Will better voting methods end the two-party system and make smaller parties truly viable? Australia offers an interesting case study. It has a bicameral legislature, and both chambers are elected via Ranked Choice Voting. But it uses a different form of RCV for each chamber: the House is elected via Instant Runoff Voting (IRV aka single-winner RCV), and the Senate via Single Transferable Vote (STV aka Proportional RCV). This provides a natural experiment on the different forms of RCV; when we see substantially different results in the Senate than in the House, we can be confident that it’s due to the difference in voting methods.
Australian Politics 101
House members are elected from single-winner districts (called “electorates”) with approximately equal populations and serve 3-year terms. As for the Senate, each state (New South Wales, Victoria, Queensland, Western Australia, South Australia, and Tasmania) and territory (Australian Capital Territory and Northern Territory) elects its senators in at-large elections using STV. Every state is represented by 12 senators regardless of population (Tasmania gets about 15 times as much representation in the Senate per resident as New South Wales), and each territory gets 2 senators. In most elections, half of the Senate seats in each state are up for grabs and the winners will serve 6-year terms (so there are six-winner elections in each state). However, the Senate is sometimes dissolved, resulting in an election in which every single seat is up for grabs (so there is a twelve-winner election in each state). Senators elected from territories serve three-year terms and are elected in two-winner elections.
Australia has two major parties: Labor on the Left, and the Liberal party on the Right. (The Right is superficially split between the Liberal party and the National party, but they don’t really compete against each other so the National party is effectively just part of the Liberal party. Warren Smith and Jan Kok go into this here, but I’ll just treat the Liberal/National coalition as a single party.) As for minor parties, the largest by far is the Greens, who received a bit over 10% of first preferences in both 2016 and 2019. (That is to say, one of their candidates was ranked first on over 10% of ballots.)
The most basic observation is that no third party won more than a single seat in the 151-member House in 2019, but the Greens won 6 of the 40 seats that were up for grabs in the Senate. Australia’s House has a two-party system, but not its Senate. The difference in representation is clearly due to having proportional representation in the Senate but not the House; the Greens won 10.40% of first preferences in the House and 10.19% in the Senate, so there was no significant difference in voter preference between the two chambers.
Could any single-winner voting method end the two-party system?
It seems clear that Australian third parties would win far more seats in the House if the House switched from IRV to STV (with multi-member districts). Presumably, any other proportional voting method would work as well. Might switching from IRV to another single-winner voting method also end the two-party system in Australia’s House? Do the weaknesses of IRV cause Australian third parties to have less success than they would under Approval, Score, or STAR voting?
One weakness of IRV is its vulnerability to the “center squeeze”. Suppose there is a centrist third party whose candidate is the first choice of 20% of voters, the Labor candidate is the first choice of 40%, and the Liberal candidate is the first choice of the final 40%. Moreover, suppose Labor voters heavily prefer this third party to the Liberals and that Liberal voters heavily prefer this candidate to Labor. Such a third-party candidate would defeat either the Labor or the Liberal candidate head-to-head — but gets eliminated in the next-to-last round under IRV due to not being the first choice of enough voters.
Alternatively, an extremist third party might be harmed by a fear of the center squeeze. Say this party is on the far Left, and virtually everyone in it prefers Labor to the Liberals. Labor voters would be more divided; some would prefer this third party to the Liberals, but others would feel the third party goes too far and prefer the Liberals. If such a third party became sufficiently powerful it could result in another center squeeze with the Labor candidate playing the role of the centrist who gets squeezed out. This could hand victory to the Liberals in an electorate in which Labor would ordinarily be favored, with the third-party candidate playing a spoiler role despite making it to the final round. Because of this possibility, it would be strategically sound for voters on the Left to rank Labor first, even if they slightly prefer this third party.
(This requires the third party to be at least somewhat viable. One clear prediction is that a fear of the center squeeze should not diminish support for parties that are not remotely viable; this is confirmed by the fact that the Animal Justice Party received 0.8% of the first preferences.)
If third parties are being harmed by either the center squeeze or a rational fear of the center squeeze, it would mean that switching to a single-winner voting method that is less susceptible to the center squeeze (such as STAR, Approval, or a Condorcet method) would make third parties more viable.
The largest third party is the Australian Greens, who received 10.4% of first preferences. Before writing this, I had expected that this total was artificially reduced by strategic voters who didn’t want to give the election to the right-wing Liberals by voting for the Greens over Labor, but the data contradict this. Nationwide, 82.2% of Green voters preferred Labor to the Liberals; however, roughly the same fraction (82.7%) of Labor voters preferred the Greens to the Liberals.
