series of rules to toughen party registration, and there are increased minimum nominator rules for Independents seeking their own column on the ballot paper.
The main issues with the reform proposal are with the ballot paper, how to avoid a giant ballot paper that can’t be printed or counted. I went through some of the issues in a previous post.
What is proposed will produce almost exactly the same result as most forms of List proportional representation with a highest remainder allocation of the final seats. List PR would be much much cheaper to count, but is not viewed as acceptable in Australia as there are no preferences. So as in other state Legislative Councils, WA will use a more complex and more expensive counting system to achieve roughly the same result.
Let me run through a series of issues that I think arise out of the government’s proposed changes.
(A quick note on terms. ATL means ‘above the line’, a vote above the ballot paper’s dividing line using the group or party voting squares, and BTL means ‘below the line’ or a vote using the candidate voting squares.)
Loss of Regional Representation
The current electoral system that builds in substantial weights was always likely to be abolished once Labor got control of the Legislative Council. Currently, two regions with 10% of the population elect 33% of the Council, while Perth with 75% of the population elects only 50% of the Council. With the electoral weights defined by geography and land use, and with relative population trends in the two smallest regions pointing downwards, the existing regional electoral system could only become more malapportioned. I ran through the existing weightings in a previous post.
I had thought a region based system that approached one-vote one-value would be the most likely recommendation, but the difficulty is the distribution of the state’s population. A proposal for a four-region by nine-member model would create three regions covering Perth plus Mandurah and Dawesville, with the fourth region running from Bunbury to Kalgoorlie to Broome. If you adopted voter equality as a principle, and ended up with one region covering 95% of the state, why not just adopt a state-wide solution?
NSW and South Australia elect their Legislative Councils state-wide with no allowance for regions. While the Senate gives each state equal representation, there is no allowance for regional representation in each state. The proposed WA model is the same as used elsewhere in Australia.
Whole Chamber Elections not Staggered Terms
When WA adopted Legislative Council proportional representation in 1989, it abandoned staggered terms and elected the entire chamber at each election using electoral regions. Victoria adopted the same method on moving to a region based PR Legislative Council in 2006. Both states replaced two member upper house provinces that elected MLCs from single member electorates at alternate elections.
The NSW and SA Legislative Councils, and the Senate, use state-wide electorates with staggered terms. At every election, half of the chamber’s membership is elected.
In adopting state-wide election, the government has chosen not to return to staggered terms, increasing the Council from 36 to 37 members so an odd number can be elected.
The main advantage of staggered terms is that it makes it harder for a party to win control of both houses of parliament at one election. Staggered terms in a state-wide electorate means a government needs two big wins to control the upper house.
The state-wide electorate and a very low quota makes it slightly harder for a party to win control of the Council at one election. A party must get very close to 50% of the vote for a one-party majority. Parties rarely get 50% of the first preference vote, with the election of the McGowan government earlier this year with 60% of the vote a remarkable exception.
How Many Votes will show Preferences?
Recommending only 1 preferences above the line simplifies the counting process, increases the importance of first preferences over further preferences, and makes the result in members elected per party more likely to reflect each party’s share of the vote.
At Senate elections, the recommendation is that voters should give at least six preferences, though savings provisions allow votes with fewer preferences to be admitted to the count as formal votes. Only 3.5% of votes were single 1 ATL votes at the 2019 Senate election. With more than 95% of ballot papers having preferences, all Senate ballot papers have to be scanned to conduct the count. This greatly complicates the process.
In NSW from 2003 to 2015, only around 15% of ATL ballot papers went beyond a single preferences. With voters having experience with the new Senate system in 2016, the number of ATL votes with preferences increased to 27.6% at the 2019 NSW election. At the 2018 South Australian election, the first under the new system, 33.4% were ATL votes with preferences.
The high number of 1-only votes greatly simplifies the count in NSW and SA. All 1-only votes are counted by hand, and only BTL votes and ATL votes with preferences are scanned. The WA system should mirror the experience in NSW and SA, avoiding the cost and complication of the Senate’s requirement to scan 100% of ballot papers.
