This post is a detailed look at Western Australia’s zonal electoral system ahead of a major review of how the Legislative Council is elected.
The malapportionment that applied to lower house boundaries was abolished with the introduction of one-vote one-value electoral boundaries at the 2008 election.
But malapportionment remains for the Legislative Council, and was in fact made worse by changes to region representation in 2008.
The bias in the electoral system against Perth has drifted out from 2.80-to-1 when the current system was adopted in 1989, to 3.07-to-1 in 2021.
But this hides another developing bias, an increased weighting against voters in South West Region. Where in 1989 average enrolment per MLC in the three non-metropolitan regions was equal, by the 2021 election, average enrolment in Agricultural Region and Mining and Pastoral Region had blown out to a ratio of 2.81-to-1 against voters in South West Region.
Western Australia’s current electoral regions defined by land usage rather than population is unsustainable given demographic trends.
The McGowan government has appointed a Ministerial Expert Committee chaired by QC and former WA Governor Malcolm McCusker to examine reform options for the Legislative Council. The existing malapportionment of the Legislative Council’s electoral system is one amongst several issues it will be addressing. (You can find details of the Committee here)
In this post I set out in detail the problems with the current malapportionment. In future posts I’ll return to other issues such as whether Western Australia should follow the Commonwealth, New South Wales and South Australia by abolishing group voting tickets for elections to the upper house.
Introduction
Of Australia’s 15 parliamentary chambers, only two are structured so that geography determines the weight and power of a vote.
The most unequal chamber, the Commonwealth Senate, gives equal representation to six unequal states. New South Wales elects the same number of Senators as Tasmania despite having 15 times the population.
The unequal Senate is a legacy of the compromise between the colonies required to achieve Federation. State representation in the House based on population was counterbalanced by the adoption of equal state representation in the Senate.
No noble origin story surrounds the nation’s other unequal chamber, the Western Australian Legislative Council. It is the last extant example of rural over-representation based on a defined zonal electoral system. Such systems were abolished for all other Australian parliaments by the 1990s.
Rural and remote parts of Western Australia have always been over-represented compared to Perth. Various formulas and zones have been applied over the decades, with the current regional structure in place since the 1989 election.
In the 1980s, the Burke government wanted to introduce proportional representation for the Legislative Council, preferably in conjunction with one-vote one-value boundaries for at least the lower house. In the end a deal was struck between the Labor and National Parties at the expense of the Liberal Party.
The deal established a two-to-one rural weighting for the lower house, and three-to-one for the upper house. Both were slight reductions on the rural over-representation then in existence.
More importantly, it established a new metropolitan boundary that effectively moved four seats to the metropolitan area in a way that helped Labor. It also resolved the situation where Labor’s two remote electorates, Kimberley and Pilbara, had four times the population of Liberal held Gascoyne and Murchison-Eyre.
(The enrolment discrepancy in remote district was in part due to the enfranchisement and enrolment of indigenous voters, but that’s a story for elsewhere.)
The map below shows the boundaries of the regions as they were in 1994 and as they largely remain today.
For the lower house, a two quota system was adopted based on 34 Perth seats and 23 non-metropolitan seats. This replaced a three zonal system, for the first time established a clear boundary for Perth, and removed the four very lower quota north-west seats.
While there remained a difference between Perth and the rest of the state, there was no differentiation between different parts of regional Western Australia in the lower house. It was a different story in the Legislative Council.
In the lower house, the deal allowed Labor to retain several low-enrolment regional city seats that would have had their Labor majorities swamped by the introduction of one-vote one-value boundaries. The new metropolitan boundary converted several low quota Liberal seats into high quota urban Liberal seats.
For the Nationals, the deal created a low quota Agricultural Region covering its traditional wheatbelt heartland and electing five members to the Legislative Council.
In 1989 the low-quota 5-member Mining and Pastoral Region also looked good for Labor representation. Subsequent changes to the structure of the state’s remote mining workforce, ending the constructions of mining towns and expanding fly-in fly-out employment, undermined Labor’s strength in the region.
On the new lower house boundaries, Labor narrowly won the 1989 election but was defeated in 1993. On returning to office in 2001, Labor tried without success to implement one-vote one-value reforms. It achieved its goal when a brief constitutional window of opportunity opened after the 2005 election.
But only for the lower house.
