The Australian Vote: Transferable Voting, Its Limitations and Strengths

TL;DR

This paper analyzes Australian transferable voting systems AV and STV, examining their strengths, limitations, and paradoxes, with detailed modeling and empirical data to understand their behavior and properties.

Contribution

It provides a comprehensive analysis of AV and STV, including their non-monotonicities, relation to Condorcet winners, and how they balance majority rule and voter choice.

Findings

Non-monotonicities occur in less than 3.5% of scenarios

AV always elects a majority candidate or a Condorcet winner

STV can resemble largest-remainder list voting in certain cases

Abstract

The voting systems known as Alternative Vote (AV) and Single Transferable Vote (STV) are extensively used for elections in Australia, possibly more than in any other jurisdiction. Often proposed as superior alternatives to Plurality and other common systems, they are also criticised by theoreticians for their vulnerability to paradoxical outcomes. It is argued these 'transferable voting' systems in fact operate to balance competing desiderata that attempt to distil the common notions of 'majority rule' and 'voter choice'. Among positive characteristics of AV is that its outcome is always a majority runoff between two candidates with substantial voter support, and that a Condorcet winner who gains at least one-third of votes is always elected. It is shown that in situations where a Condorcet winner is not elected, there always exists a monotonic shift producing a Condorcet cycle without changing the election outcome. A detailed analysis of non-monotonicity paradoxes in three-way contests is provided, aided by a two-dimensional spatial model where each candidate has a one-in-four probability of gaining an absolute majority of primary votes while there is a similar one-in-four probability that no candidate wins a majority. On such a model it is shown non-monotonicities are latent in less than 3.5% of election scenarios (and conjectured to be less than 3%), and the close link between non-monotonicities and Condorcet cycles and instabilities is studied in more detail. Properties of multi-winner STV are also reviewed including a demonstration of the standard fact that STV becomes similar to a form of largest-remainder list voting in the special case where ballots are cast strictly by party allegiance. The argument that AV and STV exemplify the principle of voter choice while balancing absolute and relative concepts of majority-rule is made by reference to practice in Australian elections.