Parties can win Seats with less than the 2.6% Quota
While 2.6% is set as the quota for election, I expect that every election will see four or five members elected with less than a quota of votes. This view is based on experience from NSW Legislative Council elections.
The proposed WA system is based on the NSW model in use since 2003. NSW elects 21 members with a ballot paper that has fully optional preferential voting for ATL votes, or 15 preferences below. The quota for election is 4.55%. At all five elections held under this system, 17 members were elected with filled quotas in the initial stages of the count.
At each of these five elections, four members required the full distribution of preferences before they were declared elected. Of these 20 members, only one reached the quota, with 19 elected with less than a quota as the only candidates remaining in the count. In 2003 and 2007, the final four vacancies were filled by candidates from parties with the highest partial quotas at the start of the count. At the 2011, 2015 and 2019 elections, three vacancies could have been filled on initial partial quotas, but the identity of the final elected member at each election was determined by preferences changing the final order.
Despite the quota being 4.55%, at the 2019 NSW LC election, the final four vacancies went to candidates with less than a quota, the final Coalition candidate with 3.55% after preferences, final Labor candidate 3.39%, Animal Justice 2.74% and second One Nation candidate 2.66%. Animal Justice had started the count with 1.95%.
In 2015 the Animal Justice Party won the final seat on an even lower vote. It polled 1.78% (0.29 quotas) and reached 2.12% (0.47 quotas) after preferences. This was the lowest first preference vote for a successful candidate since the electoral system was introduced in 2003. It was lower than the 2.05% for the elected Shooters Party in 2003, equal to the unsuccessful Australian Democrats in 2007 (1.78%) and lower than for Pauline Hanson’s unsuccessful contests in 2003 (1.92%) and 2011 (2.41%). The after preferences percentage of 2.12% was also lower than the previous lowest successful count after preferences, 2.25% for the Shooters Party in 2003, and lower than Pauline Hanson’s unsuccessful after preferences count of 2.53% in 2011.
The more groups there are on the ballot paper, the more likely that the final candidates for a WA Legislative Council election might win the final seat with a figure well below 2.7%. My rough rule of thumb is that any candidate with around half a quota, that’s 1.31%, will stand a chance of winning one of the final seats. But in the end, getting elected from 1.3% is still a lot more han getting elected from 98 votes, as the Daylight Saving Party did in 2021.
Controlling Ballot Paper Size
The committee has made a number of proposals that toughen up party registration, and mimicking rules in NSW and SA, requires parties to be registered six months before the election. There are increases in deposits, and an end to single independents with their own column on the ballot papers. Any group of Independents wanting their own column on the ballot paper must stand three five candidates and each candidate requires 200 unique nominators. Until now, candidates could self-nominate in Western Australia.
All this is to try and avoid the giant ballot papers that have caused party registration rules to be tightened in NSW, SA and Federally, and led to increases in nomination deposits. As I wrote in the post I have already linked to, there are consequences to the electoral process from having too many candidates on a ballot paper. It’s all fine and dandy to let lots of candidates on a ballot paper, but not if you end up unable to print or count the ballot papers.
WA will have a big ballot paper because parties will have to list their candidates, and a list of 20 candidates will take space. There are ways to squeeze together candidate names, including dropping the repeated use of a party name for each candidate in a party’s group.
There are two magic numbers to keep in mind. The first is the metre wide rule. Electoral Commissions buy sufficient A0 paper stock months before the election and it is 841mm by 1189mm. Once you arrange layouts and take the paper to printing, the maximum ballot paper width is 1040mm. Even if you can get larger paper stock, it is unlikely you would be able to print ballot papers in the short time frame quantities required to run an election.
The second number is 297mm for ballot paper depth, the size of A4 paper. That’s the approximate limit on high speed ballot paper scanning. If you have to go to A3 flat bed scanning it will take a very long time to finish the count.
Controlling ballot paper size so that you can conduct a sensible election is the single biggest impediment to electing 37 members by PR-STV. I’ll have to leave further comment on how the government’s bill deals with these issues until the bill is tabled in parliament. The Electoral Reform Committee has given the government a range of options.
Are the reforms justified by Wilson Tucker (Daylight Saving Party) being elected from 98 votes?
In short, no. The government has been very cute in constantly referring to Tucker’s victory as justification. Tucker was elected because of group voting tickets. Without group voting tickets, the cascade of other party preferences that were sent Tucker’s way could never have occurred.
The malapportionment that meant Perth regions had a quota of more than 50,000 compared to only 7,010 in Mining and Pastoral Region meant that 98 votes in Mining and Pastoral carried more weight. But group voting tickets were much more important to Tucker’s success than the malapportionment.
The argument for ending the current Legislative Council electoral system is that the current system is unsustainable and all population trends indicate the distortion can only get worse, as it has been getting worse since first being introduced in 1989, and was made even worse by changes to the number of members elected per region in 2008.
An advantage of moving to a state wide electorate, or even one-vote one-value regions, is it removes the weird inequities caused by drawing weighted regional boundaries. At the 2019 election, a Legislative Council vote in Muchea carried four times the weight of a vote just down the road in Bullsbrook, the start of the defined metropolitan region. And a vote in Muchea on the edge of Perth carried 2.35 times the weight of a vote of Wellstead, 500km to the south-east in the electorate of Albany.
There are plenty of reasons for changing the Legislative Council’s regional structure, but Wilson Tucker’s victory is not one of them.
Why No BTL Saving Provisions?
I understand there will be no savings provisions for BTL votes that fail to complete 22 20 preferences. An argument in favour of this is that BTL votes needed more than 50 preferences at the 2019 election so 22 20 is an improvement. I think this is a bad arguement.
WA is the only state that retains a numbering error savings provision for lower house ballot papers. In WA, if a lower house ballot paper has all squares completed, but there is a missing or duplicate numbering error, the vote can remain in the count.
In most cases these votes have a first preference for a candidate whose preferences will not be counted and so the missing preferences do not matter. In a small number of cases, these votes are for a candidate who is excluded in the count, and so the ballot paper’s preferences may exhaust.
It makes no sense to allow a ballot paper with a numbering error to count in the lower house but not count in the upper house.
Even more so when you remember that because it is possible to weight exhausted votes out of the count using transfer values, there is even more reason to allow savings provision un the upper house than in the lower house.
Why No Country other than Australia would elect 36 members this way.
The WA Legislative Council’s electoral system is one member of the electoral system category known as Proportional Representation by Single Transferable Vote, or PR-STV for short. Other examples are the Senate system, Legislative Council electorate systems in other mainland states, Hare-Clark as used in Tasmania and the ACT, and systems used for elections in Ireland and Malta. It is also used for local government in New Zealand, Scotland, and in a small number of other countries.
As the name implies, PR-STV is a proportional system, and ‘STV’ means it uses preferences. Proportional representation is used in many different forms around the world, but PR-STV is the smallest sub-category of PR. Most PR systems try to match the proportion of votes a party receives to the proportion of representatives it elects. PR-STV complicates what could be a simple formula for allocating seats by allowing preferences between candidates and parties. With PR-STV, the proportionality is to the vote after distributing preferences, and it is preferences that complicate the count.
For reasons that are hard to explain, the more members you elect under PR-STV, the more proportional the outcome and the less role that preferences play. This is especially the case if you also use optional preferential voting.
By electing 37 members, the WALC electoral system will produce a very proportional outcome against party share of vote, and preferences may only impact on one or maybe two seats. But the count will still have to be conducted by getting all the ballot paper preferences into data files.
In Australia, voters are used to choosing between candidates of the same party, and are used to having their vote transfer to another candidate or party if their earlier choices were excluded from the count. From a lifetime of experience, Australian voters distrust an electoral system that does not allow choice of candidate and preferences.
