As a follow-up to Thursday’s blog post, I wanted to look into the question of what would actually happen if the territories were granted a third or fourth Senate seat.
It’s easy enough to calculate the primary vote quotas in past elections. It’s relatively easy to make guesses about who would have won those extra seats (although not without room for debate), and thus to calculate the net impact on the left-right balance in the Senate.
It is clear that for most of the history of territorial Senate representation, the addition of extra seats would either have no impact or favour the left by net two seats.
Then finally I try to tackle a more complex question: does this expanded Senate do a better or worse job of reflecting how people vote?
This entire article assumes that the territories would continue to elect their entire Senate delegation at once every three years.
Let’s start by looking at voting patterns in the two territories. First up, ACT.
On this chart, I’ve drawn lines to indicate the new quotas. If the territory elects four senators, the quota would be 20% (solid line). If the territory elects three senators, the quota would be 25% (dashed line).
As with all of Australia, we’ve seen the major party vote decline. Until 2022, this didn’t make any difference to the seat results, but the vote trend was happening regardless.
Until 2010, Labor was steadily over two 20% quotas. The Coalition was generally below that number but relatively close to it.
The Greens primary vote from 2004 to 2019 was always at least 16%, which would be enough to win a seat if the territory elected four. In 2022, the Greens vote dropped substantially with David Pocock’s candidacy. Presumably if there were already four seats there would be an incumbent Greens senator and it’s unlikely that Pocock would have run.
If there were just three seats in the ACT, it is clear that the broad “left” would usually win two of them. In recent years that second left seat would usually have gone to the Greens, until Pocock in 2022. Prior to 2004, it would have usually been a second Labor seat although the Democrats would have won it twice.
Then let’s look at the primary votes in the NT:
The two major parties have always been in a position to clearly win one seat each in a three-seat election, and until the mid-2000s both major parties were usually in a position to clear 40% each.
Since 2010, the Labor primary vote dropped below 40%, but usually with more than the Greens. The CLP primary vote has dropped since 2016, as the vote for other parties has increased.
By my estimate, the four seats would always split 2-2 between Labor and the CLP, and even if the major party vote drops it’s likely the NT would continue to elect two each from the left and the right, unless there was a substantial swing in one way or the other.
On the other hand, if the NT elected three senators, that third seat would be more marginal and could flip back and forth between left and right, although generally that seat is more likely to go to the left – usually Labor, but the Greens may have won it at high points.
Next up, these charts show my estimates of who would have won each race in 3-seat and 4-seat scenarios.
Of course there’s a bit of guesswork here. I’ve started by awarding seats on full quotas, and usually then giving the final seats to the leading candidates, but I do look at what other preferences are available – if there’s a lot of Greens preferences, that might push Labor into the win over a slightly higher-polling Liberal.
In the 3-seat scenario, there is no option of an even split between left and right, which has tended to be the typical result in territory and state Senate elections over the last 30 years.
For the ACT, the left has won 2 out of 3 at every election since 1977. In early years Labor would have won two. In more recent years it was the Greens and then David Pocock.
For the NT, the left would have won 2 out of 3 at every election since 1987, except for the 2013 election.
Having four seats does create the option of a 2-2 tie, and it does happen, but the ACT is generally too progressive for that outcome at a majority of elections.
The NT has split 2-2 at every election since 1975 – Labor winning two at every election, and the CLP winning two at every election except 1987, when a separate Nationals party would have won the second right-wing seat.
For the ACT, the Liberal Party would have managed to win two seats at seven out of 18 elections, but not once since 2007. The Democrats would have won a seat off the Liberals on a number of occasions between 1984 and 2001, and the Greens from 2007 until 2022, when the seat would have gone to Pocock instead.
It’s worth noting that a 3-1 split in the ACT would be consistent with the result in 2022, when the left won two out of two seats. So whether a 3-1 split would be considered to be advantageous to the left partly depends on what you expect would happen in the future under the status quo 2-seat scenario. I suspect that the left would find it easier to win a 3-1 split than it currently is to win a 2-0 split, but both will be quite possible to achieve.
So for the next chart, I have compared my expected results to the actual results in both scenarios, to determine how much the left was advantaged at each election.
The only election where the right would have been favoured was in a 3-seat scenario in 1975, when the Coalition would have gained both extra seats.
Out of the 14 elections from 1983 until 2019, the 3-seat scenario would have advantaged the left at eleven elections, and the 4-seat scenario at ten elections.
Neither scenario would have shifted the left-right balance in 2022. This is because the actual result was 3 left and 1 right senators, and the alternative scenarios would have been 4-2 or 5-3.
So you can’t argue with the assertion that the left would benefit from an expanded territorial Senate representation. Perhaps not at every election, but over time it would often help the left and never help the right.
The next question I had is: would this make Senate elections fairer, or less fair? Generally the default result in Senate elections is an even split between the left and right 3-3. The ACT would be different, since it leans quite strongly to the left. The 4-seat scenario would largely neutralise any left bias in the NT, but would increase the impact of the left’s advantage in the ACT, since they would be winning 3-1 rather than 2-1.
Firstly, I decided to calculate the Gallagher index of disproportionality on the actual Senate results since 1990, and on the alternative results in 3-seat or 4-seat scenarios. This is a calculation of proportionality of the overall national result, comparing seat results and vote results per party. I’ve also included the House index as a comparison. Bear in mind that a higher score indicates a less proportional result.
Overall the index doesn’t shift much, and it isn’t consistently more or less proportional in any scenario.
The 3-seat scenario performs noticeably better in 1993, 1996 and 2004, when Labor would have gained two extra seats (or one Labor and one Greens in 2004).
All of the scenarios have been less proportional in the last decade, but the scenarios are all very similar in recent years. In 2022, the 3-seat scenario would have been more proportional than the actual 2-seat result or the 4-seat scenario.
The final chart in this long blog post compares the proportion of the vote and the proportion of the seats for each of the five main blocs who have won Senate seats over the last three decades.
Now we know that there is also a substantial vote for a wide variety of other parties which have often had no luck in winning seats in part due to the relatively low magnitude in each state, so you’d generally expect these parties to all overperform on average.
Overall it looks like the Coalition has steadily overperformed more, while Labor has generally overperformed by a smaller amount, and actually won less seats than their share of the vote in 1990, 1993, 1996 and (barely) 2013.
The Greens used to be under-represented, but since 2010 (when they started winning a seat in every state) they have usually been slightly over-represented. Generally the Greens can win Senate seats with less than a quota, giving them 1/6th of all state senators with less than 1/6th of the vote.
The first iteration of One Nation was severely under-represented in 1998 and 2001. Since their return they’ve usually been slightly under-represented, except at the 2016 double dissolution when they over-performed.
Labor and the Greens collectively outperformed Coalition and One Nation in 2016, 2019 and 2022, perhaps indicating a new trend.
Overall the evidence is unconvincing in either direction. An expansion in territory Senate numbers would often produce a boost for the left relative to the status quo, but there is not clear evidence that this would make Senate election results less proportional to how people actually vote.