What Does a Mixed‐Frequency Multivariate Beveridge‐Nelson Decomposition Tell us about the Australian Output Gap?*

Economic indicators inform the assessment of economic slack for central banks, yet traditional output gap estimates are often limited by the substantial reporting lags of quarterly GDP. This paper extends a mixed‐frequency Bayesian vector autoregressive (MF‐BVAR) framework to the Australian economy by applying a multivariate Beveridge–Nelson (BN) decomposition. Compared to existing applications to the US economy, the modelling setup introduces three distinct contributions: the incorporation of a block‐exogenous foreign sector, explicit accounting for COVID‐19 outliers and the integration of a weekly indicator in addition to monthly and quarterly indicators. I find that the estimated output gap is broadly consistent with central bank estimates. Informational decomposition results reveal that every variable in the model contributes non‐negligibly to the overall estimate, with foreign variables and the Trade Weighted Index (TWI) providing substantial shares of useful information. Domestically, aggregate hours worked provide a more significant contribution than the headline unemployment rate, suggesting more relevance of the intensive margin of the Australian labour market than its extensive margin. Furthermore, sectoral aggregation highlights the labour sector as the primary source of information for the output gap, while TWI alone provides informational value nearly equivalent to the entire financial or macroeconomic sectors combined. Shock decompositions reveal that while domestic shocks drive the majority of cyclical fluctuations on average, the Global Financial Crisis was largely attributable to foreign shocks. Finally, while weekly TWI allows for more timely updates, it does not improve estimates compared to a model utilising a monthly TWI instead.