In every other country, electing 36 members by proportional representation would be done by a form of List PR without preferences. The first decision to make would be what formula you use to allocate seats to party lists based on vote. The second would be whether the system would be open or closed list, that is whether a voter can choose a particular candidate of a party. The allocation formula gives parties a number of seats, and the seats are allocated to candidate in party list order (closed list) or according to voter choice (open list).
List PR is a simple tally of votes and allocation of who is elected. PR-STV must tally votes, then go through the complex distribution of preferences, and only at the very end of this process are all seats finally filled.
The WA Constitution states that members must be directly elected. The same provision is in the Commonwealth Constitution. The legal view is that ‘direct election’ prevents the initial allocation of seats to parties and from there the allocation of seats to candidates. This, and voter expectation of preferences and candidate choice, is why WA is about to elect 37 members by PR-STV rather than a simpler and less expensive List form of proportional representation.
Ballot paper design is going to be a challenge. The major parties will be nominating 20+ candidates, so a double-decker ballot seems unlikely. So, it may be the metre-long model to accommodate all the parties that will think they have a shot at a 2.6% quota. Another question would be whether independents can be accommodated. The Senate option of having them only below-the-line would be lip-service only, given the onerous 1-22 numbering requirement.
If we had a voting app, we wouldn’t have to worry about the length of the ballot paper. Each candidate’s name could be linked to a video, giving them 3min to state their policy, with a point my point list below.
COMMENT: There were 21 columns and more than 300 candidates on the 2019 NSW Legislative Council ballot paper. They have enough problem trying to randomise the order columns are displayed for voters to stop the order of presentation influencing voters. Magnify the problem many fold for presenting candidate videos in order.
Great idea WA… Perhaps it should be adopted federally, treat all of Australia as a single electorate, and let Melbourne and Sydney control the upper house, too.
COMMENT: That would require a referendum that would have to pass in every state.
Good Point Nathan. The Nationals will likely only get 1 member elected. Country people will just have to take what the city dishes out.
Well the city people have had to take what the country people dish out for along time now, about time the majority got their say, after all it’s meant to be a democracy.
I believe the federal Senate is a different league to the state Legislative Councils, in the fact that state borders are well defined since Federation 1901.
In contrast these ‘region’ boundaries are poorly defined. I read that Muchea and Bullsbrook are nearby localities, yet because they fall into 2 different LGA’s (Swan for Bullsbrook and Chittering for Muchea), one is considered regional and the other urban. In fact, since the recent WA federal redistribution, both localities now fall into the large rural and remote seat of Durack, so it shows how hard it is to define the border between urban and rural within each state.
COMMENT: Equal Senate representation for the six unequal states was required to achieve Federation and was written into the constitution. Moving away from equal Senate representation requires a triple lock referendum, passage requiring a national majority of votes, a majority of the vote in a majority of states, and requiring a majority of the vote in any state whose relative level of representation will be reduced.
WA’s electoral zones have always been determined by the WA Parliament, which means by politics. Before the current multi-member regions were introduced, the state had three zones, Metropolitan, Agricultural Mining and Pastoral, and North West. All three had different quotas for electorates and provinces drawn within each region. The current regional system was the first to have a clearly defined metropolitan boundary that didn’t require constant adjustment, though the growth of Mandurah suggests even the 1989 expanded metropolitan region is being overtaken by population growth.
I would have kept at least two upper house regions, metro and country, so long as they’re based on one vote, one value. Not liking the move to optional preferences either. Too close to first past the post, and designed to help the major parties. Combined with the federal changes, there seems to be a move towards favouring the major parties at the expense of minor parties and independents – not very democratic.
COMMENT: We have six decades of experience with the Senate between 1919 and 1983 of what happens with full preferential voting on large Senate ballot papers. You get high rates of informal voting, usually more than 10% prior to the 1984 changes. Group voting tickets were an answer to cut informal voting while keeping full preferential voting. But once the tickets became the plaything of preference whisperers, tickets were abandoned and the original proposal from 1984 has come back into play – some form of optional preferential voting.
If you are electing 37 members, then preferences will matter to at most only one or two of the vacancies. With such a low quota, proportionality will outweigh the impact of preferences, as has been the case in the NSW Legislative Council.
Thanks for your excellent analysis as always Antony. It seems to me this is a sensible reform in the context of WA. While I’d prefer multi-member electorates in the LC so that members represent a multi-member region fairly apportioned for number of voters it’s practically difficult to have this work in a state where the vast majority of people live in and around Perth.
Do you think this is the best model to achieve one-vote, one value, or could it have also been achieved while retaining regions?
COMMENT: If you insist on one-vote one-value, the whole state solution is the best fit. Regions would require you to match region boundaries to lower house electorates, and the large district allowance used in drawing lower house boundaries means remote seats have lower enrolments. So any matching of upper house regions to lower house seats would retain a small amount on non-metropolitan over-representation. The one problem with the whole state solution is trying to control ballot paper size, and drawing regions with close to one-vote one-value representation would have controlled the ballot paper depth problem by requiring fewer names to be printed in each column.
My preference would have been 4 regions each returning 9 members. The remote region covers 14 electorates. Each other covers 15 electorates. It isn’t pure electoral equality, but it would allow for regional representation.
There is no strict equality anyway in the Assembly with a large electorate allowance and single member electorates.
But a whole state electorate removes the malapportionment which is a mere tool to allow control by rural conservative parties. The regionality argument is a cover for the malapportionment argument.
Voters in the metropolitan Perth can now vote for the Nationals, which they were denied before.
I like voters deciding on their preferences. And voters have the power to direct more than their first preference. But at a threshold of 2.6%, if you get less than this, you are unlikely to merit being in the LC.
The LC will be a proper house of review allowing for a broad spectrum of representation.
If a week is a long time in politics, then 44 years is an eternity. Yet it was 1977 when Bob Hetherington, MLC, introduced a bill for PR, on the basis of one vote, one value, to the Legislative Council, which, inevitably, was unsuccessful. It is worthy of note that the driving force behind the adoption of this policy by the ALP State Conference, was the ALP’s Electoral Reform Committee, lead, with great energy and enthusiasm, by Graham Hawkes and Gordon Payne.
I was very pleased to note that both Messrs Hawkes and Payne made 2 submissions to the Committee, and that, after such a long and dedicated campaign, their vision is finally coming to fruition!
Doing away with rural weighting is a great step forward. The problem with having a potentially giant ballot, though, is that it runs the risk of turning the reformed system into something of a laughing stock – in much the way that people, when reminded of the first ACT Legislative Assembly election, always think immediately of the huge ballot paper that was used, full of obscure candidates. If that happens, it will play into the hands of those who favour the status quo, and will probably be used as an argument for reverting to the current system if there is a change of government. These sorts of reforms really have to work well first time round.
I doubt a future WA Coalition Government would re-introduce more than a small amount of rural weighting for the legislative Council (i.e. not significantly more than the current weighting in the Legislative Assembly), at least partly because the Liberals won`t want to give too much power to the Nats. Regions are potentially a different matter as the Nats have a reasonable likelihood of making a serious push for them.
Al large election results in ballot papers that are too large and preselections that are too centralised. Fortunately Victoria is unlikely to follow the at large trend as the number of regions and number of members per region are referendum protected.
How will vacancies be filled? Dose section 73 require that they have to use a count back of by election?
COMMENT: As currently happens, casual vacancies are filled by recount.
My cousins live in Bullsbrook, and the most annoying thing about the defunct rules is that they live on a farm. In fact, their farm is the border between Bullsbrook and Chittering, meaning that their next-door neighbours’ vote is worth four times theirs at the moment. This pain is also amplified by Moore getting to be smaller because of the LDA, which conveniently has been ignored. My cousins’ district, Swan Hills, is already over quota.
Does the new lack of regional boundaries mean LA districts can cross the edge of the metro boundary? The huge increase in voters in Baldivis means Warnbro would have to push up, meaning Mandurah could potentially move north? Also there’s a bunch of places, like non-metro Bullsbrook and beyond, who would best be in a regional district
COMMENT: It will be interesting to see how the metropolitan boundary is dealt with. I think crossing the boundary is most likely on the south with Serpentine-Jarrahdale Shire and the area just north of Mandurah.
A 2.63% quota will produce enormous ballot papers and just create more confusion. There are two much better ways of doing electoral reform in WA.
The second best option is to reduce the number of LC seats by one to 35 to create 7 * 5 member electorates (quota 16.66%) for 5 metro electorates and 2 country electorates.
The best option (I know this won’t happen) is to abolish the LC altogether, but to increase the number of LA MP’s accordingly, for a LA of 95 members. Create 19 multi member electorates, each electing 5 members. This is the ACT model, which would work well for all states and territories and would put the issue of major electoral reform to one side for a very long time.
COMMENT: Reducing the number of MLCs requires a referendum which is why the option has never been on the table.
I would like the multi member option, unicameral legislature. It would partially resolve the issue of regional representation as some of these multi member districts could have a greater weighting applied, similar to the LDA currently used.
I’ve just run an exercise using 2019 NSW LC preference data (the full data file including BTL) to see what would happen if it was used to fill 37 vacancies instead of 21.
Obviously, the parties (likely including the number of them) and vote shares will be different in WA, but my main interest was in the patterns, particularly in how the end of the count might play out – with candidates being elected on less than a quota due to exhausted votes.
In the actual NSW count (for 21 vacancies):
– As Antony notes, there were 17 candidates elected with a full quota before exclusions started (7 LIB/NAT, 6 ALP, 2 GRN, 1 PHON, 1 SFF). No further candidates reached quota on preferences from excluded candidates.
– The count ended when the field was reduced from five to four, filling the remaining vacancies. The last excluded candidate’s votes were not distributed..
– The four elected candidates (LIB/NAT #8, ALP #7, Animal Justice, PHON #2) had accumulated totals representing 3.55% to 2.66% of formal votes (0.78 to 0.59 quotas). The final excluded candidate (from CDP–Nile) had 2.54% (0.56 quotas)
– 349,365 votes (7.84% of the formal vote, 1.72 quotas worth) were exhausted.
In re-running the count for 37 vacancies:
– There were 31 candidates elected with a full quota before exclusions started (13 LIB/NAT, 11 ALP, 3 GRN, 2 PHON, 2 SFF). Again, no further candidates reached quota on preferences from excluded candidates
– The count ended when the field was reduced from seven to six, filling the remaining vacancies. The last excluded candidate’s votes were not distributed.
– The six elected candidates (CDP–Nile, LDP, Greens #4, Animal Justice, Keep Sydney Open, PHON #3) had accumulated totals representing 2.53% to 1.68% of formal votes (0.96 to 0.64 quotas). The final excluded candidate (Sustainable Australia) had 1.62% (0.62 quotas).
– 175,327 votes (3.94% of the formal vote, 1.50 quotas worth) were exhausted.
The main surprise for me was that the exhausted votes represented a broadly similar number of quotas, rather than a similar number of votes. But it shouldn’t have been a surprise. The big exhaustions happen in the final few exclusions when parties with 0.3 to 0.5 quotas are being excluded, and 60% or more of that is being exhausted. In a 37-vacancy count, those final few exclusions happen at a similar level in terms of quotas, but that amounts to less votes.
If WA ends up with a few more groups than NSW, it wouldn’t be a surprise to see a slightly higher exhaust (vote % and quotas) than in this exercise, because a few of those extra groups will gather a significant fraction of a quota, of which most will then exhaust on their exclusion.