Changes to the Legislative Council’s electoral system implemented at the 2008 election were relatively minor. One was a switch from regions of five or seven members to all regions electing six MLCs. For technical reasons, that decreased chances of Labor achieving a Legislative Council majority, though the scale of the McGowan government’s victory in March 2021 spectacularly overcome the impediment, Labor winning a Legislative Council majority for the first time in the state’s history.
Importantly, the change in 2008 did not alter the weighting of votes between Perth and Non-Metropolitan regions. When the zonal system was introduced in 1989, elector weighting was 2.80-to-1 in favour of non-metropolitan voters in the Legislative Council. That has drifted upwards to be 3.07-to-1 at the 2021 election.
But in a way not properly understood at the time, the 2008 changes altered the relative weight of votes between the non-metropolitan regions. For the Legislative Council, votes were de-valued in South West Region and increased in value for Agricultural Region and Mining and Pastoral Region.
In 1989, Agricultural Region and Mining and Pastoral Region together contained 14.7% of the state’s electors and elected 29.4% of the Legislative Council. By 2021, the number of electors in the two regions has declined to represent 10.2% of state enrolment, but the two regions now elect 33.3% of the Legislative Council.
And where in 1989 a vote in Agricultural Region or Mining and Pastoral region was equal to a vote in South West Region, in 2021 a vote in these regions carried 2.81 times the weight of a vote in South West.
To explain how all this has come about, let me start by comparing representation per region when the current system was put in place in 1989, with the position in 2021 post the changes introduce in 2008.
Representation by Region – Comparing 1989 and 2021
The table below shows the number of lower house seats in each region and the number of Members of the Legislative Council (MLCs) elected from each region. It compares the numbers at the 1989 and 2021 elections.
| Assembly | Council | |||
|---|---|---|---|---|
| Region | 1989 | 2021 | 1989 | 2021 |
| East Metropolitan | 10 | 14 | 5 | 6 |
| North Metropolitan | 14 | 14 | 7 | 6 |
| South Metropolitan | 10 | 15 | 5 | 6 |
| Total Perth | 34 | 43 | 17 | 18 |
| Agricultural | 7 | 4 | 5 | 6 |
| Mining and Pastoral | 6 | 4 | 5 | 6 |
| South West | 10 | 8 | 7 | 6 |
| Non-Metropolitan | 23 | 16 | 17 | 18 |
| Western Australia | 57 | 59 | 34 | 36 |
By allocating 34 seats to Perth and 23 to the rest of the state, the 1989 reforms established two enrolment quotas for drawing lower house electoral boundaries. The metropolitan quota was the population of Perth divided by 34 with all seats in Perth to be drawn within the permitted variation from quota. The Non-Metropolitan quota was the non-metropolitan population divided by 23. This established a vote weighting of about 2-to-1 in favour of non-metropolitan voters.
The ratios of MLCs to lower house members in 1989 were 2-to-1 in Perth and 1.35-to-1 in non-metropolitan regions. In 2021 those ratios were 2.4-to-1 in Perth and 1-to-0.9 in the rest of the state, there now being more non-metropolitan MLCs than non-metropolitan lower house members.
The old rules applying a single non-metropolitan quota allowed Redistribution Commissioners to move lower house seats between regions. The 2003 redistribution abolished a seat in Mining and Pastoral Region and created a new seat in South West Region. There was also some scope to vary boundaries between regions. The first one-vote one-value redistribution in 2007 moved Esperance from Agricultural to Mining and Pastoral Region, a move reversed by the 2015 redistribution.
The one-vote one-value reforms implemented by the 2007 redistribution established a single state-wide quota. The size of the lower house was increased from 57 to 59, and the first redistribution using the new rules abolished six non-metropolitan seats and created eight new seats in Perth. At the first election using one-vote one-value boundaries in 2008, there were 42 Perth seats and 17 non-metropolitan seats. The 2015 redistribution abolished another non-metropolitan seat and added a 15th seat to South Metropolitan Region, creating the current balance of 43 Perth seats and 16 non-metropolitan.
The new rules copied a provision from the Queensland Electoral Act establishing a single permitted variation from the one-vote one-value quota. Districts greater than 100,000 square kilometres in area are granted a Large District Allowance (LDA), a number of ‘notional’ electors equal to 1.5% of the area of the electorate in square kilometres. The LDA is added to the enrolment, and the adjusted enrolment is permitted to be up to 20% below the quota, where for all other electorates a 10% lower bound applies.
