Assessing the cohort-component method in Australia’s official small area population estimates

Abstract Small-area population estimates have long been a critical part of the work program of National Statistical Offices. The 2021 Australian census and following ‘rebasing’ process of population estimates from 1 July 2016 to 30 June 2021 marked five years of small area data produced using the cohort-component method. This period oversaw wider challenges however, with the COVID-19 pandemic interrupting established patterns of internal and overseas migrant flows between March 2020 and June 2021 of this period. This paper sets out to assess the components method after its first five years in small area population estimates. Analysis of percentage error between un-rebased and rebased population estimates was undertaken for Statistical Area Level 2 geographies at 30 June 2021. A selection of measures capturing bias, accuracy, and the effects of larger errors were analysed between higher-order geographies and across a range of demographic profiles. The findings of this analysis show that the cohort-component method performs better with medium-sized, slower growing or more stable populations, while larger errors tend to result among faster growing populations and those at each size extreme. This analysis provides important insights for users of Australian small area population estimates to understand data dynamics at finer geographical scales.