Dynamic and state-dependent rental-growth connectedness across Australian capital cities

Abstract

Purpose This study aims to examine the degree, direction, time variation and state dependence of rental-growth connectedness among Australia’s eight capital cities. Design/methodology/approach Advertised asking rents for three-bedroom houses are observed four times monthly from August 2009 to October 2024. Generalised Diebold–Yilmaz measures identify total, directional and net predictive connectedness; rolling estimation tracks changing city roles; and quantile connectedness distinguishes lower-tail, median and upper-tail states. Sensitivity checks vary lags, horizons, rolling windows, frequency, transformation and tail treatment. Findings Full-sample connectedness is 30.8%, indicating incomplete integration, with Sydney and Canberra as net transmitters. City roles change over time: Melbourne and Adelaide lead during the acute-pandemic phase, while Darwin and Hobart lead afterwards. Average connectedness declines descriptively from 30.5% before the pandemic to 29.1% during the acute-pandemic phase and 28.8% afterwards, despite a pronounced rolling peak around 2020–2021. Connectedness is substantially higher in the lower and upper tails (77.3% and 78.5%) than at the median (32.1%). Its magnitude is sensitive to lags, rolling windows, frequency and tail treatment. Research limitations/implications The measures capture predictive interdependence, including shared national influences, rather than structurally identified causal transmission. Practical implications Network-based monitoring should combine rental growth with vacancy, migration, supply and affordability indicators and adapt interventions to changing city roles and market states. Originality/value The study provides integrated mean, directional, rolling and quantile evidence for the Australian capital-city rental system.