Predicting the 2022 Australian Federal Election with Actuarial Methodologies

Abstract

Election forecasting has, for a long time, been considered a major tool used by voters and politicians alike to understand the electoral mood. In Australia, tools such as opinion polling even drive many policy decisions made by governments. Yet, these same tools have been misleading and/or wrong in recent elections. The consequences of this have meant large fiscal loss for bookmakers, rash policy decisions, inappropriate leadership changes, and, recently, erosion of trust in the democratic process. Hence, we must look at new ways to analyse elections, and use a more appropriate system to mimic the actual voting system in the country where the election is being held. For example, in representative democracies with strong minor party representation like Australia, any forecasting method used must not use two-party preferred on a national basis, and, instead, analyse preference flows, and first preferences, for each seat. This study uses actuarial techniques to analyse probabilities of parties winning each of the 151 seats in the 2022 Election, and then calculates final party totals based on these probabilities.
 
 This study aims to accurately predict the 2022 Election using a methodology which will enable a final forecast which is not misleading and is, in fact, appropriate for a representative democracy. Data used includes first preferences and Preference Flow data from the 2019 Election, Demographic data from the 2016 Census, Economic Data across 2019-22, and Opinion Polls from The Australian and Roy Morgan. Further, political insight will be used to discuss these results and the potential variances.