If Labor voters were split 50/50 between the Greens and the Liberals for their second choice, then it would be entirely reasonable for voters on the Left to worry about a center squeeze handing victory to the Liberals. However, with the same fraction of Labor voters preferring the Greens to the Liberals as Green voters preferring Labor to the Liberals, such a center squeeze is impossible. When the Green candidate makes it to the final round to face the Liberal, it means the Green would fare better in the final round than the Labor candidate would. Also, Liberals tend to prefer Labor to the Greens (62.2% to 37.8%), so there’s no way that the Greens are secretly a centrist party that is systematically getting shut out by the center squeeze.
(It is possible to have a center squeeze given such preferences, but not in a way that harms the Green party. Labor would be the “centrist” party getting squeezed out, but the Greens would still win. There would be an incentive for strategic voting, but it would be for Liberal voters to dishonestly rank the Labor candidate above the Liberal to prevent the Green from winning. There is no incentive for voters on the Left to vote dishonestly.)
Nick Xenophon Team
The other third party which received a glimmer of electoral success in the past two elections is the Nick Xenophon Team (NXT), a centrist party in South Australia in 2016. Being a centrist party, the question is whether NXT suffered defeats as a result of the center squeeze. There were three electorates in which NXT faced off against the Liberals in the final round; in those, NXT received 71.18%, 71.19%, and 79.95% of the vote from Labor supporters (NXT won the last of those races). There was also one electorate in which NXT faced off against Labor in the final round, in which 57.91% of Liberal voters preferred NXT to Labor. The electorate in which it is the most probable that NXT was the Condorcet winner but lost due to the center squeeze is Boothby. (A Condorcet winner is a candidate who would defeat any other candidate in a head-to-head matchup.)
Here are the vote totals in Boothby in the penultimate round:
- Liberal: 44.06%
- Labor: 30.51%
- NXT: 25.43%
NXT would have needed to be preferred by 55.76% of Liberal voters (over Labor) and by 80.53% of Labor voters (over the Liberal) to be the Condorcet winner. They probably would have had enough support from the Liberal voters. However, in order to have enough support from Labor voters, they would have needed to receive more support from them in Boothby than in any of the three electorates in which data are available. This is possible (80.53% isn’t that much greater than 79.95%), but still improbable, and even if NXT was the Condorcet winner in Boothby it would have been by far smaller margins than the Condorcet winner had in the center squeeze election in Burlington, Vermont. There are a couple of other electorates in which NXT might have been the Condorcet winner, but it seems extremely unlikely.
That said, the Nick Xenophon Team only won a single seat in the House (as compared to three seats in the much-smaller Senate). A small chance of losing one seat to a center squeeze doesn’t sound like much, but it could have been a relevant fraction of NXT’s overall success. And the important question is not, “Did a center squeeze occur in a specific election in Boothby?”, but rather, “What fraction of potential victories do parties like NXT miss out on as a result of the center squeeze?”. This cannot be answered with any confidence without a lot more data, but it seems unlikely that it’s greater than one-half. As it happened, the Nick Xenophon Team was not greatly harmed by the center squeeze. There’s an argument to be made that this was a matter of luck and that parties like NXT would be harmed a lot more in expectation, but it nevertheless seems inadequate as an explanation for the two-party system in Australia’s House.
IRV and third-party visibility
Another weakness of IRV is poor third-party visibility. IRV transparently shows the number of voters who have a third party as their first choice. It does not reveal how many voters think a third-party candidate is almost (but not quite) as good as a major party candidate since the second choices of voters who prefer one of the top two candidates remain invisible throughout tabulation. Other voting methods do better on this front; if I was a Labor voter who thought the Green candidate was nearly as good, I’d probably vote for both of them in an Approval election or give the Green a 4 in a STAR election. But in IRV, all I can do (without voting dishonestly) is mark the Green as my second choice which will never be revealed.
I think third-party visibility is a very important consideration in general, but I’m skeptical that it plays as large of a role in Australia as it would in the United States. That is because the Australian Senate already gives third parties (especially the Greens) far more visibility than they get in the US. Maybe it would help the Green party to come in second in Left-leaning electorates in which the Green candidate would beat the Liberal head-to-head despite the Green only receiving 10% of first preferences. It’s plausible that the Greens would become seen as the main alternative to Labor in such electorates, and that this would eventually result in them winning more seats. I find it more likely that Australian voters have reached a reasonably informed judgment that they simply prefer Labor, and I doubt that added visibility would help the Greens dramatically. But it could still help some.