For example, the 2019 redistribution established an enrolment quota of 27,573. The Mining and Pastoral Region electorate of North West Central, by far the state’s largest in area, had only 10,904 voters. This was augmented by an LDA of 12,275 notional electors, giving the electorate a notional enrolment of 23,179, 15.9% under quota but permitted by the lower 20% variation allowed for districts with an LDA.
At the 2019 redistribution, all four of the Mining and Pastoral Region electorates were granted a significant LDA. Two Agricultural Region seats, Central Wheatbelt and Roe, were also given a small LDA.
One-vote one-value boundaries have allowed the transfer of seats to Perth, something the former legislated numbers did not. Both old and new systems allowed the transfer of lower house seats between non-metropolitan regions, usually by moving seats to the more rapidly growing south-west region. One-vote one-value boundaries have in fact slowed down these transfers by using the LDA to allow large remote districts to be drawn below quota.
All these rules on quotas and regions apply to the lower house, but they do not impact on representation in the upper house, the Legislative Council. Non-metropolitan regions may have lost six lower house seats to Perth and changed the state’s balance of representation, but there was no change to the balance between Perth and the rest of the state in upper house representation.
Representation in the Legislative Council is not defined by population. It is defined by land use, and given the lowest enrolment regions are also the slowest growing, a definition of zones based on land use steadily increases the weight of votes cast in those regions.
What has not been properly understood is that the 2008 change to the number of members elected per region introduced another tilt to the system, by weighting the system against voters in South West Region. There are now voters on the edge of Perth with significantly greater Legislative Council representation than voters further from Perth along the state’s south coast.
The debate that is about to start on representation in the Legislative Council can be viewed as just the latest chapter in a century long debate on how to insure proper representation for voters in rural and regional Western Australia.
There will be change, as the last compromise implemented in the 1980s has been undermined by two trends in the three decades since.
The first is the on-going decline in the proportion of electors that live in Agricultural Region and Mining and Pastoral Region while the two regions retain the same upper house representation.
The second is the rapid growth in population in South West Region. When the current zonal system was adopted, representation in South West was equal in weight to the other non-metropolitan regions. That is no longer the case, partly due to population growth, but also due to the change in the number of MLCs per region introduced in 2008.
Average Enrolment and Vote Weighting in the Lower House
The first graph below shows the average electoral enrolment per lower house electorate at WA election from 1989 to 2021.
The solid black line is the combined value for the three metropolitan regions, the dotted black line combines the three non-metropolitan regions. The coloured lines show the separate averages for each of the non-metroplitan regions.
The huge changes in value between the 2005 and 2008 elections represents the change brought about by the introduction of one-vote one-value electoral boundaries.
One-vote one-value boundaries saw the average extra electors per seat in Perth fall dramatically from 13,287 versus the rest of the state in 2005, to just 3,634 in 2008. Both South West and Agricultural Regions saw a large increase in average enrolment. At the 2021 election, seats in South West Region effectively had the same average enrolment as Perth seats.
The orange line for Mining and Pastoral Region did not converge in 2008 because the Large District Allowance (LDA) permitted seats in the region to be drawn with fewer electors. The allocation of small LDAs to Agricultural Region seats at the 2019 redistribution produced a dip in the green line of average enrolments for the region at the 2021 election.
All of these lines slope upward representing growth in the state’s population. To remove population growth from the analysis, the enrolment data needs to be converted into a measure of vote weights.
To explain simply, say 80,000 electors return one MP for each metropolitan seat, but an equivalent Country MP takes only 40,000 electors. That means it takes only half as many country voters to elect an MP, or inverting the value, it means a country elector has twice the weight of a city elector.
This calculation of weights is used to convert the enrolment data in the first graph into a graph of elector weights by region shown below. The colouring of regions is the same as in the first graph.
The weights in other regions are measured against Perth, so Perth electors are defined as having a value of one. All other lines show relative elector weights compared to Perth. The graph shows the variation in elector weights from region to region after removing population growth.
Clearly the elector weights in non-metropolitan regions converge on Perth with the introduction of one-vote one-value boundaries. Mining and Pastoral Region does not converge due to the impact of the Large District Allowance. But this second graph makes clearer that elector weight in the region was reduced by one-vote one value reforms. Note that the LDA lifts the weight for Agricultural Region in 2021.
On another note, the dips in Mining and Pastoral elector weighting in 2005 and 2017 represent the abolition of a seat at the preceeding redistribution. In 2005 that seat was transferred to South West, increasing that regions weight. In 2017 the abolition was combined with transferring Esperance to Agricultural Region, impacting the weights on both regions.
