Lessons from the Australian election for AV in the UK

The Australian elections are heading towards an almost perfect tie. At the time of writing, the governing Labor party had won 70 seats, with the opposition Liberal-National Coalition gaining 72, while independents had won four seats and the Greens one. This meant that, with three seats still outstanding, no party would cross the threshold of overall control (76 seats) and a coalition deal would have to be struck between one of the larger parties, the independents and potentially the Greens.

The results in terms of seats belie the fact that the Coalition had obtained 43.5% of ‘primary votes’, compared with 38.6% for Labour and 11.4% for the Greens. So based on vote share alone, the Coalition [capital c] ought to be entitled to try to form a coalition [small c]. ‘Primary votes’ are what we’d call over here ‘ first-preference votes’: Australia uses essentially the same preferential voting system that we’re going to have the option of adopting in the referendum next May, and which is known in the UK as the Alternative Vote (AV). The only difference is that, in Australia, voters are obliged to express a ranked preference for all the candidates in the election; whereas, in the UK, voters will be allowed to rank only the candidates they actually want to vote for.

In my view, the Australian results demonstrate once again just how bad a system AV is and how it favours two-party politics, or two-and-a-half-party politics as it would be in the UK. This is because people’s higher-preference votes for smaller parties inevitably end up being eliminated in the counting process, and only those voters’ lower-preference votes for the major parties are ultimately used to determine the result. This tendency is exaggerated even further in Australia by the fact that you are obliged to exhaust the ballot (express a preference for all the candidates), so that almost every vote comes down to a contest between the two largest parties.

Also, the fact that the Greens achieved their best-ever result, and yet their 11.4% of votes translated into only one seat, shows how unfair and disproportional the system is. What essentially happened in this election is that first-preference votes for the Greens were transferred almost entirely to the Labor Party in the preference count, which frequently enabled the Labor Party to overtake the Coalition, which had obtained more primary votes than Labor in many seats. This is how Labor managed to almost achieve parity with the Coalition on seats despite its much lower share of primary votes.

In the UK, this mechanism is likely to favour the Tories and the Lib Dems at the expense of Labour. In Tory-Labour fights – in England, this is mainly in the Midlands and the North – it’s quite conceivable that more Lib Dem voters would put down the Tories as their second preference rather than Labour, especially if those two parties are still in a coalition. So if Labour is only narrowly ahead of the Conservatives on first-preference votes, it’s quite possible the Tories could leap-frog Labour to victory thanks to the Lib Dem second preferences. As a consequence of this threat, I’ve suggested elsewhere that Labour voters in close Tory-Labour elections held using AV should consider voting tactically and putting the Lib Dems down as their first choice, in order to ensure that the final two parties left in the count are the Conservatives and the Lib Dems, and so enable the Lib Dems to beat the Tories based on the second preferences of Labour voters. This example demonstrates how, despite what is claimed for it, AV actually encourages some rather perverse tactical-voting scenarios.

Meanwhile, in Tory-Lib Dem fights – e.g. in southern England – the Lib Dems are more likely to benefit from this mechanism as Labour voters’ second or final preferences would be expected to be for the Lib Dems, if anything, rather than the Tories. Now, you could say that this aspect of AV is actually fairer than allowing the election to be decided purely on the highest plurality (i.e. based on the largest minority of ‘first preferences’ only, which is effectively what First Past the Post does in most seats). But if more people genuinely want one party to win rather than any other, isn’t that a fairer result, even if it produces disproportional outcomes at a national level? AV is arguably better at producing the ‘Condorcet winner’ (the candidate that would be preferred by most voters overall to any other candidate in a straight one-to-one comparison) but not so good at indicating the candidate that is strongly preferred by the greatest number, which FPTP in theory does better – although FPTP results are distorted by tactical voting. These problems do not exist in either of the ARV or TMPR voting systems discussed in previous posts: ARV always awards the win to the most popular candidate overall, regardless of whether this is the Condorcet winner or not; and TMPR gives the seats to both the Condorcet winner and the party that is strongly preferred by most voters – or both to one party, if they are the same.

Be that as it may, as in Australia, we’d effectively end up with two-party politics in England using AV, except the two parties in the North and Midlands would be the Tories and Labour (unless tactical voting for the Lib Dems by Labour voters of the kind I suggested above kicked in), and the two parties in southern England would be the Tories and the Lib Dems. This would effectively consolidate the three parties’ stranglehold over English politics while squeezing out the smaller parties. The only way parties like the Greens and UKIP could win seats would be if there was a strong candidate from one of those parties that supporters of the other parties would vote for tactically, whether as their first or subsequent preference, in order to unseat the incumbent MP. This is in fact what happened in the Australian seat of Melbourne, won by the Greens yesterday, as first-preference supporters of the Coalition – with its notoriously hardline anti-Green leader – hypocritically transferred their subsequent preferences to the Greens in order to defeat the Labor candidate, who came top in terms of primary votes. This shows just how pernicious tactical voting can be under AV: the Greens benefiting from Coalition tactical votes designed to beat Labor, whereas normally Green voters transfer their vote to Labor.

So don’t believe it when people try to claim that AV eliminates tactical voting: far from it. Nor is it remotely proportional and, arguably, fair in terms of awarding the win to the most popular candidate in each constituency. You could argue that the overall result in Australia, in terms of seats, is proportional to the extent that, in most seats, it came down to a straight fight between the main left-of-centre and right-of-centre candidates, and that these two fundamental positions were evenly matched overall. But this does consolidate the dominance of only one left-of-centre and one right-of-centre party – or, in England, two left-of-centre parties and one right-of-centre party. And, on top of which, AV would perpetuate the electoral divisions between the different English ‘regions’, making Labour only a party of the Midlands and North, and the Lib Dems only a party of the South; while the Tories are the only real right-of-centre alternative nationwide.

No wonder the Tories were so keen to put AV, and not PR, into the coalition agreement! And perhaps there was some cynical calculation on the part of the Lib Dems to the effect that permanent three-party politics, which is the most likely consequence of AV, would at least assure they had a quasi-perpetual influence over Westminster’s unaccountable governance of England.