Macroeconometric Models of the Australian Economy: A Comparative Analysis

For more than a decade, macroeconometric models have been used in Australia as a tool in policy analysis and to help with the task of forecasting major economic variables. Despite this, not much work has been done to compare these models. This paper helps to fill the gap in this aspect of model evaluation. It examines the theoretical structure and dynamic adjustment mechanisms in four macroeconometric models of the Australian economy. The models examined are: two models from the Reserve Bank of Australia; the Treasury–Australian Bureau of Statistics model; and Professor J. W. Nevile’s fiscal policy model. Each model is first analysed in terms of a simple diagram which concentrates on the modelling of markets and the interactions between these markets. This diagrammatic approach helps the reader to identify the role of key theoretical mechanisms in the models. The simulation properties of the models are examined with two criteria: how well each model tracks the economy; and the dynamic adjustments within each model in response to changes in certain exogenous variables. The counterfactual simulation experiments include: 1. a sustained increase in real current government spending under three alternative regimes for monetary policy; 2. a sustained increase in real exports; 3. a sustained rise in world prices; and 4. a sustained reduction in real award wages. The responses of the models are broadly similar, when variables influencing demand are changed. However, when impulses other than conventional demand shocks are imposed, the results are very different in each model. The conclusion of the paper is that more theoretical and applied research is needed on aspects of macroeconometric model building. In particular, more work is needed on the role of: prices (including wages, interest rates and the exchange rate); financial interactions; dynamic adjustment processes; and the open economy aspects of the Australian economy. The approach of the second generation Reserve Bank model appears to be a step in this direction. In general, however, more work is needed on model validation based on a wider range of criteria than have been used in the past.