It should also be clarified that these graphs are of elector weights, not vote weights. Turnout is much lower in Mining and Pastoral Region, so the vote weight of the region’s voters compared to Perth is closer to eight. However, electoral boundaries are drawn based on enrolment data, so elector weight is the more appropriate measure.
Average Enrolment and Vote Weighting in the Upper House
The two graphs below repeat the above analysis but based on the number of electors per region compared to the number of MLCs to be elected. The analysis produces very different shaped graphs.
The first graph shows the average number of electors per MLC. As in the above graph for the lower house, the solid black line is a combined value for the three Perth regions, and the dotted black line a combined value for non-metropolitan regions. The three coloured lines break out the three non-metroplitan regions.
In 1989 the average enrolment per MLC in Perth was 41,506 compared to 14,844 for non-metropolitan MLC, a ratio of 2.80-to-1 electors. There are two very important observations to make about the shape of these lines.
- (1) The three non-metropolitan regions had roughly equal representation in 1989 but they have diverged since.
- (1) The change in MLCs per region decreased the weight of South West Region votes by reducing the number of MLCs from seven to six, but increased the weight of Agricultural Region votes by increasing the number of MLCs from five to six. The change in Mining and Pastoral was masked by the transfer of Esperance out of Agricultural Region, a move reversed ahead of the 2017 election.
The next graph converts enrolment to vote weights and shows more clearly how changing enrolment numbers and the change in the number of MLCs elected have really impacted on representation.
The same two points highlighted above are even more clear once the data is converted to elector weights. The weight of votes in South West are declining compared to Perth, down from 2.74 in 1989 to 1.78 in 2021. But the other two regions are going in the opposite direction, Agricultural Region up from 2.50 in 1989 to 4.19 in 2021, and Mining and Pastoral from 3.27 in 1989 to 6.22 in 2021.
The vote weights show even more starkly how the change in MLC numbers per region in 2008 weighted the system against South West Region and in favour of the other two regions.
The final graph below re-calculates weights based on average enrolment in South West Region, that is setting South West at value one. Where Agricultural Region enrolment per MLC was 0.91 compared to South West in 1989, by 2021 it had increased to 2.35. In Mining and Pastoral it has increased from 1.19 to 3.49. Combining the two smaller regions, enrolment weighting compared to South West Region has drifted flom equality when the current zonal system was adopted in 1989 to a weighting of 2.81-to-1 against South West Region in 2021.
Conclusion
The political debate about over-representation for non-metropolitan areas is as old as Western Australia itself. From time to time the enrolment weighting against Perth has become too extreme, resulting in adjustment to bring them back within an acceptable level.
The last change that systematically adjusted both houses took place in 1989, in conjunction with introducing proportional representation for the Legislative Council and ending the staggered terms of MLCs.
The introduction of one-vote one-value electoral boundaries ahead of the 2008 election ended the debate on weighting votes in the lower house. But associated changes to the Legislative Council have had the perverse effect of introducing a new bias into the electoral system that over time is disadvantaging voters in rapidly growing South West Region.
The McGowan government has appointed a Ministerial Expert Committee chaired by QC and former WA Governor Malcolm McCusker to examine reform options for the Legislative Council. The existing malapportionment of the Legislative Council’s electoral system is one amongst several issues it will be addressing. (You can find details of the Committee here)
One of the advantages of one-vote one-value is its simplicity. If you vary from one-vote one-value, you end up trying to define an appropriate ratio. What should it be? 1.2-to-1? 1.5-to-1? 2-to-1? 3-to-1?
As I have outlined in this post, the weighting ratio adopted for the Legislative Council in 1989 has drifted out from 2.80-to-1 to 3.07-to-1. More importantly, the equality of representation for the non-metropolitan regions adopted in 1989 has fallen apart, representation of the two-smaller regions now registering at 2.81-to-1 against South West Region.
It may be that some degree of non-metropolitan over-representation may be retained, as it has been in the Legislative Assembly.
But the current regions based on land usage will have to go. Given population trends, any attempt to base representation based on land use rather than population will be undermined by demography.
Any over-representation for non-metropolitan areas should be formula based, like the Large District Allowance used for drawing lower house boundaries.
Many options will be put to the Committee for its consideration. But the problems with the current system I’ve outlined in this post suggest any new system will be very different from the zone based system adopted in 1989 and modified haphazardly since.