Used historical data from the Australian Bureau of Statistics to develop 3 methods for predicting the future household net worth of Australia.
Considered a variety of techniques for forecasting including: log-linear regression, percentage change extrapolation, and an ARIMA time series.
Evaluated the best prediction for the next 10 years of household net worth growth until 2035.
ARIMA achieved the best in-sample fit by capturing short-term shocks like the 2008 financial crisis and COVID-19, leading to a more conservative 2035 forecast (~$23.3T) compared to the log-linear and percentage-change models (~$31T).
Highlighted the tradeoff between models that extrapolate long-run exponential growth versus those that adapt to recent structural slowdowns, and discussed implications for which forecast is most defensible for planning purposes.