- The Australian House of Representatives does not have a two-party system due to the flaws of IRV, it has a two-party system because third parties are legitimately unpopular. Third parties would have few additional victories with Approval, Score, or STAR voting.
- Center squeezes are less common than you might expect. The simplest and most intuitive model of politics is that candidates are on a one-dimensional Left-Right spectrum; in this model center squeezes are basically ubiquitous. The Australian data show that this model is largely wrong; instead, it’s extremely rare to have over 90% of a minor party’s supporters prefer one major party to another, and similarly with regards to preferences of major party supporters between a minor party and the opposing major party. Intuitively, if 45% of voters prefer a leftist, 45% prefer a rightist, and 10% prefer a centrist, we might expect the centrist to be the Condorcet winner. Wrong. In practice, enough left-wing voters will prefer the rightist (and vice versa) that the centrist should never win.
- The idea that other single-winner voting methods would break the two-party system whereas IRV wouldn’t is dubious. STAR and Approval Voting yield far more visibility for minor parties than IRV, but this isn’t as much of an issue in Australia where minor parties are represented in the Senate. Maybe third-party visibility is important in Australia insofar as the visibility gained in the Senate allows the Greens to get 10% of the vote instead of something like 3%, and it’s likely that using STAR or Approval instead of IRV could yield a similar effect. But the difference between 3% and 10% support is pretty slim in single-winner elections. Either way, you’ll lose almost everywhere.
- Insofar as third parties want to pursue electoral reforms to increase their odds of winning elections, they should focus primarily on proportional representation. In 2016, the Greens won 12% of Senate seats, the Nick Xenophon Team won 4% (a full quarter of the South Australia Senate seats), and the One Nation party won 5%. But none of these parties won more than a single seat in the 151-member House.
IRV doesn’t stop two-party domination, and other single-winner voting methods likely wouldn’t stop it either. Proportional representation does.
Appendix: Races in which third parties finished in the top two
Australia does not report the second choices of voters whose preferred candidate reached the final round, so to determine the fraction of Labor and Liberal voters who preferred the Greens (or the Nick Xenophon team) to the opposing major party I had to look at electorates in which the minor party finished in the top two and assume that voters elsewhere would behave similarly. (I checked every electorate for 2016 and 2019.) Notably, this limitation makes it impossible to identify a center squeeze in races involving independents or parties that fail to make the top two in multiple races. I have listed all the races in 2016 and 2019 in which a candidate from any party other than Labor, the Liberals, or the Nationals finished in the top two (the list does not include independents).
Here are the House races I found in which the Green candidate made it to the final round:
- Melbourne 2019: Greens beat Liberals in the final round. Greens won 80.40% of the Labor vote, Liberals won 19.60%
- Kooyong 2019: Greens lost to the Liberals in the final round. Greens won 83.38% of the Labor vote, Liberals won 16.62%
- Cooper 2019: Greens lost to Labor in the final round. Greens won 33.86% of the Liberal vote, Labor won 66.14%
- Wills 2019: Greens lost to Labor in the final round. Greens won 48.52% of the Liberal vote, Labor won 51.48%
- Grayndler 2019: Greens lost to Labor in the final round. Greens won 39.35% of the Liberal vote, Labor won 60.65%
- Grayndler 2016: Greens lost to Labor in the final round. Greens won 29.81% of the Liberal vote, Labor won 70.19%
- Warringah 2016: Greens lost to the Liberals in the final round. Greens won 82.44% of the Labor vote, Liberals won 17.56%
- Batman 2016: Greens lost to Labor in the final round. Greens won 36.38% of the Liberal vote, Labor won 63.62%
- Higgins 2016: Greens lost to the Liberals in the final round. Greens won 84.18% of the Labor vote, Liberals won 15.82%
- Melbourne 2016: Greens beat Liberals in the final round. Greens won 83.30% of the Labor vote, Liberals won 16.70%
- Wills 2016: Greens lost to Labor in the final round. Greens won 38.73% of the Liberal vote, Labor won 61.27%
On average, the Green candidate was preferred over the Liberal by 82.74% of Labor voters and was preferred over the Labor candidate by 37.78% of Liberal voters.
Nick Xenophon Team:
Center Alliance (the new name of the Nick Xenophon Team):
Katter’s Australia Party: